LAUSR

To the Editor: Adult T cell leukemia–lymphoma (ATL) is a peripheral T cell lymphoid malignancy caused by human T cell leukemia virus type 1 (HTLV-1). The clinical subtypes of ATL are closely associated with its prognosis, which is extremely poor in aggressive subtypes (acute, lymphoma, and unfavorable chronic) compared to indolent subtypes (favorable chronic and smoldering) [1]. Human leukocyte antigen (HLA) plays an important role in T cell-mediated elimination of cancer cells. Downregulation of HLA occurs in various cancers and has been linked to poor prognosis [2]. Structural defects in the HLA molecule are usually associated with loss of heterozygosity (LOH) and somatic mutations in HLA genes [3]. However, the detection of LOH and somatic mutations in cancer cells seems to be extremely difficult due to the highly polymorphic nature of HLA genes. Conventional polymerase chain reaction (PCR)-based HLA typing mainly focuses on the polymorphic exons encoding the antigen recognition domains. Therefore, the genetic variations in the non-coding regions or in the exons outside of the polymorphic exons have largely remained ignored. In addition, methods for precisely deciphering HLA alleles are limited due to chromosomal phase ambiguity. To overcome a line of difficulties, we successfully developed the super high-resolution singlemolecule sequence-based typing (SS-SBT) method [4], which combines long-range PCR amplification and next-generation sequencing (NGS). This method provides in phase highresolution typing, which includes nucleotide differences in both the coding and non-coding regions of HLA genes [5]. Here, we used the SS-SBT method to investigate the entire region of HLA genes in both ATL and non-ATL cells obtained from the same patients. We evaluated 25 patients diagnosed as ATL between 2012 and 2018. Their characteristics are summarized in Supplementary Tables 1 and 2. Of all the patients, five had chronic-type ATL; the remaining 20 had acute-type ATL. Because cell adhesion molecule 1 (CADM1) is known to be expressed ectopically in ATL cells [6], we separated peripheral blood mononuclear cells (PBMCs) obtained from patients into CADM1-positive ATL cells and CADM1-negative non-ATL cells. We analyzed eight classical HLA loci in ATL and non-ATL cells from the same patients. Further materials and methods are shown in the Supplemental Methods. Through magnetic cell sorting, more than 97% pure CADM1positive cells (ATL cells) and more than 90% CADM1-negative cells (non-ATL cells) were isolated in CADM1-positive and -negative fractions, respectively (Supplementary Table 2). After the sequencing of eight HLA loci using the genomic DNA from ATL and nonATL cells, basic sequence read information was obtained (Supplementary Table 3). The HLA typing results of all of the patients are shown in Supplementary Table 4. A total of 11 novel HLA alleles were identified in non-ATL cells. All of the novel variants were located in intronic regions or in the 3′UTR (Supplementary Table 5). To detect somatic mutations in HLA genes, HLA allele sequences were determined in ATL cells. Mutational events were observed in the ATL cells but not in the non-ATL cells. We found a total of 18 somatic mutations in ATL cells from 8 patients, including 13 single nucleotide variants (SNVs) and five insertions/ deletions (indels) (Table 1). Of the 13 SNVs, three were nonsense mutations, seven were nonsynonymous mutations, and two were mutations at splice sites. There was only one mutation located in intron 1 of HLA-A. Mutations in the splice sites in intron 2 caused frameshift-generating premature stop codons. All five indels also caused frameshift-generating premature stop codons. The localization of non-silent variants (NSVs) in the HLA genes in the ATL cells are shown in Supplementary Figure 1. All of the NSVs were found in HLA class I genes but not in HLA class II genes. NSVs occurred more often in HLA-A and HLA-B than in HLA-C and localized with the highest frequency to exon 4 (8 mutations out of 17, 47%) of the HLA class I genes. HLA-LOH was detected by the existence of an allelic imbalance in each HLA loci, as determined by calculating the normalized average depth ratio as the relative depth ratio of ATL cells to nonATL cells (Supplementary Methods, Supplementary Tables 6 and 7). We found HLA-LOH in ATL cells obtained from 8 patients