We all know the pharmaceutical industry is in trouble — what, with the precipitous patent cliff, soaring price of drug development and the death of the megablockbuster. And much ink has been spilled about the potential solutions to pharma’s problems, from mergers to academic partnerships to new research units. But what if the entire R&D enterprise is fundamentally flawed?
That’s the hypothesis of a new article in Wired magazine by science writer Jonah Lehrer postulating that the reductionist dogma of modern biomedicine has prompted scientists to desperately seek causal narratives where there are only statistical correlations to be found. In turn, Lehrer’s theory goes, such causal stories that scientists tell themselves have led drug developers on expensive and ultimately futile pharmaceutical goose chases.
It’s a provocative idea. The only problem is that Lehrer himself seems to be creating a causal explanation of his own for the industry’s woes on the basis of a reductionist extrapolation from a few flawed examples of pharma failures.
Lehrer’s primary example is torcetrapib, a much-touted experimental cholesterol drug from New York-based Pfizer that flopped in phase 3 trials. Torcetrapib works to raise levels of high density lipoprotein (HDL), a.k.a. ‘good’ cholesterol. And since preclinical and observational data had shown that higher HDL counts translate into better cardiovascular health, torcetrapib was expected to “redefine cardiovascular treatment,” according to former Pfizer CEO Jeff Kindler. That didn’t pan out. Pfizer pulled the plug on the drug’s pivotal trial after the compound was linked to higher rates of chest pain, heart failure and death than placebo.
“The story of torcetrapib is a tale of mistaken causation,” Lehrer writes. “Pfizer was operating on the assumption that raising levels of HDL cholesterol and lowering LDL would lead to a predictable outcome: Improved cardiovascular health. Less arterial plaque. Cleaner pipes. But that didn’t happen.”
He’s right that it didn’t happen for torcetrapib. But it could happen for torcetrapib’s drug class.
In his piece, Lehrer fails to
mention give due credit to several other compounds that work in much the same way as torcetrapib — namely, by blocking an enzyme called cholesteryl ester transferase protein, or CETP — yet look set to succeed where torcetrapib failed. For example, in separate phase 2 trials researchers found that people taking either anacetrapib (from New Jersey’s Merck) or dalcetrapib (from Switzerland’s Roche) experienced fewer heart attacks or strokes compared to those on placebo. These agents, along with Eli Lilly’s evacetrapib, are now in phase 3 trials, and pundits are now optimistic that the drug class will prove successful.
Outside of 18th century philosophy and psychological experiments with animated colored balls, Lehrer’s other main example of the scientific pursuit of causation gone amok relates to the treatment of chronic back pain. In the 1970s, Lehrer reports, doctors first started using magnetic resonance imaging to diagnose back pain. They spotted spinal disc abnormalities, which they figured were the driving force behind the pain, and, as a result, they started intervening with surgery and epidurals to fix the problem. But later research, published in 1994, found that such abnormalities were just as likely to be found in people with no back pain.
Yet another case of correlation — in this case between back pain and spinal bulges — leading scientists astray? Perhaps. But it seems more like plain sloppy science if more it took than a decade for researchers to compare the backs of people in pain with healthy controls.
Lehrer is correct that simply boosting HDL or operating on spinal abnormalities won’t necessarily translate into clinical benefit. But that doesn’t mean it’s time to throw out pharma’s babies with the biomedical bathwater. “We must never forget that our causal beliefs are defined by their limitations,” Lehrer writes about the biomedical research enterprise. The same could be said of biomedical reporting.