AUTHORS
Takeshi Sugio, MD, PhD, Shuyu Shi, PhD2, Atae Utsunomiya, MD3, Navika Shukla, MD4, Nick Phillips5, Andrea Garofalo6, Feng Tian, PhD7, Xiaoman Kang8, Eisaburo Sueoka9, Ki-Ryang Koh10, Junya Makiyama, MD, PhD11, Yasuhito Nanya12, Chih Long Liu, PhD13, Mari Olsen14, Yoshihisa Yamano15, Kaoru Uchimaru16, Koji Kato, MD, PhD17, Maximilian Diehn, MD, PhD18, Toshiki Watanabe19, Ash A. Alizadeh, MD, PhD20
AFFILIATIONS
1 Stanford University, Stanford, California, United States
2 Stanford University, Stanford, California, United States
3 Imamura General Hospital, Kagoshima, Japan
4 Stanford University, Stanford, California, United States
5 Stanford University, Stanford, California, United States
6 University of Washington, Seattle, United States
7 Stanford University, Stanford, California, United States
8 Stanford University, Stanford, California, United States
9 Saga University, Saga, Japan
10 Osaka General Hospital of West Japan Railway Company, Osaka, Japan
11 Sasebo City General Hospital, Sasebo, Japan
12 Institute of Medical Science, The University of Tokyo, Tokyo, Japan
13 Stanford University, Stanford, California, United States
14 Stanford University, Stanford, California, United States
15 St. Marianna University School of Medicine, Kanagawa, Japan
16 Showa Medical University, Tokyo, Japan, The University of Tokyo, Tokyo, Japan
17 Kyushu University, Fukuoka, Japan
18 Stanford University, Stanford, California, United States
19 St. Marianna University, Kanagawa, Japan, The University of Tokyo, Tokyo, Japan
20 Stanford University, Stanford, California, United States
ABSTRACT
Background: Some 2–5% of the estimated ~10 million human T-cell leukemia virus type 1 (HTLV-1) carriers worldwide develop adult T-cell leukemia/lymphoma (ATLL), yet the molecular events driving malignant progression remain poorly understood. PBMC-based longitudinal studies fail to capture tissue-resident disease, particularly lymphoma-type ATLL. In contrast, plasma cell-free DNA (cfDNA) captures tumor-derived signals from both circulating and tissue-resident disease. We applied a multimodal cfDNA platform to longitudinally reconstruct the genomic, epigenomic, immunogenomic, and viral evolution preceding ATLL development.
Methods: In this case-control study, we analyzed 336 longitudinal plasma samples from 63 HTLV-1 carrier cases enrolled in the nationwide Japanese Joint Study on Predisposing Factors of ATL Development (JSPFAD), including 211 samples from 38 carriers who developed ATLL (mean 3 pre-diagnostic samples/patient [range, 1–6]; median 0.98 years from last pre-diagnostic sample to diagnosis) and 125 samples from 25 carrier controls who remained ATLL-free for ≥10 years. Plasma cfDNA was profiled using genomic sequencing, T-cell receptor clonotyping, epigenetic profiling, and whole-HTLV-1 genome sequencing. Longitudinal phylogenetic analyses reconstructed host and viral clonal evolution.
Results: Among 38 evaluable ATLL progressors, 23 developed smoldering-, 6 acute-, 5 chronic-, and 4 lymphoma-type disease. Tumor-associated somatic mutations were identified in all patients who subsequently developed acute-, lymphoma-, or chronic-type ATLL, with median lead times of 4 years (range:1–10), 3 years (4–10), and 2 years (1–5), respectively. Among 18 evaluable smoldering-type cases, 14 (78%) harbored detectable tumor-associated mutations with a median lead time of 1 year before diagnosis. Notably, all smoldering-type cases that subsequently transformed to acute-type ATLL exhibited detectable pre-diagnostic mutations with significantly higher mutant allele fractions than non-transforming smoldering cases. HTLV-1 proviral loads were consistently higher in progressors than long-term asymptomatic carriers. Transforming smoldering-type cases exhibited significantly higher proviral loads than non-transforming controls (median 4.7x higher, P=0.0028). Recurrent 5’-long terminal repeat (LTR) deletions were identified in 6 progressors but absent in all long-term asymptomatic carrier controls.
Integrated host and viral phylogenetic analyses revealed convergent evolution. Across both indolent and aggressive ATLL, TCR/NFκB pathway alterations represented shared early evolutionary events coinciding with monoclonal expansion of tumor-specific TCR clonotypes. In contrast, DNA damage response alterations (TP53, ATM) were significantly enriched in aggressive ATLL (5/10 vs 2/28, P=0.008), with ~80% representing founder events. Immune-evasion-related alterations (B2M, FAS, CD58, CD274) were acquired later during clonal evolution, typically around the time of diagnosis. Furthermore, all patients progressing from indolent to acute-type ATLL evolved from common ancestral clones, demonstrating that aggressive transformation resulted from evolution of a pre-existing precursor. 5′ HTLV-1 LTR deletions represented late events during leukemogenesis. Identical monoclonal integration sites (IS) identified in diagnostic cfDNA were detected in pre-diagnostic plasma samples in 87.8% (29/33) of patients. Recurrent viral mutations associated with ATLL progression were identified.
Among patients with tumor-specific TCR clonotypes identified in diagnostic plasma cfDNA, identical clonotypes were retrospectively detected in pre-diagnostic plasma in 91.3% (21/23), with emergence closely paralleling founder mutation detection. Persistent expansion of specific TCR clonotypes across serial samples was not observed in long-term asymptomatic carrier controls.
In a held-out validation subset of pre-diagnostic samples, an integrated multimodal machine-learning model combining HTLV-1 proviral load, clonal IS, cfDNA epigenetic signatures, TCR clonotype profiles, and mutation AF, distinguished ATLL progressors from long-term asymptomatic carriers with an AUC of 0.94 (Sn 81%, Sp 97%, PPV 96.6%).
Conclusions: Longitudinal multimodal cfDNA profiling delineates the evolutionary trajectory preceding ATLL development and enables accurate pre-diagnostic risk stratification of HTLV-1 carriers, providing a foundation for longitudinal surveillance and early intervention.
DISCLOSURES
Takeshi Sugio, MD, PhD: No relevant conflicts of interest to declare.