As the population of male veterans grows older, prostate cancer is an increasingly common problem in the Veterans Affairs (VA) health care system. A wide spectrum exists for the risk of death at the time of diagnosis of prostate cancer, and clinicians currently have limited data and methods to predict survival accurately in individual patients. Treatment decisions, especially for patients with localized prostate cancer, are currently hindered by these limitations.
The goal of this research is to develop and internally validate a comprehensive prognostic staging system that predicts long-term survival among men diagnosed with prostate cancer. The staging system will include cogent clinical characteristics of patients (e.g., age, symptom status, comorbidity) and selected laboratory characteristics of tumors (e.g., angiogenic and tumor suppressor factors), as well as "conventional" measures of anatomic spread, histologic grade, and serum prostate specific antigen (PSA) level.
The cohort for this observational study consists of more than 70,000 men receiving ambulatory health care at any of 10 VA medical centers in New England during 1989-90. Among this population, more than 1,300 men were diagnosed with prostate cancer during 1991-95. Data are being obtained from: medical records for demographic and clinical information about the patient; search of VA pathology database for identifying prostate cancer biopsies; search of VA laboratory database for PSA assays; analysis of archived biopsy specimens for selected angiogenic, tumor suppressor, and cellular adhesion factors; and search of National Death Index and VA Beneficiary Identifier Locator System (BIRLS) files for mortality. Statistical analysis will include baseline variables measured at time of decision to treat (or not treat) each patient: demographics (age, ethnicity); clinical condition (symptom status, comorbidity, functional and cognitive status); tumor characteristics (anatomic stage, histologic grade, PSA level, angiogenic profile); and type of therapy, although treatment effects are not the major focus of this study. The outcome variable is death through 1999 (5-9 year f/u). Results will be obtained using a proportional hazards model to identify clinically and statistically significant risk factors in multivariable analysis. Subsequently, a simple staging system will be developed for clinical application.
We have requested 1,325 medical records, and are reviewing more than 1,000 records received to date. Laboratory assays have been piloted, and 1,267 biopsy specimens have been retrieved for study. BIRLS and PTF (Patient Treatment Files) are accessed for mortality every three months.
The development of a comprehensive prognostic staging system for prostate cancer will provide important information to improve clinical treatment decisions, and can also be used in research to evaluate therapeutic results, "adjust" data across studies, and stratify patients at entry in clinical trials.
- Concato J. Challenges in prognostic analysis. Cancer. 2001 Apr 15; 91(8 Suppl):1607-14.