Screening for Ovarian Cancer: A Pilot Randomised Controlled Trial

Abstract

Twenty-two thousand women in the United Kingdom were asked to participate in an ovarian screening trial in which they were randomly assigned to undergo three annual screenings or to be followed for 3 years without screening. Screening consisted of serum cancer antigen 125 (CA-125) measurement with subsequent pelvic ultrasound for CA-125 levels above 30 U/ml and surgical investigation if ultrasound findings were positive. There were 21,935 participants; 10,958 were randomly assigned to the screening group, and 10,977 served as control subjects and were followed for 3 years with no screening. When the results were equivocal, the women were followed up with repeated scans every 3 months for 1 year. Seven years after the last screening, a follow-up survey was performed; the survey included a search of the National Health Service Resister for index cancer cases and a mail questionnaire. In the screening group, 7743 women participated in all three screenings, and 9364 underwent at least one screening in the 3 years; 8732 women underwent the first screening, 8672 underwent the second screening, and 8455 were screened during the third year. Elevated CA-125 levels were found in 254 women at the first screening, 153 at the second screening, and 325 at the third screening (a total of 732 scans performed on 468 women). Results of subsequent pelvic ultrasound were abnormal and surgery was performed in 15 women after the first screening, in three after the second screening, and in 11 after the third screening. In all, 6 of the 29 women who underwent surgery were diagnosed with cancer of the ovary or fallopian tube (index cancer). Of the remaining 23 women with a false-positive result, 14 were found to have benign ovarian tumors, 4 had fibroids, 2 had an adenocarcinoma of unknown origin, and no abnormality was found in 3. There were no serious complications in the women undergoing surgery. By the end of the 7-year follow-up period, 20 women in the control group and 10 women in the screening group had developed an index cancer. The median length of survival was 73 months for the screening group and 42 months for the control group. Eighteen of the 10,977 women in the control group died of an index cancer, compared with 9 of the 10,958 women in the screening group, but this difference was not statistically significant (P = .083). There was also no significant difference between the two groups in the stage of disease at diagnosis, although more stage I and II index cancers were seen in the screening group than in the control group (31 vs. 10 percent, respectively; P = .171). Women in the screened group generally had a lower grade of disease, but this difference was also not significant. Both groups of women were similar in all clinical and demographic characteristics. The follow-up questionnaire was returned by 90 to 91 percent of women in both the control and screening groups, and documentation of death and survival was equal for both groups. Lancet 1999;353:1207–1210