Role of REMS fracture risk prediction within the DVO-based fracture risk framework: a retrospective observational study

Technical Summary: Role of REMS Fracture Risk Prediction within the DVO-based Fracture Risk Framework

Problem Statement
Osteoporosis management in Germany has undergone a paradigm shift with the 2023 update to the German DVO (Deutsches Verbund Osteologie) guidelines. These guidelines moved from a 10-year fracture risk prediction model to a more clinically oriented 3-year assessment period, integrating Dual-energy X-ray Absorptiometry (DXA) T-scores with demographic, anthropometric, and clinical risk factors to establish therapeutic thresholds (3% to <5%, 5% to <10%, and ≥10%). While DXA remains the gold standard for bone mineral density (BMD) measurement, it is limited by radiation exposure and potential artifacts that may lead to BMD overestimation. Radiofrequency Echographic Multi-Spectrometry (REMS) is a radiation-free, ultrasound-based technology that provides BMD estimates (T-scores) alongside a proprietary Fragility Score (FS) reflecting bone quality and a 5-year fracture risk stratification. However, the potential role of REMS-derived parameters within the specific, country-mandated DVO 2023 risk algorithm remained unexplored. The study aimed to determine if REMS T-scores could substitute DXA T-scores within the DVO framework to yield comparable risk categories and to evaluate the internal coherence of these REMS-based categories with REMS-specific parameters (FS and 5-year risk).

Methodology
This retrospective observational study analyzed 256 patients (128 femoral and 128 spinal acquisitions) who underwent paired DXA and REMS examinations at the Orthopaedic Institute Dr. Baron&Colleagues.

• Data Collection: Patients were postmenopausal women or men aged 50–80 years. DXA scans were performed using a Discovery W scanner, and REMS scans were conducted using an EchoStation device.
• Risk Calculation: The 2023 German DVO risk algorithm was applied twice for each patient:
  1. Standard: Using the DXA-derived T-score.
  2. Exploratory: Using the REMS-derived T-score, while keeping all other variables (age, sex, BMI, clinical risk factors) constant.
• Statistical Analysis:
  • Agreement: The agreement between DXA-based and REMS-substituted DVO risk categories was assessed using percentage agreement and weighted Cohen's kappa ($\kappa$). A $\pm 0.3$ tolerance was applied to T-score thresholds to account for borderline cases.
  • Internal Coherence: The relationship between REMS-substituted DVO categories and REMS-specific outputs (Fragility Score and 5-year fracture risk classes R1–R7) was evaluated using Kruskal-Wallis tests and Wilcoxon rank-sum tests with false discovery rate (FDR) correction.

Key Results

• Diagnostic Agreement: There was substantial agreement between DXA-based and REMS-substituted DVO risk categories.
  • Femur: 81.2% agreement ($\kappa = 0.84$), increasing to 86.7% ($\kappa = 0.87$) with the $\pm 0.3$ tolerance.
  • Spine: 80.5% agreement ($\kappa = 0.85$), increasing to 83.6% ($\kappa = 0.87$) with the $\pm 0.3$ tolerance.
  • Complete agreement was observed in the highest-risk category ($\ge 10\%$) for femoral scans. Discordant cases were predominantly reclassified into adjacent risk categories.
• Internal Coherence with Fragility Score (FS): FS values increased progressively across higher DVO risk categories for both sites.
  • Femur: Median FS rose from 14.6 in the <3% group to 55.5 in the $\ge 10\%$ group ($p < 0.001$). All pairwise comparisons were significant.
  • Spine: Median FS rose from 24.1 in the <3% group to 40.7 in the $\ge 10\%$ group ($p < 0.001$). Significant differences were found between the lowest and highest risk groups.
• Internal Coherence with 5-Year Risk: The distribution of REMS 5-year fracture risk classes (R1–R7) shifted toward higher-risk categories as the REMS-substituted DVO thresholds increased. For instance, in the femoral $\ge 10\%$ DVO group, the majority of patients fell into the higher REMS risk classes (R5–R7), whereas the <3% group was dominated by R1 and R3.

Key Contributions

• Validation of REMS in DVO Framework: The study provides the first evidence that REMS-derived T-scores can be substituted for DXA T-scores within the German DVO 2023 risk algorithm with high diagnostic agreement.
• Clinical Coherence: The study demonstrates that REMS-based risk stratification aligns with REMS-specific measures of bone quality (FS) and short-term fracture probability, suggesting that the REMS parameters are internally consistent with the DVO clinical decision-making structure.
• Insight into Discordance: The analysis of discordant cases suggests that REMS may identify higher risk in specific populations (e.g., patients with Type 2 diabetes or vertebral deformities) where DXA might overestimate BMD due to artifacts or fail to capture bone quality deficits.

Significance and Claims
The authors conclude that REMS-derived parameters show substantial coherence with DVO risk groups. The study claims that REMS technology supports a timely indication for intervention and preventive strategies within the DVO-based clinical pathway. Specifically, the technology is presented as a valuable, radiation-free alternative that allows for fracture risk stratification consistent with current German guidelines. The authors note that while REMS offers a 5-year risk horizon compared to the DVO's 3-year horizon, the internal coherence observed supports the use of REMS in clinical pathways, particularly in scenarios where DXA applicability is limited or where bone quality assessment (via FS) provides complementary information not captured by BMD alone. The study acknowledges limitations, including its retrospective, single-center design and the lack of longitudinal follow-up to validate the predictive performance of REMS-substituted estimates against actual incident fractures.