To date, there are no large published randomized clinical trials involving direct comparisons of several antidepressants from the same or different classes with large enough numbers of patients in each arm to detect small but clinically meaningful differences. In the absence of such trials, several meta-analytic techniques have been employed to detect differences between classes of antidepressant drugs (e.g., see Anderson21
This meta-analysis involved a cross-section of depressed patients from Europe and North America who took part in 10 double-blind randomized clinical trials involving escitalopram. About two-thirds of the population were women, and almost half had depression categorized as “severe,” defined by a baseline MADRS score of 30 or more. These trials reflect a balance between primary care and specialist settings (4 primary care, 4 specialist and 2 mixed), as well as a wide range of active comparator antidepressants (4 citalopram, 2 fluoxetine, 1 paroxetine, 1 sertraline and 2 venlafaxine XR). The sample was large enough to allow separate analyses of the severely depressed group on each of the contrasts defined a priori, as well as exploratory analyses of potentially influential variables such as fixed or flexible dosing schedule and the presence or absence of placebo control.
The principal finding in this meta-analysis is that escitalopram consistently demonstrated greater efficacy, as assessed by MADRS, on a series of end-point comparisons involving change in scores from baseline and in response and remission rates. The improvement in MADRS with escitalopram was 1.22 points greater than with conventional SSRIs. This magnitude of difference is comparable to the effect size of 1.2 points on the HAM-D scale found in meta-analyses of venlafaxine compared with conventional SSRIs.11
Based on the published results of 3 individual pivotal trials and preliminary data from a fourth,13
the superiority of escitalopram versus other SSRIs, particularly citalopram, has been questioned.22
A pooled analysis from 4 trials showed a significant superiority of escitalopram versus citalopram,9
and this has been confirmed in a direct comparison of escitalopram and citalopram in severely depressed patients.23
In the present meta-analysis, the estimated difference in treatment effect between escitalopram and other agents also increased with severity of depression at baseline. For example, in patients with a baseline severity score above 30 on the MADRS, the separation between escitalopram and comparators was 2.34 points in favour of escitalopram. This was also true for rates of response and remission, where the difference between escitalopram and comparators was significantly greater in the severely depressed population compared with the total population. This was a secondary analysis, so these results must be interpreted with caution, although previous studies conducted among patients with severe MDD have demonstrated that antidepressant effects and response rates are lower than those observed in less severely depressed patients.24
Placebo response rates are also lower in patients with severe MDD,25
so that randomized clinical trials conducted in this population are more sensitive in demonstrating the efficacy of an antidepressant. In these circumstances, the impact of potential confounding factors may be less important and, consequently, the observed effects may reflect more precisely the true antidepressant effect. The greater efficacy of escitalopram in the severely depressed population found in this comprehensive pooled analysis, using a consensus definition of severe depression,26
extends previous findings where escitalopram was shown to be more efficacious than citalopram.25
These results suggest that some heterogeneity exists within the class of SSRIs in terms of magnitude of antidepressant effect.27
What are the potential explanations for the apparently superior efficacy of escitalopram versus conventional SSRIs, particularly in the treatment of more severe depression? One explanation that has recently been proposed relates to an allosteric modulation of the serotonin transporter following administration of escitalopram compared with citalopram.28
In addition to a primary, high-affinity binding site that mediates the inhibition of serotonin reuptake, there is a low-affinity allosteric site that modulates the affinity of ligands at the primary site.29
Recent work has shown that escitalopram, when bound to the allosteric site, appears to potentiate its own binding to the primary binding site. R-citalopram, however, also potentiates the binding of escitalopram to the primary binding site, but to a lesser extent than escitalopram.28
These results provide an alternative explanation to the hypothesis that dual reuptake inhibition of serotonin and norepinephrine is necessarily associated with superior antidepressant efficacy of venlafaxine compared with SSRIs. Escitalopram decreases its own dissociation rate from the serotonin transporter, possibly via the allosteric site,28
leading to more prolonged inhibition of the transporter and higher extracellular serotonin levels. A persistent increase in serotonin levels may be essential for the antidepressant effect. Thus, although venlafaxine may be more efficacious than most SSRIs, the proposed mechanism of action of escitalopram may explain why it is as efficacious as venlafaxine, with the superior tolerability of an SSRI.4,5
A potential limitation of the present meta-analysis relates to an inconsistency in duration of treatment. Although most of the data come from week 8 (8/10 studies), in the 2 long-term trials week 24 and week 27 end-point data were used. Although these long-term end points were slightly better, the sensitivity analysis performed on week 8 data from all 10 studies showed that the results still hold. The same is true if the 2 long-term trials are excluded from the meta-analysis.
When evaluating the ad hoc analyses, we must consider whether receiving placebo or not could affect treatment outcome due to patient selection bias and patient expectations.30
Thus, whereas the analyses show that escitalopram is superior to active comparators in placebo-controlled studies, as well as in fixed-dose and high-dose studies, statistical comparisons between these factors have not been performed and can only be considered as explorative in nature. These factors are relevant for randomized clinical trials and may not be generalizable to treatment in primary care.
As in other published meta-analyses,10,11
there is a disproportionate weighting toward 1 or 2 comparators. In this case, citalopram (n
= 577) accounted for the majority of patients who received an active SSRI comparator as compared with fluoxetine (n
= 262), paroxetine (n
= 156) and sertraline (n
= 107), and the overall results do not necessarily reflect a significant difference between escitalopram and each SSRI. However, the potential for a disproportionate influence of one or more studies is limited, because the number of patients in each study was roughly similar, varying from 194 (study 9) to 339 (study 4). Finally, although 1 late-life depression study was included in the present meta-analysis and some patients over 65 years of age participated in the other studies, there are insufficient data to apply these results to the elderly population.
What is the clinical relevance of these results? A treatment difference between drug and placebo of at least 2 points on the MADRS is usually considered clinically significant. In this pooled analysis, the estimated mean treatment difference of 1.07 points on the MADRS is small, but statistically significant. For patients with a more severe baseline depression (MADRS ≥ 30), who accounted for almost half of the patients, the estimated mean treatment difference is 2.34 points. In the 5 placebo-controlled escitalopram trials in MDD,6–8,12,31
the adjusted mean treatment difference on the MADRS is 3.0 points, and the LOCF response rates after 8 weeks are 37.3% for placebo (n
= 738) and 52.9% for escitalopram (n
= 851). For severely depressed patients, the corresponding adjusted mean treatment difference on the MADRS is 3.3 points, with response rates of 34.5% for placebo (n
= 333) and 50.8% for escitalopram (n
= 384). This corresponds to a difference in response rates of over 15%, which is considered to be a clinically meaningful difference.32
In a trial with severely depressed patients, an adjusted mean treatment difference on the MADRS of 2.1 points corresponds to a difference of 14.6% in response rates (76.1% for escitalopram and 61.5% for citalopram).23
It has been noted that even a modest difference between treatments in the proportion of patients achieving remission is likely to be associated with advantages in important “real-world” domains.30
In conclusion, in this meta-analysis, escitalopram had greater efficacy compared with the comparators (citalopram, fluoxetine, paroxetine, sertraline and venlafaxine XR), as assessed by MADRS on a series of end-point comparisons involving change in efficacy scores from baseline and in response and remission rates. The proposed mechanism of action of escitalopram may explain its enhanced efficacy compared with conventional SSRIs. Given its favourable tolerability profile based on withdrawals due to adverse events, these results suggest that escitalopram may have an improved benefit–risk ratio compared with other antidepressant medications.