Drug-induced DNA modification in buccal cells of cancer patients receiving carboplatin and cisplatin combination chemotherapy, as determined by an immunocytochemical method: interindividual variation and correlation with disease response.

Twenty-six patients with a variety of tumor types were treated according to a phase 1 experimental treatment protocol consisting of repetitive cycles of cis-diammine(1,1-cyclobutanedicarboxylato)platinum(II) (carboplatin, 200-480 mg/m2) at day 1 and cis-diamminedichloroplatinum(II) (cisplatin, 50-100 mg/m2) at day 3. Buccal cells were collected in one or two treatment cycles prior to carboplatin, 24 h after carboplatin, just prior to cisplatin, and approximately 24 h after cisplatin administration. Drug-induced DNA modification was visualized at the single cell level by anti-serum NKI-A59 and quantitated by microdensitometry. All (39 of 39) treatments with carboplatin, and almost all (33 of 35) treatments with cisplatin resulted in an increase in nuclear stain. Interindividual variation in drug-induced, adduct-specific nuclear stain amounted to a factor of 5-8 for carboplatin and 5-12 for cisplatin. This drug-induced increase was, however, not related to the dose of either carboplatin or cisplatin, suggesting that large interindividual differences in DNA adduct formation and/or repair obscured the effects of dose variation within the relatively small range used for the drugs (2.4 for carboplatin and 2.0 for cisplatin). This explanation was strengthened by the good reproducibility of the immunocytochemical assay and by the reasonable correlation between carboplatin-induced nuclear stain in cycles 1 and 2 (correlation coefficient, 0.69; P = 0.009). Mean carboplatin-induced nuclear stain was significantly higher in the first cycle than in the second cycle (P = 0.0001) but this difference was no longer significant when drug-induced nuclear stain was corrected for carboplatin dose. Differences in cisplatin-induced nuclear stain between cycle 1 and cycle 2 were small and not significant. Carboplatin-induced nuclear stain was significantly higher in the partial responders than in the nonresponders (P < 0.0001, two cycles combined); the level of statistical significance remained the same after dose correction. Cisplatin-induced nuclear stain did not differ significantly between partial responders and nonresponders; this result might, however, be confounded to some extent by remaining carboplatin-induced nuclear stain at the moment of cisplatin administration. It is concluded that determination of the extent of platinum-induced DNA modification might be helpful in predicting the tumor response in cancer patients.