T Cell Receptor Clonal Dynamics and Hematopoietic Recovery after Immunotherapy in Severe Aplastic Anemia

Severe immune aplastic anemia (SAA) is a fatal human disease due to the destruction of marrow hematopoietic cells by cytotoxic lymphocytes. Aplastic anemia is both the paradigm of a family of marrow failure syndromes and closely related to other human autoimmune diseases, but our understanding of its pathophysiology at the cellular and molecular levels is incomplete.

We applied advanced single cell methodologies--mass cytometry by time of flight (CyTOF), single-cell RNA sequencing (scRNA-seq), and single-cell TCR/BCR V(D)J sequencing, to 93 bone marrow (BM) and sorted cell population samples from 20 patients with SAA enrolled in a prospective clinical trial who were treated with hATG, cyclosporine and eltrombopag, the standard immunosuppression (IST) regimen.

At the 6 month primary efficacy endpoint, 15 patients had achieved hematopoietic response, 4 patients had withdrawn from the protocol, and 1 was a non-responder. Both CyTOF and scRNA-seq results showed an increase in T lymphocytes and decrease in myeloid cells before treatment compared with healthy controls, and decreased lymphocytes and recovery of myeloid cells after treatment. In pre-treatment samples, early HSCs were drastically reduced but showed trilineage differentiation without lineage bias, suggesting that immune targeting of early stem cells resulted in BM failure. Post-treatment, late-stage HSCs mainly contributed to trilineage differentiation.

AA samples exhibited higher CD8+ effector memory potential and increased cell cytotoxicity compared to healthy. Effector memory CD8+ T cells were clonally expanded in SAA with TCR usage highly individual-private. T cell clone dynamics could be categorized into 3 groups: increased (> 2 fold, treatment-resistant), unchanged (0.5-2 fold, treatment-insensitive) and decreased (< 0.5 fold, treatment-sensitive). We tracked transcriptional changes of individual clones pre- and post-treatment: IFN-γ signaling in T cells was uniformly decreased post-treatment in all three groups, with more dramatically decreased cytotoxicity and increased apoptosis in groups with unchanged and decreased clones compared to treatment-resistant. Our results suggest that suppression of T cell function with IST regardless of clone size change post treatment.

Among 16 patients who were evaluable at 6 months, in 9 clone size increased and in 7 decreased. Clone size change was inversely correlated with diversity of TCR usage (by Gini index) and with robustness of blood count recovery. Novel clones, not present at baseline, predominantly contributed to clone size increase after treatment; preexisting clone size decreased in all but two cases. TCR sequences of preexisting clones and novel clones overlapped, more similar to each other compared to healthy controls. These results were concordant with clonal TCR usage in SAA patients. Preexisting and novel clones appear to target shared (unknown) antigens. Functionally, IFN-γ scores were decreased more dramatically in individuals with decreased clone size after treatment compared to those with increased clone size. Higher baseline IFN-γ cell activation, TNF-α, and exhaustion scores correlated with better blood counts at 6 months, and decreased scores post-treatment correlated with more robust hematologic recovery. Hematopoietic recovery appeared dependent on effective suppression of preexisting T cell clones in size and cytotoxicity, and conversely emergence of novel clones after treatment associated with less robust blood count improvement.

We also integrated the current scRNA-seq data with polygenic signals from GWAS, to evaluate for genome-wide heritable genetic factors in AA and disease-associated cell populations in bone marrow. A gene-level association analysis was performed using MAGMA. GWAS-based pathway enrichment analysis revealed that the top 100 genes associated with AA were enriched in immune response pathways, including IFN-γ and IFN-α, and CD8+ T and NK cells had the highest disease scores indicating most involved cell types in disease. Many of these top disease-associated genes exhibit cell-type specific expression in CD8+T, CD4+T, and NK cells, and positive cells with higher expression of immune activation genes, compared with negative cells.

Our approach offers a general guide to high resolution investigation of autoimmune diseases.