LAUSR

OBJECTIVES Colchicine forms the mainstay of treatment in familial Mediterranean fever (FMF). Around 5-10% of FMF patients are colchicine resistant and require anti-interleukin-1 drugs. We aimed to compare the characteristics of colchicine-resistant and colchicine-responsive patients and to develop a score for predicting colchicine resistance at the time of FMF diagnosis. METHODS FMF patients (0-18 years) enrolled in the TURPAID (Turkish Pediatric Autoinflammatory Diseases) registry were included. The predictive score for colchicine resistance was developed by using univariate/multivariate regression and ROC analyses. RESULTS A total of 3445 FMF patients (256 [7.4%] colchicine-resistant and 3189 colchicine-responsive) were included (F/M = 1.02; median age at diagnosis 67.4 months). Colchicine-resistant patients had longer, more frequent attacks and were younger at symptom onset and diagnosis (p< 0.05). Fever, erysipelas-like erythema, arthralgia, arthritis, myalgia, abdominal pain, diarrhea, chest pain, comorbidities, parental consanguinity, and homozygosity/compound heterozygosity for exon 10 MEFV mutations were significantly more prevalent among colchicine-resistant than colchicine-responsive patients (p< 0.05). Multivariate logistic regression analysis in the training cohort (n = 2684) showed that age at symptom onset, attack frequency, arthritis, chest pain, and having two exon 10 mutations were the strongest predictors of colchicine resistance. The score including these items had a sensitivity of 81.3% and a specificity of 49.1%. In the validation cohort (n = 671), its sensitivity was 93.5% and specificity was 53.8%. CONCLUSION We have developed a clinician-friendly and practical predictive score which could help us to identify FMF patients with a greater risk of colchicine resistance and tailor disease management individually at the time of diagnosis.