Blog | Thursday, January 26, 2012

Trials and errors in oncology, part II

So, the purpose of the Cancer Genome Atlas is to identify all mutations in the most common cancers. A massive project. Several maps have been completed including melanoma, pancreatic, ovarian, and lung cancers. Activation of these mutated genes results in the 6 characteristics of cancer previously listed. Therefore, it would stand to reason that aiming novel agents at these targets would inhibit growth and spread, possibly cure, cancer. But which targets to aim at?

Sentier au bord du Cher by via Flickr and a Creative Commons licenseFor example, pancreatic cancers contain between 50 to 60 mutations. To make order out of chaos, the most frequently identified mutated genes are inferred to be causative or "driver" mutations while the rest are "bystanders." Also, by recognizing that driver mutations tend to be found in certain "core" pathways but not others, this further reduces possible targets to a more manageable number. Between 13 to 15 pathways, an average of 13, are affected in a typical cancer type.

Getting back to the issue of causation, this month's issue of Wired contains an intriguing article by Jonah Lehrer headlined by the following statement: "Deadeend experiments, useless drugs, unnecessary surgery. Why science is failing us." The title, "TRIALS AND ERRORS", persuaded me to read on.

The story starts with Big Pharma's nightmare: a failed clinical trial. The drug, torcetrapib, appeared to be a slam dunk in that it lowered bad cholesterol (LDL) and increased the good (HDL) by inhibiting a protein that converts HDL to LDL. Inferred from these facts was that plaque formation would be reduced, which in turn would result in decreased morbidity and mortality from heart attacks and strokes.

In fact, the opposite occurred, and the Phase III trial was terminated. Pfizer had invested more than $1 billion dollars to develop torcetrapib, plus an additional $90 million to expand the manufacturing facility. The value of the company dropped by $21 billion in one week. Since 40% of drugs fail Phase II clinical trials, and 25,000 volunteers were participating in this trial alone, both the financial and human costs are staggering. $100 billion is invested in biomedical research annually.

How could torcetrapib fail? After all, the entire pathway of cholesterol metabolism had been mapped out and the drug's exact site of action was known. Sound familiar? As the author states, "It is a tale of mistaken causation." By lowering LDL and increasing HDL, it was assumed that improved cardiovascular health would result.

"This assumption-that understanding a system's constituent parts means we also understand the causes within the system-is not limited to the pharmaceutical industry or even to biology. It defines modern science. In general, we believe that the so-called problem of causation can be cured by more information, by our ceaseless accumulation of facts. Scientists refer to this process as reductionism ... Once we find the cause, of course, we can begin working on a cure."

Over the years we have learned that our attitude toward cause and effect is perceptual and that causal explanations are oversimplifications. We have learned to deal with the issue of causation through statistical correlation. The central concept is statistical significance, which "defines a significant result as any data point that would be produced by chance less than 5% of the time."

But significant correlation does not necessarily equal cause. "While correlations help us track the relationship between independent measurements, such as the link between smoking and cancer, they are much less effective at making sense of systems in which the variables cannot be isolated." The human body is extremely complex with inter-relationships between multiple pathways. Mapping one pathway and identifying all mutations does not reveal interactions between multiple pathways that are connected. This is why torcetrapib failed.

We are designing new clinical trials. We are mapping all the pathways of various cancers. By inference and statistical correlation, we think we have unearthed the driver mutations and core pathways that cause cancers, whose hallmarks have been identified. Are we setting ourselves up for another torcetrapib?

This post by Richard Just, MD, ACP Member, originally appeared at, a joint publication of Richard Just, MD, aka @chemosabe1 on Twitter and Gregg Masters, MPH, aka @2healthguru on Twitter. Dr. Just has 36 years in clinical practice of hematology and medical oncology.