Therapeutic implications of tumor heterogeneity.

At the outset of this review, we stated that we wished to raise some questions that challenge the commonly held view that tumor heterogeneity is of major significance to treatment failure. The nature of this challenge is the following: although tumor heterogeneity in sensitivity to therapeutic agents has been demonstrated repeatedly, using isolated subpopulations of cells, primarily in cell culture systems, there is very little work that has been directed toward asking the tough questions about how that heterogeneity actually impacts on the response to treatment in vivo. In our own work, when we have attempted to simulate heterogeneity, in vivo or in vitro with mixed populations of tumor cells, we have seen that the simple prediction that treatment response would reflect the sensitivities of the individual subpopulations was not valid. Tumor subpopulation interactions, influencing both growth and drug sensitivity, resulted in treatment responses that were either better or worse than would be expected. Shifts in the distribution of subpopulations under the influence of therapy did not necessarily correlate with treatment response. Marked differences in the relative proportions of subpopulations within tumors did not necessarily translate into marked differences in the behavior of whole tumors. Imposition of in vivo-like three-dimensional tissue architecture caused major changes in the overall drug sensitivity of individual subpopulations, beyond those seen as a result of heterogeneity. Of course we realize that our work is very limited, one tumor system and a few treatment protocols, but that is the challenge. Much more in-depth experimental and clinical research is necessary in order to evaluate how, and how much, tumor heterogeneity really does affect treatment. Without such work, efforts to devise more effective treatment strategies, based on common assumptions and theoretical models, rather than experimental analysis, can only be superficial and, ultimately, useless.