Lessons from cardiology: inflammation on trial

In cardiology, associations between inflammation and atherosclerosis have long been known. However, as with inflammation and psychiatry, proving that inflammation was causal rather than a bystander was difficult. In 2017, the CANTOS study showed that inflammation was a viable target in cardiovascular disease. ^{Reference Ridker, Everett, Thuren, MacFadyen, Chang and Ballantyne4} The trial enrolled 10 061 participants with previous myocardial infarction and high-sensitivity C-reactive protein (hsCRP) of ≥2 mg/L, randomising them to canakinumab (an anti-IL-1β antibody) or placebo. The trial demonstrated that reducing inflammation helped to prevent cardiovascular events without lowering lipids; however, the US Food and Drug Administration later denied canakinumab a cardiovascular indication, citing reliance on a post hoc hsCRP-responder subgroup, modest benefit and excess fatal infections. More recently, colchicine (an anti-inflammatory agent licensed for treatment of gout) was repurposed and received Food and Drug Administration approval to reduce the risk of cardiovascular events in patients with multiple risk factors. The COLCOT trial (n = 4745) paved the way for approval by showing that colchicine reduced cardiovascular events in patients with recent myocardial infarction. ^{Reference Tardif, Kouz, Waters, Bertrand, Diaz and Maggioni5}

Lessons from neurology: migraine and CGRP

In the 1980s, it was discovered that calcitonin gene-related peptide (CGRP) acts as a potent vasodilator, and that CGRP receptors are localised in the sensory nerve fibres innervating meningeal blood vessels. A breakthrough came in the 2000s, when researchers infused CGRP into humans and showed that the peptide provoked migraine attacks in people with migraine but not in healthy controls. ^{Reference Hargreaves and Olesen6} This led to the development of several therapies targeting CGRP. The first agents were the gepants, such as telcagepant and olcegepant, which act as antagonists at the CGRP receptor. Both showed efficacy but stalled before reaching the market: olcegepant because it required intravenous administration, and telcagepant because of hepatotoxicity. Rather than abandoning the path, the field developed second-generation gepants, as well as monoclonal antibodies targeting CGRP and its receptor.

In the late 2010s and early 2020s, several landmark clinical trials demonstrated that targeting CGRP or its receptor was effective in both prevention and acute treatment of migraine. For example, the ACHIEVE trials (both including >1600 participants) demonstrated the efficacy of ubrogepant in treating acute migraine attacks. ^{Reference Dodick, Lipton, Ailani, Lu, Finnegan and Trugman7} However, a recent study indicated that the role of CGRP might be more biologically complex than previously thought, as plasma levels of circulating CGRP are reduced in people with migraine and do not differ between subtypes of migraine or after preventive treatment. ^{Reference Karlsson, Christensen, Al-Khazali, Haanes, Jørgensen and Ashina8} Nonetheless, CGRP remains the most robust therapeutic target in migraine.

Translational lessons for psychiatry

Common threads in these two examples include a move from associations to more causal experiments in humans, followed by confirmation in large-scale clinical trials. In the cardiovascular case, CANTOS provided evidence that reducing inflammation without lowering lipids could reduce cardiovascular risk; in migraine, infusion of CGRP could provoke attacks. Both novel and repurposed drug candidates were then tested in trials large enough to allow robust conclusions. Importantly, concepts around diagnosis and symptoms were also sharpened. Some of these threads are being picked up in the field of (immuno)psychiatry, whereas others ought to be.

First, a recent randomised trial demonstrated that patients with depression and low-grade inflammation (CRP ≥3 mg/L) experienced increased anhedonia upon infusion of bacterial antigen lipopolysaccharide, showing how a defined symptom (anhedonia) could be provoked by an experimental exposure (lipopolysaccharide). ^{Reference Savitz, Figueroa-Hall, Teague, Yeh, Zheng and Kuplicki9}

Second, many randomised controlled trials are testing (or have recently tested) anti-inflammatory treatments in mental disorders. However, most of these have not been adequately powered to reliably test efficacy and safety. As the examples from cardiology and neurology show, studies with hundreds or thousands of patients are often needed to confirm clinical efficacy. If hypotheses regarding certain subgroups of patients (e.g. patients with immunometabolic depression) are to be tested, large pragmatic trials should ideally be conducted with adequate power to allow biomarker-selected subgroup analyses, distinguishing efficacy in those with and without the phenotype.

Third, important problems in translational efforts are illustrated by CGRP, whose plasma concentration may be unchanged in patients despite evidence of mechanism of action and therapeutic efficacy. This parallels the highly debated role of serotonin in depression. Here, questions remain regarding whether individuals responding to antidepressant treatment are ‘true’ serotonin-deficient depressive patients, whereas non-respondents constitute a subgroup with a different disease mechanism, e.g. an immune-mediated mental disorder.

Fourth, the migraine example highlights how diagnostic categories with conceptual validity are important if biological insights are to be translated into treatment. Current psychiatric diagnostic systems seem imperfect in this regard. Frameworks such as the Research Domain Criteria or approaches that measure core symptoms such as anhedonia rather than an array of non-specific symptoms may provide fertile ground.

The association between inflammation and mental disorders is too strong to simply ignore, but the next step for the field is to test hypotheses in a more causal framework. To do so, psychiatry, luckily, does not need to reinvent the wheel – and we must not be immune to evidence, allowing our hypotheses to be confirmed, refined or abandoned.