Construction Of A Prognostic Model For Patients With Rheumatoid Arthritis Responding To Leflunomide

Objective: Rheumatoid arthritis(RA)is an autoimmune disease characterized by chronic synovitis.If the condition of RA patients could not be controlled in time,it would cause joint deformities and disability,which would seriously endanger the quality of life of the patients.Leflunomide(LEF),as a biological inhibitor,was widely used in the treatment of RA in the clinic.LEF could alleviate the pain of patients in both long-term and short-term conditions,and improved patients’ clinical symptoms.Due to the difference in drug response,not all RA patients taking LEF could have a good prognosis effect,and some patients might even have adverse reactions such as diarrhea,hypertension,liver damage,etc.In order to enable patients to obtain good prognosis and avoid delay of their illness,it was of great significance to confirm whether RA patients were suitable for LEF as soon as possible.Therefore,this study analyzed various indicators that might affect the prognosis efficacy of LEF,and constructed a model that combines traditional prognostic factors and genetic factors to evaluate the prognosis efficacy of LEF in RA patients.Methods: This study collected data from four tertiary first-class hospitals(the First Affiliated Hospital of China Medical University,Shengjing Hospital Affiliated to China Medical University,Dalian Central Hospital,and the First Affiliated Hospital of Jinzhou Medical University).The study collected a total of 245 RA patients who were treated with LEF from June 2018 to June 2020,and each patient would be followed up for 6months,excluding the number of people who were lost to follow-up and withdrew from treatment due to adverse reactions.We also collected patients’ traditional prognostic indicators including baseline information,serum immunology and other clinical indicators,and evaluate the patient’s disease activity(DAS28).Each enrolled patient was followed up for 6 months,and the prognostic effect of drugs of the patient was judged based on the difference between the DAS28 score after 6 months of LEF taking and the baseline patient’s DAS28 score.In addition,this study also screened out 15 single nucleotide polymorphisms(SNPs)that were potentially related to the prognostic effect of LEF through text mining,and database screening,and the SNP microarray data was obtained by multiplex PCR sequencing.Single-factor regression analysis was performed on 15 SNPs data,and SNPs were screened with P<0.2 as the standard.The selected SNPs were analyzed using LASSO regression and Bayesian regression methods to construct risk scores representing genetic indicators.SPSS23.0 was used to perform univariate logistic regression analysis on the traditional prognostic indicators of patients,and multivariate regression analysis was performed on the indicators with P<0.2(the method used Forward: LR,Entry: 0.05 Removal: 0.10).Then the statistically significant traditional prognostic indicators with the SNP risk score were combined to construct the final prognostic model of LEF.Receiver operating characteristic curve(ROC)was used to evaluate the discrimination of the model,and the calibration and Hosmer-Lemeshow test were used to evaluate the calibration of the model.Nomogram was used to visualize the model,and decision curve analysis(DCA)and clinical impact curve were used to evaluate the application value of the model.Results: A total of 245 patients were enrolled in the study.After 6 months of follow-up,76 patients had no significant improvement in DAS28,which was defined as no response;169 patients had DAS28 decreased and their condition was relieved,which was defined as good response.The results of univariate regression on the traditional prognostic indicators of RA patients found that alcohol drinking(P=0.181),disease duration(P<0.001),immunoglobulin G(P=0.042),Complement protein 3(P=0.184),High-density lipoprotein(P=0.068),glucose(P=0.162),monocyte to lymphocyte ratio(P=0.02),erythrocyte sedimentation rate(P=0.132),baseline DAS28(P=0.109)were make sense under conditions P<0.2.The above factors were incorporated into the multivariate logistic regression model,and finally 4 indicators were screened out: disease duration(P<0.001),monocytes to lymphocytes ratio(P=0.035),immunoglobulin G(P=0.029),baseline DAS28(P=0.044).Through univariate regression analysis of 15 SNPs,we found that rs1051266(P=0.188),rs10757668(P=0.189),rs11231809(P=0.145),rs1127354(P=0.127),rs2259571(P=0.151),rs2298383(P=0.071),rs2470890(P=0.037),rs2813563(P=0.139),rs3213422(P=0.062),rs9340799(P=0.139),rs983332(P<0.001)were meaningful under the condition of P<0.2.The 11 selected SNPs were processed by LASSO regression and Bayesian regression for dimensionality reduction,and 4 SNPs were selected,namely rs2298383,rs2470890,rs3213422,and rs983332.The risk score constructed by 4 SNPs combined with the 4 selected traditional prognostic indicators(disease duration,monocytes to lymphocytes ratio,immunoglobulin G,baseline DAS28)were used to jointly construct a prognostic model of LEF.The area under the curve(AUC)of the model was 0.7872(95% confidence intervals(CI):0.7279-0.8466).Conclusion: The shorter the disease duration,the greater the ratio of monocytes to lymphocytes,the higher the immunoglobulin G,and the higher baseline DAS28 RA patients with LEF would have a better prognosis.Combining the traditional prognostic indicators of patients with genetic indicators to construct a prognostic model for RA patients takng LEF might have certain value in guiding the medication of RA patients.