Abstract
Introduction Patients initiating hemodialysis (HD) are at high risk for cardiovascular disease. The ankle-brachial index (ABI) is a simple and widely used tool to detect peripheral artery disease and reflects different patterns of vascular calcification: low ABI indicates intimal arterial calcification, while high ABI may suggest medial arterial calcification (MAC). Coronary artery calcium score (CACS), a marker of coronary atherosclerosis, may reflect both IAC and MAC, though it does not distinguish between them. However, the association between ABI and CACS at HD initiation remains unclear. Methods In this single-center cross-sectional study, we included 204 patients who initiated HD and underwent both ABI and coronary CT between November 2013 and December 2023, at the time of dialysis initiation. Patients with ischemic heart disease or prior peritoneal dialysis were excluded to reduce confounding due to coronary interventions or altered baseline parameters. ABI was measured using an automated oscillometric device; the mean of bilateral values was used. ABI was categorized as low (≤0.90), normal (0.91-1.30), or high (>1.30). CACS was compared using the Wilcoxon rank-sum test. Multivariable restricted cubic spline (RCS) models with three knots assessed nonlinear associations between ABI and log-transformed CACS. Logistic regression evaluated the association between the ABI group and high CACS (>400). Models were adjusted for age, sex, diabetes, BMI, smoking status, estimated glomerular filtration rate, serum albumin, CRP, calcium, phosphate, and statin use. Subgroup analyses by sex and diabetes status were also performed. Results Median CACS values with IQRs were 930.5 (159.3-2241.1) for the low ABI group, 315.7 (58.5-1046.6) for the normal ABI group, and 54.3 (0.0-413.5) for the high ABI group. Compared to the normal group, low ABI was significantly associated with higher CACS (p = 0.037), while high ABI was associated with lower CACS (p = 0.002). RCS analysis in the unadjusted model showed a curve similar to group-wise comparisons. In the age-adjusted and fully adjusted models, the spline curve in the low ABI range tended to decline. Logistic regression showed a similar trend: low ABI was associated with high CACS in the unadjusted model (OR 2.74, p = 0.029), but the association diminished in the fully adjusted model (OR 1.29, p = 0.651). High ABI was associated with lower odds of high CACS across models, though not statistically significant. Although statistical significance was inconsistent, subgroup analyses stratified by sex and diabetes showed directionally similar trends, and interaction terms in logistic models were not significant. Conclusions In patients at the initiation of HD, low ABI (≤0.9) was associated with higher CACS, whereas high ABI (>1.3) was associated with lower CACS. Multivariable analysis indicated that elevated CACS in the low ABI group may be largely influenced by background factors such as age. In contrast, the relatively low CACS in the high ABI group may suggest delayed progression of MAC from peripheral to central arteries, although this remains a hypothesis-generating observation. These findings underscore the potential utility of ABI and CACS as complementary tools for early cardiovascular risk stratification in incident HD patients and may inform future research on vascular calcification dynamics.