AUTHORS
Kazuki Mitani, MD, Hirohito Kubota, MD,PhD2, Kiyotaka Isobe, MD3, Shibata Hirofumi4, Masafumi Seki5, Shunsuke Kimura6, Shuro Nishio7, Shohei Azumi8, Ryunosuke Tojo9, Hirokazu Kobushi10, Satoshi Saida11, Itaru Kato12, Yoshinori Uchihara13, Katsutsugu Umeda14, Fumihiko Hayakawa15, Takahiko Yasuda16, Emiko Sakaida17, Nobuyuki Kakiuchi18, Aiko Sato-Otsubo, MS19, Kentaro Watanabe, MD, PhD20, Katsuyoshi Koh21, Souichi Suenobu22, Atsushi Sato23, Motohiro Kato, MD, PhD24, Satoru Miyano25, Seishi Ogawa, MD, PhD26, Kenichi Yoshida27, Junko Takita, MD,PhD28
AFFILIATIONS
1 Graduate School of Medicine, Kyoto University, Kyoto, Japan
2 Graduate School of Medicine, Kyoto University, Kyoto, Japan
3 National Cancer Center Hospital, Tokyo, Japan
4 Graduate School of Medicine, Kyoto University, Kyoto, Japan
5 The University of Tokyo, Tokyo, Japan
6 The University of Tokyo, Tokyo, Japan
7 Graduate School of Medicine, Kyoto University, Kyoto, Japan
8 Graduate School of Medicine, Kyoto University, Kyoto, Japan
9 Graduate School of Medicine, Kyoto University, Kyoto, Japan
10 Graduate School of Medicine, Kyoto University, Kyoto, Japan
11 Graduate School of Medicine, Kyoto University, Kyoto, Japan
12 Graduate School of Medicine, Kyoto University, Kyoto, Japan
13 Graduate School of Medicine, Kyoto University, Kyoto, Japan
14 Graduate School of Medicine, Kyoto University, Kyoto, Japan, University of Fukui, Fukui, Japan
15 Nagoya University Graduate School of Medicine, Nagoya, Japan
16 NHO Nagoya Medical Center, Nagoya, Japan
17 Chiba University Hospital, Chiba, Japan
18 Kyoto University, Kyoto, Japan
19 The University of Tokyo, Tokyo, Japan
20 The University of Tokyo, Tokyo, Japan
21 Saitama Children’s Medical Center, Saitama, Japan
22 NHO Nishibeppu National Hospital, Beppu, Japan
23 Miyagi Children’s Hospital, Sendai, Japan
24 The University of Tokyo, Tokyo, Japan
25 Institute of Science Tokyo, Tokyo, Japan
26 Graduate School of Medicine, Kyoto University, Kyoto, Japan, Kindai University, Higashiosaka, Japan, Kyoto University, Kyoto, Japan
27 National Cancer Center Research Institute, Tokyo, Japan
28 Graduate School of Medicine, Kyoto University, Kyoto, Japan
ABSTRACT
Risk-adapted therapies have substantially improved survival in pediatric T-cell acute lymphoblastic leukemia (T-ALL). However, relapsed and refractory cases remain a major clinical challenge. Comprehensive molecular characterization of high-risk T-ALL is therefore required to identify biologically distinct subgroups, refine risk stratification, and facilitate the development of biomarker-driven therapeutic strategies. Here, through integrated multi-omics profiling of a nationwide prospective cohort, we identified and characterized SPI1-like, a previously unrecognized high-risk subgroup, and developed a transcriptome-based score for its detection.
We performed integrated multi-omics analyses of diagnostic samples from 115 children with T-ALL enrolled in the Japanese nationwide prospective JCCG ALL-T11 trial. Gene expression, genomic alterations, DNA methylation, and microRNA expression profiles were integrated with clinical data to identify biologically and clinically relevant subgroups. To refine the molecular classification, we additionally integrated multi-omics data from an independent JACLS/TCCSG cohort (n=75) and two in-house cases of T-cell lymphoblastic lymphoma harboring t(14;16)(q32;q24).
Similarity network fusion (SNF) analysis of gene expression and DNA methylation identified a SPI1-fusion-negative subgroup, designated SPI1-like, that consistently clustered with SPI1-fusion-positive leukemia (hereafter referred to as the SPI1 subtype). SPI1-like encompassed subsets of ETP-like, BCL11B-activated, TME-enriched, and t(14;16)(q32;q24) (FOXF1/FENDRR) subtypes.
Within the ALL-T11 cohort, this SNF-defined cluster was associated with significantly inferior clinical overall survival (HR, 3.10; 95% CI, 1.01-9.49), an effect driven by SPI1-like cases, whereas no relapses or deaths occurred among patients with the SPI1 subtype. Despite similar end-of-induction remission rates and end-of-consolidation measurable residual disease (MRD) distributions, SPI1-like cases showed inferior survival, suggesting that molecular classification may provide prognostic information complementary to conventional MRD-based risk assessment.
SPI1-like retained the core transcriptional program of the SPI1 subtype, including strong enrichment of SPI1 target gene signatures, while exhibiting greater enrichment of immune and inflammatory Hallmark gene sets, including IFN-γ response, TNF-α signaling via NF-κβ, IL6-JAK-STAT3 signaling, and inflammatory response.
This inflammatory phenotype was reinforced at additional regulatory layers: DNA methylation analysis demonstrated global hypomethylation, including loci associated with inflammatory pathway genes, and microRNA profiling revealed reduced expression of multiple microRNAs predicted to regulate NF-κβ and STAT signaling. Together, these findings indicate coordinated dysregulation of inflammatory signaling at the transcriptional, epigenetic, and post-transcriptional levels.
In addition, deconvolution analysis revealed enrichment of innate lymphoid, myeloid, hematopoietic stem/progenitor, and B-lineage transcriptional programs together with reduced T-lineage signatures, indicating pronounced lineage ambiguity.
To enable the identification of SPI1-like in transcriptomic datasets, we next developed a transcriptome-based SPI1-like score derived from differentially expressed genes in the SPI1-like/SPI1 subtype SNF cluster. Using a predefined scaled cutoff established in the discovery cohort (integrated ALL-T11 and JACLS/TCCSG cohorts), we applied this score to an independent public pediatric T-ALL cohort reported by Pölönen et al. (Nature, 2024). The SPI1-like score-positive subgroup comprised all SPI1 subtype cases, all t(14;16)(q32;q24) cases, and subsets of the ETP-like and TME-enriched subtypes. Score positivity was associated with significantly inferior event-free survival (HR, 2.69; 95% CI, 1.51-4.82) and overall survival (HR, 3.78; 95% CI, 1.99-7.20), supporting the reproducibility and prognostic relevance of the SPI1-like subgroup across independent cohorts.
These findings identify SPI1-like as a biologically distinct, high-risk T-ALL subgroup characterized by inflammatory signaling, epigenetic and post-transcriptional dysregulation, and lineage ambiguity. Integrative multi-omics analysis enabled the identification of this subgroup beyond current molecular classification and MRD assessment, and the SPI1-like score may facilitate transcriptome-based risk stratification.
DISCLOSURES
Kazuki Mitani, MD: No relevant conflicts of interest to declare.