| Objective: This study aims to explore the relationship between blood fibrinogen levels and mayo scores in patients with ulcerative colitis.And to provide a reference basis for judging the disease activity of UC patients as soon as possible and formulating clinical treatment by analyze the predictive value of the regression model constructed by blood fibrinogen to the disease activity of UC patients.Methods: This study retrospectively analyzed the hospitalization data of patients diagnosed with UC.238 cases of UC patients hospitalized in the First Affiliated Hospital of Guangxi Medical University from July 2012 to October2019 were selected as the research object.Patients’ serological data were the blood test results within 24 hours after hospitalization.Colonoscopy was performed to evaluate the mayo score of all patients with UC.Divided disease activity into remission,mild,moderate and severe.First of all,238 patients with disease activity was divided into training set(n=187)and validation set(n=51).Through the analysis of the differences in the hematological indexes and clinical characteristics of UC patients in different disease activity,the correlation between blood fibrinogen with the disease activity of UC patients was evaluated.Established a clinical prediction model of UC disease activity.Used odds ratio(OR)and 95% confidence interval(CI)to compare statistical results.Univariate analysis was performed on UC patients in the training set,and a least absolute shrinkage and selection operato(LASSO)regression model was used to dimensionality reduction processing and feature screening.And used multi-factor logistic regression analysis to adjust the screening of confounding factors and establish the final clinical prediction model.Further used nonlinear fitting and box plot to assess whether fibrinogen is linearly related to the mayo score of patients with UC.Used scatter plots combined with linear regression or nonlinear regression to explore the correlation between fibrinogen and traditional serological inflammation indicators of UC.Receiver operating characteristic curve(ROC)analysis of the regression model constructed by fibrinogen was used to learn about predictive value on disease activity of patients with UC patients.Then,Hosmer-Lemeshow test and calibration curve were used in the training set and validation set to evaluate the calibration degree of the clinical prediction model.Finally,the nomogram is used to visualize the constructed regression model.Results: A total of 238 patients were included in this study,and 184 patients were moderate to severe in disease activity,77.3% of the total(184/238).In the training set and the validation set,patients with moderate or severe disease activity accounted for 74.87%(140/187)and 86.27%(44/51)of the data set respectively.The results of univariate logistic regression analysis in the training set showed that fibrinogen has a significant relationship with the disease activity of UC patients.Used Lasso regression analysis to reduce dimensions and filter features.The results showed that the 14 variables have a non-zero coefficient relationship with the disease activity of UC patients.The multi-factor logstic regression was further used to selected confounding factors.The results suggest a positive correlation between fibrinogen and disease activity in UC patients(OR=1.61,95% CI: 1.07 to 2.42;P<0.05).In addition,we found a negative correlation between the percentage of lymphocytes(OR=0.59,95% CI: 0.38 to0.91;P<0.05),and Hematocrit(OR=0.44,95% CI: 0.22 to 0.87;P<0.05)with the disease activity of UC patients,and a history of abdominal surgery(OR=5.96,95% CI: 1.28 to 27.82;P<0.05)was a independent risk factors for UC patients to be in moderate to severe disease activity.Nonlinear fitting showed that there was a non-linear positive correlation between fibrinogen and mayo score of UC patients.Scatter plots combined with linear regression suggested that fibrinogen was positively correlated with UC traditional serological indicators,and Spearman rank correlation suggested a correlation coefficient r = 0.674.When combined with nonlinear fitting,trends were equally significant in male and female subgroups.ROC analysis was performed to evaluate the discrimination of the clinical prediction model.The results showed that the regression model predicted the area under curve(AUC)of disease activity in UC patients is 0.811.When compared with the traditional UC serum inflammation markers erythrocyte sedimentation rate to predict disease activity in UC,The area under the curve increase significantly.The factors obtained from the multi-factor regression were further tested by Hosmer-Lemeshow test and correction curve chart to evaluate the clinical prediction model,finally it is well calibrated(P>0.05).Finally,we used the nomogram to visualize the above regression model.The results suggested that fibrinogen contribute significantly to the regression model for predicting disease activity in UC patients.Conclusion: The results of this study showed that fibrinogen level was associated with disease activity in patients with UC,and there was a non-linear positive correlation between fibrinogen and mayo scores in patients with UC.The mayo score of UC patients showed an upward trend with the rise of fibrinogen.We also had found that the mayo score of UC patients showed a downward trend with the increase of lymphocyte percentage and hematocrit.Abdominal surgery history was an independent risk factor for UC patients in the middle and severe stage of disease activity.In addition,AUC of the regression model constructed fibrinogen which was used to predict the disease activity of UC patients was 0.811.In the comparison of fibrinogen and traditional serological indicators of UC disease activity,AUC was significantly increased.Finally,the nomogram plot of regression model suggested that the model helped to predict the disease activity of UC patients,and to guide early clinical prediction of disease activity and formulate treatment plan as soon as possible.Fibrinogen played an important role in the model. |
PDF Full Text Request