Tuesday, August 4, 2026

Opparounds

🔬Psychiatric Research Article

Interleukin 6 as a Treatment Target for Depression: A Proof-of-Concept Randomized Clinical Trial

Éimear M. Foley, Nicholas Turner, Ruta Margelyte, Hannah J. Jones, Muzaffer Kaser, Glyn Lewis, Peter B. Jones, Golam M. Khandaker · JAMA Psychiatry · May 2026

Thirty adults with difficult-to-treat depression and evidence of low-grade inflammation were randomized in this four-week, double-blind, placebo-controlled proof-of-concept trial to a single infusion of tocilizumab — an IL-6 receptor antagonist used in rheumatoid arthritis — or placebo. The primary outcome, somatic symptoms at day 14, was null (mean difference -0.12, 95% CI -2.51 to 2.28), as was depression severity.

The authors are explicit that the trial was not powered for statistical significance and designed instead to estimate effect sizes for a future trial. On that basis the day-28 signals are what they report: remission 53.9 percent with tocilizumab versus 31.3 percent with placebo (NNT 5), and response 46.2 versus 18.8 percent (NNT 4), in 29 completers.

💡 Why it matters

This is a 30-person trial with a negative primary endpoint, and the honest reading is that it tells us how big a real IL-6 trial needs to be — not that IL-6 blockade treats depression. Quote it that way. What it does support is the narrower, more defensible clinical habit: in difficult-to-treat depression, check a CRP before assuming the next step is another monoamine agent, because the inflamed subgroup is where this literature keeps pointing.

Read the paper → doi:10.1001/jamapsychiatry.2026.1053

📖Psychotherapy Research Article

Baseline Social Inhibition and Age as Moderators of the Within-Patient Alliance-Outcome Link in Treatments for Depression

Aditya H. P. Mehta, Michael J. Constantino, Alice E. Coyne, R. Michael Bagby, Paula Ravitz, Carolina McBride, et al. · Psychotherapy Research · July 2026

Eighty adults with major depression received 16 sessions of either cognitive-behavioral or interpersonal psychotherapy, with alliance and symptoms measured repeatedly. The analysis was deliberately within-patient: not whether patients with better alliances do better overall, but whether a given patient's alliance rising above their own average predicts their subsequent symptom change.

It did. And the effect was moderated by baseline interpersonal style — the alliance-outcome link was strongest in patients high in social inhibition, the cold-and-avoidant octant of the interpersonal circumplex. Age, examined exploratorily, did not moderate the association.

💡 Why it matters

The moderation runs opposite to clinical intuition. The socially inhibited patient — reserved, hard to read, the one who makes you wonder whether therapy is landing — is precisely the patient for whom shifts in the alliance carry the most predictive weight. With that patient, attending to the relationship is not a detour from the protocol; on this evidence it is where the leverage is.

Read the paper → doi:10.1080/10503307.2026.2705902

💊Psychiatric Fact

Choosing an antidepressant for a woman on tamoxifen is an oncology decision

Tamoxifen is a prodrug. Its therapeutic activity comes almost entirely from endoxifen, a metabolite roughly 100 times more potent than the parent compound, and the conversion is performed by CYP2D6. Any drug that inhibits 2D6 therefore reduces endoxifen concentrations, and the reductions are not subtle: paroxetine and fluoxetine cut endoxifen levels by 55 to 70 percent, bupropion and duloxetine are moderate inhibitors with intermediate effects, and 2D6 poor metabolizers reach low endoxifen concentrations regardless of what else they take.

Whether this translates into breast cancer recurrence has been genuinely contested — a large Canadian cohort found excess breast cancer mortality with paroxetine coprescription, while other cohorts and a Danish study found no effect — and the outcome literature has never been settled. That ambiguity is exactly why the prescribing decision is easy: there is no clinical reason to accept the uncertainty when equally effective alternatives exist. Venlafaxine is the classic choice, with citalopram, escitalopram, desvenlafaxine, and mirtazapine all being weak or negligible 2D6 inhibitors. Venlafaxine and gabapentin also happen to be the best-evidenced treatments for tamoxifen-induced hot flashes, which is frequently the presenting complaint. If a patient arrives already stable on fluoxetine and about to start tamoxifen, the conversation belongs with her oncologist rather than being settled unilaterally in either direction — but it should happen before the first dose, not at the first follow-up.

🛋️Psychotherapy Teaching Pearl

A behavioral experiment that cannot fail has not tested anything

The most common way CBT is done badly by competent people is that the behavioral experiment is designed to succeed. The patient predicts catastrophe, the therapist quietly chooses a task mild enough that catastrophe will not occur, catastrophe does not occur, and everyone agrees the belief has been disconfirmed. Nothing has been learned, because the experiment never risked the prediction.

A real experiment has four elements and the first one is where the work lives. The prediction must be specific, falsifiable, and rated for belief before the fact: not 'it will go badly' but 'I will start crying, at least three people will notice, and one of them will say something' — with a percentage attached. Then the design must be an actual test, meaning that a plausible version of the feared outcome is genuinely possible. Then the observation must be recorded during or immediately after, because retrospective recall is filtered by the very belief under test. Then the debrief asks what the result means about the belief, not whether the patient feels better.

Two failure modes recur. Safety behaviors invalidate the result silently — the patient did the exposure while gripping a water bottle, rehearsing sentences, or standing near the exit, so the disconfirmation is attributed to the crutch rather than the belief. And when the feared outcome actually happens, that is not a failed experiment; it is the more valuable one, because the belief being tested is almost never 'X will happen' but 'if X happens I will not survive it.' The therapist who is visibly unruffled by the bad outcome teaches more in that moment than ten successful trials.

🗒️ Vignette

A 34-year-old man with social anxiety has avoided speaking in his team's weekly meeting for two years. His belief: 'If I speak up I will blank, everyone will see I have nothing to contribute, and my standing at work will be damaged.' Conviction: 90 percent.

The tempting experiment is to have him ask a short, pre-scripted question. He will manage it, nothing will happen, and he will conclude — correctly — that asking a rehearsed question is not the same as speaking.

So the therapist makes the experiment harder rather than easier, and makes the prediction precise first. What exactly does 'blank' look like? He says: a pause of more than five seconds where he cannot find the word. How many people will visibly react? 'Most of them — six or seven.' What will happen afterward? 'Someone will mention it.' They write all three down with his ratings.

The agreed test: contribute one unscripted point in Thursday's meeting, and — this is the part he resists — deliberately allow a pause rather than filling it, because the safety behavior of rushing is what has kept the belief intact. He is to note the actual length of the pause and count the actual reactions.

He comes back having done it. There was a pause; he timed it at about four seconds. He counted two people looking up. Nobody mentioned it afterward. Conviction drops from 90 to 55 percent — not to zero, and the therapist does not push for zero.

The following month, in a different meeting, he does blank — a genuine, ugly, eight-second silence, and someone else finishes his sentence for him. He arrives at the next session expecting the therapist to be disappointed. The therapist is delighted, and says so: this is the outcome they could not have manufactured. What actually happened to his standing at work? Nothing measurable. What happened to him? He was mortified for about an hour and then it passed. That is the datum the two years of avoidance existed to prevent him from collecting, and no successful experiment could have supplied it.