Divergent CD4+ T-Cell Fates Govern Relapse and Remission in B-Cell Acute Lymphoblastic Leukemia

Introduction: Eradicating measurable residual disease (MRD) in patients with B-ALL requires prolonged chemotherapy, allogeneic hematopoietic stem cell transplantation (HSCT) or newer approaches such as CART cells. However, relapse is common even after HSCT or CART therapy, highlighting the immunoevasive properties of residual leukemia. Neoantigen-specific CD8+ T-cells are harbored in high frequencies by patients with B-ALL, but show minimal signs of dysfunction. In contrast, the frequency of PD1+/TIM3+ CD4+ T-cells at diagnosis is a major predictor of relapse. We previously observed that PD1+/TIM3+ CD4+ T-cells expanded in a mouse model of BCR-ABL+ B-ALL, mimicking clinical observations. Adding anti-PDL1 immune checkpoint blockade (ICB) to the 2^nd-line tyrosine kinase inhibitor nilotinib improved cure rates from 10% to 70% in leukemia-bearing mice. To mechanistically understand how CD4+ T-cell states contribute to relapse or remission in B-ALL, we recently studied neoantigen-specific CD4+ T-cells from a novel mouse model, along with relevant clinical specimens.

Methods: We created a novel mouse strain, HV1, that expresses a fixed T-cell receptor with specificity for a leukemia neoantigenic peptide derived from the BCR-ABL oncogene. Naïve HV1 CD4+ T-cells (10,000/mouse) were adoptively transferred into congenically distinct naïve recipients, followed 24h later by challenge with a BCR-ABL+ cell line (“LM138”; 2,500 cells/mouse). scATACseq, scRNAseq, multiplex immunofluorescence (Cellscape; Canopy-Biosciences) and functional studies were used to analyze HV1 cells during leukemia progression or treatment. To model MRD+ remissions, mice were treated with a 5-day course of nilotinib (75 mg/kg po). Nilotinib was then combined with anti-PDL1 blockade (10 mg/kg i.p.) to model curative therapy. Observations from adoptive transfer experiments were compared to bone marrow aspirates of patients with newly diagnosed with BCR-ABL+ or BCR-ABL- B-ALL. Survival analysis using log-rank testing was applied to outcomes from the TARGET database.

Results: UMAp clustering of HV1 cells at day 9 of leukemia challenge confirmed a canonical bifurcation into early T-follicular helper (Tfh) and Th1 subsets. By a late leukemia timepoint (day 17), both Tfh- and Th1-polarized HV1s adopted the epigenetic, transcriptomic, and phenotypic states (FOXP3-/IL10+) of Type-1 regulatory cells (Tr1s). Tr1-differentiated HV1s suppressed bystander CD8+ T-cells in co-culture assays. Leukemia cells co-localized with CD4+ T-cells but not CD8+ T-cells, and were found in closer proximity, on average, to CD4+ rather than CD8+ T-cells (mean distance 117 v. 143 um, p <0.001, student T-test). Ki67- LM138s preferentially maintained MHCII expression. HV1 subset proportions were significantly altered after treatment with nilotinib + anti-PDL1 therapy, resulting in a lower frequency of Tr1 cells and a higher frequency of CX3CR1+/AHNAKhi eff/mem cells, as compared to nilotinib only. CD8+ T-cell subset distributions were unchanged. We confirmed that similar CD4+ T-cell subsets are observable in scRNAseq datasets from patient samples. We interrogated the TARGET B-ALL database, using a Tr1 gene signature; patients with samples in the highest quartile of Tr1 signature expression experienced inferior overall survival as compared to those with the lowest (p =0.02, log-rank test).

Conclusions: Here, we show that neoantigen-specific CD4+ T-cells spontaneously develop into PD1+/TIM3+ Tr1s in the leukemia microenvironment. The relative proximity of CD4+ T-cells to LM138 cells and selective preservation of MHCII positivity in Ki67- LM138 cells implies that neoantigen-specific CD4+ T-cells may adopt Tr1 states after direct instruction by quiescent leukemia cells. This agrees with recent findings that neoantigen-specific CD4+ T-cells develop into Tr1s after direct contact with mutation-harboring hematopoetic stem cells (HSCs). Thus, Tr1s protect residual leukemia cells from immune attack by maladaptively recapitulating an immune circuit of the HSC niche. Eradication of residual B-ALL occurred concurrent with therapeutic reprogramming of neoantigen-specific CD4+ T-cells from Tr1 towards unique CX3CR1+/AHNAKhi mem/eff differentiation states.