Identification of Major Risk Factors and Non-linear Effects to the Development of Left Ventricular Hypertrophy in Chronic Kidney Disease by Constructing and Validation of Nomograms

BACKGROUND: Left ventricular hypertrophy (LVH) is a common cardiovascular complication among chronic kidney disease (CKD) patients. The present study aimed to identify major independent risk factors and determine their contribution and relationship to LVH development. METHODS: Clinical and echocardiographic data of 2002 pre-dialytic CKD patients were retrospectively collected. Independent risk factors for LVH were identified using univariable and multivariable logistic regression. Nomograms together with restricted cubic splines method were employed to explore the effect size and possible non-linear relationship with regard to LVH. A simplified predictive model was constructed and its predictive ability was validated to demonstrate to which extent the identified risk factors accounted for LVH risk. RESULTS: Multivariable logistic regression identified age, body mass index (BMI), systolic blood pressure (SBP), eGFR and hemoglobin as independent influencing factors for LVH. Nomogram revealed BMI, SBP and hemoglobin concentration as the most important risk factors. Impaired renal function only showed obvious risk for LVH when eGFR declined below 30 ml/min/1.73 m(2). Significant threshold effects existed for blood pressure and obesity that the risks for LVH doubled when SBP exceeded 160 mmHg or BMI exceeded 30 kg/m(2). The predictive model constructed performed well on both the training and validation cohort using calibration curve, ROC curve and AUC value, with AUC above 0.80 for both the training cohort and the validation cohort. CONCLUSIONS: With the help of nomogram model, we identified five independent factors that explain a large proportion of LVH risk in CKD patients. Among them, major contribution to LVH development was resulted from comorbidities and complications of CKD (hypertension, anemia, obesity) rather than eGFR reduction per se. Non-linear relationship and threshold relationship between eGFR, blood pressure, obesity and LVH risk were also identified.