Relationships between reductive metabolism, DNA damage and antitumour activity of the hypoxia-activated prodrug PR-104 in preclinical models

Abstract

Hypoxia is a characteristic of solid tumours and a potentially important therapeutic target. PR-104, a prodrug designed to target hypoxic cells, is currently in phase II clinical trial. PR- 104 is converted systemically to the alcohol PR-104A which is reduced selectively under hypoxia to its hydroxylamine and amine metabolites. This thesis aims to determine the mechanism(s) of cytotoxicity of PR-104 in preclinical tumour models. Specific objectives were to identify biomarkers of response to PR-104, and to ascertain which features of tumours determine their sensitivity to PR-104 monotherapy. PR-104A demonstrated hypoxia-selective cytotoxicity for all nine human tumour cell lines investigated in culture, but with widely differing hypoxic/oxic differentials because of large differences in aerobic cytotoxicity. Hypoxic cytotoxicity correlated with DNA interstrand crosslink (ICL) frequency (measured using the alkaline comet assay), suggesting ICL are invariably responsible for hypoxic cytotoxicity. There was a similar relationship between ICL and aerobic cytotoxicity, except for four lines with very low rates of aerobic PR-104A reduction. Nonetheless, monotherapy activity of PR-104 in tumour xenografts of all nine cell lines showed a strong correlation with ICL at 24 hours (r2 = 0.733, p<0.001) indicating that DNA crosslinking is the dominant mechanism of cytotoxicity in the pharmacologically relevant dose range. A hypoxia-selective increase in the DNA double strand break marker γH2AX was also observed in all cell lines after PR-104A in vitro, with kinetics and cell cycle distribution consistent with DNA replication arrest at ICL. This marker also correlated with cytotoxicity in tumours (r2 = 0.465, p<0.001). The wide variation in γH2AX levels between cell lines at equivalent cytotoxicity suggested γH2AX is less useful than ICL as an absolute biomarker, but has possible utility for comparing response of aerobic and hypoxic cells within the same tumour. Reduced PR-104A metabolites in tumours also correlated with cytotoxicity (r2 = 0.585, p<0.05), but less well than for ICL. Unexpectedly, the results suggest that rates of intratumour activation of PR-104A are not primarily determined by hypoxia (measured using pimonidazole binding) or expression of the major hypoxic reductase cytochrome P450 oxidoreductase, and that aerobic nitroreduction is a major contributor to its monotherapy antitumour activity.