Statistical models for fertility-related issues in adjuvant treatment for breast cancer.

摘要

The statistical modeling of menstrual status of premenopausal patients undergoing adjuvant treatment for breast cancer has become increasingly important in recent years. Some adjuvant therapies cause amenorrhea, and it is of interest to describe the process by which menses discontinue and resume after treatment is completed. The process is complicated by the fact that natural menopause also occurs in the patient population, and that treatment-induced amenorrhea is not distinguishable from menopause unless menses resume after treatment end. In addition, adjuvant treatments are effective therapies for improving disease-free survival (DFS), but studies have not consistently shown that patients who experience ovarian suppression have higher DFS than those who do not. This thesis proposes three models for analyzing these data, and we illustrate each method by application to data arising from a clinical trial conducted by the International Breast Cancer Study Group (IBCSG).; The first study discusses a parametric model for the times to amenorrhea and to the recovery of menses, accounting for the possibility that treatment causes an anticipation of natural menopause. The model incorporates parametric assumptions on the distribution of the ages at entry and of natural menopause in the potential patient population before the eligibility requirement that the patients be pre-menopausal is enforced.; The second study proposes a method for estimating the failure time distribution of a masked event time that is also subject to a cure rate. Failure time data may consist of the observation of an event whose cause is unknown due to the censoring or lack of a second event that could identify the cause of the first event. Standard competing risks methodology does not apply to this setting because the cause of the event is not always identifiable. Assuming a proportional hazards regression structure for the latency, we estimate the parameters of the cure model via the EM algorithm.; The third study examines a joint cure model for disease-free survival and for the times to cessation and recovery of menses in the population of interest. Parameters are estimated using a Monte Carlo EM algorithm.