On the normality of growth dynamics of neoplasms in vivo: a data base analysis

The in vivo population dynamics of 248 literature data bases of normal and neoplastic growth processes were examined by computer velocity analysis. The growth of most tumors was remarkably normal. When abnormalities did occur, they tended to be infrequent and subtle. Tumor growth was highly regulated, not unregulated Exponential growth was rare, and the kinetics of both normal tissues and tumors were predominantly deceleratory. With both normal tissues and tumors, the most intense period of growth inhibition usually occurred early in development. In general, growth was rapid only at small size. There was no evidence that tumors as a group grew faster than normal tissues. A small subclass of tumors exhibited a specific proliferative abnormality. Their growth did not stop entirely at large size, but instead continued indefinitely at a very slow basal rate. The majority of tumors did not exhibit this anomaly, however. Eight of the data bases permitted a kinetic analysis of the inhibitory mechanisms underlying growth inhibition. Six behaved as if their growth were governed by a tissue sizer, that is, by a negative feedback inhibition that strongly correlated with tumor size. The seventh behaved as if its growth were regulated by a biological time-keeper, and the final tumor as if by a stochastic mechanism such as random arrest or terminal differentiation. The analysis indicated that many tumors were completely normal in their growth governance policies and control mechanisms, and suggested that in some instances neoplastic transformation might be a disease of tissue neogenesis resulting from an altered or inappropriate cellular recognition process. The ability of 18 different equations to model in vivo growth data was examined. Accurate modeling required an equation to be objectively selected from a menu of alternatives by statistical criteria. All but one of the equations provided a best-fit to at least one data base, and all of the equations were inferior models of some data bases. The Spillman and inverse Nth root equations provided the greatest number of best-fits. The Gompertz equation was a good growth model for normal tissues, but mediocre for tumors.