By Shalom Feinberg, M.D.

Most psychiatrists would agree that the first step in treating depression is to start one of SSRIs, because they are effective, safe, have a broad spectrum of both anti-anxiety and antidepressant activity, and are often cheap, since all SSRIs but Lexapro are now available generically. How cheap is cheap? At Walmart pharmacies, you can now purchase fluoxetine for about 13 cents a pill, or only $4 per month for the original “miracle” antidepressant.

However, we know that many patients will not respond or will not tolerate an SSRI, and the STAR*D study was supposed to provide guidance on what to do next. But, as you have read elsewhere in this issue, STAR*D’s results fell short of these expectations. Underlining the inadequacy of the existing database on this issue, a just published systematic review of all available research on antidepressant switching options for SSRI nonresponders concluded there are no distinct recommendations on how to next treat these patients (Ruhe et al J Clin Psychiatry 2006;67:1836-1855). We’re left with our clinical experiences and preferences, along with a smattering of relevant research that has been published over the years.

Here are TCPR’s guidelines for how to deal with SSRI non-response:

Patience is a virtue for some patients, but counterproductive for others.
One useful clinical nugget from STAR*D is the reminder that chronically ill or recurrently depressed patients may take many weeks to get well. Patients on Celexa (citalopram) in Step One of STAR*D required an average of 5.7 weeks to respond, and 6.7 weeks to fully remit, despite a fairly aggressive dosing titration schedule leading to an average dose of 41.7 mg/day (Trivedi et al, Am J Psychiatry 2006;163:28-40).

On the other hand, recent studies have shown that up to 60% of acutely depressed will show onset of a response to an antidepressant within the first two weeks, which challenges the clinical lore that antidepressants take four weeks to kick in (see, for example, the recent review in Katz et al, J Clin Psychopharmacology 2006; 26:549-553). Thus, if you are treating a patient who is fairly new to psychiatric treatment, and there is absolutely no response within two weeks on a therapeutic dose , chances are low that they will ever respond to that agent, and you should seriously consider changing strategies. The STAR*D patients probably fit into a very different category, since they were chronically depressed, with an average continuous period of depression of two years prior to study entry.

Bump up the dose? Maybe.

Just about every review article you will read on treatment resistant depression will suggest “dose optimization,” meaning that you should increase the dose until there is either a response or limiting side effects. However, the actual data in support of this practice is meager. For example, a recent systematic review of eight controlled trials that involved dose escalation of SSRIs found no value in increasing the SSRI dose within the first four weeks. After eight weeks, increasing the dose yielded some limited benefit, but only in patients who had already shown at least a partial response (Ruhe et al, Br J Psychiatry, 2006, 189:309-316.)

So where does this leave us? The fact is that we all have experience with patients who tell us they have improved when we have increased their doses, no matter what the studies say. A reasonable approach, then, is to escalate the SSRI dose by about half the “unit dose” (eg., the unit dose of fluoxetine is 20 mg, of sertraline 50 mg, etc….) if there is incomplete response after 2 to 4 weeks, and to keep escalating until you achieve either: A) A satisfactory response; or B) Intolerable side effects.

Switch or Augment?

This was the major disappointment of STAR*D, namely, that it provided no guidance on when to make this crucial decision. So we fall back on clinical lore, which offers the following:

Switch meds for patients who:
-Have had no response whatsoever,
-Are unable to tolerate a given SSRI, or
-Prefer a simpler (and cheaper) single drug regimen rather then taking two agents.

Augment for patients who:
-Have only a partial response,
-Have failed a number of individual drug switches, or
-Have more severe depressive illness.

In this article we don’t have room to delve into the vast world of augmentation strategies, but we recommend an excellent recent article by Debattista for those who are interested (Debattista, J Psychopharmacology 2006 ;20, No. 3 suppl: 11-18).

Switch to What?

Is it worthwhile to switch to a different SSRI or should we go into a different class immediately? There are two common scenarios here: Switching because of intolerance to the first SSRI, and switching due to lack of response.

It is clear that in both the STAR*D trial and in other studies, some patients intolerant to an initial SSRI trial may in fact tolerate and respond to a second SSRI (see review by Ruhe et al in J Clin Psychiatry referred to above). Other studies have shown that patients who tolerate but do not respond to the first SSRI will often respond to second. For instance, in one of the better performed SSRI switch studies, 57 acutely depressed patients with at least 6 weeks of nonresponse to Prozac (fluoxetine, avg. length of nonresponse, 5.9 months, avg. dose, 31 mg/day) were immediately switched to Celexa (citalopram, up to 60 mg/day, avg., 39 mg/day) . In this trial, 63% of patients responded to Celexa over 12 weeks (Thase, J Clin Psychiatry 2001;62: 683-687)

But this study, like all other switch studies in the literature, was neither controlled nor double-blind. Because of this, we have no way of knowing for sure whether switching to a different SSRI is any more effective than simply staying the course on the original medication.
Switching out of class.

When switching out of class, most psychiatrists choose one of the SNRIs (serotonin norepinephrine reuptake inhibitors), either Effexor XR (venlafaxine XR) or Cymbalta (duloxetine). There is more data supporting the efficacy of switching to Effexor (see, for example, Poirier et al, Br J Psychiatry 1999;175:12-16 and Baldomero et al, Depression Anxiety 2005; 22:68-76) but this may simply reflect the fact that Effexor’s been around longer than Cymbalta. Cymbalta has some advantages over Effexor, including an easier titration schedule and a lower rate of severe discontinuation symptoms (Perahia et al , J Affective Dis 2005;89:207-212).

While we would all like to believe that switching to Wellbutrin works well (because its lack of sexual side effects), there’s not much data to support its efficacy in treatment-resistant depression. In STAR*D, the Wellbutrin switch did as well as Effexor XR and Zoloft, but the lack of blinding or placebo makes all of STAR*D’s results questionable. Switching to Remeron (mirtazepine) is also an option, although its side effects of weight gain and sedation make it unappealing to many.

If the SNRI switch is ineffective, we recommend going to one of the “bigger guns”—either a tricyclic (TCA) or a monoamine oxidase inhibitor (MAOI). Of course, these switches can be tricky because of drug interactions. Both Prozac and Paxil can increase serum levels of tricyclics, whereas the other SSRIs are safe. All SSRIs interact with MAOIs, necessitating the dreaded “antidepressant gap” of at least two weeks (five weeks for Prozac). Anecdotally, some patients weather this transition better if prescribed lithium, lamotrigine, an atypical antipsychotic, or a benzodiazepine—all of which are perfectly safe when combined with MAOIs.

Of course, if switching isn’t working, you should re-evaluate the diagnosis. Look for subtle signs of psychosis, bipolar disorder, substance abuse or any other psychiatric disorder that would alter the treatment plan.

TCPR Verdict:

SSRI Nonresponse: No clear guidelines, but plenty of options.