Leukemia relapse via genetic immune escape after allogeneic hematopoietic cell transplantation

Graft-versus-leukemia (GvL) reactions are responsible for the effectiveness of allogeneic hematopoietic cell transplantation as a treatment modality for myeloid neoplasia, whereby donor T- effector cells recognize leukemia neoantigens. However, a substantial fraction of patients experiences relapses because of the failure of the immunological responses to control leukemic outgrowth. Here, through a broad immunogenetic study, we demonstrate that germline and somatic reduction of human leucocyte antigen (HLA) heterogeneity enhances the risk of leukemic recurrence. We show that preexistent germline-encoded low evolutionary divergence of class II HLA genotypes constitutes an independent factor associated with disease relapse and that acquisition of clonal somatic defects in HLA alleles may lead to escape from GvL control. Both class I and II HLA genes are targeted by somatic mutations as clonal selection factors potentially impairing cellular immune responses and response to immunomodulatory strategies. These findings define key molecular modes of post-transplant leukemia escape contributing to relapse.

All the supplementary tables are provided as separate Excel files. Myeloablative conditioning regimens (MAC) included busulfan (3.2 mg/kg/day for 4 days) combined with cyclophosphamide (60 mg/kg/day for 2 days) 2 or with fludarabine (30 mg/m2/day for 5 days), 3 or total body irradiation (TBI) of 1200 cGy combined with cyclophosphamide (60 mg/kg/day for 2 days), or busulfan (3.2 mg/kg/day for 2-3 days) combined with thiotepa (5 mg/kg/day for 2 days) and fludarabine (40 mg/m2/day for 4 days). 4 Reduced-intensity regimens (RIC) included fludarabine-based protocols, according to the disease and age of the recipient. Standard protocols of immunosuppression including cyclosporine (CSA) or tacrolimus (FK) and short-term methotrexate (MTX) or CSA/FK with mycophenolic mofetil (MMF) were used for GvHD prophylaxis. In addition, recipients of unrelated donor transplants received rabbit anti-thymocyte globulin (Thymoglobuline 2.5 mg/kg/day for 2-4 days before transplantation) or anti-lymphocyte globulin (Grafalon 10 mg/kg/day for 1-3 days) according to disease and donor HLA matching.
Severity of acute GvHD was graded according to Glucksberg's criteria 5 whereas the assessment of all patients developing chronic GVHD was made according to the National Institute of Health (NIH) consensus criteria. 6,7 DNA isolation Genomic DNA was isolated directly from cryopreserved unfractionated peripheral or bone marrow blood mononuclear cells with the Nuclei Lysis Solution (Promega) according to manufacturer's instructions.

HLA mutational analysis
Details of the bioinformatic approach to investigate somatic HLA mutational status have been described elsewhere. 8 In brief, after obtaining a confident full 4 th field typing through NovoHLA typing algorithm (Novocraft Technologies), paired-end reads from either targeting sequencing were directly aligned on a per-patient HLA reference using Novoalign (Novocraft Technologies Sdn Bhd). Allelic loss was imputed computing the number of reads covering each called heterozygous allele within a given locus. The following formula is used: With Ci and Cz representing the read coverage for each allele belonging to the same locus. For structurally similar alleles, we included an adjustment taking into account sequence variation defined as "Variant coverage", directly computed by the NovoHLA pipeline.
All mean Log2 ratios <-1.5 were retained as confident allelic losses, based on a previous internal validation study on 234 healthy controls and cell lines that did not showed altered regions. 12 It is noteworthy to clarify that since this algorithm is based on the coverage information for each reference base of each allele, differently from SNP-array platforms, does not allow a proper estimation of the genomic boundaries for 6p loss events but can only identify heterozygous allelic losses. A visualization of this process is reported in Supplementary Figure   11.

Whole exome sequencing and analysis
After quality control checks sequencing libraries were generated using Agilent SureSelect Human All Exon Kit (Agilent Technologies, Santa Clara, CA, USA) following manufacturer's instructions. After capture and enrichment with index tags, products were purified and quantified using AMPure XP system (Beckman Coulter, Beverly, MA, USA) and the Agilent high sensitivity DNA assay on the Agilent Bioanalyzer 2100 system. The prepped libraries were hybridized in the buffer with biotin-labeled probes, and magnetic beads with streptavidin were used to capture the exons of genes. Subsequently, non-hybridized fragments were washed out and probes were digested. The captured libraries were enriched by PCR amplification. 7 The library was checked with Qubit and real-time PCR for quantification and bioanalyzer for size distribution detection. Quantified libraries will be pooled and sequenced on Illumina platforms with PE150 strategy, according to effective library concentration and data amount required.
After discarding low quality reads (i.e. containing sequencing artifacts, including adapterrelated sequences or harboring >10% of uncertain bases) and checking for GC content, Burrows-Wheeler Aligner (BWA, v0.7.17) 13  A) Kaplan-Meyer estimates of overall survival according to HLA-C KIR ligand group. B) Cumulative incidence of relapse according to HLA-C KIR ligand group. C) Cumulative incidence grade II-IV acute GvHD according to HLA-C KIR ligand group. D) Cumulative incidence chronic extensive GvHD according to HLA-C KIR ligand group. Shaded bands represent 95% confident interval. E) Multivariable cox cause specific models of impact of class I on relapse adjusted for confounding factors, including recipient/donor KIR ligand status. F) Multivariable cox cause specific models of impact of HLA-C HED on relapse adjusted for confounding factors, including recipient/donor KIR ligand status.
Black squares indicate the odd ratio and error bars the 95% confident intervals. Non-adjusted p-values indicate the significance of the log-rank test.