TNFi and RA-ILD – The Pendulum Swings Again Save

Tumour necrosis factor inhibitors (TNFi) revolutionised the treatment of rheumatoid arthritis. They heralded the wave of advanced therapies, which has transformed the treatment landscape in rheumatic disease. While highly effective for the joint manifestations of the disease, a note of caution was sounded regarding the potential for worsening of ILD with these agents.

The original randomised controlled trials of TNFi showed no evidence of an excess of ILD compared to placebo. The most striking report of the potential risk of TNFi was from a BSR biologics registry study, which reported a doubling of mortality rates for RA-ILD patients treated with TNFi compared to rituximab. The mechanistic explanations for why TNFi would be associated with worse pulmonary outcomes always felt slightly tenuous. The biologic rationale for why a medication which worked so well for all other aspects of rheumatoid disease would be associated with worse pulmonary rheumatoid disease was also difficult to conceptualise. It has become the generally perceived wisdom that TNFi should be preferentially avoided in RA-ILD in favor of agents such as rituximab and abatacept.

Within this framework, a study by England et al presented at ACR23 was highly relevant. This was Abstract 1582 presented at the second plenary session. The authors performed a target trial emulation study evaluating TNFi compared to non-TNFi biologic and targeted synthetic DMARDS. This was a new-user propensity score matched study. The study was conducted within the Veterans Health Administration system with the usual caveats which go along with this setting (92% male, mean age 68 years). 474 patients were recruited, split equally between the two groups. Vitally the non-TNFi group was largely composed of our 2 favored RA-ILD drugs with 53% receiving rituximab and 28% abatacept. Essentially TNFi were going up against the standard of care. The study was conducted over 3 years of follow-up.

There was no significant difference in respiratory-related hospitalisation, aHR 1.22 (0.92, 1.60). There was also no difference in all cause or respiratory mortality, aHR 1.12 (0.79, 1.59) and aHR 1.36 (0.75, 2.46) respectively. One could argue that the study may be underpowered for a clinically relevant difference, the point estimate is of 22% more respiratory hospitalisation, 12% more all cause and 36% more respiratory mortality, all likely clinically relevant differences. But the clincher here is that all of these favor the TNFi group – if any difference became evident by increasing power it would likely be in favor of TNFi. As with any retrospective database type study like this we need to be cautious about the interpretation. There may be residual confounders that the propensity score matching did not fully account for. 

This data does not support the avoidance of TNFi in RA-ILD. It certainly doesn’t support the cessation of TNFi in a patient because of an incidental finding of ILD. Does it support starting TNFi to treat RA-ILD? I’m not sure I would go quite that far, I would still favor rituximab and abatacept, but it certainly gets us close to clinical equipoise. I think this provides further rationale for controlled prospective comparative effectiveness studies of TNFi and other biologic and targeted synthetic DMARDs in RA-ILD. This is a common clinical scenario where we need a robust evidence base.