Renal MRI radiomics in Beckwith–Wiedemann syndrome: a novel imaging approach for genotype identification

TL;DR

This study explores how MRI radiomics can help identify genetic subtypes of Beckwith–Wiedemann syndrome by analyzing kidney features.

Contribution

The paper introduces a novel MRI-based radiomics approach for genotype identification in Beckwith–Wiedemann syndrome.

Findings

The BWSUPD+IC1 subtype showed larger total renal volume and weaker age-volume correlation compared to BWSIC2.

Radiomics models using T2WI, ADC, and combined data achieved high accuracy in distinguishing BWS subtypes.

91.84% of BWS patients had renal volumes exceeding the normal population's upper limit.

Abstract

To valuate the role of nonmalignant nephrological findings and renal MRI radiomics in differentiating molecular subtypes of Beckwith–Wiedemann syndrome (BWS).

Clinical data and abdominal MRI scans of 49 patients who underwent partial glossectomy between July 2019 and March 2024 were retrospectively analysed. Patients were categorized into two subtypes: BWSUPD+IC1 (24 cases, with a predisposition to renal involvement) and BWSIC2 (25 cases, with a lower risk of renal involvement), based on genetic testing. Pearson correlation analysis was conducted to evaluate the relationship between patients’ age and renal volume. Radiomic features derived from the T2WI sequence and the ADC map were selected to construct single-sequence and combined models. Delong test was used to compare the performance of the models.

Clinically, the BWSUPD+IC1 subtype exhibited a lower incidence of ear creases/pits (P = 0.048) and omphalocele/umbilical hernia (P = 0.032) compared to the BWSIC2 subtype. Abdominal MRI findings indicated the BWSUPD+IC1 subtype had larger total renal volume (P = 0.017) and a weaker correlation between total renal volume and patients’ age (r = 0.38). Notably, 91.84% (45/49) of BWS patients exhibited a total renal volume exceeding the normal population’s upper limit, with the IC1 subtype demonstrating the largest mean volume. The BWSUPD+IC1 subtype showed higher incidences of nonmalignant renal (P = 0.013) and non-renal abdominal abnormalities. The T2WI, ADC, and combined models achieved the highest area under the receiver operating characteristic (ROC) curves (AUCs) of 0.837, 0.882 and 0.954 (P > 0.05), respectively.

Nonmalignant renal abnormalities and MRI radiomics models have potential as alternative imaging tools for the identification of renal predisposition genotypes and the surveillance of renal size change in BWS patients.

The online version contains supplementary material available at 10.1186/s13023-025-03841-x.

Figures