On Thursday an appellate court in New Jersey reversed the decision to let me testify in the case of Palazzolo v Hoffman La Roche, although they upheld the decision to not permit the results of our brain imaging study showing that Accutane affected function of the orbitofrontal cortex. In doing so they correctly noted that the results of the imaging study were just one part of the body of evidence needed to conclude that Accutane can cause depression, not the linchpin of the argument, as the prior judge had noted.

This announcement caused an outpouring of vitriol (e.g. calling my study “faked” and “junk science”), first from lawyers working on behalf of pharma and device makers, writing in the Drug and Device Law blog, then from an MD who works as writer and marketer for pharma and pharma-sponsored CME. I felt I need to set the record straight on a number of points that were made.

First of all, it is not true that the study was “commissioned for the litigation”. Eighty percent of the study was paid for by money donated by Liam Grant of Ireland. Brain scans aren’t free. Roche refused to do a study. In fact, back in 1999, I met personally with John McLane, who unbenownst to me was a senior executive for Roche at the time, who refused to even provide medication for the study. I am not an expert in Liam Grant’s litigation (in Ireland you cannot get more than $50,000 in litigation, so he is hardly doing it for the money). I agreed to be an expert after the study was done, not before. Half of all research is supported by pharma. They “commission” research to serve their marketing goals all the time. The bias that introduces has been well documented. When a study is done that is not controlled by them, they go bananas.

Some of the missing data referred to as “bmax” was actually an erroneous term introduced in the course of the 15 depositions (8 hours each) I went through with Roche over this study. The inability to retrieve the numbers was temporary due to problems accessing old media (not an uncommon problem in imaging research). By the time the data was retrieved a court deadline had passed. But it is inaccurate to imply that they were never retrieved.

The reference to not following the study methodology refers to a questionnaire about satisfaction with skin condition (called the Skindex) that was not part of the original protocol, and that was added late, and was not the primary focus of the study.  The original article stated that it was given before and after treatment, but the after treatment results were not presented. As I wrote in a correction later the questionnaire was not given to all of the subjects after treatment.

Roche spent a lot of time and effort trying to debunk this study, in the course of which some data entry errors were found. A re-analysis of the study with corrected data continued to show a reduction in function of the orbitofrontal brain function, in fact the results were more statistically significant than before. Roche next accused me of fraud and asked the journal to retract the paper, which led to an inquiry at my university where the committee had access to all the data and legal documents, and cleared me of the charges of fraud, recommending a letter of correction based on the corrected data, which led to this single sentence correction published in the journal.

The conclusion that Accutane can cause depression shouldn’t rest on a single study, which it doesn’t, as the evidence that retinoids play a role in affective disorders continues to grow.