Ultrasound-based transient elastography combined with metabolic biomarkers enhances the diagnostic efficacy for pediatric MASLD.

Nonalcoholic fatty liver disease, now encompassed within the updated metabolic dysfunction-associated steatotic liver disease (MASLD) nomenclature, is a common chronic liver disorder in children. This retrospective study evaluated the diagnostic performance of ultrasound-based transient elastography (TE) combined with metabolic biomarkers for clinically defined pediatric MASLD. A total of 334 children (278 in the MASLD group and 56 in the non-MASLD control group) were enrolled. MASLD status was determined retrospectively according to the Chinese Expert Consensus on the Diagnosis of Pediatric Nonalcoholic Fatty Liver Disease rather than liver biopsy. Liver stiffness measurement (LSM) and controlled attenuation parameter (CAP) were obtained using FibroTouch TE. Metabolic biomarkers covering liver function, lipid and glucose metabolism, inflammatory status, and uric acid metabolism were assessed. Receiver operating characteristic (ROC) analysis was used to evaluate discrimination, and exploratory combined models were constructed using logistic regression. Children with clinically defined MASLD showed significantly elevated LSM (7.52 ± 2.17 vs 5.76 ± 1.07 kPa, P < .001) and CAP (288.46 ± 46.76 vs 218.59 ± 28.16 dB/m, P < .001). CAP had the highest area under the ROC curve among single indicators (0.901, 95% confidence interval 0.868-0.935). The exploratory model incorporating CAP, LSM, C-reactive protein, uric acid, and homeostasis model assessment of insulin resistance yielded an apparent area under the ROC curve of 0.982 (95% confidence interval: 0.970-0.994), with sensitivity of 92.1% and specificity of 94.6% at the data-derived optimal cutoff. In this single-center retrospective cohort, combining TE parameters with metabolic biomarkers showed high apparent discrimination for clinically defined pediatric MASLD. Because the reference diagnosis was non-histologic and partly imaging-based, and because the model was evaluated in the same dataset in which it was developed without external validation, these estimates may be optimistic. Prospective multicenter validation against an independent reference standard is required before clinical implementation.