The two major limitations of our study are (1) the open label design; and (2) the small number of subjects in each group, which may have precluded finding an accelerating effect of either medication. We estimated to require 20 patients per treatment arm based on a priori power analysis to detect a 15% difference in the primary outcome (MADRS) scores from the mean with 80% power and an alpha of 0.05. However, due to the stringent exclusion criteria (i.e. first episode, not on antidepressant, and no active comorbid axis I), the enrollment was slow and was terminated before a minimum number of intended-to-treat (ITT) subjects were enrolled. However, this pilot study with treatment group, combined allowed us to evaluate the relationship of thyroid status at baseline to treatment outcome. The study suggests that optimal thyroid function may be associated with faster response to citalopram and perhaps more so in men than in women. Low baseline TSH was associated with shorter time to response, while patients with baseline TSH above the mean were less likely to reach full remission. Moreover, higher baseline free T4 was inversely correlated with the time to response in males. We (Gitlin et al. 2004
) and others (Amsterdam et al. 1996
; Berlin and Corruble 2002
) have reported that lower serum TSH values are associated with better responses to SSRI antidepressants and that high baseline FT4I is associated with a better antidepressant response in men as measured by a shorter length of stay in hospital for male patients (Abulseoud et al. 2007
The significant inverse correlation between baseline free T4 and response time only in males is in agreement with our previous report (Abulseoud et al. 2007
). However, the exact mechanism for this gender discrepancy is not entirely clear. Part of the mixed signals (i.e. heterogeneity) could be attributed to the use of T3 for acceleration and T4 for augmentation. T3, having a short half-life, initiated at the time of starting antidepressant treatment shortens the antidepressant response time, while T4 initiated during antidepressant treatment in refractory cases could augment antidepressant efficacy. Changes in female sex hormone during menstrual cycle could also account for some of subtle thyroid dysregulation as estrogen is known to increase the levels of thyroid-binding globulin with subsequent increase in total and decrease in free thyroxine levels (Tahboub and Arafah 2009
). It could be speculated that males, compared to females, are able to activate the HPT axis and produce more thyroid hormone during a depressive episode, and T3 acceleration effect is noted more in females (Altshuler et al. 2001
). However, further research is needed to better understand if the relative activation in HPT axis is pathologic or compensatory.
Perhaps due to the small sample size, coinitiating T3 or pindolol with citalopram in patients with major depression did not show a significant difference from placebo in reducing the time to response. However, Papakostas et al. (2009
) reviewed five double-blind acceleration studies with T3 and found no statistically significant difference in terms of remission rates or response rates at week 1, week 2, or at endpoint between the SSRI +T3 coinitiation therapy versus SSRI monotherapy in patients with major depression.
This observation is in contrast with the significant effect of T3 in accelerating the antidepressant effect of TCAs (Altshuler et al. 2001
), and the effect of pindolol in accelerating SSRIs. The median survival time until first response in Portella et al. (2011
) meta-analysis was 65% less in the pindolol group (22 days vs. 30 days; P
= 0.03). One explanation for this disparity is that T3 may shorten the response time to TCAs and not SSRIs. Of course, the small sample size of our study may have resulted in a Type II error in evaluating the accelerating response of both T3 and pindolol.
The mean time to response in our sample was only two weeks despite the relatively high baseline MADRS scores (29.7 ± 5.8). This interesting observation is in line with other published meta-analyses of double-blind randomized clinical trials reporting statistically significantly greater response to fluoxetine (Tollefson and Holman 1994
) and both mirtazapine and amitriptyline (Bech 2001
) compared with placebo starting from second week of treatment. This intriguing finding is difficult to fully comprehend in the face of clinical practice. Baldwin et al. (2009
) and others (Stassen et al. 2007
) suggested that the antidepressants differ from placebo in their ability to engender preparedness for recovery, but once triggered, recovery occurs at the same rate, regardless of the nature of the pharmacological treatment intervention.
Despite the negative acceleration results in this study, the main finding suggests that hyperactive hypothalamic pituitary thyroid axis may be associated with better antidepressant response. Further research is needed to further explore this potential marker.