MON-811 Novel Subclassification Of Low Bone Mass And Its Association With Comorbidities In Korean Older Adults: A Data-driven Cluster Analysis

Abstract Disclosure: S. Lee: None. T. Lee: None. G. Yoo: None. J. Lim: None. Osteoporosis is commonly observed in older adults and is often accompanied by other chronic conditions. Although the bone mineral density (BMD) T-score is the gold standard for diagnosing osteoporosis, fracture risk varies by age and sex even at the same T-score. Moreover, more fractures occur in individuals with osteopenia than in those with osteoporosis. Since T-score alone does not fully account for fracture risk, various approaches have been developed, including the Fracture Risk Assessment Tool; however, their clinical applicability remains limited. In this study, we aimed to classify patients with low bone mass (LBM) based on several clinical parameters, thereby identifying subgroups that may benefit more from osteoporosis treatment. We conducted a data-driven cluster analysis using K-means and hierarchical clustering in patients with LBM (T-score < -1.0) from the Korea National Health and Nutrition Examination Survey cohort. Individuals with normal BMD (T-score ≥ -1.0), age < 50 years, those with extreme values of alkaline phosphatase (ALP) or parathyroid hormone (PTH), and those with an estimated glomerular filtration rate < 45 mL/min/1.73 m2 were excluded. Clustering was performed using nine variables: age, body mass index (BMI), lumbar spine, femur neck, and total hip T-scores, delta lumbar spine/ total hip areal BMD, ALP, PTH, and 25-hydroxyvitamin D levels. To distinguish individuals with suspected vertebral compression fracture, the presence of marked height loss, which defined as a reduction of more than 4 cm from peak adult height, was also investigated. As a result, a total of 2,803 participants were included in the study, with a mean age of 64.8 years. Five distinct clusters were identified, each with significantly different clinical characteristics and comorbidities. The clusters were classified as follows: cluster 1 (mild vertebrae-dominant LBM), cluster 2 (mild hip-dominant LBM), cluster 3 (moderate LBM with relatively high bone turnover), and cluster 4 (severe LBM). As the cluster number increased from 1 to 4, participants tended to be older. Cluster 3, in particular, had a higher BMI, along with elevated systolic blood pressure and PTH levels, and lower 25-hydroxyvitamin D levels, compared to other clusters. Notably, individuals in the cluster 3 had the highest prevalence of cardiovascular disease and chronic kidney disease; and those in the cluster 4 showed the highest prevalence of marked height loss. In conclusion, we stratified individuals with LBM into four subgroups with clinically distinct features, including comorbidities. This data-driven classification may help guide individualized management strategies beyond bone health in the elderly by compensating for the limitations of BMD-based diagnostics. Presentation: Monday, July 14, 2025