DEVELOPMENT AND EVALUATION OF CLINICAL RISK PREDICTION MODELS USING BONE TURNOVER MARKER FOR OSTEOPOROSIS SCREENING IN CHRONIC HEMODIALYSIS PATIENTS

Introduction:

Patients with chronic kidney disease (CKD), particularly those on hemodialysis (HD), are at elevated risk for fractures, which contribute to significant morbidity and mortality. Reduced bone mineral density (BMD) is a critical predictor of fracture risk and is recommended by international guidelines for assessment in this population. However, BMD testing is expensive, resource-intensive, and not widely reimbursed. This study aimed to develop and validate new clinical screening tools to identify HD patients at risk of osteoporosis, potentially reducing the reliance on BMD testing.

Methods:

This prospective observational study enrolled stable adult HD patients between January 2022 and May 2024. BMD of the lumbar spine, total hip, femoral neck, and distal one-third of the radius were measured using dual-energy X-ray absorptiometry. Serum intact parathyroid hormone (PTH) and bone turnover markers, including bone-specific alkaline phosphatase (BAP) and tartrate-resistant acid phosphatase type 5b (TRAP 5b), were measured concurrently with BMD testing. Osteoporosis was defined by a BMD T-score < -2.5, and osteopenia by a BMD T-score between -1 to -2.5. Participants were randomly split into training (90%) and validation (10%) sets. A prediction model was developed using multivariable logistic regression with variable selection via LASSO method.

Results:

A total of 353 chronic HD patients underwent blood and BMD testing. The mean age was 53.7 ± 15 years, with 49% female; of these, 70% were post-menopausal. The mean BMI was 23.5 ± 4.4 kg/m². Median serum intact PTH was 306.5 pg/mL (IQR: 132.6 - 616.6 pg/mL), with 48% of patients within the KDIGO recommended range. Interestingly, 87% of patients were diagnosed with either osteopenia or osteoporosis (T-score < -1), and 49% met the WHO criteria for osteoporosis. The distal radius was the most common site of osteoporosis (40%), followed by the femoral neck (28%), lumbar spine (16%), and total hip (14%). However, when using BMD testing for general population based on age, sex, BMI, and menopausal status, only 137 patients (38.8%) met these conventional indications. The new prediction models based on CKD-MBD markers were explored. The final prediction model from four covariates: log age, log BMI, menopausal status (if female), and log BAP was selected. Within the validation cohort, the model achieved an area under the ROC curve of 0.84 (95% CI 0.71 - 0.97). At a high-sensitivity cutoff of 0.42, the model demonstrated a sensitivity of 92.9% and specificity of 52.4%, with likelihood ratios of 0.14 for a negative test and 1.95 for a positive test.

Conclusions:

The newly developed clinical risk prediction model shows promise as a tool for identifying HD patients at risk of osteoporosis, with reasonable sensitivity and specificity. Only conventional indication for BMD testing in general population is not adequate even patients who had serum PTH levels within KDIGO guideline suggestion. This model could serve as a cost-effective alternative to BMD testing, particularly in resource-limited settings. Further refinement and external validation are needed before widespread implementation.

I have no potential conflict of interest to disclose.

I used generative AI and AI-assisted technologies in the writing process.
During the preparation of this work the authors used ChatGPT 4o in order to assist with refining the abstract. After using this tool, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication.