Statistical shape modelling of the first carpometacarpal joint: A cross-sectional analysis of an osteoarthritis initiative cohort

Introduction: Statistical shape modelling (SSM) has been used extensively to investigate the relationship between joint shape and hip, knee and spine osteoarthritis (OA), but first carpometacarpal joint (CMCJ) shape and thumb base OA is less understood. We established a statistical shape model of the first CMCJ to investigate the ability of SSM to evaluate CMCJ morphology and determine its ability to investigate association with thumb base OA.
Methods: 100 participants' bilateral hand and wrist radiographs were selected at random from the OAI database. A model was developed using SSM software to characterise first CMCJ shape and applied to radiographs. Radiographic OA severity was graded using Kellgren-Lawrence. Independent modes of variation in shape within the cohort were identified. Relationship with clinical symptoms and radiological severity was assessed using multivariate regression models adjusted for age, sex, height and weight.
Results: Ten and nine modes of variation were identified in right and left models respectively. Concurrent ipsilateral hand pain and stiffness was associated with right-mode-1 (RM1) (multivariate model, co-efficient 0.104, 95 % CI 0.011–0.197, p = 0.028), which anatomically represents greater prominence of ulnar aspect of trapezium joint surface. Stratified radiological OA severity showed CMCJ severity was associated with RM1,6,10 and LM 1,6,7 and STJ severity was associated with RM1 and LM1,6,7.
Conclusions: We demonstrate the application of SSM in a novel context, showing its ability to evaluate variation in CMCJ morphology from radiographs. This study identifies and justifies the requirement for novel automation techniques in SSM to facilitate longitudinal studies required to investigate clinical applications.