Collections > Electronic Theses and Dissertations > Evaluation of hysterectomy as both exposure and outcome: Association with epithelial ovarian cancer and prediction of premenopausal hysterectomy with ovarian conservation
pdf

Hysterectomy is the most common non-obstetric, surgical procedure among women in the United States. Older studies have generally reported that hysterectomy is inversely associated with risk of epithelial ovarian cancer (EOC). One of the goals of this dissertation was to investigate this relationship, focusing on subtypes of EOC. Since a majority of ovarian-conserving hysterectomies occur prior to menopause, we also sought to develop a predictive model for premenopausal hysterectomy with ovarian conservation as a way of identifying traits that may help identify women for clinical monitoring and potentially earlier and less invasive treatment than hysterectomy for their conditions. We estimated study-specific odds ratios by conditional logistic regression for almost 30,000 EOC cases and controls from 15 studies in the Ovarian Cancer Association Consortium and pooled them using random-effects models. Hysterectomy was associated with an average relative increase of 19% in the odds for invasive EOC. The average odds ratios were above the null for all subtypes of EOC, except invasive clear cell. Hysterectomies prior to age 40 or 30 years or more in the past were also positively associated with invasive serous EOC. We also fit unconditional logistic regression models including fourteen demographic, behavioral, and reproductive and medical history variables to nearly one thousand premenopausal women with and without hysterectomy enrolled in the Prospective Research on Ovarian Function Study. The predictive model exhibited very good discriminatory power and was well-calibrated. Family history of benign gynecologic conditions did not contribute to the prediction model and in fact, resulted in a net worsening of classification of hysterectomy. Our results, from the largest set of EOC cases and controls to date, do not support the long-held belief that hysterectomy is protective against EOC. Considerable heterogeneity of results, potential biases in previous and/or current studies, and/or a changing association between hysterectomy and EOC may explain our results and should be explored further. Though we were able to predict premenopausal hysterectomy with ovarian conservation with very good discriminatory power, additional model development should be pursued in larger and more diverse populations of premenopausal women as we try to better understand premenopausal women who undergo hysterectomy.