Currently both the American College of Obstetricians and Gynecologists and the Society for Gynecologic Oncology state that the best method to diagnose ovarian cancer is for clinicians to have a high index of suspicion in a symptomatic patient.1,2 At this time, no organization endorses screening for ovarian cancer using currently available tests because screening has not been shown to reduce morbidity or mortality.1–3 Many mainstream media organizations encourage women to keep symptom diaries and to report symptoms to physicians; some physicians may solicit symptom reports from patients; however, the safety of this practice has not been evaluated. Concerns have been raised that a focus on symptoms may result in high rates of unnecessary surgery with secondary complications. This study sought to examine the extent to which this is likely to be a problem.
Previous case–control studies by our group have identified a symptom index algorithm, which is associated with ovarian cancer (sensitivity 87.0%, specificity 87.9%).4–8 The symptom index is positive when a patient has any one of six symptoms (bloating, increased abdominal size, abdominal pain, pelvic pain, difficulty eating, feeling full quickly) occurring more than 12 times a month and present for less than 1 year.
Our group previously reported a pilot study of symptom-triggered screening to assess feasibility as well as patient and health care provider acceptability.9 The goal of our current study was to evaluate the potential harms (unnecessary surgery and testing) and ovarian cancer outcomes associated with proactively assessing symptoms with a symptom index followed up with routine referral for diagnostic testing.
PATIENTS AND METHODS
This prospective cohort study was approved by the institutional review boards at the University of Washington and Fred Hutchinson Cancer Research Center. Between April 2008 and July 2011, women attending one of the three primary care clinics at the University of Washington were approached by a research nurse to determine eligibility and interest in participating in a research trial to evaluate symptom-triggered diagnostic evaluation for ovarian cancer. The participating clinics are the three main primary care clinics for the University of Washington serving the local community and are located in the same building as radiology, ultrasonography, and phlebotomy services. An initial 1,001 women were enrolled as an unscreened pilot group.10 In our initial pilot study, 80% of eligible women agreed to participate in the study.10 From September 2008 to July 2011, 5,012 women participated in symptom index assessment with referral for CA 125 testing and transvaginal ultrasonography when the symptom index was positive.
Women were eligible to participate if they were 40 years of age or older, had at least one ovary, were not pregnant, and were able to give informed consent. Those interested in participating signed an informed consent form and filled out a baseline health questionnaire and a symptom index assessment form. The consent form allowed for review of medical records for 12 months after participation to identify any studies or procedures performed as a result of study participation. Women then had their regularly scheduled appointment with their health care provider. Physicians were informed of participation in the study but were instructed not to alter their clinical management and to order any tests that the patient's symptoms or examination warranted. At the time of the clinic visit, the research nurse scored the symptom index as positive or negative. Those with a positive symptom index were offered CA 125 testing and transvaginal ultrasonography as diagnostic tests ordered as part of the study and any subsequent tests or procedures ordered by their physicians were recorded and coded as related to the study. Patients and health care providers were informed of the results of the study-ordered transvaginal ultrasound and CA 125 testing. Women with abnormal results on these tests were managed according to standard of care by their physicians consistent with earlier studies.9 CA 125 results were considered normal if less than 35 units/mL. All transvaginal ultrasound scans were performed in the radiology department at the clinic by a trained ultrasonographer and radiologist as they would be for routine clinical care. Radiologists knew patients had a positive symptom index and so there was special focus on the ovaries, but no specific protocol was used for scans nor was Doppler, because we wanted the intervention to mimic standard of care. Transvaginal ultrasonography was considered normal if the uterus, adnexa, and ovaries were normal with no ascites or other abnormalities. This conservative definition of normal meant that transvaginal ultrasound scans were considered abnormal even if problems were identified on the scan that were clearly not ovarian, and these ultrasound findings led to appropriate treatment for nonovarian conditions that were causing symptoms.
Approximately 1 year (June 2012) after study completion, all participant names (positive and negative symptom index) were linked to the Western Washington Surveillance, Epidemiology, and End Results database to assess if any of the participants developed ovarian cancer as had been done in our pilot study.9 Sufficient time was allowed to be certain that all ovarian cancer cases were captured. Use of the Surveillance, Epidemiology, and End Results database assured us access to a comprehensive listing of patients with cancer, thus providing data on women who were diagnosed with cancer after completing a symptom index, including women who received a negative result and did not have further interactions with study staff.
The STATA 10.0 statistical software package was used for analyses. Descriptive statistics characterizing the study population included calculation of the median and range for continuous variables and the frequency and percentiles for categorical variables. Associations between patient characteristics and the results of the symptom index were assessed using the χ2 and Fisher's exact test. All statistical tests were two-sided and considered to be statistically significant at P≤.05.
A total of 6,013 women were prospectively enrolled in the study (Fig. 1) The first 1,001 were part of a feasibility pilot and used to refine the symptom index instrument.10 These women participated in the earliest pilot activities and filled out the symptom index and baseline health questionnaires but were not offered transvaginal ultrasound or CA 125 testing. From September 2008 to July 2009, 1,261 women participated in symptom-triggered diagnostic evaluation, with those being symptom index–positive being offered transvaginal ultrasound and CA 125 testing.7 The focus of this portion of the study was to establish patient and health care provider acceptability as well as rates of positive symptom index and subsequent testing. From July 2009 to July 2011, an additional 3,751 women were enrolled in the protocol to assess the consequences of proactive assessment of symptoms. All 6,013 women who filled out a symptom index had their cancer outcomes assessed in Surveillance, Epidemiology, and End Results are reported here.
In the 5,012 women who participated in the symptom-triggered diagnostic evaluation, 241 (4.8%) were symptom index–positive. Table 1 shows the demographics of the women who participated in the study. Of the 241 women who screened positive on the symptom index, 211 (88%) participated in additional testing with transvaginal ultrasonography and CA 125 (12% declined).
Tables 2 and 3 show the results of the transvaginal ultrasound and CA 125 tests performed in women with a positive symptom index. Of the 211 transvaginal ultrasound tests ordered, 80 (38%) were also ordered as part of routine clinical care. Most (70%) were ordered because of pain with or without postmenopausal bleeding and the rest resulting from postmenopausal bleeding alone or suspicious examination. There were 24 indeterminate transvaginal ultrasound scans (13 thickened endometrial stripes11 and 11 adnexal masses) and six scans that were reported as suspicious for cancer. There were six CA 125 tests that were elevated above 35 units/mL; none was highly elevated.
Although some of the follow-up procedures done on study participants were not based on concern or suspicion of ovarian cancer (eg, the thickened endometrial stripes with postmenopausal bleeding), all subsequent procedures are shown in Table 4. In total, there were 20 laparoscopies, laparotomies, or vaginal hysterectomies performed in study participants within 6 months of a positive symptom index (0.4% of participating women). However, only six (0.12%) were done for a suspicious ovarian mass. Overall, among study participants with positive symptom index results, there was a false-positive rate resulting in a major surgical procedure of 0.4% (20/5,012) or one in 250 women (95% confidence interval [CI] 1/162–1/410). Because 14 of these surgeries were performed for clearly nonovarian conditions in the absence of a transvaginal ultrasound or CA 125 result deemed suspicious for ovarian cancer, the rate for surgical procedures associated with a positive follow-up test result in a symptom index–positive woman was 0.012% (6/5,012) or one woman out of 835 (95% CI 1/418–1/2,005).
Ovarian cancer outcomes were assessed for all 6,013 women through the Western Washington Surveillance, Epidemiology, and End Results Cancer Registry. Because studies have shown that most patients with ovarian cancer do not recognize symptoms until less than 1 year and more reliably 6 months before diagnosis, we were primarily interested in women who developed ovarian cancer within 6 months from participating in a symptom index. Table 5 shows the ovarian cancer outcomes for the study. There were two women who developed ovarian cancer within 6 months of completion of the symptom index. One was positive and was diagnosed 31 days later with distant disease (part of the 1,001 unscreened pilot group) and one who was symptom index-negative was diagnosed with local disease. Of note, this symptom index-negative patient had a family history of ovarian cancer and was being worked up as part of routine care for a pelvic mass at the time of her study participation. The other six patients with ovarian cancer were diagnosed 281–843 days (median 569 days) after participation in the study. Half of these patients (3/6) were diagnosed with regional disease.
Also of potential interest, 80 (38%) of the transvaginal ultrasound scans and 26 (12%) of the CA 125 tests done in the symptom index–positive women also were ordered by clinic physicians as part of routine clinical care based on their personal clinical judgment separate from our study activities. Although there were 20 major surgical procedures performed, only three laparoscopies and one laparotomy were performed solely as a result of participation in the study. Of the four procedures, one dermoid, two endometriomas, and one atypical leiomyoma were diagnosed. One patient who underwent a laparoscopy developed a postoperative wound infection. Overall, this corresponds to a false-positive rate for major surgery of 0.08% (4/5,012), or 1 of every 1,253 women evaluated (95% CI 1/489–1/4,599).
Currently there are no organizations that recommend screening for ovarian cancer in the general population.1–3 The U.S. Preventive Services Task Force reissued their grade D recommendation in 2012 based largely on the results from the Prostate Lung Colorectal and Ovarian cancer screening trial, which was reported in 2011.12 A second large screening trial from the United Kingdom has also shown that using transvaginal ultrasonography as part of the initial screen results in a significant false-positive rate and high rates of probable unindicated surgeries,13 reinforcing that transvaginal ultrasonography should be used as a diagnostic test but not as a screening test.
In our study population, symptom-triggered diagnostic evaluation resulted in 0.4% (20/5,012) of women undergoing surgery, 80% (16/20) of these were done for clinical indications, and only 4 of 20 (20%) were performed solely as a result of participation in the study. This rate of false-positive surgery 0.08% (1 of every 1,253 women evaluated) is important because concerns have been raised that monitoring for symptoms associated with ovarian cancer could lead to high rates of unindicated procedures. Although there is a low rate of potentially unindicated surgery, the possibility that symptom screening could lead to a 20% (4/20) increase in surgery must be considered a potentially negative consequence. However, although these were false-positives for ovarian cancer, each of these patients did have pelvic pathology and the surgery may have prevented future problems.
The efficacy of a symptom-triggered approach cannot be adequately assessed in this study because of the low number of ovarian cancers. There were only two cases of ovarian cancer diagnosed within 6 months of symptom index completion. This is the time window when most nonrecall-biased studies show that patients begin to experience and report symptoms.14–17 Only one of the patients diagnosed within 6 months had a positive symptom index and she had advanced-stage disease; therefore, no patients were diagnosed with early-stage disease within 6 months. The other six patients were diagnosed beyond 6 months. If we had used 12 months as a cutoff, only one of four (25%) would have been identified through symptoms. Fortunately, only 4.8% of participants were symptom index–positive; therefore, if universal symptoms screening was found to be an effective method to identify women who should receive diagnostic evaluations for ovarian cancer, more than 95% of women would not require any testing. However, if symptoms were to be used as the only means of screening for ovarian cancer, the sensitivity of a symptom-based screening program would be low and would need to be improved, possibly through more frequent screening or the addition of biomarkers to a multimodal strategy.
Long-term follow-up of study participants identified six women who developed ovarian cancer more than 6 months after their participation in the study. Of these six women, three had their ovarian cancers identified at the regional stage. Although it is impossible to determine the details leading to this better than would be expected stage distribution in women who had previously participated in our study, and it is possible that with such a small sample this result is coincidental, it may be possible that study participation provided these women with valuable education about ovarian cancer symptoms spurring them to seek evaluation of subsequent symptoms that were associated with the cancers they eventually developed. Such a result would be consistent with the findings of Gilbert et al,14 who found that prompt aggressive evaluation of approximately 1,500 women with symptoms resulted in an ovarian cancer prevalence 10 times greater than the general population and an improved prognosis in a majority of cases.
In the Diagnosing Ovarian Cancer Early study, women were recruited through radio, newsprint, and flyers.18 Women who met symptom entry inclusion criteria were offered CA 125 and transvaginal ultrasound testing within 2 weeks. If the CA 125 result was indeterminant, it was repeated in 4 weeks. If it was normal and women persisted in having symptoms, it was repeated in 4 months. Of the 1,455 women enrolled, 11 or 1/132 women had ovarian cancer. Of the 11 cases diagnosed, four (36%) were early-stage disease, four (36%) were small-volume stage III disease that was optimally cytoreduced, and three (27%) were advanced suboptimally cytoreduced disease. Comparison to the general clinic population in Montreal found that 56% of the ovarian cancer population was advanced suboptimally cytoreduced at the time of diagnosis. Also, in the study population, 11 endometrial cancers were diagnosed as a result of a thickened endometrial stripe11 and postmenopausal bleeding; therefore, 1 of every 66 symptomatic women in this study was diagnosed with a gynecologic malignancy. Both the Gilbert et al and our study emphasize the need to further study the prompt response to symptoms as a low-cost method to triage women for diagnostic tests.
Limitations to our study include the low number of cancers, which limits the ability to evaluate sensitivity, specificity, and the potential for early diagnosis. However, with 5,000 screens, we only expected two to three cancers to be diagnosed within 6 months of a positive symptom index. Other limitations include the fact that ultrasonographers knew patients were in a study and may have done a more thorough evaluation of adnexa. Health care providers were also aware of patient participation, which may reduce the external validity of the study. In a different setting, physician and patient concerns could drive the rate of surgical intervention higher. In addition, we did not have a control group who did not participate in a symptom-triggered evaluation so that surgical interventions could be compared.
Overall, we found that proactive symptom-triggered diagnostic evaluation would increase rates of potentially unindicated surgery by one patient in 1,253 evaluated. We found that 50% of the ovarian cancers detected in our study population were diagnosed in local or regional stages. Although our numbers are small, based on Surveillance, Epidemiology, and End Results statistics, only 20–30% are expected to be found in this more favorable stage category. Although the actual symptom-triggered evaluation process may not have led to earlier diagnosis, it is speculated that some of these cancers may have been diagnosed early because women had increased awareness of symptoms, which may have prompted them to seek care earlier resulting in identification of ovarian cancer at an earlier stage. The real value of a symptom index may lie in its ability to act indirectly as an educational tool.
1. The role of the obstetrician-gynecologist in the early detection of the epithelial ovarian cancer. Committee Opinion No. 477. American College of Obstetricians and Gynecologists. Obstet Gynecol 2011;117:742–6.
2. Schorge JO, Modesitt SWC, Coleman RL, Cohn DE, Kauff ND, Duska LR, et al.. SGO White Paper on ovarian cancer: etiology, screening and surveillance. Gynecol Oncol 2010;119:7–17.
3. U.S. Preventive Services Task Force Agency for Healthcare Research and Quality. Screening for ovarian cancer, topic page. Rockville (MD): U.S. Department of Health and Human Services; 2004.
4. Goff BA, Mandel L, Melancon CH, Muntz HG. Ovarian carcinoma diagnosis. Cancer 2000;89:2068–75.
5. Goff BA, Mandel LS, Melancon CH, Muntz HG. Frequency of symptoms of ovarian cancer in women presenting to primary care clinics. JAMA 2004;291:2705–12.
6. Goff BA, Mandel LS, Drescher SW, Urban N, Gough S, Schurman KM, et al.. Development of an ovarian cancer symptom index: possibilities for earlier detection. Cancer 2007;109:221–7.
7. Andersen MR, Goff BA, Lowe KA, Scholler N, Bergan L, Dresher CW, et al.. Combining a symptoms index with CA125 to improve detection of ovarian cancer. Cancer 2008;113:484–9.
8. Andersen MR, Goff BA, Lowe KA, Scholler N, Bergan L, Drescher CW, et al.. Use of a symptom index, CA1215, and HE4 to predict ovarian cancer. Gynecol Oncol 2010;116:378–83.
9. Goff A, Lowe KA, Kane JC, Robertson MD, Gaul MA, Andersen MR. Symptom triggered screening for ovarian cancer: a pilot study of feasibility and acceptability. Gynecol Oncol 2012;124:230–5.
10. Andersen RM, Goff B, Lowe K. Development of an instrument to identify symptoms potentially indicative of ovarian cancer in a primary care clinic setting. Open J Obstet Gynecol 2012;2:183–91.
11. The role of transvaginal ultrasonography in the evaluation of postmenopausal bleeding. ACOG Committee Opinion No. 426. American College of Obstetricians and Gynecologists. Obstet Gynecol 2009;113:462–4.
12. Buys SS, Partridge E, Black A, Johnson CC, Lamerato L, Isaacs C, et al.. Effect of screening on ovarian cancer mortality: the Prostate, Lung, Colorectal and Ovarian (PLCO) cancer screening randomized controlled trial. JAMA 2011;305:2295–303.
13. Menon U, Gentry-Maharaj A, Hallett R, Ryan A, Burnell M, Sharma A, et al.. Sensitivity and specificity of multimodal and ultrasound screening for ovarian cancer, and stage distribution of detected cancers: results of the prevalence screen of the UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS). Lancet Oncol 2009;10:327–40.
14. Hamilton W, Peters TJ, Bankhead C, Sharp D. Risk of ovarian cancer in women with symptoms in primary care: population based case-control study. BMJ 2009;339:b2998.
15. Smith LH, Morris CR, Yasmeen S, Parikh-Patrl A, Cress RD, Romano PS. Ovarian cancer: can we make the clinical diagnosis earlier? Cancer 2005;104:1398–407.
16. Freidman GD, Skilling JS, Udalstove NV, Smith LH. Early symptoms of ovarian cancer: a case-control study without recall bias. Fam Pract 2005;22:548–53.
17. Rossing MA, Wicklund KG, Cushing-Haugen KL, Weiss NS. Predictive value of symptoms for early detection of ovarian cancer. J Natl Cancer Inst 2010;102:222–9.
18. Gilbert L, Basso O, Sampalis J, Karp I, Martins C, Feng J, et al.. Assessment of symptomatic women for early diagnosis of ovarian cancer: results from the prospective DOvE pilot project. Lancet Oncol 2012;13:285–91.