Letters To The Editor

To the Editor: We previously addressed whether VKORC1 and CYP2C9 pharmacogenomic testing should be considered when prescribing warfarin.¹ Our recommendation, based on available evidence at that time, was that physicians should consider pharmacogenomic testing for any patient who is started on warfarin therapy.

Since the publication of this recommendation, two major trials, COAG (Clarification of Optimal Anticoagulation Through Genetics)² and EU-PACT (European Pharmacogenetics of Anticoagulant Therapy-Warfarin),³ were published along with commentaries debating the clinical utility of warfarin pharmacogenomics.^4–15 Based on these publications, we would like to update our recommendations for pharmacogenomic testing for warfarin therapy.

COAG compared the efficacy of a clinical algorithm or a clinical algorithm plus VKORC1 and CYP2C9 genotyping to guide warfarin dosage. At the end of 4 weeks, the mean percentage of time within the therapeutic international normalized ratio (INR) range was 45.4% for those in the clinical algorithm arm and 45.2% for those in the genotyping arm (95% confidence interval [CI] –3.4 to 3.1, P = .91). For both treatment groups, clinical data that included body surface area, age, target INR, concomitantly prescribed drugs, and smoking status were used to predict warfarin dose, with the genotyping arm including VKORC1 and CYP2C9. Although VKORC1 and CYP2C9 genotyping offered no additional benefit, caution should be used when extrapolating this conclusion to clinical settings in which warfarin therapy is initiated using a standardized starting dose (eg, 5 mg daily) instead of a clinical dosing algorithm.

Of interest, in the COAG trial, among black patients, the mean percentage of time in the therapeutic INR range was significantly less for those in the genotype-guided arm than for those in the clinically guided arm—ie, 35.2% vs 43.5% (95% CI –15.0 to –2.0, P = .01). The percentage of time with therapeutic INR has been identified as a surrogate marker for poor outcomes such as death, stroke, or major hemorrhage, with those with a lower percentage of time in therapeutic INR being at greater risk of an adverse event.¹⁶ Wan et al¹⁷ demonstrated that a 6.9% improvement of time in therapeutic INR decreased the risk of major hemorrhage by one event per 100 patient-years.¹⁷ Therefore, black patients in the COAG genotyping arm may have been at greater risk for an adverse event because of a lower observed percentage of time within the therapeutic INR range.

In the COAG trial, genotyping was done for only one VKORC1 variant and for two CYP2C9 alleles (CYP2C9*2, and CYP2C9*3). Other genetic variants are of clinical importance for warfarin dosing in black patients, and the lack of genotyping for these additional variants may explain why black patients in the genotyping arm performed worse.^5,7,11 In particular, CYP2C9*8 may be an important predictor of warfarin dose in black patients.¹⁸

EU-PACT compared the efficacy of standardized warfarin dosing and that of a clinical algorithm.³ Patients in the standardized dosing arm were prescribed warfarin 10 mg on the first day of treatment (5 mg for those over age 75), and 5 mg on days 2 and 3, with subsequent dosing adjustments based on INR. Patients in the genotyping arm were prescribed warfarin based on an algorithm that incorporated clinical data that included body surface area, age, and concomitantly prescribed drugs, as well as VKORC1 and CYP2C9 genotypes. At the end of 12 weeks, the mean percentage of time in the therapeutic INR range was 60.3% for those in the standardized-dosing arm and 67.4% for those in the genotyping arm (95% CI 3.3 to 10.6, P < .001).² The approximate 7% improvement in percentage of time in the therapeutic INR range may predict a lower risk of hemorrhage for those in the genotyping arm.¹⁷ Although patients in the genotyping arm had a higher percentage of time in the therapeutic INR range, it is unclear whether genotyping alone is superior to standardized dosing because the dosing algorithm used both clinical data and genotype data.

There are substantial differences between the COAG and EU-PACT trials, including dosing schemes, racial diversity, and trial length, and these differences could have contributed to the conflicting results. Based on these two trials, a possible conclusion is that genotype-guided warfarin dosing may be superior to standardized dosing, but may be no better than utilizing a clinical algorithm in white patients. For black patients, additional studies are needed to determine which genetic variants are of importance for guiding warfarin dosing.

We would like to update the recommendations we made in our previously published article,¹ to state that genotyping for CYP2C9 and VKORC1 may be of clinical utility in white patients depending on whether standardized dosing or a clinical algorithm is used to initiate warfarin therapy. Routine genotyping in black patients is not recommended until further studies clarify which genetic variants are of importance for guiding warfarin dosing.

The ongoing Genetics Informatics Trial of Warfarin to Prevent Venous Thrombosis may bring much needed clarity to the clinical utility of warfarin pharmacogenomics. We hope to publish a more detailed update of our 2013 article after completion of that trial.

To the Editor: We previously addressed whether VKORC1 and CYP2C9 pharmacogenomic testing should be considered when prescribing warfarin.¹ Our recommendation, based on available evidence at that time, was that physicians should consider pharmacogenomic testing for any patient who is started on warfarin therapy.

Since the publication of this recommendation, two major trials, COAG (Clarification of Optimal Anticoagulation Through Genetics)² and EU-PACT (European Pharmacogenetics of Anticoagulant Therapy-Warfarin),³ were published along with commentaries debating the clinical utility of warfarin pharmacogenomics.^4–15 Based on these publications, we would like to update our recommendations for pharmacogenomic testing for warfarin therapy.

COAG compared the efficacy of a clinical algorithm or a clinical algorithm plus VKORC1 and CYP2C9 genotyping to guide warfarin dosage. At the end of 4 weeks, the mean percentage of time within the therapeutic international normalized ratio (INR) range was 45.4% for those in the clinical algorithm arm and 45.2% for those in the genotyping arm (95% confidence interval [CI] –3.4 to 3.1, P = .91). For both treatment groups, clinical data that included body surface area, age, target INR, concomitantly prescribed drugs, and smoking status were used to predict warfarin dose, with the genotyping arm including VKORC1 and CYP2C9. Although VKORC1 and CYP2C9 genotyping offered no additional benefit, caution should be used when extrapolating this conclusion to clinical settings in which warfarin therapy is initiated using a standardized starting dose (eg, 5 mg daily) instead of a clinical dosing algorithm.

Of interest, in the COAG trial, among black patients, the mean percentage of time in the therapeutic INR range was significantly less for those in the genotype-guided arm than for those in the clinically guided arm—ie, 35.2% vs 43.5% (95% CI –15.0 to –2.0, P = .01). The percentage of time with therapeutic INR has been identified as a surrogate marker for poor outcomes such as death, stroke, or major hemorrhage, with those with a lower percentage of time in therapeutic INR being at greater risk of an adverse event.¹⁶ Wan et al¹⁷ demonstrated that a 6.9% improvement of time in therapeutic INR decreased the risk of major hemorrhage by one event per 100 patient-years.¹⁷ Therefore, black patients in the COAG genotyping arm may have been at greater risk for an adverse event because of a lower observed percentage of time within the therapeutic INR range.

In the COAG trial, genotyping was done for only one VKORC1 variant and for two CYP2C9 alleles (CYP2C9*2, and CYP2C9*3). Other genetic variants are of clinical importance for warfarin dosing in black patients, and the lack of genotyping for these additional variants may explain why black patients in the genotyping arm performed worse.^5,7,11 In particular, CYP2C9*8 may be an important predictor of warfarin dose in black patients.¹⁸

EU-PACT compared the efficacy of standardized warfarin dosing and that of a clinical algorithm.³ Patients in the standardized dosing arm were prescribed warfarin 10 mg on the first day of treatment (5 mg for those over age 75), and 5 mg on days 2 and 3, with subsequent dosing adjustments based on INR. Patients in the genotyping arm were prescribed warfarin based on an algorithm that incorporated clinical data that included body surface area, age, and concomitantly prescribed drugs, as well as VKORC1 and CYP2C9 genotypes. At the end of 12 weeks, the mean percentage of time in the therapeutic INR range was 60.3% for those in the standardized-dosing arm and 67.4% for those in the genotyping arm (95% CI 3.3 to 10.6, P < .001).² The approximate 7% improvement in percentage of time in the therapeutic INR range may predict a lower risk of hemorrhage for those in the genotyping arm.¹⁷ Although patients in the genotyping arm had a higher percentage of time in the therapeutic INR range, it is unclear whether genotyping alone is superior to standardized dosing because the dosing algorithm used both clinical data and genotype data.

There are substantial differences between the COAG and EU-PACT trials, including dosing schemes, racial diversity, and trial length, and these differences could have contributed to the conflicting results. Based on these two trials, a possible conclusion is that genotype-guided warfarin dosing may be superior to standardized dosing, but may be no better than utilizing a clinical algorithm in white patients. For black patients, additional studies are needed to determine which genetic variants are of importance for guiding warfarin dosing.

We would like to update the recommendations we made in our previously published article,¹ to state that genotyping for CYP2C9 and VKORC1 may be of clinical utility in white patients depending on whether standardized dosing or a clinical algorithm is used to initiate warfarin therapy. Routine genotyping in black patients is not recommended until further studies clarify which genetic variants are of importance for guiding warfarin dosing.

The ongoing Genetics Informatics Trial of Warfarin to Prevent Venous Thrombosis may bring much needed clarity to the clinical utility of warfarin pharmacogenomics. We hope to publish a more detailed update of our 2013 article after completion of that trial.

To the Editor: We previously addressed whether VKORC1 and CYP2C9 pharmacogenomic testing should be considered when prescribing warfarin.¹ Our recommendation, based on available evidence at that time, was that physicians should consider pharmacogenomic testing for any patient who is started on warfarin therapy.

Since the publication of this recommendation, two major trials, COAG (Clarification of Optimal Anticoagulation Through Genetics)² and EU-PACT (European Pharmacogenetics of Anticoagulant Therapy-Warfarin),³ were published along with commentaries debating the clinical utility of warfarin pharmacogenomics.^4–15 Based on these publications, we would like to update our recommendations for pharmacogenomic testing for warfarin therapy.

COAG compared the efficacy of a clinical algorithm or a clinical algorithm plus VKORC1 and CYP2C9 genotyping to guide warfarin dosage. At the end of 4 weeks, the mean percentage of time within the therapeutic international normalized ratio (INR) range was 45.4% for those in the clinical algorithm arm and 45.2% for those in the genotyping arm (95% confidence interval [CI] –3.4 to 3.1, P = .91). For both treatment groups, clinical data that included body surface area, age, target INR, concomitantly prescribed drugs, and smoking status were used to predict warfarin dose, with the genotyping arm including VKORC1 and CYP2C9. Although VKORC1 and CYP2C9 genotyping offered no additional benefit, caution should be used when extrapolating this conclusion to clinical settings in which warfarin therapy is initiated using a standardized starting dose (eg, 5 mg daily) instead of a clinical dosing algorithm.

Of interest, in the COAG trial, among black patients, the mean percentage of time in the therapeutic INR range was significantly less for those in the genotype-guided arm than for those in the clinically guided arm—ie, 35.2% vs 43.5% (95% CI –15.0 to –2.0, P = .01). The percentage of time with therapeutic INR has been identified as a surrogate marker for poor outcomes such as death, stroke, or major hemorrhage, with those with a lower percentage of time in therapeutic INR being at greater risk of an adverse event.¹⁶ Wan et al¹⁷ demonstrated that a 6.9% improvement of time in therapeutic INR decreased the risk of major hemorrhage by one event per 100 patient-years.¹⁷ Therefore, black patients in the COAG genotyping arm may have been at greater risk for an adverse event because of a lower observed percentage of time within the therapeutic INR range.

In the COAG trial, genotyping was done for only one VKORC1 variant and for two CYP2C9 alleles (CYP2C9*2, and CYP2C9*3). Other genetic variants are of clinical importance for warfarin dosing in black patients, and the lack of genotyping for these additional variants may explain why black patients in the genotyping arm performed worse.^5,7,11 In particular, CYP2C9*8 may be an important predictor of warfarin dose in black patients.¹⁸

EU-PACT compared the efficacy of standardized warfarin dosing and that of a clinical algorithm.³ Patients in the standardized dosing arm were prescribed warfarin 10 mg on the first day of treatment (5 mg for those over age 75), and 5 mg on days 2 and 3, with subsequent dosing adjustments based on INR. Patients in the genotyping arm were prescribed warfarin based on an algorithm that incorporated clinical data that included body surface area, age, and concomitantly prescribed drugs, as well as VKORC1 and CYP2C9 genotypes. At the end of 12 weeks, the mean percentage of time in the therapeutic INR range was 60.3% for those in the standardized-dosing arm and 67.4% for those in the genotyping arm (95% CI 3.3 to 10.6, P < .001).² The approximate 7% improvement in percentage of time in the therapeutic INR range may predict a lower risk of hemorrhage for those in the genotyping arm.¹⁷ Although patients in the genotyping arm had a higher percentage of time in the therapeutic INR range, it is unclear whether genotyping alone is superior to standardized dosing because the dosing algorithm used both clinical data and genotype data.

There are substantial differences between the COAG and EU-PACT trials, including dosing schemes, racial diversity, and trial length, and these differences could have contributed to the conflicting results. Based on these two trials, a possible conclusion is that genotype-guided warfarin dosing may be superior to standardized dosing, but may be no better than utilizing a clinical algorithm in white patients. For black patients, additional studies are needed to determine which genetic variants are of importance for guiding warfarin dosing.

We would like to update the recommendations we made in our previously published article,¹ to state that genotyping for CYP2C9 and VKORC1 may be of clinical utility in white patients depending on whether standardized dosing or a clinical algorithm is used to initiate warfarin therapy. Routine genotyping in black patients is not recommended until further studies clarify which genetic variants are of importance for guiding warfarin dosing.

The ongoing Genetics Informatics Trial of Warfarin to Prevent Venous Thrombosis may bring much needed clarity to the clinical utility of warfarin pharmacogenomics. We hope to publish a more detailed update of our 2013 article after completion of that trial.

To the Editor: I read with avid interest the IM Board Review by Pagán et al on a man with pheochromocytoma.¹ The article stated that the classic triad is headache, hypertension, and hyperglycemia. I found this to be incorrect. And Harrison’s Principles of Internal Medicine² states that the classic triad is palpitations, headaches, and profuse sweating. I hope you will clarify this in the interest of the readers as it is a board review article.

To the Editor: I read with avid interest the IM Board Review by Pagán et al on a man with pheochromocytoma.¹ The article stated that the classic triad is headache, hypertension, and hyperglycemia. I found this to be incorrect. And Harrison’s Principles of Internal Medicine² states that the classic triad is palpitations, headaches, and profuse sweating. I hope you will clarify this in the interest of the readers as it is a board review article.

To the Editor: I read with avid interest the IM Board Review by Pagán et al on a man with pheochromocytoma.¹ The article stated that the classic triad is headache, hypertension, and hyperglycemia. I found this to be incorrect. And Harrison’s Principles of Internal Medicine² states that the classic triad is palpitations, headaches, and profuse sweating. I hope you will clarify this in the interest of the readers as it is a board review article.

In Reply: Thanks to Dr. Belur for his observation. He is correct in that the classic triad includes headaches, palpitations, and diaphoresis, although hypertension and hyperglycemia have been described in the literature as frequently occurring, and the clinical presentation can be extremely variable.

In Reply: Thanks to Dr. Belur for his observation. He is correct in that the classic triad includes headaches, palpitations, and diaphoresis, although hypertension and hyperglycemia have been described in the literature as frequently occurring, and the clinical presentation can be extremely variable.

In Reply: Thanks to Dr. Belur for his observation. He is correct in that the classic triad includes headaches, palpitations, and diaphoresis, although hypertension and hyperglycemia have been described in the literature as frequently occurring, and the clinical presentation can be extremely variable.

To the Editor: The article by Dr. Catherine Curley,¹ “Rule out pulmonary tuberculosis: clinical and radiographic clues for the internist,” was very well written, but we would like to point out two facts regarding the diagnosis of pulmonary tuberculosis, especially in high-prevalence countries like India, that might make the article more informative.

First, it has been shown conclusively that good-quality microscopy of two consecutive sputum specimens identifies the majority (95%–98%) of smear-positive tuberculosis patients. The World Health Organization (WHO) therefore revised its policy on case detection by microscopy² in 2007 to recommend a reduction in the number of specimens examined, from three to two in settings with appropriate external quality assurance and documented good-quality microscopy. This approach greatly reduces the workload of laboratories, a considerable advantage in countries with a high proportion of smear-negative tuberculosis patients because of human immunodeficiency virus (HIV), extrapulmonary disease, or both.

Moreover, in 2011, the WHO recommended in a policy statement that countries that have implemented the current WHO policy for two-specimen case-finding consider switching to same-day diagnosis, especially in settings where patients are likely to default from the diagnostic pathway.³

Second, regarding the interferon-gamma-release assay, the 2011 WHO policy stated that there are not only insufficient data and low-quality evidence on the performance of this assay in low- and middle-income countries, typically those with a high tuberculosis and HIV burden, but also that the interferon-gamma-release assay and the tuberculin skin test cannot accurately predict the risk of infected individuals developing active tuberculosis. Moreover, neither the assay nor the skin test should be used for the diagnosis of active tuberculosis disease. The interferon-gamma-release assay is more costly and technically complex than the skin test. Given comparable performance but the increased cost, replacing the skin test with the interferon-gamma-release assay is not recommended as a public health intervention in resource-constrained settings.⁴ The majority of tuberculosis cases (on average 85.8%) were detected with the first sputum specimen. With the second sputum specimen, the average incremental yield was 11.9%, while the incremental yield of the third specimen, when the first two specimens were negative, was 3.1%.⁵

To the Editor: The article by Dr. Catherine Curley,¹ “Rule out pulmonary tuberculosis: clinical and radiographic clues for the internist,” was very well written, but we would like to point out two facts regarding the diagnosis of pulmonary tuberculosis, especially in high-prevalence countries like India, that might make the article more informative.

First, it has been shown conclusively that good-quality microscopy of two consecutive sputum specimens identifies the majority (95%–98%) of smear-positive tuberculosis patients. The World Health Organization (WHO) therefore revised its policy on case detection by microscopy² in 2007 to recommend a reduction in the number of specimens examined, from three to two in settings with appropriate external quality assurance and documented good-quality microscopy. This approach greatly reduces the workload of laboratories, a considerable advantage in countries with a high proportion of smear-negative tuberculosis patients because of human immunodeficiency virus (HIV), extrapulmonary disease, or both.

Moreover, in 2011, the WHO recommended in a policy statement that countries that have implemented the current WHO policy for two-specimen case-finding consider switching to same-day diagnosis, especially in settings where patients are likely to default from the diagnostic pathway.³

Second, regarding the interferon-gamma-release assay, the 2011 WHO policy stated that there are not only insufficient data and low-quality evidence on the performance of this assay in low- and middle-income countries, typically those with a high tuberculosis and HIV burden, but also that the interferon-gamma-release assay and the tuberculin skin test cannot accurately predict the risk of infected individuals developing active tuberculosis. Moreover, neither the assay nor the skin test should be used for the diagnosis of active tuberculosis disease. The interferon-gamma-release assay is more costly and technically complex than the skin test. Given comparable performance but the increased cost, replacing the skin test with the interferon-gamma-release assay is not recommended as a public health intervention in resource-constrained settings.⁴ The majority of tuberculosis cases (on average 85.8%) were detected with the first sputum specimen. With the second sputum specimen, the average incremental yield was 11.9%, while the incremental yield of the third specimen, when the first two specimens were negative, was 3.1%.⁵

To the Editor: The article by Dr. Catherine Curley,¹ “Rule out pulmonary tuberculosis: clinical and radiographic clues for the internist,” was very well written, but we would like to point out two facts regarding the diagnosis of pulmonary tuberculosis, especially in high-prevalence countries like India, that might make the article more informative.

First, it has been shown conclusively that good-quality microscopy of two consecutive sputum specimens identifies the majority (95%–98%) of smear-positive tuberculosis patients. The World Health Organization (WHO) therefore revised its policy on case detection by microscopy² in 2007 to recommend a reduction in the number of specimens examined, from three to two in settings with appropriate external quality assurance and documented good-quality microscopy. This approach greatly reduces the workload of laboratories, a considerable advantage in countries with a high proportion of smear-negative tuberculosis patients because of human immunodeficiency virus (HIV), extrapulmonary disease, or both.

Moreover, in 2011, the WHO recommended in a policy statement that countries that have implemented the current WHO policy for two-specimen case-finding consider switching to same-day diagnosis, especially in settings where patients are likely to default from the diagnostic pathway.³

Second, regarding the interferon-gamma-release assay, the 2011 WHO policy stated that there are not only insufficient data and low-quality evidence on the performance of this assay in low- and middle-income countries, typically those with a high tuberculosis and HIV burden, but also that the interferon-gamma-release assay and the tuberculin skin test cannot accurately predict the risk of infected individuals developing active tuberculosis. Moreover, neither the assay nor the skin test should be used for the diagnosis of active tuberculosis disease. The interferon-gamma-release assay is more costly and technically complex than the skin test. Given comparable performance but the increased cost, replacing the skin test with the interferon-gamma-release assay is not recommended as a public health intervention in resource-constrained settings.⁴ The majority of tuberculosis cases (on average 85.8%) were detected with the first sputum specimen. With the second sputum specimen, the average incremental yield was 11.9%, while the incremental yield of the third specimen, when the first two specimens were negative, was 3.1%.⁵

In Reply: Thank you for your interesting and appropriate comments. The workup and testing of patients with suspected tuberculosis is clearly different in countries with a higher prevalence of tuberculosis than in countries with a lower prevalence. I appreciate that both the purified protein derivative and the interferon-gamma-release assay have very limited utility in the evaluation for active tuberculosis when there is a very high background prevalence of latent tuberculosis infection. In low-prevalence countries like the United States, tuberculosis is often considered in the differential diagnosis even when other infections or lung cancer is more likely. The tests for latent tuberculosis are considered quite important in the workup of active tuberculosis in this setting.

In Reply: Thank you for your interesting and appropriate comments. The workup and testing of patients with suspected tuberculosis is clearly different in countries with a higher prevalence of tuberculosis than in countries with a lower prevalence. I appreciate that both the purified protein derivative and the interferon-gamma-release assay have very limited utility in the evaluation for active tuberculosis when there is a very high background prevalence of latent tuberculosis infection. In low-prevalence countries like the United States, tuberculosis is often considered in the differential diagnosis even when other infections or lung cancer is more likely. The tests for latent tuberculosis are considered quite important in the workup of active tuberculosis in this setting.

In Reply: Thank you for your interesting and appropriate comments. The workup and testing of patients with suspected tuberculosis is clearly different in countries with a higher prevalence of tuberculosis than in countries with a lower prevalence. I appreciate that both the purified protein derivative and the interferon-gamma-release assay have very limited utility in the evaluation for active tuberculosis when there is a very high background prevalence of latent tuberculosis infection. In low-prevalence countries like the United States, tuberculosis is often considered in the differential diagnosis even when other infections or lung cancer is more likely. The tests for latent tuberculosis are considered quite important in the workup of active tuberculosis in this setting.

To the Editor: Recently, Drs. Zimmerman and Pantalone¹ cited the Diabetes Control and Complications Trial (DCCT)² and the United Kingdom Prospective Diabetes Study (UKPDS)³ as evidence that glycemic control lowers cardiac risk in type 2 diabetes. And in a related counterpoint article, Drs. Menon and Aggarwal⁴ also discussed the UKPDS.

These studies should not be cited in this context, since the DCCT is a study of type 1 and not type 2 diabetic patients, and the UKPDS was performed in an era when statins were not available. The UKPDS was launched in 1977 and completed in 1997, and statins were not available until 1987. Indeed, the UKPDS showed that the strongest risk factor for myocardial infarction was an elevated level of low-density lipoprotein cholesterol, followed by a low level of high-density lipoprotein cholesterol.⁵ It is therefore not surprising that in the initial UKPDS report the incidence of myocardial infarction was not increased in the group with a 0.9% higher hemoglobin A_1c, but that in the 10-year follow-up, when statins were probably used by most patients, myocardial infarction was reduced by a significant 15% (P = .01).^3,6 As would be expected in the more modern studies, ie, the Action to Control Cardiovascular Risk (ACCORD),⁷ the Action in Diabetes and Vascular Disease (ADVANCE),⁸ and the Veteran Affairs Diabetes Trial (VADT),⁹ cardiovascular events were not reduced with improved glycemic control.

While the UKPDS clearly demonstrated a decrease in microvascular disease due to improved glycemic control, it should not be used as evidence that improved glycemic control in type 2 diabetes decreases cardiac events.^3,6

To the Editor: Recently, Drs. Zimmerman and Pantalone¹ cited the Diabetes Control and Complications Trial (DCCT)² and the United Kingdom Prospective Diabetes Study (UKPDS)³ as evidence that glycemic control lowers cardiac risk in type 2 diabetes. And in a related counterpoint article, Drs. Menon and Aggarwal⁴ also discussed the UKPDS.

These studies should not be cited in this context, since the DCCT is a study of type 1 and not type 2 diabetic patients, and the UKPDS was performed in an era when statins were not available. The UKPDS was launched in 1977 and completed in 1997, and statins were not available until 1987. Indeed, the UKPDS showed that the strongest risk factor for myocardial infarction was an elevated level of low-density lipoprotein cholesterol, followed by a low level of high-density lipoprotein cholesterol.⁵ It is therefore not surprising that in the initial UKPDS report the incidence of myocardial infarction was not increased in the group with a 0.9% higher hemoglobin A_1c, but that in the 10-year follow-up, when statins were probably used by most patients, myocardial infarction was reduced by a significant 15% (P = .01).^3,6 As would be expected in the more modern studies, ie, the Action to Control Cardiovascular Risk (ACCORD),⁷ the Action in Diabetes and Vascular Disease (ADVANCE),⁸ and the Veteran Affairs Diabetes Trial (VADT),⁹ cardiovascular events were not reduced with improved glycemic control.

While the UKPDS clearly demonstrated a decrease in microvascular disease due to improved glycemic control, it should not be used as evidence that improved glycemic control in type 2 diabetes decreases cardiac events.^3,6

To the Editor: Recently, Drs. Zimmerman and Pantalone¹ cited the Diabetes Control and Complications Trial (DCCT)² and the United Kingdom Prospective Diabetes Study (UKPDS)³ as evidence that glycemic control lowers cardiac risk in type 2 diabetes. And in a related counterpoint article, Drs. Menon and Aggarwal⁴ also discussed the UKPDS.

These studies should not be cited in this context, since the DCCT is a study of type 1 and not type 2 diabetic patients, and the UKPDS was performed in an era when statins were not available. The UKPDS was launched in 1977 and completed in 1997, and statins were not available until 1987. Indeed, the UKPDS showed that the strongest risk factor for myocardial infarction was an elevated level of low-density lipoprotein cholesterol, followed by a low level of high-density lipoprotein cholesterol.⁵ It is therefore not surprising that in the initial UKPDS report the incidence of myocardial infarction was not increased in the group with a 0.9% higher hemoglobin A_1c, but that in the 10-year follow-up, when statins were probably used by most patients, myocardial infarction was reduced by a significant 15% (P = .01).^3,6 As would be expected in the more modern studies, ie, the Action to Control Cardiovascular Risk (ACCORD),⁷ the Action in Diabetes and Vascular Disease (ADVANCE),⁸ and the Veteran Affairs Diabetes Trial (VADT),⁹ cardiovascular events were not reduced with improved glycemic control.

While the UKPDS clearly demonstrated a decrease in microvascular disease due to improved glycemic control, it should not be used as evidence that improved glycemic control in type 2 diabetes decreases cardiac events.^3,6

In Reply: We appreciate Dr. Bell’s interest in and comments regarding our recent article. Dr. Bell contends that the DCCT¹ and UKPDS² studies should not be cited since the DCCT is a study of type 1 and not type 2 diabetic patients, and the UKPDS was performed in an era when statins were not available.

While we can appreciate his point of view, we disagree with his interpretation of the available data. These studies, and their respective observational follow-up reports,^3,4 provide evidence that early intervention may reduce cardiovascular risk, and that our approach to examining cardiovascular risk reduction in high-risk cardiovascular patients, as in ACCORD,⁵ ADVANCE,⁶ and VADT,⁷ may be short-sighted. There is an important difference between reducing long-term cardiovascular risk by treating younger and healthier patients with diabetes (type 1 or type 2) early in the disease course, before the development of complications (including cardiovascular disease), as was the case in DCCT and UKPDS, vs treating older patients with diabetes who have established cardiovascular disease or who have numerous risk factors substantially increasing their cardiovascular risk, as in ACCORD, ADVANCE, and VADT.

To his second point, that the UKPDS did not demonstrate cardiovascular risk reduction until after the 10-year follow-up when statins were probably utilized by the vast majority of patients, there would not have been a difference in cardiac events between treatment and control groups during this observational period if the statins were the cause of the reduced rate of cardiac events. The control and treatment groups would have had the same reduction in events. That was not the case. The finding of a lower risk of myocardial infarction at the completion of the follow-up period, despite ubiquitous statin use by both the treatment and control groups during this 10-year period, suggests another variable—ie, that the early differences in glycemic control achieved between the treatment and control groups during the UKPDS was responsible for the observed reduction in the risk of myocardial infarction.

In Reply: We appreciate Dr. Bell’s interest in and comments regarding our recent article. Dr. Bell contends that the DCCT¹ and UKPDS² studies should not be cited since the DCCT is a study of type 1 and not type 2 diabetic patients, and the UKPDS was performed in an era when statins were not available.

While we can appreciate his point of view, we disagree with his interpretation of the available data. These studies, and their respective observational follow-up reports,^3,4 provide evidence that early intervention may reduce cardiovascular risk, and that our approach to examining cardiovascular risk reduction in high-risk cardiovascular patients, as in ACCORD,⁵ ADVANCE,⁶ and VADT,⁷ may be short-sighted. There is an important difference between reducing long-term cardiovascular risk by treating younger and healthier patients with diabetes (type 1 or type 2) early in the disease course, before the development of complications (including cardiovascular disease), as was the case in DCCT and UKPDS, vs treating older patients with diabetes who have established cardiovascular disease or who have numerous risk factors substantially increasing their cardiovascular risk, as in ACCORD, ADVANCE, and VADT.

To his second point, that the UKPDS did not demonstrate cardiovascular risk reduction until after the 10-year follow-up when statins were probably utilized by the vast majority of patients, there would not have been a difference in cardiac events between treatment and control groups during this observational period if the statins were the cause of the reduced rate of cardiac events. The control and treatment groups would have had the same reduction in events. That was not the case. The finding of a lower risk of myocardial infarction at the completion of the follow-up period, despite ubiquitous statin use by both the treatment and control groups during this 10-year period, suggests another variable—ie, that the early differences in glycemic control achieved between the treatment and control groups during the UKPDS was responsible for the observed reduction in the risk of myocardial infarction.

In Reply: We appreciate Dr. Bell’s interest in and comments regarding our recent article. Dr. Bell contends that the DCCT¹ and UKPDS² studies should not be cited since the DCCT is a study of type 1 and not type 2 diabetic patients, and the UKPDS was performed in an era when statins were not available.

While we can appreciate his point of view, we disagree with his interpretation of the available data. These studies, and their respective observational follow-up reports,^3,4 provide evidence that early intervention may reduce cardiovascular risk, and that our approach to examining cardiovascular risk reduction in high-risk cardiovascular patients, as in ACCORD,⁵ ADVANCE,⁶ and VADT,⁷ may be short-sighted. There is an important difference between reducing long-term cardiovascular risk by treating younger and healthier patients with diabetes (type 1 or type 2) early in the disease course, before the development of complications (including cardiovascular disease), as was the case in DCCT and UKPDS, vs treating older patients with diabetes who have established cardiovascular disease or who have numerous risk factors substantially increasing their cardiovascular risk, as in ACCORD, ADVANCE, and VADT.

To his second point, that the UKPDS did not demonstrate cardiovascular risk reduction until after the 10-year follow-up when statins were probably utilized by the vast majority of patients, there would not have been a difference in cardiac events between treatment and control groups during this observational period if the statins were the cause of the reduced rate of cardiac events. The control and treatment groups would have had the same reduction in events. That was not the case. The finding of a lower risk of myocardial infarction at the completion of the follow-up period, despite ubiquitous statin use by both the treatment and control groups during this 10-year period, suggests another variable—ie, that the early differences in glycemic control achieved between the treatment and control groups during the UKPDS was responsible for the observed reduction in the risk of myocardial infarction.

To the Editor: We read with interest the article by Ching Sun et al¹ on the relationship between diabetes therapy and cancer risk. We noted that there was no reference in the text to the long-acting insulins detemir and degludec, and we would like to add some relevant information.

With regard to detemir, a meta-analysis published in 2009 showed that patients treated with this insulin had a lower or similar rate of occurrence of a cancer compared with patients treated with neutral protamine Hagedorn insulin or insulin glargine.² In addition, in a cohort study, no difference in cancer risk between insulin detemir users and nonusers was reported.³

Insulin detemir has a lower binding affinity for human insulin receptor isoform A  (IR-A) relative to human insulin, and a much lower affinity for isoform B  (IR-B). The binding affinity ratio of insulinlike growth factor-1 (IGF-1) receptor to insulin receptor for detemir is less than or equal to 1 relative to human insulin and displays a dissociation pattern from the insulin receptor that is similar to or faster than that of human insulin. Consequently, the relative mitogenic potency of detemir in cell types predominantly expressing either the IGF-1 receptor or the insulin receptor is low and corresponds to its IGF-1 receptor and insulin receptor affinities.⁴

Regarding insulin degludec, its affinity for both IR-A and IR-B, as well as for the IGF-1 receptor, has been found to be lower than human insulin. Its mitogenic response, in the absence of albumin, was reported to range from 4% to 14% relative to human insulin.⁵ Furthermore, in cellular assays, in which no albumin was added, the in vitro metabolic potency was determined to be in the range of 8% to 20%, resulting in a mitogenic-to-metabolic potency ratio of 1 or lower.⁵

It appears that insulins detemir and degludec have low mitogenic potential. However, additional studies are needed, especially with degludec, to further determine long-term safety.

To the Editor: We read with interest the article by Ching Sun et al¹ on the relationship between diabetes therapy and cancer risk. We noted that there was no reference in the text to the long-acting insulins detemir and degludec, and we would like to add some relevant information.

With regard to detemir, a meta-analysis published in 2009 showed that patients treated with this insulin had a lower or similar rate of occurrence of a cancer compared with patients treated with neutral protamine Hagedorn insulin or insulin glargine.² In addition, in a cohort study, no difference in cancer risk between insulin detemir users and nonusers was reported.³

Insulin detemir has a lower binding affinity for human insulin receptor isoform A  (IR-A) relative to human insulin, and a much lower affinity for isoform B  (IR-B). The binding affinity ratio of insulinlike growth factor-1 (IGF-1) receptor to insulin receptor for detemir is less than or equal to 1 relative to human insulin and displays a dissociation pattern from the insulin receptor that is similar to or faster than that of human insulin. Consequently, the relative mitogenic potency of detemir in cell types predominantly expressing either the IGF-1 receptor or the insulin receptor is low and corresponds to its IGF-1 receptor and insulin receptor affinities.⁴

Regarding insulin degludec, its affinity for both IR-A and IR-B, as well as for the IGF-1 receptor, has been found to be lower than human insulin. Its mitogenic response, in the absence of albumin, was reported to range from 4% to 14% relative to human insulin.⁵ Furthermore, in cellular assays, in which no albumin was added, the in vitro metabolic potency was determined to be in the range of 8% to 20%, resulting in a mitogenic-to-metabolic potency ratio of 1 or lower.⁵

It appears that insulins detemir and degludec have low mitogenic potential. However, additional studies are needed, especially with degludec, to further determine long-term safety.

To the Editor: We read with interest the article by Ching Sun et al¹ on the relationship between diabetes therapy and cancer risk. We noted that there was no reference in the text to the long-acting insulins detemir and degludec, and we would like to add some relevant information.

With regard to detemir, a meta-analysis published in 2009 showed that patients treated with this insulin had a lower or similar rate of occurrence of a cancer compared with patients treated with neutral protamine Hagedorn insulin or insulin glargine.² In addition, in a cohort study, no difference in cancer risk between insulin detemir users and nonusers was reported.³

Insulin detemir has a lower binding affinity for human insulin receptor isoform A  (IR-A) relative to human insulin, and a much lower affinity for isoform B  (IR-B). The binding affinity ratio of insulinlike growth factor-1 (IGF-1) receptor to insulin receptor for detemir is less than or equal to 1 relative to human insulin and displays a dissociation pattern from the insulin receptor that is similar to or faster than that of human insulin. Consequently, the relative mitogenic potency of detemir in cell types predominantly expressing either the IGF-1 receptor or the insulin receptor is low and corresponds to its IGF-1 receptor and insulin receptor affinities.⁴

Regarding insulin degludec, its affinity for both IR-A and IR-B, as well as for the IGF-1 receptor, has been found to be lower than human insulin. Its mitogenic response, in the absence of albumin, was reported to range from 4% to 14% relative to human insulin.⁵ Furthermore, in cellular assays, in which no albumin was added, the in vitro metabolic potency was determined to be in the range of 8% to 20%, resulting in a mitogenic-to-metabolic potency ratio of 1 or lower.⁵

It appears that insulins detemir and degludec have low mitogenic potential. However, additional studies are needed, especially with degludec, to further determine long-term safety.