Assessing the value of antidepressants in depression

Effects of antidepressant administration on negative biases in emotional processing

Higher depression severity scores are correlated with lower numbers of positive items recalled on a memory task.¹ As such, Dr Murphy questioned whether negative biases in emotional processing are a marker of vulnerability to future depression. Indeed, an online study found that increased recognition of angry faces at baseline significantly predicted increased levels of anxiety and depression 10 months later.²

Neurologically, negative affective biases are associated with corticolimbic circuity imbalances. These can be reversed with a single dose of, or short-term treatment with, an antidepressant³ in both people with depression⁴ and healthy controls.⁵ Negative affective biases can also be induced in animals and a positive affective bias can be shown after antidepressant administration.⁶

Clinical response to antidepressant therapy might be predicted by early changes in emotional processing

Dr Murphy highlighted how early positive improvements in emotional processing might be predictive of clinical response to antidepressant therapy.⁷ In a multi-center study, patients were randomised to treatment as usual (TAU) or to have treatment guided by a predictive algorithm (PreDicT). This took into consideration performance on a facial expression task following 8 weeks treatment then recommended treatment maintenance or change. At Week 24, the PreDicT group showed significant reduction in anxiety symptoms compared to TAU.⁸

These observations led to a neuropsychological theory of antidepressant action whereby early in treatment there is a positive shift in emotional information processing then over time, with repeated exposure to social interactions, increased positive associations improve mood.³

Do rating scales adequately measure depression severity and other factors?

Testing for clinical validity of the 17-item Hamilton Depression Rating scale (HAM-D₁₇⁹) revealed that only six items correlate with clinical assessment of depression and severity distinction: depressed mood, guilt, activities and interests, psychomotor retardation, psychic anxiety and general somatic symptoms.^10,11

Better clinical validity for assessing core depressive symptoms with the HAM-D₆ than the HAM-D₁₇

These findings are important, stressed Dr Østergaard, when measuring efficacy of antidepressants as much lower, or even negative, effect sizes in the ‘stress-related arousal items’ of the HAM-D₁₇ (including insomnia, somatic anxiety, weight loss and psychomotor agitation)^11,12 might be due to the adverse events (AEs) associated with antidepressants, meaning that participants that were not shown to respond on HAM-D₁₇ in clinical trials were actually responders. Indeed, in another study where the more AEs reported, the higher the HAM-D₁₇ rating at the endpoint, this relationship did not hold when HAM-D₆ or the HAM-D item of depressed mood was assessed.¹³

Reappraisal of effectiveness of antidepressant treatment at the symptom level

Dr Näslund showed that the efficacy of selective serotonin reuptake inhibitors (SSRI) compared to placebo is not evidenced in a number of trials when using the HAM-D₁₇ results. However, re-analysis using only the items reflecting core symptoms of depression show significant treatment effects, compared to placebo, in many more of these trials.¹²

Assessment SSRI efficacy in clinical trials might also be dependent on baseline depression severity when utilizing the HAM-D₁₇.¹⁴ However, positive effects of SSRIs might seem to be lower in patients with non-severe depression when assessed by HAM-D₁₇ as, Dr Näslund suggested, many of the assessed items, such as insomnia and weight changes, might not be present in these cases. Considering only the depressed mood item of the HAM-D₁₇, or using the HAM-D₆, reveals similar positive outcomes to patients with severe depression.¹⁵

Efficacy of SSRIs on the HAM-D₁₇ may be masked by impingement of adverse effects on item scores

Dr Näslund also discussed how some SSRI effects may occur within the first 1−6 weeks of therapy,¹⁶ including, on the HAM-D ‘psychic anxiety.’¹⁷ There are also indications of a positive effect of SSRIs on the HAM-D measure of suicidality in adults (aged ≥25 years), with a significant negative effect of placebo treatment on suicidal ideation.¹⁸

Evidencing the value of antidepressant therapy

Dr Hieronymus discussed how in many trials there needs to be a 7-point difference on HAM-D₁₇ scores to note a rating of ‘minimal improvement’ on the Clinical Global Impressions-Improvement scale (CGI-I), an effect size of 0.875.¹⁹ However, he asked, is this a reasonable cutoff as these “are practically unattainable and would require near 100% remission.”²⁰

One issue may be where a clinician-rated HAM-D₁₇ score of ≥22 is needed in a trial and leads to ‘over-rating,’ as evidenced when patients are asked to rate themselves and many score <22.²¹ This, according to Dr Hieronymus, can lead to inflated physician-related change scores. Dichotomizing scores into remission and non-remission may also be problematic as this creates an artificial boundary whereby patients minimally below the cutoff will be deemed non-responders even though scores are very close to responders.²²

Individual variability in response and clinician overrating needs to be considered in antidepressant clinical trial analysis

Another confounding factor is individual variability in response to antidepressants. A meta-analysis found variability ratio to be similar in treatment and placebo groups, concluding that there are no real responders to antidepressants.²³ However, Dr Hieronymus posited that this is an erroneous interpretation of this analysis method as such analyses are not comparing variability but standard deviations. When Dr Hieronymus and his team analyzed data at a patient level, instead of a group level, they found that between group distributions do differ.²⁴