these tainted articles…

Posted on Saturday 10 January 2015

Sometimes, I’m asked why I keep looking at studies long passed – like GSK’s Paxil Study 329 [2001] or AstraZeneca’s Seroquel Study 15 [1993-1995: still unpublished]. The usual answer is that the only way I know to aim for a better future is to thoroughly understand the misadventures of the past. But there’s another simpler answer. These tainted articles, once in the literature, are not really in the past. They’re sitting there in a journal as if they’re hot off the press long after they’re published – still used as references. Here’s a prime example from a recent New York Times blog:
New York Times
By Richard A. Friedman, M.D.
January 8, 2015

…Dr. Helen Mayberg, a professor of psychiatry at Emory University, recently published a study in JAMA Psychiatry that identified a potential biomarker in the brain that could predict whether a depressed patient would respond better to psychotherapy or antidepressant medication.

Using PET scans, she randomized a group of depressed patients to either 12 weeks of treatment with the S.S.R.I. antidepressant Lexapro or to cognitive behavior therapy, which teaches patients to correct their negative and distorted thinking.

Over all, about 40 percent of the depressed subjects responded to either treatment. But Dr. Mayberg found striking brain differences between patients who did well with Lexapro compared with cognitive behavior therapy, and vice versa. Patients who had low activity in a brain region called the anterior insula measured before treatment responded quite well to C.B.T. but poorly to Lexapro; conversely, those with high activity in this region had an excellent response to Lexapro, but did poorly with C.B.T.

What might explain these different responses?
This post isn’t about Dr. Mayberg, a neurologist at Emory best known for studying brain stimulation in refractory depression, or her contemporary study reported here [Toward a Neuroimaging Treatment Selection Biomarker for Major Depressive Disorder: full text on-line], or even only about Dr. Friedman’s NYT speculative blog. It’s about this next reference he mentions:
It turns out that other clinical factors may also help patients get the best treatment. For example, there is intriguing evidence that depressed patients who have a history of childhood trauma, such as the early loss of a parent or sexual or physical abuse, do not respond as well to an antidepressant as they do to psychotherapy.

In a large, multicenter study, Dr. Charles Nemeroff, then a professor of psychiatry at Emory and now at the University of Miami, found that for depressed adults without a history of abuse, there was a clear ranking order of treatment efficacy: Combined psychotherapy [using a form of cognitive behavior therapy] and an antidepressant [in this case, Serzone] was superior to either treatment alone. But for those who had a history of childhood trauma, the results were strikingly different: 48 percent of these patients achieved remission with psychotherapy alone, but only 33 percent of these patients responded to an antidepressant alone. The combination of psychotherapy and a drug was not significantly better than psychotherapy alone…
And so begins our remembrance of things past…
2003…
If you read the referenced article [Differential responses to psychotherapy versus pharmacotherapy in patients with chronic forms of major depression and childhood trauma: full text on-line], you’ll find that it says exactly what Dr. Friedman says it says – bolstered by this graph:
But if you look at that article in PubMed, you’ll notice this:
Erratum in
    Proc Natl Acad Sci U S A. 2005 Nov 8;102(45):16530.
2005…
And if you chase down that Erratum reference, you’ll find this…
Erratum in Proceedings of the National Academy of Science. 2005 102[45):16530.

MEDICAL SCIENCES. For the article "Differential responses to psychotherapy versus pharmacotherapy in patients with chronic forms of major depression and childhood trauma," which appeared in the Proc. Natl. Acad. Sci. in November 13, 2003, the authors note the following. "Results of the analyses of variance comparing change in Hamilton Rating Scale for Depression scores as a function of treatment type and early life trauma histories as well as Fig. 1A reflect change relative to the first week of treatment instead of baseline. When change scores relative to baseline are used, the interaction effects between treatment type and childhood trauma histories are not statistically significant…
… which invalidates their conclusion in spite of their attempt to evoke a plan B.
This discrepancy is due to marked changes in depression scores during the first week of treatment. Note that all analyses comparing the more conservative outcome measure of remission as a function of treatment type and childhood trauma as well as Fig. 1B are correct. Thus, consideration of treatment response relative to baseline does not detect the effect of childhood trauma on final remission, whereas consideration of final response relative to first response does detect the effect."
2012…
… and in spite of  that Erratum, here’s that same graph used in a presentation Dr. Nemeroff made at NYU in 2012 [see has to stop…]. Does he think everyone will just forget?


2003                                           2012

I originally got onto this thread reading a posting on Healthcare Renewal [PROFESSOR NEMEROFF GOES TO LONDON] by veteran Nemeroff watcher Dr. Bernard Carroll as Dr. Nemeroff was about to redo his NYU Grand Rounds in London last year. But then I got curious about the origin of the data in Differential responses to psychotherapy versus pharmacotherapy in patients with chronic forms of major depression and childhood trauma, and I found myself back at the turn of the century.
2000…
It came from a Serzone® Clinical Trial financed by Bristol-Myers Squibb:
by Martin B. Keller, M.D., James P. McCullough, Ph.D., Daniel N. Klein, Ph.D., Bruce Arnow, Ph.D., David L. Dunner, M.D., Alan J. Gelenberg, M.D., John C. Markowitz, M.D., Charles B. Nemeroff, M.D., Ph.D., James M. Russell, M.D., Michael E. Thase, M.D., Madhukar H. Trivedi, M.D., Janice A. Blalock, Ph.D., Frances E. Borian, R.N., Darlene N. Jody, M.D., Charles DeBattista, D.M.H., M.D., Lorrin M. Koran, M.D., Alan F. Schatzberg, M.D., Jan Fawcett, M.D., Robert M.A. Hirschfeld, M.D., Gabor Keitner, M.D., Ivan Miller, Ph.D., James H. Kocsis, M.D., Susan G. Kornstein, M.D., Rachel Manber, Ph.D., Philip T. Ninan, M.D., Barbara Rothbaum, Ph.D., A. John Rush, M.D., Dina Vivian, Ph.D., and John Zajecka, M.D.
New England Journal of Medicine. 2000 342[20]:1462-1470.

Conclusions: Although about half of patients with chronic forms of major depression have a response to short-term treatment with either nefazodone or a cognitive behavioral-analysis system of psychotherapy, the combination of the two is significantly more efficacious than either treatment alone.
It’s not the Clinical Trial that grabs the reader’s attention. It’s the list of 29 authors and what we now know about their future. Among other things, we’ll see Martin Keller leading the Paxil Study 329 team the next year [2001]; we see the leaders of the TMAP project [Rush and Trivedi]; we’ll see four who will make Senator Grassley’s list of COI offenders [Keller, Nemeroff, Schatzberg, and Rush]; we’ll see the principals in an attempt to get into the PHARMA business [Corcept: DeBattista and Schatzberg]; there are three department chairs who stepped down early [Keller @Brown, Nemerroff @Emory, Schatzberg @Stanford]; and others of note. It’s a veritable convention of KOLs in the salad days of 2000 [the year clinicaltrials.gov launched] on a Serzone® Clinical Trial [Serzone®, an SSRI, was pulled from the market in 2003 because of potential fatal hepatotoxicity]. The COI declaration list was so long that the NEJM published it separately on their web-site. But it was also the dawn of something else. In the same issue, then NEJM Editor Marcia Angell wrote an editorial focused on this specific article and all of the Conflicts of Interest, the start of what would later become her general campaign against Conflicts of Interest. It began:
by MARCIA ANGELL, MD
New England Journal of Medicine. 342[20]:1516-1518.

In 1984 the Journal became the first of the major medical journals to require authors of original research articles to disclose any financial ties with companies that make products discussed in papers submitted to us. We were aware that such ties were becoming fairly common, and we thought it reasonable to disclose them to readers. Although we came to this issue early, no one could have foreseen at the time just how ubiquitous and manifold such financial associations would become. The article by Keller et al. in this issue of the Journal provides a striking example. The authors’ ties with companies that make antidepressant drugs were so extensive that it would have used too much space to disclose them fully in the Journal. We decided merely to summarize them and to provide the details on our Web site.

Finding an editorialist to write about the article presented another problem. Our conflict-of-interest policy for editorialists, established in 1990, is stricter than that for authors of original research papers. Since editorialists do not provide data, but instead selectively review the literature and offer their judgments, we require that they have no important financial ties to companies that make products related to the issues they discuss. We do not believe disclosure is enough to deal with the problem of possible bias. This policy is analogous to the requirement that judges recuse themselves from hearing cases if they have financial ties to a litigant. Just as a judge’s disclosure would not be sufficiently reassuring to the other side in a court case, so we believe that a policy of caveat emptor is not enough for readers who depend on the opinion of editorialists.

But as we spoke with research psychiatrists about writing an editorial on the treatment of depression, we found very few who did not have financial ties to drug companies that make antidepressants. [Fortunately, Dr. Jan Scott, who is eminently qualified to write the editorial, met our standards with respect to conflicts of interest.] The problem is by no means unique to psychiatry. We routinely encounter similar difficulties in finding editorialists in other specialties, particularly those that involve the heavy use of expensive drugs and devices…
She later came to be less forgiving, and saw psychiatry as the leader of the COI pack [see in the name of ethics…]. I see her editorial as an awakening, one that I wouldn’t have for a number of years [when Senator Chuck Grassley and right hand, Paul Thacker, woke up a lot of us with their investigation]…

And so to my favorite quote:
    We shall not cease from exploration 
    And the end of all our exploring 
    Will be to arrive where we started 
    And know the place for the first time…
    Little Gidding, 1942, T.S. Eliot
So we’re back around to Dr. Friedman’s blog in the New York Times last Thursday. It’s a perfectly reasonable topic, Treat Depression, Drugs or Therapy?, something of general interest. In the discussion, he evokes and links an older study, speculating on the meaning of the neuroimaging findings reported by Dr. Mayberg et al recently. A general reader following the link in the blog would find that study with no clue that the central scientific claim had been retracted [for what may well have been a bit of sleight of hand – discovered how?].
 
Should Dr. Friedman have known about all of that? or about the Conflicts of Interest in the original study? Probably. Even if he doesn’t keep up with the blogs here at the edge of the galaxy, quoting Dr. Nemeroff, particularly from a paper back in 2000 or 2003, is always risky business. And that’s a widely known bit of information in the psychiatric community and elsewhere. But that’s not my central point. A paper like that should have been retracted from the literature, or at the least, retrospectively annotated on the Journal’s web-site by the Journal itself. The medical literature endures. It’s one thing for a study to become outdated and replaced by newer information. But it’s quite another if the original paper is as seriously flawed as this one. As time passes and there’s more distance from the scene of the crime, such papers may even become more credible as the contemporary criticisms [and critics] fade away. In psychiatry, particularly in the realm of industry funded Clinical Trials with guest authors [and/or ghost writers], there are still a number of thoroughly discredited articles that occupy the exact same space as legitimate publications, and we owe it to future medical and general readers to make their presence obvious or eliminate them from consideration altogether. This is a clear example of why!
  1.  
    Steve Lucas
    January 11, 2015 | 8:51 AM
     

    I questioned a doctor about their prescribing habits and was told bluntly that when I did a study, like the drug company, then I could comment on a certain drug. My reality is that when a new drug is announced there is a lead/lag in information. The drug company knows the only literature will be what they publish and it will take years for a proper meta analysis.

    This gives them an opportunity to sell the drug, or device, and try to make it part of the standard of care for that issue. Drug companies are use to working with very long time horizons so a retraction years down the road will be met with silence in the hope that some of the literature will survive giving life to a discredited product.

    Today we can look at the conflicts of various authors and question their bias in any new studies, remembering there is a pipeline of new KOL’s waiting in the wings. We can also look at the studies themselves and see the language of ghostwriting and that science has been kicked to the curb in favor of marketing.

    Old studies are important since they are the only ones we can analyze and then use as a base line for behavior and truthfulness in the newer material.

    A different doctor told me that you get what you pay for, so they relied on drug reps and journal articles for their prescribing information.

    Fool me once….

    Steve Lucas

  2.  
    Steve Lucas
    January 13, 2015 | 7:34 AM
     

    Here is an example of how a simple idea can make it into common practice:

    http://www.forbes.com/sites/larryhusten/2015/01/12/millions-of-americans-taking-aspirin-when-they-shouldnt/

    Steve Lucas

  3.  
    Susan Molchan, MD
    January 13, 2015 | 6:14 PM
     

    Thanks so much for the perspective on the Nemeroff/Serzone article. I’d commented on the Friedman piece as I thought it was poor journalism in general–saying we’d be using imaging “soon” to direct antidepressant tx. (I’m a psychiatrist & nuc med doc, and had done PET research).
    The study reported on is a research study & it’s a huge jump to think that it’s anywhere near clinical usefulness–For one thing 82 randomized patients but data used from only 38 of them (??!!). And we’re talking about activation in the insula on one side of the brain. What happened in Dr Mayberg’s former favorite–area 25 in the cingulate–oh, nothing : (
    And then of all studies to pick to provide perspective–Nemeroff and say he was still at Emory, when the NYT had reported some of his stomach-turning “misadventures.”
    Anyway, I’d like to bring to attention of Health News Review and cite 1BOM’s research if OK

    Susan

  4.  
    January 13, 2015 | 6:23 PM
     

    Susan

    Be my guest, and thanks for the comment.

    Mickey

  5.  
    Bernard Carroll
    January 13, 2015 | 8:23 PM
     

    Susan Molchan is right to point out the inconsistencies and overstatements from the Mayberg group at Emory. A pointedly critical letter appeared in JAMA Psychiatry in response to the 2013 Mayberg report that Richard Friedman discussed. That link will give a further link to the full text. As well, in 2014 Mayberg published a somewhat confusing second report on the predictors of response in the same 82 patients. In that follow-up study there was no positive finding for the anterior insula which had been highlighted in the earlier study. In both studies there is reason to ask whether the authors were not HARKing – at the very least they were analyzing data selectively, as Dr. Molchan points out. We don’t know whether Helen Mayberg talked with Richard Friedman but if she did then she doesn’t seem to have mentioned the follow-up study.

    Beware the Principal Investigator promoting a narrative in The New York Times and on the Charlie Rose show!

  6.  
    James O'Brien, M.D.
    January 14, 2015 | 3:16 PM
     

    As twisted as all of this is, the conclusions of the NYT article does sort of comport to clinical experience and common sense.

  7.  
    Joseph Arpaia
    January 16, 2015 | 1:23 AM
     

    Pretty pictures from brain scans — Phrenology anyone?

  8.  
    Susan Molchan, MD
    January 20, 2015 | 12:04 PM
     

    Hi Mickey–I got a blog posted on Health News Watch Dog re the Friedman NYT piece–you will enjoy, as will your readers, esp Dr. Carroll ; ) Thanks to both of you for your comments. Love your blog

    http://www.healthnewsreview.org/2015/01/potential-biomarker-that-could-predict-caveats-about-psychiatric-brain-imaging-blogging-about-it/

    Susan

Sorry, the comment form is closed at this time.