A Meta-analysis Of Risk Factors For Pulmonary Interstitial Lesions In Patients With Rheumatoid Arthritis And Establishment Of A Risk Prediction Model

Part Ⅰ:Risk factors of the incidence of Interstitial Lung Disease in rheumatoid arthritis patients:A systematic review and meta-analysisObjective:This study aimed at identifying risk factors of the incidence of interstitial lung disease in rheumatoid arthritis patients(RA-ILD)by a systematic review and meta-analysis.Methods:Studies published by March,2021 were searched in PubMed,Web of Science,MEDLINE,EMBASE,Cochrane Library and Scopus databases.Two researchers independently screened the studies according to inclusion and exclusion criteria,and extracted data from these studies.Quality assessments of included studies were performed using Newcastle-Ottawa Scale.Based on the result of heterogeneity,random-effects model or fixed-effects model was chosen in the meta-analysis.Furthermore,a sensitivity analysis was conducted to identify the origins of heterogeneity,and publication bias was evaluated for the factors with more than 5 included studies by funnel plots and Egger’s test.Results:Among 3075 identified articles,12 studies met the inclusion criteria.17 risk factors were included in the meta-analysis.Male(OR 2.18,p<0.001),age>60-year-old(OR 1.53,p=0.02),old RA onset age(OR 1.06,p=0.008),smoking(OR 1.63,p<0.001),lung complications(OR 2.60,p<0.001),positive rheumatoid factor(OR 1.30,p=0.007).high ESR(OR 1.01,p=0.01),leflunomide usage(OR 1.41,p=0.04)were shown as risk factors of RA-ILD.Conclusion:Physicians should be aware that RA patients with above risk factors are likely to develop RA-ILD,and perform close ILD screening for the patients during follow-ups,so that the patients can be early diagnosed and treated,and achieve improved prognosis.Part Ⅱ:Development of the risk prediction model of rheumatoid arthritis associated interstitial lung diseaseObjective:This study aimed at investigating predictors of rheumatoid arthritis associated interstitial lung disease(RA-ILD)and developing a multidimensional risk prediction model for RA-ILD occurrence.Methods:Patients in Chinese registry of rheumatoid arthritis(CREDIT)cohort with complete baseline data and at least 1-year follow-up until March,2021 were included in the study.Baseline characteristics of these patients were extracted,including demographic data,antibody condition,organ involvements and treatment strategies.We used Lasso regression to screen candidate variables according to the accessibility,stability and literature knowledge of the factors,and developed the risk prediction model by multiple COX regression.The model was presented by nomogram and forest plot and was tested for accuracy and robustness by concordance-index,calibration curve and Brier score.We further identified the risk stratification for RA-ILD and its clinical value through receiver operator characteristic(ROC)curve and decision analysis(DCA)curve.Results:3856 RA patients were enrolled in the study,including 96 patients who developed ILD in follow-up period.Through Lasso regression and COX regression,a prediction model including 5 variables(sex,age,rheumatoid factor,anti-cyclic citrullinated peptide antibodies and methotrexate)was developed.The model was tested for discriminability and stability by c-index(0.796),calibration curve and brier score(0.009).According to ROC curve and DCA curve,we recommended 141points in nomogram(2.2%for RA-ILD risk in next year)as the cut-off for patients with high risk and low risk.Conclusion:This risk prediction model can conveniently and accurately predict the RA-ILD risk in clinic.It may help physicians to make early and individual screenings,diagnosis and treatment strategies.