Germ Cell Tumors (GCT) are a curable malignancy in which the physician must discriminate between good risk and poor risk attributes and choose a treatment program designed for good risk or poor risk patients. Although several allocation criteria for good and poor risk categories exist, the currently available models based on a patient's pretreatment clinical characteristics are imprecise. About 10% of good risk patients still die, and about one-third of poor risk patients survive with standard treatment despite poor prognostic indicators. Recent data largely generated at our institution suggest that one or more additional tumor markers may be better, or additional, prognostic indicators.
The Specific Aims of this proposal are to relate four putative markers of prognosis to tumor response and patient survival: (1) the rate of clearance (half-life) of alphafetoprotein (AFP) and/or human chorionic gonadotropin (HCG) after initiation of chemotherapy using weekly marker assays; (2) the 12p copy number in the primary tumor determined by 12p painting probes; (3) TP53 expression in the primary tumor; and (4) Hst1/kFGF expression in primary tumors. The specific hypotheses are that a prolonged half-life clearance of AFP and/or HCG, high 12p copy number, mutant TP53 mutation, and Hst-l/kFGF expression are poor prognostic findings which may be independent predictors of prognosis. Good risk patients from Memorial Hospital and the Southwest Oncology Group and poor risk patients from Memorial Hospital will be accrued onto prospective clinical trials specific to risk status. Weekly assays of AFP and HCG will be obtained after the initiation of chemotherapy to determine the marker half-life clearance. Primary tumors will be obtained and studied for 12p copy number, mutant TP53 expression, and Hst-1/kFGF expression. Biostatistical analyses will be performed using logistic and Cox regression techniques to determine the significance of each of these four variables in the presence of standard prognostic variables. If one or more of these four new variables independently predicts treatment outcome and patient survival, then a better algorithm for the selection of treatment for the individual patient can be developed.
