Abstract 5283: Cloud-based computational framework for individualized genomic analysis in pediatric acute lymphoblastic leukemia: A nationwide multi-center real-world clinical study

Abstract Acute lymphoblastic leukemia (ALL) is the most common childhood cancer. While genomic studies have identified key molecular subtypes and aberrations in ALL, it requires integrating multi-omics data to complete complex and time-consuming analyses in large retrospective cohorts. It is challenging to perform individualized clinical genomic analysis in real-world. We present a nationwide precision genomic study as part of the Chinese Children Cancer Group ALL 2020 clinical trial. Between 2020 and 2023, 6486 pediatric ALL patients were enrolled from 25 medical centers across 15 provinces in China. RNA-seq was performed for 5103 patients during diagnosis. We developed the National Children's Medical Center ALL Bio-Cloud (NCMC-ABC), an automated, cloud-based framework for real-time RNA-seq data process. NCMC-ABC is designed to analyze multiple clinically relevant genomic aberrations from single RNA-seq data, including molecular subtypes, coding and noncoding driver mutations, fusions and CNVs. The median turnaround time from sample collection to clinical reporting was 14 days across all hospitals, aligning with clinical treatment timelines. We established a molecular subtype classification framework for pediatric ALL, and successfully classified 94.94% of B-ALLs into 20 subtypes and 86.38% of T-ALLs into 11 subtypes. This framework significantly improved the traditional MICM approach, which classified only 48.58% of B-ALLs and did not account for T-ALL subtypes. The enhanced classification is due to the improved detection of key fusions (DUX4, PAX5, ZNF384, MEF2D rearrangements) and mutations (PAX5 P80R and IKZF1 N159Y). Meanwhile, we achieved more precise subtyping of HYPO, HYPER and KMT2A BALLs. The refined subtypes unveiled a distinct profile of Chinese B-ALL patients, with higher frequencies of HYPER, ETV6, DUX4 and PH subtypes, and lower frequencies of Ph-like, iAMP21 and HYPO, compared to Western cohorts. Importantly, the refined framework directly improved the risk stratification of patients. We identified a median of 2.44 pathogenic SNPs/indels and 1.28 fusions per patient. The driver mutations were detected in 259 genes in B-ALL and 156 in T-ALL. We observed different driver mutation profiles in our cohort compared to the Western cohort. Mutations in RAS pathway (NRAS, KRAS and PTPN11) were more frequent in Chinese patients, whereas the JAK-STAT (JAK2, IL7R, SH2B3 and CRLF2) pathway was more frequently mutated in Western cohort. We observed direct clinical relevance of these aberrations. For example, patients with TP53 and NR3C1 mutations showed inferior treatment response. The implementation of NCMC-ABC in a nationwide multicenter pediatric ALL clinical trial demonstrated its effectiveness and feasibility in real-world, improving risk stratification and therapeutic decision making in clinic. Citation Format: Han Wang, Jiaoyang Cai, Jie Yu, Shaoyan Hu, Yongjun Fang, Ju Gao, Jian Li, Hua Jiang, Xiuli Ju, Sixi Liu, Wenyong Kuang, Runming Jin, Liangchun Yang, Xuedong Wu, Xiaowen Zhai, Qun Hu, Hui Jiang, Ningling Wang, Chi Kong Li, Lirong Sun, Jiao Jin, Chun Li, Changda Liang, Yan Dai, Kaili Pan, Hao Xiong, Ching-Hon Pui, Shuhong Shen, Yu Liu. Cloud-based computational framework for individualized genomic analysis in pediatric acute lymphoblastic leukemia: A nationwide multi-center real-world clinical study [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 5283.