Clinical Trials and Personalized Medicine - Interpreting Studies
March 04, 2013 12:00:00 AM
Medicine does not search for truth. It searches for cure. It does not look for the universal, it tries to create exceptions.
Medicine emerged from witchcraft. It has always utilized the most advanced technology of its day. Medical models and reasoning always evolve and that evolution makes the previous model obsolete. One of the foundation models of modern medicine is the randomized controlled clinical trial.
The principal of the randomized controlled clinical trial is that a single observation needs to be validated and reproduced. The clinical trial provides an estimate of how often a particular observation will occur. It tells us that chemotherapy improves survival for patients with non-small cell lung cancer at one year from 20% to 29%. It tells us that FOLFOX treatment for advanced colon cancer gives a median time to progression of 8.7 months, response rate of 45%, and median survival time of 19.5 Months. This is accurate information about populations. It's use for the individual is a difficult problem.
Every person is a unit, no one is 20% or 29% or 45%. The question is who will respond and to what therapy.
A mathematically rigorous approach to the application of statistical data to the individual is (theoretically) available through the application of Bayes’ theorem, but this has rarely been done clinically. With the greater availability of computing power, we hope that a mathematical alternative to clinical judgment will be more widely available.
The heterogeneity of cancers makes clinical trial data very hard to interpret. For a long time, we called malignancies that are clearly different the same, and studied them together in clinical trials. We did not distinguish between BCR-ABL rearranged CML (which almost always responds to ABL kinase inhibitors) from myeloproliferative syndromes driven by JAK mutuations. Cytogenetic analysis separated the retinoic acid responsive promyelocytic leukemia from other acute leukemias.
Now, the internal heterogeneity of cancers is recognized as a challenge to therapy.
Theory and statistics must serve practice. Practice looks for the best possible result for the individual patient.