Bone metabolism in complex regional pain syndrome

Abstract

Introduction: Patients with complex regional pain syndrome (CRPS) often show disturbed bone metabolism, assessed using three-phase bone scintigraphy (TPBS). However, current methods lack automation and standardisation. Bone serum markers have been proposed as biomarkers, but their utility is unclear.

Objectives: This study aimed to evaluate bone metabolism in CRPS using TPBS and bone serum markers.

Methods: A deep learning model for automated segmentation quantified tracer enhancement in TPBS images. Serum markers analysed included alkaline phosphatase (AP), 25-OH vitamin D, osteoprotegerin, procollagen type I N-terminal propeptide (PINP), and β-C-terminal telopeptide, compared to 48 healthy controls (HC). The study included 114 patients with CRPS, 41 of whom underwent TPBS.

Results: Of the 41 patients with CRPS with TPBS, 39 (95.1%) displayed radiotracer enhancement in the bone phase across CRPS subtypes. Serum markers of 114 patients did not significantly differ between patients and HC, nor did they correlate with TPBS enhancement, except in warm CRPS. In these patients, TPBS accumulation in the metacarpophalangeal region correlated with PINP (Spearman ρ = 0.63, P = 0.038), and AP levels were elevated at 78 U/L (interquartile range 64-88) compared to cold CRPS at 66 U/L (51-77; P = 0.003) and HC at 60 U/L (53-69; P < 0.001).

Conclusion: Automated TPBS quantification revealed widespread bone metabolism alterations, common in CRPS and detectable beyond qualitative assessment. Although most serum markers remained unchanged, patients with warm CRPS exhibited unique features, suggesting distinct pathophysiological profiles. Integrating novel image analysis with other biomarkers may enhance diagnostic precision and patient stratification for targeted therapies.

Keywords: Biomarkers; Bone markers; Complex regional pain syndrome; Three-phase bone scintigraphy.

Figure 1.

Schematic of the automated image segmentation process. (1) Illustration of image acquisition in three-phase bone scintigraphy for all 3 phases. (2) Annotation process for the hands (with their regions and subregions) and feet. (3) Example of iterative training for image segmentation and prediction of prox. interphalangeal region, and quantification of radiotracer accumulation by calculating the ratio between the mean pixel intensity values of the ipsilateral and contralateral sides. Phases 1–3, p1–p3; metacarpophalangeal and prox. interphalangeal subregions of the hands, MCP and PIP.

Figure 2.

Increased bone turnover distal to the injury in almost all patients, independent of CRPS subtype and injury location. The ipsilateral/contralateral scintigraphy ratios are shown for all regions and subregions, and the dotted line indicates a ratio of 1, which is equivalent to no difference. Ratios >1 correspond to accumulation in the injured limb. Lines indicate the medians and first and third quartile. (A) Upper extremity ratios (N = 28) and (B) lower extremity ratios (N = 13). (C) No statistically significant differences were observed between the warm and cold temperature subtypes of the hands. (D) No statistically significant differences between cases with (N = 9) and without (N = 19) injury to the distal hand. Pairwise comparisons with Mann–Whitney U test. Carpal, metacarpophalangeal, and prox. interphalangeal subregions of the hands, carpal, MCP, and PIP.

Figure 3.

Elevated alkaline phosphatase levels in patients with CRPS, but no difference in specific bone serum markers. (A) Schematic representation of bone serum markers. (B–F) Pairwise comparisons between bone serum markers in healthy controls (HC, n = 48) and patients with CRPS (n = 114 for alkaline phosphatase). Values are the median and the first and third quartile. Pairwise comparisons with Mann–Whitney U test.

Figure 4.

Higher AP levels in patients with warm CRPS who also demonstrated correlations between PINP levels and scintigraphy ratios. (A) Pairwise comparisons between AP levels in healthy controls (HC) and subgroups of patients with cold and warm CRPS. Lines indicate values as medians and the first and third quartile. Differences between groups were calculated using the Kruskal–Wallis test and Dunn post-hoc test with Bonferroni correction. (B and C) Correlations between scintigraphy ratios and MCP and PIP regions in the warm CRPS subtype. Calculations with Spearman correlations were adjusted for multiple testing using Bonferroni correction. PINP, procollagen type 1 N-terminal propeptide; metacarpophalangeal and prox. interphalangeal subregions of the hands, MCP, PIP.