Follow-up Bone Mineral Density Testing: 2023 Official Positions of the International Society for Clinical Densitometry

Characterizing low femoral neck BMD in Qatar Biobank participants using machine learning models

BACKGROUND

Identifying determinants of low bone mineral density (BMD) is crucial for understanding the underlying pathobiology and developing effective prevention and management strategies. Here we applied machine learning (ML) algorithms to predict low femoral neck BMD using standard demographic and laboratory parameters.

METHODS

Data from 4829 healthy individuals enrolled in the Qatar Biobank were studied. The cohort was split 60% and 40% for training and validation, respectively. Logistic regression algorithms were implemented to predict femoral neck BMD, and the area under the curve (AUC) was used to evaluate model performance. Features associated with low femoral neck BMD were subjected the statistical analysis to establish associated risk.

RESULTS

The final predictive model had an AUC of 86.4% (accuracy 79%, 95%CI: 77.98-80.65%) for the training set and 85.9% (accuracy 78%, 95% CI: 75.92-80.61%) for the validation set. Sex, body mass index, age, creatinine, alkaline phosphatase, total cholesterol, and magnesium were identified as informative features for predicting femoral neck BMD. Age (odds ratio (OR) 0.945, 95%CI: 0.945-0.963, p < 0.001), alkaline phosphatase (OR 0.990, 95%CI: 0.986-0.995, p < 0.001), total cholesterol (OR 0.845, 95%CI: 0.767-0.931, p < 0.001), and magnesium (OR 0.136, 95%CI: 0.034-0.571, p < 0.001) were inversely associated with BMD, while BMI and creatinine were positively associated with BMD (OR 1.116, 95%CI: 1.140-1.192, p < 0.001 and OR 1.031, 95%CI: 1.022-1.039, p < 0.001, respectively).

CONCLUSION

Several biological determinants were found to have a significant global effect on BMD with a reasonable effect size. By combining standard demographic and laboratory variables, our model provides proof-of-concept for predicting low BMD. This approach suggests that, with further validation, an ML-driven model could complement or potentially reduce the need for imaging when assessing individuals at risk for low BMD, which is an important component of fracture risk prediction.

CLINICAL TRIAL NUMBER

Not applicable.

Evaluating lower limits of body fat percentage in athletes using DXA

Body fat percentage (BF%) is routinely measured in athletes to monitor training and dietary interventions. Dual energy x-ray absorptiometry (DXA) is widely considered the gold standard body composition measurement technique, but DXA BF% values measure consistently higher than other techniques. Therefore, the main purpose of this study is to determine the lowest DXA-estimated whole-body fat mass in free-living athletes with unrestricted access to food. In this cross-sectional analyses, 732 participants across 18 competitive sports (643 athletes; 88 %) plus two (male and female) "non-athlete" male and female cohorts (89 non-athletes; 12 %) underwent a whole-body DXA scan using a Horizon A Hologic™ device (software version 13.6.0.5; TBAR 1209), performed and analyzed by a single operator. The average BF % in 454 males (20.9 ± 3.8 years) was 18.2 ± 4.9 % (range 10.3 - 37.2 %), with basketball players having the lowest BF% (15.3 ± 2.6 %) and non-athletes having the highest BF% (21.6 ± 5.4 %) (1-way ANOVA between 10 teams: F = 12.3; p < 0.0001). The average BF% in 278 females (21.3 ± 5.5 years) was 27.1 ± 5.1 % (range 16.2 - 45.3 %), with runners having the lowest BF% (23.5 ± 3.5 %) and non-athletes having the highest BF% (31.7 ± 6.3 %) (1-way ANOVA between 10 teams F = 12.9; p < 0.0001). In absolute values (kg), the average body fat for males was 19.9 ± 8.0 kg (range 6.7 - 62.0 kg) and 18.9 ± 6.4 kg (range 9.2 - 49.7 kg) for females. These data suggest that the lower limits of whole-body fat mass in free-living competitive athletes is approximately 10 % for males and 16 % for females. Whether these DXA-derived fat thresholds represent "healthy" levels, or how much of these DXA-derived fat depots represent essential fat stores located within lean soft tissue mass, remains unclear.

Integrating radiomics with clinical data for enhanced prediction of vertebral fracture risk

Introduction

Osteoporotic vertebral fractures are a major cause of morbidity, disability, and mortality among the elderly. Traditional methods for fracture risk assessment, such as dual-energy X-ray absorptiometry (DXA), may not fully capture the complex factors contributing to fracture risk. This study aims to enhance vertebral fracture risk prediction by integrating radiomics features extracted from computed tomography (CT) scans with clinical data, utilizing advanced machine learning techniques.

Methods

We analyzed CT imaging data and clinical records from 124 patients, extracting a comprehensive set of radiomics features. The dataset included shape, texture, and intensity metrics from segmented vertebrae, alongside clinical variables such as age and DXA T-values. Feature selection was conducted using a Random Forest model, and the predictive performance of multiple machine learning models-Random Forest, Gradient Boosting, Support Vector Machines, and XGBoost-was evaluated. Outcomes included the number of fractures (N_Fx), mean fracture grade, and mean fracture shape. Incorporating radiomics features with clinical data significantly improved predictive accuracy across all outcomes. The XGBoost model demonstrated superior performance, achieving an R2 of 0.7620 for N_Fx prediction in the training set and 0.7291 in the validation set. Key radiomics features such as Dependence Entropy, Total Energy, and Surface Volume Ratio showed strong correlations with fracture outcomes. Notably, Dependence Entropy, which reflects the complexity of voxel intensity arrangements, was a critical predictor of fracture severity and number.

Discussion

This study underscores the potential of radiomics as a valuable tool for enhancing fracture risk assessment beyond traditional clinical methods. The integration of radiomics features with clinical data provides a more nuanced understanding of vertebral bone health, facilitating more accurate risk stratification and personalized management in osteoporosis care. Future research should focus on standardizing radiomics methodologies and validating these findings across diverse populations.

Estimating Lumbar Spine Least Significant Change for Fewer than Four Vertebrae: The Manitoba BMD Registry

INTRODUCTION

The International Society of Clinical Densitometry recommends omitting lumbar vertebrae affected by structural artifact from spine BMD measurement. Since reporting fewer than 4 vertebrae reduces spine BMD precision, least significant change (LSC) needs to be adjusted upwards when reporting spine BMD change based on fewer than 4 vertebrae.

METHODOLOGY

In order to simplify estimating LSC from combinations of vertebrae other than L1-L4 (denoted LSCL1-4 ), we analyzed 879 DXA spine scan-pairs from the Manitoba BMD Program's ongoing precision evaluation. The additional impact on the LSC of performing the second scan on the same day vs different day was also assessed.

RESULTS

LSC progressively increased when fewer vertebrae were included, and also increased when the scans were performed on different days. We estimated that the LSCL1-4 should be adjusted upwards by 7 %, 24 % and 65 % to approximate the LSC for 3, 2, or 1 vertebral body, respectively. To additionally capture the greater LSC when the precision study was done on different days, LSCL1-4 derived from a precision study where scans were done on the same day should be adjusted upwards by 39 %, 60 % and 112 % for 3, 2, or 1 vertebral body, respectively.

CONCLUSION

LSCL1-4 derived from a precision study where scans are performed on the same day can be used to estimate LSC for fewer than 4 vertebrae and for scans performed on different days.