Reversal of T Cell Exhaustion in Pre-Treatment Marrow T Cells Is Associated with Effective Graft-Versus-Leukemia Responses to Donor Lymphocyte Infusion

Abstract Abstract 1903 Donor lymphocyte infusion (DLI) can provide curative treatment for relapsed hematologic malignancies following allogeneic hematopoietic stem cell transplant (HSCT). However, the precise mechanism by which DLI eliminates leukemia cells in vivo has not been established. We hypothesized that marrow-infiltrating immune populations play a critical role in DLI responses since marrow is the primary site of disease and a reservoir of high-avidity antigen-specific memory T cells that can recognize tumor antigens, therefore potentially mediating graft-versus-leukemia (GvL) responses. We performed immunohistochemical staining of immune cells in serial marrow biopsies collected before and after DLI from 29 patients with relapsed CML. Twenty-two patients achieved cytogenetic remission within twelve months (‘responders’) while 7 patients demonstrated persistent disease (‘non-responders’). While no significant changes in the numbers of circulating T cells were seen between patient groups following DLI, the median number of marrow-infiltrating CD8+ T cells increased 2-fold in responders but remained unchanged in non-responders (P=0.02), demonstrating that clinical response to DLI is associated with T cell responses at the site of disease that may not be apparent in the peripheral blood. To investigate whether immune cell infiltration of the marrow could predict DLI response, we compared pre-treatment samples from both patient groups. Responders exhibited significantly higher proportions of CD8+ T cells in pre-DLI marrow compared to non-responders (5% vs 2.5%; P=0.01). Because disease burden is a known risk factor for ineffectual DLI response, we evaluated the interaction between disease burden and pre-existing CD8+ T cell infiltrate through the clinical course of 8 patients with high (≥70%) pre-treatment marrow cellularity. Three of 8 had ≥5% CD8+ T cell marrow infiltrates, and all 3 subsequently achieved cytogenetic remission. In contrast, 5 of 8 patients had <5% CD8+ T cell infiltrates, and all failed to show a cytogenetic response. ROC analysis of the entire original patient cohort using disease burden and CD8+ T cell infiltrate criteria (pre-DLI cellularity ≤70% or CD8+ T cell ≥5%) revealed a sensitivity and specificity of 100% in predicting responsiveness to DLI. These findings highlight the use of CD8+ T cell marrow infiltration as a new predictive marker for DLI response and suggest that immune status can overcome the adverse influence of high disease burden in therapeutic response. To identify candidate mechanisms underlying T cell responses to DLI, we performed mRNA expression profiling of CD3+ T cells isolated from matched pre- and post-treatment marrows of 4 responders and 2 non-responders (U133+ Affymetrix cDNA microarrays). We first compared global gene expression patterns between pre-treatment marrow-infiltrating T cells of both groups. Notably, 28% of significantly upregulated genes in responders play critical roles in T cell exhaustion. This finding was strengthened by unbiased gene set enrichment analysis (GSEA) using 880 sets from the Molecular Signatures Database spiked with 17 additional specifically curated T cell exhaustion sets. Four of 15 positively enriched sets were T cell exhaustion-specific. We next compared differential gene expression in marrow-infiltrating T cells before and after therapy in responders compared to non-responders and found 21% of significantly down-regulated genes to be components within T cell exhaustion pathways along with repression of multiple, distinct T cell exhaustion gene sets. The robust reversibility of T cell exhaustion signatures after DLI therapy underscores this gene module as a likely therapeutic target of DLI. In conclusion, CD8+ T cell marrow infiltration may serve as a novel predictive marker of response to DLI, including patients with high disease burden. In addition, these data implicate T cell exhaustion in distinguishing responders from non-responders and, provocatively, propose reversal of T cell exhaustion as a potential marker of DLI responsiveness in patients with relapsed CML after HSCT. These studies illustrate the importance of local immune responses at the site of disease and suggest a potential tool to predict DLI response in other hematologic malignancies. The clinical debut of newer agents that reverse T cell exhaustion suggests their use in lieu of DLI to promote GvL responses after allogeneic HSCT. Disclosures: No relevant conflicts of interest to declare.