Between half and two thirds of patients started on methotrexate do not reach remission by six months, and there is still no usable way to know in advance which ones. This study tested a quantitative methylation-sensitive PCR assay measuring DNA methylation of the TNFA gene, in early rheumatoid arthritis and in psoriatic arthritis.
Pre-treatment methylation was reduced in both diseases compared with controls. In rheumatoid arthritis, higher methylation predicted methotrexate-induced remission with an area under the curve of 0.721, and adding it to a model built on clinical data raised that model from 0.699 to 0.760. In psoriatic arthritis it predicted nothing.
Methylation fell over time in rheumatoid arthritis regardless of whether the patient responded, which matters for interpretation: the assay looks useful before treatment and not as a monitoring tool afterwards.
This is not ready for clinic. An AUC of 0.72 is modest, the gain over clinical data alone is about six percentage points, and the assay needs prospective validation in an independent cohort before it could change a prescription. It is worth knowing about because it is one of the few methotrexate response markers that has improved on clinical prediction at all.
- Not a test to request: this is research-stage and not validated prospectively.
- If a biomarker for methotrexate response reaches your practice, ask what it adds over clinical prediction, not what its AUC is alone.
- The negative result in psoriatic arthritis is a reminder that markers do not transfer between diseases that share a drug.
- Methylation fell irrespective of response, so this would be a baseline test, not a monitoring one.
The statistics, in plain English
Area under the curve is the chance that the test ranks a patient who will remit above one who will not. 0.5 is a coin toss and 1.0 is perfect, so 0.721 is modest — useful for sorting a cohort, not for deciding about one person. The number that matters is the comparison, not the absolute: clinical data alone reached 0.699 and adding the assay reached 0.760, so the marker contributes about six percentage points. A test that scores well on its own but adds nothing to what the clinician already knows is not worth running, and this one clears that bar narrowly.
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.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for rheumatology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free