MO905: Calcification of Branches is Superior to Abdominal Aortic Calcification in Predicting Mortality of Hemodialysis Patients

Abstract

Abstract BACKGROUND AND AIMS Vascular calcification (VC) is considered to be an important initiating risk factor of cardiovascular events, which contributes to the leading cause of death in maintenance hemodialysis (MHD) patients. VC has a high incidence and has been reported to predict the mortality of MHD patients. Inspired by an extreme example of VC in dialysis patients, calciphylaxis (which involves extensive calcification of small- and medium- arteries with poor prognosis), we hypothesized that VC of branches was a better predictor of mortality than abdominal aortic calcification. Therefore, a cohort study based on MHD patients was constructed to investigate the distribution characteristics of VC, compare the predictive value of VC on mortality among different arteries, and explore a comprehensive model predicting mortality. METHOD A total of 1583 patients in our Blood Purification Center from January 2015 to September 2017 were screened, and 237 eligible patients were finally included. Abdominal-pelvic computed tomography images were collected and an open-source software 3D Slicer (version 4.11) was used to calculate the calcification volume in abdominal aorta, celiac trunk, superior and inferior mesenteric artery, renal artery, common iliac artery and internal iliac artery. Demographic and clinical data including medical history and some laboratory indexes were collected. Patients were followed-up until all-cause death or September 2020. Cause of death and survival time were recorded in detail. The area under the curve (AUC) of Time-dependent receiver operating curves (ROC) was calculated to evaluate the predictive value of arterial calcification. 80% of cases were randomly assigned to the training dataset and the remaining were assigned to the validation dataset. Lasso regression was performed to screen variables. Cox proportional hazards regression was performed to construct the predicting model and nomogram. C-index, calibration and decision curve analysis were used to evaluate the comprehensive model. SPSS software (version 26.0) and R software (version 4.1.2) perform statistical analysis. RESULTS The prevalence of total VC is 95.36%. Among the abdominal arteries, abdominal aortic calcification is the most prevalent (88.61%), followed by calcification of internal iliac artery (85.65%) and common iliac artery (80.17%). Median survival time was 37.00 (1.00–70.00) months. The 1-year survival rate, 3-year survival rate and 5-year survival rate were 73.42%, 50.01% and 35.43%, respectively. Cardiovascular disease accounts for 54.01% of the death cause. AUC of Time-dependent ROC by internal iliac artery and mesenteric artery calcification for predicting mortality is significantly higher than that of other arterial calcification. Internal iliac artery calcification (IIAC) was screened out as the most representative predictor for all-cause death among the calcification of arterial calcification by lasso regression. When analyzing IIAC and other clinical data, age (hazard ratio 1.03, P < 0.001), serum albumin level (hazard ratio 0.919, P < 0.001) and IIAC (hazard ratio 1.36, P = 0.003) were independent factor of mortality in HD patients. The nomogram based on Cox proportional hazards model is shown in Figure 1. C-index of the model was 74.09%. Decision curves of Cox proportional hazards model from training dataset and validation dataset showed the reliable predictive value of Cox model (Figure 2). CONCLUSION This is the first study evaluating the calcification volume of abdominal aortic branches and comparing their predictive values. In conclusion, calcification of abdominal aortic branches, especially internal iliac artery, is a better predictor for mortality of HD patients than abdominal aortic calcification. Our concern with calcification of large arteries in MHD patients needs to be shifted to smaller branch arteries.