Design: updated systematic review and random-effects meta-analysis
Population: 24 studies of adults with major depressive disorder starting at least 4 weeks of pharmacotherapy
Primary outcome: baseline peripheral biomarker difference between antidepressant responders and non-responders
Effect: no marker significant in primary pooled analysis; interleukin-2 in recent studies SMD −0.81 (95% CI −1.59 to −0.03)

An updated systematic review extended a 2022 meta-analysis through January 2026, pooling 24 studies that measured baseline peripheral biomarkers in adults with major depressive disorder before at least four weeks of pharmacotherapy and compared responders with non-responders. Eight of the studies were new.

In the primary pooled analyses, nothing separated the two groups — including C-reactive protein and interleukin-8, the markers most often invoked. Exploratory analysis found a nominal interaction by publication period for interleukin-2, and in studies published from 2021 onwards baseline interleukin-2 was lower in responders (standardised mean difference −0.81, 95% CI −1.59 to −0.03; P=0.04). Restricting recent studies to selective serotonin and serotonin-noradrenaline reuptake inhibitors, baseline interleukin-6 was also lower in responders (−0.37, 95% CI −0.69 to −0.05; P=0.02). Leave-one-out analysis showed interleukin-6 was moderately sensitive to single studies.

The practical conclusion is negative and worth stating plainly: there is no inflammatory blood test to order before choosing an antidepressant. C-reactive protein in particular has acquired a clinical reputation that this pooling does not support.

The secondary finding is genuinely interesting as science — if newer studies detect a signal older ones did not, either the assays or the populations have changed — but it is a subgroup of a subgroup, defined after the fact.

Do not order C-reactive protein or cytokine panels to guide antidepressant choice
Where inflammatory markers are already raised, treat the cause rather than reading them as a prescribing signal
Keep prediction where the evidence is: previous response, family history, tolerability, comorbidity
If you quote the interleukin findings to colleagues, quote them as exploratory
Expect a wait — biomarker-guided prescribing in depression is not close

Don't overread it

The interleukin-2 and interleukin-6 results are post-hoc subgroup findings in an exploratory analysis, not evidence that measuring them helps.

The statistics, in plain English

The primary analysis is the one to weigh: pooled across all 24 studies, no marker distinguished responders. The positive results come from splitting by publication period and drug class after the data were seen, and both confidence intervals nearly touch zero — the interleukin-2 interval runs to −0.03. 'Nominal' in this paper means unadjusted for multiple comparisons, and with this many subgroups that adjustment would almost certainly remove the finding.