OBJECTIVE: To estimate the risk of ovarian malignancy among asymptomatic women with abnormal transvaginal ultrasound scans or CA 125 and to provide guidance to physicians managing these women.
METHODS: A cohort of women from the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial with abnormal ovarian results at the initial (T0) and subsequent (T1+) screens were analyzed to estimate which findings were associated with high risk of ovarian cancer. Cancer risks more than 10% were designated as high and risks of 3% or less were designated as low.
RESULTS: For the T0 screen, two high-risk categories were identified: CA 125 of 70 or more with negative transvaginal ultrasound scan (positive predictive value [PPV] 15.9%, CI 14.7–17.7%); and positive for both CA 125 and transvaginal ultrasound scan (PPV 25.0%, CI 23.3–27.3%). For T1+ screens, three high-risk categories were identified: negative transvaginal ultrasound scan with change in CA 125 of 45 or more (PPV 29.0%, CI 28.3–30.3%); increase in size of cyst 6 cm or more with negative CA 125 (PPV 13.3%, CI 10.5–18.0%); and positive for both tests (PPV 42.9%, CI 40.0–46.0%). High-risk criteria for T0 provide a sensitivity of 60%, specificity of 96.2%, PPV of 19.7%, and a negative predictive value (NPV) of 99.3%. T1+ criteria yielded a sensitivity of 85.3%, specificity of 95.6%, PPV of 29.6%, and NPV of 99.7%.
CONCLUSIONS: High-risk categories for predicting risk of cancer in women with abnormal CA 125, transvaginal ultrasound scan, or both at initial and subsequent screens have been identified. The large number of women in this study, the 4-year complete follow-up, and small number of invasive cancers in the low-risk categories provide guidance for clinical decisions regarding need for surgery in these women.
LEVEL OF EVIDENCE: II
Risk of ovarian malignancy in asymptomatic women with abnormal CA 125, transvaginal ultrasound scans, or both can be estimated, providing guidance for operative intervention.
University of Alabama at Birmingham, Birmingham, Alabama; Marshfield Clinic Research, Foundation, Marshfield, Wisconsin; Information Management Services, Westat, and National Cancer Institute, Rockville, Maryland; and Huntsman Cancer Institute at the University of Utah, Salt Lake City, Utah.
Corresponding author: Edward E. Partridge, MD, UAB Comprehensive Cancer Center, 1802 6th Avenue South, Birmingham, AL 35294-3300; e-mail:email@example.com.
The National Cancer Institute funded the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial. This research also was supported by contracts from the Division of Cancer Prevention of the National Cancer Institute and by the Intramural Research Program of the Division of Cancer Epidemiology and Genetics National Cancer Institute, National Institutes of Health, Department of Health and Human Services (NIH–NO1-CN-75022 and NCI–NO1-CA-75022-70).
Financial Disclosure The authors did not report any potential conflicts of interest.
Carcinoma of the ovary is the leading cause of death from gynecologic malignancies. More than 21,900 women in the United States will have development of ovarian cancer annually and 15,460 will die.1 This high case-fatality rate is largely attributable to the fact that most ovarian cancer is diagnosed in an advanced stage. Because of improved survival in women with early stage cancer, there has been interest in early detection by screening asymptomatic women.
A major problem in screening for a disease with a low prevalence, such as ovarian cancer, is false-positive results. A particular problem for ovarian cancer is that evaluation of an abnormal screen includes, in many cases, an operative procedure for definitive diagnosis. The dilemma in an asymptomatic woman with an abnormal test is two-fold. First, distinguishing between a benign and malignant condition is important to avoid surgery in women with benign, asymptomatic conditions. Second, if malignancy is suspected, referral to surgeons with specialized training or experience, or both, in the operative management of this disease is possible.
To evaluate the efficacy of screening for ovarian cancer (as well as prostate, lung, and colorectal cancers), the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial was initiated in 1992. For ovaries, the Prostate, Lung, Colorectal, and Ovarian trial has the following objective: to determine in healthy women aged 55–74 years at entry whether screening with CA 125 and transvaginal ultrasonography can reduce mortality from ovarian cancer.2 Results of the initial and subsequent screens and the effect on mortality have been published.3–5 The primary analysis randomized 34,253 women to undergo screening (intervention) and 34,304 to not undergo screening (usual care). Ovarian cancer was diagnosed in 212 women in the intervention group and in 176 in the usual care group (relative risk [RR] 1.21, 95% confidence interval [CI] 0.99–1.48). There were 118 deaths caused by ovarian cancer in the intervention group and 100 deaths in the usual care group (mortality RR 1.18, 95% CI 0.82–1.71). Of 3,285 women with false-positive results, 1,080 underwent surgical follow-up; 163 of these women (15%; 95% CI 14.99–15.22%) experienced at least one serious complication, but no known surgically related deaths were recorded. The conclusions were that simultaneous screening with CA 125 and transvaginal ultrasonography compared with usual care did not reduce ovarian cancer mortality, and that diagnostic evaluation after a false-positive screening test result was associated with complications.5 Despite this, it is likely that CA 125 and transvaginal ultrasonography will continue to be used by some practitioners until dissemination of this information is fully integrated into practice and until other trials of ovarian cancer are completed. In the Prostate, Lung, Colorectal, and Ovarian trial, the participant’s primary physician was responsible for follow-up diagnostic tests, including need for referral or surgical intervention or both. No specific guidance was provided. Because these women were asymptomatic, the low rate of cancer implies that a substantial number of surgeries may have been unnecessary.
The purpose of this study was to estimate the risk of ovarian malignancy among asymptomatic postmenopausal women with abnormal findings on screening tests at initial or subsequent screening years. To provide physicians with some guidance for managing women in this situation, an additional goal was to estimate the risk using clinically relevant thresholds for surgery.
MATERIALS AND METHODS
Design of the Prostate, Lung, Colorectal, and Ovarian Trial has been described in detail elsewhere.2 All study sites for the trial obtained Institutional Review Board approval. In the Prostate, Lung, Colorectal, and Ovarian trial, a positive screen result for ovarian cancer was defined as CA 125 more than 35 units/mL or a transvaginal ultrasound scan showing ovarian or cyst volume more than 10 cm3 or solid areas, papillary projections, or mixed components within a cystic ovarian tumor of any size. The study population for this analysis included women who had both the ovarian screening tests and had an abnormal screening result for at least one.
The study evaluated the relationship of personal, demographic, and screening result characteristics with ovarian cancer detection, defined as a cancer discovered within 2 years of a positive screen. Because of the inherent differences in the risk of cancer detection between the baseline (T0) screen and the subsequent annual screens (T1+), analyses were conducted separately. The following characteristics were evaluated as potential correlates of cancer after a positive T0 screen: age; race; family history of ovarian cancer; personal or family history of breast cancer; CA 125 level; maximum ovary or cyst diameter; presence of mixed, irregular, or papillary features; or a solid mass. The analysis of positive screens at T1+ considered these in addition to the change in CA 125 level and the change in ovary or cyst size from the previous screen.
We constructed distribution tables for each of the characteristics evaluated at initial and subsequent screens (Table 1), including the number and proportion of participants in each classification, the number and proportion with surgical follow-up, and the number and proportion with cancer. Exact CIs (95%) were determined for the estimated probability of cancer detection in each of the characteristic classifications by using the Clopper-Pearson method. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) for malignancy among women with a positive screen result were calculated using standard methods.6
The underlying principle for this analysis was to use clinically relevant thresholds to distinguish between high and low risks for cancer. It was first suggested by Jacobs7 and generally accepted that a rate of one cancer detected for every 10 surgeries would be an acceptable PPV for an ovarian cancer screen. A model for surgical intervention that could, within a reasonable degree of accuracy, have a PPV of 10% or more and a corresponding NPV of 97% or more would be clinically useful. We have designated risks of malignancy of 10% or more as high. Also, from a clinical point of view, a risk of cancer of less than 3% may not warrant surgical intervention but close observation instead, particularly if the patient has comorbidities and given the 15% surgical complication rate in the Prostate, Lung, Colorectal, and Ovarian trial false-positive results. We designated a risk of malignancy of 3% or less as low. For categories of patients with estimated probability of detecting a malignancy between 3% and 10%, or when data limitations leave confidence limits overlapping into this range, we have considered the risk of malignancy as uncertain.
For the intervention arm, 39,105 women were randomized. Women who had their first abnormal suspicious screen results, who received both tests in that same year, and who had at least 2 years of follow-up were included; 3,224 women met these criteria (Fig. 1). At T0 screen, 1,611 women had an abnormal result; 543 underwent surgical intervention and 25 invasive cancers were detected. During the T1+ screening years, 1,613 additional women had an abnormal result. Of these, 329 underwent surgical evaluation and 34 additional invasive cancers were detected. None of the 34 women in whom cancer was detected in the T1+ screen had positive screens at T0. Five of these women had abnormal nonsuspicious transvaginal ultrasound results at T0, and three women did not have CA 125 performed at T0. The overall probability of cancer detection after an abnormal screen result was 1.6% (CI 0.9–2.2) at T0 and 2.1% (CI 1.4–2.8) during T1+.
Table 2 shows an analysis of the relationship of size and morphology of abnormalities found on transvaginal ultrasonography to the positive prediction of cancer at T0. If no solid or mixed, irregular, or papillary components are present, the risk of cancer is low, independent of the size of the abnormality. Similarly, a cyst less than 5 cm has a low risk of being cancer, regardless of the morphology. However, a cyst of 5 cm or more with mixed, papillary, or irregular components places a woman at uncertain risk. Size and morphology alone do not identify a high-risk category.
Predictive values for the combination of CA 125 level and transvaginal ultrasound status are shown in Table 3. If the transvaginal ultrasound scan is negative, a CA 125 less than 60 is in the range of low risk (RR 0.0%, CI 0.0–0.3%), 60–70 indicates uncertain risk (RR 4.6%, CI 2.9–7.9%), and more than 70 is considered high risk (RR 15.9%, CI 14.7–17.7%). A positive transvaginal ultrasound result with a negative CA 125 result is low risk (RR 0.7%, CI 0.7–0.8%). A combination of abnormal transvaginal ultrasound scan and abnormal CA 125 results indicates high risk (RR 25.0%, CI 23.3–27.3%).
During T1+ we were able to compare current with previous ovarian and cyst size and CA 125 levels. Table 4 compares transvaginal ultrasound scan and CA 125 changes. As in T0, the combination of abnormal results for both CA 125 and transvaginal ultrasonography predicts a high risk for cancer (RR 42.9%, CI 40.9–46.0%). With a negative transvaginal ultrasound result, the magnitude in change in CA 125 is predictive. A change of 25 or less predicts a low risk (RR 0.3%, CI 0.2–0.5%), whereas a change of 45 or more predicts a high risk (RR 29.0%, CI 28.3–30.3%). A change of 25–45 was low for risk of cancer (RR 1.7%, CI 1.4–2.4%). If the transvaginal ultrasound scan is positive, increases in CA 125 may indicate an elevated risk for ovarian cancer even when the CA 125 remains normal (less than 35). In this situation, an increase of more than 10 trends toward a high risk (RR 10.0%, CI 7.9–13.9%), although the lower limit of CI is less than 10%. A CA 125 change less than 10 predicts a low probability of cancer (RR 0.2%, CI 0.15–0.3%).
Table 5 takes into account changes in cyst or ovarian size in women with abnormal transvaginal ultrasound results, but normal CA 125 results. A change in size of 6 cm or more suggests the possibility of high risk (RR 13.3%, CI 10.5–18.0%), whereas a change in size less than 6 cm suggests low risk.
For T0, using the high-risk criteria of a negative transvaginal ultrasound result with a CA 125 level of 70 or more or a CA 125 and a positive transvaginal ultrasound result, the sensitivity for detection of cancer within the Prostate, Lung, Colorectal, and Ovarian data is 60%, the specificity is 96.2%, PPV is 19.7%, and NPV is 99.3%.
For T1+ using three high-risk criteria—negative transvaginal ultrasound result with a change in CA 125 of more than 45, CA 125 negative result but a change in diameter of 6 cm or more, or both CA 125 and transvaginal ultrasound scan with positive results–the sensitivity for detection of cancer within the Prostate, Lung, Colorectal, and Ovarian data is 85.3%, specificity is 95.6%, PPV is 29.6%, and NPV is 99.7%.
The Prostate, Lung, Colorectal, and Ovarian trial demonstrated that screening for ovarian cancer using simultaneous CA 125 and transvaginal ultrasonography did not reduce ovarian cancer mortality and was associated with complications after diagnostic evaluation of false-positive screen results.5 Despite this, it is likely that these tests will continue to be recommended because of slow dissemination of findings or resistance to incorporate into practice for various reasons.8
A screening recommendation for disease with a low prevalence can be problematic because of high false-positive test results. In the Prostate, Lung, Colorectal, and Ovarian trial of 3,285 women with false-positive results, 1,080 underwent surgery (32.9%) as part of diagnostic work-up. Among these 1,080 women, there were 20.6 complications per 100 surgical procedures.5
We reviewed screening and follow-up results from the first 3 years of the Prostate, Lung, Colorectal, and Ovarian trial to provide some guidance to physicians faced with an abnormal CA 125 or transvaginal ultrasound result in asymptomatic women. We set arbitrary but clinically meaningful thresholds of cancer risk, with 3% or less being low, 10% or more being high, and between 3% and 10% being uncertain. Women who can confidently be placed in a low-risk category could be followed-up with repeat testing and surgery could be avoided, especially with a 15% serious complication risk from surgery for false-positive results. However, women confidently placed in a high-risk category should be considered for surgery unless major comorbidities preclude this. The uncertain category or CIs that cross risk categories are more problematic and require physician judgment and an informed patient. Although our study cannot definitively estimate the cancer risk in other populations of women, we believe the empirical assessment of risk in this large population of asymptomatic women 55 years and older may provide some useful insights and potentially may avoid unnecessary surgery in women inappropriately screened.
Because of the importance of distinguishing benign from malignant conditions in women, a number of scoring systems have been developed to evaluate adnexal masses. These have included systems that incorporate combined information or data from multiple sources such as imaging, risk factors, laboratory data, and physician assessment.7,9–16
A common scoring system is the Risk of Malignancy Index17, which incorporates a clinical prediction rule based on ultrasonography, CA 125, and menopausal status. This system reported a sensitivity of 85% and specificity of 97% in a population of 143 women undergoing surgical exploration for an adnexal mass. Subsequent studies18–20 have reported similar performance, whereas others have failed to equal this performance.21–24 Several more recent studies have provided guidelines for assessment of adnexal masses for risk of malignancy. Dearking et al13 evaluated the guidelines published by the American College of Obstetricians and Gynecologists and Society of Gynecologic Oncologists for referral of adnexal masses and noted these perform well in advanced-stage disease, especially in premenopausal women. The group from the University of Kentucky most recently has proposed a multivariate index assay that performed better compared with CA 125 and physician assessment14 and also enhanced the American College of Obstetricians and Gynecologists guidelines for referral when the multivariate index assay replaced CA 125 in the guidelines.15 However, the relevance of these studies to asymptomatic women undergoing screening with CA 125 and transvaginal ultrasonography compared with evaluation of an adnexal mass is unclear.
This study evaluated criteria to distinguish benign from malignant disease among asymptomatic postmenopausal women who have abnormal transvaginal ultrasound scans or CA 125 as a result of screening. It is important to determine the risk of malignancy to potentially avoid surgery in an asymptomatic woman and to refer appropriately if there is a suspicion of malignancy. No scoring system to determine risk of malignancy has been developed and validated in an exclusively postmenopausal asymptomatic population.
We have shown that it may be possible to establish criteria based on screening test results that could reduce the number of arguably unnecessary surgeries while still remaining sensitive to cancer discovery. However, conclusions from this descriptive analysis are not definitive, and specific data points should be interpreted with caution. For example, the categorization of covariate data in this analysis to reflect low and high risk (eg, CA 125 less than 60 and more than 70, respectively) was empirical, and predictive values observed in these data likely will be overly optimistic when applied in a new population. Even in this large clinical trial, there was limited precision for estimating risk and identifying optimal cut points in risk for many of the categories with small numbers of ovarian cancer cases. We therefore strongly encourage validation and refinement of these results in other populations of postmenopausal women.
However, the large number of women in this study, the 4-year complete follow-up, and the small percentage of invasive cancers discovered among the low-risk categories should provide some guidance for clinical decisions regarding the need for surgery in asymptomatic postmenopausal women with abnormal findings on CA 125 or transvaginal ultrasonography. In conclusion, it should be emphasized that the use of combination CA 125 and transvaginal ultrasonography in asymptomatic women to screen for ovarian cancer using the protocol used in the Prostate, Lung, Colorectal, and Ovarian trial does not reduce mortality, does have associated harms, and should not be recommended.
1. Siegel R, Ward E, Brawley O, Jemal A. Cancer Statistics 2011, the impact of eliminating socioeconomic and racial disparities on premature cancer deaths. CA Cancer J 2011;5:212–36.
2. Gohagan JK, Prorok PC, Hayes RB, Kramer BS; PLCO Project Team. The Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial of the National Cancer Institute: history, organization, and status. Control Clin Trials 2000;21:251S–72S.
3. Buys SS, Partridge EE, Greene MH, Prorok PC, Reding D, Riley TL, et al..; PLCO Project Team. Ovarian cancer screening in the Prostate, Lung, Colorectal and Ovarian (PLCO) cancer screening trial: findings from the initial screen of a randomized trial. Am J Obstet Gynecol 2005;193:1630–9.
4. Partridge EE, Kreimer AR, Greenlee RT, Williams C, Xu JL, Church TR, et al..; PLCO Project Team. Results from four rounds of ovarian cancer screening in a randomized trial. Obstet Gynecol 2009;113:775–82.
5. Buys SS, Partridge EE, Black A, Johnson CC, Lamerato L, Isaacs C, et al..; PLCO Project Team. Effect of screening on ovarian cancer mortality, the Prostate, Lung, Colorectal and Ovarian (PLCO) cancer screening randomized controlled trial. JAMA 2011;305:2295–303.
6. Weiss NS. Clinical epidemiology: the study of the outcome of illness. 2nd ed. New York (NY): Oxford University Press; 1996:10–11.
7. Jacobs I, Stabile I, Bridges J, Kemsley P, Reynolds C, Grudzinskas J, et al.. Multimodal approach to screening for ovarian cancer. Lancet 1988:1:268–71.
8. Baldwin LM, Trivers KF, Matthews B, Andrilla CH, Miller JW, Berry DL, et al.. Vignette-based study of ovarian cancer screening: do U.S. physicians report adhering to evidence-based recommendations? Ann Intern Med 2012;156:182–94.
9. Maggino T, Gadduci A, D’Addario V, Pecorelli S, Lissoni A, Stella M, et al.. Prospective multicenter study on CA 125 in postmenopausal pelvic masses. Gynecol Oncol 1994;54:117–23.
10. Schutter EM, Kenemans P, Sohn C, Kristen P, Crombach G, Westermann R, et al.. Diagnostic value of pelvic examination, ultrasound, and serum CA 125 in postmenopausal women with a pelvic mass. An international multicenter study. Cancer 1994;74:1398–406.
11. Schutter EM, Sohn C, Kirsten P, Mobus V, Crombach G, Kaufmann M, et al.. Estimation of probability of malignancy using a logistic model combining physical examination, ultrasound, serum CA 125, and serum CA 72-4 in postmenopausal women with a pelvic mass: an international multicenter study. Gynecol Oncol 1998;69:56–63.
12. Adonakis GL, Paraskevaidis E, Tsiga S, Seferiadis K, Lolis DE. A combined approach for the early detection of ovarian cancer in asymptomatic women. Eur J Obstet Gynecol Reprod Biol 1996;65:221–5.
13. Dearking AC, Aletti GD, McGee ME, Weaver AL, Sommerfield MK, Cliby WA. How relevant are ACOG and SGO guidelines for referral of adnexal mass? Obstet Gynecol 2007;110:841–8.
14. Ueland FR, Desimone CP, Seamon LG, Miller RA, Goodrich S, Podzielinski I, et al.. Effectiveness of multivariate index assay in the preoperative assessment of ovarian tumors. Obstet Gynecol 2011;117:1289–97.
15. Miller RW, Smith A, DeSimone CP, Seamon LG, Goodrich S, Podzielinski I, et al.. Performance of the American College of Obstetricians and Gynecologists’ ovarian tumor referral guidelines with a multivariate index assay. Obstet Gynecol 2011;117:1298–306.
16. Moore RG, Miller MC, Disilvestro P, Landrum LM, Gajewski W, Ball JJ, et al.. Evaluation of the diagnostic accuracy of the risk of ovarian malignancy algorithm in women with a pelvic mass. Obstet Gynecol 2011;118:280–8.
17. Jacobs I, Oram D, Fairbanks J, Turner J, Frost C, Grudzinskas JG. A risk of malignancy index incorporation CA 125, ultrasound and menopausal status for the accurate preoperative diagnosis of ovarian cancer. BJOG 1990;97:922–9.
18. Asif N, Sattar A, Dawood MM, Rafi T, Aamir M, Anwar M. Pre-operative evaluation of ovarian mass: risk of malignancy index. J Coll Physicians Surg Pak 2004;14:128–31.
19. Davies AP, Jacobs I, Woolas R, Fish A, Oram D. The adnexal mass: benign or malignant? Evaluation of a risk of malignancy index. BJOG 1993;100:927–31.
20. Obeidat BR, Amarin ZO, Latimer JA, Crawford RA. Risk of malignancy index in the preoperative evaluation of pelvic masses. Int J Gynaecol Obstet 2004;85:255–8.
21. Lu C, Van Gestel T, Suykens JA, VanHuffel S, Vergote I, Timmerman D. Preoperative prediction of malignancy of ovarian tumors using least squares support vector machines. Artif Intell Med 2003;28:281–306.
22. Manjunath AP, Pratapkumar, Sujatha K, Vani R. Comparison of three risk of malignancy indices in evaluation of pelvic masses. Gynecol Oncol 2001;81:225–9.
23. Tingulstad S, Hagen B, Skjeldestad FE, Onsrud M, Kiserud T, Halvorsent T, et al.. Evaluation of a risk of malignancy index based on serum CA 125, ultrasound findings and menopausal status in the pre-operative diagnosis of pelvic masses. BJOG 1996;103:826–31.
24. Timmerman D, Bourne TH, Tailor A, Collins WP, Verrelst H, Vandenberghe K, et al.. A comparison of methods for preoperative discrimination between malignant and benign adnexal masses: the development of a new logistic regression model. Am J Obstet Gynecol 1999;181:57–65.