Summary

Three trials evaluating four screening strategies showed no effect of screening on ovarian cancer mortality on the basis of a priori per protocol statistical analyses testing. The screening tests evaluated were annual transvaginal ultrasound, annual transvaginal ultrasound and CA-125 serum testing, annual CA-125 testing, and annual CA-125 serum testing interpreted with an algorithm (ROCA) that incorporates changes over time to inform triage and rescreening intervals.

Description of Included Studies

Three large, good-quality trials reported ovarian cancer incidence and mortality for women randomized to ovarian cancer screening versus no screening or usual care. The largest (n=202,546) and most recent trial is the UKCTOCS,⁹² which began enrolling trial participants in 2001 through 13 National Health Service centers in England, Wales, and Northern Ireland. Identification of women was centralized using primary care registers, and eligible women received personalized invitations. A smaller included trial (n=21,935) was conducted in the United Kingdom by the same research group in preparation for UKCTOCS (U.K. Pilot).⁹⁸ This trial recruited women who had participated in a previous ovarian cancer screening study.⁹⁹ The third included trial is the PLCO⁹¹ cancer screening randomized trial, begun in 1993 at 10 clinical screening centers in the United States (n = 68,557). The 10 screening centers developed individual recruitment methods. Women were primarily recruited via direct mail; however, other methods included community outreach and mass media.¹⁰⁰

The demographic characteristics of randomized trial participants are described in Table 2. In the UKCTOCS and U.K. Pilot trials, the race/ethnic composition of the study population was over 95 percent white.^{92, 98} Eighty-eight percent of women in the PLCO⁹¹ were white, non-Hispanic (6% were black, non-Hispanic, 4% were Asian or Pacific Islander, and very few were Hispanic or Native American). The PLCO⁹¹ and UKCTOCS⁹² had similar participant age distributions, with over half of UKCTOCS participants between the ages of 56 and 66 at baseline and 65% in the PLCO between the ages of 55 and 64. The U.K. trials^{92, 98} included post-menopausal women as young as 45 or 50, whereas the PLCO⁹¹ did not include any women younger than age 55. Nineteen percent of women in UKCTOCS⁹² and 27 percent of PLCO⁹¹ participants had had a prior hysterectomy at baseline (without bilateral oophorectomy). Although women with a known elevated familial ovarian cancer risk were excluded, 1.6 percent of women in UKCTOCS⁹² reported maternal history of ovarian cancer, and 6.4 percent a maternal history of breast cancer. 17 percent of women in the PLCO ⁹¹ reported any family history of breast or ovarian cancer.

Table 2

Baseline Characteristics of Ovarian Cancer Screening Trial Participants.

All three trials tested annual screening for ovarian cancer with CA-125 and/or transvaginal ultrasound (Table 3). The UKCTOCS⁹² had two intervention arms and a no-screening control arm (randomized 1:1:2, respectively). Women assigned to CA-125 risk algorithm screening received CA-125 serum testing, with triage and followup determined by application of the Risk of Ovarian Cancer Algorithm (ROCA). Three levels of followup, depending on the ROCA estimate were assigned: continue annual screening (low risk), repeat CA-125 in 12 weeks (intermediate risk), or repeat CA-125 testing and transvaginal ultrasound in 6 weeks (elevated risk). Women with persistent abnormalities were referred for surgical investigation at tertiary care health centers in the U.K. National Health Service system. Women in the ultrasound intervention arm had yearly transvaginal ultrasound testing, with repeat ultrasounds scheduled for unsatisfactory or abnormal results and referral to surgical diagnostic testing and recommended treatment through usual care in the U.K., as in the CA-125 screening arm. Women in the intervention arms of the UKCTOCS underwent a maximum of 11 screening rounds with a median of 11.1 years of follow up.⁹² The U.K. Pilot⁹⁸ trial compared annual CA-125 screening tests to no screening. Women with CA-125 serum levels of 30 U/mL or greater were considered screening test positive, and further evaluated using ultrasonography. Women in the intervention arm of the U.K. Pilot study underwent a maximum of 3 rounds of screening with up to 8 years of follow up.⁹⁸ Women in the screening intervention arm of the PLCO⁹¹ received both CA-125 testing and ultrasonography. The cutoff designating an abnormal result on the CA-125 test was serum level of 35 U/mL or greater. Women with abnormal results on either the CA-125 test or ultrasound were referred for additional evaluation by notifying the woman and her usual community physician.^{91, 101} The original trial design for PLCO called for screening for only four annual ovarian cancer screening rounds. A modification of the protocol in 1998 extended screening to 6 annual rounds, with the purpose of increasing the power of the trial. During this 2-round extension, only CA-125 was used for screening.¹⁰² Therefore, women in the intervention arm of PLCO underwent a maximum of 6 rounds of screening (4 with CA-125 and ultrasound, 2 with CA-125 alone) with a median of 12.4 years of follow up. Additionally, in the first 5 years of the PLCO, palpation of the ovaries with a bimanual pelvic examination was also included in the screening protocol. This component of the intervention was dropped, however, as no cancers were identified solely on the basis of this screening component.⁷⁰

Table 3

Ovarian Cancer Screening Trial Protocols.

Screening Adherence and Contamination

In the UKCTOCS⁹², 95 percent of women in the ultrasound arm completed at least one screen, and 99 percent of women in the CA-125 ROCA intervention arm completed at least one screen (Table 4). Overall 81 percent of screens were attended in the CA-125 ROCA intervention arm and 78 percent were attended in the ultrasound intervention arm. Adherence varied by round, ranging from 47 to 98 percent in the CA-125 ROCA group and 36 to 95 percent in the ultrasound group.⁹² In the U.K. Pilot⁹⁸ trial, 86 percent of randomized participants completed at least one screen and 71 percent completed all three screening rounds. Adherence in the PLCO⁹¹ was similar, with 78 to 84 percent of participants attending ultrasound screening and 73 to 85 percent of women attending CA-125 screening depending on the screening round. Rates of screening contamination in the control group were not reported in the U.K. Pilot⁹⁸ trial, but were below 5 percent in the UKCTOCS⁹² and PLCO⁹¹.

Table 4

Adherence to Screening in Ovarian Cancer Screening Trials.

Definition of Outcomes in the Trials

The following results from the PLCO⁹¹ and UKCTOCS⁹² include cases of ovarian, fallopian, and primary peritoneal cancer cases, the prioritized outcome for this review. Data from the U.K. Pilot trial did not capture information related to peritoneal cancer⁹⁸; therefore, these results are limited to primary cancer of the ovary and fallopian tubes.

Ovarian Cancer Incidence

The incidence of ovarian cancer did not differ significantly between study arms in any of the included trials. In the UKCTOCS⁹² ovarian cancer was diagnosed 0.7 percent of those in the CA-125 ROCA arm (354 cases), and 0.6 percent of those in the ultrasound arm (324 cases) and control arm (645 cases). The incidence rate of ovarian cancer was 6.4 per 10,000 person-years in the CA-125 ROCA screening arm, and 5.9 per 10,000 person-years in both the ultrasound and no screening control arms. In the PLCO,⁹¹ ovarian cancer was diagnosed in 0.6 percent (212 cases) of those in the intervention arm and 0.5 percent (176 cases) of those in the usual care arm (RR 1.21 95% CI, 0.99 to 1.48). In the U.K. Pilot trial, ovarian cancer was diagnosed in 0.1 percent (16 cases) of women in intervention arm and 0.2 percent (20 cases) in the no-screening arm. The low incidence may be due to smaller sample sizes, or the fact that women underwent a prevalence screen 10 years prior to the study. Incidence rates were not reported for the U.K. Pilot trial.⁹⁸

Ovarian Cancer Mortality

CA-125 Screening

The UKCTOCS⁹² and U.K. Pilot⁹⁸ trial included intervention arms that compared CA-125 screening to no screening (Table 5). Although both of the trials used an initial CA-125 screening test to determine followup, the U.K. Pilot study used a single cutpoint level to refer participants to further evaluation (CA-125 ≥ 30 u/ml), and the UKCTOCS used the ROCA algorithm¹⁰³ to triage intermediate and high CA-125 ROCA results for repeat screening and ultrasound. The ROCA CA-125 screening arm was described as multimodal screening in the UKCTOCS trial publications, and included a standard protocol for all followup screening tests. The ROCA algorithm is more complex than single cutpoint CA-125 testing approaches because it incorporates changes in CA-125 changes over time for individual women. In the UKCTOCS, ovarian cancer mortality (calculated to include peritoneal cancers) in the intervention arm and control arm was similar (IG: 0.32% versus CG: 0.35%) and in survival analysis was 2.9 ovarian cancer deaths per 10,000 person-years in the intervention group and 3.3 ovarian cancer deaths per 10,000 person-years in the control group. This difference was not statistically significant based on a Cox proportional hazards model (HR = 0.89 [95% CI, 0.74 to 1.08]).⁹²

Table 5

Benefits Reported in Ovarian Cancer Screening Trials: Ovarian Cancer Mortality.

The smaller U.K. Pilot⁹⁸ trial (n = 21,935) was designed to assess feasibility and performance of screening and was not powered to test mortality differences. There were 9 ovarian cancer deaths in the intervention group (0.08%) and 18 in the no-screening comparison group (0.16%); the difference was not statistically significant e (RR 0.50 [95% CI, 0.22 to 1.11]). A statistically significant difference in survival between women with index cancers in the IG and the CG was observed when computed from the date of randomization (IG median 72.9 months, CG median 41.8 months; p=.01). This finding was based on a small number of events, and only 6 of 16 index cancers identified in the intervention arm were screen detected. Survival in the control group noted by the study authors as being unusually poor.

Quality of Life

Two studies addressed changes in quality of life associated with ovarian cancer screening,^{97, 104}both with regard to potential anxiety associated with screening. These results are therefore reported for Key Question 2 as potential harms.

Ovarian Cancer Morbidity

There were no differences in treatment approach by study arm in the PLCO; 81 percent in the intervention group received surgery plus systemic therapy, compared with 80 percent in the usual care group.⁹¹

Summary

Four trials reported on the harms of ovarian cancer screening. False positive rates and surgical harms were highest for screening programs using transvaginal ultrasound as an initial test. Major surgical complications as estimated in the two largest trials occurred in women with investigations from screening that did not lead to a cancer diagnosis, ranging from 3 to 15 percent of surgeries. The screening tests themselves resulted in minor complications, at rates widely ranging based on study specific definitions from 0.86 to 58.3 per 10,000 screens/women for CA-125 test blood draws (e.g., fainting, bruising) and from 1.86 to 3.3 per 10,000 screens/women for ultrasound testing (e.g., pain, discomfort, infection, bruising).

Description of Included Studies

Evidence on false positive rates and surgical harms of screening were included from the three trials^{91, 92, 98} included for KQ1. A substudy of the UKCTOCS¹⁰⁴ (n=23,374) and an additional fair-quality trial (QUEST)⁹⁷ (n=549) aimed at evaluating the effects of ovarian cancer screening on quality of life (QoL) and psychological outcomes are described below.

False Positive Rates and Complications

CA-125 Screening

False positive rates calculated for screening with CA-125 across all rounds of the U.K. Pilot trial were 4.2 percent.⁹⁸ In the UKCTOCS the false positive rate in the prevalence round of screening was 9.0 percent⁶⁸ and cumulatively across all subsequent incidence rounds of the UKCTOCS the false positive rate rose to 44.2 percent with a total of 44.3 percent of women screened receiving a positive test result at least once⁶⁸ (Table 6). Screening test complications were minor for CA-125 screening, including bruising, fainting, and pain. These complications were reported to occur for 0.86 per 10,000 screens in UKCTOCS⁹² and 58.3 per 10,000 women in the PLCO.⁹¹

Table 6

Harms Reported in Ovarian Cancer Screening Trials: Positive Testing, False Positive Testing, and Surgical Complications in Ovarian Cancer Screening.

Table 7Harms Reported in Ovarian Cancer Screening Trials: Psychological Effects of Screening

View in own window

Trial, Author, Year of publication	UKCTOCS, 92 Barrett, 2014104	QUEST,97 Andersen, 200797
Quality	Good	Fair
Population	Random sample from UKCTOCS: CA-125 ROCA: 301 ultrasound: 283 CG: 755 Event sample*: CA-125 ROCA: 12,357 ultrasound: 9,678	All screened patients participating in Quality of Life, Education, and Screening Trial (QUEST): IG: 292, CG: 150
Measures	Spielberger State/Trait Anxiety Inventory, General Health Questionnaire 12	QoL: SF-36 Mental and Physical Health scores Distress: Impact of Events Scale Cancer worry: Modified Lerman cancer worry scale
Psychological effect of screening	Random sample: no evidence of difference in state anxiety between screening and control groups	QoL, Distress, Cancer Worry: No statistically significant differences between study arms
Psychological effects of positive test results or repeat screening tests	Random sample: No evidence of change in anxiety or psychological morbidity due to repeat screenings compared with annual screen. Event sample: Compared with a single repeat screen: evidence of higher anxiety for multiple repeat scans (p<0.010) (small absolute effect); greater odds of psychological morbidity (GHQ-12: score ≥4) with higher level referral screening: OR 1.28 (95% CI, 1.18 to 1.39)	Women with abnormal test results (n=32) compared with women with no abnormal results more likely to report cancer worry at 2 year followup (OR 2.8; 95% CI, 1.1 to 7.2)

*

All women in the screening arms recalled for repeat screening (excluding those in the random sample)

Abbreviations: CG = control group; CI = confidence interval; GHQ-12 = General Health Questionnaire 12; OR = odds ratio; QoL = quality of life; QUEST = Quality of life, Education, and Screening Trial; SF-36 = 36-Item Short Form Survey; UKCTOCS = U.K. Collaborative Trial of Ovarian Cancer Screening

Table 8Summary of Evidence

View in own window

Test	# studies (k), sample size (n) Design	Summary of Findings by Outcome	Consistency/Precision	Reporting Bias	Quality	Body of Evidence Limitations	EPC Assessment of Strength of Evidence for the KQ	Applicability
KQ1: Effectiveness
Annual screening with CA-125 testing	k=2 n=173,858 RCT	Ovarian cancer mortality (k=2, n=173,858). Screening with CA-125 did not result in improved ovarian cancer mortality compared with no screening (UKCTOCS92 HR 0.89, 95% CI, 0.74 to 1.08, U.K. Pilot98 RR=0.50, 95% CI, 0.22 to 1.11)	Reasonably consistent Reasonably precise	Undetected	Good	Followup not available beyond 10 years for a substantial proportion of UKCTOCS trial participants. Reporting bias undetected.	Moderate	Trial evidence from the U.K. where screening occurred in specialized trial settings and cancer treatment was provided through the National Health Service, which is a more centralized health system relative to the U.S. Study enrolled mostly white women UKCTOCS92 began in 2001 FDA does not support ROCA screening algorithm
Annual screening with transvaginal ultrasound examination	k=1 n=151,922 RCT	Ovarian cancer mortality (k=1, n=151,922). TVU screening did not result in improved ovarian cancer mortality compared with usual care (UKCTOCS92 HR 0.91, 95% CI, 0.76 to 1.09)	Consistency NA Reasonably precise	Undetected	Good	Followup data incomplete beyond 10 years for a substantial proportion of trial participants.	Moderate	Trial evidence from the U.K. where screening occurred in specialized trial settings and cancer treatment provided through the National Health Service, which is a more centralized health system relative to the U.S. Study enrolled few nonwhite participants
Annual screening with CA-125 testing and transvaginal ultrasound examination	k=1 n=68,557 RCT	Ovarian cancer mortality (k=1, n=68,557). No reduction was found in ovarian cancer mortality from combined TVU and CA-125 screening compared with usual care (PLCO91 RR 1.18, 95% CI, 0.82 to 1.71).	Consistency NA Reasonably precise	Undetected	Good	Changes to protocol, ovarian palpation dropped after first 4 trial years	Moderate	U.S. multisite trial with usual care control condition and referral to community clinicians for screen positives Majority white, non-Hispanic study participants Trial begun in 1993
KQ2: Harms
Annual screening with CA-125 testing	k=3 n=242,415 RCT	False positive rate from screening (k=2, n=173,858). False positive rate over multiple rounds of screening in the largest trial was 44%. Complications from screening (k=2, n=220,480). Complications from CA-125 testing were generally minor and ranged from 0.86 per 10,000 screens to 58.3 per 10,000 women. False positive surgery (k=2, n=173,858). False positive surgeries occurred in 0.2% to 1% of those screened with CA-125. Complications from false positive surgery (k=2, n=173,858). 1 larger trial (n=151,923) reported complications in 3.1% of false positive surgeries. 1 smaller trial (n= n=21,935) reported no surgical complications. Psychological effects of screening (k=1, n=13,413). Psychological harms reported in a subset of 1 trial. No statistically significant differences were found in psychological outcomes between the screening and no screening arms; increased psychological morbidity risk among women recalled for higher level screening.	Reasonably consistent or NA Reasonably precise	Undetected	Good	Psychological harms measured only for subsets of trial participants	Moderate (Low for psychological harms)	Trial evidence from the U.K. where screening occurred in specialized trial settings and cancer treatment provided through the National Health Service, which is a more centralized health system relative to the U.S.
Annual screening with transvaginal ultrasound examination	k=2 n =220,479 RCT	False positive rate and complications from screening (k=1, n=151,922). A false positive rate of 11.9% was reported in initial screening round. Complications from screening (k=2, n=220,479). Complications from screening with TVU ranged from 1.86 per 10,000 screens to 3.3 per 10,000 women. False positive surgery (k=1, n=151,922). False positive surgeries occurred in 3.2% of those screened with TVU. Complications from false positive surgery (k=1, n=151,922). Complications occurred in 3.5% of false positive surgeries. Psychological effects of screening (k=1, n=10,716). Psychological harms were reported in a subset of 1 trial. No statistically significant differences were found in psychological outcomes between the screening and no screening arms.	Reasonably consistent or NA Reasonably precise	Undetected	Good	Psychological harms measured only for subsets of trial participants Data on cumulative false positive rate not reported	Moderate (Low for psychological harms)	Screening conducted inspecialized trial centers Treatment for cancer (in all study arms) was through the centralized National Health Service system in U.K. and in community care settings in U.S.
Annual screening with CA-125 testing and transvaginal ultrasound examination	k=2 n=69,106 RCT	False positive rate and complications from screening (k=1, n=68,557). False positive screening rate of 9.8% over the course of the screening program. Complications from screening (see complication rates for individual components). False positive rate for screen positive surgery (k=1, n=68,557). False positive surgeries occurred in 3.17% of those screened. Complications from false positive surgery (k=1, n=68,557). Complications occurred in 15.09% of false positive surgeries. Psychological effects of screening (k=1, n=549). Women with abnormal test results (n=32) compared with women with no abnormal results more likely to report cancer worry at 2 year followup (OR 2.8; 95% CI, 1.1 to 7.2).	Consistency NA Reasonably precise (except psychological harms imprecise)	Undetected	Fair to Good	Psychological harms measured only for subsets of trial participants	Moderate (Low for psychological harms)	U.S. based, multisite trial Pragmatic trial with usual care control condition and referral to community clinicians for screen positives Majority white, non-Hispanic participants

Abbreviations: CA-125 = cancer antigen 125; CI = confidence interval; EPC = evidence-based practice center; FDA = U.S. Food and Drug Administration; HR = hazard ratio; KQ = key question; NA = not applicable; OR = odds ratio; PLCO = Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial; QUEST = Quality of life, Education, and Screening Trial; RCT = randomized controlled trial; RR = rate ratio; TVU = transvaginal ultrasound; UKCTOCS = U.K. Collaborative Trial of Ovarian Cancer Screening; U.S. = United States

False positive surgeries, defined as surgery following a positive screen among women found to have normal or benign adnexa, occurred in nearly 1 percent of those in the CA-125 ROCA arm of the UKCTOCS⁹² (n = 488) and 0.2 percent of those screened in the U.K. Pilot (n=23).⁹⁸ In the UKCTOCS, one or more major complication occurred in 3.1 percent of false positive surgeries in the CA-125 ROCA arm and included: infection, injury to hollow viscus, anesthetic complications, and cardiovascular and pulmonary events.⁹² No surgical complications were reported among the women undergoing surgical investigations in the U.K. Pilot trial.⁹⁸

Deaths From Other Causes

The UKCTOCS provided data on causes of death other than ovarian cancer, and the PLCO provided data on causes of death other than ovarian, colorectal, and lung cancers. There were no statistically significant differences in causes of death in screening arms compared with control arms.

Psychological Harms of Screening

A study of the psychological morbidity associated with ovarian cancer screening was undertaken within the UKCTOCS (n=23,374).¹⁰⁴ A random sample of women was drawn at baseline from each trial arm (n=1,339) and a survey was administered annually for 7 years to obtain variables for evaluating psychological effects of screening. Similar data were also collected for women in the screening arms who were at any time recalled for followup testing (CA-125 ROCA: n = 12,357; ultrasound: n = 9,678). For the event group, upon abnormal results and recall, questionnaires were administered, and thereafter on an annual basis.¹⁰⁴ The main measures were the Spielberger State/Trait Anxiety Inventory (STAI)¹⁰⁵ and the General Health Questionnaire 12 (GHQ-12)¹⁰⁶ for evaluating psychological morbidity. A small but statistically significant difference in the education levels of women between the ROCA screening and ultrasound group of the random samples was observed, and in the event sample more women reported hormone replacement therapy use than in the random sample (21% versus 14%, p<.001).¹⁰⁴ In light of these unexpected imbalances, the study authors recommended cautious interpretation of findings. In adjusted analyses with linear and logistic regression, no statistically significant differences in mean STAI or the risk of psychological morbidity (GHQ-12 ≥ 4) were observed between the control and intervention arms in the random sample. In the analysis of women with recall screening events, there was a statistically significant increased risk of psychological morbidity among women recalled for higher-level screening (adjusted OR 1.28 [95% CI, 1.18 to 1.39]).¹⁰⁴

The QUEST⁹⁷ trial analyzed 549 average-risk women age 30 years or older in the United States to examine the effect of ovarian cancer screening on cancer worry and QoL. Women were randomized to ovarian cancer screening, risk counseling, or a screening/risk counseling combination compared to a usual care-only protocol consisting of annual pelvic examination and routine education by a woman's primary care physician. We report only on results from the screening and usual care control arms of the trial (n = 442). Ovarian cancer screening consisted of alternating CA-125 measurement and TVU every 6 months for a maximum of 4 screening rounds. Overall, women had a high level of education, with 99 percent completing high school and 95 percent attending college. The QUEST⁹⁷ study found no statistically significant differences for QoL measured with the SF-36, distress measured with the Impact of Events Scale, or cancer worry measured with a modified Lerman cancer worry scale. The only significant effect observed in participants was a higher level of cancer worry after 2 years among those who had experienced any abnormal test results.⁹⁷