Predicting steroid-induced osteonecrosis of the femoral head: role of lipid metabolism biomarkers and radiomics in young and middle-aged adults

Femoral head necrosis is a common orthopedic disease that results in significant physical disability in patients. Early prediction and diagnosis of steroid-induced osteonecrosis of the femoral head (SONFH) are crucial for the prevention and treatment of this condition.

In this study, initial CT images and clinical data of patients with SONFH, admitted from January 2019 to December 2022, were collected. Patients were grouped as follows: (1) those diagnosed with SONFH at the initial diagnosis (control group), and (2) those with high-risk factors but no symptoms at first diagnosis, who developed SONFH two years later (experimental group). CT imaging histological features, clinical characteristics, and transcriptome screening for differentially expressed genes, pathway enrichment, and immune infiltration analyses were performed.

Significant differences were found in triglyceride (TG) levels between the training and validation groups. Age, sex, alkaline phosphatase (ALP), and hemoglobin levels differed between the training and internal validation groups, while HDL and red blood cell counts varied between the training and external validation groups. Univariate analysis showed that age, TG, HDL, and Radiomics scores influenced SONFH, while multivariate analysis revealed TG, HDL, and Radiomics scores were closely related to SONFH. Transcriptomic analysis showed associations with sphingolipid and adipocyte signaling pathways, along with immune cell involvement, linking SONFH to lipid metabolism and atherosclerosis.

These findings indicate a significant association between steroid-induced osteonecrosis of the femoral head and age, with TG and HDL serving as indicators of lipid metabolism closely correlated with the occurrence of SONFH. Radiomics scores were also found to correlate with SONFH occurrence, supported by transcriptomic and CT imaging findings. However, this study has limitations, including its retrospective design and a relatively limited sample size, which may impact the generalizability of the results. Further prospective studies with larger, more diverse populations are needed to validate and enhance the predictive model.