medwireNews: A risk score that incorporates sex, age at rheumatoid arthritis (RA) onset, disease activity, and the MUC5B rs35705950 genetic variant may be useful for the prediction of subclinical RA-associated interstitial lung disease (ILD), research suggests.
Philippe Dieudé (Université de Paris, France) and co-investigators say that the risk score could “help clinicians in their daily clinical practice to identify high-risk patients with RA without pulmonary symptoms that would benefit from ILD screening by chest HRCT [high-resolution computed tomography] scan.”
The researchers developed the score using data from the prospective ESPOIR cohort, which included 163 people (21.5% men) with RA (median age 49 years at onset) and no ILD, pulmonary signs, or respiratory symptoms.
All of the participants were genotyped for the MUC5B rs35705950 genetic variant, which has previously been associated with a threefold increased risk for RA-ILD, and 11.3% carried the T risk allele.
The cohort also underwent chest HRCT, with subclinical RA-ILD detected in 19.0%.
Dieudé and co-authors report in Arthritis & Rheumatology that, on multivariate analysis, four factors were significantly associated with subclinical RA-ILD risk: male sex (odds ratio [OR]=3.93); the MUC5B rs35705950 T risk allele (OR=3.74); higher mean DAS28-ESR (OR=2.03 per unit); and older age at RA onset (OR=1.10 per year).
The team used these variables to create a patient-specific subclinical RA-ILD risk score by summing weighted scores for each independent risk factor.
In the development cohort the risk score predicted subclinical RA-ILD with an accuracy of 81% on area under the receiver operating characteristic curve (AUC) analysis. At a cutoff of 51 points, the sensitivity was 71.0% and specificity 79.6%.
Furthermore, when the researchers used the more easily available mean of the four most recent DAS28-ESR values rather than a mean of all values, the performance of the risk score was unaffected, with AUC analysis demonstrating an accuracy of 80%. Similarly, when only the most recent DAS28-ESR value was included, the accuracy was 82%.
Given that obtaining more recent DAS28-ESR values is more convenient in daily practice than an overall average, the researchers validated these two models in the TRANSLATE2 cohort (n=89), which had a subclinical RA-ILD rate of 16.9%. In this cohort the accuracy was 79% when the last four DAS28-ESR values were included and 78% when only the most recent value was used.
Dieudé et al also acknowledge that MUC5B rs35705950 genotyping is not routinely available and therefore created a simplified model excluding this variable.
In the discovery cohort, the simplified model had identical sensitivity to the full model that included only the most recent DAS28-ESR value (75.0% in both) but lower specificity (69.0 vs 85.0%). And the same was true in the validation cohort, with sensitivity at 86.7% in both analyses and specificity at 47.3% and 62.2%, respectively.
Furthermore, excluding MUC5B rs35705950 from the model provided a significantly lower goodness of fit, which the researchers say supports “an important contribution of MUC5B rs35705950 to subclinical RA-ILD risk.”
The authors conclude that their “findings are in good agreement with the results of the FinnGen epidemiological cohort and support a pivotal role of the MUC5B rs35705950 variant for both clinical and subclinical RA-ILD risk stratification and of possible interest for genotyping the risk variant in future clinical practice.”
medwireNews is an independent medical news service provided by Springer Healthcare Ltd. © 2022 Springer Healthcare Ltd, part of the Springer Nature Group