Ovarian cancer screening has more ‘harms’ than ‘benefit’

Ovarian cancer remains the leading cause of death from gynecologic malignancy in the United States. The poor survival rate associated with ovarian cancer is largely because of the advanced stage of the cancer at the time of diagnosis in the majority of patients. As with other cancers, survival is significantly increased when ovarian cancer is detected at an early stage. For example, for women with stage I cancer, the 5-year survival rate is 89%; 66% for stage II; 34% for stage III, and 18% for stage IV. Consequently, there has been a significant amount of research aimed at early detection and the development of screening strategies. Screening methods evaluated include serum tumor markers and ultrasonography. An ovarian cancer symptom index has been developed which has been used in combination with tumor markers. Nevertheless, the current evidence strongly argues against routine screening in average-risk women with CA-125 and/or ultrasonography (JAMA 2011;305:2295-303). We will briefly discuss the current evidence as well as the current guidelines for screening inpatients with and without a strong family history.

Tumor markers, most specifically CA-125, have received significant attention as they are noninvasive, easy to repeat, and relatively inexpensive. While serum CA-125 is elevated in 50% of women with early stage ovarian cancer, it is nonspecific and can be elevated in up to 1% of healthy women as well as in women with other benign and malignant conditions. Therefore, trials looking at annual screening CA-125 levels have failed to show sufficient specificity for the test to be used in an average risk population. In addition, the lifetime risk of developing ovarian cancer in the United States is 1.4%. As a consequence of this low prevalence, CA-125 has shown unacceptably low positive predictive values as a stand-alone test, ranging from 2.6%-3.7% (Am. J. Obstet. Gynecol. 2005; 193:163-9; Obstet. Gynecol. 2009;113:775-82).

A second tumor marker, with similar sensitivity to CA-125, is HE4, human epididymis protein 4. In a validated algorithm, using both tests appears to be more sensitive than either test alone, correctly classifying 93.8% of masses as high risk of being an epithelial ovarian cancer (Gynecol. Oncol. 2009;112:40-6).

Nevertheless, the HE4 assay has not been studied for the purpose of screening. The improved sensitivity seen with this combination suggests that such a strategy may provide improved detection rates as the first step in screening protocols. However, in order for any of these tumor markers to serve as screening tests, they should be able to predict disease before the clinical diagnosis has been established, and this has not been the case.

Both of these tumor markers have been evaluated in combination with the ovarian cancer symptom index. The index screens women for symptoms related to ovarian cancer such as bloating, increased abdominal girth or early food satiety occurring more than 12 times per month for less than a year. The symptom index, when used in combination with CA-125, improves the sensitivity over CA-125 alone. Furthermore, if HE4 is included, the sensitivity is 84% when two of the three tests are positive. However, because of the lack of specificity of the symptoms included in the index, the test should only be used in a stepwise fashion, and women who test positive need follow-up with transvaginal ultrasound.

Unfortunately, the sensitivity of transvaginal ultrasonography (TVUS) is user dependent, and the test has not been shown to be much better than CA-125 when used alone or in combination with the tumor marker. In the UKCTOCS (U.K. Collaborative Trial of Ovarian Cancer Screening) trial, TVUS detected 25 invasive carcinomas but only 12 of those were early-stage disease. The UKCTOCS trial also enrolled patients into a multimodal screening arm. In that arm, the Risk of Ovarian Cancer Algorithm (ROCA) was utilized as the first screening step and patients that were deemed high risk based on the algorithm had a TVUS. The ROCA is a computer algorithm that essentially compares changes that happen in a woman’s baseline CA-125 levels over time with the baseline changes seen in women who developed ovarian cancer. Patients are then deemed low risk, intermediate or high risk based on their ROCA score. For those who are low risk, a CA-125 is repeated in a year; patients at intermediate risk have a CA-125 repeated within 3 months. Each time the CA-125 is checked, the patient undergoes a new risk stratification using ROCA and her risk may be upgraded or downgraded, based on the new score. Lastly, if a patient is deemed to be at high risk, she undergoes a TVUS for further evaluation. Using this algorithm, the UKCTOCS demonstrated that multimodal screening with TVUS and CA-125 had better sensitivity, specificity, and positive predictive for detection of invasive primary cancer than TVUS alone (Lancet Oncol. 2009;10:327-40).

Similarly, a single-arm prospective study in the United States looking at a two-stage screening strategy using ROCA and TVUS showed comparable specificity and positive predictive values to those reported on UKCTOCS (Cancer 2013;119:3454-61). Consequently, multimodal screening resulted in significantly fewer surgeries performed to detect a case of cancer in both studies. However, survival studies are needed to supplement the U.S.-based trial, and the UKCTOCS study has not yet shown an improved survival in the "screened" population. The final results for the latter trial will not be available until 2015.

Ideally, a good screening test would decrease unnecessary operations because of the risks for serious complications that they carry. This was demonstrated by the randomized PLCO (Prostate, Lung, Colorectal, and Ovarian Cancer Screening) trial, in which multimodal screening (CA-125 and TVUS) was evaluated. Complications ("harms") from screening were one of the secondary outcomes. Screening led to surgery for almost a third of the women who had positive findings. Because the positive predictive value in this study was only 3.7%, many surgeries were performed for benign conditions. A total of 15% of the women who had surgery for a false-positive result experienced at least one serious complication. Furthermore, the increased harm did not come with improved detection or survival rates. The women in the screening arm had similar disease-specific and all-cause mortality as did the women in the nonintervention arm, and there was no difference in ovarian cancer stage at the time of diagnosis.

In summary, based on the significant potential harms without significant benefits, there is no role for the use of either CA-125 or ultrasound in screening for ovarian cancer in women without a suspected family history of hereditary ovarian cancer syndromes. This position has been endorsed by the U.S. Preventive Services Task Force, the Society of Gynecologic Oncology (SGO), and the American College of Obstetricians and Gynecologists (ACOG).

The benefit, or lack thereof, of ovarian cancer screening in women with BRCA1, BRCA2 mutations, or Lynch syndrome is less well defined, mainly because of the paucity of data. Randomized trials addressing screening are unlikely to be performed in this population because the low likelihood that any patient would agree to be in the no-screening arm. The small amount of evidence that exists does not show improved early detection and is consequently not very reassuring. However, given the high lifetime risk of these patients developing ovarian cancer, SGO, AGOG, and the National Comprehensive Cancer Network recommend routine screening with CA-125 and TVUS every 6 months starting at age 30-35 years old or 5-10 years before the age of first diagnosis in the family.

In conclusion, the current evidence suggests that the harm of screening for ovarian cancer with CA-125 and ultrasound outweighs the benefits in women at average risks for the development of the disease. Thus, screening in this group is neither effective nor indicated (N. Engl. J. Med. 2009;361:170-7). In women with hereditary ovarian cancer syndromes, screening is encouraged, although we currently lack evidence that early detection or survival will be improved. Further research is needed in order to develop safer, more reliable, and cost-effective screening strategies.

Dr. Clarke-Pearson is the chair and the Robert A. Ross Distinguished Professor of Obstetrics and Gynecology, and a professor of in the division of gynecologic oncology at the University of North Carolina, Chapel Hill. Dr. Roque is a fellow in the gynecologic oncology program at the University of North Carolina. Neither Dr. Clarke-Pearson nor Dr. Roque has any relevant financial disclosures. E-mail them at [email protected].

Ovarian cancer remains the leading cause of death from gynecologic malignancy in the United States. The poor survival rate associated with ovarian cancer is largely because of the advanced stage of the cancer at the time of diagnosis in the majority of patients. As with other cancers, survival is significantly increased when ovarian cancer is detected at an early stage. For example, for women with stage I cancer, the 5-year survival rate is 89%; 66% for stage II; 34% for stage III, and 18% for stage IV. Consequently, there has been a significant amount of research aimed at early detection and the development of screening strategies. Screening methods evaluated include serum tumor markers and ultrasonography. An ovarian cancer symptom index has been developed which has been used in combination with tumor markers. Nevertheless, the current evidence strongly argues against routine screening in average-risk women with CA-125 and/or ultrasonography (JAMA 2011;305:2295-303). We will briefly discuss the current evidence as well as the current guidelines for screening inpatients with and without a strong family history.

Tumor markers, most specifically CA-125, have received significant attention as they are noninvasive, easy to repeat, and relatively inexpensive. While serum CA-125 is elevated in 50% of women with early stage ovarian cancer, it is nonspecific and can be elevated in up to 1% of healthy women as well as in women with other benign and malignant conditions. Therefore, trials looking at annual screening CA-125 levels have failed to show sufficient specificity for the test to be used in an average risk population. In addition, the lifetime risk of developing ovarian cancer in the United States is 1.4%. As a consequence of this low prevalence, CA-125 has shown unacceptably low positive predictive values as a stand-alone test, ranging from 2.6%-3.7% (Am. J. Obstet. Gynecol. 2005; 193:163-9; Obstet. Gynecol. 2009;113:775-82).

A second tumor marker, with similar sensitivity to CA-125, is HE4, human epididymis protein 4. In a validated algorithm, using both tests appears to be more sensitive than either test alone, correctly classifying 93.8% of masses as high risk of being an epithelial ovarian cancer (Gynecol. Oncol. 2009;112:40-6).

Nevertheless, the HE4 assay has not been studied for the purpose of screening. The improved sensitivity seen with this combination suggests that such a strategy may provide improved detection rates as the first step in screening protocols. However, in order for any of these tumor markers to serve as screening tests, they should be able to predict disease before the clinical diagnosis has been established, and this has not been the case.

Both of these tumor markers have been evaluated in combination with the ovarian cancer symptom index. The index screens women for symptoms related to ovarian cancer such as bloating, increased abdominal girth or early food satiety occurring more than 12 times per month for less than a year. The symptom index, when used in combination with CA-125, improves the sensitivity over CA-125 alone. Furthermore, if HE4 is included, the sensitivity is 84% when two of the three tests are positive. However, because of the lack of specificity of the symptoms included in the index, the test should only be used in a stepwise fashion, and women who test positive need follow-up with transvaginal ultrasound.

Unfortunately, the sensitivity of transvaginal ultrasonography (TVUS) is user dependent, and the test has not been shown to be much better than CA-125 when used alone or in combination with the tumor marker. In the UKCTOCS (U.K. Collaborative Trial of Ovarian Cancer Screening) trial, TVUS detected 25 invasive carcinomas but only 12 of those were early-stage disease. The UKCTOCS trial also enrolled patients into a multimodal screening arm. In that arm, the Risk of Ovarian Cancer Algorithm (ROCA) was utilized as the first screening step and patients that were deemed high risk based on the algorithm had a TVUS. The ROCA is a computer algorithm that essentially compares changes that happen in a woman’s baseline CA-125 levels over time with the baseline changes seen in women who developed ovarian cancer. Patients are then deemed low risk, intermediate or high risk based on their ROCA score. For those who are low risk, a CA-125 is repeated in a year; patients at intermediate risk have a CA-125 repeated within 3 months. Each time the CA-125 is checked, the patient undergoes a new risk stratification using ROCA and her risk may be upgraded or downgraded, based on the new score. Lastly, if a patient is deemed to be at high risk, she undergoes a TVUS for further evaluation. Using this algorithm, the UKCTOCS demonstrated that multimodal screening with TVUS and CA-125 had better sensitivity, specificity, and positive predictive for detection of invasive primary cancer than TVUS alone (Lancet Oncol. 2009;10:327-40).

Similarly, a single-arm prospective study in the United States looking at a two-stage screening strategy using ROCA and TVUS showed comparable specificity and positive predictive values to those reported on UKCTOCS (Cancer 2013;119:3454-61). Consequently, multimodal screening resulted in significantly fewer surgeries performed to detect a case of cancer in both studies. However, survival studies are needed to supplement the U.S.-based trial, and the UKCTOCS study has not yet shown an improved survival in the "screened" population. The final results for the latter trial will not be available until 2015.

Ideally, a good screening test would decrease unnecessary operations because of the risks for serious complications that they carry. This was demonstrated by the randomized PLCO (Prostate, Lung, Colorectal, and Ovarian Cancer Screening) trial, in which multimodal screening (CA-125 and TVUS) was evaluated. Complications ("harms") from screening were one of the secondary outcomes. Screening led to surgery for almost a third of the women who had positive findings. Because the positive predictive value in this study was only 3.7%, many surgeries were performed for benign conditions. A total of 15% of the women who had surgery for a false-positive result experienced at least one serious complication. Furthermore, the increased harm did not come with improved detection or survival rates. The women in the screening arm had similar disease-specific and all-cause mortality as did the women in the nonintervention arm, and there was no difference in ovarian cancer stage at the time of diagnosis.

In summary, based on the significant potential harms without significant benefits, there is no role for the use of either CA-125 or ultrasound in screening for ovarian cancer in women without a suspected family history of hereditary ovarian cancer syndromes. This position has been endorsed by the U.S. Preventive Services Task Force, the Society of Gynecologic Oncology (SGO), and the American College of Obstetricians and Gynecologists (ACOG).

The benefit, or lack thereof, of ovarian cancer screening in women with BRCA1, BRCA2 mutations, or Lynch syndrome is less well defined, mainly because of the paucity of data. Randomized trials addressing screening are unlikely to be performed in this population because the low likelihood that any patient would agree to be in the no-screening arm. The small amount of evidence that exists does not show improved early detection and is consequently not very reassuring. However, given the high lifetime risk of these patients developing ovarian cancer, SGO, AGOG, and the National Comprehensive Cancer Network recommend routine screening with CA-125 and TVUS every 6 months starting at age 30-35 years old or 5-10 years before the age of first diagnosis in the family.

In conclusion, the current evidence suggests that the harm of screening for ovarian cancer with CA-125 and ultrasound outweighs the benefits in women at average risks for the development of the disease. Thus, screening in this group is neither effective nor indicated (N. Engl. J. Med. 2009;361:170-7). In women with hereditary ovarian cancer syndromes, screening is encouraged, although we currently lack evidence that early detection or survival will be improved. Further research is needed in order to develop safer, more reliable, and cost-effective screening strategies.

Dr. Clarke-Pearson is the chair and the Robert A. Ross Distinguished Professor of Obstetrics and Gynecology, and a professor of in the division of gynecologic oncology at the University of North Carolina, Chapel Hill. Dr. Roque is a fellow in the gynecologic oncology program at the University of North Carolina. Neither Dr. Clarke-Pearson nor Dr. Roque has any relevant financial disclosures. E-mail them at [email protected].

Ovarian cancer remains the leading cause of death from gynecologic malignancy in the United States. The poor survival rate associated with ovarian cancer is largely because of the advanced stage of the cancer at the time of diagnosis in the majority of patients. As with other cancers, survival is significantly increased when ovarian cancer is detected at an early stage. For example, for women with stage I cancer, the 5-year survival rate is 89%; 66% for stage II; 34% for stage III, and 18% for stage IV. Consequently, there has been a significant amount of research aimed at early detection and the development of screening strategies. Screening methods evaluated include serum tumor markers and ultrasonography. An ovarian cancer symptom index has been developed which has been used in combination with tumor markers. Nevertheless, the current evidence strongly argues against routine screening in average-risk women with CA-125 and/or ultrasonography (JAMA 2011;305:2295-303). We will briefly discuss the current evidence as well as the current guidelines for screening inpatients with and without a strong family history.

Tumor markers, most specifically CA-125, have received significant attention as they are noninvasive, easy to repeat, and relatively inexpensive. While serum CA-125 is elevated in 50% of women with early stage ovarian cancer, it is nonspecific and can be elevated in up to 1% of healthy women as well as in women with other benign and malignant conditions. Therefore, trials looking at annual screening CA-125 levels have failed to show sufficient specificity for the test to be used in an average risk population. In addition, the lifetime risk of developing ovarian cancer in the United States is 1.4%. As a consequence of this low prevalence, CA-125 has shown unacceptably low positive predictive values as a stand-alone test, ranging from 2.6%-3.7% (Am. J. Obstet. Gynecol. 2005; 193:163-9; Obstet. Gynecol. 2009;113:775-82).

A second tumor marker, with similar sensitivity to CA-125, is HE4, human epididymis protein 4. In a validated algorithm, using both tests appears to be more sensitive than either test alone, correctly classifying 93.8% of masses as high risk of being an epithelial ovarian cancer (Gynecol. Oncol. 2009;112:40-6).

Nevertheless, the HE4 assay has not been studied for the purpose of screening. The improved sensitivity seen with this combination suggests that such a strategy may provide improved detection rates as the first step in screening protocols. However, in order for any of these tumor markers to serve as screening tests, they should be able to predict disease before the clinical diagnosis has been established, and this has not been the case.

Both of these tumor markers have been evaluated in combination with the ovarian cancer symptom index. The index screens women for symptoms related to ovarian cancer such as bloating, increased abdominal girth or early food satiety occurring more than 12 times per month for less than a year. The symptom index, when used in combination with CA-125, improves the sensitivity over CA-125 alone. Furthermore, if HE4 is included, the sensitivity is 84% when two of the three tests are positive. However, because of the lack of specificity of the symptoms included in the index, the test should only be used in a stepwise fashion, and women who test positive need follow-up with transvaginal ultrasound.

Unfortunately, the sensitivity of transvaginal ultrasonography (TVUS) is user dependent, and the test has not been shown to be much better than CA-125 when used alone or in combination with the tumor marker. In the UKCTOCS (U.K. Collaborative Trial of Ovarian Cancer Screening) trial, TVUS detected 25 invasive carcinomas but only 12 of those were early-stage disease. The UKCTOCS trial also enrolled patients into a multimodal screening arm. In that arm, the Risk of Ovarian Cancer Algorithm (ROCA) was utilized as the first screening step and patients that were deemed high risk based on the algorithm had a TVUS. The ROCA is a computer algorithm that essentially compares changes that happen in a woman’s baseline CA-125 levels over time with the baseline changes seen in women who developed ovarian cancer. Patients are then deemed low risk, intermediate or high risk based on their ROCA score. For those who are low risk, a CA-125 is repeated in a year; patients at intermediate risk have a CA-125 repeated within 3 months. Each time the CA-125 is checked, the patient undergoes a new risk stratification using ROCA and her risk may be upgraded or downgraded, based on the new score. Lastly, if a patient is deemed to be at high risk, she undergoes a TVUS for further evaluation. Using this algorithm, the UKCTOCS demonstrated that multimodal screening with TVUS and CA-125 had better sensitivity, specificity, and positive predictive for detection of invasive primary cancer than TVUS alone (Lancet Oncol. 2009;10:327-40).

Similarly, a single-arm prospective study in the United States looking at a two-stage screening strategy using ROCA and TVUS showed comparable specificity and positive predictive values to those reported on UKCTOCS (Cancer 2013;119:3454-61). Consequently, multimodal screening resulted in significantly fewer surgeries performed to detect a case of cancer in both studies. However, survival studies are needed to supplement the U.S.-based trial, and the UKCTOCS study has not yet shown an improved survival in the "screened" population. The final results for the latter trial will not be available until 2015.

Ideally, a good screening test would decrease unnecessary operations because of the risks for serious complications that they carry. This was demonstrated by the randomized PLCO (Prostate, Lung, Colorectal, and Ovarian Cancer Screening) trial, in which multimodal screening (CA-125 and TVUS) was evaluated. Complications ("harms") from screening were one of the secondary outcomes. Screening led to surgery for almost a third of the women who had positive findings. Because the positive predictive value in this study was only 3.7%, many surgeries were performed for benign conditions. A total of 15% of the women who had surgery for a false-positive result experienced at least one serious complication. Furthermore, the increased harm did not come with improved detection or survival rates. The women in the screening arm had similar disease-specific and all-cause mortality as did the women in the nonintervention arm, and there was no difference in ovarian cancer stage at the time of diagnosis.

In summary, based on the significant potential harms without significant benefits, there is no role for the use of either CA-125 or ultrasound in screening for ovarian cancer in women without a suspected family history of hereditary ovarian cancer syndromes. This position has been endorsed by the U.S. Preventive Services Task Force, the Society of Gynecologic Oncology (SGO), and the American College of Obstetricians and Gynecologists (ACOG).

The benefit, or lack thereof, of ovarian cancer screening in women with BRCA1, BRCA2 mutations, or Lynch syndrome is less well defined, mainly because of the paucity of data. Randomized trials addressing screening are unlikely to be performed in this population because the low likelihood that any patient would agree to be in the no-screening arm. The small amount of evidence that exists does not show improved early detection and is consequently not very reassuring. However, given the high lifetime risk of these patients developing ovarian cancer, SGO, AGOG, and the National Comprehensive Cancer Network recommend routine screening with CA-125 and TVUS every 6 months starting at age 30-35 years old or 5-10 years before the age of first diagnosis in the family.

In conclusion, the current evidence suggests that the harm of screening for ovarian cancer with CA-125 and ultrasound outweighs the benefits in women at average risks for the development of the disease. Thus, screening in this group is neither effective nor indicated (N. Engl. J. Med. 2009;361:170-7). In women with hereditary ovarian cancer syndromes, screening is encouraged, although we currently lack evidence that early detection or survival will be improved. Further research is needed in order to develop safer, more reliable, and cost-effective screening strategies.

Dr. Clarke-Pearson is the chair and the Robert A. Ross Distinguished Professor of Obstetrics and Gynecology, and a professor of in the division of gynecologic oncology at the University of North Carolina, Chapel Hill. Dr. Roque is a fellow in the gynecologic oncology program at the University of North Carolina. Neither Dr. Clarke-Pearson nor Dr. Roque has any relevant financial disclosures. E-mail them at [email protected].