Predicting the maximum-tolerated dose of PNU-159548 (4-demethoxy-3'-deamino-3'-aziridinyl-4'-methylsulphonyl-daunorubicin) in humans using CFU-GM clonogenic assays and prospective validation.

A haematotoxicity model was proposed by Parchment in 1998 to predict the maximum-tolerated dose (MTD) in humans of myelosuppressive antitumour agents by combining data from in vitro clonogenic assays on haematopoietic progenitors and in vivo systemic exposure data in animals. A prospective validation of this model in humans was performed with PNU-159548, a novel agent showing selective dose-limiting myelosuppression in animals. PNU-159548 and its main metabolite, PNU-169884, were tested in vitro on murine, canine and human colony forming units-granulocyte macrophages (CFU-GM) and in vivo on mice and dogs. The IC(90x) ratios (IC(x)=concentration inhibiting x% of colony growth) for CFU-GM and drug plasma protein binding were used to adjust the target plasma concentrations versus time curve (AUC) and predict the human MTD. The predicted MTD was compared with values achieved in phase I studies. Canine CFU-GM were 6-fold more sensitive (P<0.01) and murine CFU-GM 1.7-fold less sensitive (P<0.05) to PNU-159548 treatment than the human progenitors. PNU-169884 behaved similarly to PNU-159548. The predicted MTDs in humans calculated from data in mice and dogs were 15 and 38 mg/m(2), respectively. Overall, 61 patients were treated in two phase I studies, at doses ranging from 1.0 to 16 mg/m(2). Thrombocytopenia was dose-limiting with a MTD of 14 and 16 mg/m(2) in heavily and minimally pretreated/non-pretreated patients, respectively. Adjusting animal MTD data by means of the CFU-GM ratio between species can predict the human MTD with a good quantitative accuracy. Inhibition of common haemopoietic progenitors by PNU-159548 induced neutropenia/thrombocytopenia in animals and thrombocytopenia in patients, probably due to the higher sensitivity to the compound observed in human colony forming units-megakaryocyte (CFU-MK).