Radiographic severity patterns of spinal deformity in osteogenesis imperfecta: a latent class analysis

Spinal deformities are common in osteogenesis imperfecta (OI) and may contribute to functional impairment, pulmonary compromise, and reduced quality of life. However, conventional classifications do not adequately capture the multidimensional nature of spinal involvement in OI. This exploratory study aimed to identify radiographic severity patterns of spinal deformity in patients with OI using latent class analysis (LCA).

A retrospective cross-sectional study was conducted in 62 patients with OI who underwent panoramic spinal radiographic evaluation in the coronal and sagittal planes. Thirteen categorical radiographic indicators, including scoliosis, Cobb angle patterns, thoracolumbar kyphosis, vertebral deformities, biconcave vertebrae, thoracic cage deformity, and sagittal alignment abnormalities, were analyzed using LCA. Competing two-, three-, and four-class models were evaluated using Akaike's Information Criterion, Bayesian Information Criterion, sample-size adjusted Bayesian Information Criterion, entropy, likelihood ratio tests, model stability, parsimony, and radiographic interpretability.

The three-class solution demonstrated superior statistical fit; however, the additional class showed substantial overlap with the remaining classes and predominantly represented an intermediate level of structural involvement. The two-class solution was therefore retained as the primary exploratory model based on parsimony and radiographic interpretability. The model-estimated class proportions were 40.6% for the lower-severity profile and 59.4% for the higher-severity profile. Overall model entropy was 0.929. Scoliosis and the main Cobb angle showed the greatest discriminatory contribution between classes, followed by vertebral deformity, biconcave vertebrae, and thoracic cage deformity. No significant associations were observed between class membership and age or sex.

Latent class analysis identified two exploratory radiographic severity profiles of spinal deformity in OI, reflecting lower and higher cumulative burdens of structural abnormalities. This probabilistic approach provides a multidimensional framework for describing radiographic patterns of spinal involvement but should not yet be interpreted as a prognostic or clinically validated classification. External and longitudinal validation are required before clinical application.