Responders and non-responders to a TNF inhibitor had separated in the blood by week two

Design

longitudinal blood transcriptomic profiling at three timepoints during treatment

Population

86 patients with plaque psoriasis in a Chinese cohort receiving a TNF-α inhibitor

Primary outcome

time-course drug-response gene expression and its relationship with treatment response

Effect

3,351 drug-response genes identified; 'red' co-expression module negatively correlated with outcome, hub gene TNFSF10 associated with poor response; module pattern changed by week 2 in responders but not non-responders

Eighty-six patients with plaque psoriasis starting a TNF-α inhibitor had blood transcriptomes profiled at three time points through treatment. The analysis identified 3,351 time-course drug-response genes, enriched in immune signalling and cell activation pathways and expressed in both lesional skin and blood, of which 259 differed between at least one pair of timepoints and 962 were shared with inflammatory bowel disease.

Responders and non-responders differed at the gene level - responders had higher OLIG1, LRRK1 and KSR1 and lower TNFAIP3 and MPP7. A co-expression module, labelled 'red' and enriched in inflammation and TNF-related pathways, correlated negatively with treatment outcome, and high expression of its hub gene TNFSF10 was associated with poor response. The observation that matters clinically is about timing: by week two, the red module's co-expression pattern had changed significantly in responders and not in non-responders.

None of this is a test you can order, and it should not be presented as one - these are research-grade transcriptomic modules in 86 patients from a single cohort, with no external validation and no demonstration that acting on them improves outcomes. What the timing observation does support is something already actionable without any assay: the biology of response is settling far earlier than the standard 12- to 16-week assessment point, which means a patient showing nothing at all by week four is not merely early. That is a reason to check adherence and injection technique then, rather than to wait three more months for the scheduled review.

• Check adherence, dosing interval and injection technique at four weeks in a patient with no response at all, rather than waiting for the formal assessment
• Keep the formal response assessment where guidelines put it - this does not justify switching early on no evidence
• Do not order or request transcriptomic response testing; it does not exist as a clinical assay
• Note the overlap with inflammatory bowel disease gene sets, which is consistent with shared TNF biology rather than a finding about either disease
• Treat the named genes as research findings, not as a panel

Why it matters

It says the decision about whether a biologic is working is biologically settled long before the appointment at which we ask the question.

Don't overread it

An 86-patient single-cohort transcriptomic study with no external validation - the markers are not clinical tests, and no trial has shown that earlier assessment improves outcomes.

The statistics, in plain English

Three thousand three hundred and fifty-one genes identified in 86 patients is far more variables than people, which is normal for transcriptomics and means the specific gene lists will not replicate exactly - the module-level findings are more robust than the individual genes named. No external validation cohort is reported, so the 'red' module and TNFSF10 should be read as generated by this dataset rather than confirmed in another. The week-two divergence is the most transferable observation because it is about timing rather than about any particular gene.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.