Main

Enteropathy-associated T-cell lymphoma (EATL) is a rare type of T-cell lymphoma with extranodal origin, representing <1% of malignant lymphomas and <5% of digestive tract malignant lymphomas.1, 2 EATL is frequently associated with intestinal perforations, and diagnosis for EATL is many times delayed because of tumor development in the small intestines, which is difficult to explore by routine endoscopic examination and screening. Moreover, effective chemotherapy for EATL has not yet been developed3 and prognosis of EATL is very poor.1, 2, 3

EATL is subclassified into two types according to its clinicopathological characteristics. Type I EATL occurs in association with celiac disease, a gluten intolerance disease. Cases are commonly observed in Northern European countries, particularly in Scandinavia. On the other hand, Type II EATL, which occurs de novo, is found all over the world and with higher frequency in Eastern countries.1, 2, 3, 4 Celiac disease is quite a rare disease in Asia, and only a few cases of Type I EATL have been reported in Japan before the recent introduction of a gluten-containing diet.5, 6 Prevalence of the HLA-DQ2 and DQ8 haplotypes that are frequently found in celiac disease was reported to be low in the Eastern Hemisphere.7 In Asia, almost all EATL cases are classified as Type II,6, 7, 8, 9 and because of its low incidence, its pathological mechanism and immunohistological findings remain undefined and poorly understood. However, the association of Type II EATL with autoimmune hemolytic anemia is suggestive of a yet unidentified immunological background.10

Intraepithelial lymphocytes are thought to be the cell of origin for EATL, and may have a key role in protection from foreign antigens, detachment/elimination of virally infected villous epithelia, and turnover of epithelial cells.11 Intraepithelial lymphocytes consist of phenotypically diverse and complex T-cell populations, which differ from the conventional peripheral T-cells. Intraepithelial lymphocytes can express either αβ-T-cell receptor (TCR) or γδ-TCR, and are capable of expressing CD8, granzyme B, and CD103, which confer enterocyte destruction and villous atrophy capabilities.12, 13

Recently, numerous monoclonal and polyclonal antibodies for formalin-fixed paraffin-embedded tissue sections have been developed. Pan-T-cell antigens such as CD3 and CD5, as well as helper T-cell, cytotoxic T-cell, and αβ- and γδ-T-cell markers have been made available, aiding in the study of T-cells and their neoplasms; hence, a detailed immunohistochemical analysis of EATL may be performed. Chan et al.8 reported that Type II EATL frequently expressed γδ-T-cell markers. Additionally genomic alterations characteristic of EATL in Western countries have been previously reported. By array comparative genomic hybridization, recurrent minimal regions of chromosomal alteration were identified. EATL was initially characterized by 9q gains (minimal region 9q33-q34).14 A recent study of western EATL has reported frequent complex gains of 9q31.3-qter (70%) and loss of 16q12.1 (23%); and the subtype of EATL Type I was characterized by gains of 1q and 5q, while Type II EATL was associated with gains in 8q.15 In Japan, one study using comparative genomic hybridization analysis identified gains of 8q2 (47%), Xp (53%), and Xq (73%), but none of the cases had any gains of 1q3, 5q3, 7q2, or 9q3, which are typically found in Caucasian EATL patients.6 To the best of our knowledge, there have been no reports on the association between αβ- and γδ-T-cell phenotype markers and genomic alterations.

In the present study, we analyzed the genomic and immunohistochemical profiles of EATL from formalin-fixed paraffin-embedded tissue using an oligonucleotide array-based comparative genomic hybridization technique. We demonstrated that a 9q34 gain was frequently found in Type II EATL derived from both αβ-T-cells and γδ-T-cells.

Materials and methods

Case Selection, Lymphoma Samples, and Clinical Data

A total of 22 samples from Japanese patients with an initial diagnosis of EATL from 2000 to 2012 were retrieved from the Departments of Pathology of Tokai University, Okayama University, and the Kanagawa Cancer Center. Inclusion criteria were the presence of a lymphoma with EATL morphology and phenotype as established by the 4th edition 2008 WHO classification of lymphomas, which is characterized by a monomorphic diffuse proliferation of lymphoma cells in the lamina propria with infiltration and destruction of different intestinal compartments, and the presence of intraepithelial lymphocytes. Lymphoid cells also must be positive for cytoplasmic CD3 (cCD3; T-cell surface glycoprotein CD3 epsilon chain). Cases of extranodal NK/T-cell lymphoma nasal type were carefully excluded by noting the absence of intraepithelial lymphocytes in cases positive for Epstein-Barr virus Small RNA (EBER) genes by EBER RNA in situ hybridization. Two cases were reclassified, one as extranodal NK/T-cell lymphoma nasal type and another as peripheral T-cell lymphoma not otherwise specified, and therefore excluded. As a result, 20 EATL cases were employed in the present study (Table 1).

Table 1 Clinical features of the patients
Full size table

The formalin-fixed paraffin-embedded samples were obtained from surgical resection in 15 patients and tumor biopsy in five patients. Clinical findings were collected from written description. Of note, none of the EATL cases had a history of celiac disease. This study was approved by the institutional review board of the participating institutions where required (12R-084), was conducted in accordance with the Helsinki Declaration of 1975 as revised in 2008 and with accordance with the ethical standards on human experimentation. There is compliance with BRISQ, MIAME and REMARK guidelines.

Histological Evaluation

Lymphoma cells were categorized by size (small, medium, or large) and nuclear configuration (round or irregular), and investigated for the presence of intraepithelial lymphocytes by hematoxylin-eosin (HE) stain.

In Situ Hybridization of EBV-Encoded RNA (EBER In Situ Hybridization) and Immunohistochemistry

EBER in situ hybridization was performed using Bond ready-to-use in situ hybridization EBER fluorescein-conjugated oligonucleotide probe (PB0589, Leica Biosystems, Novocastra (NV), Newcastle upon Tyne, UK) and the purified IgG fraction of a mouse monoclonal anti-fluorescein antibody (NV, AR0833). For immunohistochemistry, mouse monoclonal antibodies against CD2 (NV, clone LFA), CD3ɛ (NV, clone LN10), CD4 (NV, clone 4B12), CD5 (NV, clone 4C7), CD7 (NV, clone LP15), CD8 (NV, clone 4B11), CD10 (NV, clone 56C6), CD20 (Nichirei, clone L26), CD56 (NV, clone CD564), TIA-1 (Beckman Coulter, clone 2G9A10F5), Granzyme B (NV, clone 11F1), TNFRSF (Abcam, ab47677, rabbit polyclonal) TCRβ (Human TCR beta chain constant region, Santa Cruz, G-11), TCRγ (Human TCR gamma chain constant region, Thermo Scientific, TCR C gamma M1 antibody clone γ3.20), and TCRγδ (Human TCR Pan TCR gamma delta, Thermo Scientific, TCR gamma+delta antibody clone 5A6.E9) were purchased as primary antibodies. BTLA was provided from Spanish National Cancer Research Center. Detection of EBER in situ hybridization and immunohistochemistry signals was performed using the Leica BOND-MAX fully automatic immunohistochemistry system with the BOND Polymer Refine detection kit (DS9800) according to the manufacturer’s instructions. BOND Epitope Retrieval Solution 2 (AR9640) was used for 20 min for CD2, CD3, CD4, CD5, CD7, CD8, CD20, CD56, TIA-1, Granzyme B, TNFRSF, BTLA and TCRβ. Epitope retrieval of TCRγ was performed with a pressure cooker (Asahi e-range pressure cooker RC-11, Japan) inside a microwave (Panasonic NE-EA212 high power 750 W) in EDTA (pH9.0) antigen retrieval solution for 4 min, and for TCRγδ by microwaving to boiling point in EDTA (pH9.0) for 20 min; these were then stained by Leica BOND-MAX system. The cutoff value for sample positivity was set at 30% of positive lymphoma cells.

DNA Extraction and Array Comparative Genomic Hybridization

Slides containing at least 70% lymphoma cells based on histology and immunohistochemistry were selected for crude micro-dissection and DNA extraction. DNA was extracted from formalin-fixed paraffin-embedded slides of surgically resected tissue using a silica-membrane-based DNA purification method (QIAamp DNA Micro Kit, Cat. no. 56304, QIAGEN K.K., Japan). All extracted DNA samples were assessed for quality by PCR amplification as previously described by BIOMED-2 guidelines,16 and eight cases with optimal genomic fragments larger than 200 bp were subjected to array comparative genomic hybridization. Array comparative genomic hybridization was performed hybridizing 1.5 μg of test DNA of a sex-matched reference on a SurePrint G3 Human array comparative genomic hybridization. Array comparative genomic hybridization microarray 1M (Agilent Technologies, USA), following the protocol ‘Oligonucleotide Array-Based for Genomic DNA Analysis’ (Agilent ULS labeling for blood, cells, tissues or formalin-fixed paraffin-embedded tissues). Array comparative genomic hybridization was outsourced to Quantitative Genomic Medicine Laboratories (qGenomics, Spain). As previously described by Salaverria et al.17 ‘Protocol Version 3.1′ was followed with slight modifications. Briefly, after digestion, the DNA was fragmented (99 °C for 40 min) and labeled. After hybridization, the slides were washed and fluorescence was assessed using a DNA microarray scanner (G2565CA, Agilent Technologies). Raw data were generated from scanned images using the Agilent Feature Extraction Software (v10.7). Log2 ratios of background-corrected tumor over normal DNA values were calculated. Normalization was carried out on Agilent's comparative genomic hybridization analytics, integrated on the Genomic Workbench suite (v5.0). Detection of copy number alterations was performed using the Aberration Detection Method-2 (ADM-2) algorithm, implemented within Agilent's genomics suite Genomic Workbench v5.0 and Nexus Copy Number software v6.0 Discovery Edition (BioDiscovery, USA). The results were concordant with both algorithms. All alterations were confirmed by visual inspection by two different observers. Regions of alterations (ROAs) were identified as previously described by Deleeuw et al.15 based on the frequency and exclusivity of genes or losses and the minimal common regions identified. The Waves array comparative genomic hybridization Correction Algorithm (WACA), based on GC content and fragment size adjustment was applied in several cases.18 WACA efficiently removes the wave artifact, thereby greatly improving the accuracy of array comparative genomic hybridization data analysis. Copy number variations/polymorphisms were identified and excluded from further analyses. Regions showing aberrant copy number changes were mapped according to the human reference sequence (NCBI36/hg18).

FISH Analyses

Based on the results obtained by array comparative genomic hybridization, the minimal common regions for copy number gains or losses were examined, including a search of potential relevant biomarkers. The resulting loci and targets were tested by fluorescence in situ hybridization (FISH), using either commercial or in-house FISH probes with the corresponding chromosome enumeration probe (CEP). The copy number change for 9q34 was validated with the commercial LSI 9q34 Spectrum Aqua Probe (Vysis) and CEP9 Spectrum Orange Probe (Vysis). In-house FISH probes targeted 1q21.3 (RP11-307C12, CKS1B as the target), 6q16.3 (RP11-460L11 - RP11-809N15, HACE1), 7p22.3 (RP11-16P10, MAFK), 9q33.3 (RP11-258M22, PPP6C), and 9q34.3 (RP11-413M3, CARD9). Generation of FISH probes and the FISH technique were previously described.19, 20, 21 BAC clones were labeled with Spectrum Green by nick translation (Abbott Laboratories, UK). Commercial probes were used with the Vysis paraffin pretreatment kit II (7J02-02, Vysis). Experiments were performed blinded to the array comparative genomic hybridization data. Image acquisition and processing was performed using a fluorescence microscope (Olympus BX51, Olympus K.K., Japan), Olympus DP70 digital camera system, and Olympus PD Controller v.1.1.1.65 and PD Manager v.1.1.1.71.

FISH signals were counted in at least 100 cells for each case and results were expressed by mean and standard deviation. Each of the FISH probes were investigated in ten reactive lymphoid tissues (tonsil and reactive lymph node) and the positivity threshold was set at mean±2 s.d. (2xSTD).

Results

Histology, EBER In Situ Hybridization and Immunohistochemistry

Histological findings of EATL by HE staining in the 20 cases are shown in Table 2 and Figures 1 and 2. A diffuse proliferation of small- to medium-sized cells were observed in five EATL cases (25%), medium-sized cells in two cases (10%), and medium- to large-sized cells in 13 cases (65%). All EATL cases had increased numbers of intraepithelial lymphocytes. Nevertheless, histologic evidence of celiac disease in the adjacent mucosa of the lymphoma such as villous blunting/atrophy and crypt hyperplasia were not found. In seven cases (35%) the intraepithelial lymphocytes were seen not only in the vicinity of the lymphoma but also at a distance from the main lesion in a greater or lesser degree. Necrosis was found in the tumor in three cases (15%). Reactive small T-cells showing a CD3ɛ+/CD5+ phenotype admixing with the neoplastic cells were observed in all cases but eosinophils, macrophages, and plasma cells were scarce.

Table 2 Histological findings and immunophenotype
Full size table
Figure 1
figure 1

Low-magnification view of enteropathy-associated T-cell lymphoma and the adjacent wall of the intestinal tract. The lymphoma cells are proliferating in all layers. Villous blunting in villi is not evident.

Full size image
Figure 2
figure 2

Phenotypic characteristics of enteropathy-associated T-cell lymphoma in Japan. Enteropathy-associated T-cell lymphoma in Japan is comprised of a heterogeneous lymphoid population from medium- to large-sized cells (a) with increased numbers of IELs (b), and with a characteristic phenotype similar to a cytotoxic CD8+ CD56+ lymphocyte. This figure shows a characteristic case where the lymphoma cells are CD3+ (c), CD5− (d), CD8+ (e), CD56+ (f), TIA-1+ (g), TCRβ− (h), TCRγδ+ (i), TCRγ− (j) and Epstein-Barr virus small RNA in situ hybridization-negative (k).

Full size image

Results of EBER in situ hybridization and immunohistochemistry are shown in Table 2 and Figure 2. Expression of markers (positive cases/examined cases; percentage) in lymphoma cells were as follows: EBER (0/20, 0%); CD2 (12/20, 60%); CD3ɛ (20/20, 100%); CD4 (2/20, 10%); CD5 (0/20, 0%); CD7 (19/20, 95%); CD8 (16/20, 80%); CD56 (17/20, 85%); TIA-1 (20/20, 100%); Granzyme B (5/20, 25%); TCRβ (2/20, 10%); TCRγ (7/20, 35%); TCRγδ (10/20, 50%); TNFRSF14 (19/20, 95%); and BTLA (4/20; 20%). The Ki-67 index ranged from 20 to 90% with a mean±STD of 50%±21.7.

Genomic Profile

Samples of good-quality DNA control gene PCR band >200 bp were obtained from eight formalin-fixed paraffin-embedded specimens. All cases except one showed multiple genetic alterations throughout the genome (Figure 3a). Table 3 shows the detailed genomic profile of each case. Recurrent gains were found at 1q32.3 (4/8, 50%), 4p15.1 (5/8, 63%), 5q34 (3/8, 38%), 7q34 (5/8, 63%), 8p11.23 (5/8, 63%), 8q24 (3/8, 38%), 9q22.31 (5/8, 63%), 9q33.2 (5/8, 63%), and 9q34.13 (6/8, 75%), and losses at 7p14.1 (6/8, 75%), 8p23.3-p11.21 (3/8, 38%) and 16q (4/8, 50%).

Figure 3
figure 3

(a) Summary view of the copy number alterations across the genome. The karyoview of enteropathy-associated T-cell lymphoma in Japan showed multiple regions of gains and losses such as gain of 4p15.1 (5/8, 63%), 7q34 (5/8, 63%), 8p11.23 (5/8, 63%), 9q22.31 (5/8, 63%), 9q33.2 (5/8, 63%), 9q34.13 (6/8, 75%) and losses of 7p14.1 (6/8, 75%). (b) Genomic profile of enteropathy-associated T-cell lymphoma case 11. This figure depicts the copy number changes for the enteropathy-associated T-cell lymphoma case 11. Deviations of the log2 ratios to the right represent gains and to the left losses; above them a line or dot is plotted depending on the copy number length. Below the karyoview the results for case 11 of fluorescence in situ hybridization targeting band 1q21.3, 6q16.3, 7p22.3, and 9q34.3 are presented. CEP1, 6, 7, and 9 (orange); 1q, 6q, and 7 (green); and 9q34 (aqua).

Full size image
Table 3 Genomic copy number changes in each case by human genome CGH microarray
Full size table

Region of alteration 1q21 ranged from 150853058 to 201504239, containing 34 genes, the most relevant of which were TIPRL, CKS1B, RGS2, CDC73, and PRCC (region filtered by cell cycle and oncogenesis). Region 6q16 ranged from 95601978 to 95641552 containing only one gene: MANEA (post-translational protein modification). ROA 7p22 ranged from 293816 to 20689516, containing 37 genes and the most relevant of which were DGKB, GNA12, ETV1, RPA3, GPER1, RAC1, MAFK, ADAP1, RBAK, C1GALT1, PDGFA, HDAC9, and CARD11 (DNA-binding transcription factor, nucleoplasm). ROA 9q34 ranged from 133531096 to 139816430, containing seven genes: FCN1; FCN2; C8G; TUBB4B; TRAF2; VAV2; and CARD9 (innate immune response).

Fluorescence In Situ Hybridization

A target gene with biological sense was selected from each of the ROAs for the FISH analysis (Figure 4). Cutoff values for each FISH probe for copy number gains and losses are shown in Table 4. Cutoff values for gains were 4%, 19%, 13%, 20%, and 17% for the 1q21, 6q16, 7p22, 9q33, and 9q34 bands, respectively. For losses, the cutoff values were 73%, 62%, 60%, 70%, and ND, respectively. In the eight cases with array comparative genomic hybridization results, good correlation between FISH and array comparative genomic hybridization was observed (64% of agreement). Of note, the agreement reached a 82% when a minimum of 30% of EATL cells with FISH alterations were set up as threshold for array comparative genomic hybridization technique sensitivity; reaching perfect 100% correlation for 9q34 locus.

Figure 4
figure 4

Correlation between microarray and fluorescence in situ hybridization findings. This figure shows the correlation between microarray and fluorescence in situ hybridization at four different locations. In-house probes (1q21.3, CSK1B; 6q16.3, HACE1; and 7p22.3, MAFK; green signals) and commercial probe (9q34.3, ASS1, aqua signals) with the corresponding chromosome enumeration probe (orange). (a) Chromosomal gain at 1q21 (three green and two orange signals); (b) no change at 6q16 (two green and two orange signals); (c) loss at 7p22 (one green and two orange signals); and (d) gain of 9q34 (three aqua and three orange signals). Of note, good correlation with the array comparative genomic hybridization results was observed for those cases (the figure depicts the waves array comparative genomic hybridization correction algorithm plots).

Full size image
Table 4 Correlation between microarray and FISH findings
Full size table

The FISH analysis was expanded to the 12 additional cases with no array comparative genomic hybridization results available. Overall, FISH showed 15/20 patients (75%) with a gain of 9q34 (ASS1), with good correlation between array comparative genomic hybridization results (88%). Gains of 1q21.3 (CKS1B), 6q16.3 (HACE1), 7p22.3 (MAFK), 9q33.3 (PPP6C), and 9q34.4 (CARD9) were observed in 12/18 (67%), 6/16 (38%), 1/13 (8%), 6/15 (40%), and 12/17 (71%) cases, respectively. Losses of 6q16 and 7p22 were observed in 1/16 (6%) and 5/13 (39%) cases (positive/ examined cases). No cases showed any loss of 1q21 or 9q33.

No differences were observed between the different EATL TCR phenotypes (TCRβ, γδ and −/−) regarding the genomic profiles of array comparative genomic hybridization and FISH.

Discussion

In the present study, we investigated the genomic and immunohistochemical profiles of EATL cases in Japan using formalin-fixed paraffin-embedded samples. EATL is an uncommon lymphoma of intestinal intraepithelial lymphocytes associated with celiac disease, with a higher frequency of incidence in Northern Europe. Data of EATL occurrence in Asian countries as well as its clinicopathological characteristics are limited. Types I and II EATLs differ in terms of histological appearance, whereas in Western countries EATL Type II is characterized by small- to medium-sized tumor cells,1, 2 the majority of our Japanese cases showed medium to large tumor cell proliferation (13/20, 65%). Chan et al.8 described that there was some degree of nuclear size variation in tumor cells of Chinese EATL. Our cases showed similarity to those cases. Therefore, size variation in tumor cells is characteristics of Asian cases and that may be one of differences from Western cases. Because no history of celiac disease was present in our series, we conclude that there may be some different immunological background characteristics in Japanese cases of EATL.

The phenotypic profile of Japanese EATL corresponded to cytotoxic CD8-positive and CD56-positive intraepithelial lymphocytes. Immunohistochemistry in our series revealed that the tumor cells frequently expressed CD8 (16/20, 80%) and CD56 (17/20, 85%). These results are consistent with the immunophenotype of Type II EATL as previously described.1, 2, 3, 4, 5, 6 In the present study, a high proportion of cases expressed γδ-TCR (10/20, 50%), and six cases were negative for both αβ and γδ-TCR, a finding that differs from the description in the WHO classification, which states that EATL Type II cases express αβ-TCR. Tan et al.7 reported predominant expression of αβ-TCR over γδ-TCR, and Chan et al.8 reported that 78% of Type II EATL cases expressed γδ-TCR. Our results confirm that the expression of γδ-TCR is frequently associated to the EATL Type II in Asia. Three cases (12, 17, and 19) expressed γ-TCR but were negative for γδ-TCR. Estimation of the γδ-TCR antibody is controversial. A study using the same antibody (TCR gamma+delta antibody clone 5A6.E9) can successfully recognize γδ-T-cells with membranous staining in formalin-fixed paraffin-embedded tissue,22 but there was another one reporting that the antibody was not so robust.23 Although membranous positive-staining of γδ-TCR is acceptable as positive, we need further study for γδ-TCR.

The genomic profile of Japanese EATL was characterized by multiple regions of copy number gains and losses as shown in the karyoview (Figure 3). From the most common and/or characteristic abnormal loci, the ROAs and minimal common regions and gene candidates were found and selected genes validated by FISH. It has been reported that EATL is characterized by recurrent gains of 1q, 5q, 7q, and recurrent losses of 8p, 9p, and 13q, with gains of 9q33-q34 being the most frequent alteration (60% of frequency).15 In line with those previous findings, the present series of EATL was characterized by multiple common and distinct imbalances. In contrast to other types of tumors, the EATL profile did not have a single or unique alteration, but rather have multiple regions, an observation that supports the view that EATL is a multifactorial and polygenic (complex) disorder. It was reported that Type I EATL frequently shows chromosomal gains of 5q34-q35.2 and 1q32.2-q41, while Type II is more often characterized by amplification of 8q24. Deletion of 16q12.1 and amplification of 9q31.3-qter chromosomal regions is frequently observed in both Type I and II EATLs. We demonstrated a large number of genetic abnormalities in our Type II EATLs: gains of 1q, 5q, 7q, and 9q, and losses of 8p, 9p, and 13q, with the most common gain of 9q33-q34 (58%). The general profile is in concordance with previously reported data from BAC-arrays of Deleeuw et al.15 Our high-resolution array comparative genomic hybridization and expanded FISH analysis in a series of EATL Type II in Japan has identified a genomic profile similar to EATL of Western countries with characteristics of both Types I and II. Despite that our cases are phenotypically more similar to EATL Type II the histology of the tumor cells is more heterogeneous and range from small to large; and the genomic profile lacks the EATL Type II characteristic genomic copy number gain of MYC oncogene locus at 8q24 but has the 1q32 and 5q34 gains previously described as characteristic of EATL Type I. The results of this study, therefore, raise the question of whether in Asian countries the distinction between EATL Type I and II may not be as clear as previously thought. Of note, the gain at 9q34 loci was the most frequent copy number change confirmed by FISH (ASS1) in 75% of the cases.

Takeshita et al.6 previously reported that only 7% of cases had an amplification of the 9q band in EATL cases in Japan by metaphase comparative genomic hybridization. Our series has refined and expand these findings observing a gain of 9q34 in 75% of cases and confirmed in 75% (ASS1) and 71% (CARD9) of cases by FISH. Nevertheless, there was no deletion of 16q12.1. Gain of the 9q33 band was observed in 50% (4/8) of cases by array comparative genomic hybridization and in 43% (6/14) by FISH. A good correlation was found at the 9q34 band between array comparative genomic hybridization and FISH. Therefore, we propose that gain of 9q34 is frequent in Japanese EATL cases, similar to the European series.14, 15 Among other important gene targets, the target gene CARD9 was selected for FISH validation in the 9q34 region. CARD9 is an oncogene regulator of the ITAM-mediated signaling pathway, forming a complex with BCL10–MALT1, and is a regulatory mechanism for NFKB and MAPK activation through innate and adaptive immunereceptors.14 Importantly, CARD9 is abundantly expressed in MALT lymphoma and DLBCL.24, 25, 26 At 9q34 band we also investigated by FISH by commercial probe the locus containing the tumor suppressor gene of ASS1.

Gain at the 1q region has been described in both refractory celiac disease IEL cell lines and EATL by metaphase27 and array comparative genomic hybridization (Carreras and co-workers, unpublished observations). By array comparative genomic hybridization, our series showed 1q21 gain in 38% (3/8) of cases with CKS1B as the target gene, a cell cycle regulator. Subsequent FISH screening of the entire EATL series confirmed gain in 67% (12/18) of cases. CKS1B is known to have an oncogenic role in the pathogenesis of hematological malignancies such as mantle cell lymphoma and follicular lymphoma.28, 29 Therefore, CKS1B may similarly be associated with EATL lymphomagenesis.

At 6q16, the region of interest contained only one gene: MANEA. However, we selected HACE1 as the target marker due to its interaction with HRAS and function in degrading cellular proteins as well as because its genomic implications for intestinal T-cell lymphomas (Carreras J and co-workers, unpublished observations). As a tumor suppressor, HACE1 inactivation in mice leads to the generation of cancer, a process that is enhanced with the addition of ‘second hits’ such as mutations in p53. HACE1 is frequently downregulated in human tumors.30, 31 Methylation of HACE1 has been identified in gastric and colorectal cancers and its expression is markedly reduced in nasal type extranodal NK/T-cell lymphoma.32 Recently, the 6q21-22 region, which contains HACE1, has been confirmed as a celiac disease susceptibility locus (possibly mutated).33 In our series, loss could not be identified by array comparative genomic hybridization, and loss frequency was observed in only 6% (1/16) of cases by FISH. Therefore, the loss of 6q is not characteristic of EATL cases in Japan and may be useful for exclusion of peripheral T-cell lymphoma, not otherwise specified and Extranodal NK/T lymphoma, nasal type.

We observed a gain of 8q24 in 25% (2/8) of cases by array comparative genomic hybridization. This data is in agreement with previously reported data of EATL Type I (27%).15 In addition, there have been no reports on the loss of 19q33.33 and gain of 4p15, 8p11, 12p13, and 16q23, which we have observed to be minimal common regions.

In conclusion, EATL cases in Japan have a phenotype of CD8+CD56+ cytotoxic lymphocytes and increased number of intraepithelial lymphocytes, yet patients lack history of celiac disease. The results of this study point out a more blurred distinction between EATL Type I and II in Asian countries. Although it remains to be clarified what immunological background triggers EATL pathogenesis, the genomic profiles of EATL in Japan is characterized by multiple regions of copy number gains and losses, with the most characteristic being the 9q34 gain, irrespective of the cell origin of αβ-T-cells and γδ-T-cells.