Decoding Clonal Evolution in Relapsed T Cell Acute Lymphoblastic Leukemia at Single Cell Resolution

T-lineage acute lymphoblastic leukemia (T-ALL) comprises approximately 10-15% of pediatric ALL cases with distinct feature in biology and largely inferior outcome compared to B-ALL. Growing evidence has reflected pivotal roles of clonal evolution in T-ALL recurrence, but bulk sequencing may not serve as the perfect model to reliably infer clonal heterogeneities and their immunomodulatory milieu during leukemia development. In this study, single-cell sequencing was applied to uncover leukemic clonal relationships with relapse throughout chemotherapy in T-ALL at a more accurate resolution.

We performed bulk whole-exome sequencing for sorted CD7 ⁺ BMMCs from 5 pairs of diagnosis-relapse (Dx_Rel) samples, revealing a series of well-reported hotspot mutations in T-ALL. Among those, we observed diagnosis-specific variations and relapse-emerged variations, suggesting the putative correlations with chemo-resistance. Transcriptomic sequencing highlighted additional stemness and metabolic abnormalities underlying leukemic re-occurrence. Incorporated Dx_Rel paired ATAC-seq depicted relapse-specific activated chromatin regions, such as ELK1, ELK4, RUNX1.

To dissect clonal diversities within and across the 5 Dx_Rel T-ALL pairs, we carried out high-throughput droplet-based 5'-single-cell RNA-seq (scRNA-seq) and paired T cell receptor sequencing (scTCR-seq). By performing unsupervised clustering of scRNA-seq profiles encompassing 10 samples, we identified 23 distinct T-lineage clusters (Cluster 0-22) based on the two-dimensional UMAP visualization. In 2 out of 5 patients (T593 and T788), diffusion map of T-lineage sub-clusters between diagnostic and relapsed samples appeared to be almost identical, while distinct shifts from diagnosis to relapse in the compositions have been observed in the other 3 out of 5 patients (T956, T723 and T856). Besides, it was noteworthy that two T-cell sub-clusters were concluded as "normal" T cells (Cluster 9 and 12) uniformly presented in both diagnostic and relapsed diffusion of T-cell sub-clusters across 5 Dx_Rel, from which TCR repertoires and expression profiles could well discriminate leukemic cells.

Next, we sought to further deconvolute the clonal evolution patterns for T-ALL Dx_Rel pairs. We observed that except in T788 lacking of clonal TCRs, dominant diagnostic clones of the other 4 patients diminished (T593) or vanished (T956, T723, T856) at relapse, sparing newly emerged subclones predominantly substituted at relapse. We clearly depicted two distinct patterns of evolutionary trajectories in these 4 Dx_Rel pairs by comprehensively mapping hierarchical TCR clonotypes onto leukemic clonotypes at single cell levels. Specifically, in T956 and T723, we observed significant outgrowth of incidental diagnostic sub-clones at relapse, whereby surrogate TCR repertoires correspondingly enumerated, suggestive of dynamic shifts in dominant clone over continuous chemo-exposure. Whereas in T593 and T856, expanding clones at relapse were showed up with completely different gene signatures from the diagnostic ones, but dominant clones at diagnosis and relapse were surprisingly presented with identical TCR repertoires. This was undoubtedly informative of leukemic "clonal drift" within which hypothetical intrinsic transformation happened to the same subclones over persistent chemotherapy.

Besides, we took advantage of our well-discriminated model to fully delineated the involvement of "normal" T subclusters in leukemic latency and chemo-responsiveness. By analyzing TCR repertoires in combined with expression profiles, we noted that "normal" T cells infiltrated by T-ALL were majorly distributed in CD8-effector sub-clusters compared to those from healthy donor, suggesting a robust leukemic stimulation on effector CD8 signaling in T-ALL microenvironment.

Collectively, our presented study accurately distinguished leukemic cells from normal T cells in T-ALL at a single-cell resolution. By tracking transcriptomic profiles within and across Dx_Rel T-ALL pairs, we further identified distinct clonal evolutionary patterns, which may determine diversified fates of leukemic clones in response to therapeutic pressures. In the meantime, we provided a comprehensive phenotypic view on "normal" T cells under leukemic prevalence and re-occurrence, extending significant implications for future precise immunotherapies.