This is the fifth blog in a series of blogs on neoplasia.
In the past few blogs, I've been trying to explain the disconnect between cancer survival data and cancer death rate data. The cancer survival data seems to indicate that we're making enormous improvements in cancer treatment. The cancer death rate indicates that Americans are dying from cancer at about the same rate as they had been a half-century ago.
Yesterday, I listed 15 biases in cancer survival data that contribute to an overly optimistic sense of medical progress.
In this and the next few blogs, I thought I'd review some of these biases. The purpose of this exercise is to explain that the interpretation of survival data is enormously complex and that survival data is probably not the best way to gauge progress in the field of cancer research.
Of the sources of bias listed yesterday, stage assignment bias is the most subtle and the most difficult to understand. It is also the most important bias introduced by the most modern cancer trials, which are often aimed at developing stage-specific treatments for groups of patients with common cancers.
Suppose every patient with X type of cancer is staged into one of two groups (I and II). The stage I group has no evidence of distant metastases at the time of diagnosis and has a 40% chance of 5-year survival under the standard treatment protocol. Patients are put into the Stage II group if they have distant metastases at the time of diagnosis. Their chance of having a 5-year survival under the standard treatment is 2%.
Professor Rads, at the University of Goodcare, has recently developed a very sensitive imaging device that can detect small metastases that would be undetectable by less sophisticated devices. In the next clinical trial for treatment of cancer X, Professor Rads tests each trial candidate with his new device. He finds that about half of the patients who would otherwise be assigned to Stage I (no metastases) are found to have metastases with his sensitive machine. With this information, these erstwhile Stage I patients are re-assigned into Stage II.
A clinical trial is conducted with the standard treatment. The Stage I group is now much smaller than the Stage II group. When the trial is complete, we find that the 5-year survival for the Stage I group is now 80% (up from 40%). The 5-year survival for the Stage II group is now 2% (the same as before).
The newspaper headline following the trial is: "New imaging discovery yields 100% improvement in survival for Stage I Cancer X"
Actually, the clinical trial, as described, yielded no improved survival for any patients. All it accomplished was to correctly re-assign some of the Stage I patients into the low-survival Stage II group. The apparent improvement in survival in Stage I cancer patients was simply the result of more accurate staging of patients in the low-risk category of disease.
What does this mean? Was the clinical trial a fraud? Did it accomplish nothing at all? No, accurate staging of patients with cancer is an absolutely crucial step in the development of new, effective anti-cancer regimens. It is impossible to assess the effects of a new chemotherapeutic agent on a group with heterogeneous disease. The beneficial effects of a new drug on Stage I cancer patients might be lost in a study where the Stage I group is mixed with patients with Stage II cancers (that might not be responsive to the regimen).
By accurately staging patients with cancer, trialists can test and develop drugs that are most likely to benefit patients in a particular stage of disease. In fact, that is the current approach to cancer trials: subdividing cancer patients into well-defined clinical subsets and fine-tuning new anti-cancer regimens to maximize their effectiveness for each subset of disease.
So we see, in this instance, that cancer survival data is important, but it can be misleading.
We'll continue discussing the biases in survival data in the next blog.
-Copyright (C) 2008 Jules J. Berman
key words: cancer, tumor, tumour, carcinogen, neoplasia, neoplastic development, classification, biomedical informatics, tumor development, precancer, benign tumor, ontology, classification, developmental lineage classification and taxonomy of neoplasms