Abdominal aortic calcification detected during bone imaging tied to CV risk
Key takeaways:
- A machine learning algorithm can assess abdominal aortic calcification level during a vertebral fracture assessment.
- The percentage of adults with major adverse CV events rose with higher calcification score.
Machine learning may be able to identify older adults most at risk for major adverse cardiovascular events by assessing abdominal aortic calcification in vertebral fracture assessment images, according to study data.
Researchers in the province of Manitoba, Canada, used a machine learning model to categorize older adults undergoing osteoporosis screening as having low, medium or high abdominal aortic calcification (AAC). During a mean 3.9 years of follow-up, rates of major adverse CV events were highest among those with high AAC and were smaller for adults with low AAC.
“No community-based screening strategy currently exists to assess vascular calcification,” William D. Leslie, MD, MSc, FRCPC, professor of medicine and radiology at the University of Manitoba in Winnipeg, Canada, and colleagues wrote in a study published in the Journal of Bone and Mineral Research. “In a routine clinical setting with vertebral fracture assessment imaging, we provide robust estimates for the risk of major adverse CV events and its components in the next 3.9 years. Our data suggest that you only need to screen four people who undergo routine vertebral fracture assessment to identify one high-risk person who may benefit from preventative lifestyle approaches ... or more rigorous CVD risk assessment and management.”
The researchers noted that AAC has not been examined during vertebral fracture assessments in the past due to cost, the time it took to read images and the lack of health care professionals trained to assess the images. They wrote that machine learning can overcome these barriers and provide AAC analysis for more patients undergoing vertebral fracture assessments in the future.
Researchers obtained vertebral fracture assessment images from adults aged 70 years or older who had a bone mineral density T-score of –1.5 or lower at the lumbar spine, total hip or femoral neck; or adults aged 50 to 69 years with historical height loss of more than 5 cm, measured height loss of more than 2.5 cm or those using glucocorticoids for at least 3 months in the prior year. A machine learning model scored each participant’s image for AAC. A score of less than 2 was considered low AAC, a score of 2 to 5 was defined as medium AAC and a score of 6 or greater was categorized as high AAC. Major adverse CV events were obtained from medical records.
Of 10,250 adults in the study, 42% had low AAC, 32.8% had moderate AAC and 25.3% had a high AAC score.
During follow-up, 12.3% of adults had a major adverse CV event. The percentage of participants with a major adverse CV event in the high AAC group was 19.2% compared with 13.1% for those with moderate AAC and 7.6% for adults with low AAC.
Among 7,327 adults not using statins, 46.2% had low AAC, 32.1% had moderate AAC and 21.7% had high AAC. Among nonstatin users, the proportion of adults having a major adverse CV event was 17.8% in the low AAC group, 36.1% for those with moderate AAC and 59.2% for adults with high AAC
Compared with adults with a low AAC, those younger than 80 years with a moderate or high AAC score had greater risks for major adverse CV events than adults aged 80 years or older (P for interaction = .013). Researchers also found a borderline interaction by sex. Compared with those with a low AAC score, women with a moderate or high AAC score had higher hazard ratios for major adverse CV events than men (P for interaction = .064).
“Opportunistic AAC assessment at the time of DXA (specifically vertebral fracture assessments) represents a novel community-based screening strategy to identify high-risk women for CVD and could help to address the sex disparity in CVD,” the researchers wrote.
Collapse
Read more about