Blood samples from 36 germ cell tumor patients receiving chemotherapy with either cisplatin or carboplatin in combination with other drugs [etoposide or vinblastine, cyclophosphamide, dactinomycin, and bleomycin (VAB-6)] were analyzed for the presence of 7 different biological markers. The biomarkers included platinum-protein adducts, platinum-DNA adducts, sister chromatid exchange (SCE), micronuclei (MN), and somatic gene mutation at the hypoxanthine phosphoribosyl transferase (HPRT) locus and the glycophorin A (GPA) loci (NO and NN). Patients were asked to donate 9 serial blood samples: a pretreatment sample followed by another drawn 12-24 h after each of four cycles of treatment and a final sample provided 3-6 months after the last cycle. Most individuals gave 7-8 samples; 7 individuals donated all 9. Because of limited amounts of cells in some cases, it was not possible to carry out all 7 assays on every sample. Pt-protein adducts, Pt-DNA adducts, and SCE showed a direct and consistent effect of treatment and were very highly correlated. A significant correlation was also seen between both Pt-protein and Pt-DNA adducts and HPRT mutation. All of the posttreatment samples were significantly elevated compared to the baseline sample. These markers also remained elevated 3-6 months after the end of treatment. By contrast, MN, HPRT mutation, and GPA mutation (both NO and NN variants) showed varying patterns of dose response, probably reflective of the differing biology of these markers scored in lymphocytes (MN and HPRT) and erythrocytes (GPA). MN were significantly elevated in the posttreatment samples drawn at cycles 2 and 3. Although induction of HPRT mutation was only of marginal significance, results here are for the mutant frequency determination assay only. In progress is the potentially more informative analysis of the type of mutations by Southern blot and the sequencing of mutations to look for characteristic mutational spectra. The GPA assay showed a significant increase over baseline in samples drawn after cycles 3 and 4 (NO variants) and after cycles 2, 3, and 4 (NN variants). The level of GPA mutation (both NO and NN variants) was clearly elevated even 3-6 months after the last cycle of chemotherapy. Correlations were seen between HPRT and MN as well as between GPA NO and GPA NN variants. Analysis of biomarkers by treatment group does not reveal a consistent pattern or trend across all cycles.(ABSTRACT TRUNCATED AT 400 WORDS)