Supplementary Material for: Novel Approach to Cardiovascular Outcome Prediction in Haemodialysis Patients

Background: Cardiovascular mortality is high in haemodialysis (HD) patients. Arterial stiffness and global longitudinal strain (GLS) are important non-atheromatous cardiovascular risk predictors. No study has encompassed both parameters in a combined model for prediction of outcomes in HD patients. This is important because left ventricular (LV) dysfunction can result from fibrotic remodelling secondary to increased arterial stiffness. Methods: Two hundred and nineteen HD patients had pulse wave velocity (PWV) and echocardiography (including GLS) assessments. Patients were followed-up until death, transplantation or November 16, 2015, whichever happened first. Pearson's correlation coefficient was used to determine factors associated with PWV and GLS. A multivariable Cox regression model investigated factors associated with all-cause, cardiac death and events. Results: One hundred and ninety eight HD patients had full datasets (median age 64.2, 68.7% males) with a mean LV ejection fraction (LVEF) of 61.7 ± 10.1% and GLS -13.5 ± 3.3%; 51% had LV hypertrophy. Forty eight deaths (15 cardiac) and 44 major cardiac events occurred during a median follow-up of 27.6 (25th-75th percentile, 17.3-32.7) months. In separate survival models, PWV and GLS were independently associated with all-cause mortality; however, in a combined model, LV mass indexed to height^2.7 (LVMI/HT^2.7; adjusted hazard ratio (HR) 1.02, 95% CI 1.00-1.04) and PWV (adjusted HR 1.23, 95% CI 1.03-1.47) were significant. PWV was neither associated with cardiac death nor associated with related cardiac events. However, GLS was associated with cardiac death (adjusted HR 1.24, 95% CI 1.00-1.54) and cardiac events (adjusted HR 1.13, 95% CI 1.03-1.25). Conclusions: PWV and LVMI/HT^2.7 were superior to GLS in prediction of all-cause mortality. However, GLS was associated with cardiac death and events even when accounting for LVEF and LVMI/HT^2.7.