INTRODUCTION: The potent graft versus leukemia (GVL) effect of allogeneic stem cell transplantation (allo-SCT) is considered as a blueprint for cellular immunotherapy. However, failure of GVL leads to relapse of underlying leukemia, the major cause of death after allo-SCT. In solid tumors, higher tumor mutation burdens are associated with better response to check point inhibitors which implies the importance of neoantigen specific T-cell functions in cancer immunity. In contrast, the frequencies of somatic mutations in acute leukemia are generally low, therefore the role of neoantigens in GVL remains undetermined. Here, we developed a platform to screen for potential neoantigens by performing whole exome sequencing (WES) and RNA sequencing (RNAseq) in matched samples: leukemic blasts at relapse after allo-SCT, recipient T cell controls, and donor cells.

METHODS: Leukemic blasts from patients in relapse were enriched by flow sorting from bone marrow aspirate or peripheral blood samples. Recipient T cells were isolated from pre-transplant peripheral blood as germline controls, and donor monocytes or CD34-positive cells were used as hematopoietic-lineage cell controls. WES was performed to 100X coverage, paired with RNAseq 40M reads per sample. Somatic mutations were detected with mutect and mpileup, followed by annotation with SnpEff. High confidence somatic mutations were subjected to pVAC-seq for neoantigen predictions.

RESULTS: Six patients with relapsed acute leukemia (AML 5, ALL 1) after allo-SCT and their transplant donors (matched sibling 3, haplo-identical 3) had suitable samples available for analysis. On average, somatic mutations were identified in 297 genes (range 108- 609) by comparing leukemic blasts and germline control T cells. Among those mutations, potential candidates of neoantigen were identified in five out of six subjects. Allele frequencies of mutant genes varied. Most of neoantigens were predicted to bind HLA of both class I (median 5, range 0-15) and class II (median 6, range 0-12). One subject had only HLA class II restricted peptides as predicted neoantigens. Of interest majority of antigens were derived from molecules known to play important roles in leukemia or tumor biology which include ETV6, CCNY, IDH2, PTPN11, SF3B1, and TP53. Evolution analysis of neoantigen showed an emergence of new antigens in relapsed leukemia while a few driver gene mutations persisted after allo-SCT (Figure).

CONCLUSION: Our in-silico analysis demonstrated the possibility that somatic mutation in acute leukemia could serve as putative neoantigens applicable for novel immunotherapy after allo-SCT. The binding capacity of mutant peptides to class I and II HLA implies the importance of both CD4 and CD8 contributions to anti-neoantigen immunity. Next, we will search for neoantigen specific T cells exerting an anti-leukemia effect to validate the GVL potential of these mutations in allo-SCT.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.