Brown University Center for Statistical Sciences Seminar
Abstract: Early detection of ovarian cancer is an appealing approach to reducing mortality due to this disease. The blood marker CA125 was discovered in 1979 at DFCI by Bast and colleagues, who published in 1983 a study demonstrating the clinical utility of CA125 for monitoring ovarian cancer patients for recurrence. This paper established the current clinical paradigm for interpreting CA125; a positive test resulted if the CA125 exceeded 35 U/nL. Clinical screening trials for ovarian cancer using CA125 blood tests have been ongoing since 1986. We report on the sequence of trials and the statistical innovations made as trials proceeded from one phase to the next. In particular, methods were devekioed based ib hierarchical longitudinal change- point models which interpreted the annual CA125 value for a particular subject in light of previous values from the same subject. In contrast to using a constant reference level of 35 U/mL. for all subject, these approaches individualized decisions by accounting for an individual's previous CA125 values in the interpretation of the current CA125 value. Screening decisions include obtaining another blood test in 3 months rather than waiting one year, or referral to ultrasound. Retrospective analyses suggests the resulting screening program is more sensitive because it identifies cases with a rising CA125 above a low baseline earlier, and more specific because it rules out subjects with high yet stable CA125 values. This is joint work with Donna Pauler Ankerst at FHCRC, and Ian Jacobs MD, Professor of Gynecologic Oncology at University College London.
**Sponsored by the Charles P. Sisson II Memorial Lecturship
***Co Sponsored by The Marshall Woods Lectureships Foundation of
Fine Arts
Brown University -
Graduate School, Dissertation Defense Information
<--- 2004 Index