Objectives: To assess the use of transvaginal sonography as a screening test for familial ovarian cancer and, secondarily, to determine the value of a family history of malignant disease and the potential role of serum CA 125 levels in the screening procedure.
Methods: Two thousand five hundred self-referred women were studied at the Ovarian Screening Clinic at King's College Hospital, London. These symptom-free women with at least one close relative who had developed ovarian cancer were studied prospectively. Each woman was scanned for the presence of a persistent ovarian lesion and a sample of peripheral blood was taken for the retrospective analysis of serum CA 125. Women with a positive screen result were referred for surgical investigations; those with a negative result but considered to be at high risk were rescreened. The main outcome measures were findings at surgery, the histological classification of ovarian lesions, and cancer reported at follow-up.
Results: The women were aged 17 to 78 (mean, 48) years; 65% were premenopausal, 26% were postmenopausal and 9% had undergone hysterectomy. Seven hundred and thirty-eight women (29.5%) had a family history of multiple site cancers and 279 (11.2%) reported cancer specific to the ovary. There were 4231 screenings (2500 first screens, 998 second screens and 733 third or higher order screens). One hundred and four screens gave a positive result (2.5%); 11 cancers were detected (seven (64%) at stage I, four of which were of borderline malignancy). One additional cancer was reported within 12 months of the last scan and classified as a false-negative screen result. Eight cancers (including two peritoneal) were reported at follow-up (> 1-9 years after the last scan). All these women presented at an advanced stage (stage III). Fifteen of 20 cancers occurred in premenopausal women. The overall sensitivity of ultrasound screening was 92% (95% confidence interval, 76-100); the specificity was 97.8%. The prior odds of any woman having a screen-detectable ovarian malignancy during the study period were 1 : 207. The posterior odds subsequent to a positive screen result were 1 : 8.5 (1 : 12.7 at Screen 1; 1 : 3.7 at Screen 2; 1 : 3.0 at subsequent screens); the value was 1 : 11.4 for women with one family relative with ovarian cancer and 1 : 5.0 for women with the site-specific ovarian cancer syndrome. The prior use of a decision level for serum CA 125 >or= 20 U/mL would have reduced the number of women undergoing sonography by 78%; seven of the 12 cancers (58%) would have been detected (63% of all stage I disease, 75% of invasive stage I disease). An alternative cut-off value of 35 U/mL would have resulted in a detection rate of 33%.
Conclusions: Transvaginal sonography can effectively detect intraovarian cancer and tumors of borderline malignancy in women with a family history of the disease. Screening efficacy is related to the type of family history. The level of serum CA 125 can be used to select women for sonography, but the detection rate for early cancers would be reduced.
Copyright 2003 ISUOG. Published by John Wiley & Sons, Ltd.