A Nomogram For Predicting Nonalcoholic Fatty Liver Disease In Obstructive Sleep Apnea Adults

Objective: Non-alcoholic fatty liver disease(NAFLD)is becoming more prevalent in patients with obstructive sleep apnea(OSA),but NAFLD is not easily detectable.The aim of this study was to explore the risk factors for the development of NAFLD in OSA patients and to develop and validate a visual nomogram evaluation tool to aid clinical prediction of NAFLD in OSA patients,so as to save more medical resources and reduce missed diagnoses.Methods: This study was a cross-sectional study,retrospectively included 466 OSA patients admitted to the Second Affiliated Hospital of Kunming Medical University from June 2019 to June 2022.All individuals were randomly divided into training dataset and validation dataset at 70%(n = 328)and 30%(n = 138),using the classification and regression training(caret)package in R software.The least absolute shrinkage and selection operator(LASSO)for reducing the data dimension and preliminary selecting risk factors in the training dataset.The risk variables selected in LASSO regression analysis were incorporated into multivariate Logistic regression analysis to determine the independent influencing factors for NAFLD with OSA patients.A nomogram incorporating the selected independent risk factors in the training dataset was constructed.Then,we used the receiver operating gharacteris(ROC)area under the curve(AUC),Hosmer-Lemeshow test,calibration curve,and decision curve analysis(DCA)to test the discrimination,calibration,and clinical meaning of the nomogram.At last,internal validation was used in the validation dataset.Results: Among 466 OSA patients,175 had no NAFLD and 291 had NAFLD.A total of five predictors,namely type 2 diabetes(OR=1.825,95CI%: 1.01 ～ 3.30,P=0.047),BMI(OR=1.174,95CI%: 1.09 ～ 1.26,P<0.001),ALT(OR=1.037,95CI%:1.02 ～ 1.06,P<0.001),TG(OR=1.467,95CI%: 1.12 ～ 1.92,P=0.005),AHI(OR=1.016,95CI%: 1.00 ～ 1.03,P= 0.024),were identified by LASSO regression analysis and multivariate Logistic regression analysis from a total of 19 variables studied.The model constructed using these five predictors displayed medium prediction ability.The AUC of the nomogram in the training dataset and validation dataset were 0.811 [95%CI(0.765 ～ 0.856)] and 0.796 [95%CI(0.717 ～ 0.876)].The Hosmer-Lemeshow test results were P>0.05,indicating that the model had a good fit.The calibration curves showed good agreement between the predicted and observed results in the training and validation datasets.The DCA curve showed that the nomogram could obtain a net clinical benefit if the risk threshold was morn than0.23 in the training dataset and morn than 0.24 in the validation dataset.Conclusion: 1.Type 2 diabetes,BMI,ALT,TG,and AHI are independent influencing factors of NAFLD in OSA patients based on LASSO regression and multivariate Logistic regression analysis.2.Based on the type 2 diabetes,BMI,ALT,TG,and AHI five predictors,this study developed a simple,noninvasive,effective and convenient nomogram that achieved an optimal detection of NAFLD in OSA patients.Using this nomogram,it helps clinicians to early identify the condition of NAFLD in OSA patients,intervene early,and improve the prognosis of NAFLD patients.