Analysis of growth delay data: potential pitfalls.

Abstract Several theoretical aspects of growth delay curves which can affect the relationship between delay and cell killing have been considered and compared with examples of real data. (1) Damage to the tumour "bed", presumably vasculature, can result in slower growth compared with untreated tumours; the so called "tumour bed effect". As a consequence the delay in growth depends on the end-point size. In addition, agents which modify tumour and stromal cell damage to a different degree may lead to different conclusions depending upon the assay. (2) A slow rate of clearance of dead cells and debris after treatment can lead to slower growth of tumours in the initial stages of regrowth above treatment size. This can appear to be a tumour bed effect even when stromal damage is absent. (3) Average growth curves for a group of tumours should not be used instead of the delay for each individual animal because this can also lead to an apparently slower regrowth in the absence of a tumour bed effect. (4) Tumour size at treatment should be standardized. If the normal growth pattern before and after treatment is Gompertzian, treatments which produce equal killing will produce more delay in larger tumours.