1 Nutrition & Health Innovation Research Institute, School of Medical and Health Science, Edith Cowan University, Perth, Australia; College of Health Sciences, Mekelle University, Mekelle, Ethiopia.
2 Nutrition & Health Innovation Research Institute, School of Medical and Health Science, Edith Cowan University, Perth, Australia. Electronic address: a.gebre@ecu.edu.au.
3 Nutrition & Health Innovation Research Institute, School of Medical and Health Science, Edith Cowan University, Perth, Australia; Medical School, University of Western Australia, Perth, Australia.
4 Nutrition & Health Innovation Research Institute, School of Medical and Health Science, Edith Cowan University, Perth, Australia; School of Science, Edith Cowan University, Perth, Australia; Department of Computer Science and Software Engineering, University of Western Australia, Perth, Australia.
5 Nutrition & Health Innovation Research Institute, School of Medical and Health Science, Edith Cowan University, Perth, Australia; School of Science, Edith Cowan University, Perth, Australia.
6 Department of Internal Medicine and Radiology, University of Manitoba, Winnipeg, Canada.
7 HealthPartners, Park Nicollet Clinic and HealthPartners Institute, Minneapolis, USA; Division of Health Policy and Management, University of Minnesota, Minneapolis, USA.
8 Department of Medicine, University of Wisconsin, Madison, USA.
9 Department of Computer Science, Concordia University, Montreal, Canada.
10 George and Fay Yee Centre for Healthcare Innovation, University of Manitoba, Winnipeg, Canada.

Background: Sagittal abdominal diameter (SAD), a measure of visceral adiposity, has been linked to major adverse cardiovascular events (MACE). However, the relationship between SAD and abdominal aortic calcification (AAC), a marker of subclinical vascular disease, and whether they independently and jointly predict MACE remains unclear.

Objective: To investigate whether weight-normalized SAD and AAC scored using a validated machine learning algorithm (ML-AAC24) are independently and jointly associated with incident MACE.

Methods: SAD and ML-AAC24 were measured from dual-energy X-ray absorptiometry (DXA) posteroanterior and lateral spine images, respectively, from the Manitoba Bone Density registry.

Results: Among 8806 individuals (mean age 75.1 ± 6.6 years, 93.9% women), 11.3% experienced MACE during a mean follow-up of 3.8 years. SAD/weight and ML-AAC24 were positively correlated (Spearman r = 0.11, P < 0.001). Individuals with moderate and high ML-AAC24 had 1.1% and 3.0% higher mean SAD/weight, respectively, than those with low ML-AAC24. Both ML-AAC24 and SAD/weight were independently associated with higher risk of MACE. Adjusted hazard ratios [HRs] for MACE were 1.45, 95%CI 1.24-1.71 and 1.99, 95%CI 1.67-2.35 for moderate and high ML-AAC24, respectively, vs. low. The HR for the highest vs. lowest tertile of SAD/weight was 1.37, 95%CI 1.16-1.61. Individuals who had both high ML-AAC24 and were in the highest SAD/weight tertile had the highest MACE risk (HR 2.63, 95% CI 2.02-3.44).

Conclusion: Higher baseline SAD/weight was associated with higher ML-AAC24 scores. Both measures independently and jointly associated with MACE. Their combined use may potentially help identify individuals at high risk for cardiovascular disease during routine bone density testing.