Pristiq (desvenlafaxine extendedrelease tablet) was approved by the FDA in February of 2008 for the treatment of major depression. In the April 2008 issue of TCPR, we reviewed the efficacy data available at that time. We concluded that it is a marginally effective antidepressant with no clear advantages over its precursor molecule, venlafaxine extended release (Effexor XR).

How has Pristiq fared over the past year and half? Not very well. As of May 2009 (over one year after its launch) Pristiq had captured 1.3% of the U.S. antidepressant market versus 9.1% for Effexor, 9.8% for Cymbalta and 16.8% for Lexapro (http://bit.ly/NW7P5). In October of 2008, Wyeth decided to stop trying to get European approval of Pristiq, reportedly because the European Medicines Agency believed that Pristiq was less effective than Effexor for either depression or for hot flashes of menopause (http://bit.ly/10nXzo).

When we published our earlier review of Pristiq in April 2008, the two key Wyeth-funded studies of Pristiq were available only in poster form. They have since been published.

In the first study, conducted in various sites in the U.S., 447 patients with major depression were randomly assigned to Pristiq 50 mg/day, Pristiq 100 mg/day, or placebo. After eight weeks of treatment, patients taking 50 mg/day were in slightly better shape than patients on placebo (mean improvement on the Hamilton Depression Scale 11.5 points on Pristiq versus 9.5 points on placebo), but those assigned to 100 mg/day improved by only 11 points, not statistically superior to placebo (Liebowitz MR et al., Curr Med Res Opin 2008;24(7):1877-90). In the second study, conducted in Europe and South Africa, Pristiq fared somewhat better.Using the same research design, 483 patients were randomized to the three groups, and after eight weeks, both doses of Pristiq statistically beat placebo, though not by much (Pristiq 50 mg/day, -13.2; Pristiq 100 mg/day, -13.7; placebo, -10.7) (Boyer P et al., Int Clin Psychopharmacol 2008;23(5):243-53).

Of course, the question foremost in most psychiatrists’ minds is whether Pristiq has any advantages over its mother compound, Effexor. In the only available comparative studies (funded by Wyeth), Pristiq did poorly.

According to Lieberman DZ et al., (Int Clin Psychopharm 2008;23:188-197), a total of 738 depressed patients were enrolled in two separate studies, each with three arms: Pristiq, Effexor XR, and placebo. Patients were randomly assigned thusly: 250 to placebo, 239 to Pristiq (200-400 mg/day, mean dose 302 mg/day or 336 mg/day, depending on the study), 249 to Effexor XR (75-225 mg/day, mean dose 118 mg/day or 206 mg/day). The primary outcome measure was the change in the Hamilton depression score (HAM-D17).

Pristiq failed to outperform placebo in either of the two trials, while Effexor XR beat placebo in the study in which it was dosed at 206 mg/day (it did not separate from placebo when its average dose was 118 mg/day). On the secondary outcome measures of response and remission rates, Pristiq again failed to separate from placebo, whereas Effexor XR did (see table for figures)(Data from Lieberman DZ et al., Int Clin Psychopharm 2008;23:188-197).

In a post-hoc analysis, the researchers re-crunched the numbers using a controversial approach called MMRM (mixed-effect model for repeated measures). The method is complex, but the bottom line is that it tends to yield larger differences between drug and placebo groups than the standard LOCF (last observation carried forward) method (for a review, see Prakash A et al., Int J Clin Pract 2008;62 (8):1147–1158). Indeed, the authors were able to salvage their data this way, reporting that Pristiq beat placebo by a slim 2.34 points on the Hamilton (p<0.001) according to MMRM.

A tweak alert is in order here, because these post-hoc statistics are frequently done by companies when initial results look bad. In this case, Pristiq may have performed poorly because the high doses used caused a 38% rate of nausea and therefore a higher drop-out rate than Effexor XR. The LOCF method is more likely than MMRM to treat such early drop outs as “non-responders.” Ultimately, the company submitted only the 50 mg dose for FDA approval.

Aside from the efficacy studies, Wyeth has conducted studies showing that Pristiq is unlikely to cause any drug-drug interactions. In two studies of similar design, small numbers of healthy volunteers were coadministered Pristiq along with desipramine (DMI), and were then crossed over to receive Paxil with DMI (Nichols AI et al., J Clin Pharm 2009 ;49(2):219-28) or Cymbalta with DMI (Patroneva A et al., Drug Metab Dispos. 2008;36(12):2484-91). Both Paxil and Cymbalta increased DMI levels whereas Pristiq did not. Pristiq, like several other antidepressants (i.e., Celexa, Effexor XR, Lexapro, Remeron, and Zoloft) is unlikely to cause significant drug-drug interactions.

The bottom line pretty much remains the same as the last time we looked at Pristiq. At the recommended dose of 50 mg per day, Pristiq is marginally more effective than placebo but there is no data comparing this low dose with any other antidepressant. Very high doses of Pristiq appear to be less effective than therapeutic doses of Effexor XR. Pristiq’s documented lack of a dose response relationship is probably the drug’s greatest liability, because it is common practice among psychiatrists to increase the dose of antidepressants when a patient is not responding. Therefore, starting a patient on a medication that yields no benefit at higher doses (but substantially more side effects) seems inadvisable given the plentiful array of other antidepressants available, many of them either in generic form or approaching patent expiration.

TCPR Verdict:

Thumbs down on Pristiq