Personalized Antidepressant Therapy: Controversy and Clarification

Unrecognizable mental health professional holds an anti depressant. The male patient is in the background with his head in his hands.
Personalized medicine “promises to move beyond data regarding the average effectiveness of treatments to identify the best treatment for any individual.” However, personalized medicine for depression necessitates “identifying characteristics of individuals that reliably predict differences in benefits and/or adverse effects of alternative depression treatments, including both biological and psychosocial treatments.”

According to the World Health Organization (WHO) depression affects over 264 million people worldwide.1 In the United States, an estimated 17.3 million adults (7.1%) had at least one major depressive episode during 2017.2 Moreover, rates of depression have been increasing3—a trend that was already taking place even before the COVID-19 pandemic. Now, during the pandemic, rates have dramatically risen, according to a new report issued by Mental Health America.4

These staggering numbers are continuing to climb, despite a growing armamentarium of treatments for major depressive disorder (MDD). In fact, only about approximately one-third of patients with MDD reach full remission after one treatment trial and only two thirds reaching remission after four treatment trials.5 

Although antidepressants show similar efficacy, patients vary in response to different medications and may respond to a second medication despite not responding to the first.6 However, many patients do not have the opportunity to receive second-line treatments,6 so accurate selection of the best initial antidepressant is critical.

Personalized medicine “promises to move beyond data regarding the average effectiveness of treatments to identify the best treatment for any individual.”6 However, personalized medicine for depression necessitates “identifying characteristics of individuals that reliably predict differences in benefits and/or adverse effects of alternative depression treatments, including both biological and psychosocial treatments.”6

Madukhar Trivedi, MD, who is the Betty Jo Hay Distinguished Chair in mental health, UT Southwestern Medical Center, Dallas, Texas, has long been an advocate for precision medicine in treating depression.

“We have spent the last 25 years as a field using clinical measures and clinical approaches to depression treatment, but that has not been very fruitful because we have not matched specific patients with specific treatments,” he told Psychiatry Advisor.

This lack of precision has led to a “cycle of trial and error,” in which patients might “wait as long as several weeks for a drug to kick in, and then may not have an adequate response, so they will be switched to a different drug and have to repeat the same process,” said Dr Trivedi, who is also the Julie K. Hersh Chair for depression research and clinical care.

“My work has focused on bringing biological measures into this inquiry, which can then be coupled with clinical markers,” he added.

Controversies in Precision Medicine for MDD

Some research has called into question the feasibility of precision medicine for MDD. For example, a meta-analysis7 of treatment outcomes between drug arms and placebo arms in 163 randomized, placebo-controlled trials (n = 51,396 patients) investigated whether patient-by-treatment interaction led to larger variances in drug arms than placebo arms. The researchers ran “simple simulations” that assumed different definitions and rates of patients who respond particularly well to antidepressants. The variance ratios and coefficient of variance ratios of the individual trials were combined. The meta-analysis looked not only at antidepressants as a whole but also at specific classes of antidepressants and specific antidepressants.

The researchers found a lack of increased treatment-outcome variance in the antidepressant vs the placebo groups, indicating “that no or only very small subgroups of patients respond particularly well to antidepressants,” suggesting that “the scope for personalized treatment with antidepressants seems to be limited.”7

A recent meta-analysis8 of 169 randomized controlled trials (n=58,687 patients) utilized effect sizes log variability ratio and log coefficient of variation ratio to “analyze the difference in variability of active and placebo response,” and used Bayesian random-effects meta-analyses to “estimate the treatment effects variability between antidepressants and placebo.” The findings suggested that “substantial treatment effect heterogeneity…seems rather unlikely.”8

However, another meta-analysis9 of 87 randomized placebo-controlled trials (n=17,540 patients) also analyzed coefficients of variation for antidepressants and placebo, and then calculated their ratios to compare outcome variability between the antidepressant and placebo, using a random-effects model.

The researchers found variability in response to antidepressants, compared to placebo, which was moderated by baseline severity of depression. Variability in response to selective serotonin reuptake inhibitors (SSRIs) was found to be lower than variability on response to noradrenergic agents.

The analysis was subsequently retracted10 by the publication because the analysis was based on “an incorrect analysis (coefficient of variation ratios).” However, “when a proper analysis is used (random-slope mixed-effects model), the original findings are no longer valid.”10

Benoit Musant, MD, MS, Labatt Family Chair, department of psychiatry and professor of geriatric psychiatry, University of Toronto and senior author of the analysis told Psychiatry Advisor that he and his colleagues “conducted the analysis and published the paper in good faith.”

“To our knowledge, this was only the third time this analytical approach was used with psychiatric data (and the second time with treatment data) [but] after the paper was made available online, two different groups sent letters to the editor pointing out potential flaws in our analytical approach,” he continued.

“After conducting additional analyses, we agree that the analytical approach we used yielded incorrect results. We communicated this to the editor and he made the decision to retract the paper,” Dr Musant stated, adding, “Honest errors happen in science; it is to the benefit of the scientific community that these errors are identified and corrected.”

Dr Trivedi regards the analysis as having been affected by statistical flaws but emphasized that it does not detract from the evidence pointing to the validity and role of personalized medicine for MDD. In fact, his research lends support to the stance that depression is a heterogeneous disease with multiple subtypes that vary considerably from one patient to the next, and treatment cannot therefore be a “one-size-fits-all” process.

A Simple Blood Test Can Provide Clues to Depressive Subtype

Trivedi and colleagues have demonstrated the potential of blood tests to identify biomarkers that may flag differential response to antidepressants.

For example, there are patients whose depression is associated with inflammation, “and this can be evaluated through a simple blood test that can detect inflammatory biomarkers,” Dr Trivedi said.

Dr Trivedi has honed in on C-reactive protein (CRP), a plasma protein synthesized by the liver, which is “sensitive to inflammatory cytokines and increases markedly…in acute response to serious infection or tissue injury.”11 Patients with elevated inflammatory markers have shown a lower likelihood of response both to psychotherapy and SSRIs.11

Dr Trivedi and colleagues studied patients with depression who were randomized to receive either SSRI monotherapy (escitalopram plus placebo) or bupropion plus escitalopram (n=51 and n=55, respectively).11

They found that the treatment arms did not differ in depressive symptom or side effect outcomes; however, higher baseline CRP levels were associated with lower depression severity in patients treated with the bupropion-SSRI combination but not with SSRI monotherapy.11

Dr Trivedi called CRP “a clinically pragmatic measure of inflammation” because it is “readily available, inexpensive, relatively stable in stored biological specimens, unaffected by day or meal intake, and has a stable year-to-year in individual subjects, in the absence of acute factors.11

“The risk is low and the benefit is significant,” he said.

He noted that patients who have higher baseline levels of CRP also fare better when a dopaminergic agent is combined with an SSRI than when they are treated with SSRI monotherapy.11

Another marker under investigation by Dr Trivedi’s team is platelet-derived growth factor (PDGF), which is integral to the maintenance of the blood-brain barrier (BBB). PDGF increases when the BBB is disrupted and is associated with neuroinflammation.12 In a study comparing antidepressants, they discovered that bupropion-plus-escitalopram selectively improved anhedonia, which in turn resulted in improved overall depression severity in depressed patients with elevated PGDF levels, as compared with selective serotonin reuptake inhibitor monotherapy or venlafaxine-plus-mirtazapine.12

The potential role of inflammation in informing antidepressant therapy is not confined only to pharmacologic selection. Tumor necrosis factor-α (TNF-α), another inflammatory marker, which is elevated in patients with rheumatoid arthritis, is also elevated in some patients with MDD.13 Dr Trivedi reported that he and his team have studied exercise as a potential treatment for depression and found that it alleviated symptoms in patients with MDD who had elevated TNF-α levels, but not in MDD patients with normal TNF-α levels.14

“An Exciting Time”

To date, Dr Trivedi’s research has identified additional domains of investigation to narrow down the particular subtype of a given patient and try to match it to the most appropriate treatment. The Establishing Moderators and Biosignatures of Antidepressant Response in Clinical Care (EMBARC) study investigates several of these domains.15,16

The EMBARC study takes into account sociodemographic as well as “clinical moderators,” such as anxious depression, early trauma, gender, melancholic and atypical depression, the presence of an Axis II disorder, hypersomnia/fatigue, and chronicity of depression.15 Additionally, it studies “biological moderators and mediators,” including cerebral cortical thickness, task-based fMRI (reward and emotion conflict), resting connectivity, diffusion tensor imaging (DTI), arterial spin labeling (ASL), electroencephalography (EEG), cortical evoked potentials, and behavioral/cognitive tasks.15

“This is an exciting time to do research because we have identified these domains and are beginning to get very strong signals,” Dr Trivedi said.

“I would say that currently, we are on the threshold of being able to use several of these measurements, but that is still in its initial phase,” he continued. “Nevertheless, it is currently clear that blood tests, imaging, and EEG seem to be tools that are becoming much more ready to use although more research needs to be done.”


1.  World Health Organization (WHO). Depression. Updated January 30, 2020. Accessed: June 30, 2020.

2.  National Institute of Mental Health (NIMH). Depression. Updated February 2019. Accessed: June 30, 2020.

3.  Weinberger AH, Gbedemah M, Martinez AM, Nash D, Galea S, Goodwin RD. Trends in depression prevalence in the USA from 2005 to 2015: widening disparities in vulnerable groups. Psychol Med. 2018;48(8):1308-1315.

4.  Mental Health America. Mental Health and COVID-19: More than 169,000 people impacted by anxiety and depression. Updated July 1, 2020. Accessed: July 17, 2020.

5.  Trivedi MH, Daly EJ. Treatment strategies to improve and sustain remission in major depressive disorder. Dialogues Clin Neurosci. 2008;10(4):377-384.

6.  Simon GE, Perlis RH. Personalized medicine for depression: can we match patients with treatments?. Am J Psychiatry. 2010;167(12):1445-1455. doi:10.1176/appi.ajp.2010.09111680

7.  Plöderl M, Hengartner MP. What are the chances for personalised treatment with antidepressants? Detection of patient-by-treatment interaction with a variance ratio meta-analysis. BMJ Open. 2019;9(12):e034816.

8.  Volkmann C, Volkmann A, Muller CA. On the treatment effect heterogeneity of antidepressants in major depression. A Bayesian meta-analysis. BMJ Yale. Published online April 7, 2020.

9.  Maslej MM, Furukawa TA, Cipriani A, Andrews PW, Mulsant BH. Individual Differences in Response to Antidepressants: A Meta-analysis of Placebo-Controlled Randomized Clinical Trials. JAMA Psychiatry. 2020;77(6):1-12. 

10.  Öngür D, Bauchner H. Notice of Retraction: Maslej et al. Individual Differences in Response to Antidepressants: A Meta-analysis of Placebo-Controlled Randomized Clinical Trials. JAMA Psychiatry. 2020;77(6):607-617.

11.  Jha MK, Minhajuddin A, Gadad BS, et al. Can C-reactive protein inform antidepressant medication selection in depressed outpatients? Findings from the CO-MED trial. Psychoneuroendocrinology. 2017;78:105-113.

12.  Jha MK, Minhajuddin A, Gadad BS, Trivedi MH. Platelet-Derived Growth Factor as an Antidepressant Treatment Selection Biomarker: Higher Levels Selectively Predict Better Outcomes with Bupropion-SSRI Combination. Int J Neuropsychopharmacol. 2017;20(11):919-927.

13.  Jha MK, Trivedi MH. Personalized Antidepressant Selection and Pathway to Novel Treatments: Clinical Utility of Targeting Inflammation. Int J Mol Sci. 2018;19(1):233.

14.  Rethorst CD, Toups MS, Greer TL, et al. Pro-inflammatory cytokines as predictors of antidepressant effects of exercise in major depressive disorder. Mol Psychiatry. 2013;18(10):1119-1124. doi:10.1038/mp.2012.125

15.  Trivedi MH, McGrath PJ, Fava M, et al. Establishing moderators and biosignatures of antidepressant response in clinical care (EMBARC): Rationale and design. J Psychiatr Res. 2016;78:11-23. 

16.  Webb CA, Trivedi MH, Cohen ZD, et al. Personalized prediction of antidepressant v. placebo response: evidence from the EMBARC study. Psychol Med. 2019;49(7):1118-1127.