Vascular

The sound bites about these trials in the news have confused physicians and patients alike. But, as we have all experienced during this election year, to understand complex problems requires an in-depth analysis instead of a sound bite.

I was troubled by the results of the Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial,² in which more patients who were treated with an intense hemoglobin A_1c-lowering strategy died (mostly of macrovascular events) than those treated with a standard strategy. Older data showing a beneficial effect of glucose-lowering on the microvascular complications of diabetes are solid. I did not understand the mechanistic basis of the ACCORD results, unless the very aggressive therapy caused many hypoglycemic events with catecholamine surges, resulting in stroke or myocardial infarction, or whether a problem with a specific drug arose more often in the intensive-treatment group. There has been similar dialogue surrounding intensity of glucose control in critically ill inpatients³; here, the data suggest that hypoglycemic episodes may limit other benefits of aggressive treatment in the intensive care unit, such as reduced infection rates.

Not to be ignored is that the patients in all arms of the ACCORD trial fared far better than historical diabetic controls. The meticulous attention to management of blood pressure and LDL-C that all patients in the ACCORD trial received paid off. (If only we could do as well in our practices!) But what do we do about the sugar?

This large, well-done, ongoing trial deserves a detailed analysis for those of us who need to translate the conclusions regarding glucose control to our patients. This month in the Journal, I have invited Byron Hoogwerf, a clinical diabetologist, former internal medicine program director, well-published clinical trialist, and ACCORD investigator, to provide this analysis.⁴ His discussion is more detailed than what we often print purposefully, and it is well worth reading. Some issues simply can’t be understood as a sound bite.

The sound bites about these trials in the news have confused physicians and patients alike. But, as we have all experienced during this election year, to understand complex problems requires an in-depth analysis instead of a sound bite.

I was troubled by the results of the Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial,² in which more patients who were treated with an intense hemoglobin A_1c-lowering strategy died (mostly of macrovascular events) than those treated with a standard strategy. Older data showing a beneficial effect of glucose-lowering on the microvascular complications of diabetes are solid. I did not understand the mechanistic basis of the ACCORD results, unless the very aggressive therapy caused many hypoglycemic events with catecholamine surges, resulting in stroke or myocardial infarction, or whether a problem with a specific drug arose more often in the intensive-treatment group. There has been similar dialogue surrounding intensity of glucose control in critically ill inpatients³; here, the data suggest that hypoglycemic episodes may limit other benefits of aggressive treatment in the intensive care unit, such as reduced infection rates.

Not to be ignored is that the patients in all arms of the ACCORD trial fared far better than historical diabetic controls. The meticulous attention to management of blood pressure and LDL-C that all patients in the ACCORD trial received paid off. (If only we could do as well in our practices!) But what do we do about the sugar?

This large, well-done, ongoing trial deserves a detailed analysis for those of us who need to translate the conclusions regarding glucose control to our patients. This month in the Journal, I have invited Byron Hoogwerf, a clinical diabetologist, former internal medicine program director, well-published clinical trialist, and ACCORD investigator, to provide this analysis.⁴ His discussion is more detailed than what we often print purposefully, and it is well worth reading. Some issues simply can’t be understood as a sound bite.

The sound bites about these trials in the news have confused physicians and patients alike. But, as we have all experienced during this election year, to understand complex problems requires an in-depth analysis instead of a sound bite.

I was troubled by the results of the Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial,² in which more patients who were treated with an intense hemoglobin A_1c-lowering strategy died (mostly of macrovascular events) than those treated with a standard strategy. Older data showing a beneficial effect of glucose-lowering on the microvascular complications of diabetes are solid. I did not understand the mechanistic basis of the ACCORD results, unless the very aggressive therapy caused many hypoglycemic events with catecholamine surges, resulting in stroke or myocardial infarction, or whether a problem with a specific drug arose more often in the intensive-treatment group. There has been similar dialogue surrounding intensity of glucose control in critically ill inpatients³; here, the data suggest that hypoglycemic episodes may limit other benefits of aggressive treatment in the intensive care unit, such as reduced infection rates.

Not to be ignored is that the patients in all arms of the ACCORD trial fared far better than historical diabetic controls. The meticulous attention to management of blood pressure and LDL-C that all patients in the ACCORD trial received paid off. (If only we could do as well in our practices!) But what do we do about the sugar?

This large, well-done, ongoing trial deserves a detailed analysis for those of us who need to translate the conclusions regarding glucose control to our patients. This month in the Journal, I have invited Byron Hoogwerf, a clinical diabetologist, former internal medicine program director, well-published clinical trialist, and ACCORD investigator, to provide this analysis.⁴ His discussion is more detailed than what we often print purposefully, and it is well worth reading. Some issues simply can’t be understood as a sound bite.

How extensive a workup does a patient with high blood pressure need?

On one hand, we would not want to start therapy on the basis of a single elevated reading, as blood pressure fluctuates considerably during the day, and even experienced physicians often make errors in taking blood pressure that tend to falsely elevate the patient’s readings. Similarly, we would not want to miss the diagnosis of a potentially curable cause of hypertension or of a condition that increases a patient’s risk of cardiovascular disease. But considering that nearly one-third of adults in the United States have hypertension and that another one-fourth have prehypertension (formerly called high-normal blood pressure),¹ if we were to launch an intensive workup for every patient with high blood pressure, the cost and effort would be enormous.

Fortunately, for most patients, it is enough to measure blood pressure accurately and repeatedly, perform a focused history and physical examination, and obtain the results of a few basic laboratory tests and an electrocardiogram, with additional tests in special cases.

In this review we address four fundamental questions in the evaluation of patients with a high blood pressure reading, and how to answer them.

ANSWERING FOUR QUESTIONS

The goal of the hypertension evaluation is to answer four questions:

• Does the patient have sustained hypertension? And if so—
• Is the hypertension primary or secondary?
• Does the patient have other cardiovascular risk factors?
• Does he or she have evidence of target organ damage?

DOES THE PATIENT HAVE SUSTAINED HYPERTENSION?

It is important to measure blood pressure accurately, for several reasons. A diagnosis of hypertension has a measurable impact on the patient’s quality of life.² Furthermore, we want to avoid undertaking a full evaluation of hypertension if the patient doesn’t actually have high blood pressure, ie, systolic blood pressure greater than 140 mm Hg or diastolic pressure greater than 90 mm Hg. However, many people have blood pressures in the prehypertensive range (ie, 120–139 mm Hg systolic; 80–89 mm Hg diastolic). Many people in this latter group can expect to develop hypertension in time, as the prevalence of hypertension increases steadily with age unless effective preventive measures are implemented, such as losing weight, exercising regularly, and avoiding excessive consumption of sodium and alcohol.

The best position to use is sitting, as the Framingham Heart Study and most randomized clinical trials that established the value of treating hypertension used this position for diagnosis and follow-up.⁶

Proper patient positioning, the correct cuff size, calibrated equipment, and good inflation and deflation technique will yield the best assessment of blood pressure levels. But even if your technique is perfect, blood pressure is a dynamic vital sign, so it is necessary to repeat the measurement, average the values for any particular day, and keep in mind that the pressure is higher (or lower) on some days than on others, so that the running average is more important than individual readings. This leads to two final points about blood pressure measurement:

• Take it right, at least two times on any occasion
• Take it on at least two (preferably three) separate days.

Following up on blood pressure

After measuring the blood pressure, it is necessary to plan for follow-up readings, guided by both the blood pressure levels (Table 2) and your clinical judgment.

If the systolic and diastolic blood pressures fall into different categories, you should follow the recommendations for the shorter follow-up time.

IS THE HYPERTENSION PRIMARY OR SECONDARY?

Most patients with hypertension have primary (“essential”) hypertension and are likely to remain hypertensive for life. However, some have secondary hypertension, ie, high blood pressure due to an identifiable cause. Some of these conditions (and the hypertension that they cause) can be cured. For example, pheochromocytoma can be cured if found and removed. Other causes of secondary hypertension, such as parenchymal renal disease, are infrequently cured, and the goal is usually to control the blood pressure with drugs.

The sudden onset of severe hypertension in a patient previously known to have had normal blood pressure raises the suspicion of a secondary form of hypertension, as does the onset of hypertension in a young person (< 25 years) or an older person (> 55 years). However, these ages are arbitrary; with the increasing body mass index in young people, essential hypertension is now more commonly diagnosed in the third decade. And since systolic pressure increases throughout life, we can expect many older patients to develop essential hypertension.⁷ Indeed, current guidelines are urging us to pay more attention to systolic pressure than in the past.

WHAT IS THE PATIENT’S CARDIOVASCULAR RISK?

The relationship between blood pressure and risk of cardiovascular disease is linear, continuous, and independent of (though additive to) other risk factors.¹ For people 40 to 70 years old, each increment of either 20 mm Hg in systolic blood pressure or 10 mm Hg in diastolic blood pressure doubles the risk of cardiovascular disease across the entire range from 115/75 to 185/115 mm Hg.¹ If the patient smokes or has elevated cholesterol, other cardiovascular risk factors, or the metabolic syndrome, the risk is even higher.⁸

The usual goal of antihypertensive treatment is systolic pressure less than 140 mm Hg and diastolic pressure less than 90 mm Hg. However, the target is lower—less than 130/80 mm Hg—for those with diabetes⁹ or target organ damage such as heart failure or renal disease.^1,10 Thus, it is important to try to detect these conditions in the evaluation of the hypertensive patient.

Another reason it is important is that reducing such risk sometimes calls for using (or avoiding) antihypertensive drugs that are likely to alter these factors. For example, the use of beta-blockers in patients with a low level of high-density lipoprotein cholesterol (HDL-C) can lower HDL-C further.¹¹

DOES THE PATIENT HAVE TARGET ORGAN DAMAGE?

Target organ damage is very important to detect because it changes the goal of treatment from primary prevention of adverse target organ outcomes into the more challenging realm of secondary prevention. For example, if a patient has had a stroke, his or her chance of having another stroke in the next 5 years is about 20%. This is much higher than the risk in an average hypertensive patient without such a history. For such patients, the current guidelines¹ recommend the combination of a diuretic and an angiotensin-converting enzyme inhibitor, a combination shown to reduce the risk of a second stroke.¹² Thus, we need to discover whether the patient had a stroke in the first place.

HISTORY

• The duration (if known) and severity of the hypertension
• The degree of blood pressure fluctuation
• Concomitant medical conditions, especially cardiovascular or renal problems
• Dietary habits
• Alcohol consumption
• Tobacco use
• Level of physical activity
• A family history of hypertension, renal disease, cardiovascular problems, or diabetes mellitus
• Past medications, with particular attention to their side effects and their efficacy in controlling blood pressure
• Current medications, including over-the-counter preparations. One reason: non-steroidal anti-inflammatory drugs other than aspirin can decrease the efficacy of antihypertensive drugs, presumably through mechanisms that inhibit the effects of vasodilatory and natriuretic prostaglandins and potentiate those of angiotensin II.¹³

PHYSICAL EXAMINATION

The physical examination starts with measurement of height, weight, waist circumference, and blood pressure—in both arms and the leg if coarctation of the aorta is suspected. Measurements with the patient supine, sitting, and standing are usually taken at the first visit, though such an approach is more suited to a hypertension specialty clinic than a primary care setting, in which time constraints usually limit the blood pressure readings to two or three seated values. Most prospective data on the benefits of hypertension treatment are based on a seated blood pressure, so we favor that measurement for follow-up.

Special attention in the physical examination is directed to:

The retina (to assess the vascular impact of the high blood pressure). Look for arteriolar narrowing (grade 1), arteriovenous compression (grade 2), hemorrhages or exudates (grade 3), and papilledema.² Such findings not only relate to severity (higher grade = more severe blood pressure) but also predict future cardiovascular disease.¹⁴

The blood vessels. Bruits in the neck may indicate carotid stenosis, bruits in the abdomen may indicate renovascular disease, and femoral bruits are a sign of general atherosclerosis. Bruits also signal vascular stenosis and irregularity and may be a clue to vascular damage or future loss of target organ function. However, bruits may simply result from vascular tortuosity, particularly with significant flow in the vessel.

Also check the femoral pulses: poor or delayed femoral pulses are a sign of aortic coarctation. The radial artery is about as far away from the heart as the femoral artery; consequently, when palpating both sites simultaneously the pulse should arrive at about the same moment. In aortic coarctation, a palpable delay in the arrival of the femoral pulse may occur, and an interscapular murmur may be heard during auscultation of the back. In these instances, a low leg blood pressure (usually measured by placing a thigh-sized adult cuff on the patient’s thigh and listening over the popliteal area with the patient prone) may confirm the presence of aortic obstruction. When taking a leg blood pressure, the large cuff and the amount of pressure necessary to occlude the artery may be uncomfortable, and one should warn the patient about the discomfort before taking the measurement.

Poor or absent pedal pulses are a sign of peripheral arterial disease.

The heart (to detect gallops, enlargement, or both). Palpation may reveal a displaced apical impulse, which can indicate left ventricular enlargement. A sustained apical impulse may indicate left ventricular hypertrophy. Listen for a fourth heart sound (S4), one of the earliest physical findings of hypertension when physical findings are present. An S4 indicates that the left atrium is working hard to overcome the stiffness of the left ventricle. An S3 indicates an impairment in left ventricular function and is usually a harbinger of underlying heart disease. In some cases, lung rales can also be heard, though the combination of an S3 gallop and rales is an unusual office presentation in the early management of the hypertensive patient.

The lungs. Listen for rales (see above).

The lower extremities should be examined for peripheral arterial pulsations and edema. The loss of pedal pulses is a common finding, particularly in smokers, and is a clue to increased cardiovascular risk.

Strength, gait, and cognition. Perform a brief neurologic examination for evidence of remote stroke. We usually observe our patients’ gait as they enter or leave the examination room, test their bilateral grip strength, and assess their judgment, speech, and memory during the history and physical examination.

A great deal of research has linked high blood pressure to future loss of cognitive function,¹⁵ and it is useful to know that impairment is present before beginning treatment, since some patients will complain of memory loss after starting antihypertensive drug treatment.

LABORATORY EVALUATION

Routine tests

The routine evaluation of hypertensive patients should include, at a minimum:

• A hemoglobin or hematocrit measurement
• Urinalysis with microscopic examination
• Serum electrolyte concentrations
• 
  Serum creatinine concentrations
• Serum glucose concentration
• A fasting lipid profile
• A 12-lead electrocardiogram (Table 5).

Nonroutine tests

In some cases, other studies may be appropriate, depending on the clinical situation, eg:

• Serum uric acid in those with a history of gout, since some antihypertensive drugs (eg, diuretics) may increase serum uric acid and predispose to further episodes of gout
• Serum calcium in those with a personal or family history of kidney stones, to detect subtle parathyroid excess
• Thyroid-stimulating hormone or other thyroid studies if the history suggests thyroid excess, or if a thyroid nodule is discovered
• Limited echocardiography, which is more sensitive than electrocardiography for detecting left ventricular hypertrophy.

We sometimes use echocardiography if the patient is overweight but seems motivated to lose weight. In these cases we might not start drug therapy right away, choosing rather to wait and see if the patient can lose some weight (which might lower the blood pressure and make drug therapy unnecessary)—but only if the echocardiogram shows that he or she does not have left ventricular hypertrophy.

We also use echocardiography in patients with white-coat hypertension (see below), in whom office pressures are consistently high but whom we have elected to either not treat or not alter treatment. In these cases the echocardiogram serves as a “second opinion” about the merits of not altering therapy and supports this decision when the left ventricular wall thicknesses are normal (and remain so during long-term follow-up). In cases of suspected white-coat hypertension, home or ambulatory blood pressure monitoring is valuable to establish or exclude this diagnosis.¹

Urinary albumin excretion. Microalbuminuria is an early manifestation of diabetic nephropathy and hypertension. Although routine urine screening for microalbuminuria is typically done in the management of diabetes, it is still not considered a standard of care, though the growing literature on its role as a cardiovascular risk predictor^16–18 and its value as a therapeutic target in diabetes^19,20 make it an attractive aid in the overall assessment of patients with hypertension.

Plasma renin activity and serum aldosterone concentrations are useful in screening for aldosterone excess, but are usually reserved as follow-up tests in patients with either hypokalemia or failure to achieve blood pressure control on a three-drug regimen in which at least one drug is a diuretic.^1,21

Of note, primary aldosteronism is not as rare as previously thought. In a study of patients referred to hypertension centers, 11% had primary aldosteronism according to prospective diagnostic criteria, almost 5% had curable aldosterone-producing adenomas, and 6% had idiopathic hyperaldosteronism.²²

A search for secondary forms of hypertension is usually considered in patients with moderate or severe hypertension that does not respond to antihypertensive agents. Another situation is in hypertensive patients younger than 25 years, since curable forms of hypertension are more common in this age group. In older patients, the prevalence of secondary hypertension is lower and does not justify the costs and effort of routine elaborate workups unless there is evidence from the history, physical examination, or routine laboratory work for suspecting its presence. An exception to this rule is the need to exclude atherosclerotic renovascular hypertension in an elderly patient. This cause of secondary hypertension is common in the elderly and may be amenable to therapeutic intervention.²⁶

WHEN TO CONSIDER HOME OR AMBULATORY MONITORING

Suspected white-coat hypertension

Blood pressure can be influenced by an environment such as an office or hospital clinic. This has led to the development of ambulatory blood pressure monitors and more use of self-measurement of blood pressure in the home. Blood pressure readings with these techniques are generally lower than those measured in an office or hospital clinic. These methods make it possible to screen for white-coat hypertension. In 10% to 20% of people with hypertensive readings, the blood pressure may be elevated persistently only in the presence of a physician.²⁸ When measured elsewhere, including at work, the blood pressure is not elevated in those with the white-coat effect. Although this response may become less prominent with repeated measurements, it occasionally persists in the office setting, sometimes for years in our experience.

Suspected nocturnal hypertension (’nondipping’ status)

Ambulatory blood pressure is also helpful to screen for nocturnal hypertension. Evidence is accumulating to suggest that hypertensive patients whose pressure remains relatively high at night (“nondippers,” ie, those with less than a 10% reduction at night compared with daytime blood pressure readings) are at greater risk of cardiovascular morbidity than “dippers” (those whose blood pressure is at least 10% lower at night than during the day).²⁹

An early morning surge

Ambulatory monitoring can also detect morning surges in systolic blood pressure,³⁰ a marker of cerebrovascular risk. Generally, these patients have an increase of more than 55 mm Hg in systolic pressure between their sleeping and early-hour waking values, and we may wish to start or alter treatment specifically to address these high morning systolic values.³¹

‘PIPESTEM’ VESSELS AND PSEUDOHYPERTENSION

Occasionally, one encounters patients with vessels that are stiff and difficult to compress. If the pressure required to compress the brachial artery and stop audible blood flow with a standard blood pressure cuff is greater than the actual blood pressure within the artery as measured invasively, the condition is called pseudohypertension. The stiffness is thought to be due to calcification of the arterial wall.

A way to check for this condition is to inflate the cuff to at least 30 mm Hg above the palpable systolic pressure and then try to “roll” the brachial or radial artery underneath your fingertips, a procedure known as Osler’s maneuver.³² If you feel something that resembles a stiff tube reminiscent of the stem of a tobacco smoker’s pipe (healthy arteries are not palpable when empty), the patient may have pseudohypertension. However, the specificity of Osler’s maneuver has been questioned, particularly in hospitalized elderly patients.³³

Pseudohypertension is important because the patients in whom it occurs, usually the elderly or the chronically ill (with diabetes or chronic kidney disease), are prone to orthostatic or postural hypotension, which may be aggravated by increasing their antihypertensive treatment on the basis of a cuff pressure that is actually much higher than the real blood pressure.³³

How extensive a workup does a patient with high blood pressure need?

On one hand, we would not want to start therapy on the basis of a single elevated reading, as blood pressure fluctuates considerably during the day, and even experienced physicians often make errors in taking blood pressure that tend to falsely elevate the patient’s readings. Similarly, we would not want to miss the diagnosis of a potentially curable cause of hypertension or of a condition that increases a patient’s risk of cardiovascular disease. But considering that nearly one-third of adults in the United States have hypertension and that another one-fourth have prehypertension (formerly called high-normal blood pressure),¹ if we were to launch an intensive workup for every patient with high blood pressure, the cost and effort would be enormous.

Fortunately, for most patients, it is enough to measure blood pressure accurately and repeatedly, perform a focused history and physical examination, and obtain the results of a few basic laboratory tests and an electrocardiogram, with additional tests in special cases.

In this review we address four fundamental questions in the evaluation of patients with a high blood pressure reading, and how to answer them.

ANSWERING FOUR QUESTIONS

The goal of the hypertension evaluation is to answer four questions:

• Does the patient have sustained hypertension? And if so—
• Is the hypertension primary or secondary?
• Does the patient have other cardiovascular risk factors?
• Does he or she have evidence of target organ damage?

DOES THE PATIENT HAVE SUSTAINED HYPERTENSION?

It is important to measure blood pressure accurately, for several reasons. A diagnosis of hypertension has a measurable impact on the patient’s quality of life.² Furthermore, we want to avoid undertaking a full evaluation of hypertension if the patient doesn’t actually have high blood pressure, ie, systolic blood pressure greater than 140 mm Hg or diastolic pressure greater than 90 mm Hg. However, many people have blood pressures in the prehypertensive range (ie, 120–139 mm Hg systolic; 80–89 mm Hg diastolic). Many people in this latter group can expect to develop hypertension in time, as the prevalence of hypertension increases steadily with age unless effective preventive measures are implemented, such as losing weight, exercising regularly, and avoiding excessive consumption of sodium and alcohol.

The best position to use is sitting, as the Framingham Heart Study and most randomized clinical trials that established the value of treating hypertension used this position for diagnosis and follow-up.⁶

Proper patient positioning, the correct cuff size, calibrated equipment, and good inflation and deflation technique will yield the best assessment of blood pressure levels. But even if your technique is perfect, blood pressure is a dynamic vital sign, so it is necessary to repeat the measurement, average the values for any particular day, and keep in mind that the pressure is higher (or lower) on some days than on others, so that the running average is more important than individual readings. This leads to two final points about blood pressure measurement:

• Take it right, at least two times on any occasion
• Take it on at least two (preferably three) separate days.

Following up on blood pressure

After measuring the blood pressure, it is necessary to plan for follow-up readings, guided by both the blood pressure levels (Table 2) and your clinical judgment.

If the systolic and diastolic blood pressures fall into different categories, you should follow the recommendations for the shorter follow-up time.

IS THE HYPERTENSION PRIMARY OR SECONDARY?

Most patients with hypertension have primary (“essential”) hypertension and are likely to remain hypertensive for life. However, some have secondary hypertension, ie, high blood pressure due to an identifiable cause. Some of these conditions (and the hypertension that they cause) can be cured. For example, pheochromocytoma can be cured if found and removed. Other causes of secondary hypertension, such as parenchymal renal disease, are infrequently cured, and the goal is usually to control the blood pressure with drugs.

The sudden onset of severe hypertension in a patient previously known to have had normal blood pressure raises the suspicion of a secondary form of hypertension, as does the onset of hypertension in a young person (< 25 years) or an older person (> 55 years). However, these ages are arbitrary; with the increasing body mass index in young people, essential hypertension is now more commonly diagnosed in the third decade. And since systolic pressure increases throughout life, we can expect many older patients to develop essential hypertension.⁷ Indeed, current guidelines are urging us to pay more attention to systolic pressure than in the past.

WHAT IS THE PATIENT’S CARDIOVASCULAR RISK?

The relationship between blood pressure and risk of cardiovascular disease is linear, continuous, and independent of (though additive to) other risk factors.¹ For people 40 to 70 years old, each increment of either 20 mm Hg in systolic blood pressure or 10 mm Hg in diastolic blood pressure doubles the risk of cardiovascular disease across the entire range from 115/75 to 185/115 mm Hg.¹ If the patient smokes or has elevated cholesterol, other cardiovascular risk factors, or the metabolic syndrome, the risk is even higher.⁸

The usual goal of antihypertensive treatment is systolic pressure less than 140 mm Hg and diastolic pressure less than 90 mm Hg. However, the target is lower—less than 130/80 mm Hg—for those with diabetes⁹ or target organ damage such as heart failure or renal disease.^1,10 Thus, it is important to try to detect these conditions in the evaluation of the hypertensive patient.

Another reason it is important is that reducing such risk sometimes calls for using (or avoiding) antihypertensive drugs that are likely to alter these factors. For example, the use of beta-blockers in patients with a low level of high-density lipoprotein cholesterol (HDL-C) can lower HDL-C further.¹¹

DOES THE PATIENT HAVE TARGET ORGAN DAMAGE?

Target organ damage is very important to detect because it changes the goal of treatment from primary prevention of adverse target organ outcomes into the more challenging realm of secondary prevention. For example, if a patient has had a stroke, his or her chance of having another stroke in the next 5 years is about 20%. This is much higher than the risk in an average hypertensive patient without such a history. For such patients, the current guidelines¹ recommend the combination of a diuretic and an angiotensin-converting enzyme inhibitor, a combination shown to reduce the risk of a second stroke.¹² Thus, we need to discover whether the patient had a stroke in the first place.

HISTORY

• The duration (if known) and severity of the hypertension
• The degree of blood pressure fluctuation
• Concomitant medical conditions, especially cardiovascular or renal problems
• Dietary habits
• Alcohol consumption
• Tobacco use
• Level of physical activity
• A family history of hypertension, renal disease, cardiovascular problems, or diabetes mellitus
• Past medications, with particular attention to their side effects and their efficacy in controlling blood pressure
• Current medications, including over-the-counter preparations. One reason: non-steroidal anti-inflammatory drugs other than aspirin can decrease the efficacy of antihypertensive drugs, presumably through mechanisms that inhibit the effects of vasodilatory and natriuretic prostaglandins and potentiate those of angiotensin II.¹³

PHYSICAL EXAMINATION

The physical examination starts with measurement of height, weight, waist circumference, and blood pressure—in both arms and the leg if coarctation of the aorta is suspected. Measurements with the patient supine, sitting, and standing are usually taken at the first visit, though such an approach is more suited to a hypertension specialty clinic than a primary care setting, in which time constraints usually limit the blood pressure readings to two or three seated values. Most prospective data on the benefits of hypertension treatment are based on a seated blood pressure, so we favor that measurement for follow-up.

Special attention in the physical examination is directed to:

The retina (to assess the vascular impact of the high blood pressure). Look for arteriolar narrowing (grade 1), arteriovenous compression (grade 2), hemorrhages or exudates (grade 3), and papilledema.² Such findings not only relate to severity (higher grade = more severe blood pressure) but also predict future cardiovascular disease.¹⁴

The blood vessels. Bruits in the neck may indicate carotid stenosis, bruits in the abdomen may indicate renovascular disease, and femoral bruits are a sign of general atherosclerosis. Bruits also signal vascular stenosis and irregularity and may be a clue to vascular damage or future loss of target organ function. However, bruits may simply result from vascular tortuosity, particularly with significant flow in the vessel.

Also check the femoral pulses: poor or delayed femoral pulses are a sign of aortic coarctation. The radial artery is about as far away from the heart as the femoral artery; consequently, when palpating both sites simultaneously the pulse should arrive at about the same moment. In aortic coarctation, a palpable delay in the arrival of the femoral pulse may occur, and an interscapular murmur may be heard during auscultation of the back. In these instances, a low leg blood pressure (usually measured by placing a thigh-sized adult cuff on the patient’s thigh and listening over the popliteal area with the patient prone) may confirm the presence of aortic obstruction. When taking a leg blood pressure, the large cuff and the amount of pressure necessary to occlude the artery may be uncomfortable, and one should warn the patient about the discomfort before taking the measurement.

Poor or absent pedal pulses are a sign of peripheral arterial disease.

The heart (to detect gallops, enlargement, or both). Palpation may reveal a displaced apical impulse, which can indicate left ventricular enlargement. A sustained apical impulse may indicate left ventricular hypertrophy. Listen for a fourth heart sound (S4), one of the earliest physical findings of hypertension when physical findings are present. An S4 indicates that the left atrium is working hard to overcome the stiffness of the left ventricle. An S3 indicates an impairment in left ventricular function and is usually a harbinger of underlying heart disease. In some cases, lung rales can also be heard, though the combination of an S3 gallop and rales is an unusual office presentation in the early management of the hypertensive patient.

The lungs. Listen for rales (see above).

The lower extremities should be examined for peripheral arterial pulsations and edema. The loss of pedal pulses is a common finding, particularly in smokers, and is a clue to increased cardiovascular risk.

Strength, gait, and cognition. Perform a brief neurologic examination for evidence of remote stroke. We usually observe our patients’ gait as they enter or leave the examination room, test their bilateral grip strength, and assess their judgment, speech, and memory during the history and physical examination.

A great deal of research has linked high blood pressure to future loss of cognitive function,¹⁵ and it is useful to know that impairment is present before beginning treatment, since some patients will complain of memory loss after starting antihypertensive drug treatment.

LABORATORY EVALUATION

Routine tests

The routine evaluation of hypertensive patients should include, at a minimum:

• A hemoglobin or hematocrit measurement
• Urinalysis with microscopic examination
• Serum electrolyte concentrations
• 
  Serum creatinine concentrations
• Serum glucose concentration
• A fasting lipid profile
• A 12-lead electrocardiogram (Table 5).

Nonroutine tests

In some cases, other studies may be appropriate, depending on the clinical situation, eg:

• Serum uric acid in those with a history of gout, since some antihypertensive drugs (eg, diuretics) may increase serum uric acid and predispose to further episodes of gout
• Serum calcium in those with a personal or family history of kidney stones, to detect subtle parathyroid excess
• Thyroid-stimulating hormone or other thyroid studies if the history suggests thyroid excess, or if a thyroid nodule is discovered
• Limited echocardiography, which is more sensitive than electrocardiography for detecting left ventricular hypertrophy.

We sometimes use echocardiography if the patient is overweight but seems motivated to lose weight. In these cases we might not start drug therapy right away, choosing rather to wait and see if the patient can lose some weight (which might lower the blood pressure and make drug therapy unnecessary)—but only if the echocardiogram shows that he or she does not have left ventricular hypertrophy.

We also use echocardiography in patients with white-coat hypertension (see below), in whom office pressures are consistently high but whom we have elected to either not treat or not alter treatment. In these cases the echocardiogram serves as a “second opinion” about the merits of not altering therapy and supports this decision when the left ventricular wall thicknesses are normal (and remain so during long-term follow-up). In cases of suspected white-coat hypertension, home or ambulatory blood pressure monitoring is valuable to establish or exclude this diagnosis.¹

Urinary albumin excretion. Microalbuminuria is an early manifestation of diabetic nephropathy and hypertension. Although routine urine screening for microalbuminuria is typically done in the management of diabetes, it is still not considered a standard of care, though the growing literature on its role as a cardiovascular risk predictor^16–18 and its value as a therapeutic target in diabetes^19,20 make it an attractive aid in the overall assessment of patients with hypertension.

Plasma renin activity and serum aldosterone concentrations are useful in screening for aldosterone excess, but are usually reserved as follow-up tests in patients with either hypokalemia or failure to achieve blood pressure control on a three-drug regimen in which at least one drug is a diuretic.^1,21

Of note, primary aldosteronism is not as rare as previously thought. In a study of patients referred to hypertension centers, 11% had primary aldosteronism according to prospective diagnostic criteria, almost 5% had curable aldosterone-producing adenomas, and 6% had idiopathic hyperaldosteronism.²²

A search for secondary forms of hypertension is usually considered in patients with moderate or severe hypertension that does not respond to antihypertensive agents. Another situation is in hypertensive patients younger than 25 years, since curable forms of hypertension are more common in this age group. In older patients, the prevalence of secondary hypertension is lower and does not justify the costs and effort of routine elaborate workups unless there is evidence from the history, physical examination, or routine laboratory work for suspecting its presence. An exception to this rule is the need to exclude atherosclerotic renovascular hypertension in an elderly patient. This cause of secondary hypertension is common in the elderly and may be amenable to therapeutic intervention.²⁶

WHEN TO CONSIDER HOME OR AMBULATORY MONITORING

Suspected white-coat hypertension

Blood pressure can be influenced by an environment such as an office or hospital clinic. This has led to the development of ambulatory blood pressure monitors and more use of self-measurement of blood pressure in the home. Blood pressure readings with these techniques are generally lower than those measured in an office or hospital clinic. These methods make it possible to screen for white-coat hypertension. In 10% to 20% of people with hypertensive readings, the blood pressure may be elevated persistently only in the presence of a physician.²⁸ When measured elsewhere, including at work, the blood pressure is not elevated in those with the white-coat effect. Although this response may become less prominent with repeated measurements, it occasionally persists in the office setting, sometimes for years in our experience.

Suspected nocturnal hypertension (’nondipping’ status)

Ambulatory blood pressure is also helpful to screen for nocturnal hypertension. Evidence is accumulating to suggest that hypertensive patients whose pressure remains relatively high at night (“nondippers,” ie, those with less than a 10% reduction at night compared with daytime blood pressure readings) are at greater risk of cardiovascular morbidity than “dippers” (those whose blood pressure is at least 10% lower at night than during the day).²⁹

An early morning surge

Ambulatory monitoring can also detect morning surges in systolic blood pressure,³⁰ a marker of cerebrovascular risk. Generally, these patients have an increase of more than 55 mm Hg in systolic pressure between their sleeping and early-hour waking values, and we may wish to start or alter treatment specifically to address these high morning systolic values.³¹

‘PIPESTEM’ VESSELS AND PSEUDOHYPERTENSION

Occasionally, one encounters patients with vessels that are stiff and difficult to compress. If the pressure required to compress the brachial artery and stop audible blood flow with a standard blood pressure cuff is greater than the actual blood pressure within the artery as measured invasively, the condition is called pseudohypertension. The stiffness is thought to be due to calcification of the arterial wall.

A way to check for this condition is to inflate the cuff to at least 30 mm Hg above the palpable systolic pressure and then try to “roll” the brachial or radial artery underneath your fingertips, a procedure known as Osler’s maneuver.³² If you feel something that resembles a stiff tube reminiscent of the stem of a tobacco smoker’s pipe (healthy arteries are not palpable when empty), the patient may have pseudohypertension. However, the specificity of Osler’s maneuver has been questioned, particularly in hospitalized elderly patients.³³

Pseudohypertension is important because the patients in whom it occurs, usually the elderly or the chronically ill (with diabetes or chronic kidney disease), are prone to orthostatic or postural hypotension, which may be aggravated by increasing their antihypertensive treatment on the basis of a cuff pressure that is actually much higher than the real blood pressure.³³

How extensive a workup does a patient with high blood pressure need?

On one hand, we would not want to start therapy on the basis of a single elevated reading, as blood pressure fluctuates considerably during the day, and even experienced physicians often make errors in taking blood pressure that tend to falsely elevate the patient’s readings. Similarly, we would not want to miss the diagnosis of a potentially curable cause of hypertension or of a condition that increases a patient’s risk of cardiovascular disease. But considering that nearly one-third of adults in the United States have hypertension and that another one-fourth have prehypertension (formerly called high-normal blood pressure),¹ if we were to launch an intensive workup for every patient with high blood pressure, the cost and effort would be enormous.

Fortunately, for most patients, it is enough to measure blood pressure accurately and repeatedly, perform a focused history and physical examination, and obtain the results of a few basic laboratory tests and an electrocardiogram, with additional tests in special cases.

In this review we address four fundamental questions in the evaluation of patients with a high blood pressure reading, and how to answer them.

ANSWERING FOUR QUESTIONS

The goal of the hypertension evaluation is to answer four questions:

• Does the patient have sustained hypertension? And if so—
• Is the hypertension primary or secondary?
• Does the patient have other cardiovascular risk factors?
• Does he or she have evidence of target organ damage?

DOES THE PATIENT HAVE SUSTAINED HYPERTENSION?

It is important to measure blood pressure accurately, for several reasons. A diagnosis of hypertension has a measurable impact on the patient’s quality of life.² Furthermore, we want to avoid undertaking a full evaluation of hypertension if the patient doesn’t actually have high blood pressure, ie, systolic blood pressure greater than 140 mm Hg or diastolic pressure greater than 90 mm Hg. However, many people have blood pressures in the prehypertensive range (ie, 120–139 mm Hg systolic; 80–89 mm Hg diastolic). Many people in this latter group can expect to develop hypertension in time, as the prevalence of hypertension increases steadily with age unless effective preventive measures are implemented, such as losing weight, exercising regularly, and avoiding excessive consumption of sodium and alcohol.

The best position to use is sitting, as the Framingham Heart Study and most randomized clinical trials that established the value of treating hypertension used this position for diagnosis and follow-up.⁶

Proper patient positioning, the correct cuff size, calibrated equipment, and good inflation and deflation technique will yield the best assessment of blood pressure levels. But even if your technique is perfect, blood pressure is a dynamic vital sign, so it is necessary to repeat the measurement, average the values for any particular day, and keep in mind that the pressure is higher (or lower) on some days than on others, so that the running average is more important than individual readings. This leads to two final points about blood pressure measurement:

• Take it right, at least two times on any occasion
• Take it on at least two (preferably three) separate days.

Following up on blood pressure

After measuring the blood pressure, it is necessary to plan for follow-up readings, guided by both the blood pressure levels (Table 2) and your clinical judgment.

If the systolic and diastolic blood pressures fall into different categories, you should follow the recommendations for the shorter follow-up time.

IS THE HYPERTENSION PRIMARY OR SECONDARY?

Most patients with hypertension have primary (“essential”) hypertension and are likely to remain hypertensive for life. However, some have secondary hypertension, ie, high blood pressure due to an identifiable cause. Some of these conditions (and the hypertension that they cause) can be cured. For example, pheochromocytoma can be cured if found and removed. Other causes of secondary hypertension, such as parenchymal renal disease, are infrequently cured, and the goal is usually to control the blood pressure with drugs.

The sudden onset of severe hypertension in a patient previously known to have had normal blood pressure raises the suspicion of a secondary form of hypertension, as does the onset of hypertension in a young person (< 25 years) or an older person (> 55 years). However, these ages are arbitrary; with the increasing body mass index in young people, essential hypertension is now more commonly diagnosed in the third decade. And since systolic pressure increases throughout life, we can expect many older patients to develop essential hypertension.⁷ Indeed, current guidelines are urging us to pay more attention to systolic pressure than in the past.

WHAT IS THE PATIENT’S CARDIOVASCULAR RISK?

The relationship between blood pressure and risk of cardiovascular disease is linear, continuous, and independent of (though additive to) other risk factors.¹ For people 40 to 70 years old, each increment of either 20 mm Hg in systolic blood pressure or 10 mm Hg in diastolic blood pressure doubles the risk of cardiovascular disease across the entire range from 115/75 to 185/115 mm Hg.¹ If the patient smokes or has elevated cholesterol, other cardiovascular risk factors, or the metabolic syndrome, the risk is even higher.⁸

The usual goal of antihypertensive treatment is systolic pressure less than 140 mm Hg and diastolic pressure less than 90 mm Hg. However, the target is lower—less than 130/80 mm Hg—for those with diabetes⁹ or target organ damage such as heart failure or renal disease.^1,10 Thus, it is important to try to detect these conditions in the evaluation of the hypertensive patient.

Another reason it is important is that reducing such risk sometimes calls for using (or avoiding) antihypertensive drugs that are likely to alter these factors. For example, the use of beta-blockers in patients with a low level of high-density lipoprotein cholesterol (HDL-C) can lower HDL-C further.¹¹

DOES THE PATIENT HAVE TARGET ORGAN DAMAGE?

Target organ damage is very important to detect because it changes the goal of treatment from primary prevention of adverse target organ outcomes into the more challenging realm of secondary prevention. For example, if a patient has had a stroke, his or her chance of having another stroke in the next 5 years is about 20%. This is much higher than the risk in an average hypertensive patient without such a history. For such patients, the current guidelines¹ recommend the combination of a diuretic and an angiotensin-converting enzyme inhibitor, a combination shown to reduce the risk of a second stroke.¹² Thus, we need to discover whether the patient had a stroke in the first place.

HISTORY

• The duration (if known) and severity of the hypertension
• The degree of blood pressure fluctuation
• Concomitant medical conditions, especially cardiovascular or renal problems
• Dietary habits
• Alcohol consumption
• Tobacco use
• Level of physical activity
• A family history of hypertension, renal disease, cardiovascular problems, or diabetes mellitus
• Past medications, with particular attention to their side effects and their efficacy in controlling blood pressure
• Current medications, including over-the-counter preparations. One reason: non-steroidal anti-inflammatory drugs other than aspirin can decrease the efficacy of antihypertensive drugs, presumably through mechanisms that inhibit the effects of vasodilatory and natriuretic prostaglandins and potentiate those of angiotensin II.¹³

PHYSICAL EXAMINATION

The physical examination starts with measurement of height, weight, waist circumference, and blood pressure—in both arms and the leg if coarctation of the aorta is suspected. Measurements with the patient supine, sitting, and standing are usually taken at the first visit, though such an approach is more suited to a hypertension specialty clinic than a primary care setting, in which time constraints usually limit the blood pressure readings to two or three seated values. Most prospective data on the benefits of hypertension treatment are based on a seated blood pressure, so we favor that measurement for follow-up.

Special attention in the physical examination is directed to:

The retina (to assess the vascular impact of the high blood pressure). Look for arteriolar narrowing (grade 1), arteriovenous compression (grade 2), hemorrhages or exudates (grade 3), and papilledema.² Such findings not only relate to severity (higher grade = more severe blood pressure) but also predict future cardiovascular disease.¹⁴

The blood vessels. Bruits in the neck may indicate carotid stenosis, bruits in the abdomen may indicate renovascular disease, and femoral bruits are a sign of general atherosclerosis. Bruits also signal vascular stenosis and irregularity and may be a clue to vascular damage or future loss of target organ function. However, bruits may simply result from vascular tortuosity, particularly with significant flow in the vessel.

Also check the femoral pulses: poor or delayed femoral pulses are a sign of aortic coarctation. The radial artery is about as far away from the heart as the femoral artery; consequently, when palpating both sites simultaneously the pulse should arrive at about the same moment. In aortic coarctation, a palpable delay in the arrival of the femoral pulse may occur, and an interscapular murmur may be heard during auscultation of the back. In these instances, a low leg blood pressure (usually measured by placing a thigh-sized adult cuff on the patient’s thigh and listening over the popliteal area with the patient prone) may confirm the presence of aortic obstruction. When taking a leg blood pressure, the large cuff and the amount of pressure necessary to occlude the artery may be uncomfortable, and one should warn the patient about the discomfort before taking the measurement.

Poor or absent pedal pulses are a sign of peripheral arterial disease.

The heart (to detect gallops, enlargement, or both). Palpation may reveal a displaced apical impulse, which can indicate left ventricular enlargement. A sustained apical impulse may indicate left ventricular hypertrophy. Listen for a fourth heart sound (S4), one of the earliest physical findings of hypertension when physical findings are present. An S4 indicates that the left atrium is working hard to overcome the stiffness of the left ventricle. An S3 indicates an impairment in left ventricular function and is usually a harbinger of underlying heart disease. In some cases, lung rales can also be heard, though the combination of an S3 gallop and rales is an unusual office presentation in the early management of the hypertensive patient.

The lungs. Listen for rales (see above).

The lower extremities should be examined for peripheral arterial pulsations and edema. The loss of pedal pulses is a common finding, particularly in smokers, and is a clue to increased cardiovascular risk.

Strength, gait, and cognition. Perform a brief neurologic examination for evidence of remote stroke. We usually observe our patients’ gait as they enter or leave the examination room, test their bilateral grip strength, and assess their judgment, speech, and memory during the history and physical examination.

A great deal of research has linked high blood pressure to future loss of cognitive function,¹⁵ and it is useful to know that impairment is present before beginning treatment, since some patients will complain of memory loss after starting antihypertensive drug treatment.

LABORATORY EVALUATION

Routine tests

The routine evaluation of hypertensive patients should include, at a minimum:

• A hemoglobin or hematocrit measurement
• Urinalysis with microscopic examination
• Serum electrolyte concentrations
• 
  Serum creatinine concentrations
• Serum glucose concentration
• A fasting lipid profile
• A 12-lead electrocardiogram (Table 5).

Nonroutine tests

In some cases, other studies may be appropriate, depending on the clinical situation, eg:

• Serum uric acid in those with a history of gout, since some antihypertensive drugs (eg, diuretics) may increase serum uric acid and predispose to further episodes of gout
• Serum calcium in those with a personal or family history of kidney stones, to detect subtle parathyroid excess
• Thyroid-stimulating hormone or other thyroid studies if the history suggests thyroid excess, or if a thyroid nodule is discovered
• Limited echocardiography, which is more sensitive than electrocardiography for detecting left ventricular hypertrophy.

We sometimes use echocardiography if the patient is overweight but seems motivated to lose weight. In these cases we might not start drug therapy right away, choosing rather to wait and see if the patient can lose some weight (which might lower the blood pressure and make drug therapy unnecessary)—but only if the echocardiogram shows that he or she does not have left ventricular hypertrophy.

We also use echocardiography in patients with white-coat hypertension (see below), in whom office pressures are consistently high but whom we have elected to either not treat or not alter treatment. In these cases the echocardiogram serves as a “second opinion” about the merits of not altering therapy and supports this decision when the left ventricular wall thicknesses are normal (and remain so during long-term follow-up). In cases of suspected white-coat hypertension, home or ambulatory blood pressure monitoring is valuable to establish or exclude this diagnosis.¹

Urinary albumin excretion. Microalbuminuria is an early manifestation of diabetic nephropathy and hypertension. Although routine urine screening for microalbuminuria is typically done in the management of diabetes, it is still not considered a standard of care, though the growing literature on its role as a cardiovascular risk predictor^16–18 and its value as a therapeutic target in diabetes^19,20 make it an attractive aid in the overall assessment of patients with hypertension.

Plasma renin activity and serum aldosterone concentrations are useful in screening for aldosterone excess, but are usually reserved as follow-up tests in patients with either hypokalemia or failure to achieve blood pressure control on a three-drug regimen in which at least one drug is a diuretic.^1,21

Of note, primary aldosteronism is not as rare as previously thought. In a study of patients referred to hypertension centers, 11% had primary aldosteronism according to prospective diagnostic criteria, almost 5% had curable aldosterone-producing adenomas, and 6% had idiopathic hyperaldosteronism.²²

A search for secondary forms of hypertension is usually considered in patients with moderate or severe hypertension that does not respond to antihypertensive agents. Another situation is in hypertensive patients younger than 25 years, since curable forms of hypertension are more common in this age group. In older patients, the prevalence of secondary hypertension is lower and does not justify the costs and effort of routine elaborate workups unless there is evidence from the history, physical examination, or routine laboratory work for suspecting its presence. An exception to this rule is the need to exclude atherosclerotic renovascular hypertension in an elderly patient. This cause of secondary hypertension is common in the elderly and may be amenable to therapeutic intervention.²⁶

WHEN TO CONSIDER HOME OR AMBULATORY MONITORING

Suspected white-coat hypertension

Blood pressure can be influenced by an environment such as an office or hospital clinic. This has led to the development of ambulatory blood pressure monitors and more use of self-measurement of blood pressure in the home. Blood pressure readings with these techniques are generally lower than those measured in an office or hospital clinic. These methods make it possible to screen for white-coat hypertension. In 10% to 20% of people with hypertensive readings, the blood pressure may be elevated persistently only in the presence of a physician.²⁸ When measured elsewhere, including at work, the blood pressure is not elevated in those with the white-coat effect. Although this response may become less prominent with repeated measurements, it occasionally persists in the office setting, sometimes for years in our experience.

Suspected nocturnal hypertension (’nondipping’ status)

Ambulatory blood pressure is also helpful to screen for nocturnal hypertension. Evidence is accumulating to suggest that hypertensive patients whose pressure remains relatively high at night (“nondippers,” ie, those with less than a 10% reduction at night compared with daytime blood pressure readings) are at greater risk of cardiovascular morbidity than “dippers” (those whose blood pressure is at least 10% lower at night than during the day).²⁹

An early morning surge

Ambulatory monitoring can also detect morning surges in systolic blood pressure,³⁰ a marker of cerebrovascular risk. Generally, these patients have an increase of more than 55 mm Hg in systolic pressure between their sleeping and early-hour waking values, and we may wish to start or alter treatment specifically to address these high morning systolic values.³¹

‘PIPESTEM’ VESSELS AND PSEUDOHYPERTENSION

Occasionally, one encounters patients with vessels that are stiff and difficult to compress. If the pressure required to compress the brachial artery and stop audible blood flow with a standard blood pressure cuff is greater than the actual blood pressure within the artery as measured invasively, the condition is called pseudohypertension. The stiffness is thought to be due to calcification of the arterial wall.

A way to check for this condition is to inflate the cuff to at least 30 mm Hg above the palpable systolic pressure and then try to “roll” the brachial or radial artery underneath your fingertips, a procedure known as Osler’s maneuver.³² If you feel something that resembles a stiff tube reminiscent of the stem of a tobacco smoker’s pipe (healthy arteries are not palpable when empty), the patient may have pseudohypertension. However, the specificity of Osler’s maneuver has been questioned, particularly in hospitalized elderly patients.³³

Pseudohypertension is important because the patients in whom it occurs, usually the elderly or the chronically ill (with diabetes or chronic kidney disease), are prone to orthostatic or postural hypotension, which may be aggravated by increasing their antihypertensive treatment on the basis of a cuff pressure that is actually much higher than the real blood pressure.³³

Soon, the checklist for internists seeing patients about to undergo surgery may include prescribing one of the lipid-lowering hydroxymethylglutaryl-CoA reductase inhibitors, also called statins.

Statins? Not long ago, we were debating whether patients who take statins should stop taking them before surgery, based on the manufacturers’ recommendations.¹ The discussion, however, has changed to whether patients who have never received a statin should be started on one before surgery to provide immediate prophylaxis against cardiac morbidity, and how much harm long-term statin users face if these drugs are withheld perioperatively.

The evidence is still very preliminary and based mostly on studies in animals and retrospective studies in people. However, an expanding body of indirect evidence suggests that these drugs are beneficial in this situation.

In this review, we discuss the mechanisms by which statins may protect the heart in the short term, drawing on data from animal and human studies of acute myocardial infarction, and we review the current (albeit limited) data from the perioperative setting.

FEW INTERVENTIONS DECREASE RISK

Each year, approximately 50,000 patients suffer a perioperative cardiovascular event; the incidence of myocardial infarction during or after noncardiac surgery is 2% to 3%.² The primary goal of preoperative cardiovascular risk assessment is to predict and avert these events.

But short of canceling surgery, few interventions have been found to reduce a patient’s risk. For example, a landmark study in 2004 cast doubt on the efficacy of preoperative coronary revascularization.³ Similarly, although early studies of beta-blockers were promising^4,5 and although most internists prescribe these drugs before surgery, more recent studies have cast doubt on their efficacy, particularly in patients at low risk undergoing intermediate-risk (rather than vascular) surgery.^6–8

This changing clinical landscape has prompted a search for new strategies for perioperative risk-reduction. Several recent studies have placed statins in the spotlight.

POTENTIAL MECHANISMS OF SHORT-TERM BENEFIT

Statins have been proven to save lives when used long-term, but how could this class of drugs, designed to prevent the accumulation of arterial plaques by lowering low-density lipoprotein cholesterol (LDL-C) levels, have any short-term impact on operative outcomes? Although LDL-C reduction is the principal mechanism of action of statins, not all of the benefit can be ascribed to this mechanism.⁹ The answer may lie in their “pleiotropic” effects—ie, actions other than LDL-C reduction.

The more immediate pleiotropic effects of statins in the proinflammatory and prothrombotic environment of the perioperative period are thought to include improved endothelial function (both antithrombotic function and vasomotor function in response to ischemic stress), enhanced stability of atherosclerotic plaques, decreased oxidative stress, and decreased vascular inflammation.^10–12

EVIDENCE FROM ANIMAL STUDIES

Experiments in animals suggest that statins, given shortly before or after a cardiovascular event, confer benefit before any changes in LDL-C are measurable.

Lefer et al¹³ found that simvastatin (Zocor), given 18 hours before an ischemic episode in rats, blunted the inflammatory response in cardiac reperfusion injury. Not only was reperfusion injury significantly less in the hearts of the rats that received simvastatin than in the saline control group, but the simvastatin-treated hearts also expressed fewer neutrophil adhesion molecules such as P-selectin, and they had more basal release of nitric oxide, the potent endothelial-derived vasodilator with antithrombotic, anti-inflammatory, and antiproliferative effects.¹⁴ These results suggest that statins may improve endothelial function acutely, particularly during ischemic stress.

Osborne et al¹⁵ fed rabbits a cholesterol-rich diet plus either lovastatin (Mevacor) or placebo. After 2 weeks, the rabbits underwent either surgery to induce a myocardial infarction or a sham procedure. Regardless of the pretreatment, biopsies of the aorta did not reveal any atherosclerosis; yet the lovastatin-treated rabbits sustained less myocardial ischemic damage and they had more endothelium-mediated vasodilatation.

Statin therapy also may improve cerebral ischemia outcomes in animal models.^14,16

Sironi et al¹⁶ induced strokes in rats by occluding the middle cerebral artery. The rats received either simvastatin or vehicle for 3 days before the stroke or immediately afterwards. Even though simvastatin did not have enough time to affect the total cholesterol level, rats treated with simvastatin had smaller infarcts (as measured by magnetic resonance imaging) and produced more nitric oxide.

Comment. Taken together, these studies offer tantalizing evidence that statins have short-term, beneficial nonlipid effects and may reduce not only the likelihood of an ischemic event, but—should one occur—the degree of tissue damage that ensues.

EFFECTS OF STATINS IN ACUTE CORONARY SYNDROME

The National Registry of Myocardial Infarction¹⁷ is a prospective, observational database of all patients with acute myocardial infarction admitted to 1,230 participating hospitals throughout the United States. In an analysis from this cohort, patients were divided into four groups: those receiving statins before and after admission, those receiving statins only before admission, those receiving statins only after admission, and those who never received statins.

Compared with those who never received statins, fewer patients who received them both before and after admission died while in the hospital (unadjusted odds ratio 0.23, 95% confidence interval [CI] 0.22–0.25), and the odds ratio for those who received statins for the first time was 0.31 (95% CI 0.29–0.33). Patients who stopped receiving a statin on admission were more likely to die than were patients who never received statins (odds ratio 1.09, 95% CI 1.03–1.15). These trends held true even when adjustments were made for potential confounding factors.

Comment. Unmeasured confounding factors (such as the inability to take pills due to altered mental status or the different practice styles of the providers who chose to discontinue statins) might have affected the results. Nevertheless, these results suggest that the protective effects of statins stop almost immediately when these drugs are discontinued, and that there may even be an adverse “rebound” effect when patients who have been taking these drugs for a long time stop taking them temporarily.

The Platelet Receptor Inhibition in Ischemic Syndrome Management trial,¹⁸ in a subgroup analysis, had nearly identical findings. In the main part of this trial, patients with coronary artery disease and chest pain at rest or accelerating pain in the last 24 hours were randomized to receive tirofiban (Aggrastat) or heparin. Complete data on statin use were available for 1,616 (50%) of the 3,232 patients in this trial, and the rate of the primary end point (death, myocardial infarction, or recurrent ischemia) was analyzed on the basis of statin therapy in this subgroup.

Comment. Together, these data lead to the conclusion that, when admitted for either acute myocardial infarction or acute coronary syndrome, patients already receiving statins should not have them stopped, and those who had not been receiving statins should receive them immediately. The safety of these medications in the acute setting appears excellent: in the Myocardial Ischemia Reduction With Acute Cholesterol Lowering (MIRACL)¹² and the Pravastatin or Atorvastatin Evaluation and Infection Therapy (PROVE-IT)¹¹ trials, fewer than 5% of statin-treated patients had transient elevations in transaminase levels, and no cases of rhabdomyolysis were reported.

PERIOPERATIVE STATIN STUDIES

The data on perioperative statin use are mostly observational and retrospective and fall into essentially four surgical categories: coronary artery bypass grafting (CABG), carotid endarterectomy,^19,20 noncardiac vascular surgery, and major noncardiac surgery. Two meta-analyses have also evaluated the data.^21,22 The only randomized controlled trial (performed by Durazzo et al²³) was small and was carried out at a single center in vascular surgery patients, and the event rate was low.

Current recommendations from the National Cholesterol Education Program (NCEP)²⁴ say that patients who need CABG, have peripheral arterial disease, have an abdominal aortic aneurysm, or have cerebrovascular disease should already be on a statin to achieve an LDL-C goal level of less than 100 mg/dL, with an optional goal of less than 70 mg/dL, independent of surgery.

Since not all patients who should be on statins are actually on them, questions arise:

• Is it important (and safe) to start statin treatment preoperatively?
• Will patients with cardiovascular risk factors but without known cardiovascular disease benefit from statins perioperatively?

Noncardiac vascular surgery

Multiple retrospective studies have evaluated the effect of statins in patients undergoing major noncardiac vascular surgery.^25–32

Kertai et al²⁵ evaluated 570 patients in Holland who underwent elective open surgery for infrarenal abdominal aortic aneurysms between 1991 and 2001, looking for an association between statin use and the incidence of perioperative death from myocardial infarction. Only 162 of the 570 patients had been on long-term statin therapy before the surgery. The use of statins was only one of many known baseline characteristics that were significantly different between the two groups, including age, body mass index, known coronary artery disease, and use of angiotensin-converting enzyme inhibitors and beta-blockers. In univariate analysis, statins appeared to be protective: 6 (3.7%) of the patients in the statin group died of a myocardial infarction, compared with 45 (11%) of those in the nostatin group. A multivariate analysis yielded similar findings, with an odds ratio of 0.24 (95% CI 0.11–0.54).

Ward et al²⁷ performed a very similar retrospective study, with similar findings. In 446 patients who underwent surgery for infrarenal abdominal aortic aneurysm, statin therapy was associated with a significantly lower incidence of the combined end point of death, myocardial infarction, stroke, and major peripheral vascular complications, with an adjusted odds ratio of 0.36 (95% CI 0.14–0.93).

Poldermans et al²⁶ noted similar findings in a case-control study of noncardiac vascular surgery patients. Statin users had a much lower perioperative risk of death than did nonusers, with an adjusted odds ratio of 0.22 (95% CI 0.10–0.47).

O’Neil-Callahan et al,²⁸ in a cohort study, found that statin users had fewer perioperative cardiac complications, with an adjusted odds ratio of 0.49 (95% CI 0.28–0.84, P = .009).

Le Manach et al³³ reviewed the outcomes for all patients of a single hospital in Paris who underwent nonemergency infrarenal aortic procedures between January 2001 and December 2004. In January 2004, the hospital instituted guidelines to ensure that patients on statins continue taking them up to the evening before surgery and that statins be restarted on the first postoperative day (via nasogastric tube if necessary). Before 2004, there had been no specific guidelines, and patients on statins did not receive them for a median of 4 days postoperatively. Types of procedures were similar during the two time periods, as were the rates of beta-blocker use, preoperative revascularization, venous thromboembolism prophylaxis, and perioperative blood pressure control. After surgery, topononin I levels were measured in all patients as surveillance for cardiac events, and were defined as elevated when greater than 0.2 ng/mL.

Compared with patients not on statins at all, those treated with statins continuously throughout the perioperative period (after January 2004) had a lower rate of elevated troponin (relative risk 0.38). In contrast, those who had their statins transiently discontinued perioperatively (prior to 2004) had troponin elevations more often than those who had never been treated (relative risk 2.1). This suggested an over fivefold risk reduction (P < .001) conferred by not discontinuing statins in the immediate postoperative period. This finding was maintained after multivariate adjustment: statin withdrawal was associated with a 2.9-fold (95% CI 1.6–5.5) increase in the risk of cardiac enzyme elevations postoperatively. No fewer deaths were noted, but the study was not powered to detect a mortality difference.

Comment. Although secular trends cannot be entirely discounted as contributing to these findings, the prompt increase in cardiac events after just 4 days of statin withdrawal adds to the growing body of evidence suggesting that statin discontinuation can have harmful acute effects. It also brings up the question: Can starting statins benefit patients in the same time period?

Should statins be started before vascular surgery?

Schouten et al³² evaluated the effects of newly started or continued statin treatment in patients undergoing major elective vascular surgery. Patients were screened before surgery and started on statins if they were not already receiving them and their total cholesterol levels were elevated; new users received the medication for about 40 days before surgery. Of the 981 screened patients, 44 (5%) were newly started on statins and 182 (19%) were continued on their therapy. Perioperative death or myocardial infarction occurred in 22 (8.8%) of the statin users and 111 (14.7%) of the nonusers, a statistically significant difference. Temporary discontinuation (median 1 day) of statins in this study due to the inability to take an oral medication did not appear to affect the likelihood of a myocardial infarction.

Durazzo et al²³ performed a single-center, randomized, prospective, placebo-controlled, double-blind clinical trial of atorvastatin (Lipitor) 20 mg daily vs placebo in 100 patients undergoing noncardiac arterial vascular surgery. Patients were excluded if they had previously used medications to treat dyslipidemia, recently had a cardiovascular event, or had contraindications to statin treatment such as a baseline creatinine level greater than 2.0 mg/dL or severe hepatic disease. The intervention group received atorvastatin starting at least 2 weeks before surgery for a total of 45 days. Patients were then continued or started on a statin after surgery if their LDL-C level was greater than 100 mg/dL. Beta-blocker use was recommended “on the basis of current guidelines.”

One month after surgery, the LDL-C level was statistically significantly lower in the atorvastatin group. Since most patients did not continue or start statin therapy after the 45-day treatment period, the LDL-C levels were not statistically different at 3 and 6 months after surgery.

At 6 months, the rate of the primary end point (death from cardiovascular causes, nonfatal acute myocardial infarction, ischemic stroke, or unstable angina) was 26.0% in the placebo group and 8.0% in the atorvastatin group, a statistically significant difference. Three patients in the atorvastatin group had cardiac events in the first 10 days after surgery, compared with 11 patients in the placebo group. Thirteen of the 17 total cardiac events took place within 10 days after surgery.

One of the atorvastatin patients developed rhabdomyolysis and elevated aminotransferase levels.

Major noncardiac surgery

Lindenauer et al² performed a retrospective cohort study of surgical patients who were at least 18 years old and survived beyond the second hospital day. Patients were divided into a group receiving any form of lipid-lowering treatment (of whom more than 90% were taking statins) and a group that had never never received a lipid-lowering drug or only started one on the third day of the hospitalization or later. The period of study was from January 1, 2000, to December 31, 2001.

In all, 780,591 patients from 329 hospitals throughout the United States were included, of whom only 77,082 (9.9%) received lipid-lowering therapy. Eight percent of the patients underwent vascular surgery. Not surprisingly, the treated patients were more likely to have a history of hypertension, diabetes, ischemic heart disease, or hyperlipidemia. They also were more likely to have a vascular procedure performed, to have two or more cardiac risk factors (high-risk surgery, ischemic heart disease, congestive heart failure, cerebrovascular disease, renal insufficiency, or diabetes mellitus), and to be treated with beta-blockers and angiotensin-converting enzyme inhibitors, but they were less likely to have high-risk and emergency surgery performed.

The primary end point, perioperative death, occurred in 2.13% of the treated patients and 3.05% of the nontreated group. Compared with the rate in a propensity-matched cohort, the odds ratio adjusted for unbalanced covariates was 0.62 (95% CI 0.58–0.67) in favor of lipid treatment. Stratification by cardiac risk index revealed a number needed to treat of 186 for those with no risk factors, 60 for those with two risk factors, and 30 for those with four or more risk factors.

Unfortunately, this analysis was not able to take into account whether and for how long patients were receiving lipid-lowering therapy before hospitalization. It therefore does not answer the questions of whether starting lipid-lowering therapy before surgery is beneficial or whether stopping it is harmful. It also does not shed light on whether perioperative lipid-lowering increases the risk of rhabdomyolysis or liver disease.

Two recent retrospective cohort studies evaluated the outcomes in patients undergoing carotid endarterectomy.^19,20

Kennedy et al¹⁹ found that patients on a statin at the time of admission who had symptomatic carotid disease had lower rates of inhospital death (adjusted odds ratio 0.24, 95% CI 0.06–0.91) and ischemic stroke or death (adjusted odds ratio 0.55, 95% CI 0.31–0.97). However, cardiac outcomes among these symptomatic patients were not significantly improved (odds ratio 0.82, 95% CI 0.45–1.50), nor was there benefit for asymptomatic patients, raising the possibility that the positive findings were due to chance or that patients at lower baseline risk for vascular events may have less benefit.

McGirt et al²⁰ performed a similar study; they did not, however, distinguish whether patients had symptomatic vs asymptomatic carotid disease. The 30-day risk of perioperative stroke was lower in patients treated with a statin, with an odds ratio of 0.41 (95% CI 0.18–0.93); the odds ratio for death was 0.21 (95% CI 0.05–0.96). Cardiac outcomes were not significantly affected.

Coronary artery bypass graft surgery

According to the NCEP recommendations, nearly all patients undergoing CABG should already be on a statin before surgery since they all have known coronary artery disease. Multiple observational studies have offered confirmatory evidence that statins are beneficial in this setting.^34–38

Liakopoulos et al³⁹ evaluated whether the anti-inflammatory effects of statins may, in part, account for their beneficial effect in the perioperative period. The authors prospectively matched 18 patients who were taking statins and were referred for elective CABG with 18 patients who were not prescribed statins previously. The only major measured baseline characteristic that differed between the two groups was a statistically significantly lower LDL-C level in the statin group. The operative characteristics did not differ, and cytokine levels at baseline were similar.

Tumor necrosis factor alpha levels increased significantly in the control group but did not change significantly in the statin group. Interleukin 8 increased in both groups by a similar amount. Interleukin 6 (the major inducer of C-reactive protein) increased from baseline in both groups but did not increase nearly as much in the statin group as in the control group; the intergroup difference was statistically significant. The anti-inflammatory cytokine interleukin 10 increased minimally from baseline in the control group, while the statin group’s levels increased significantly above baseline and those of the control group.

Christenson⁴⁰ also found that inflammatory markers were improved with pre-CABG statin treatment in a small randomized trial in which patients received simvastatin 20 mg 4 weeks prior to CABG surgery vs no statin. Interestingly, far fewer statin-treated patients developed thrombocytosis (platelet count > 400 × 10⁹/L) than did control patients (3% vs 81%, P < .0001).

RISKS OF PERIOPERATIVE STATINS

The risks associated with statin therapy in general appear low, but specific perioperative risks have not been well studied.

Baigent et al,⁴¹ in a meta-analysis of randomized trials of nonperioperative statin therapy, found that rhabdomyolysis occurred in 9 (0.023%) of 39,884 patients receiving statins vs 6 (0.015%) of the 39,817 controls, with a number needed to harm of 12,500. Moreover, the rates of nonvascular death and cancer did not increase. It is plausible that the risk is somewhat greater in the perioperative setting but is likely not enough to outweigh the potential benefits, especially since the risk of ischemic vascular events is particularly high then.

Some of the perioperative studies cited above specifically addressed potential risks. For example, in the study by Schouten et al,³² mild creatine kinase elevations were more common in the statin-treated group, but the incidence of moderate and severe creatine kinase elevations did not differ significantly. No case of rhabdomyolysis occurred, and length of surgery was the only predictor of myopathy. MIRACL and PROVE-IT revealed similar safety profiles; aminotransferase levels normalized when statins were stopped, and no cases of rhabdomyolysis occurred.^11,12 In the vascular surgery study by Durazzo et al,²³ 1 (2%) of the 50 atorvastatin-treated patients developed both rhabdomyolysis and elevated aminotransferase levels that prompted discontinuation of the statin.

Overall, the observational studies do not indicate that statin continuation or treatment is harmful in perioperative patients. However, these studies did not specifically evaluate patients with acute insults from surgery such as sepsis, renal failure, or hepatitis. It is unknown what effect statin therapy would have in those patients and whether statins should be selectively discontinued in patients who develop major hepatic, musculoskeletal, or renal complications after surgery.

OUR RECOMMENDATIONS

Before CABG or vascular surgery

Given the NCEP recommendations, existing primary and secondary prevention studies, observational studies of CABG and noncardiac vascular surgery patients, and the one randomized trial of vascular surgery patients, data support the use of statins in nearly all patients undergoing cardiac or vascular surgery. We advocate starting statins in the perioperative period to take advantage of their rapid-acting pleiotropic effects, and continuing them long-term to take advantage of their lipid-lowering effects. This recommendation is in line with the recently released American College of Cardiology/American Heart Association (ACC/AHA) 2007 perioperative guidelines that state “for patients undergoing vascular surgery with or without clinical risk factors, statin use is reasonable.”⁴²

Although the ideal time to start statins is not certain, the study by Durazzo et al²³ suggests that they should be started at least 2 weeks before surgery if possible. Moreover, patients already taking statins should definitely not have their statins discontinued if at all possible.

Before major nonvascular surgery

For patients undergoing major nonvascular (intermediate-risk) surgery, physicians should first ascertain if the patient has an indication for statin therapy based on current nonsurgical lipid level recommendations. However, even if there is no clear indication for statin therapy based on NCEP guidelines, we endorse the recently released ACC/AHA perioperative guidelines that state that statin therapy can be considered in patients with a risk factor who are undergoing intermediate-risk procedures. Moreover, we wholeheartedly support the ACC/AHA’s strongest recommendation that patients who are already receiving statins and are undergoing noncardiac surgery should not have their statins discontinued.

When to discontinue statins?

The risk of harm overall appears to be minimal and certainly less than the likelihood of benefit. It is reasonable to observe patients postoperatively for adverse clinical events that may increase the risk of perioperative statin treatment, such as acute renal failure, hepatic failure, or sepsis, but whether statins should be stopped in patients with these complications remains unknown; we advocate individualizing the decision.

More studies needed

We need more data on whether moderate-risk patients undergoing moderate-risk surgery benefit from perioperative statin therapy, when therapy should be started, whether therapy should be started on the day of surgery if it was not started earlier, which statin and what doses are optimal, how long therapy should be continued, and what degree of risk is associated with perioperative statin therapy.

Fortunately, important data should be forthcoming in the next few years: the Dutch Echocardiographic Cardiac Risk Evaluation Applying Stress Echocardiography (DECREASE-IV) study⁴³ is a 4-year two-by-two factorial placebo-controlled study evaluating the use of fluvastatin (Lescol) and bisoprolol (Zebeta, a beta-blocker) separately and together in patients who are older than 40 years, are undergoing elective noncardiac surgery, have an estimated risk of cardiovascular death of more than 1%, have not used statins previously, and do not have elevated cholesterol.

Soon, the checklist for internists seeing patients about to undergo surgery may include prescribing one of the lipid-lowering hydroxymethylglutaryl-CoA reductase inhibitors, also called statins.

Statins? Not long ago, we were debating whether patients who take statins should stop taking them before surgery, based on the manufacturers’ recommendations.¹ The discussion, however, has changed to whether patients who have never received a statin should be started on one before surgery to provide immediate prophylaxis against cardiac morbidity, and how much harm long-term statin users face if these drugs are withheld perioperatively.

The evidence is still very preliminary and based mostly on studies in animals and retrospective studies in people. However, an expanding body of indirect evidence suggests that these drugs are beneficial in this situation.

In this review, we discuss the mechanisms by which statins may protect the heart in the short term, drawing on data from animal and human studies of acute myocardial infarction, and we review the current (albeit limited) data from the perioperative setting.

FEW INTERVENTIONS DECREASE RISK

Each year, approximately 50,000 patients suffer a perioperative cardiovascular event; the incidence of myocardial infarction during or after noncardiac surgery is 2% to 3%.² The primary goal of preoperative cardiovascular risk assessment is to predict and avert these events.

But short of canceling surgery, few interventions have been found to reduce a patient’s risk. For example, a landmark study in 2004 cast doubt on the efficacy of preoperative coronary revascularization.³ Similarly, although early studies of beta-blockers were promising^4,5 and although most internists prescribe these drugs before surgery, more recent studies have cast doubt on their efficacy, particularly in patients at low risk undergoing intermediate-risk (rather than vascular) surgery.^6–8

This changing clinical landscape has prompted a search for new strategies for perioperative risk-reduction. Several recent studies have placed statins in the spotlight.

POTENTIAL MECHANISMS OF SHORT-TERM BENEFIT

Statins have been proven to save lives when used long-term, but how could this class of drugs, designed to prevent the accumulation of arterial plaques by lowering low-density lipoprotein cholesterol (LDL-C) levels, have any short-term impact on operative outcomes? Although LDL-C reduction is the principal mechanism of action of statins, not all of the benefit can be ascribed to this mechanism.⁹ The answer may lie in their “pleiotropic” effects—ie, actions other than LDL-C reduction.

The more immediate pleiotropic effects of statins in the proinflammatory and prothrombotic environment of the perioperative period are thought to include improved endothelial function (both antithrombotic function and vasomotor function in response to ischemic stress), enhanced stability of atherosclerotic plaques, decreased oxidative stress, and decreased vascular inflammation.^10–12

EVIDENCE FROM ANIMAL STUDIES

Experiments in animals suggest that statins, given shortly before or after a cardiovascular event, confer benefit before any changes in LDL-C are measurable.

Lefer et al¹³ found that simvastatin (Zocor), given 18 hours before an ischemic episode in rats, blunted the inflammatory response in cardiac reperfusion injury. Not only was reperfusion injury significantly less in the hearts of the rats that received simvastatin than in the saline control group, but the simvastatin-treated hearts also expressed fewer neutrophil adhesion molecules such as P-selectin, and they had more basal release of nitric oxide, the potent endothelial-derived vasodilator with antithrombotic, anti-inflammatory, and antiproliferative effects.¹⁴ These results suggest that statins may improve endothelial function acutely, particularly during ischemic stress.

Osborne et al¹⁵ fed rabbits a cholesterol-rich diet plus either lovastatin (Mevacor) or placebo. After 2 weeks, the rabbits underwent either surgery to induce a myocardial infarction or a sham procedure. Regardless of the pretreatment, biopsies of the aorta did not reveal any atherosclerosis; yet the lovastatin-treated rabbits sustained less myocardial ischemic damage and they had more endothelium-mediated vasodilatation.

Statin therapy also may improve cerebral ischemia outcomes in animal models.^14,16

Sironi et al¹⁶ induced strokes in rats by occluding the middle cerebral artery. The rats received either simvastatin or vehicle for 3 days before the stroke or immediately afterwards. Even though simvastatin did not have enough time to affect the total cholesterol level, rats treated with simvastatin had smaller infarcts (as measured by magnetic resonance imaging) and produced more nitric oxide.

Comment. Taken together, these studies offer tantalizing evidence that statins have short-term, beneficial nonlipid effects and may reduce not only the likelihood of an ischemic event, but—should one occur—the degree of tissue damage that ensues.

EFFECTS OF STATINS IN ACUTE CORONARY SYNDROME

The National Registry of Myocardial Infarction¹⁷ is a prospective, observational database of all patients with acute myocardial infarction admitted to 1,230 participating hospitals throughout the United States. In an analysis from this cohort, patients were divided into four groups: those receiving statins before and after admission, those receiving statins only before admission, those receiving statins only after admission, and those who never received statins.

Compared with those who never received statins, fewer patients who received them both before and after admission died while in the hospital (unadjusted odds ratio 0.23, 95% confidence interval [CI] 0.22–0.25), and the odds ratio for those who received statins for the first time was 0.31 (95% CI 0.29–0.33). Patients who stopped receiving a statin on admission were more likely to die than were patients who never received statins (odds ratio 1.09, 95% CI 1.03–1.15). These trends held true even when adjustments were made for potential confounding factors.

Comment. Unmeasured confounding factors (such as the inability to take pills due to altered mental status or the different practice styles of the providers who chose to discontinue statins) might have affected the results. Nevertheless, these results suggest that the protective effects of statins stop almost immediately when these drugs are discontinued, and that there may even be an adverse “rebound” effect when patients who have been taking these drugs for a long time stop taking them temporarily.

The Platelet Receptor Inhibition in Ischemic Syndrome Management trial,¹⁸ in a subgroup analysis, had nearly identical findings. In the main part of this trial, patients with coronary artery disease and chest pain at rest or accelerating pain in the last 24 hours were randomized to receive tirofiban (Aggrastat) or heparin. Complete data on statin use were available for 1,616 (50%) of the 3,232 patients in this trial, and the rate of the primary end point (death, myocardial infarction, or recurrent ischemia) was analyzed on the basis of statin therapy in this subgroup.

Comment. Together, these data lead to the conclusion that, when admitted for either acute myocardial infarction or acute coronary syndrome, patients already receiving statins should not have them stopped, and those who had not been receiving statins should receive them immediately. The safety of these medications in the acute setting appears excellent: in the Myocardial Ischemia Reduction With Acute Cholesterol Lowering (MIRACL)¹² and the Pravastatin or Atorvastatin Evaluation and Infection Therapy (PROVE-IT)¹¹ trials, fewer than 5% of statin-treated patients had transient elevations in transaminase levels, and no cases of rhabdomyolysis were reported.

PERIOPERATIVE STATIN STUDIES

The data on perioperative statin use are mostly observational and retrospective and fall into essentially four surgical categories: coronary artery bypass grafting (CABG), carotid endarterectomy,^19,20 noncardiac vascular surgery, and major noncardiac surgery. Two meta-analyses have also evaluated the data.^21,22 The only randomized controlled trial (performed by Durazzo et al²³) was small and was carried out at a single center in vascular surgery patients, and the event rate was low.

Current recommendations from the National Cholesterol Education Program (NCEP)²⁴ say that patients who need CABG, have peripheral arterial disease, have an abdominal aortic aneurysm, or have cerebrovascular disease should already be on a statin to achieve an LDL-C goal level of less than 100 mg/dL, with an optional goal of less than 70 mg/dL, independent of surgery.

Since not all patients who should be on statins are actually on them, questions arise:

• Is it important (and safe) to start statin treatment preoperatively?
• Will patients with cardiovascular risk factors but without known cardiovascular disease benefit from statins perioperatively?

Noncardiac vascular surgery

Multiple retrospective studies have evaluated the effect of statins in patients undergoing major noncardiac vascular surgery.^25–32

Kertai et al²⁵ evaluated 570 patients in Holland who underwent elective open surgery for infrarenal abdominal aortic aneurysms between 1991 and 2001, looking for an association between statin use and the incidence of perioperative death from myocardial infarction. Only 162 of the 570 patients had been on long-term statin therapy before the surgery. The use of statins was only one of many known baseline characteristics that were significantly different between the two groups, including age, body mass index, known coronary artery disease, and use of angiotensin-converting enzyme inhibitors and beta-blockers. In univariate analysis, statins appeared to be protective: 6 (3.7%) of the patients in the statin group died of a myocardial infarction, compared with 45 (11%) of those in the nostatin group. A multivariate analysis yielded similar findings, with an odds ratio of 0.24 (95% CI 0.11–0.54).

Ward et al²⁷ performed a very similar retrospective study, with similar findings. In 446 patients who underwent surgery for infrarenal abdominal aortic aneurysm, statin therapy was associated with a significantly lower incidence of the combined end point of death, myocardial infarction, stroke, and major peripheral vascular complications, with an adjusted odds ratio of 0.36 (95% CI 0.14–0.93).

Poldermans et al²⁶ noted similar findings in a case-control study of noncardiac vascular surgery patients. Statin users had a much lower perioperative risk of death than did nonusers, with an adjusted odds ratio of 0.22 (95% CI 0.10–0.47).

O’Neil-Callahan et al,²⁸ in a cohort study, found that statin users had fewer perioperative cardiac complications, with an adjusted odds ratio of 0.49 (95% CI 0.28–0.84, P = .009).

Le Manach et al³³ reviewed the outcomes for all patients of a single hospital in Paris who underwent nonemergency infrarenal aortic procedures between January 2001 and December 2004. In January 2004, the hospital instituted guidelines to ensure that patients on statins continue taking them up to the evening before surgery and that statins be restarted on the first postoperative day (via nasogastric tube if necessary). Before 2004, there had been no specific guidelines, and patients on statins did not receive them for a median of 4 days postoperatively. Types of procedures were similar during the two time periods, as were the rates of beta-blocker use, preoperative revascularization, venous thromboembolism prophylaxis, and perioperative blood pressure control. After surgery, topononin I levels were measured in all patients as surveillance for cardiac events, and were defined as elevated when greater than 0.2 ng/mL.

Compared with patients not on statins at all, those treated with statins continuously throughout the perioperative period (after January 2004) had a lower rate of elevated troponin (relative risk 0.38). In contrast, those who had their statins transiently discontinued perioperatively (prior to 2004) had troponin elevations more often than those who had never been treated (relative risk 2.1). This suggested an over fivefold risk reduction (P < .001) conferred by not discontinuing statins in the immediate postoperative period. This finding was maintained after multivariate adjustment: statin withdrawal was associated with a 2.9-fold (95% CI 1.6–5.5) increase in the risk of cardiac enzyme elevations postoperatively. No fewer deaths were noted, but the study was not powered to detect a mortality difference.

Comment. Although secular trends cannot be entirely discounted as contributing to these findings, the prompt increase in cardiac events after just 4 days of statin withdrawal adds to the growing body of evidence suggesting that statin discontinuation can have harmful acute effects. It also brings up the question: Can starting statins benefit patients in the same time period?

Should statins be started before vascular surgery?

Schouten et al³² evaluated the effects of newly started or continued statin treatment in patients undergoing major elective vascular surgery. Patients were screened before surgery and started on statins if they were not already receiving them and their total cholesterol levels were elevated; new users received the medication for about 40 days before surgery. Of the 981 screened patients, 44 (5%) were newly started on statins and 182 (19%) were continued on their therapy. Perioperative death or myocardial infarction occurred in 22 (8.8%) of the statin users and 111 (14.7%) of the nonusers, a statistically significant difference. Temporary discontinuation (median 1 day) of statins in this study due to the inability to take an oral medication did not appear to affect the likelihood of a myocardial infarction.

Durazzo et al²³ performed a single-center, randomized, prospective, placebo-controlled, double-blind clinical trial of atorvastatin (Lipitor) 20 mg daily vs placebo in 100 patients undergoing noncardiac arterial vascular surgery. Patients were excluded if they had previously used medications to treat dyslipidemia, recently had a cardiovascular event, or had contraindications to statin treatment such as a baseline creatinine level greater than 2.0 mg/dL or severe hepatic disease. The intervention group received atorvastatin starting at least 2 weeks before surgery for a total of 45 days. Patients were then continued or started on a statin after surgery if their LDL-C level was greater than 100 mg/dL. Beta-blocker use was recommended “on the basis of current guidelines.”

One month after surgery, the LDL-C level was statistically significantly lower in the atorvastatin group. Since most patients did not continue or start statin therapy after the 45-day treatment period, the LDL-C levels were not statistically different at 3 and 6 months after surgery.

At 6 months, the rate of the primary end point (death from cardiovascular causes, nonfatal acute myocardial infarction, ischemic stroke, or unstable angina) was 26.0% in the placebo group and 8.0% in the atorvastatin group, a statistically significant difference. Three patients in the atorvastatin group had cardiac events in the first 10 days after surgery, compared with 11 patients in the placebo group. Thirteen of the 17 total cardiac events took place within 10 days after surgery.

One of the atorvastatin patients developed rhabdomyolysis and elevated aminotransferase levels.

Major noncardiac surgery

Lindenauer et al² performed a retrospective cohort study of surgical patients who were at least 18 years old and survived beyond the second hospital day. Patients were divided into a group receiving any form of lipid-lowering treatment (of whom more than 90% were taking statins) and a group that had never never received a lipid-lowering drug or only started one on the third day of the hospitalization or later. The period of study was from January 1, 2000, to December 31, 2001.

In all, 780,591 patients from 329 hospitals throughout the United States were included, of whom only 77,082 (9.9%) received lipid-lowering therapy. Eight percent of the patients underwent vascular surgery. Not surprisingly, the treated patients were more likely to have a history of hypertension, diabetes, ischemic heart disease, or hyperlipidemia. They also were more likely to have a vascular procedure performed, to have two or more cardiac risk factors (high-risk surgery, ischemic heart disease, congestive heart failure, cerebrovascular disease, renal insufficiency, or diabetes mellitus), and to be treated with beta-blockers and angiotensin-converting enzyme inhibitors, but they were less likely to have high-risk and emergency surgery performed.

The primary end point, perioperative death, occurred in 2.13% of the treated patients and 3.05% of the nontreated group. Compared with the rate in a propensity-matched cohort, the odds ratio adjusted for unbalanced covariates was 0.62 (95% CI 0.58–0.67) in favor of lipid treatment. Stratification by cardiac risk index revealed a number needed to treat of 186 for those with no risk factors, 60 for those with two risk factors, and 30 for those with four or more risk factors.

Unfortunately, this analysis was not able to take into account whether and for how long patients were receiving lipid-lowering therapy before hospitalization. It therefore does not answer the questions of whether starting lipid-lowering therapy before surgery is beneficial or whether stopping it is harmful. It also does not shed light on whether perioperative lipid-lowering increases the risk of rhabdomyolysis or liver disease.

Two recent retrospective cohort studies evaluated the outcomes in patients undergoing carotid endarterectomy.^19,20

Kennedy et al¹⁹ found that patients on a statin at the time of admission who had symptomatic carotid disease had lower rates of inhospital death (adjusted odds ratio 0.24, 95% CI 0.06–0.91) and ischemic stroke or death (adjusted odds ratio 0.55, 95% CI 0.31–0.97). However, cardiac outcomes among these symptomatic patients were not significantly improved (odds ratio 0.82, 95% CI 0.45–1.50), nor was there benefit for asymptomatic patients, raising the possibility that the positive findings were due to chance or that patients at lower baseline risk for vascular events may have less benefit.

McGirt et al²⁰ performed a similar study; they did not, however, distinguish whether patients had symptomatic vs asymptomatic carotid disease. The 30-day risk of perioperative stroke was lower in patients treated with a statin, with an odds ratio of 0.41 (95% CI 0.18–0.93); the odds ratio for death was 0.21 (95% CI 0.05–0.96). Cardiac outcomes were not significantly affected.

Coronary artery bypass graft surgery

According to the NCEP recommendations, nearly all patients undergoing CABG should already be on a statin before surgery since they all have known coronary artery disease. Multiple observational studies have offered confirmatory evidence that statins are beneficial in this setting.^34–38

Liakopoulos et al³⁹ evaluated whether the anti-inflammatory effects of statins may, in part, account for their beneficial effect in the perioperative period. The authors prospectively matched 18 patients who were taking statins and were referred for elective CABG with 18 patients who were not prescribed statins previously. The only major measured baseline characteristic that differed between the two groups was a statistically significantly lower LDL-C level in the statin group. The operative characteristics did not differ, and cytokine levels at baseline were similar.

Tumor necrosis factor alpha levels increased significantly in the control group but did not change significantly in the statin group. Interleukin 8 increased in both groups by a similar amount. Interleukin 6 (the major inducer of C-reactive protein) increased from baseline in both groups but did not increase nearly as much in the statin group as in the control group; the intergroup difference was statistically significant. The anti-inflammatory cytokine interleukin 10 increased minimally from baseline in the control group, while the statin group’s levels increased significantly above baseline and those of the control group.

Christenson⁴⁰ also found that inflammatory markers were improved with pre-CABG statin treatment in a small randomized trial in which patients received simvastatin 20 mg 4 weeks prior to CABG surgery vs no statin. Interestingly, far fewer statin-treated patients developed thrombocytosis (platelet count > 400 × 10⁹/L) than did control patients (3% vs 81%, P < .0001).

RISKS OF PERIOPERATIVE STATINS

The risks associated with statin therapy in general appear low, but specific perioperative risks have not been well studied.

Baigent et al,⁴¹ in a meta-analysis of randomized trials of nonperioperative statin therapy, found that rhabdomyolysis occurred in 9 (0.023%) of 39,884 patients receiving statins vs 6 (0.015%) of the 39,817 controls, with a number needed to harm of 12,500. Moreover, the rates of nonvascular death and cancer did not increase. It is plausible that the risk is somewhat greater in the perioperative setting but is likely not enough to outweigh the potential benefits, especially since the risk of ischemic vascular events is particularly high then.

Some of the perioperative studies cited above specifically addressed potential risks. For example, in the study by Schouten et al,³² mild creatine kinase elevations were more common in the statin-treated group, but the incidence of moderate and severe creatine kinase elevations did not differ significantly. No case of rhabdomyolysis occurred, and length of surgery was the only predictor of myopathy. MIRACL and PROVE-IT revealed similar safety profiles; aminotransferase levels normalized when statins were stopped, and no cases of rhabdomyolysis occurred.^11,12 In the vascular surgery study by Durazzo et al,²³ 1 (2%) of the 50 atorvastatin-treated patients developed both rhabdomyolysis and elevated aminotransferase levels that prompted discontinuation of the statin.

Overall, the observational studies do not indicate that statin continuation or treatment is harmful in perioperative patients. However, these studies did not specifically evaluate patients with acute insults from surgery such as sepsis, renal failure, or hepatitis. It is unknown what effect statin therapy would have in those patients and whether statins should be selectively discontinued in patients who develop major hepatic, musculoskeletal, or renal complications after surgery.

OUR RECOMMENDATIONS

Before CABG or vascular surgery

Given the NCEP recommendations, existing primary and secondary prevention studies, observational studies of CABG and noncardiac vascular surgery patients, and the one randomized trial of vascular surgery patients, data support the use of statins in nearly all patients undergoing cardiac or vascular surgery. We advocate starting statins in the perioperative period to take advantage of their rapid-acting pleiotropic effects, and continuing them long-term to take advantage of their lipid-lowering effects. This recommendation is in line with the recently released American College of Cardiology/American Heart Association (ACC/AHA) 2007 perioperative guidelines that state “for patients undergoing vascular surgery with or without clinical risk factors, statin use is reasonable.”⁴²

Although the ideal time to start statins is not certain, the study by Durazzo et al²³ suggests that they should be started at least 2 weeks before surgery if possible. Moreover, patients already taking statins should definitely not have their statins discontinued if at all possible.

Before major nonvascular surgery

For patients undergoing major nonvascular (intermediate-risk) surgery, physicians should first ascertain if the patient has an indication for statin therapy based on current nonsurgical lipid level recommendations. However, even if there is no clear indication for statin therapy based on NCEP guidelines, we endorse the recently released ACC/AHA perioperative guidelines that state that statin therapy can be considered in patients with a risk factor who are undergoing intermediate-risk procedures. Moreover, we wholeheartedly support the ACC/AHA’s strongest recommendation that patients who are already receiving statins and are undergoing noncardiac surgery should not have their statins discontinued.

When to discontinue statins?

The risk of harm overall appears to be minimal and certainly less than the likelihood of benefit. It is reasonable to observe patients postoperatively for adverse clinical events that may increase the risk of perioperative statin treatment, such as acute renal failure, hepatic failure, or sepsis, but whether statins should be stopped in patients with these complications remains unknown; we advocate individualizing the decision.

More studies needed

We need more data on whether moderate-risk patients undergoing moderate-risk surgery benefit from perioperative statin therapy, when therapy should be started, whether therapy should be started on the day of surgery if it was not started earlier, which statin and what doses are optimal, how long therapy should be continued, and what degree of risk is associated with perioperative statin therapy.

Fortunately, important data should be forthcoming in the next few years: the Dutch Echocardiographic Cardiac Risk Evaluation Applying Stress Echocardiography (DECREASE-IV) study⁴³ is a 4-year two-by-two factorial placebo-controlled study evaluating the use of fluvastatin (Lescol) and bisoprolol (Zebeta, a beta-blocker) separately and together in patients who are older than 40 years, are undergoing elective noncardiac surgery, have an estimated risk of cardiovascular death of more than 1%, have not used statins previously, and do not have elevated cholesterol.

Soon, the checklist for internists seeing patients about to undergo surgery may include prescribing one of the lipid-lowering hydroxymethylglutaryl-CoA reductase inhibitors, also called statins.

Statins? Not long ago, we were debating whether patients who take statins should stop taking them before surgery, based on the manufacturers’ recommendations.¹ The discussion, however, has changed to whether patients who have never received a statin should be started on one before surgery to provide immediate prophylaxis against cardiac morbidity, and how much harm long-term statin users face if these drugs are withheld perioperatively.

The evidence is still very preliminary and based mostly on studies in animals and retrospective studies in people. However, an expanding body of indirect evidence suggests that these drugs are beneficial in this situation.

In this review, we discuss the mechanisms by which statins may protect the heart in the short term, drawing on data from animal and human studies of acute myocardial infarction, and we review the current (albeit limited) data from the perioperative setting.

FEW INTERVENTIONS DECREASE RISK

Each year, approximately 50,000 patients suffer a perioperative cardiovascular event; the incidence of myocardial infarction during or after noncardiac surgery is 2% to 3%.² The primary goal of preoperative cardiovascular risk assessment is to predict and avert these events.

But short of canceling surgery, few interventions have been found to reduce a patient’s risk. For example, a landmark study in 2004 cast doubt on the efficacy of preoperative coronary revascularization.³ Similarly, although early studies of beta-blockers were promising^4,5 and although most internists prescribe these drugs before surgery, more recent studies have cast doubt on their efficacy, particularly in patients at low risk undergoing intermediate-risk (rather than vascular) surgery.^6–8

This changing clinical landscape has prompted a search for new strategies for perioperative risk-reduction. Several recent studies have placed statins in the spotlight.

POTENTIAL MECHANISMS OF SHORT-TERM BENEFIT

Statins have been proven to save lives when used long-term, but how could this class of drugs, designed to prevent the accumulation of arterial plaques by lowering low-density lipoprotein cholesterol (LDL-C) levels, have any short-term impact on operative outcomes? Although LDL-C reduction is the principal mechanism of action of statins, not all of the benefit can be ascribed to this mechanism.⁹ The answer may lie in their “pleiotropic” effects—ie, actions other than LDL-C reduction.

The more immediate pleiotropic effects of statins in the proinflammatory and prothrombotic environment of the perioperative period are thought to include improved endothelial function (both antithrombotic function and vasomotor function in response to ischemic stress), enhanced stability of atherosclerotic plaques, decreased oxidative stress, and decreased vascular inflammation.^10–12

EVIDENCE FROM ANIMAL STUDIES

Experiments in animals suggest that statins, given shortly before or after a cardiovascular event, confer benefit before any changes in LDL-C are measurable.

Lefer et al¹³ found that simvastatin (Zocor), given 18 hours before an ischemic episode in rats, blunted the inflammatory response in cardiac reperfusion injury. Not only was reperfusion injury significantly less in the hearts of the rats that received simvastatin than in the saline control group, but the simvastatin-treated hearts also expressed fewer neutrophil adhesion molecules such as P-selectin, and they had more basal release of nitric oxide, the potent endothelial-derived vasodilator with antithrombotic, anti-inflammatory, and antiproliferative effects.¹⁴ These results suggest that statins may improve endothelial function acutely, particularly during ischemic stress.

Osborne et al¹⁵ fed rabbits a cholesterol-rich diet plus either lovastatin (Mevacor) or placebo. After 2 weeks, the rabbits underwent either surgery to induce a myocardial infarction or a sham procedure. Regardless of the pretreatment, biopsies of the aorta did not reveal any atherosclerosis; yet the lovastatin-treated rabbits sustained less myocardial ischemic damage and they had more endothelium-mediated vasodilatation.

Statin therapy also may improve cerebral ischemia outcomes in animal models.^14,16

Sironi et al¹⁶ induced strokes in rats by occluding the middle cerebral artery. The rats received either simvastatin or vehicle for 3 days before the stroke or immediately afterwards. Even though simvastatin did not have enough time to affect the total cholesterol level, rats treated with simvastatin had smaller infarcts (as measured by magnetic resonance imaging) and produced more nitric oxide.

Comment. Taken together, these studies offer tantalizing evidence that statins have short-term, beneficial nonlipid effects and may reduce not only the likelihood of an ischemic event, but—should one occur—the degree of tissue damage that ensues.

EFFECTS OF STATINS IN ACUTE CORONARY SYNDROME

The National Registry of Myocardial Infarction¹⁷ is a prospective, observational database of all patients with acute myocardial infarction admitted to 1,230 participating hospitals throughout the United States. In an analysis from this cohort, patients were divided into four groups: those receiving statins before and after admission, those receiving statins only before admission, those receiving statins only after admission, and those who never received statins.

Compared with those who never received statins, fewer patients who received them both before and after admission died while in the hospital (unadjusted odds ratio 0.23, 95% confidence interval [CI] 0.22–0.25), and the odds ratio for those who received statins for the first time was 0.31 (95% CI 0.29–0.33). Patients who stopped receiving a statin on admission were more likely to die than were patients who never received statins (odds ratio 1.09, 95% CI 1.03–1.15). These trends held true even when adjustments were made for potential confounding factors.

Comment. Unmeasured confounding factors (such as the inability to take pills due to altered mental status or the different practice styles of the providers who chose to discontinue statins) might have affected the results. Nevertheless, these results suggest that the protective effects of statins stop almost immediately when these drugs are discontinued, and that there may even be an adverse “rebound” effect when patients who have been taking these drugs for a long time stop taking them temporarily.

The Platelet Receptor Inhibition in Ischemic Syndrome Management trial,¹⁸ in a subgroup analysis, had nearly identical findings. In the main part of this trial, patients with coronary artery disease and chest pain at rest or accelerating pain in the last 24 hours were randomized to receive tirofiban (Aggrastat) or heparin. Complete data on statin use were available for 1,616 (50%) of the 3,232 patients in this trial, and the rate of the primary end point (death, myocardial infarction, or recurrent ischemia) was analyzed on the basis of statin therapy in this subgroup.

Comment. Together, these data lead to the conclusion that, when admitted for either acute myocardial infarction or acute coronary syndrome, patients already receiving statins should not have them stopped, and those who had not been receiving statins should receive them immediately. The safety of these medications in the acute setting appears excellent: in the Myocardial Ischemia Reduction With Acute Cholesterol Lowering (MIRACL)¹² and the Pravastatin or Atorvastatin Evaluation and Infection Therapy (PROVE-IT)¹¹ trials, fewer than 5% of statin-treated patients had transient elevations in transaminase levels, and no cases of rhabdomyolysis were reported.

PERIOPERATIVE STATIN STUDIES

The data on perioperative statin use are mostly observational and retrospective and fall into essentially four surgical categories: coronary artery bypass grafting (CABG), carotid endarterectomy,^19,20 noncardiac vascular surgery, and major noncardiac surgery. Two meta-analyses have also evaluated the data.^21,22 The only randomized controlled trial (performed by Durazzo et al²³) was small and was carried out at a single center in vascular surgery patients, and the event rate was low.

Current recommendations from the National Cholesterol Education Program (NCEP)²⁴ say that patients who need CABG, have peripheral arterial disease, have an abdominal aortic aneurysm, or have cerebrovascular disease should already be on a statin to achieve an LDL-C goal level of less than 100 mg/dL, with an optional goal of less than 70 mg/dL, independent of surgery.

Since not all patients who should be on statins are actually on them, questions arise:

• Is it important (and safe) to start statin treatment preoperatively?
• Will patients with cardiovascular risk factors but without known cardiovascular disease benefit from statins perioperatively?

Noncardiac vascular surgery

Multiple retrospective studies have evaluated the effect of statins in patients undergoing major noncardiac vascular surgery.^25–32

Kertai et al²⁵ evaluated 570 patients in Holland who underwent elective open surgery for infrarenal abdominal aortic aneurysms between 1991 and 2001, looking for an association between statin use and the incidence of perioperative death from myocardial infarction. Only 162 of the 570 patients had been on long-term statin therapy before the surgery. The use of statins was only one of many known baseline characteristics that were significantly different between the two groups, including age, body mass index, known coronary artery disease, and use of angiotensin-converting enzyme inhibitors and beta-blockers. In univariate analysis, statins appeared to be protective: 6 (3.7%) of the patients in the statin group died of a myocardial infarction, compared with 45 (11%) of those in the nostatin group. A multivariate analysis yielded similar findings, with an odds ratio of 0.24 (95% CI 0.11–0.54).

Ward et al²⁷ performed a very similar retrospective study, with similar findings. In 446 patients who underwent surgery for infrarenal abdominal aortic aneurysm, statin therapy was associated with a significantly lower incidence of the combined end point of death, myocardial infarction, stroke, and major peripheral vascular complications, with an adjusted odds ratio of 0.36 (95% CI 0.14–0.93).

Poldermans et al²⁶ noted similar findings in a case-control study of noncardiac vascular surgery patients. Statin users had a much lower perioperative risk of death than did nonusers, with an adjusted odds ratio of 0.22 (95% CI 0.10–0.47).

O’Neil-Callahan et al,²⁸ in a cohort study, found that statin users had fewer perioperative cardiac complications, with an adjusted odds ratio of 0.49 (95% CI 0.28–0.84, P = .009).

Le Manach et al³³ reviewed the outcomes for all patients of a single hospital in Paris who underwent nonemergency infrarenal aortic procedures between January 2001 and December 2004. In January 2004, the hospital instituted guidelines to ensure that patients on statins continue taking them up to the evening before surgery and that statins be restarted on the first postoperative day (via nasogastric tube if necessary). Before 2004, there had been no specific guidelines, and patients on statins did not receive them for a median of 4 days postoperatively. Types of procedures were similar during the two time periods, as were the rates of beta-blocker use, preoperative revascularization, venous thromboembolism prophylaxis, and perioperative blood pressure control. After surgery, topononin I levels were measured in all patients as surveillance for cardiac events, and were defined as elevated when greater than 0.2 ng/mL.

Compared with patients not on statins at all, those treated with statins continuously throughout the perioperative period (after January 2004) had a lower rate of elevated troponin (relative risk 0.38). In contrast, those who had their statins transiently discontinued perioperatively (prior to 2004) had troponin elevations more often than those who had never been treated (relative risk 2.1). This suggested an over fivefold risk reduction (P < .001) conferred by not discontinuing statins in the immediate postoperative period. This finding was maintained after multivariate adjustment: statin withdrawal was associated with a 2.9-fold (95% CI 1.6–5.5) increase in the risk of cardiac enzyme elevations postoperatively. No fewer deaths were noted, but the study was not powered to detect a mortality difference.

Comment. Although secular trends cannot be entirely discounted as contributing to these findings, the prompt increase in cardiac events after just 4 days of statin withdrawal adds to the growing body of evidence suggesting that statin discontinuation can have harmful acute effects. It also brings up the question: Can starting statins benefit patients in the same time period?

Should statins be started before vascular surgery?

Schouten et al³² evaluated the effects of newly started or continued statin treatment in patients undergoing major elective vascular surgery. Patients were screened before surgery and started on statins if they were not already receiving them and their total cholesterol levels were elevated; new users received the medication for about 40 days before surgery. Of the 981 screened patients, 44 (5%) were newly started on statins and 182 (19%) were continued on their therapy. Perioperative death or myocardial infarction occurred in 22 (8.8%) of the statin users and 111 (14.7%) of the nonusers, a statistically significant difference. Temporary discontinuation (median 1 day) of statins in this study due to the inability to take an oral medication did not appear to affect the likelihood of a myocardial infarction.

Durazzo et al²³ performed a single-center, randomized, prospective, placebo-controlled, double-blind clinical trial of atorvastatin (Lipitor) 20 mg daily vs placebo in 100 patients undergoing noncardiac arterial vascular surgery. Patients were excluded if they had previously used medications to treat dyslipidemia, recently had a cardiovascular event, or had contraindications to statin treatment such as a baseline creatinine level greater than 2.0 mg/dL or severe hepatic disease. The intervention group received atorvastatin starting at least 2 weeks before surgery for a total of 45 days. Patients were then continued or started on a statin after surgery if their LDL-C level was greater than 100 mg/dL. Beta-blocker use was recommended “on the basis of current guidelines.”

One month after surgery, the LDL-C level was statistically significantly lower in the atorvastatin group. Since most patients did not continue or start statin therapy after the 45-day treatment period, the LDL-C levels were not statistically different at 3 and 6 months after surgery.

At 6 months, the rate of the primary end point (death from cardiovascular causes, nonfatal acute myocardial infarction, ischemic stroke, or unstable angina) was 26.0% in the placebo group and 8.0% in the atorvastatin group, a statistically significant difference. Three patients in the atorvastatin group had cardiac events in the first 10 days after surgery, compared with 11 patients in the placebo group. Thirteen of the 17 total cardiac events took place within 10 days after surgery.

One of the atorvastatin patients developed rhabdomyolysis and elevated aminotransferase levels.

Major noncardiac surgery

Lindenauer et al² performed a retrospective cohort study of surgical patients who were at least 18 years old and survived beyond the second hospital day. Patients were divided into a group receiving any form of lipid-lowering treatment (of whom more than 90% were taking statins) and a group that had never never received a lipid-lowering drug or only started one on the third day of the hospitalization or later. The period of study was from January 1, 2000, to December 31, 2001.

In all, 780,591 patients from 329 hospitals throughout the United States were included, of whom only 77,082 (9.9%) received lipid-lowering therapy. Eight percent of the patients underwent vascular surgery. Not surprisingly, the treated patients were more likely to have a history of hypertension, diabetes, ischemic heart disease, or hyperlipidemia. They also were more likely to have a vascular procedure performed, to have two or more cardiac risk factors (high-risk surgery, ischemic heart disease, congestive heart failure, cerebrovascular disease, renal insufficiency, or diabetes mellitus), and to be treated with beta-blockers and angiotensin-converting enzyme inhibitors, but they were less likely to have high-risk and emergency surgery performed.

The primary end point, perioperative death, occurred in 2.13% of the treated patients and 3.05% of the nontreated group. Compared with the rate in a propensity-matched cohort, the odds ratio adjusted for unbalanced covariates was 0.62 (95% CI 0.58–0.67) in favor of lipid treatment. Stratification by cardiac risk index revealed a number needed to treat of 186 for those with no risk factors, 60 for those with two risk factors, and 30 for those with four or more risk factors.

Unfortunately, this analysis was not able to take into account whether and for how long patients were receiving lipid-lowering therapy before hospitalization. It therefore does not answer the questions of whether starting lipid-lowering therapy before surgery is beneficial or whether stopping it is harmful. It also does not shed light on whether perioperative lipid-lowering increases the risk of rhabdomyolysis or liver disease.

Two recent retrospective cohort studies evaluated the outcomes in patients undergoing carotid endarterectomy.^19,20

Kennedy et al¹⁹ found that patients on a statin at the time of admission who had symptomatic carotid disease had lower rates of inhospital death (adjusted odds ratio 0.24, 95% CI 0.06–0.91) and ischemic stroke or death (adjusted odds ratio 0.55, 95% CI 0.31–0.97). However, cardiac outcomes among these symptomatic patients were not significantly improved (odds ratio 0.82, 95% CI 0.45–1.50), nor was there benefit for asymptomatic patients, raising the possibility that the positive findings were due to chance or that patients at lower baseline risk for vascular events may have less benefit.

McGirt et al²⁰ performed a similar study; they did not, however, distinguish whether patients had symptomatic vs asymptomatic carotid disease. The 30-day risk of perioperative stroke was lower in patients treated with a statin, with an odds ratio of 0.41 (95% CI 0.18–0.93); the odds ratio for death was 0.21 (95% CI 0.05–0.96). Cardiac outcomes were not significantly affected.

Coronary artery bypass graft surgery

According to the NCEP recommendations, nearly all patients undergoing CABG should already be on a statin before surgery since they all have known coronary artery disease. Multiple observational studies have offered confirmatory evidence that statins are beneficial in this setting.^34–38

Liakopoulos et al³⁹ evaluated whether the anti-inflammatory effects of statins may, in part, account for their beneficial effect in the perioperative period. The authors prospectively matched 18 patients who were taking statins and were referred for elective CABG with 18 patients who were not prescribed statins previously. The only major measured baseline characteristic that differed between the two groups was a statistically significantly lower LDL-C level in the statin group. The operative characteristics did not differ, and cytokine levels at baseline were similar.

Tumor necrosis factor alpha levels increased significantly in the control group but did not change significantly in the statin group. Interleukin 8 increased in both groups by a similar amount. Interleukin 6 (the major inducer of C-reactive protein) increased from baseline in both groups but did not increase nearly as much in the statin group as in the control group; the intergroup difference was statistically significant. The anti-inflammatory cytokine interleukin 10 increased minimally from baseline in the control group, while the statin group’s levels increased significantly above baseline and those of the control group.

Christenson⁴⁰ also found that inflammatory markers were improved with pre-CABG statin treatment in a small randomized trial in which patients received simvastatin 20 mg 4 weeks prior to CABG surgery vs no statin. Interestingly, far fewer statin-treated patients developed thrombocytosis (platelet count > 400 × 10⁹/L) than did control patients (3% vs 81%, P < .0001).

RISKS OF PERIOPERATIVE STATINS

The risks associated with statin therapy in general appear low, but specific perioperative risks have not been well studied.

Baigent et al,⁴¹ in a meta-analysis of randomized trials of nonperioperative statin therapy, found that rhabdomyolysis occurred in 9 (0.023%) of 39,884 patients receiving statins vs 6 (0.015%) of the 39,817 controls, with a number needed to harm of 12,500. Moreover, the rates of nonvascular death and cancer did not increase. It is plausible that the risk is somewhat greater in the perioperative setting but is likely not enough to outweigh the potential benefits, especially since the risk of ischemic vascular events is particularly high then.

Some of the perioperative studies cited above specifically addressed potential risks. For example, in the study by Schouten et al,³² mild creatine kinase elevations were more common in the statin-treated group, but the incidence of moderate and severe creatine kinase elevations did not differ significantly. No case of rhabdomyolysis occurred, and length of surgery was the only predictor of myopathy. MIRACL and PROVE-IT revealed similar safety profiles; aminotransferase levels normalized when statins were stopped, and no cases of rhabdomyolysis occurred.^11,12 In the vascular surgery study by Durazzo et al,²³ 1 (2%) of the 50 atorvastatin-treated patients developed both rhabdomyolysis and elevated aminotransferase levels that prompted discontinuation of the statin.

Overall, the observational studies do not indicate that statin continuation or treatment is harmful in perioperative patients. However, these studies did not specifically evaluate patients with acute insults from surgery such as sepsis, renal failure, or hepatitis. It is unknown what effect statin therapy would have in those patients and whether statins should be selectively discontinued in patients who develop major hepatic, musculoskeletal, or renal complications after surgery.

OUR RECOMMENDATIONS

Before CABG or vascular surgery

Given the NCEP recommendations, existing primary and secondary prevention studies, observational studies of CABG and noncardiac vascular surgery patients, and the one randomized trial of vascular surgery patients, data support the use of statins in nearly all patients undergoing cardiac or vascular surgery. We advocate starting statins in the perioperative period to take advantage of their rapid-acting pleiotropic effects, and continuing them long-term to take advantage of their lipid-lowering effects. This recommendation is in line with the recently released American College of Cardiology/American Heart Association (ACC/AHA) 2007 perioperative guidelines that state “for patients undergoing vascular surgery with or without clinical risk factors, statin use is reasonable.”⁴²

Although the ideal time to start statins is not certain, the study by Durazzo et al²³ suggests that they should be started at least 2 weeks before surgery if possible. Moreover, patients already taking statins should definitely not have their statins discontinued if at all possible.

Before major nonvascular surgery

For patients undergoing major nonvascular (intermediate-risk) surgery, physicians should first ascertain if the patient has an indication for statin therapy based on current nonsurgical lipid level recommendations. However, even if there is no clear indication for statin therapy based on NCEP guidelines, we endorse the recently released ACC/AHA perioperative guidelines that state that statin therapy can be considered in patients with a risk factor who are undergoing intermediate-risk procedures. Moreover, we wholeheartedly support the ACC/AHA’s strongest recommendation that patients who are already receiving statins and are undergoing noncardiac surgery should not have their statins discontinued.

When to discontinue statins?

The risk of harm overall appears to be minimal and certainly less than the likelihood of benefit. It is reasonable to observe patients postoperatively for adverse clinical events that may increase the risk of perioperative statin treatment, such as acute renal failure, hepatic failure, or sepsis, but whether statins should be stopped in patients with these complications remains unknown; we advocate individualizing the decision.

More studies needed

We need more data on whether moderate-risk patients undergoing moderate-risk surgery benefit from perioperative statin therapy, when therapy should be started, whether therapy should be started on the day of surgery if it was not started earlier, which statin and what doses are optimal, how long therapy should be continued, and what degree of risk is associated with perioperative statin therapy.

Fortunately, important data should be forthcoming in the next few years: the Dutch Echocardiographic Cardiac Risk Evaluation Applying Stress Echocardiography (DECREASE-IV) study⁴³ is a 4-year two-by-two factorial placebo-controlled study evaluating the use of fluvastatin (Lescol) and bisoprolol (Zebeta, a beta-blocker) separately and together in patients who are older than 40 years, are undergoing elective noncardiac surgery, have an estimated risk of cardiovascular death of more than 1%, have not used statins previously, and do not have elevated cholesterol.

The pendulum of expert opinion is swinging away from routinely recommending beta-blockers to prevent cardiac events in non-cardiac surgery patients. We won’t be abandoning the perioperative use of beta-blockers altogether, but we will probably be using them more selectively than in the past.

The latest factor driving the trend is the online publication in May 2008 of the results of the Perioperative Ischemic Evaluation (POISE) trial,¹ the largest placebo-controlled trial of perioperative beta-blocker use to date. In brief, in a cohort of patients with atherosclerotic disease or at risk for it who were undergoing noncardiac surgery, fewer patients who received extended-release metoprolol succinate had a myocardial infarction, but more of them died or had a stroke compared with those receiving placebo. (Extended-release metoprolol succinate is available in the United States as Toprol-XL and generically.)

Not so long ago, the pendulum was going the other way. After two small trials in the 1990s concluded that beta-blockers reduced the risk of perioperative cardiac events in selected patients with known or suspected coronary disease,^2,3 their perioperative use was subsequently endorsed by the Leapfrog Group and the Agency for Healthcare Research and Quality. The National Quality Forum included perioperative beta-blockade in its “Safe Practices for Better Healthcare 2006 update,”^4,5 and the Physician Consortium for Performance Improvement and the Surgical Care Improvement Project both listed it as a quality measure.

Since then, this practice has been closely studied, especially as concomitant research has failed to demonstrate that pre-operative coronary revascularization improves outcomes, even in the presence of ischemic disease. But evidence has been accumulating that routine use of beta-blockers may not benefit as many patients as was hoped, and may actually cause harm. The 2007 joint American College of Cardiology (ACC) and American Heart Association (AHA) guidelines on perioperative cardiovascular evaluation and care for noncardiac surgery gives its strongest recommendation (class I: benefit clearly outweighs risk) for perioperative beta-blocker use only for patients at high risk: those with known ischemic heart disease undergoing vascular surgery and those who are already on these drugs before surgery.⁶

However, there are still gaps in our knowledge. Perhaps, with proper implementation, we may be able to use beta-blockers to improve outcomes in patients at intermediate risk as well. In this paper, we review the rationale and the evidence for and against perioperative use of beta-blockers and provide practical guidance for internists and hospitalists.

WHY CARDIAC EVENTS OCCUR AFTER SURGERY

Adverse cardiovascular events such as myocardial infarction and unstable angina are the leading causes of death after surgery.⁷ Such events occur in approximately 1% of patients older than 50 years undergoing elective inpatient surgery, but this number may be higher (approximately 5%) in those with known or suspected coronary disease.^8,9 Perioperative cardiac events can also be harbingers of further complications, dramatically increasing hospital length of stay.¹⁰

Some ischemic events are caused by physiologic derangements involving the balance between inflammatory mediators, sympathetic tone, and oxygen supply and demand that occur under the stress of surgery. Others are more “traditional” in etiology, involving acute plaque rupture, thrombosis, and occlusion. Studies have consistently found a correlation between perioperative ischemia and cardiac events (both in-hospital and long-term) and death.^11–17 Other studies suggest that most perioperative cardiac infarcts are non-Q-wave events,¹⁸ and most events occur within the first few days after surgery, particularly the first 48 hours, when the effects of anesthetics, pain, fluid shifts, and physiologic derangements are greatest.

Factors that may trigger acute occlusion in the perioperative period include abrupt changes in sympathetic tone, increased levels of cortisol and catecholamines, and tissue hypoxia. Other potential triggers activated or increased by the stress of surgery include coagulation factors such as alterations in platelet function; inflammatory factors such as tumor necrosis factor alpha, interleukin 1, interleukin 6, and C-reactive protein; and metabolism of free fatty acids (which contribute to increased oxygen demand as well as endothelial dysfunction).^9,19,20

A 1996 autopsy study found that 38 (90%) of 42 patients who died of a perioperative infarct had evidence of acute plaque rupture or plaque hemorrhage on coronary sectioning, findings corroborated in another, similar study.^21,22 These studies suggest that multiple causes contribute to perioperative myocardial infarction, and a single strategy may not suffice for prevention.

IF BETA-BLOCKERS PROTECT, HOW DO THEY DO IT?

Beta-blockers have several effects that should, in theory, protect against cardiac events during and after surgery.²³ They reduce cardiac oxygen demand by reducing the force of contraction and the heart rate, and they increase the duration of diastole, when the heart muscle is perfused. They are also antiarrhythmic, and they may limit free radical production, metalloproteinase activity, and myocardial plaque inflammation.²⁴

Some researchers have speculated that using beta-blockers long-term may alter intra-cellular signaling processes, for example decreasing the expression of receptors that receive signals for cell death, which in turn may affect the response to reperfusion cell injury and death. If this is true, there may be an advantage to starting beta-blockers well in advance of surgery.²⁵

EARLY CLINICAL EVIDENCE IN FAVOR OF PERIOPERATIVE BETA-BLOCKER USE

Evidence in patients at high risk

Mangano et al,² in a study published in 1996, randomized 200 patients with known coronary disease or established risk factors for it who were undergoing noncardiac surgery to receive the beta-blocker atenolol orally and intravenously or placebo in the immediate perioperative period. Fewer patients in the atenolol group died in the first 6 months after hospital discharge (0 vs 8%, P < .001), the first year (3% vs 14%, P = .005), and the first 2 years (10% vs 21%, P = .019). However, there was no difference in short-term outcomes, and the study excluded patients who died in the immediate postoperative period. If these patients had been included in the analysis, the difference in the death rate at 2 years would not have been statistically significant.²⁶ Other critical findings: more patients in the atenolol group were using angiotensin-converting enzyme inhibitors and beta-blockers when they were discharged, and the placebo group had slightly more patients with prior myocardial infarction or diabetes.²⁷ (Atenolol is available in the United Sates as Tenormin and generically.)

Poldermans et al,³ in a study published in 1999, randomized 112 vascular surgery patients to receive either oral bisoprolol or placebo. These patients were selected from a larger cohort of 1,351 patients on the basis of high-risk clinical features and abnormal results on dobutamine echocardiography. Bisoprolol was started at least 1 week before surgery (range 7–89 days, mean 37 days), and patients were reevaluated before surgery so that the dose could be titrated to a goal heart rate of less than 60 beats per minute. After surgery, the drug was continued for another 30 days. The study was stopped early because the bisoprolol group had a 90% lower rate of non-fatal myocardial infarction and cardiac death at 30 days. Despite the study’s limitation (eg, enrolling selected patients and using an unblinded protocol), these compelling findings made a strong case for the use of beta-blockers perioperatively in patients at high risk, ie, those with ischemic heart disease who are undergoing major vascular surgery. (Bisoprolol is available in the United States as Zebeta and generically)

Evidence in patients at intermediate risk

Boersma et al²⁸ performed a follow-up to the study by Poldermans et al, published in 2001, in which they analyzed characteristics of all 1,351 patients who had been originally considered for enrollment. Using regression analysis, they identified seven clinical risk factors that predicted adverse cardiac events: angina, prior myocardial infarction, congestive heart failure, prior stroke, diabetes, renal failure, and age 70 years or older. Furthermore, for the entire cohort, patients receiving beta-blockers had a lower risk of cardiac complications (0.8%) than those not receiving beta-blockers (2.3%). In particular, the patients at intermediate risk (defined as having one or two risk factors) had a very low event rate regardless of stress test results, provided they were on beta-blockers: their risk of death or myocardial infarction was 0.9%, compared with 3.0% for those not on beta-blockers.

The authors concluded that dobutamine stress testing may not be necessary in patients at intermediate risk if beta-blockers are appropriately prescribed. However, others took issue with their data and conclusions, arguing that there have been so few trials that the data are still inconclusive and inadequate to ascertain the benefit of perioperative beta-blockade, particularly in patients not at high risk.^26,29

The Revised Cardiac Risk Index. Although the Boersma risk-factor index is not used in general practice, numerous experts^27,20–32 recommend a similar one, the Revised Cardiac Risk Index, devised by Lee et al.⁸ This index consists of six risk factors, each of which is worth one point:

• Congestive heart failure, based on history or examination
• Myocardial infarction, symptomatic ischemic heart disease, or a positive stress test
• Renal insufficiency (ie, serum creatinine level > 2 mg/dL)
• History of stroke or transient ischemic attack
• Diabetes requiring insulin
• High-risk surgery (defined as intrathoracic, intra-abdominal, or suprainguinal vascular surgery).

Patients with three or more points are considered to be at high risk, and those with one or two points are considered to be at intermediate risk. The ACC/AHA 2007 guidelines⁶ use a modified version of this index that considers the issue of surgical risk separately from the other five clinical conditions.

Devereaux et al³³ performed a meta-analysis, published in 2005, of 22 studies of perioperative beta-blockade. They concluded that beta-blockers had no discernable benefit in any outcome measured, including deaths from any cause, deaths from cardiovascular causes, other cardiac events, hypotension, bradycardia, and bronchospasm. However, they based this conclusion on the use of a 99% confidence interval for each relative risk, which they believed was justified because the trials were small and the numbers of events were only moderate. When the outcomes are assessed using the more common 95% confidence interval, benefit was detected in the combined end point of cardiovascular death, nonfatal myocardial infarction, and nonfatal cardiac arrest.

Yang et al,³⁴ Brady et al,³⁵ and Juul et al³⁶ performed three subsequent randomized trials that added to the controversy. Most of the patients in these trials were at intermediate or low risk, and none of the trials found a significant benefit with perioperative beta-blocker use. However, the protocols in these studies were different from the one in the study by Poldermans et al,³ which had found perioperative beta-blockade to be beneficial. Whereas patients in that earlier study started taking a beta-blocker at least 1 week before surgery (and on average much earlier), had their dose aggressively titrated to a target heart rate, and continued taking it for 30 days afterward, the protocols in the later trials called for the drug to be started within 24 hours before surgery and continued for only a short time afterward.

Lindenauer et al,³⁷ in a retrospective study published in 2005, found that fewer surgical patients who received beta-blockers in the hospital died in the hospital. The researchers used an administrative database of more than 780,000 patients who underwent noncardiac surgery, and they used propensity-score matching to compare the postoperative mortality rates of patients who received beta-blockers and a matched group in the same large cohort who did not. Beta-blockers were associated with a lower morality rate in patients in whom the Revised Cardiac Risk Index score was 3 or greater. However, although there was a trend toward a lower rate with beta-blocker use in patients whose score was 2 (ie, at intermediate risk), the difference was not statistically significant, and patients with a score of 0 or 1 saw no benefit and were possibly harmed.

The authors admitted that their study had a number of limitations, including a retrospective design and the use of an administrative database for information regarding risk index conditions and comorbidities. In addition, because they assumed that any patient who received a beta-blocker on the first 2 hospital days was receiving appropriate perioperative treatment, they may have incorrectly estimated the number of patients who actually received these drugs as a risk-reduction strategy. For instance, some patients at low risk could have received beta-blockers for treatment of a specific event, which would be reflected as an increase in event rates for this group. They also had no data on what medications the patients received before they were hospitalized or whether the dose was titrated effectively. The study excluded all patients with congestive heart failure and chronic obstructive pulmonary disease, who may be candidates for beta-blockers in actual practice. In fact, a recent observational study in patients with severe left ventricular dysfunction suggested that these drugs substantially reduced the incidence of death in the short term and the long term.³⁸ Finally, half the surgeries were nonelective, which makes extrapolation of their risk profile by the Revised Cardiac Risk Index difficult, since Lee et al excluded patients undergoing emergency surgery from the cohorts from which they derived and validated their index criteria.

Nevertheless, the authors concluded that patients at intermediate risk derive no benefit from perioperative beta-blocker use, and that the odds ratio for death was actually higher in patients with no risk factors who received a beta-blocker.

DOES PERIOPERATIVE BETA-BLOCKER USE CAUSE HARM?

The published data on whether perioperative beta-blocker use harms patients are conflicting and up to now have been limited.

Stone et al³⁹ reported a substantial incidence of bradycardia requiring atropine in patients treated with a single dose of a beta-blocker preoperatively, but the complications were not clearly characterized.

The Perioperative Beta-Blockade trial.³⁵ Significantly more patients given short-acting metoprolol had intraoperative falls in blood pressure and heart rate, and more required inotropic support during surgery, although the treating anesthesiologists refused to be blinded in that study. (Short-acting metoprolol is available in the United States as Lopressor and generically.)

Devereaux et al,³³ in their meta-analysis, found a higher risk of bradycardia requiring treatment (but not a higher risk of hypotension) in beta-blocker users in nine studies, including the study by Stone et al and the Perioperative Beta-Blockade trial (relative risk 2.27, 95% confidence interval 1.36–3.80).

Conversely, at least three other studies found no difference in rates of intraoperative events.^36,40,41 There are few data on the incidence of other complications such as perioperative pulmonary edema and bronchospasm.

POISE: THE FIRST LARGE RANDOMIZED TRIAL

In May 2008, results were published from POISE, the first large randomized controlled trial of perioperative beta-blockade.¹ An impressive 8,351 patients—most of them at intermediate risk—were randomized to receive extended-release metoprolol succinate or placebo starting just before surgery and continuing for 30 days afterward.

Although the incidence of the primary composite end point (cardiovascular death, nonfatal myocardial infarction, and nonfatal cardiac arrest) was lower at 30 days in the metoprolol group than in the placebo group (5.8% vs 6.9%, hazard ratio 0.83, P = .04), other findings were worrisome: more metoprolol recipients died of any cause (3.1% vs 2.3%, P = .03) or had a stroke (1.0% vs 0.5%, P = .005). The major contributor to the higher mortality rate in this group appears to have been sepsis.

How beta-blockers might promote death by sepsis is unclear. The authors offered two possible explanations: perhaps beta-blocker-induced hypotension predisposes patients to infection and sepsis, or perhaps the slower heart rate and lower force of contraction induced by beta-blockers could mask normal responses to systemic infection, which in turn could delay recognition and treatment or impede the normal immune response. These mechanisms, like others, are speculative.

The risks of other adverse outcomes such as bradycardia and hypotension were substantially higher in the metoprolol group. The authors also pointed out that most of the patients who suffered nonfatal strokes were subsequently disabled or incapacitated, while most of those who suffered nonfatal cardiac events did not progress to further cardiac problems.

This new study has not yet been rigorously debated, but it will likely come under scrutiny for its dosing regimen (extended-release metoprolol succinate 100 mg or placebo 2–4 hours before surgery; another 100 mg or placebo 6 hours after surgery or sooner if the heart rate was 80 beats per minute or more and the systolic blood pressure 100 mm Hg or higher; and then 200 mg or placebo 12 hours after the second dose and every 24 hours thereafter for 30 days). This was fairly aggressive, especially for patients who have never received a beta-blocker before. In contrast, the protocol for the Perioperative Beta-Blockade trial called for only 25 to 50 mg of short-acting metoprolol twice a day. Another criticism is that the medication was started only a few hours before surgery, although there is no current standard practice for either the dose or when the treatment should be started. The population had a fairly high rate of cerebrovascular disease (perhaps predisposing to stroke whenever blood pressure dropped), and 10% of patients were undergoing urgent or emergency surgery, which carries a higher risk of morbidity.

ANY ROLE FOR BETA-BLOCKERS IN THOSE AT INTERMEDIATE RISK?

Thus, in the past decade, the appropriate perioperative use of beta-blockers, which, after the findings by Mangano et al and Poldermans et al, were seen as potentially beneficial for any patient at risk of coronary disease, with little suggestion of harm, has become more clearly defined, and the risks are more evident. The most compelling evidence in favor of using them comes from patients with ischemic heart disease undergoing vascular surgery; the 2007 ACC/AHA guidelines recommend that this group be offered beta-blockers in the absence of a contraindication (class I recommendation: benefit clearly outweighs risk).⁶ The guidelines also point out that these drugs should be continued in patients already taking them for cardiac indications before surgery, because ischemia may be precipitated if a beta-blocker is abruptly discontinued.^42,43

Additionally, the guidelines recommend considering beta-blockers for vascular surgery patients at high cardiac risk (with a Revised Cardiac Risk Index score of 3 or more), even if they are not known to have ischemic heart disease. This is a class IIa recommendation (the benefit outweighs the risk, but more studies are required).

The guidelines also recommend that beta-blockers be considered for patients who have a score of 0 if they are undergoing vascular surgery (class IIb recommendation) or a score of 1 if they are undergoing vascular surgery (class IIa recommendation) or intermediate-risk surgery (class IIb recommendation). However, in view of the POISE results, these recommendations need to be carefully scrutinized.

These data notwithstanding, beta-blockers still might be beneficial in perioperative patients at intermediate risk.

Start beta-blockers sooner?

To help patients at intermediate risk (such as those with diabetes without known heart disease), we may need to do what Poldermans et al did³: instead of seeing patients only once a day or two before surgery, we may need to do the preoperative assessment as much as a month before and, if necessary, start a beta-blocker at a low dose, titrate it to a goal heart rate, and follow the patient closely up until surgery and afterward.

The importance of heart-rate control was illustrated in a recent cohort study of perioperative beta-blockers in vascular surgery patients,⁴⁴ in which higher beta-blocker doses, carefully monitored, were associated with less ischemia and cardiac enzyme release. In addition, long-term mortality rates were lower in patients with lower heart rates. And Poldermans et al⁴⁵ recently performed a study in more than 700 intermediate-risk patients who were divided into two groups, one that underwent preoperative stress testing and one that did not. Beta-blockers were given to both groups, and doses were titrated to a goal heart rate of less than 65. The patients with optimally controlled heart rates had the lowest event rates.

However, the logistics of such a program would be challenging. For the most part, internists and hospitalists involved in perioperative assessment do not control the timing of referral or surgery, and adjustments cannot be made for patients whose preoperative clinic visit falls only a few days before surgery. Instituting a second or third visit to assess the efficacy of beta-blockade burdens the patient and may not be practical.

Are all beta-blockers equivalent?

An additional factor is the choice of agent. While the most significant studies of perioperative beta-blockade have used beta-1 receptor-selective agents (ie, metoprolol, atenolol, and bisoprolol), there is no prospective evidence that any particular agent is superior. However, a recent retrospective analysis of elderly surgical patients did suggest that longer-acting beta-blockers may be preferable: patients who had been on atenolol in the year before surgery had a 20% lower risk of postoperative myocardial infarction or death than those who had been on short-acting metoprolol, with no difference in noncardiac outcomes.⁴⁶

The pendulum of expert opinion is swinging away from routinely recommending beta-blockers to prevent cardiac events in non-cardiac surgery patients. We won’t be abandoning the perioperative use of beta-blockers altogether, but we will probably be using them more selectively than in the past.

The latest factor driving the trend is the online publication in May 2008 of the results of the Perioperative Ischemic Evaluation (POISE) trial,¹ the largest placebo-controlled trial of perioperative beta-blocker use to date. In brief, in a cohort of patients with atherosclerotic disease or at risk for it who were undergoing noncardiac surgery, fewer patients who received extended-release metoprolol succinate had a myocardial infarction, but more of them died or had a stroke compared with those receiving placebo. (Extended-release metoprolol succinate is available in the United States as Toprol-XL and generically.)

Not so long ago, the pendulum was going the other way. After two small trials in the 1990s concluded that beta-blockers reduced the risk of perioperative cardiac events in selected patients with known or suspected coronary disease,^2,3 their perioperative use was subsequently endorsed by the Leapfrog Group and the Agency for Healthcare Research and Quality. The National Quality Forum included perioperative beta-blockade in its “Safe Practices for Better Healthcare 2006 update,”^4,5 and the Physician Consortium for Performance Improvement and the Surgical Care Improvement Project both listed it as a quality measure.

Since then, this practice has been closely studied, especially as concomitant research has failed to demonstrate that pre-operative coronary revascularization improves outcomes, even in the presence of ischemic disease. But evidence has been accumulating that routine use of beta-blockers may not benefit as many patients as was hoped, and may actually cause harm. The 2007 joint American College of Cardiology (ACC) and American Heart Association (AHA) guidelines on perioperative cardiovascular evaluation and care for noncardiac surgery gives its strongest recommendation (class I: benefit clearly outweighs risk) for perioperative beta-blocker use only for patients at high risk: those with known ischemic heart disease undergoing vascular surgery and those who are already on these drugs before surgery.⁶

However, there are still gaps in our knowledge. Perhaps, with proper implementation, we may be able to use beta-blockers to improve outcomes in patients at intermediate risk as well. In this paper, we review the rationale and the evidence for and against perioperative use of beta-blockers and provide practical guidance for internists and hospitalists.

WHY CARDIAC EVENTS OCCUR AFTER SURGERY

Adverse cardiovascular events such as myocardial infarction and unstable angina are the leading causes of death after surgery.⁷ Such events occur in approximately 1% of patients older than 50 years undergoing elective inpatient surgery, but this number may be higher (approximately 5%) in those with known or suspected coronary disease.^8,9 Perioperative cardiac events can also be harbingers of further complications, dramatically increasing hospital length of stay.¹⁰

Some ischemic events are caused by physiologic derangements involving the balance between inflammatory mediators, sympathetic tone, and oxygen supply and demand that occur under the stress of surgery. Others are more “traditional” in etiology, involving acute plaque rupture, thrombosis, and occlusion. Studies have consistently found a correlation between perioperative ischemia and cardiac events (both in-hospital and long-term) and death.^11–17 Other studies suggest that most perioperative cardiac infarcts are non-Q-wave events,¹⁸ and most events occur within the first few days after surgery, particularly the first 48 hours, when the effects of anesthetics, pain, fluid shifts, and physiologic derangements are greatest.

Factors that may trigger acute occlusion in the perioperative period include abrupt changes in sympathetic tone, increased levels of cortisol and catecholamines, and tissue hypoxia. Other potential triggers activated or increased by the stress of surgery include coagulation factors such as alterations in platelet function; inflammatory factors such as tumor necrosis factor alpha, interleukin 1, interleukin 6, and C-reactive protein; and metabolism of free fatty acids (which contribute to increased oxygen demand as well as endothelial dysfunction).^9,19,20

A 1996 autopsy study found that 38 (90%) of 42 patients who died of a perioperative infarct had evidence of acute plaque rupture or plaque hemorrhage on coronary sectioning, findings corroborated in another, similar study.^21,22 These studies suggest that multiple causes contribute to perioperative myocardial infarction, and a single strategy may not suffice for prevention.

IF BETA-BLOCKERS PROTECT, HOW DO THEY DO IT?

Beta-blockers have several effects that should, in theory, protect against cardiac events during and after surgery.²³ They reduce cardiac oxygen demand by reducing the force of contraction and the heart rate, and they increase the duration of diastole, when the heart muscle is perfused. They are also antiarrhythmic, and they may limit free radical production, metalloproteinase activity, and myocardial plaque inflammation.²⁴

Some researchers have speculated that using beta-blockers long-term may alter intra-cellular signaling processes, for example decreasing the expression of receptors that receive signals for cell death, which in turn may affect the response to reperfusion cell injury and death. If this is true, there may be an advantage to starting beta-blockers well in advance of surgery.²⁵

EARLY CLINICAL EVIDENCE IN FAVOR OF PERIOPERATIVE BETA-BLOCKER USE

Evidence in patients at high risk

Mangano et al,² in a study published in 1996, randomized 200 patients with known coronary disease or established risk factors for it who were undergoing noncardiac surgery to receive the beta-blocker atenolol orally and intravenously or placebo in the immediate perioperative period. Fewer patients in the atenolol group died in the first 6 months after hospital discharge (0 vs 8%, P < .001), the first year (3% vs 14%, P = .005), and the first 2 years (10% vs 21%, P = .019). However, there was no difference in short-term outcomes, and the study excluded patients who died in the immediate postoperative period. If these patients had been included in the analysis, the difference in the death rate at 2 years would not have been statistically significant.²⁶ Other critical findings: more patients in the atenolol group were using angiotensin-converting enzyme inhibitors and beta-blockers when they were discharged, and the placebo group had slightly more patients with prior myocardial infarction or diabetes.²⁷ (Atenolol is available in the United Sates as Tenormin and generically.)

Poldermans et al,³ in a study published in 1999, randomized 112 vascular surgery patients to receive either oral bisoprolol or placebo. These patients were selected from a larger cohort of 1,351 patients on the basis of high-risk clinical features and abnormal results on dobutamine echocardiography. Bisoprolol was started at least 1 week before surgery (range 7–89 days, mean 37 days), and patients were reevaluated before surgery so that the dose could be titrated to a goal heart rate of less than 60 beats per minute. After surgery, the drug was continued for another 30 days. The study was stopped early because the bisoprolol group had a 90% lower rate of non-fatal myocardial infarction and cardiac death at 30 days. Despite the study’s limitation (eg, enrolling selected patients and using an unblinded protocol), these compelling findings made a strong case for the use of beta-blockers perioperatively in patients at high risk, ie, those with ischemic heart disease who are undergoing major vascular surgery. (Bisoprolol is available in the United States as Zebeta and generically)

Evidence in patients at intermediate risk

Boersma et al²⁸ performed a follow-up to the study by Poldermans et al, published in 2001, in which they analyzed characteristics of all 1,351 patients who had been originally considered for enrollment. Using regression analysis, they identified seven clinical risk factors that predicted adverse cardiac events: angina, prior myocardial infarction, congestive heart failure, prior stroke, diabetes, renal failure, and age 70 years or older. Furthermore, for the entire cohort, patients receiving beta-blockers had a lower risk of cardiac complications (0.8%) than those not receiving beta-blockers (2.3%). In particular, the patients at intermediate risk (defined as having one or two risk factors) had a very low event rate regardless of stress test results, provided they were on beta-blockers: their risk of death or myocardial infarction was 0.9%, compared with 3.0% for those not on beta-blockers.

The authors concluded that dobutamine stress testing may not be necessary in patients at intermediate risk if beta-blockers are appropriately prescribed. However, others took issue with their data and conclusions, arguing that there have been so few trials that the data are still inconclusive and inadequate to ascertain the benefit of perioperative beta-blockade, particularly in patients not at high risk.^26,29

The Revised Cardiac Risk Index. Although the Boersma risk-factor index is not used in general practice, numerous experts^27,20–32 recommend a similar one, the Revised Cardiac Risk Index, devised by Lee et al.⁸ This index consists of six risk factors, each of which is worth one point:

• Congestive heart failure, based on history or examination
• Myocardial infarction, symptomatic ischemic heart disease, or a positive stress test
• Renal insufficiency (ie, serum creatinine level > 2 mg/dL)
• History of stroke or transient ischemic attack
• Diabetes requiring insulin
• High-risk surgery (defined as intrathoracic, intra-abdominal, or suprainguinal vascular surgery).

Patients with three or more points are considered to be at high risk, and those with one or two points are considered to be at intermediate risk. The ACC/AHA 2007 guidelines⁶ use a modified version of this index that considers the issue of surgical risk separately from the other five clinical conditions.

Devereaux et al³³ performed a meta-analysis, published in 2005, of 22 studies of perioperative beta-blockade. They concluded that beta-blockers had no discernable benefit in any outcome measured, including deaths from any cause, deaths from cardiovascular causes, other cardiac events, hypotension, bradycardia, and bronchospasm. However, they based this conclusion on the use of a 99% confidence interval for each relative risk, which they believed was justified because the trials were small and the numbers of events were only moderate. When the outcomes are assessed using the more common 95% confidence interval, benefit was detected in the combined end point of cardiovascular death, nonfatal myocardial infarction, and nonfatal cardiac arrest.

Yang et al,³⁴ Brady et al,³⁵ and Juul et al³⁶ performed three subsequent randomized trials that added to the controversy. Most of the patients in these trials were at intermediate or low risk, and none of the trials found a significant benefit with perioperative beta-blocker use. However, the protocols in these studies were different from the one in the study by Poldermans et al,³ which had found perioperative beta-blockade to be beneficial. Whereas patients in that earlier study started taking a beta-blocker at least 1 week before surgery (and on average much earlier), had their dose aggressively titrated to a target heart rate, and continued taking it for 30 days afterward, the protocols in the later trials called for the drug to be started within 24 hours before surgery and continued for only a short time afterward.

Lindenauer et al,³⁷ in a retrospective study published in 2005, found that fewer surgical patients who received beta-blockers in the hospital died in the hospital. The researchers used an administrative database of more than 780,000 patients who underwent noncardiac surgery, and they used propensity-score matching to compare the postoperative mortality rates of patients who received beta-blockers and a matched group in the same large cohort who did not. Beta-blockers were associated with a lower morality rate in patients in whom the Revised Cardiac Risk Index score was 3 or greater. However, although there was a trend toward a lower rate with beta-blocker use in patients whose score was 2 (ie, at intermediate risk), the difference was not statistically significant, and patients with a score of 0 or 1 saw no benefit and were possibly harmed.

The authors admitted that their study had a number of limitations, including a retrospective design and the use of an administrative database for information regarding risk index conditions and comorbidities. In addition, because they assumed that any patient who received a beta-blocker on the first 2 hospital days was receiving appropriate perioperative treatment, they may have incorrectly estimated the number of patients who actually received these drugs as a risk-reduction strategy. For instance, some patients at low risk could have received beta-blockers for treatment of a specific event, which would be reflected as an increase in event rates for this group. They also had no data on what medications the patients received before they were hospitalized or whether the dose was titrated effectively. The study excluded all patients with congestive heart failure and chronic obstructive pulmonary disease, who may be candidates for beta-blockers in actual practice. In fact, a recent observational study in patients with severe left ventricular dysfunction suggested that these drugs substantially reduced the incidence of death in the short term and the long term.³⁸ Finally, half the surgeries were nonelective, which makes extrapolation of their risk profile by the Revised Cardiac Risk Index difficult, since Lee et al excluded patients undergoing emergency surgery from the cohorts from which they derived and validated their index criteria.

Nevertheless, the authors concluded that patients at intermediate risk derive no benefit from perioperative beta-blocker use, and that the odds ratio for death was actually higher in patients with no risk factors who received a beta-blocker.

DOES PERIOPERATIVE BETA-BLOCKER USE CAUSE HARM?

The published data on whether perioperative beta-blocker use harms patients are conflicting and up to now have been limited.

Stone et al³⁹ reported a substantial incidence of bradycardia requiring atropine in patients treated with a single dose of a beta-blocker preoperatively, but the complications were not clearly characterized.

The Perioperative Beta-Blockade trial.³⁵ Significantly more patients given short-acting metoprolol had intraoperative falls in blood pressure and heart rate, and more required inotropic support during surgery, although the treating anesthesiologists refused to be blinded in that study. (Short-acting metoprolol is available in the United States as Lopressor and generically.)

Devereaux et al,³³ in their meta-analysis, found a higher risk of bradycardia requiring treatment (but not a higher risk of hypotension) in beta-blocker users in nine studies, including the study by Stone et al and the Perioperative Beta-Blockade trial (relative risk 2.27, 95% confidence interval 1.36–3.80).

Conversely, at least three other studies found no difference in rates of intraoperative events.^36,40,41 There are few data on the incidence of other complications such as perioperative pulmonary edema and bronchospasm.

POISE: THE FIRST LARGE RANDOMIZED TRIAL

In May 2008, results were published from POISE, the first large randomized controlled trial of perioperative beta-blockade.¹ An impressive 8,351 patients—most of them at intermediate risk—were randomized to receive extended-release metoprolol succinate or placebo starting just before surgery and continuing for 30 days afterward.

Although the incidence of the primary composite end point (cardiovascular death, nonfatal myocardial infarction, and nonfatal cardiac arrest) was lower at 30 days in the metoprolol group than in the placebo group (5.8% vs 6.9%, hazard ratio 0.83, P = .04), other findings were worrisome: more metoprolol recipients died of any cause (3.1% vs 2.3%, P = .03) or had a stroke (1.0% vs 0.5%, P = .005). The major contributor to the higher mortality rate in this group appears to have been sepsis.

How beta-blockers might promote death by sepsis is unclear. The authors offered two possible explanations: perhaps beta-blocker-induced hypotension predisposes patients to infection and sepsis, or perhaps the slower heart rate and lower force of contraction induced by beta-blockers could mask normal responses to systemic infection, which in turn could delay recognition and treatment or impede the normal immune response. These mechanisms, like others, are speculative.

The risks of other adverse outcomes such as bradycardia and hypotension were substantially higher in the metoprolol group. The authors also pointed out that most of the patients who suffered nonfatal strokes were subsequently disabled or incapacitated, while most of those who suffered nonfatal cardiac events did not progress to further cardiac problems.

This new study has not yet been rigorously debated, but it will likely come under scrutiny for its dosing regimen (extended-release metoprolol succinate 100 mg or placebo 2–4 hours before surgery; another 100 mg or placebo 6 hours after surgery or sooner if the heart rate was 80 beats per minute or more and the systolic blood pressure 100 mm Hg or higher; and then 200 mg or placebo 12 hours after the second dose and every 24 hours thereafter for 30 days). This was fairly aggressive, especially for patients who have never received a beta-blocker before. In contrast, the protocol for the Perioperative Beta-Blockade trial called for only 25 to 50 mg of short-acting metoprolol twice a day. Another criticism is that the medication was started only a few hours before surgery, although there is no current standard practice for either the dose or when the treatment should be started. The population had a fairly high rate of cerebrovascular disease (perhaps predisposing to stroke whenever blood pressure dropped), and 10% of patients were undergoing urgent or emergency surgery, which carries a higher risk of morbidity.

ANY ROLE FOR BETA-BLOCKERS IN THOSE AT INTERMEDIATE RISK?

Thus, in the past decade, the appropriate perioperative use of beta-blockers, which, after the findings by Mangano et al and Poldermans et al, were seen as potentially beneficial for any patient at risk of coronary disease, with little suggestion of harm, has become more clearly defined, and the risks are more evident. The most compelling evidence in favor of using them comes from patients with ischemic heart disease undergoing vascular surgery; the 2007 ACC/AHA guidelines recommend that this group be offered beta-blockers in the absence of a contraindication (class I recommendation: benefit clearly outweighs risk).⁶ The guidelines also point out that these drugs should be continued in patients already taking them for cardiac indications before surgery, because ischemia may be precipitated if a beta-blocker is abruptly discontinued.^42,43

Additionally, the guidelines recommend considering beta-blockers for vascular surgery patients at high cardiac risk (with a Revised Cardiac Risk Index score of 3 or more), even if they are not known to have ischemic heart disease. This is a class IIa recommendation (the benefit outweighs the risk, but more studies are required).

The guidelines also recommend that beta-blockers be considered for patients who have a score of 0 if they are undergoing vascular surgery (class IIb recommendation) or a score of 1 if they are undergoing vascular surgery (class IIa recommendation) or intermediate-risk surgery (class IIb recommendation). However, in view of the POISE results, these recommendations need to be carefully scrutinized.

These data notwithstanding, beta-blockers still might be beneficial in perioperative patients at intermediate risk.

Start beta-blockers sooner?

To help patients at intermediate risk (such as those with diabetes without known heart disease), we may need to do what Poldermans et al did³: instead of seeing patients only once a day or two before surgery, we may need to do the preoperative assessment as much as a month before and, if necessary, start a beta-blocker at a low dose, titrate it to a goal heart rate, and follow the patient closely up until surgery and afterward.

The importance of heart-rate control was illustrated in a recent cohort study of perioperative beta-blockers in vascular surgery patients,⁴⁴ in which higher beta-blocker doses, carefully monitored, were associated with less ischemia and cardiac enzyme release. In addition, long-term mortality rates were lower in patients with lower heart rates. And Poldermans et al⁴⁵ recently performed a study in more than 700 intermediate-risk patients who were divided into two groups, one that underwent preoperative stress testing and one that did not. Beta-blockers were given to both groups, and doses were titrated to a goal heart rate of less than 65. The patients with optimally controlled heart rates had the lowest event rates.

However, the logistics of such a program would be challenging. For the most part, internists and hospitalists involved in perioperative assessment do not control the timing of referral or surgery, and adjustments cannot be made for patients whose preoperative clinic visit falls only a few days before surgery. Instituting a second or third visit to assess the efficacy of beta-blockade burdens the patient and may not be practical.

Are all beta-blockers equivalent?

An additional factor is the choice of agent. While the most significant studies of perioperative beta-blockade have used beta-1 receptor-selective agents (ie, metoprolol, atenolol, and bisoprolol), there is no prospective evidence that any particular agent is superior. However, a recent retrospective analysis of elderly surgical patients did suggest that longer-acting beta-blockers may be preferable: patients who had been on atenolol in the year before surgery had a 20% lower risk of postoperative myocardial infarction or death than those who had been on short-acting metoprolol, with no difference in noncardiac outcomes.⁴⁶

The pendulum of expert opinion is swinging away from routinely recommending beta-blockers to prevent cardiac events in non-cardiac surgery patients. We won’t be abandoning the perioperative use of beta-blockers altogether, but we will probably be using them more selectively than in the past.

The latest factor driving the trend is the online publication in May 2008 of the results of the Perioperative Ischemic Evaluation (POISE) trial,¹ the largest placebo-controlled trial of perioperative beta-blocker use to date. In brief, in a cohort of patients with atherosclerotic disease or at risk for it who were undergoing noncardiac surgery, fewer patients who received extended-release metoprolol succinate had a myocardial infarction, but more of them died or had a stroke compared with those receiving placebo. (Extended-release metoprolol succinate is available in the United States as Toprol-XL and generically.)

Not so long ago, the pendulum was going the other way. After two small trials in the 1990s concluded that beta-blockers reduced the risk of perioperative cardiac events in selected patients with known or suspected coronary disease,^2,3 their perioperative use was subsequently endorsed by the Leapfrog Group and the Agency for Healthcare Research and Quality. The National Quality Forum included perioperative beta-blockade in its “Safe Practices for Better Healthcare 2006 update,”^4,5 and the Physician Consortium for Performance Improvement and the Surgical Care Improvement Project both listed it as a quality measure.

Since then, this practice has been closely studied, especially as concomitant research has failed to demonstrate that pre-operative coronary revascularization improves outcomes, even in the presence of ischemic disease. But evidence has been accumulating that routine use of beta-blockers may not benefit as many patients as was hoped, and may actually cause harm. The 2007 joint American College of Cardiology (ACC) and American Heart Association (AHA) guidelines on perioperative cardiovascular evaluation and care for noncardiac surgery gives its strongest recommendation (class I: benefit clearly outweighs risk) for perioperative beta-blocker use only for patients at high risk: those with known ischemic heart disease undergoing vascular surgery and those who are already on these drugs before surgery.⁶

However, there are still gaps in our knowledge. Perhaps, with proper implementation, we may be able to use beta-blockers to improve outcomes in patients at intermediate risk as well. In this paper, we review the rationale and the evidence for and against perioperative use of beta-blockers and provide practical guidance for internists and hospitalists.

WHY CARDIAC EVENTS OCCUR AFTER SURGERY

Adverse cardiovascular events such as myocardial infarction and unstable angina are the leading causes of death after surgery.⁷ Such events occur in approximately 1% of patients older than 50 years undergoing elective inpatient surgery, but this number may be higher (approximately 5%) in those with known or suspected coronary disease.^8,9 Perioperative cardiac events can also be harbingers of further complications, dramatically increasing hospital length of stay.¹⁰

Some ischemic events are caused by physiologic derangements involving the balance between inflammatory mediators, sympathetic tone, and oxygen supply and demand that occur under the stress of surgery. Others are more “traditional” in etiology, involving acute plaque rupture, thrombosis, and occlusion. Studies have consistently found a correlation between perioperative ischemia and cardiac events (both in-hospital and long-term) and death.^11–17 Other studies suggest that most perioperative cardiac infarcts are non-Q-wave events,¹⁸ and most events occur within the first few days after surgery, particularly the first 48 hours, when the effects of anesthetics, pain, fluid shifts, and physiologic derangements are greatest.

Factors that may trigger acute occlusion in the perioperative period include abrupt changes in sympathetic tone, increased levels of cortisol and catecholamines, and tissue hypoxia. Other potential triggers activated or increased by the stress of surgery include coagulation factors such as alterations in platelet function; inflammatory factors such as tumor necrosis factor alpha, interleukin 1, interleukin 6, and C-reactive protein; and metabolism of free fatty acids (which contribute to increased oxygen demand as well as endothelial dysfunction).^9,19,20

A 1996 autopsy study found that 38 (90%) of 42 patients who died of a perioperative infarct had evidence of acute plaque rupture or plaque hemorrhage on coronary sectioning, findings corroborated in another, similar study.^21,22 These studies suggest that multiple causes contribute to perioperative myocardial infarction, and a single strategy may not suffice for prevention.

IF BETA-BLOCKERS PROTECT, HOW DO THEY DO IT?

Beta-blockers have several effects that should, in theory, protect against cardiac events during and after surgery.²³ They reduce cardiac oxygen demand by reducing the force of contraction and the heart rate, and they increase the duration of diastole, when the heart muscle is perfused. They are also antiarrhythmic, and they may limit free radical production, metalloproteinase activity, and myocardial plaque inflammation.²⁴

Some researchers have speculated that using beta-blockers long-term may alter intra-cellular signaling processes, for example decreasing the expression of receptors that receive signals for cell death, which in turn may affect the response to reperfusion cell injury and death. If this is true, there may be an advantage to starting beta-blockers well in advance of surgery.²⁵

EARLY CLINICAL EVIDENCE IN FAVOR OF PERIOPERATIVE BETA-BLOCKER USE

Evidence in patients at high risk

Mangano et al,² in a study published in 1996, randomized 200 patients with known coronary disease or established risk factors for it who were undergoing noncardiac surgery to receive the beta-blocker atenolol orally and intravenously or placebo in the immediate perioperative period. Fewer patients in the atenolol group died in the first 6 months after hospital discharge (0 vs 8%, P < .001), the first year (3% vs 14%, P = .005), and the first 2 years (10% vs 21%, P = .019). However, there was no difference in short-term outcomes, and the study excluded patients who died in the immediate postoperative period. If these patients had been included in the analysis, the difference in the death rate at 2 years would not have been statistically significant.²⁶ Other critical findings: more patients in the atenolol group were using angiotensin-converting enzyme inhibitors and beta-blockers when they were discharged, and the placebo group had slightly more patients with prior myocardial infarction or diabetes.²⁷ (Atenolol is available in the United Sates as Tenormin and generically.)

Poldermans et al,³ in a study published in 1999, randomized 112 vascular surgery patients to receive either oral bisoprolol or placebo. These patients were selected from a larger cohort of 1,351 patients on the basis of high-risk clinical features and abnormal results on dobutamine echocardiography. Bisoprolol was started at least 1 week before surgery (range 7–89 days, mean 37 days), and patients were reevaluated before surgery so that the dose could be titrated to a goal heart rate of less than 60 beats per minute. After surgery, the drug was continued for another 30 days. The study was stopped early because the bisoprolol group had a 90% lower rate of non-fatal myocardial infarction and cardiac death at 30 days. Despite the study’s limitation (eg, enrolling selected patients and using an unblinded protocol), these compelling findings made a strong case for the use of beta-blockers perioperatively in patients at high risk, ie, those with ischemic heart disease who are undergoing major vascular surgery. (Bisoprolol is available in the United States as Zebeta and generically)

Evidence in patients at intermediate risk

Boersma et al²⁸ performed a follow-up to the study by Poldermans et al, published in 2001, in which they analyzed characteristics of all 1,351 patients who had been originally considered for enrollment. Using regression analysis, they identified seven clinical risk factors that predicted adverse cardiac events: angina, prior myocardial infarction, congestive heart failure, prior stroke, diabetes, renal failure, and age 70 years or older. Furthermore, for the entire cohort, patients receiving beta-blockers had a lower risk of cardiac complications (0.8%) than those not receiving beta-blockers (2.3%). In particular, the patients at intermediate risk (defined as having one or two risk factors) had a very low event rate regardless of stress test results, provided they were on beta-blockers: their risk of death or myocardial infarction was 0.9%, compared with 3.0% for those not on beta-blockers.

The authors concluded that dobutamine stress testing may not be necessary in patients at intermediate risk if beta-blockers are appropriately prescribed. However, others took issue with their data and conclusions, arguing that there have been so few trials that the data are still inconclusive and inadequate to ascertain the benefit of perioperative beta-blockade, particularly in patients not at high risk.^26,29

The Revised Cardiac Risk Index. Although the Boersma risk-factor index is not used in general practice, numerous experts^27,20–32 recommend a similar one, the Revised Cardiac Risk Index, devised by Lee et al.⁸ This index consists of six risk factors, each of which is worth one point:

• Congestive heart failure, based on history or examination
• Myocardial infarction, symptomatic ischemic heart disease, or a positive stress test
• Renal insufficiency (ie, serum creatinine level > 2 mg/dL)
• History of stroke or transient ischemic attack
• Diabetes requiring insulin
• High-risk surgery (defined as intrathoracic, intra-abdominal, or suprainguinal vascular surgery).

Patients with three or more points are considered to be at high risk, and those with one or two points are considered to be at intermediate risk. The ACC/AHA 2007 guidelines⁶ use a modified version of this index that considers the issue of surgical risk separately from the other five clinical conditions.

Devereaux et al³³ performed a meta-analysis, published in 2005, of 22 studies of perioperative beta-blockade. They concluded that beta-blockers had no discernable benefit in any outcome measured, including deaths from any cause, deaths from cardiovascular causes, other cardiac events, hypotension, bradycardia, and bronchospasm. However, they based this conclusion on the use of a 99% confidence interval for each relative risk, which they believed was justified because the trials were small and the numbers of events were only moderate. When the outcomes are assessed using the more common 95% confidence interval, benefit was detected in the combined end point of cardiovascular death, nonfatal myocardial infarction, and nonfatal cardiac arrest.

Yang et al,³⁴ Brady et al,³⁵ and Juul et al³⁶ performed three subsequent randomized trials that added to the controversy. Most of the patients in these trials were at intermediate or low risk, and none of the trials found a significant benefit with perioperative beta-blocker use. However, the protocols in these studies were different from the one in the study by Poldermans et al,³ which had found perioperative beta-blockade to be beneficial. Whereas patients in that earlier study started taking a beta-blocker at least 1 week before surgery (and on average much earlier), had their dose aggressively titrated to a target heart rate, and continued taking it for 30 days afterward, the protocols in the later trials called for the drug to be started within 24 hours before surgery and continued for only a short time afterward.

Lindenauer et al,³⁷ in a retrospective study published in 2005, found that fewer surgical patients who received beta-blockers in the hospital died in the hospital. The researchers used an administrative database of more than 780,000 patients who underwent noncardiac surgery, and they used propensity-score matching to compare the postoperative mortality rates of patients who received beta-blockers and a matched group in the same large cohort who did not. Beta-blockers were associated with a lower morality rate in patients in whom the Revised Cardiac Risk Index score was 3 or greater. However, although there was a trend toward a lower rate with beta-blocker use in patients whose score was 2 (ie, at intermediate risk), the difference was not statistically significant, and patients with a score of 0 or 1 saw no benefit and were possibly harmed.

The authors admitted that their study had a number of limitations, including a retrospective design and the use of an administrative database for information regarding risk index conditions and comorbidities. In addition, because they assumed that any patient who received a beta-blocker on the first 2 hospital days was receiving appropriate perioperative treatment, they may have incorrectly estimated the number of patients who actually received these drugs as a risk-reduction strategy. For instance, some patients at low risk could have received beta-blockers for treatment of a specific event, which would be reflected as an increase in event rates for this group. They also had no data on what medications the patients received before they were hospitalized or whether the dose was titrated effectively. The study excluded all patients with congestive heart failure and chronic obstructive pulmonary disease, who may be candidates for beta-blockers in actual practice. In fact, a recent observational study in patients with severe left ventricular dysfunction suggested that these drugs substantially reduced the incidence of death in the short term and the long term.³⁸ Finally, half the surgeries were nonelective, which makes extrapolation of their risk profile by the Revised Cardiac Risk Index difficult, since Lee et al excluded patients undergoing emergency surgery from the cohorts from which they derived and validated their index criteria.

Nevertheless, the authors concluded that patients at intermediate risk derive no benefit from perioperative beta-blocker use, and that the odds ratio for death was actually higher in patients with no risk factors who received a beta-blocker.

DOES PERIOPERATIVE BETA-BLOCKER USE CAUSE HARM?

The published data on whether perioperative beta-blocker use harms patients are conflicting and up to now have been limited.

Stone et al³⁹ reported a substantial incidence of bradycardia requiring atropine in patients treated with a single dose of a beta-blocker preoperatively, but the complications were not clearly characterized.

The Perioperative Beta-Blockade trial.³⁵ Significantly more patients given short-acting metoprolol had intraoperative falls in blood pressure and heart rate, and more required inotropic support during surgery, although the treating anesthesiologists refused to be blinded in that study. (Short-acting metoprolol is available in the United States as Lopressor and generically.)

Devereaux et al,³³ in their meta-analysis, found a higher risk of bradycardia requiring treatment (but not a higher risk of hypotension) in beta-blocker users in nine studies, including the study by Stone et al and the Perioperative Beta-Blockade trial (relative risk 2.27, 95% confidence interval 1.36–3.80).

Conversely, at least three other studies found no difference in rates of intraoperative events.^36,40,41 There are few data on the incidence of other complications such as perioperative pulmonary edema and bronchospasm.

POISE: THE FIRST LARGE RANDOMIZED TRIAL

In May 2008, results were published from POISE, the first large randomized controlled trial of perioperative beta-blockade.¹ An impressive 8,351 patients—most of them at intermediate risk—were randomized to receive extended-release metoprolol succinate or placebo starting just before surgery and continuing for 30 days afterward.

Although the incidence of the primary composite end point (cardiovascular death, nonfatal myocardial infarction, and nonfatal cardiac arrest) was lower at 30 days in the metoprolol group than in the placebo group (5.8% vs 6.9%, hazard ratio 0.83, P = .04), other findings were worrisome: more metoprolol recipients died of any cause (3.1% vs 2.3%, P = .03) or had a stroke (1.0% vs 0.5%, P = .005). The major contributor to the higher mortality rate in this group appears to have been sepsis.

How beta-blockers might promote death by sepsis is unclear. The authors offered two possible explanations: perhaps beta-blocker-induced hypotension predisposes patients to infection and sepsis, or perhaps the slower heart rate and lower force of contraction induced by beta-blockers could mask normal responses to systemic infection, which in turn could delay recognition and treatment or impede the normal immune response. These mechanisms, like others, are speculative.

The risks of other adverse outcomes such as bradycardia and hypotension were substantially higher in the metoprolol group. The authors also pointed out that most of the patients who suffered nonfatal strokes were subsequently disabled or incapacitated, while most of those who suffered nonfatal cardiac events did not progress to further cardiac problems.

This new study has not yet been rigorously debated, but it will likely come under scrutiny for its dosing regimen (extended-release metoprolol succinate 100 mg or placebo 2–4 hours before surgery; another 100 mg or placebo 6 hours after surgery or sooner if the heart rate was 80 beats per minute or more and the systolic blood pressure 100 mm Hg or higher; and then 200 mg or placebo 12 hours after the second dose and every 24 hours thereafter for 30 days). This was fairly aggressive, especially for patients who have never received a beta-blocker before. In contrast, the protocol for the Perioperative Beta-Blockade trial called for only 25 to 50 mg of short-acting metoprolol twice a day. Another criticism is that the medication was started only a few hours before surgery, although there is no current standard practice for either the dose or when the treatment should be started. The population had a fairly high rate of cerebrovascular disease (perhaps predisposing to stroke whenever blood pressure dropped), and 10% of patients were undergoing urgent or emergency surgery, which carries a higher risk of morbidity.

ANY ROLE FOR BETA-BLOCKERS IN THOSE AT INTERMEDIATE RISK?

Thus, in the past decade, the appropriate perioperative use of beta-blockers, which, after the findings by Mangano et al and Poldermans et al, were seen as potentially beneficial for any patient at risk of coronary disease, with little suggestion of harm, has become more clearly defined, and the risks are more evident. The most compelling evidence in favor of using them comes from patients with ischemic heart disease undergoing vascular surgery; the 2007 ACC/AHA guidelines recommend that this group be offered beta-blockers in the absence of a contraindication (class I recommendation: benefit clearly outweighs risk).⁶ The guidelines also point out that these drugs should be continued in patients already taking them for cardiac indications before surgery, because ischemia may be precipitated if a beta-blocker is abruptly discontinued.^42,43

Additionally, the guidelines recommend considering beta-blockers for vascular surgery patients at high cardiac risk (with a Revised Cardiac Risk Index score of 3 or more), even if they are not known to have ischemic heart disease. This is a class IIa recommendation (the benefit outweighs the risk, but more studies are required).

The guidelines also recommend that beta-blockers be considered for patients who have a score of 0 if they are undergoing vascular surgery (class IIb recommendation) or a score of 1 if they are undergoing vascular surgery (class IIa recommendation) or intermediate-risk surgery (class IIb recommendation). However, in view of the POISE results, these recommendations need to be carefully scrutinized.

These data notwithstanding, beta-blockers still might be beneficial in perioperative patients at intermediate risk.

Start beta-blockers sooner?

To help patients at intermediate risk (such as those with diabetes without known heart disease), we may need to do what Poldermans et al did³: instead of seeing patients only once a day or two before surgery, we may need to do the preoperative assessment as much as a month before and, if necessary, start a beta-blocker at a low dose, titrate it to a goal heart rate, and follow the patient closely up until surgery and afterward.

The importance of heart-rate control was illustrated in a recent cohort study of perioperative beta-blockers in vascular surgery patients,⁴⁴ in which higher beta-blocker doses, carefully monitored, were associated with less ischemia and cardiac enzyme release. In addition, long-term mortality rates were lower in patients with lower heart rates. And Poldermans et al⁴⁵ recently performed a study in more than 700 intermediate-risk patients who were divided into two groups, one that underwent preoperative stress testing and one that did not. Beta-blockers were given to both groups, and doses were titrated to a goal heart rate of less than 65. The patients with optimally controlled heart rates had the lowest event rates.

However, the logistics of such a program would be challenging. For the most part, internists and hospitalists involved in perioperative assessment do not control the timing of referral or surgery, and adjustments cannot be made for patients whose preoperative clinic visit falls only a few days before surgery. Instituting a second or third visit to assess the efficacy of beta-blockade burdens the patient and may not be practical.

Are all beta-blockers equivalent?

An additional factor is the choice of agent. While the most significant studies of perioperative beta-blockade have used beta-1 receptor-selective agents (ie, metoprolol, atenolol, and bisoprolol), there is no prospective evidence that any particular agent is superior. However, a recent retrospective analysis of elderly surgical patients did suggest that longer-acting beta-blockers may be preferable: patients who had been on atenolol in the year before surgery had a 20% lower risk of postoperative myocardial infarction or death than those who had been on short-acting metoprolol, with no difference in noncardiac outcomes.⁴⁶

Ezetimibe (Zetia) was licensed by the US Food and Drug Administration in 2002 on the basis of its ability to reduce low-density lipoprotein cholesterol (LDL-C) levels. The reductions are mild, approximately 15%,¹ which is comparable to the effects of a stringent diet and exercise or of a statin in titrated doses.

See related commentary

However, there was no evidence that ezetimbe, which has a unique mechanism of action, delivers a benefit in terms of clinical outcomes. Despite this, the use of ezetimibe (alone or in fixed-dose combination with simvastatin, a preparation sold as Vytorin) grew rapidly, generating annual sales of $5.2 billion. Clinicians and the manufacturer (Merck/Schering-Plough) broadly assumed that LDL-C reduction would carry ezetimibe’s day as clinical trials emerged.

The assumption seemed reasonable, since evidence from the past 3 decades has established a clear link between lowering LDL-C levels via diverse mechanisms and positive clinical outcomes, particularly lower rates of cardiovascular disease and death. Indeed, LDL-C measurement is now a focus of cardiovascular risk assessment and management, as reflected in national treatment guidelines.

THE ENHANCE TRIAL: EZETIMIBE FAILS A KEY TEST

Unexpectedly, ezetimibe failed its first step in clinical trial validation, the Ezetimibe and Simvastatin in Hypercholesterolemia Enhances Atherosclerosis Regression (ENHANCE) trial.² Apart from the scientifically irrelevant political regulatory intrigue generated by the sponsor’s conduct in this trial, ENHANCE’s findings challenge us to confront issues of what we assume vs what we really know, and how to interpret the complex results of clinical trials.

To be fair to the trial’s investigators, ENHANCE achieved its objective of enrolling a population with a very high LDL-C level, which is ezetimibe’s target and has been widely used in the study of atherosclerosis progression as a marker of potential drug benefit. Nevertheless, and even though the LDL-C level 2 years later was 52 mg/dL lower in the group receiving ezetimibe/simvastatin than in the group receiving simvastatin alone (Zocor), at LDL-C levels that are typically associated with atherosclerosis progression (140–190 mg/dL), ezetimibe failed to reduce the progression of atherosclerosis.

Supplementary appendix to Kastelein JJ, et al. Simvastatin with or without ezetimibe in familial hypercholesterolemia. N Engl J Med 2008; 358:1431–1443. doi:10.1056/NEJMoa0800742. Copyright 2008, Massachusetts Medical Society.

These trends are particularly worrisome, given that the ezetimibe/simvastatin group achieved a greater reduction in C-reactive protein levels, which typically has resulted in superior outcomes in atherosclerosis³ and clinical effects⁴ in combination with LDL-C reduction.

In view of these findings, should clinicians stand firm and continue to use ezetimibe? Or should we reevaluate our position and await more data about this unique, first-in-class compound?

WISHFUL POST HOC HYPOTHESES

In this issue of the Cleveland Clinic Journal of Medicine, Dr. Michael Davidson,⁵ a respected lipid expert but one invested in ezetimibe’s development, assures us that all is in order and that the results of ENHANCE can be explained away by several arguments, most notably that most of the trial’s participants had previously received lipid-lowering treatment, which obscured the effects of ezetimibe. Moreover, he argues that ezetimibe’s mechanism of action is well understood and that the drug is safe and well tolerated and thus should remain a first-line treatment for hyperlipidemia.

These arguments may eventually prove to be correct, but as of now they are merely wishful post hoc hypotheses awaiting more data apart from ENHANCE. Negative clinical trials do occur as a matter of chance, but we should be cautious in any attempts to explain away a trial that was designed, executed, and reported as conceived simply because the results do not match our expectations.

Confronted with ENHANCE, the astute clinician should ask three questions: Do we really understand ezetimibe’s mechanism of action? Do other lines of evidence indicate the drug is beneficial? And how reliable is the arterial thickness as a surrogate end point?

DO WE UNDERSTAND EZETIMIBE’S MECHANISM OF ACTION?

Do we understand ezetimibe’s full mechanism of action? Not really.

True, ezetimibe inhibits cholesterol transport, a process that is integral both to cholesterol’s enteric absorption and to its systemic clearance. But although Dr. Davidson asserts that ezetimibe has cellular effects similar to those of statins, in fact it has the opposite effect on HMG-coA reductase, and no effects on LDL receptors.⁶

Furthermore, although initial studies suggested that ezetimibe inhibits enteric cholesterol absorption by inhibiting the Niemann-Pick C1L1 (NPC1L1) receptor, more recent investigations call this into serious question and point more definitively at a receptor known as scavenger receptor-B1 (SR-B1). As stated in a recent editorial, “SR-B1 in the apical site of enterocytes is the primary high-affinity site of cholesterol uptake and ezetimibe can inhibit this process. Moreover, the [possibility is ruled out] of NPC1L1 being a major player in this cholesterol uptake. This is at variance with the view of the colleagues from Schering-Plough who claim the same for NPC1L1.”⁷

SR-B1 is also a high-affinity receptor for high-density lipoprotein⁸ and thus is active in the antiatherosclerotic process of reverse cholesterol transport, inhibition of which significantly accelerates the development of atherosclerosis.⁹

Additionally, in vitro and thus unrelated to the effects of changing cholesterol concentration, ezetimibe down-regulates SR-B1 and another key cholesterol transporter protein called ABCA1.¹⁰ Further, ezetimibe induces down-regulation of raft protein domains, including CD36,¹¹ another effect opposite to that of statins.

These little-recognized effects of ezetimibe are among many that are completely unrelated to enteric cholesterol absorption. Yet, they are likely to be active within the liver and systemically where these proteins reside, and they are putatively proatherosclerotic. Contrary to often-cited opinion, ezetimibe is systemically absorbed, with 11% of the compound excreted in the urine.¹² Thus, the compound is systemically available to exert these same actions in the liver and elsewhere. Moreover, the absorbed drug is glucuronidated and is extensively recirculated in the liver in a form (its glucuronide) that is more potent than the parent compound.

In sum, present opinion is that ezetimibe inhibits lipid transport and interacts with a variety of receptors, not only in the gut but also systemically at the cell membrane and also inside the cell, focally disrupting several tightly regulated biologic processes.⁷ Thus, although ezetimibe reduces serum LDL-C levels via its effect in the gut, this effect may well be offset or even overridden systemically by other, unmeasurable effects, leading to counterintuitive results in terms of atherosclerosis or clinical events.

This would not be the first time a lipid-lowering drug has disappointed us: torcetrapib, another transport inhibitor, dramatically raises serum high-density lipoprotein cholesterol levels and reduces LDL-C but was found not only to have no effect on atherosclerosis, but also to potentiate adverse clinical outcomes.

The net impact of these other actions of ezetimibe is not known. We will discover its true clinical effects only through studies of endothelial function, atherosclerosis, and clinical cardiovascular outcomes. ENHANCE, which looked at atherosclerosis, is thus our strongest signal to date on the net effect of ezetimibe.

DO OTHER LINES OF EVIDENCE INDICATE EZETIMIBE IS BENEFICIAL?

Can we be reassured that ENHANCE’s results are spurious on the basis of other lines of evidence? Again, not really.

Experiments in animals, particularly in mice,¹³ have shown that ezetimibe may be antiatherosclerotic, although mice are considered the “worst model”⁷ for the study of ezetimibe, and notably, LDL-C levels were lowered far more in these experiments than they are clinically. Enthusiasm for these animal models should be tempered by interspecies variability in ezetimibe’s “off-target” effects and in the recent failure of other lipid transport drugs in human trials (torcetrapib and ACAT inhibitors) that had shown initial success in animals. No animal model is established for evaluating drugs of ezetimibe’s class, given its complex mechanism of action.

In human studies, the only other surrogate of the net effect of ezetimibe is endothelial function. Among several randomized clinical trials of ezetimibe,^14–18 only one was designed to compare the effects of ezetimibe alone, ezetimibe plus a statin, and a statin by itself in titrated or in maximum doses.¹⁵ After 4 weeks of therapy, all groups had lower LDL-C levels. However, ezetimibe monotherapy and ezetimibe/simvastatin combination therapy had no detectable effect on the arterial response to acetylcholine, but atorvastatin (Lipitor) monotherapy did. To be fair, the other (very small) trials showed mixed results, thus keeping the hypothesis of ezetimibe’s benefit alive, but with nothing close to a clear signal of benefit.

IS ARTERIAL THICKNESS RELIABLE AS A SURROGATE END POINT?

Was the principal problem in ENHANCE the use of carotid intima-media thickness as the primary end point? No.

This issue has received a lot of attention, much of which I believe is misinformed. No trial end point is infallible, including carotid intima-media thickness, and one must remain open to the possibility of chance findings. However, it has been a relatively reasonable end point in trials of diverse cardiovascular preventive strategies, including lipid-lowering, blood-pressure-lowering, and lifestyle interventions and as a directional biomarker of clinical atherosclerotic events.

We should be cautious about comparing data on carotid intima-media thickness from different trials, as Dr. Davidson attempts to do, in view of methodologic and population differences: each trial must be considered independently. Of greatest concern in ENHANCE is the consistency among intima-media thickness end points, including strong trends toward adverse effects in the most diseased carotid and femoral segments.

Moreover, ENHANCE’s detractors contend that the carotid intima-media thickness of the studied population was normal, citing this as evidence of delipidation from prior treatment. Although not impossible (as shown by the work of Zhao and colleagues in the setting of prolonged, intense lipid-lowering therapy¹⁹), at the moment this hypothesis is a matter of conjecture in the ENHANCE participants, particularly because their LDL-C levels were still quite elevated during the trial and conceivably even before randomization.

But these patients were not normal: they were typical patients with familial hypercholesterolemia with extremely elevated LDL-C levels and abnormally thick arteries for their age. Population screening estimates show that, for age and sex, the carotid intima-media thickness values in ENHANCE would lie in the upper quartile of those in the general population.²⁰ Moreover, their mean value is consistent with that in similar-aged groups of patients with familial hypercholesterolemia, even with lower rates of prior statin pretreatment.²¹

The most convincing evidence for the validity of the ENHANCE findings comes from the published subgroup data (Figure 1). In participants whose baseline carotid intima-media thickness was above the median at baseline, the thickness increased more with ezetimibe/simvastatin than with simvastatin alone. The same was true in the subgroup with above-average LDL-C levels at baseline. The subgroups with no prior statin treatment, low-dose prior statin treatment, and high-dose prior statin showed no heterogeneity of response: their carotid intima-media thickness increased more with ezetimibe/simvastatin than with simvastatin alone. None of these differences was statistically significant; however, these prespecified subgroup data seemingly invalidate arguments against the ENHANCE results based on carotid intima-media thickness findings.

In this context, ENHANCE can only be interpreted as a strong initial negative signal, a “red flag” about ezetimibe’s net health benefits.

WHAT NEXT?

The proper present focus of this debate is not on LDL-C but rather on ezetimibe, its unique mechanism of action, and on the need for more evidence about this complex compound.

At present, ezetimibe’s mechanism of action is not fully understood, and its benefit—for now, only mild LDL-C reduction—is too uncertain for us to be spending $5.2 billion a year for it. Its manufacturer is fortunate that the drug is even licensed, given the current and seemingly appropriate regulatory changes under which drugs introducing new therapeutic classes are scrutinized more closely for benefits and risks. “Safe and well tolerated,” as contended by Dr. Davidson, is not nearly enough: drugs must show clinically important benefits. We still know too little about this drug, the manufacturer of which has invested far more in marketing than in science, a point on which Dr. Davidson and I agree.

In 2008, ezetimibe is an appropriate candidate for testing in clinical trials, and in years to come it may be worthy of clinical attention—if rigorous and objectively conducted clinical trials prove its worth. At present, clinical equipoise dictates that ezetimibe is not an appropriate alternative to a statin in titrated doses, to the addition of other lipid-lowering drugs to a statin, to greater attention to drug adherence, or to lifestyle modification.

For the moment, given the ENHANCE results, the clinical usefulness of ezetimibe still remains to be proven. Much more evidence is needed before we can confidently reembrace the clinical use of ezetimibe.

Ezetimibe (Zetia) was licensed by the US Food and Drug Administration in 2002 on the basis of its ability to reduce low-density lipoprotein cholesterol (LDL-C) levels. The reductions are mild, approximately 15%,¹ which is comparable to the effects of a stringent diet and exercise or of a statin in titrated doses.

See related commentary

However, there was no evidence that ezetimbe, which has a unique mechanism of action, delivers a benefit in terms of clinical outcomes. Despite this, the use of ezetimibe (alone or in fixed-dose combination with simvastatin, a preparation sold as Vytorin) grew rapidly, generating annual sales of $5.2 billion. Clinicians and the manufacturer (Merck/Schering-Plough) broadly assumed that LDL-C reduction would carry ezetimibe’s day as clinical trials emerged.

The assumption seemed reasonable, since evidence from the past 3 decades has established a clear link between lowering LDL-C levels via diverse mechanisms and positive clinical outcomes, particularly lower rates of cardiovascular disease and death. Indeed, LDL-C measurement is now a focus of cardiovascular risk assessment and management, as reflected in national treatment guidelines.

THE ENHANCE TRIAL: EZETIMIBE FAILS A KEY TEST

Unexpectedly, ezetimibe failed its first step in clinical trial validation, the Ezetimibe and Simvastatin in Hypercholesterolemia Enhances Atherosclerosis Regression (ENHANCE) trial.² Apart from the scientifically irrelevant political regulatory intrigue generated by the sponsor’s conduct in this trial, ENHANCE’s findings challenge us to confront issues of what we assume vs what we really know, and how to interpret the complex results of clinical trials.

To be fair to the trial’s investigators, ENHANCE achieved its objective of enrolling a population with a very high LDL-C level, which is ezetimibe’s target and has been widely used in the study of atherosclerosis progression as a marker of potential drug benefit. Nevertheless, and even though the LDL-C level 2 years later was 52 mg/dL lower in the group receiving ezetimibe/simvastatin than in the group receiving simvastatin alone (Zocor), at LDL-C levels that are typically associated with atherosclerosis progression (140–190 mg/dL), ezetimibe failed to reduce the progression of atherosclerosis.

Supplementary appendix to Kastelein JJ, et al. Simvastatin with or without ezetimibe in familial hypercholesterolemia. N Engl J Med 2008; 358:1431–1443. doi:10.1056/NEJMoa0800742. Copyright 2008, Massachusetts Medical Society.

These trends are particularly worrisome, given that the ezetimibe/simvastatin group achieved a greater reduction in C-reactive protein levels, which typically has resulted in superior outcomes in atherosclerosis³ and clinical effects⁴ in combination with LDL-C reduction.

In view of these findings, should clinicians stand firm and continue to use ezetimibe? Or should we reevaluate our position and await more data about this unique, first-in-class compound?

WISHFUL POST HOC HYPOTHESES

In this issue of the Cleveland Clinic Journal of Medicine, Dr. Michael Davidson,⁵ a respected lipid expert but one invested in ezetimibe’s development, assures us that all is in order and that the results of ENHANCE can be explained away by several arguments, most notably that most of the trial’s participants had previously received lipid-lowering treatment, which obscured the effects of ezetimibe. Moreover, he argues that ezetimibe’s mechanism of action is well understood and that the drug is safe and well tolerated and thus should remain a first-line treatment for hyperlipidemia.

These arguments may eventually prove to be correct, but as of now they are merely wishful post hoc hypotheses awaiting more data apart from ENHANCE. Negative clinical trials do occur as a matter of chance, but we should be cautious in any attempts to explain away a trial that was designed, executed, and reported as conceived simply because the results do not match our expectations.

Confronted with ENHANCE, the astute clinician should ask three questions: Do we really understand ezetimibe’s mechanism of action? Do other lines of evidence indicate the drug is beneficial? And how reliable is the arterial thickness as a surrogate end point?

DO WE UNDERSTAND EZETIMIBE’S MECHANISM OF ACTION?

Do we understand ezetimibe’s full mechanism of action? Not really.

True, ezetimibe inhibits cholesterol transport, a process that is integral both to cholesterol’s enteric absorption and to its systemic clearance. But although Dr. Davidson asserts that ezetimibe has cellular effects similar to those of statins, in fact it has the opposite effect on HMG-coA reductase, and no effects on LDL receptors.⁶

Furthermore, although initial studies suggested that ezetimibe inhibits enteric cholesterol absorption by inhibiting the Niemann-Pick C1L1 (NPC1L1) receptor, more recent investigations call this into serious question and point more definitively at a receptor known as scavenger receptor-B1 (SR-B1). As stated in a recent editorial, “SR-B1 in the apical site of enterocytes is the primary high-affinity site of cholesterol uptake and ezetimibe can inhibit this process. Moreover, the [possibility is ruled out] of NPC1L1 being a major player in this cholesterol uptake. This is at variance with the view of the colleagues from Schering-Plough who claim the same for NPC1L1.”⁷

SR-B1 is also a high-affinity receptor for high-density lipoprotein⁸ and thus is active in the antiatherosclerotic process of reverse cholesterol transport, inhibition of which significantly accelerates the development of atherosclerosis.⁹

Additionally, in vitro and thus unrelated to the effects of changing cholesterol concentration, ezetimibe down-regulates SR-B1 and another key cholesterol transporter protein called ABCA1.¹⁰ Further, ezetimibe induces down-regulation of raft protein domains, including CD36,¹¹ another effect opposite to that of statins.

These little-recognized effects of ezetimibe are among many that are completely unrelated to enteric cholesterol absorption. Yet, they are likely to be active within the liver and systemically where these proteins reside, and they are putatively proatherosclerotic. Contrary to often-cited opinion, ezetimibe is systemically absorbed, with 11% of the compound excreted in the urine.¹² Thus, the compound is systemically available to exert these same actions in the liver and elsewhere. Moreover, the absorbed drug is glucuronidated and is extensively recirculated in the liver in a form (its glucuronide) that is more potent than the parent compound.

In sum, present opinion is that ezetimibe inhibits lipid transport and interacts with a variety of receptors, not only in the gut but also systemically at the cell membrane and also inside the cell, focally disrupting several tightly regulated biologic processes.⁷ Thus, although ezetimibe reduces serum LDL-C levels via its effect in the gut, this effect may well be offset or even overridden systemically by other, unmeasurable effects, leading to counterintuitive results in terms of atherosclerosis or clinical events.

This would not be the first time a lipid-lowering drug has disappointed us: torcetrapib, another transport inhibitor, dramatically raises serum high-density lipoprotein cholesterol levels and reduces LDL-C but was found not only to have no effect on atherosclerosis, but also to potentiate adverse clinical outcomes.

The net impact of these other actions of ezetimibe is not known. We will discover its true clinical effects only through studies of endothelial function, atherosclerosis, and clinical cardiovascular outcomes. ENHANCE, which looked at atherosclerosis, is thus our strongest signal to date on the net effect of ezetimibe.

DO OTHER LINES OF EVIDENCE INDICATE EZETIMIBE IS BENEFICIAL?

Can we be reassured that ENHANCE’s results are spurious on the basis of other lines of evidence? Again, not really.

Experiments in animals, particularly in mice,¹³ have shown that ezetimibe may be antiatherosclerotic, although mice are considered the “worst model”⁷ for the study of ezetimibe, and notably, LDL-C levels were lowered far more in these experiments than they are clinically. Enthusiasm for these animal models should be tempered by interspecies variability in ezetimibe’s “off-target” effects and in the recent failure of other lipid transport drugs in human trials (torcetrapib and ACAT inhibitors) that had shown initial success in animals. No animal model is established for evaluating drugs of ezetimibe’s class, given its complex mechanism of action.

In human studies, the only other surrogate of the net effect of ezetimibe is endothelial function. Among several randomized clinical trials of ezetimibe,^14–18 only one was designed to compare the effects of ezetimibe alone, ezetimibe plus a statin, and a statin by itself in titrated or in maximum doses.¹⁵ After 4 weeks of therapy, all groups had lower LDL-C levels. However, ezetimibe monotherapy and ezetimibe/simvastatin combination therapy had no detectable effect on the arterial response to acetylcholine, but atorvastatin (Lipitor) monotherapy did. To be fair, the other (very small) trials showed mixed results, thus keeping the hypothesis of ezetimibe’s benefit alive, but with nothing close to a clear signal of benefit.

IS ARTERIAL THICKNESS RELIABLE AS A SURROGATE END POINT?

Was the principal problem in ENHANCE the use of carotid intima-media thickness as the primary end point? No.

This issue has received a lot of attention, much of which I believe is misinformed. No trial end point is infallible, including carotid intima-media thickness, and one must remain open to the possibility of chance findings. However, it has been a relatively reasonable end point in trials of diverse cardiovascular preventive strategies, including lipid-lowering, blood-pressure-lowering, and lifestyle interventions and as a directional biomarker of clinical atherosclerotic events.

We should be cautious about comparing data on carotid intima-media thickness from different trials, as Dr. Davidson attempts to do, in view of methodologic and population differences: each trial must be considered independently. Of greatest concern in ENHANCE is the consistency among intima-media thickness end points, including strong trends toward adverse effects in the most diseased carotid and femoral segments.

Moreover, ENHANCE’s detractors contend that the carotid intima-media thickness of the studied population was normal, citing this as evidence of delipidation from prior treatment. Although not impossible (as shown by the work of Zhao and colleagues in the setting of prolonged, intense lipid-lowering therapy¹⁹), at the moment this hypothesis is a matter of conjecture in the ENHANCE participants, particularly because their LDL-C levels were still quite elevated during the trial and conceivably even before randomization.

But these patients were not normal: they were typical patients with familial hypercholesterolemia with extremely elevated LDL-C levels and abnormally thick arteries for their age. Population screening estimates show that, for age and sex, the carotid intima-media thickness values in ENHANCE would lie in the upper quartile of those in the general population.²⁰ Moreover, their mean value is consistent with that in similar-aged groups of patients with familial hypercholesterolemia, even with lower rates of prior statin pretreatment.²¹

The most convincing evidence for the validity of the ENHANCE findings comes from the published subgroup data (Figure 1). In participants whose baseline carotid intima-media thickness was above the median at baseline, the thickness increased more with ezetimibe/simvastatin than with simvastatin alone. The same was true in the subgroup with above-average LDL-C levels at baseline. The subgroups with no prior statin treatment, low-dose prior statin treatment, and high-dose prior statin showed no heterogeneity of response: their carotid intima-media thickness increased more with ezetimibe/simvastatin than with simvastatin alone. None of these differences was statistically significant; however, these prespecified subgroup data seemingly invalidate arguments against the ENHANCE results based on carotid intima-media thickness findings.

In this context, ENHANCE can only be interpreted as a strong initial negative signal, a “red flag” about ezetimibe’s net health benefits.

WHAT NEXT?

The proper present focus of this debate is not on LDL-C but rather on ezetimibe, its unique mechanism of action, and on the need for more evidence about this complex compound.

At present, ezetimibe’s mechanism of action is not fully understood, and its benefit—for now, only mild LDL-C reduction—is too uncertain for us to be spending $5.2 billion a year for it. Its manufacturer is fortunate that the drug is even licensed, given the current and seemingly appropriate regulatory changes under which drugs introducing new therapeutic classes are scrutinized more closely for benefits and risks. “Safe and well tolerated,” as contended by Dr. Davidson, is not nearly enough: drugs must show clinically important benefits. We still know too little about this drug, the manufacturer of which has invested far more in marketing than in science, a point on which Dr. Davidson and I agree.

In 2008, ezetimibe is an appropriate candidate for testing in clinical trials, and in years to come it may be worthy of clinical attention—if rigorous and objectively conducted clinical trials prove its worth. At present, clinical equipoise dictates that ezetimibe is not an appropriate alternative to a statin in titrated doses, to the addition of other lipid-lowering drugs to a statin, to greater attention to drug adherence, or to lifestyle modification.

For the moment, given the ENHANCE results, the clinical usefulness of ezetimibe still remains to be proven. Much more evidence is needed before we can confidently reembrace the clinical use of ezetimibe.

Ezetimibe (Zetia) was licensed by the US Food and Drug Administration in 2002 on the basis of its ability to reduce low-density lipoprotein cholesterol (LDL-C) levels. The reductions are mild, approximately 15%,¹ which is comparable to the effects of a stringent diet and exercise or of a statin in titrated doses.

See related commentary

However, there was no evidence that ezetimbe, which has a unique mechanism of action, delivers a benefit in terms of clinical outcomes. Despite this, the use of ezetimibe (alone or in fixed-dose combination with simvastatin, a preparation sold as Vytorin) grew rapidly, generating annual sales of $5.2 billion. Clinicians and the manufacturer (Merck/Schering-Plough) broadly assumed that LDL-C reduction would carry ezetimibe’s day as clinical trials emerged.

The assumption seemed reasonable, since evidence from the past 3 decades has established a clear link between lowering LDL-C levels via diverse mechanisms and positive clinical outcomes, particularly lower rates of cardiovascular disease and death. Indeed, LDL-C measurement is now a focus of cardiovascular risk assessment and management, as reflected in national treatment guidelines.

THE ENHANCE TRIAL: EZETIMIBE FAILS A KEY TEST

Unexpectedly, ezetimibe failed its first step in clinical trial validation, the Ezetimibe and Simvastatin in Hypercholesterolemia Enhances Atherosclerosis Regression (ENHANCE) trial.² Apart from the scientifically irrelevant political regulatory intrigue generated by the sponsor’s conduct in this trial, ENHANCE’s findings challenge us to confront issues of what we assume vs what we really know, and how to interpret the complex results of clinical trials.

To be fair to the trial’s investigators, ENHANCE achieved its objective of enrolling a population with a very high LDL-C level, which is ezetimibe’s target and has been widely used in the study of atherosclerosis progression as a marker of potential drug benefit. Nevertheless, and even though the LDL-C level 2 years later was 52 mg/dL lower in the group receiving ezetimibe/simvastatin than in the group receiving simvastatin alone (Zocor), at LDL-C levels that are typically associated with atherosclerosis progression (140–190 mg/dL), ezetimibe failed to reduce the progression of atherosclerosis.

Supplementary appendix to Kastelein JJ, et al. Simvastatin with or without ezetimibe in familial hypercholesterolemia. N Engl J Med 2008; 358:1431–1443. doi:10.1056/NEJMoa0800742. Copyright 2008, Massachusetts Medical Society.

These trends are particularly worrisome, given that the ezetimibe/simvastatin group achieved a greater reduction in C-reactive protein levels, which typically has resulted in superior outcomes in atherosclerosis³ and clinical effects⁴ in combination with LDL-C reduction.

In view of these findings, should clinicians stand firm and continue to use ezetimibe? Or should we reevaluate our position and await more data about this unique, first-in-class compound?

WISHFUL POST HOC HYPOTHESES

In this issue of the Cleveland Clinic Journal of Medicine, Dr. Michael Davidson,⁵ a respected lipid expert but one invested in ezetimibe’s development, assures us that all is in order and that the results of ENHANCE can be explained away by several arguments, most notably that most of the trial’s participants had previously received lipid-lowering treatment, which obscured the effects of ezetimibe. Moreover, he argues that ezetimibe’s mechanism of action is well understood and that the drug is safe and well tolerated and thus should remain a first-line treatment for hyperlipidemia.

These arguments may eventually prove to be correct, but as of now they are merely wishful post hoc hypotheses awaiting more data apart from ENHANCE. Negative clinical trials do occur as a matter of chance, but we should be cautious in any attempts to explain away a trial that was designed, executed, and reported as conceived simply because the results do not match our expectations.

Confronted with ENHANCE, the astute clinician should ask three questions: Do we really understand ezetimibe’s mechanism of action? Do other lines of evidence indicate the drug is beneficial? And how reliable is the arterial thickness as a surrogate end point?

DO WE UNDERSTAND EZETIMIBE’S MECHANISM OF ACTION?

Do we understand ezetimibe’s full mechanism of action? Not really.

True, ezetimibe inhibits cholesterol transport, a process that is integral both to cholesterol’s enteric absorption and to its systemic clearance. But although Dr. Davidson asserts that ezetimibe has cellular effects similar to those of statins, in fact it has the opposite effect on HMG-coA reductase, and no effects on LDL receptors.⁶

Furthermore, although initial studies suggested that ezetimibe inhibits enteric cholesterol absorption by inhibiting the Niemann-Pick C1L1 (NPC1L1) receptor, more recent investigations call this into serious question and point more definitively at a receptor known as scavenger receptor-B1 (SR-B1). As stated in a recent editorial, “SR-B1 in the apical site of enterocytes is the primary high-affinity site of cholesterol uptake and ezetimibe can inhibit this process. Moreover, the [possibility is ruled out] of NPC1L1 being a major player in this cholesterol uptake. This is at variance with the view of the colleagues from Schering-Plough who claim the same for NPC1L1.”⁷

SR-B1 is also a high-affinity receptor for high-density lipoprotein⁸ and thus is active in the antiatherosclerotic process of reverse cholesterol transport, inhibition of which significantly accelerates the development of atherosclerosis.⁹

Additionally, in vitro and thus unrelated to the effects of changing cholesterol concentration, ezetimibe down-regulates SR-B1 and another key cholesterol transporter protein called ABCA1.¹⁰ Further, ezetimibe induces down-regulation of raft protein domains, including CD36,¹¹ another effect opposite to that of statins.

These little-recognized effects of ezetimibe are among many that are completely unrelated to enteric cholesterol absorption. Yet, they are likely to be active within the liver and systemically where these proteins reside, and they are putatively proatherosclerotic. Contrary to often-cited opinion, ezetimibe is systemically absorbed, with 11% of the compound excreted in the urine.¹² Thus, the compound is systemically available to exert these same actions in the liver and elsewhere. Moreover, the absorbed drug is glucuronidated and is extensively recirculated in the liver in a form (its glucuronide) that is more potent than the parent compound.

In sum, present opinion is that ezetimibe inhibits lipid transport and interacts with a variety of receptors, not only in the gut but also systemically at the cell membrane and also inside the cell, focally disrupting several tightly regulated biologic processes.⁷ Thus, although ezetimibe reduces serum LDL-C levels via its effect in the gut, this effect may well be offset or even overridden systemically by other, unmeasurable effects, leading to counterintuitive results in terms of atherosclerosis or clinical events.

This would not be the first time a lipid-lowering drug has disappointed us: torcetrapib, another transport inhibitor, dramatically raises serum high-density lipoprotein cholesterol levels and reduces LDL-C but was found not only to have no effect on atherosclerosis, but also to potentiate adverse clinical outcomes.

The net impact of these other actions of ezetimibe is not known. We will discover its true clinical effects only through studies of endothelial function, atherosclerosis, and clinical cardiovascular outcomes. ENHANCE, which looked at atherosclerosis, is thus our strongest signal to date on the net effect of ezetimibe.

DO OTHER LINES OF EVIDENCE INDICATE EZETIMIBE IS BENEFICIAL?

Can we be reassured that ENHANCE’s results are spurious on the basis of other lines of evidence? Again, not really.

Experiments in animals, particularly in mice,¹³ have shown that ezetimibe may be antiatherosclerotic, although mice are considered the “worst model”⁷ for the study of ezetimibe, and notably, LDL-C levels were lowered far more in these experiments than they are clinically. Enthusiasm for these animal models should be tempered by interspecies variability in ezetimibe’s “off-target” effects and in the recent failure of other lipid transport drugs in human trials (torcetrapib and ACAT inhibitors) that had shown initial success in animals. No animal model is established for evaluating drugs of ezetimibe’s class, given its complex mechanism of action.

In human studies, the only other surrogate of the net effect of ezetimibe is endothelial function. Among several randomized clinical trials of ezetimibe,^14–18 only one was designed to compare the effects of ezetimibe alone, ezetimibe plus a statin, and a statin by itself in titrated or in maximum doses.¹⁵ After 4 weeks of therapy, all groups had lower LDL-C levels. However, ezetimibe monotherapy and ezetimibe/simvastatin combination therapy had no detectable effect on the arterial response to acetylcholine, but atorvastatin (Lipitor) monotherapy did. To be fair, the other (very small) trials showed mixed results, thus keeping the hypothesis of ezetimibe’s benefit alive, but with nothing close to a clear signal of benefit.

IS ARTERIAL THICKNESS RELIABLE AS A SURROGATE END POINT?

Was the principal problem in ENHANCE the use of carotid intima-media thickness as the primary end point? No.

This issue has received a lot of attention, much of which I believe is misinformed. No trial end point is infallible, including carotid intima-media thickness, and one must remain open to the possibility of chance findings. However, it has been a relatively reasonable end point in trials of diverse cardiovascular preventive strategies, including lipid-lowering, blood-pressure-lowering, and lifestyle interventions and as a directional biomarker of clinical atherosclerotic events.

We should be cautious about comparing data on carotid intima-media thickness from different trials, as Dr. Davidson attempts to do, in view of methodologic and population differences: each trial must be considered independently. Of greatest concern in ENHANCE is the consistency among intima-media thickness end points, including strong trends toward adverse effects in the most diseased carotid and femoral segments.

Moreover, ENHANCE’s detractors contend that the carotid intima-media thickness of the studied population was normal, citing this as evidence of delipidation from prior treatment. Although not impossible (as shown by the work of Zhao and colleagues in the setting of prolonged, intense lipid-lowering therapy¹⁹), at the moment this hypothesis is a matter of conjecture in the ENHANCE participants, particularly because their LDL-C levels were still quite elevated during the trial and conceivably even before randomization.

But these patients were not normal: they were typical patients with familial hypercholesterolemia with extremely elevated LDL-C levels and abnormally thick arteries for their age. Population screening estimates show that, for age and sex, the carotid intima-media thickness values in ENHANCE would lie in the upper quartile of those in the general population.²⁰ Moreover, their mean value is consistent with that in similar-aged groups of patients with familial hypercholesterolemia, even with lower rates of prior statin pretreatment.²¹

The most convincing evidence for the validity of the ENHANCE findings comes from the published subgroup data (Figure 1). In participants whose baseline carotid intima-media thickness was above the median at baseline, the thickness increased more with ezetimibe/simvastatin than with simvastatin alone. The same was true in the subgroup with above-average LDL-C levels at baseline. The subgroups with no prior statin treatment, low-dose prior statin treatment, and high-dose prior statin showed no heterogeneity of response: their carotid intima-media thickness increased more with ezetimibe/simvastatin than with simvastatin alone. None of these differences was statistically significant; however, these prespecified subgroup data seemingly invalidate arguments against the ENHANCE results based on carotid intima-media thickness findings.

In this context, ENHANCE can only be interpreted as a strong initial negative signal, a “red flag” about ezetimibe’s net health benefits.

WHAT NEXT?

The proper present focus of this debate is not on LDL-C but rather on ezetimibe, its unique mechanism of action, and on the need for more evidence about this complex compound.

At present, ezetimibe’s mechanism of action is not fully understood, and its benefit—for now, only mild LDL-C reduction—is too uncertain for us to be spending $5.2 billion a year for it. Its manufacturer is fortunate that the drug is even licensed, given the current and seemingly appropriate regulatory changes under which drugs introducing new therapeutic classes are scrutinized more closely for benefits and risks. “Safe and well tolerated,” as contended by Dr. Davidson, is not nearly enough: drugs must show clinically important benefits. We still know too little about this drug, the manufacturer of which has invested far more in marketing than in science, a point on which Dr. Davidson and I agree.

In 2008, ezetimibe is an appropriate candidate for testing in clinical trials, and in years to come it may be worthy of clinical attention—if rigorous and objectively conducted clinical trials prove its worth. At present, clinical equipoise dictates that ezetimibe is not an appropriate alternative to a statin in titrated doses, to the addition of other lipid-lowering drugs to a statin, to greater attention to drug adherence, or to lifestyle modification.

For the moment, given the ENHANCE results, the clinical usefulness of ezetimibe still remains to be proven. Much more evidence is needed before we can confidently reembrace the clinical use of ezetimibe.

The Ezetimibe and Simvastatin in Hypercholesterolemia Enhances Atherosclerosis Regression (ENHANCE) trial¹ was probably the most widely publicized clinical study of the past decade. How did a 720-patient imaging trial with a neutral result in patients with severe hypercholesterolemia rise to a level warranting massive media attention, a congressional investigation, and a recommendation to curtail the use of a drug widely used to reduce levels of low-density-lipoprotein cholesterol (LDL-C)?

See related editorial

The reaction to the ENHANCE trial reveals more about the political climate and the relationship between the pharmaceutical industry and the American public than it does about the effects of ezetimibe (available combined with simvastatin as Vytorin and by itself as Zetia) on the progression of atherosclerosis.

SOME SELF-DISCLOSURE

Before I discuss the clinical implications of the ENHANCE trial, I must describe both my financial conflicts and intellectual biases. I am a paid consultant, speaker, and researcher on behalf of Merck/Schering-Plough, the sponsor of the ENHANCE trial. I was a principal investigator in the first phase II trial of ezetimibe and have conducted more than 10 clinical trials of either ezetimibe or ezetimibe/simvastatin. I also have been a strong advocate for imaging trials to assist in the clinical development of novel therapeutic agents and to support regulatory approval.

Therefore, I believe that the thickness of the intima and media layers of the carotid arteries is a useful surrogate to evaluate the potential antiatherosclerotic effects of drugs (more on this topic below). Also, I believe that the LDL-C-lowering hypothesis has been proven: ie, that all drugs that lower LDL-C safely, without off-target adverse effects, should reduce cardiovascular events. I support the goal levels of LDL-C and non-high-density-lipoprotein cholesterol set by the National Cholesterol Education Program’s third Adult Treatment Panel (ATP III) guidelines,^2,3 which specify LDL-C targets rather than the use of specific drugs. In spite of these conflicts and potential biases, I believe I have always served the best interests of patient care.

HISTORY OF THE ENHANCE TRIAL