Common pitfalls in preclinical cancer target validation

Key Points Scientific robustness refers to the ability of a finding to withstand experimental variation. Results that are reproducible, but only under an extremely narrow set of conditions, are unlikely to make predictions that will be true (robust) under real-world conditions, such as in the clinic. Whether an elevated level of a particular protein is associated with a poor prognosis in a given cancer provides very little information as to whether that protein would be a good target in that cancer. Being associated with a poor prognosis is neither necessary nor sufficient to be a good cancer target. The fact that A correlates with B and that it is biologically plausible that A causes B does not formally prove that A causes B. For example, observing that high expression of a gene correlates with poor survival in cancer patients and knowing that that gene regulates malignant cell behaviour would not formally prove that the high expression of that gene is responsible for the poor survival. Similarly, observing that a drug is having its expected pharmacodynamic effect on its intended target and knowing that its intended target is important for cancer cell survival would not formally prove that the cytotoxicity of the drug is on-target. Most of the cellular assays used in cancer pharmacology are 'down' rather than 'up' assays, which is problematic because there are far more uninteresting ways to make a complex system, such as a cell, perform less well than there are to make it work better. Cellular phenotypes caused by a chemical or genetic perturbant should be considered to be off-target until proved otherwise, especially when the phenotypes were detected in a down assay and therefore could reflect a nonspecific loss of cellular fitness. It is only by performing rescue experiments that one can formally address whether the effects of a perturbant are on-target. The basis for the therapeutic indices of the currently available cancer drugs, including cytotoxic and targeted agents, is still poorly understood. Most successful drugs do not inhibit their targets completely and continuously at their therapeutically useful doses and accurately predicting, a priori, the therapeutic index for inhibition of a new cancer target is virtually impossible. This Perspective discusses some of the causes of the robustness and reproducibility problem in preclinical cancer research and suggests solutions. An alarming number of papers from laboratories nominating new cancer drug ta