Abstract
Introduction. Adult T-cell leukemia/lymphoma (ATLL) is an aggressive, rare malignancy with poor survival despite anthracycline-based, salvage, antiviral-immunomodulatory, and novel targeted therapies. Predictive biomarkers for regimen selection are lacking. We present the first systematic integration of NGS mutation data with regimen-specific clinical outcomes in ATLL to identify actionable sensitivity and resistance markers.
Methods. We identified 47 ATLL patients diagnosed and treated at Montefiore Medical Center from 2015 to 2025 with available next-generation sequencing (NGS) data. All-line therapies were backbone-classified as:
C-AC: anthracycline combinations
C-SALV: cytarabine/platinum/ifosfamide salvage
AV/IM: antiviral-immunomodulatory
TT/IT: targeted/immuno-trial agents
stratum denoting the entire cohort across backbones
Responses were defined as Complete/Partial Remission (CR/PR), Stable Disease (SD), Progression of Disease (PD), and Death, collected from manual chart review. Endpoints were Progression-Free Survival (PFS) (days; 6-/12-month), Objective Response Rate (ORR = CR+PR), Disease Control Rate (DCR = CR+PR+SD), and all-cause death during available follow-up (non-time-to-event). Effect sizes were absolute differences (mutant — wild-type; days for PFS, percentage points for binary outcomes). Confidence Intervals (CI): Newcombe (risk-difference), Wilson (group). PFS was measured from regimen start to progression/death. Analyses were exploratory; given the number of gene-endpoint tests and small mutant counts, we did not apply formal multiplicity adjustment and interpret results cautiously, flagging sparse mutant cells (n_mut ≤ 3) as exploratory.
Results. We identified 36 significant single-gene associations across subgroups (ALL = 24, C-AC = 8, AV/IM = 4) and 16 significant multi-gene combinations.
Benefit markers (ALL): DIS3 was associated with PFS +1,439.7 days (2124 vs 684.3; p=0.010; variance not estimable from available summaries) and increased 6-/12-month PFS by +76.6 pp (95% CI +6.6 to +86.4; p=0.0156) and +85.1 pp (95% CI +16.2 to +92.6; p=0.0046). PDGFRB was associated with PFS +652.6 days (1347.8 vs 695.2; p=0.0455) and 12-month PFS +44.8 pp (95% CI −3.0 to +80.1; p=0.0364). DNMT3A was associated with 12-month PFS +59.9 pp (95% CI +3.0 to +87.6; p=0.0164). ARID1B was associated with an improved ORR +56.7 pp (95% CI +6.4 to +87.6; p=0.0013), DCR +52.4 pp (95% CI +5.2 to +85.3; p=0.0073), 6-month PFS +57.2 pp (95% CI +2.3 to +82.5; p=0.015), 12-month PFS +44.8 pp (95% CI −3.0 to +80.1; p = 0.0364), and PFS +145.7 days (p = 0.011), however in C-AC, correlated with reduced durability (PFS −243.7 days; 1140 vs 1383.7; p=0.0351). In C-AC, KLF2 suggested DCR +56.4 pp (95% CI −2.8 to +89.8; p=0.0488; n_mut=3). BRCA1 was associated with large effects in AV/IM (2949.5 vs 208.3 days; Δ +2741.2) and ALL (2949.5 vs 682.1; Δ+2267.4); 6-/12-month PFS +93.3/+84.2 pp; all p ≤ 0.0336; n_mut=2 (exploratory).
Resistance markers. IRF4 was associated with shorter PFS in ALL (−750.8 days; 32.1 vs 783.0; p = 0.0014; n_mut=7) and C-AC (−1433.0 days; 35.7 vs 1468.7; p = 0.0199). In ALL, CARD11 (−756.7 days; 57.9 vs 814.6; p = 0.0172; n_mut = 11), CIC (−505.7 days; 270 vs 775.7; p = 0.0273; n_mut = 9), and ATG2B was associated with more adverse outcomes (−724.8 days; 41.4 vs 766.2; p = 0.0311) . Additional ALL signals - BTK, GNB1, PTPRC, TNFAIP3 - each were associated with shortened PFS by −730.7 days (20.7 vs 751.4; p=0.0148; n_mut=3). ATM was linked to higher Death in C-AC (+92.5 pp; 95% CI +14.4 to +97.4; p=0.0116) and ALL (+86.3 pp; 95% CI +8.5 to +93.2; p = 0.0226), mut = 2 (exploratory).
Combinations. Of 16 multi-gene hits, 13/16 were explained by single-gene effects (often IRF4, CARD11, CIC, ATG2B; effect-size ratios 0.05–0.17). PDGFRB+DNMT3A (ALL) demonstrated a true incremental signal (PFS + 2822 days; 2912 vs 90; p = 0.0298), with 12-month PFS +85.6 pp (95% CI +4.9 to +93.7; p = 0.0292; n_mut = 2). PLCG1+NUP98 (ALL) was a novel adverse pair (PFS −85 days; 5 vs 90; p = 0.015).
Conclusions. Regimen-stratified genomics in ATLL nominates DIS3, PDGFRB, DNMT3A, ARID1B as genes associated with benefit and IRF4, CARD11, CIC, ATG2B, and ATM as mutations associated with resistance; ARID1B is context-dependent under C-AC; PDGFRB+DNMT3A is associated with incremental benefit. These findings support that exploration of NGS-guided treatment strategies may enable more personalized, regimen-optimized care in ATLL.
