A stochastic model for dual label experiments: an analysis of the heterogeneity of S phase duration

We develop a statistical approach for the study of S phase duration in experiments using sequential pulses of two thymidine analogues. Cell entrance into S phase is assumed to follow a possibly nonhomogeneous Poisson process, and the duration in S phase independently follows an unspecified distribution, thus allowing the possibility of variable S phase duration times for cells. Several conclusions regarding experimental design considerations are reached. The availability of three labelled cell subgroups comprising cells receiving exactly one of the two thymidine analogues or receiving both thymidine analogues at least doubles the efficiency of the mean S phase duration estimate compared to conventional estimates based on two labelled groups. Increasing the duration between the two thymidine analogue pulses can also dramatically increase the efficiency. The modelling technique was applied to fourteen uveal melanomas from patients who received in vivo injections of two thymidine analogues, bromodeoxyuridine (BrdUrd) and iododeoxyuridine (IdUrd). Counts of labelled cells were consistent with steady-state time homogeneous entry into S phase and nonhomogeneous spatial entry into S phase.