| Background and ObjectiveCrohn’s disease(CD)is a non-specific inflammatory disease that often occurs in the gastrointestinal tract and occurs repeatedly.Therefore,it is very important to choose different biological agents according to the different characteristics of patients with Crohn’s disease and judge whether they are effective after treatment,especially whether they achieve endoscopic remission after treatment.Endoscopic remission,also defined as mucosal healing,is judged by the simple endoscopic score of Crohn’s disease(SES-CD).SES-CD≤2 is defined as endoscopic remission.Endoscopic remission is significantly related to long-term prognosis.However,in clinical practice,endoscopy is invasive,uncomfortable,time-consuming and expensive.Or for various reasons,some patients with Crohn’s disease can’t carry out endoscopy in time or even refuse to carry out endoscopy after 14-22 weeks of biological agent treatment.Based on the above background,this study mainly explores the role of albumin to globulin ratio(AGR)and neural network model based on AGR,C-reactive protein(CRP)and blood routine in identifying endoscopic healing of Crohn’s disease after 14-22 weeks of biological treatment,and transforms it into a simple and practical clinical decision-making tool,which will make endoscopic monitoring of Crohn’s disease easier and more effective.MethodThis is a retrospective study.Include Patients with Crohn’s disease treated anti TNF-α in our hospital from January 2018 to December 2020 Patients with Crohn’s disease and patients with Crohn’s disease treated with Vidozumab(VDZ)and Ustekinumab(UST)from November 2020 to December 2021,who have underwent endoscopy and the examination including blood routine albumin,globulin and Creactive protein within the 14th to 22nd weeks after treatment.Patients were divided into two groups according to SES-CD.SES-CD ≤2 was defined as endoscopic remission group,and SES-CD>2 was defined as non endoscopic remission group.First,univariate logistic regression and multivariate logistic regression were used to evaluate the correlation between blood indexes and endoscopic remission.Secondly,according to the patient’s serological indexes(blood routine,albumin,globulin,AGR,C-reactive protein)and whether endoscopic remission is achieved after treatment,the diagnostic efficacy of relevant indexes and common monitoring indexes CRP,ALB and PLT for endoscopic remission are evaluated through receiver operating characteristic curve(ROC).The ROC curve was used to determine the optimal threshold of relevant indicators to distinguish patients with endoscopic remission from patients with non endoscopic remission.Thirdly,neural network is used to evaluate the correlation between agr combined with C-reactive protein,blood routine and endoscopic remission.The diagnostic efficacy of neural network model for endoscopic remission is evaluated through receiver operating characteristic curve(ROC),and the importance of each serological index to neural network model is determined.Result1.A total of 236 patients were included.It is divided into three parts according to the different biological agents.The number of patients treated with Anti TNF-α is 191,The number of patients treated with VDZ is 33 and The number of patients treated with UST is 12.1)Anti TNF-α:191 patients met the inclusion and exclusion criteria,including 102 in the endoscopic remission group and 89 in the non endoscopic remission group.There was no significant difference in clinical phenotype and drug use between the two groups.2)VdZ:33 patients met the inclusion and exclusion criteria,including 11 in the endoscopic remission group and 22 in the non endoscopic remission group.There was no significant difference in clinical phenotype and drug use between the two groups.3)UST:12 patients met the inclusion and exclusion criteria and were selected into this study,including 6 in endoscopic remission group and 6 in non endoscopic remission group.There was no significant difference in clinical phenotype and drug use between the two groups.2.Correlation and diagnostic efficacy between serological indexes and endoscopic remission after treatment.1)Anti TNF-α:Multivariate logistic regression analysis showed that there was no significant difference between serological characteristics and endoscopic remission after treatment between endoscopic remission group and non endoscopic remission group,and there was no significant difference in intestinal SES-CD score between the two groups.After treatment,multivariate logistic regression analysis showed that AGR was an independent predictor of endoscopic remission(OR=15.400,95%CI 3.418-69.377),but univariate logistic analysis showed that CRP,PLT,ALB commonly used to monitor disease activity were statistically different beteen endoscopic remission and non endoscopic remission;After treatment,the AUC value of AGR for identifying endoscopic remission was 0.778(95%CI:0.711-0.845),the sensitivity and specificity were 0.794 and 0.674,respectively,and the cut-off point value for identifying endoscopic remission was 1.42.The AUC values of ALB,G,CRP and PLT commonly used to monitor disease activity to identify endoscopic remission were 0.744(95%CI:0.674-0.814),0.675(95%CI:0.597-0.753),0.727(95%CI:0.6550.800)and 0.641(95%CI:0.560-0.721),respectively;The cutoff values of ALB,G,CRP and PLT to identify endoscopic remission were 38.20,30.20,1.63 and 306.00 respectively.2)VDZ:The correlation between AGR and endoscopic remission after vdz treatment was statistically different.The AUC of AGR in identifying endoscopic remission was 0.717(0.552-1.000,P=0.045),and the best cut-off point for diagnosis was 1.44.The AUC values of HGB,CRP and PLT commonly used to monitor disease activity after treatment to identify endoscopic remission were 0.767(95%CI:0.531-1.000),0.835(95%CI:0.691-0.978)and 0.901(95%CI:0.787-1.000).The cut-off points of HGB,CRP and PLT to identify endoscopic remission were 113.5,3.45 and 224.5,respectively.3)UST:there was no significant difference between AGR and endoscopic remission after ust treatment(P=0.051).The AUC of agr in identifying endoscopic remission was 0.819(0.552-1.000,P=0.66),and the best cut-off point was 1.43.The AUC values of ALB,CRP and PLT commonly used to monitor disease activity after treatment to identify endoscopic remission were 0.806(95%CI:0.531-1.000),0.833(95%CI:0.590-1.000),0.806(95%CI:0.550-1.000),and the cut-off values of ALB,CRP and PLT to identify endoscopic remission were 38.95,3.45 and 269.50 respectively.3.The neural network model based on blood routine,C-reactive protein and AGR after treatment has high diagnostic ability for endoscopic remission.1)Anti TNF-α:Consistent with multivariate logistic analysis,AGR plays the most important role in the neural network model for predicting endoscopic remission based on serological indicators.2)Vdz:AGR after vdz treatment plays an important role in the neural network model with high accuracy(AUC=0.901)in predicting endoscopic remission based on serological indicators,only second to the most important PLT.3)UST:AGR after ust treatment plays an important role in the neural network model with high accuracy(AUC=0.917)in predicting endoscopic remission based on serological indicators,only second to the most important PLT.Conclusion1.Main conclusion:AGR may be a potentially effective means to identify endoscopic remission in the process of monitoring biological agents in the treatment of Crohn’s disease,which is expected to eliminate unnecessary endoscopy.1)AGR after Anti TNF-α treatment is an independent predictor of endoscopic remission.There is a significant correlation between AGR after Anti TNF-α treatment and endoscopic remission 14-22 weeks after treatment,and the optimal threshold of agr is 1.42,which can better stratify Crohn’s disease patients with endoscopic remission and non endoscopic remission.2)There is an association between AGR and endoscopic remission 14-22 weeks after VDZ treatment.The optimal threshold of AGR is 1.44,which can better stratify Crohn’s disease patients with endoscopic remission and non endoscopic remission.3)There may be an association between AGR and endoscopic remission 14-22 weeks after ust treatment(P=0.051),and the optimal threshold of agr is 1.43,which can better stratify Crohn’s disease patients with endoscopic remission and non endoscopic remission.2.Secondary conclusion:the neural network model based on AGR,CRP and blood routine has high accuracy in identifying endoscopic remission after 14-22 weeks of biological treatment,but its model is difficult to be transformed into a practical clinical decision-making tool. |
PDF Full Text Request