In Vivo Mechanical Demands on Vertebral Body Replacements During Rehabilitation Exercises: A Multidimensional and Longitudinal Analysis

Background: Mechanical complications remain a concern after vertebral body replacement (VBR), especially during rehabilitation. Yet exercise prescription is often guided by body posture or single loading measures. This study characterized mechanical demands during rehabilitation exercises after VBR and examined the effects of posture and postoperative time. Methods: Telemetric in vivo load data from instrumented VBRs in the OrthoLoad database were analyzed. A total of 119 trials across 21 exercises, performed in supine, prone, seated, and standing positions, were collected from five patients over 0.1–63 postoperative months. Mechanical demand was quantified across six biomechanical domains and integrated into a composite SafetyIndex. Posture- and time-related effects were assessed using linear mixed-effects models. Worst-case demand was defined as the 95th percentile of SafetyIndex values. Results: SafetyIndex showed a right-skewed distribution (median 8.5, IQR 3.7–14.1), with marked inter-exercise variability. Composite SafetyIndex did not differ between postures (all p > 0.13). However, posture-dependent effects emerged at the domain level: peak shear ratio was greater in prone than in sitting, standing, and supine positions (all p < 0.05); peak force was greater in standing than prone (p = 0.007 and p = 0.013 in unadjusted and adjusted models); and peak resultant moment was smaller in supine than prone (p = 0.036 and p = 0.046). Postoperative time was positively associated with peak force (β = +0.40 %BW/month, p = 0.042), peak resultant moment (β = +0.025 Nm/month, p < 0.001), and SafetyIndex (β = +0.25/month, p = 0.011), but not peak shear ratio (p = 0.879). Worst-case SafetyIndex_P95 values ranged from 0.6 to 85.0, with stable ranking across percentile thresholds (Spearman’s ρ = 0.995–0.997). Conclusions: Mechanical demand after VBR is task-specific and domain-dependent and cannot be inferred from posture alone. Axial and bending-related components increased over postoperative time, whereas shear-related loading remained task-dependent.