Treatment Choice After Two Biologics of Different Mechanisms Both Fail
A third failed mechanism looks like straightforward treatment resistance until the actual framework built for exactly this situation asks whether some of what's being treated is inflammation at all.
Frank D., a 55-year-old man, has supervised the same warehouse floor for eighteen years and has grown increasingly worried, these past few months, about how much longer his employer's disability leave policy will hold his job open. His rheumatoid arthritis, ten years old, has now failed three biologics of genuinely different mechanisms in sequence — adalimumab, tocilizumab, and most recently abatacept — each given an adequate trial before being judged inadequate. His DAS28-CRP today is 5.2, and a hand radiograph taken this visit shows two new erosions that weren't present a year ago.
That combination — failure of at least two biologic or targeted synthetic DMARDs of different mechanisms, after an adequate csDMARD trial, with signs of ongoing disease — is exactly the population EULAR's difficult-to-treat RA points-to-consider document was written to address, and the framework it lays out is not simply 'try the next drug.' It calls, as an explicit first step, for confirming that the persistent symptoms actually reflect active synovitis rather than a secondary process — central sensitization, a genuine and well-recognized amplification of pain signaling that can develop alongside long-standing inflammatory disease and inflate a composite score like DAS28 without reflecting more active joint inflammation, since two of that score's four components, tender-joint count and patient global assessment, are known to correlate only loosely with objective measures like swelling or acute-phase reactants. Ten years of disease, three failed biologics, and the accumulating fatigue of a job he's worried about losing are exactly the kind of history that predisposes toward this secondary amplification, on top of whatever inflammatory disease remains genuinely active underneath it. His new erosions are a harder, more objective finding that doesn't fit that alternate explanation cleanly, though — a real signal that something is still actively destroying his joints, whatever else may also be contributing to how he scores on paper.
After three mechanisms, what actually comes next
I'd start a JAK inhibitor now — it's a genuinely untried mechanism, and his new erosions are an objective finding, not a subjective symptom report. Joint damage doesn't wait for a diagnostic workup, and I don't want to spend more months confirming what the imaging has already shown us.
I'd want to run a central sensitization screen before we do that. EULAR's own difficult-to-treat RA framework, written for exactly his situation — two or more biologic mechanisms of different classes failed after an adequate csDMARD trial — explicitly calls for confirming that persistent symptoms reflect ongoing synovitis before assuming a fourth mechanism is the answer. His DAS28 is being driven substantially by tender-joint count and his own global assessment, both of which correlate poorly with actual inflammation — a real, well-recognized pattern in long-standing disease.
I'd take the erosions seriously too — I'm not saying his disease is inactive. But two new erosions and a central-sensitization-inflated DAS28 aren't mutually exclusive; both can be true in the same patient, and only one of them tells you whether the next biologic will actually help the part of his score driven by amplified pain rather than active joints.
Before either question gets settled, the difficult-to-treat framework's own first step is more basic than a sensitization workup: adherence and access. He's raised cost and job-security concerns directly, and neither has been formally checked against whether he's actually been able to take every dose of the last three regimens consistently. If access has been inconsistent, that changes what 'treatment failure' even means for any of the three drugs already tried.
Agreed: upadacitinib started given the objective erosive progression, run alongside, not instead of, a formal central sensitization screen and an adherence and access review, with the plan to reassess his response using ultrasound and acute-phase reactants rather than DAS28 alone, given the suspicion that his score is not a clean read on inflammation.
Not agreed, and left explicitly open: how much of his current DAS28 reflects active disease versus amplified pain, and whether his prior three treatment failures were true pharmacologic failures or were affected by inconsistent access. Neither question resolves before his response to the new drug can be assessed.