Smoking Doesn’t Impact TNFi Response in axSpA


Smoking in axial spondyloarthritis doesn't appear to impact treatment with TNF inhibitors despite the status of disease severity, a study shows.


Smoking in axial spondyloarthritis doesn't appear to impact treatment with TNF inhibitors despite the status of disease severity, a study shows. (©

It doesn’t matter whether an axial spondyloarthritis (axSpA) patient is a current smoker or an ex-smoker, they respond just as well to treatment with tumor necrosis factor inhibitors (TNFi) as never smokers, even though the severity of their condition is worse than patients who never smoked, shows a study published in Arthritis Care and Research.

After accounting for bias the authors found a relationship between smoking and more severe disease in axSpA, but no increase in first-line TNFi failure. The authors suggest, “prescribers should dispel any subconscious bias that smokers may not respond as well to treatment.”

The study included 627 patients (69 percent male, mean age 46 years) of which 33 percent were current smokers and 30 percent were former smokers.

In fact, all three groups―former smokers, current smokers and never smokers―had similar responses in BASDAI and ASDAS scores at three months, researchers reported.

Researchers highlighted bias often found in observational studies and they identified methodological differences that may be responsible for variations in reports.

Determining causality out of observational registry data can be challenging, but worthwhile especially in the case of treatment failure in axSpA since half of all patients do not respond to TNFi therapy. Currently, conflicting data exist with regards to the role of smoking in TNFi failure with some studies linking the two while others have found no association.

Nicola Goodson M.D., Ph.D., and colleagues from multiple international centers found that methodological differences in conflicting studies such as baseline disease severity measures, sample-restriction, and variable enforcement and response definitions are responsible for confounding results.

Dr. Goodson and colleagues suggest techniques for reducing bias in future investigations. “We demonstrated the importance of several methodological considerations for future studies of non-interventional exposures on treatment response, and offer inverse-probability weighting as a solution to reduce potential bias,” they write.


The study is based on data from the British Society for Rheumatology Biologics Register for Ankylosing Spondylitis (BSRBR-AS), a large UK-wide prospective cohort of patients with axSpA. Patients were limited to those who started TNFi therapy between December 2012 and June 2017. They were evaluated at months three, six and 12, and then each year after with smoking habits being self-reported via questionnaire.  Response in the first three months was assessed separately to account for non-random dropout.

Outcomes were based on patient-reported disease activity, functional impairment, and other aspects of disease severity based on established criteria. For each variable, the authors compared its change over time according to smoking status.

Compared to never smokers, disease activity (BASDAI) increased by 0.07 units more (95%CI -0.11 to 0.24) for ex- smokers and 0.04 units more (95%CI -0.13 to 0.22) for current smokers, per six-month period.

After six months, 136 participants discontinued treatment due to adverse events, inefficacy or other reasons. Discontinuation of treatment was not different across all smoking classifications.


Bias in research is nearly impossible to eliminate. As scientists we like to see words like randomization, blinding and prospective in the methods section as these factors give us confidence that the evidence is sound and as free from error as possible. The unfortunate truth is that collecting large amounts of data in rigorous, randomized ways is often impractical and even unethical. Large, retrospective, observational databases can be a treasure trove of information, if it is processed properly.

The authors state, “Disease registries are important resources for observational research. They provide high quality data for large numbers of patients generalizable to clinical practice, making them invaluable for comparing effectiveness of treatments. “

Dr. Goodson and colleagues set out to accomplish two tasks in this work: 

1) Limit bias in observational data.
2) Determine the association between smoking, disease activity, and TNFi treatment in patients with axSpA.

By limiting informative censoring and utilizing weighted logistic models the authors were able to minimize sample bias and report on causality.

Clinicians should always encourage smoking cessation in all of their patients and can counsel those with axSpA that smoking may affect longer-term treatment response while not increasing the chances of initial TNFi failure.


Gregory M. Weiss, M.D., is a cardiothoracic anesthesiologist practicing in Virginia. He is a frequent contributor to Rheumatology Network.


Sizheng Zhao MD; Kazuki Yoshida MD; PhD, Gareth T Jones PhD, et al. “The impact of smoking on response to TNF inhibitors in axial spondyloarthritis: methodological considerations for longitudinal observational studies.” Arthritis Care Res. Accepted Author Manuscript. doi:10.1002/acr.23851

Related Videos
© 2024 MJH Life Sciences

All rights reserved.