A Risk Prediction Model For Abdominal Aortic Aneurysm Base On Uric Acid/high-density Lipoprotein Cholesterol Ratio

BackgroundsStudies have shown that inflammation plays a key role in the development and progression of abdominal aortic aneurysm(AAA).Recently,the serum uric acid(UA)/high-density lipoprotein cholesterol(HDL-C)ratio(UHR)has been suggested as a novel biomarker in response to inflammatory and anti-inflammatory interactions.However,there are no studies on whether UHR is associated with AAA.Given that the prevalence of AAA in China is lower than that in European and American countries,and that there is no clear recommendation for AAA screening strategies in China,it is important to explore predictors of AAA,construct an efficient and easy-to-use model for predicting AAA,make individual risk predictions and risk stratification for the population,and accurately carry out AAA screening and prevention.PurposeThe purpose of the study was to explore the relationship between UHR and AAA through AAA screening;and to provide a basis for developing screening strategies for AAA high-risk groups by establishing risk prediction model of AAA and conducting internal and external validation.MethodsSection 1:Prospective AAA ultrasound screening was performed in 4800 patients hospitalized in the department of cardiology in our hospital.The relationship between UHR and AAA was explored by using multivariate logistic regression.The generalized additive model(GAM)was used to explore the dose-response relationship between UHR and maximum abdominal aortic diameter and associated inflammatory biomarkers and the relationship between UHR and AAA was visualized using the restricted cubic spline(RCS).To adjust for baseline differences and reduce selective bias,propensity score matching(PSM)analysis was performed,along with subgroup analysis to evaluate the consistency of findings.Section 2:The above 4800 patients were used as the training group to construct the AAA risk prediction model,and multivariate logistic regression was applied to determine the independent predictors of AAA,and the final model was visualized as a nomogram.The area under the operating characteristic curve(AUC)was used to assess the discrimination of the prediction model.The Hosmer-Lemeshow(H-L)goodnessof-fit test and calibration curve were used to assess the calibration of the prediction model.The decision curve analysis(DCA)was used to assess whether the model could benefit patients.Bootstrap resampling method was used for internal validation.AAA ultrasound screening data from 4264 residents in the community were selected for external validation to test the performance of the model in different populations.ResultsSection 1:The prevalence of AAA was 3.56%in this study.The study population was divided into a low UHR group(UHR<17.0%)and a high UHR group(UHR ≥17.0%)according to the operating characteristic curve(ROC).The prevalence of AAA was significantly higher in the high UHR group than in the low UHR group(5.07%vs.2.32%,P<0.001).After further adjustment for relevant covariates,multivariate logistic regression showed a significant positive association between UHR and presence of AAA either as a continuous variable(OR 1.02,95%CI 1.01-1.05,P=0.022)or as a categorical variable(OR 1.48,95%CI 1.02-2.16,P=0.041).The RCS curve showed a nonlinear dose-response relationship between UHR and presence of AAA.Moreover,the positive correlation between UHR and presence of AAA remained significant after PSM analysis and subgroup analysis.Section 2:Independent predictors of AAA were screened in the training group based on multivariate logistic regression,and finally age,sex,smoking history,diabetes,hemoglobin,low-density lipoprotein-cholesterol and UHR were included in the construction of the nomogram prediction model.The ROC curve AUC=0.757(95%CI 0.712-0.781),indicating that the prediction model has good discrimination.The Hosmer-Lemeshow goodness-of-fit test(P=0.924)and calibration curve,suggested that the prediction model has good calibration.In the internal validation,the C-statistic was 0.748 and the calibration curve suggested that the prediction model still has good discrimination and calibration.The DCA curves show that the model provides a net benefit to patients when they are assessed for the risk of AAA.In the external validation,the ROC curve AUC=0.762(95%CI 0.702-0.821)and the calibration curve,suggested that the prediction model still has good discrimination and calibration in the external validation.ConclusionsUHR is positively associated with presence of AAA and there is a nonlinear doseresponse relationship between them.UHR may serve as a novel and reliable biomarker for presence of AAA.In this study,we constructed a nomogram risk prediction model of AAA based on age,sex,smoking history,diabetes,hemoglobin,low-density lipoprotein-cholesterol and UHR.The prediction model was validated internally and externally with good discrimination and calibration.The model can be applied to predict the individual risk of AAA and risk stratification in the population,so as to accurately carry out AAA ultrasound screening and prevention,and provide an important basis for the formulation of AAA screening strategies in China.