John Stamos - Public Service Announcement
The Realities of Cancer Statistics
Having recognized the important role of assay endpoint (cell death vs. cell growth), in accurate drug selection, the next issue is a statistical one.
Scientific tests provide data within a probabilistic context. The performance characteristics of the cytotoxicity assay technique, established in over 500 published clinical correlations, provides a Sensitivity of 94% and Specificity of 71% (Principles and Practice of Oncology Updates: Vol. 7, No. 12, 1993). Yet, treating physicians must depend upon the more clinically relevant statistical measures, positive and negative predictive accuracies, which are profoundly influenced by the true responsiveness of the disease in question (e.g. childhood leukemia is by nature more chemo-responsive than adult colon cancer). This explains the all-too-familiar but inaccurate saying, "Chemosensitivity tests are good for selecting what won't work, but are not good for selecting what will!"
In point of fact, cytotoxicity tests are very good at finding effective drugs, but only if effective drugs are there to be found! It is therefore incumbent upon treating physicians to examine laboratory results within a realistic clinical context. This is described mathematically by what is known as Receiver-Operator-Curve (ROC) or Bayesian analysis. This constitutes an important advantage of Rational Therapeutics' approach over prior investigations in this field.
|
©2010 Rational Therapeutics. All rights reserved.