Models for dose intensity (Q1117155)

The authors introduce a concept of a time-dependent dose intensity function as a determinant of the therapeutic outcome of chemotherapy in cancer treatment related to tumor size, generalizing a proposal by Hryniak [\textit{W. M. Hryniak} and \textit{M. Bush}, The importance of dose intensity in the chemotherapy of metastatic breast cancer J. Clin. Oncol. 2, 1281-1288 (1984)]. The number N(t) of tumor cells at time t is modelled as \[ d(\ln N(t))/dt=h_ 1(N(t))[1-h_ 2(C(t)] \] where \(h_ 1\) and \(h_ 2\) are integrable functions referred to growth and kill, and where c(t) is the drug concentration at time t; \(N(0)=1\), \(h_ 1(N)>0\), and \(h_ 2(0)=0\). The dose intensity for the interval (t',t) is defined as \[ I(t',t)=[\int^{t}_{t'}h_ 2(C(u))du]/(t-t') \] represnting a mean kill over that interval. The possible range of application is discussed and illustrated by two examples where \(h_ 2=k C\). For chemotherapy administred as a series of cycles repeated a fixed number of times at regular intervals, a ranking of protocols by their dose intensity function is defined and the case \(h_ 2(C(t))=k C(t)\) is considered. The model is finally extended to the development of drug resistance by introducing the probability \(P_ 0(t)\) that there are no resistant cells at time t as solution of \[ d(\ln P_ 0(t))/dt=-\alpha h_ 1(N(t))N(t). \] Calculation of the conditional probability \(P(t',t)=P_ 0(t)/P_ 0(t')\) that resistance develops for the first time in the inteval [t',t] shows that protocols can then no longer be ranked by the dose intensity alone but have to be complemented by information from the tumor's growth.