Fluoxetine’s introduction in 1987 marked a watershed year for psychiatry. Not only was it an antidepressant with a novel mechanism of action, but it also opened the door to antidepressant combinations—a strategy that was too dangerous to undertake when tricyclics and MAOIs were the only available pairings. It wasn’t long before pilot studies emerged touting the possible benefits of combining tricyclics with fluoxetine for treatment-resistant depression.

But despite the interest, research into this strategy never really got off the ground. The reason? Wyeth introduced venlafaxine in 1993, a serotonin-norepinephrine reuptake inhibitor (SNRI) with a pharmacodynamic effect that mirrored the dual reuptake showcased in the SSRI-tricyclic combination. Wyeth invested heavily in an idea that everyone assumed would be true: that two neurotransmitters would be better than one.

Are two neurotransmitters better than one?

With three decades of research now behind us, we can see how well the hope of dual-neurotransmitter action held up.

At best, SNRIs are only marginally more effective than SSRIs. At most, only about one in 24 patients see a meaningful difference with the dual mechanism. Most famously, patients who failed to respond to citalopram in the STAR*D study fared no better when switched to venlafaxine (n=250) than those switched to bupropion (n=239) or sertraline (n=238) (Rush AJ et al, N Eng J Med 2006;354(12):1231–1242). Venlafaxine also has a higher rate of side effects and a greater propensity for withdrawal problems than many of the SSRIs, which arguably outweighs any marginal benefits it may offer. In sum, two neurotransmitters didn’t work much better than one. How about two antidepressants?

Combining antidepressants at the onset of a trial

Clinicians rarely start two antidepressants at the same time, but this strategy could be useful if it were to bring about greater recovery or faster response. Two large, NIMH-sponsored trials have tested that possibility.

The Combining Medications to Enhance Depression Outcomes (CO-MED) study was a rigorous single-blind trial that randomized patients with depression to receive escitalopram + placebo (n=224), escitalopram + bupropion (n=221), or venlafaxine + mirtazapine (n=220) over the course of 12 weeks. The results were unequivocally negative: Remission rates in the three cohorts were 39%, 39%, and 38%, respectively (Rush AJ et al, Am J Psychiatry 2011;168(7):689–701). A similar randomized trial looked at whether the escitalopram-bupropion combination was more effective than either of those antidepressants alone, but also had negative results (n=245) (Stewart JW et al, J Psychiatr Res 2014;52:7–14).

Combining antidepressants when the initial antidepressant doesn’t work

Combining antidepressants from the start clearly does not confer any advantages, but what about adding a second antidepressant after the first one proves ineffective? In addressing this question, we are confronted with numerous challenges. There are countless combinations that have been tried, but most of these studies have small sample sizes, are industry sponsored, and have widely varying methodologies. Therefore, I’ll focus on the largest and most rigorous studies.

Bupropion is the most popular augmentation option. Intuitively, this makes sense. Bupropion enjoys a benign side effect profile, has a unique mechanism of action, and can improve energy and concentration. But intuition does not always line up with science. Although three early trials were positive, they were small (n<60) and did not hold up to the rigor of a large controlled trial. In an unpublished trial sponsored by the manufacturer of bupropion, patients responded similarly to bupropion augmentation and placebo augmentation after failing a single antidepressant trial (n=325) (referenced in Gulrez G et al, Basic Clin Pharmacol Toxicol 2012;110(3):227–230).

In the government-funded VAST-D trial, outpatients who failed to respond to an initial trial of an SSRI, an SNRI, or mirtazapine were randomized to switch to bupropion alone (n=511) or to have their initial antidepressant augmented with bupropion (n=506) over 12 weeks. Remission rates favored bupropion augmentation (27% vs 22%), but this difference did not reach statistical significance (p=0.09).

Mirtazapine is the second most popular augmentation agent, and its unique mechanism of action makes this strategy appealing. Mirtazapine also has sedating and appetite-stimulating effects that would seem ideally suited for depressed patients with insomnia and decreased appetite.

Two adequately powered, blinded mirtazapine augmentation studies have been published to date. Kessler and colleagues randomized 480 depressed outpatients who had failed to respond to at least six weeks of treatment with an SSRI or SNRI to augmentation with mirtazapine or placebo. After 12 weeks, depression scores were lower in the mirtazapine cohort, but this advantage was not statistically significant (p=0.09) and faded further as the study continued for another nine months (Kessler DS et al, BMJ 2018;363:k4218).

In the second study, Kato and colleagues randomized 1,647 patients who had not responded to three weeks of treatment with sertraline to continue the SSRI (n=551), switch to mirtazapine (n=558), or augment with mirtazapine (n=538) for an additional six weeks. Response rates were 43%, 50%, and 52%, respectively. The SSRI-mirtazapine combination was superior to SSRI monotherapy (p<0.01), but not to mirtazapine monotherapy (p=0.79) (Kato T et al, BMC Med 2018;16(1):103).

Conclusion

It is rare to see so many studies align in the same direction in psychiatry. There is a signal that two antidepressants (or two neurotransmitters) are better than one, but this signal remains far too faint to be clinically discernible. This is an unsettling finding that doesn’t match what most of us do in clinical practice. If we add these results to other paradoxical findings that have recently emerged from the literature—for example, that neither switching antidepressants (Bschor T et al, J Clin Psychiatry 2018;79(1):16r10749), nor increasing the dosage of an antidepressant (Rink L et al, J Clin Psychiatry 2018;79(3):17r11693) have been found to be effective—we are forced to confront the question: Why not?

One possible explanation is that antidepressants work “downstream,” such as through changes in dopamine or BDNF, and that their initial effects on serotonin or norepinephrine simply don’t matter. This would explain why combining or switching antidepressants isn’t helpful for many. Another explanation is that antidepressants improve various symptoms of depression, but do not correct any biochemical abnormality.

Seven decades after imipramine was introduced, we still know very little about the biological underpinnings of depression. Now that the seductive theories that antidepressants correct some type of chemical imbalance have long since been abandoned, the original christening of imipramine as an “antidepressant,” a disease-modifying agent, may have lulled our minds and sent us down the wrong path. As an analogy, Tylenol reduces our fever during an infection and helps us feel better, but nobody argues that Tylenol has any impact on the disease course itself.

CARLAT VERDICT

Two antidepressants (or two neurotransmitters) are rarely better than one, and any advantage is far too faint to be clinically discernible. With the exception of adding sedating antidepressants for patients with insomnia, you should steer clear of this practice.