Improving neuroblastoma risk prediction through a polygenic risk score derived from genome-wide association...

Neuroblastoma is the most common extracranial solid tumor in children and has complex genetic underpinnings. Previous genome-wide association studies (GWASs) have identified many loci associated with neuroblastoma susceptibility; however, their application in risk prediction for Chinese children has not been systematically explored. This study seeks to enhance neuroblastoma risk prediction by validating these loci and evaluating their performance in polygenic risk models.

We validated 35 GWAS-identified neuroblastoma susceptibility loci in a cohort of Chinese children, consisting of 402 neuroblastoma patients and 473 healthy controls. Genotyping these polymorphisms was conducted via the TaqMan method. Univariable and multivariable logistic regression analyses revealed the genetic loci significantly associated with neuroblastoma risk. We constructed polygenic risk models by combining these loci and assessed their predictive performance via area under the curve (AUC) analysis. We also established a polygenic risk scoring (PRS) model for risk prediction by adopting the PLINK method.

Fourteen loci, including ten protective polymorphisms from CASC15, BARD1, LMO1, HSD17B12, and HACE1, and four risk variants from BARD1, RSRC1, CPZ and MMP20 were significantly associated with neuroblastoma risk. Compared with single-gene model, the 8-gene model (AUC=0.72) and 13-gene model (AUC=0.73) demonstrated superior predictive performance. Additionally, a PRS incorporating six significant loci achieved an AUC of 0.66, effectively stratifying individuals into distinct risk categories regarding neuroblastoma susceptibility. A higher PRS was significantly associated with advanced International Neuroblastoma Staging System (INSS) stages, suggesting its potential for clinical risk stratification.

Our findings validate multiple loci as neuroblastoma risk factors in Chinese children and demonstrate the utility of polygenic risk models, particularly the PRS, in improving risk prediction. These results suggest that integrating multiple genetic variants into a PRS can enhance neuroblastoma risk stratification and potentially improve early diagnosis by guiding targeted screening programs for high-risk children.