• Division of Cancer Epidemiology and Genetics (CB, RMP, AB, MES) and Division of Cancer Prevention (MES), National Cancer Institute, Rockville, MD; Huntsman Cancer Institute, University of Utah Health Sciences Center, Salt Lake City, UT (SSB). clara.bodelon@nih.gov.
• Division of Cancer Epidemiology and Genetics (CB, RMP, AB, MES) and Division of Cancer Prevention (MES), National Cancer Institute, Rockville, MD; Huntsman Cancer Institute, University of Utah Health Sciences Center, Salt Lake City, UT (SSB).

BACKGROUND: Accumulating evidence suggests that many ovarian cancers represent metastases from occult fallopian tube carcinomas or tumors arising within ovarian endometriosis. We hypothesized that small ovarian volumes, as reflected in nonvisualization by transvaginal ultrasound (TVU), would be a marker of lower ovarian cancer risk, whereas enlargement on serial examinations would indicate higher risk.

METHODS: To address these hypotheses, we analyzed serial ovarian volume measurements determined by TVU for 29321 women (102787 scans) performed in the ovarian cancer screening arm of the Prostate, Lung, Colorectal, and Ovarian (PLCO) trial. Cox models were used to compute hazard ratios (HRs) and 95% confidence intervals (CIs) as measures of association between TVU results and ovarian cancer. We assessed whether increasing ovarian volume preceded diagnosis of ovarian cancer in a nested case-control analysis comparing case patients and control patients matched on age, center, and screening year (1:4 ratio). All statistical tests were two-sided.

RESULTS: Visualization of normal-appearing ovaries at the last TVU scan was associated with marginally higher ovarian cancer risk compared with nonvisualization of the ovaries (HR = 1.42, 95% CI = 1.00 to 2.01). Ovarian volume increased statistically significantly in ovarian cancer case patients one to two years before diagnosis (P < .001), but not in matched control patients.

CONCLUSION: Our analysis of TVU data suggests that increasing ovarian volume is associated with greater ovarian cancer risk, but it is only detectable one to two years before diagnosis.