OBJECTIVE

To assess the performance of ultrasonography in a multimodal ovarian cancer screening strategy.

DESIGN

Prospective ovarian cancer screening trial between December 1986 and June 1993.

SETTING

General practice, occupational health departments and an ovarian cancer screening clinic at a London teaching hospital.

POPULATION

Postmenopausal women, > or = 45 years with a raised CA125.

METHODS

Volunteers with a CA125 > or = 30 U/mL underwent a pelvic ultrasound. Scans were classified as normal, abnormal (ovarian volume > or = 8.8 mL) or equivocal (normal volume with abnormal morphology). Abnormal ovarian morphology was subclassified as simple cyst (single, thin walled cyst with no septa or papillary projections) or complex (all other abnormalities). Volunteers with abnormal scans were referred for a gynaecological opinion. Follow up was via the cancer registry and postal questionnaires.

MAIN OUTCOME MEASURES

Sensitivity, specificity and positive predictive value of different ultrasound criteria for detection of index cancer (e.g. primary invasive epithelial carcinoma of the ovary and fallopian tube).

RESULTS

Seven hundred and forty-one women underwent 1,219 scans and 20 index cancers occurred during a median follow up of 6 x 8 years. The sensitivity for detection of ovarian cancer of different ultrasound criteria was 100% for abnormal morphology, 89 x 5% for abnormal volume and 84% for complex morphology. The highest specificity (97%) and positive predictive value (37 x 2%) was achieved using complex morphology.

CONCLUSION

A variety of ultrasound criteria can achieve high sensitivity, specificity and positive predictive value for index cancers in postmenopausal women with an elevated CA125. Use of ovarian morphology to interpret ultrasound may increase sensitivity and use of complex ovarian morphology may increase the positive predictive value.