Therapeutic Drug Monitoring for a Partial Adalimumab Responder After a Negative Monitoring Trial
A trial built to answer whether drug-level testing helps has already said no on average, which doesn't settle what one specific test result would still be useful for in this specific patient.
Walter B., a 67-year-old man, spends most weekend mornings in his garage refinishing furniture neighbors and grandkids drop off, work he took up after retiring from twenty-eight years as a bus mechanic and has since turned into something closer to a second career. Adalimumab controlled his rheumatoid arthritis well for the first two years of a now five-year course, but the response has drifted: his DAS28-CRP, which held near 2.4 for two years, has sat at 3.9 for the past several months despite his rheumatologist already having shortened his dosing interval from the labeled every other week to weekly, an empiric adjustment made eight months ago that brought only partial improvement.
The next decision — switch to a different TNF inhibitor, add methotrexate, or move to a different mechanism entirely — is exactly the kind of choice a trough level and anti-drug antibody test is meant to sharpen: a high antibody titer with an undetectable drug level points toward switching mechanism, since the same class is likely to be neutralized the same way, while a clean antibody result with a low trough points toward simply escalating dose further or trying a different molecule within the same class. But that reasoning has now been tested head-on, and it did not hold. ADDORA-switch, a blinded randomized test-treatment trial reported by Wientjes and colleagues in 2025, enrolled RA patients who had failed adalimumab — failure defined as a DAS28-CRP above 2.9, a threshold his own 3.9 clears — and randomized them to a switch chosen by trough level or a switch chosen at random. Mean time-weighted DAS28-CRP came out at 3.15 in both arms. Its own receiver-operating-characteristic analysis went further than the headline: adalimumab levels carried no predictive value for whether a patient would do better on another TNF inhibitor or on a different mechanism, which is exactly the inference the test is ordinarily ordered to support. NOR-DRUM B, the trial usually cited on the other side, did find real benefit from proactive monitoring — but in patients on infliximab, an intravenous TNF inhibitor with different clearance and immunogenicity dynamics entirely. Whether that positive result generalizes to Walter's own subcutaneous adalimumab is the question RA-DRUM was built to answer, and RA-DRUM is still recruiting.
Whether a drug level still tells you something
I'd still order the trough level and antibody titer. ADDORA-switch tested the average effect of a level-guided switch across a whole trial population and found none — that's a real result, but it's a claim about the average patient, not a claim that no individual result would change his management. I still don't know whether to switch him within the TNF class or move to a different mechanism, and this test is the most direct way to find out.
I'd want to hear exactly what decision the result changes before ordering it. ADDORA-switch tested precisely this drug and precisely this decision — patients failing adalimumab, randomized to a trough-guided switch or a random one — and both arms landed on the same time-weighted DAS28-CRP, 3.15. The part that cuts hardest against ordering it here is the receiver-operating-characteristic analysis: adalimumab levels had no predictive value for response to either a TNF inhibitor or a non-TNF agent. That isn't a null on average with a signal hiding inside it for the right patient — it's the test failing at the specific inference you'd be ordering it to make. NOR-DRUM B's positive finding was for infliximab, given intravenously with very different clearance and immunogenicity dynamics; the trial built to test whether that transfers to a subcutaneous TNF inhibitor, RA-DRUM, is still recruiting and has not reported.
If the plan is to test, get a result, and then make whatever decision we'd have made anyway regardless of what the number says, we're not actually using the test to decide anything — we're just paying for reassurance. An empiric switch to a different mechanism, given he's already had one empiric dose escalation fail to fully work, is simpler and just as evidence-based given what ADDORA-switch actually found.
There's a use for this test that ADDORA-switch's finding doesn't touch at all. He's on weekly adalimumab, which is already an unlicensed escalation from the labeled schedule — more drug, more cost, more injections than the label supports. If his antibody titer comes back undetectable and his trough level is genuinely therapeutic, that specific result justifies stepping him back down to the labeled every-other-week interval regardless of whether we ultimately switch mechanisms — a real, concrete decision ADDORA-switch's null finding on switch selection simply never asked about.
Agreed: the trough level and antibody test are ordered, but explicitly to answer only the de-escalation question the pharmacist raised, not to decide switch-versus-continue, which proceeds toward a mechanism switch regardless of the result.
Not agreed: whether ordering the test at all was the right call given ADDORA-switch's finding. The pharmacologist's position is that testing without a decision riding on the result is not real individualization; the rheumatologist's position is that the de-escalation question the pharmacist identified is itself sufficient justification, even if it wasn't the original reason for ordering it. Both agreed the test was worth running once that specific use was named, without resolving whether it would have been worth running otherwise.