Friday, May 14, 2010

Eli Lilly: Our Drug Failed, So it Has Serious Potential

ResearchBlogging.org
These folks at Lilly must think we are exceptionally stupid. As in can't tie our own shoes. A study in the Journal of Psychiatric Research recently found that their experimental antidepressant LY2216684 was no better than placebo. Here are a couple of quotes from the abstract:
LY2216684 did not show statistically significant improvement from baseline compared to placebo in the primary analysis of the Hamilton depression rating scale (HAM-D17) total score. Escitalopram demonstrated significant improvement compared to placebo on the HAM-D17 total score, suggesting adequate assay sensitivity.
On the primary outcome measure, the experimental drug failed whereas Lexapro worked to some extent. I know what you're thinking - "the sample size was probably too small to find a significant effect." Um, you're wrong. How about 269 people on the Lilly drug, 138 on placebo, and 62 on Lexapro.

But wait, here comes the good news...
Both LY2216684 and escitalopram showed statistically significant improvement from baseline on the patient-rated QIDS-SR total score compared to placebo... The results of this initial investigation of LY2216684’s efficacy suggest that it may have antidepressant potential.
The good news for Lilly is that most people who claim to "read journal articles" really just browse the abstract without actually looking at the full text of the paper. For the select few who have nothing better to do than read Lilly propaganda, take a look at Table 2. A total of 12 secondary outcome measures are listed. The Lilly drug beat placebo on... ONE of them. Lilly doesn't say much about how much better their drug was than placebo on the QIDS-SR measure beside throwing around that often meaningless term of "statistically significant." People on the drug improved by 10.2 points whereas placebo patients improved 8.3 points. So about a 20% difference. If you bother to calculate an effect size, it is d = .24, which is quite small and clinically insignificant. So on the ONE measure where the drug was better than placebo, it was by a small margin, and it missed the mark on 11 other secondary measures as well as on the primary outcome measure. But "it may have antidepressant potential." Hell yes, I've never been so exited about a new drug.

By the way, Lilly is apparently trying this wonder drug out in at least five trials. The journal in which this article appeared has published other dubious Eli Lilly research in the past. The editorial review process is clearly working wonders over at the Journal of Psychiatric Research. Sad, really. The journal publishes some really good work, but then runs this kind of junk as well.

Depression Self-Report Sidebar: The self-reported measure on which the drug had an advantage, the Quick Inventory of Depressive Symptoms (QIDS) - it's really awesome, according to Lilly. Remember, it's the only measure on which their experimental failure drug had an advantage over placebo. So the authors wrote "Self-reported depression symptoms, such as those obtained by the QIDS-SR, may be more sensitive than clinician-administered scales for signal detection in clinical studies of depression."

What does Bristol-Myers Squibb think? In three trials of Abilify for depression, self-reports of depression were unfavorable. So the publications for these studies made sure to downplay these depression self-reports by saying that these measures were not sensitive, that they weren't picking up improvements in depression.

So if a self-report provided positive results, then BAM, it's an awesome measure of depression. But if it provided negative results, then it's a horrendously inaccurate measure and should never have been used in the first place.

Citation below. Yes, one of the authors' last names is Kielbasa.

Dubé, S., Dellva, M., Jones, M., Kielbasa, W., Padich, R., Saha, A., & Rao, P. (2010). A study of the effects of LY2216684, a selective norepinephrine reuptake inhibitor, in the treatment of major depression Journal of Psychiatric Research, 44 (6), 356-363 DOI: 10.1016/j.jpsychires.2009.09.013