Highlights
– A viral clonality evenness (VCE) score, based on the Shannon Evenness Index applied to HTLV‑1 integration-site sequencing, discriminated carriers who progressed to adult T‑cell leukaemia (ATL) from those who did not in a retrospective Japanese cohort.
– In 56 asymptomatic carriers followed longitudinally, low baseline VCE (<0.694) correlated strongly with later ATL (14/17 progressors), yielding a classification with no false positives in this sample and markedly better individual-level performance than proviral load thresholds.
– VCE captures clonal dominance (early malignant expansion) complementary to proviral load and could enable targeted surveillance or pre-emptive interventions after validation.
Background: disease burden and unmet need
Human T‑lymphotropic virus type 1 (HTLV‑1) is a retrovirus that establishes life‑long infection and is endemic in clusters worldwide, including parts of Japan. A minority of infected individuals—classically estimated at 2–7%—develop adult T‑cell leukaemia/lymphoma (ATL), an aggressive T‑cell malignancy that typically appears decades after initial infection and carries poor prognosis once advanced. Current clinical management of asymptomatic HTLV‑1 carriers is predominantly observational because accurate predictors of who will progress to ATL are lacking. Proviral load (the proportion of peripheral blood mononuclear cells carrying integrated HTLV‑1 DNA) has been associated with increased risk, but it has limited precision at the individual level and gives rise to false positives when used alone for targeting interventions.
Study design
Karpe and colleagues conducted a retrospective longitudinal cohort analysis nested within the nationwide Joint Study on Predisposing Factors of ATL Development (JSPFAD) in Japan. Eligible participants were asymptomatic HTLV‑1 carriers at enrolment with available archived peripheral blood mononuclear cell (PBMC) DNA at baseline and at least one follow‑up sample. Participants were selected to represent three groups defined at enrolment and during follow‑up: (1) carriers who subsequently progressed to ATL (n=17), (2) carriers with high proviral load (≥4%) who did not progress to ATL (n=18), and (3) carriers with low proviral load (<4%) who did not progress (n=21).
DNA samples were analysed by HTLV‑1 clonality sequencing to map virus integration sites and quantify the relative abundance of infected clones. From these data the authors calculated a viral clonality evenness (VCE) score using a Shannon Evenness Index transformation that ranges from 0 (perfect monoclonality: one clone dominates) to 1 (fully polyclonal: many clones of similar low abundance). The analytical goals were to evaluate VCE and proviral load as predictors of progression to ATL and to compare classification performance using AUC, accuracy, and Matthews correlation coefficient (MCC). VCE values were compared across outcome groups with nonparametric testing.
Key findings
Fifty‑six participants were included (follow‑up means: progressors 8.3 years, high‑PVL non‑progressors 9.7 years, low‑PVL non‑progressors 7.5 years). Clonality sequencing in the 39 non‑progressors revealed hundreds to thousands of distinct HTLV‑1 integration sites at both timepoints, indicating many low‑abundance clones and high VCE scores (≥0.694) at baseline. By contrast, most later progressors (14 of 17) demonstrated either a single predominant clone or two to four dominant clones at both baseline and follow‑up, with significantly lower baseline VCE scores than non‑progressors (p<0.0001).
Discriminatory performance as measured by area under the receiver operating characteristic curve (AUC) was similar for binary proviral load thresholds and VCE scoring (both reported as 91; proviral load 95% CI 80–98; VCE 95% CI 78–100). However, metrics that treat each individual equally revealed major differences in utility. Using individual‑level measures, VCE scoring outperformed proviral load thresholds: accuracy for proviral load 0.76 (95% CI 0.76–0.77) versus VCE 1.00 (95% CI 0.99–1.00); Matthews correlation coefficient for proviral load 0.23 (95% CI 0.19–0.24) versus VCE 0.91 (95% CI 0.80–1.00). In this study VCE scoring produced no false positives (i.e., no individuals were misclassified as high risk who did not later develop ATL), whereas proviral load produced ~20% false positives; VCE had a slightly higher false‑negative rate (0.3% vs 0.1%) in the dataset as reported.
Interpretation and biological plausibility
The findings are biologically plausible. ATL arises from malignant transformation and clonal expansion of HTLV‑1‑infected T cells. Proviral load quantifies the aggregate burden of infected cells but does not report on clonal architecture: a high proviral load can reflect many small clones or one expanding pre‑malignant clone. Sequencing‑based clonality analysis directly captures clonal dominance. A low VCE reflects unequal clonal distribution with a dominant clone—an early indicator of neoplastic evolution—whereas a high VCE reflects a polyclonal landscape less suggestive of imminent transformation.
Strengths
– Use of longitudinal samples and a well‑characterised national cohort with long follow‑up.
– Application of high‑throughput HTLV‑1 integration‑site mapping to quantify clonal architecture objectively and reproducibly using an index (VCE) with intuitive bounds (0–1).
– Direct head‑to‑head comparison with proviral load and use of per‑individual performance metrics relevant to clinical decision‑making.
Limitations and cautions
– The study is retrospective and involved selected participants sampled from a larger cohort based on proviral load and outcome, which may introduce selection bias and reduce generalisability.
– The sample size is modest (n=56), and although differences were large, validation in independent, larger cohorts (including non‑Japanese populations) is essential before clinical implementation.
– Laboratory pipelines for clonality sequencing are more complex and costly than quantitative PCR for proviral load; assay standardisation, reproducibility across laboratories, and defined minimal sequencing depth need establishing.
– The optimal timing and frequency for VCE testing, and the temporal dynamics of VCE in the months/years before ATL diagnosis, were not fully characterised and will determine clinical utility.
– The study did not evaluate how VCE‑based risk stratification should change management in practice—whether by intensified surveillance, enrolment in prevention trials, or early therapeutic intervention—and the safety and efficacy of such strategies remain unknown.
Clinical implications and proposed next steps
VCE scoring, if validated, has potential to transform care for HTLV‑1 carriers by enabling precise identification of individuals at high imminent risk of ATL. For clinicians and health systems, the immediate implications are:
- Consider replication of these findings in independent, prospective cohorts with standardised sampling intervals to confirm sensitivity, specificity, and timing relative to transformation.
- Develop and validate standard operating procedures for HTLV‑1 integration‑site sequencing and centralized quality control to ensure consistent VCE calculation across sites.
- Integrate VCE with other candidate biomarkers—somatic mutation profiling of dominant clones, immune markers, and clinical features—to build multivariable risk models and decision aids.
- Design early‑intervention studies or surveillance trials (for example, intensified clinical review, early lymphocyte phenotyping, and enrolment into prevention trials) for carriers with low VCE, rather than applying toxic or unproven therapies broadly.
Research priorities
Key research tasks include: external validation in geographically and genetically diverse populations, prospective evaluation to determine lead time from VCE change to clinical ATL, cost‑effectiveness modelling for sequencing‑based surveillance, and trials to test whether earlier interventions guided by VCE improve clinical outcomes. Mechanistic studies could profile the genomic landscape of dominant clones (driver mutations, integration sites) to better understand malignant evolution and identify therapeutic targets.
Conclusion
Karpe et al. provide compelling retrospective evidence that a viral clonality evenness score derived from HTLV‑1 integration‑site sequencing markedly improves individual‑level prediction of progression to ATL compared with proviral load alone in this Japanese cohort. If independently validated and operationalised with robust laboratory and clinical pathways, VCE has the potential to focus surveillance and preventive strategies on the small subset of carriers at greatest risk of life‑threatening disease, while sparing most carriers from unnecessary interventions.
Funding
The study was supported by Association Jules Bordet, FNRS-Télévie, FCC, WALInnov, FLF, JSPS-KAKENHI, and CoBiA.
References
1. Karpe SD, Artesi M, Wayet J, et al. A viral clonality evenness score to predict progression to adult T‑cell leukaemia in asymptomatic carriers of human T‑lymphotropic virus type 1 in Japan: a retrospective longitudinal cohort study. Lancet Microbe. 2025 Nov;6(11):101198. doi:10.1016/j.lanmic.2025.101198. PMID: 41213277.
2. Gessain A, Cassar O. Epidemiological aspects and world distribution of HTLV‑1 infection. Front Microbiol. 2012;3:388. (Note: comprehensive review on HTLV‑1 epidemiology and disease associations.)
AI-friendly thumbnail prompt
Clinical conceptual graphic: a laboratory bench with a blood collection tube and a next‑generation sequencing readout on a screen; overlay of two circular clonal pie charts—one showing a single dominant slice and labeled “low VCE / high risk,” the other showing many small equal slices labeled “high VCE / low risk.” Soft clinical blue and white palette, subtle map silhouette of Japan in the background, modern medical infographic style.