Adiposity-lipid-glycemic clusters as potential warning signals of bone mass reduction in Asia's largest urban communities - based bone health assessment via ultrasound.

Among 15,052 individuals, the majority had some degree of reduced bone mass, with osteopenia being most prevalent.

• 55.7% had normal bone mass, 43.0% had osteopenia, and 1.3% had osteoporosis
• Study population was drawn from routine health check-ups at Beijing Tsinghua Changgung Hospital from 2017 to 2024
• BMD was assessed through ultrasound rather than DXA
• All findings were statistically significant (P < 0.001)

Age at diagnosis increased with decreasing bone mass status across the three bone mass categories.

• Median age was 43 years for normal bone mass, 53 years for osteopenia, and 65 years for osteoporosis
• Bone mass was significantly associated with age (P < 0.001)
• ROC analysis identified optimal age cutoffs of 47 years for males and 49 years for females as predictors of bone mass reduction
• Bone mass reduction rose sharply from 27.2% to 53.4% between ages 30-59

XGBoost achieved the highest predictive performance for bone mass reduction in the overall population.

• XGBoost achieved an AUC of 0.734 in the overall model
• Models compared included Logistic Regression, Support Vector Machine (SVM), Random Forest, and XGBoost
• All models were evaluated with 5-fold cross-validation
• LASSO regression identified age, gender, HDL-C, TG, HbA1c, and FMI as key predictive factors
• The best model was deployed in a Shiny web application for real-time prediction

Gender-stratified analysis revealed different key predictors of bone mass reduction for males and females.

• In males, age, HDL-C, FMI, and fasting blood glucose (FBG) were significant factors, with XGBoost AUC = 0.687
• In females, age, TG, and FMI were key factors, with XGBoost AUC = 0.770
• Female model outperformed male model by AUC (0.770 vs. 0.687)
• Bone mass was significantly associated with gender (P < 0.001)

Osteopenia and osteoporosis were linked to higher fat mass index (FMI) and metabolic markers in the body composition subsample.

• Body composition was measured in 4,999 of the 15,052 participants using multi-frequency bioelectrical impedance analysis
• FMI was identified as a key factor by LASSO regression
• Group medians in the osteopenia/osteoporosis groups fell within reference ranges, yet some individuals had values close to either limit
• Bone mass was significantly associated with BMI and metabolic markers (P < 0.001)

Metabolic markers including HDL-C, LDL-C, TG, and HbA1c were associated with bone mass reduction even when their values were within normal ranges but near reference limits.

• Variations in HDL-C, LDL-C, TG, and HbA1c near reference limits but within normal ranges were associated with bone mass reduction
• These near-limit normal values may serve as early warning indicators of bone mass reduction
• No associations were found between smoking or drinking status and BMD
• LASSO regression identified HDL-C, TG, and HbA1c as key factors alongside age, gender, and FMI

Ultrasound-based BMD assessment was used as an alternative to DXA for community-level bone health screening.

• The study cited limitations of dual-energy X-ray absorptiometry (DXA) in early-stage bone mineral density assessment as motivation for using ultrasound
• The cross-sectional study included 15,052 individuals from routine health check-ups
• The study aimed to identify the age at which bone mass reduction begins and associated risk factors
• The community-based approach was noted as potentially useful for monitoring bone health in comparable aging populations