Single-threshold cancer antigen 125 (CA-125) levels and transvaginal ultrasound (TVU)

There is solid evidence to indicate that screening women aged 55 to 74 years at average risk of developing ovarian cancer with the serum marker CA-125 (at a fixed threshold for a positive result of 35 U/mL) annually for 6 years and TVU for 4 years does not result in a decrease in ovarian cancer mortality, after a median follow-up of 14.7 years.

Magnitude of Effect: The ovarian cancer mortality rate was 3.8 deaths per 10,000 women in the screened group and 3.6 deaths per 10,000 person-years in the usual-care group, yielding a mortality rate ratio of 1.06 (95% confidence interval [CI], 0.87–1.30).[1]

• Study Design: Evidence obtained from one randomized controlled trial.
• Internal Validity: Good.
• Consistency: One trial has evaluated the impact on mortality from ovarian cancer.
• External Validity: Good.

Screening with TVU alone or with multimodal screening with CA-125 levels, assessed using the Risk of Ovarian Cancer Algorithm (ROCA), with TVU

Screening with TVU alone or with multimodal screening with CA-125 levels, assessed using the ROCA, combined with TVU in the United Kingdom Collaborative Trial of Ovarian Cancer Screening (UKCTOCS) did not show a mortality benefit of screening with either approach based on a predetermined primary endpoint among women undergoing 7 to 11 screens and a median of 11.1 years of follow-up.[2]

Magnitude of Effect: Multimodal screening was associated with a nonsignificantly lower mortality than with no screening (15% lower mortality; 95% CI, -3% to 30%; P = .10). Ultrasound only screening also resulted in nonsignificantly lower mortality (11% lower mortality; 95% CI, -7% to 27%; P = .21).[2]

• Screening complications were less than 1% for both TVU only and multimodality screening strategies.
• Study Design: Evidence obtained from one randomized controlled trial.
• Internal Validity: Good.
• Consistency: One trial has evaluated the impact on mortality from ovarian cancer using this specific approach.
• External Validity: Good—based on data from other studies assessing complementary endpoints.

Statement of Harms

Based on solid evidence, screening for ovarian cancer results in false-positive test results. Screened women had higher rates of oophorectomy and other minor complications such as fainting and bruising.

Magnitude of Effect:

• Of screened women, 9.6% had false-positive results, resulting in 6.2% undergoing surgery. The surgical complication rate was 1.2% for all screened women.
• Oophorectomy rates were 85.7 per 10,000 person-years among screened women and 64.2 per 10,000 person-years among usual-care women.
• Minor complications with screening: 58.3 cases per 10,000 women screened with CA-125 and 3.3 cases per 10,000 women screened with transvaginal sonogram (TVS).
• Study Design: Evidence obtained from one randomized controlled trial.
• Internal Validity: Good.
• Consistency: Not applicable (N/A).
• External Validity: Good.

In the TVU-only arm of the UKCTOCS trial, there were 50 false-positive surgical procedures, and in the multimodality arm, there were 14 false-positive operations per 10,000 screens.[2]

In the general population, screening is targeted to postmenopausal women, and the major complications are related to surgery. Among younger women, the potential target group among BRCA1/2 mutation carriers, oophorectomy at younger than 45 years may increase mortality secondary to cardiovascular disease. Oophorectomy, if performed among younger women, may also reduce risk of estrogen receptor–positive breast cancers, which occur with elevated frequency among carriers of BRCA2 mutations.

Incidence and mortality

Ovarian carcinoma is the fifth leading cause of cancer death among women in the United States and has the highest mortality rate of all gynecologic cancers. It is estimated that 22,440 new cases of ovarian cancer will be diagnosed in the United States in 2017, and 14,080 women will die of this disease. Over the last 20 years, incidence rates decreased by 1% per year in white women and by 0.4% per year in African American women. From 2005 to 2014, mortality rates decreased by 2% per year in white women and by 1% per year in African American women.[1] The prognosis for survival from ovarian carcinoma largely depends on stage, which is strongly associated with histopathologic type. Approximately 75% of women with ovarian carcinoma have high-grade serous carcinoma, of which 90% present with stage III or IV disease, leading to poor survival.[1,2]

Ovarian cancer is rare; the lifetime risk of being diagnosed with ovarian cancer is 1.31%.[2]

The United Kingdom Collaborative Trial of Ovarian Cancer Screening (UKCTOCS)

TVU (or transvaginal sonogram [TVS]) has been proposed as a screening method for ovarian cancer because of its ability to reliably measure ovarian size and detect small masses.[10]

In the UKCTOCS (NCT00058032), outcomes among 50,623 postmenopausal women aged 50 to 74 years who were randomly assigned to 7 to 11 rounds of annual screening with TVU alone and 50,264 who underwent multimodal screening with CA-125 testing and TVU (see below) were compared with results among 101,299 women who were not screened, which served as a comparison group. Women were recruited in 13 trial centers across the United Kingdom from 2001 to 2005. After trial initiation, but before analysis, the protocol was amended twice: 1) the study was extended to achieve greater power and 2) criteria for referral in the multimodal arm were liberalized to increase the percentage of positive screens.[11]

TVUs were scored as normal, resulting in continued annual screening; intermediate, leading to repeat CA-125 and TVU at 3 months; or abnormal, requiring repeat testing within 6 weeks. In the TVU arm, 314 cancers were diagnosed and 154 ovarian cancer–related deaths occurred compared with the nonscreened arm, in which 630 cancers were diagnosed and 347 ovarian cancer–related deaths occurred. Mortality was nonsignificantly lower with TVU screening (11%; 95% CI, -7% to 27%; P = .21). TVU screening resulted in 50 surgeries per 10,000 women for a false-positive screen. Complications resulted from less than 1% of screens and 3.4% of operations. Over a median of 11.1 years, ovarian cancer deaths occurred among 0.30% of screened women and 0.34% of unscreened women.[11,12]

CA-125 serum testing for ovarian cancer screening

CA-125 is a tumor-associated antigen that is used clinically to monitor patients with epithelial ovarian carcinomas.[13,14] Measurement of CA-125 concentrations has been proposed as a potential marker for the early detection of ovarian cancer, either as a single test with a threshold cutpoint or in algorithms examining the change in levels over time. The two randomized trials have included CA-125 either in parallel or sequentially with TVU for multimodality screening. The most commonly reported CA-125 reference value that designates a positive screening test is 35 U/mL, and this was the reference value used in the Prostate, Lung, Colorectal, and Ovarian (PLCO) Screening Trial (NCT01696994) to define an abnormal test result. The measurement of CA-125 levels, in parallel combination with TVU,[15] is the ovarian screening intervention evaluated in the PLCO trial.[16,17] Elevated CA-125 levels are not specific to ovarian cancer and have been observed in patients with nongynecological cancers,[14] endometriosis,[18] liver disease, congestive heart failure, pleural or peritoneal fluid accumulation, and in the first trimester of pregnancy.[19,20] The sensitivity of the CA-125 test for the detection of ovarian cancer was estimated in two nested case-control studies using serum banks.[21,22] The sensitivity for CA-125 levels of at least 35 U/mL ranged from 20% to 57% for cases occurring within the first 3 years of follow-up; the specificity was 95%.

A phase II/III biomarker study was conducted to evaluate the sensitivity of several markers of ovarian cancer, including CA-125 concentrations, using specimens collected from ovarian cancer patients at four sites. The estimated sensitivity for early-stage disease (stage I and II ovarian cancer) was 56% (95% confidence interval [CI], 49%–72%) for a cutpoint set to obtain a fixed specificity of 95%. The threshold for the cutpoint for CA-125 at 95% specificity was 24 U/mL. For all cases (56% of cases had stage III or IV disease at diagnosis), the sensitivity was 73% (95% CI, 64%–84%). When the clinical cutpoint of 35 U/mL was used, the sensitivity decreased.[23]

The Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial: Single-threshold CA-125 levels and TVU

The PLCO trial evaluated the effect of screening on ovarian cancer mortality among women aged 55 to 74 years on the basis of six annual screens with serum CA-125 testing at a threshold for positive of 35 U/mL and four annual screens with TVU. Women were randomly assigned to screening (n = 39,105) or usual care (n = 39,111) at ten screening centers across the United States between November 1993 and July 2001. Participants and their health care practitioners received the screening test results and managed evaluation of abnormal results. The usual-care group was not offered screening with CA-125 or TVU but received their usual medical care. Participants were initially monitored for a maximum of 13 years (median, 12.4 years; range, 10.9–13.0 years) for cancer diagnoses and death until February 28, 2010. Mortality from ovarian cancer, including primary peritoneal and fallopian tube cancers, was the main outcome measure. Secondary outcomes included ovarian cancer incidence and complications associated with screening examinations and diagnostic procedures.[24] After transition of PLCO trial participants to a centralized follow-up process, follow-up for mortality was extended until the end of 2012, for a maximum of 19.2 years and a median of 14.7 years.[25]

Compliance with screening ranged from 85% at the initial round to 73% at the sixth round, while contamination in the usual-care group ranged from about 3.0% for CA-125 to 4.6% for TVU. Across the first four screening rounds, 11.1% of women had at least one positive test, 8.1% had at least one positive TVU, and 3.4% had at least one positive CA-125 test. The yields of both tests were similar. Ovarian cancer was diagnosed in 212 women (5.7 per 10,000 person-years) in the intervention group and 176 women (4.7 per 10,000 person-years) in the usual-care group (rate ratio, 1.21; 95% CI, 0.99–1.48). The stage distributions were similar by study group, with stage III and IV cancers comprising the majority of cases in both the intervention group (163 cases, 77%) and the usual-care group (137 cases, 78%). The cancer case treatment distributions were very similar between groups within each stage. Through the original period of follow-up (maximum, 13 years), 118 deaths were caused by ovarian cancer (3.1 per 10,000 person-years) in the intervention group and 100 deaths were caused by ovarian cancer (2.6 per 10,000 person-years) in the usual-care group (mortality rate ratio, 1.18; 95% CI, 0.82–1.71). Through the extended follow-up period (median, 14.7 years; maximum, 19.2 years), 187 deaths were caused by ovarian cancer (3.8 per 10,000 person-years) in the intervention group and 176 deaths were caused by ovarian cancer (3.6 per 10,000 person-years) in the usual-care group (mortality rate ratio, 1.06; 95% CI, 0.87–1.30).[25] Of the 3,285 women with false-positive results, 1,080 underwent surgical follow-up. Of the 1,080 women who underwent surgical follow-up, 163 women experienced at least one serious complication (15%). A total of 1,771 women in the intervention group (7.7%) and 1,304 in the usual-care group (5.8%) reported oophorectomy. There were 2,924 deaths resulting from other causes (excluding ovarian, colorectal, and lung cancer) (76.6 per 10,000 person-years) in the intervention group and 2,914 such deaths (76.2 per 10,000 person-years) in the usual-care group (rate ratio, 1.01; 95% CI, 0.96–1.06).[24,26]

In the PLCO trial, women with visualized ovaries were at slightly higher risk of ovarian cancer (hazard ratio, 1.42; 95% CI, 1.00–2.01) than were women who had nonvisualized ovaries.[27] A nested analysis found that ovarian volume increases were detectable 1 to 2 years before diagnosis, but the magnitude of change did not appear translatable to clinical management. Thus, among women in the general U.S. population, simultaneous screening with CA-125 and TVU did not reduce ovarian cancer mortality when compared with usual care.[24]

The Shizuoka Cohort Study of Ovarian Cancer Screening randomly assigned women to either a screening group (n = 41,668) or a control group (n = 40,799) between 1985 and 1999 at 212 hospitals in the Shizuoka prefecture of Japan. The screening protocol comprised ultrasound and CA-125 tests annually. Women with abnormal findings were referred to a gynecological oncologist. Ovarian cancer diagnoses were determined by record linkage to the Shizuoka Cancer Registry in 2002. The annual death certificate file in Shizuoka was checked to ascertain vital status. The mean follow-up time was 9.2 years, and the mean number of screens per woman was 5.4. There were 35 ovarian cancers detected in the screening group and 32 in the control group, with a nonstatistically significant difference in the stage distribution. Nine percent of regular screening attendees had at least one false-positive result.[28] Mortality results from this trial are not available.

Ovarian cancer screening using CA-125 analyzed according to the Risk of Ovarian Cancer Algorithm (ROCA) for a positive test result in combination with TVU

The UKCTOCS trial evaluated longitudinal CA-125 measurements using the ROCA, which defines a positive test as a statistically significant increase in a woman’s serial measurements based on the algorithm, irrespective of the absolute level. The goal of this approach is to detect cancers earlier by identifying subtle within-person changes. The UKCTOCS multimodal screen incorporated a two-stage approach with ROCA as the primary screen and TVS as a secondary screen (based on results of the ROCA) for its impact on ovarian cancer mortality compared with observation without screening. Of 50,078 women in the multimodality screening arm, 338 were diagnosed with ovarian cancer and 148 died of the disease. Multimodal screening detected 40% of ovarian cancers at stage I, II, or IIIA, whereas in the arm without screening, 26% were diagnosed at these stages (P < .0001). In this arm, less than 1% of screens led to complications and 3.1% of women developed surgical complications. Using multimodal screening, 14 per 10,000 screens led to surgery for a false-positive result. At the median follow-up of 11.1 years, there were 0.29% deaths with this screening method as compared with 0.34% in the observation arm. The primary endpoint of mortality reduction with screening based on Cox regression analysis indicated a statistically nonsignificant mortality reduction of 15% (95% CI, -3% to 30%; P = .10). The trial conducted a pre-specified analysis in which cases considered prevalent based on CA-125 levels at baseline were omitted. This subset analysis found a significant reduction in mortality of 20% (95% CI, -2% to 40%, P = .021), compared with observation.[11]

A post hoc nested study was conducted within the PLCO trial to determine if the use of ROCA could potentially improve the identification of early-stage (stage I and stage II) ovarian cancer.[29] The study evaluated the potential impact under two scenarios: best case and stage shift. Best-case scenario assumed that all cancers that would have been detected earlier with ROCA than with single-threshold CA-125 concentrations would have avoided mortality. The stage-shift scenario applied the observed PLCO early-stage survival rates to cases detected at an earlier stage with ROCA. The risk of death from ovarian cancer with ROCA was lower but estimates were not statistically significant (relative risk [RR] of 0.90 for best-case scenario [95% CI, 0.69–1.17] and RR of 0.95 for stage-shift scenario [95% CI, 0.74–1.23]).

Another retrospective study using annual CA-125 concentrations from the PLCO trial examined the potential impact of parametric empirical Bayes (PEB) longitudinal algorithm for the earlier detection of 44 incident ovarian cancers identified in the PLCO trial. Setting the specificity at 99%, PEB signaled “abnormal” CA-125 concentrations on average 10 months earlier than with the single-threshold cutpoint.[30] Whether this translated into a mortality benefit could not be determined.

CA-125 velocity has also been examined using a multiple logistic regression model within the PLCO trial as a predictor for the development of ovarian cancer.[31] Both CA-125 velocity and time intervals between screening tests were associated with the development of ovarian cancer. The risk of ovarian cancer increased as velocity (measured as U/mL per month) increased, and the risk of ovarian cancer decreased when the time intervals between screening tests increased.