Original Article | Published:


Type II EATL (epitheliotropic intestinal T-cell lymphoma): a neoplasm of intra-epithelial T-cells with predominant CD8αα phenotype

Leukemia volume 27, pages 16881696 (2013) | Download Citation


In this multicentre study, we examined 60 cases of Type II enteropathy-associated T-cell lymphoma (EATL) from the Asia-Pacific region by histological review, immunohistochemistry and molecular techniques. Patients were mostly adult males (median age: 58 years, male:female 2.6:1), presenting with abdominal pain (60%), intestinal perforation (40%) and weight loss (28%). None had a history of coeliac disease and the median survival was only 7 months. Histologically, these tumours could be divided into (i) central tumour zone comprising a monotonous population of neoplastic lymphocytes, (ii) peripheral zone featuring stunted villi and morphologically atypical lymphocytes showing epitheliotropism, and (iii) distant mucosa with normal villous architecture and cytologically normal intra-epithelial lymphocytes (IELs). Characterized by extensive nuclear expression of Megakaryocyte-associated tyrosine kinase (MATK) (87%) and usually a CD8+CD56+ (88%) cytotoxic phenotype, there was frequent aberrant expression of CD20 (24%). T-cell receptor (TCR) expression was silent or not evaluable in 40% but of the remainder, there was predominant expression of TCRαβ over TCRγδ (1.6:1). In keeping with the normal ratio of IEL subsets, CD8+ cases showed predominant CD8αα homodimer expression (77%), regardless of TCR lineage. These tumours constitute a distinct entity from classical EATL, and the pathology may reflect tumour progression from IEL precursors, remnants of which are often seen in the distant mucosa.


Enteropathy-associated T-cell lymphoma (EATL) constitutes 5.4% of peripheral T-cell lymphoma and comprises two subtypes.1, 2 So named for its association with coeliac disease, classical (Type I) EATL is more common in Western populations. By contrast, Type II EATL constitutes 10–20% of EATL in Western populations2 whereas it is the predominant subtype in Asia3, 4 where coeliac disease is rare and the prevalence of the associated HLA-DQ2 and DQ8 haplotypes is low.5, 6, 7, 8, 9 The tumour features monomorphic T-lymphocytes and lacks the prominent inflammatory infiltrate that accompanies the classical EATL. Neoplastic cells typically display the CD8+CD56+ cytotoxic phenotype and are often CD30; whereas Type I EATL is usually CD4CD8 and at least focally positive for CD30.

There are a number of conflicting opinions about Type II EATL. For example, a study from the International T-cell Lymphoma Project concluded that both the types of EATL are associated with coeliac disease,1 although multiple reports from Asia indicated no such association.3, 4, 10, 11, 12, 13, 14 There is also considerable debate about the phenotype of this neoplasm. Type II EATL was originally described as an intestinal T-cell lymphoma with a CD8+CD56+ phenotype,2, 15 but neither marker is mandatory for diagnosis and cases have been reported that lack either or both these molecules.3, 4, 12, 16 Lymphocytes in the intra-epithelial compartment are described as having a phenotype that is similar to2 or consistently discordant17 from that of the invasive tumour. The World Health Organization (WHO) 2008 classification describes Type II EATL as a tumour of intra-epithelial αβ T-cells,2 but others have reported that a significant proportion displays the T-cell receptor-γδ (TCRγδ) phenotype.3 Finally, the precursor cell for Type II EATL remains uncertain.

We have reported the expression of a novel marker,Megakaryocyte-associated tyrosine kinase (MATK), in Type II EATL.18 During screening of lymphoid neoplasms, we identified four older cases of ‘EBV (Epstein–Barr virus negative) natural killer (NK)/T-cell lymphoma’ with high-level (>70%) nuclear expression of MATK. This prompted histological review and subsequent re-classification as Type II EATL. To gain a better understanding of the pathology of this neoplasm, we performed a larger study using MATK and other markers on cases contributed by multiple investigators in the Asia-Pacific region. A subset of these (cases 1–16, 20–33, 44–56) had been reported12, 17, 18 in earlier studies.

Given that this neoplasm is rare and known by different names (for example, CD56+ intestinal lymphoma,10, 15 EATL,17 primary intestinal NK-like cytotoxic T-cell lymphoma11, 14 before the 2008 WHO Classification of haematolymphoid neoplasms2), it is likely to be under-reported. In addition, with cases gathered from multiple centres over more than a decade, there was heterogeneity of management, and clinical data for many cases were not available. Therefore, the intention of this study is to focus on the pathology rather than epidemiology and clinical aspects.

Materials and Methods

Case selection

Collaborators from seven countries were asked to contribute cases of Type II EATL with available pathological material for review and further investigations. A total of 60 cases of Type II EATL over the period 1999–2012 were selected from nine centres in Singapore, Taiwan, Korea, Hong Kong, Malaysia, Australia and China. Whenever possible, clinical findings and survival data were retrieved from the hospital records. A clinical history of coeliac disease was specifically sought. Although it is known that a proportion of coeliac disease may be subclinical,19 in a region where this disease is rare (apart from Australia), more sensitive investigations such as anti-endomysial and anti-transglutaminase antibodies were not routinely performed in patients who had no previous gastrointestinal symptoms.

Histology review and diagnostic criteria

Representative diagnostic material, including haematoxylin and eosin-stained slides and unstained paraffin-embedded sections, were submitted for histology review and more detailed immunophenotyping.

A diagnosis of Type II EATL was considered when the patient presented with an intestinal lymphoma composed of EBVT-cells showing intra-epithelial lymphocytosis and with an appropriate phenotype in accordance with the 2008 WHO classification.2 For the purpose of this study, cases must display a cytotoxic phenotype (T-cell intracellular antigen (TIA1), granzyme B or perforin) and lack Epstein–Barr virus-encoded RNA (EBER) expression. They must also express some pan-T-cell markers such as CD2, CD3, CD5, CD7, TCRβ or TCRγ. Expression of CD8 and CD56, although typical for this neoplasm, was not considered mandatory for the diagnosis.3, 4, 12, 16 Prospective cases were selected by contributing pathologists from their respective hospital archives and reviewed by the study pathologist (TSY) following further immunostains. A case was entered into the study if there was consensus between the contributing pathologist and the study pathologist. Forty-three cases had been reported in previous studies, including 13 from Taiwan,11, 17 Korea,12 Singapore,18 whereas five other cases were excluded because they were EBER-positive.

Immunohistochemistry and in situ hybridization

Immunohistochemical analysis of paraffin-embedded tumour sections was performed with a variety of antibodies (Supplementary Table S1, Supplementary information), using the BONDMAX automated staining machine (Leica Microsystems, Wetzlar, Germany). In situ hybridization for EBER (Leica Microsystems) was also performed with known positive control tissue(s) on the same platform.

Interphase fluorescence in situ hybridization (FISH)

FISH was performed using the C-MYC break-apart and chromosome 8 centromeric probes (Abbott Laboratories, Abbott Park, IL, USA) and FISH ancillary kit (DAKO A/S, Glostrup, Denmark) according to the manufacturer’s instructions.20

T-cell clonality analysis by PCR

PCR for T-cell clonality analysis was performed in samples from two institutions using Biomed II primers (TCRG Gene Clonality Assay, InVivoScribe Technologies, San Diego, CA, USA), one with capillary and the other with gel electrophoresis. Briefly, DNA was isolated from formalin-fixed, paraffin-embedded tissue and PCR amplification was carried out using Tube A (VGIf and VG10+JG1.1/2.1 and JG1.3/2.3) and Tube B (VG9 and VG11+JG1.1/2.1 and JG1.3/2.3) primers. In another institution, TCRG gene rearrangement studies were performed on DNA extracted from paraffin-embedded tissues using a published protocol.21 Two oligonucleotide primer sets were used, targeting TCRJγ (5′-CAA-GTG-TTG-TTC-CAC-TGC-C-3′) and TCRJPγ (5′-GTT-ACT-ATG-AGC-YTA-GTC-C-3′). Each PCR study was carried out in duplicate and included positive, negative and no-template controls. PCR products were analysed by electrophoresis on a 16% polyacrylamide gel.


Clinical data

The clinical data are presented in Supplementary Table S2 (Supplementary information) and summarized in Table 1. The median age at presentation was 58 (range: 23–83) years with male predominance (male: female ratio 2.6:1). Apart from 3 Caucasians, all the remaining 57 patients were Asians (42 ethnic Chinese, 8 Koreans, 6 Malays, 1 Eurasian). In the 40 patients with clinical data, abdominal pain (60%), bowel perforation (40%), weight loss (28%) and diarrhoea (28%) were among the commonest presenting features. Of 35 cases with staging data, 18 cases (51%) were in Stage I, 9 (26%) in Stage II, 1 (3%) in Stage III and 7 (20%) in Stage IV. Histologically proven extra-gastrointestinal sites of involvement were observed in eight patients, affecting the lung,3 omentum2 and cervical lymph nodes.2 Of note, none had a history of coeliac disease. No anti-endomysial (0/4) and anti-transglutaminase antibodies (0/3) were detected in the cases tested. HLA typing identified one case with HLA DQ2 and two cases with DQ8; whereas one other case lacked both haplotypes.

Table 1: Summary of the clinical data

The small intestine was the single most commonly involved site (48/60, 80%), mainly affecting the jejunum and ileum. In 6 cases (10%), the neoplasm was restricted to the large intestine while both the small and large intestines were simultaneously involved in another 6 cases (10%).

Therapeutic data were available in 42 cases, and of these, chemotherapy was administered in 32 patients while 10 had surgical resections only. Two patients received consolidative autologous stem cell transplants and four patients were transplanted after salvage therapy following relapse. Follow-up data were available for 44 patients and at last review (follow-up period up to 85 months), 32 had died of disease, 5 were alive with disease and 7 were alive without evidence of disease. Tumour relapse after surgical resection with or without chemotherapy occurred in 10 cases. The longest relapse-free survival was 61 months in a transplanted patient and the median overall survival was 7 months (range: 0.5–85 months). Two patients given upfront autologous transplants were alive without disease at last follow-up, with overall survival of 15 and 61 months, respectively.

Pathological findings

Of the 60 cases recruited into the study, 29 had been previously listed as other entities (19 peripheral T-cell lymphoma, unspecified; 5 extranodal NK-/T-cell lymphoma; 1 gamma-delta T-cell lymphoma, 3 classical enteropathy associated T-cell lymphoma, 1 lymphocytic colitis) in the archives of the originating institutions and were re-classified by the contributing pathologists after histological review.

The neoplasm could be anatomically divided into a central tumour zone comprising the lymphomatous infiltrate, an adjacent peripheral zone of intramucosal lymphoma containing cytologically atypical intra-epithelial lymphocytes (IELs) showing prominent epitheliotropism and, beyond that, a distant mucosa zone containing morphologically normal IELs.3 Although IELs were seen in the peripheral zone and distant mucosa, the former were cytologically atypical, resembling cells in the central tumour zone, while the latter were morphologically normal.

Central tumour zone

The tumour was often ulcerated and comprised a monotonous population of medium-sized lymphoid cells with slightly dispersed chromatin, inconspicuous nucleoli and ample clear cytoplasm, imparting a monocytoid appearance (Figure 1). Occasional cases displayed a greater degree of nuclear pleomorphism with larger cells, but unlike Type I EATL, there was a paucity of reactive inflammatory cells and mitoses. Necrosis was not a usual feature. In the 51 cases with available data, there was transmural invasion extending to subserosa (6 cases, 12%), serosa without tumour perforation (4 cases, 8%), serosa with tumour perforation (33 cases, 65%) and invasion of adjacent structures (loops of bowel, soft tissue) in 8 cases (16%).

Figure 1
Figure 1

Morphological, phenotypic and genetic features of Type II EATL. Row 1: (a) Morphologically normal IELs in distant mucosa featuring small nuclei with condensed chromatin. Notice the intact brush border and goblet cells. (b) Cytologically atypical neoplastic cells in the peripheral zone with few residual enterocytes overlying the lymphomatous infiltrate. (c) Neoplastic cells in the central tumour zone show mild nuclear pleomorphism. Rows 2 and 3: Typical phenotype of Type II EATL, being (a–f, row 2) CD79a, CD20, CD2+, CD3+, CD5, CD7+, (a–f, row 3) CD4, CD8α+, CD8β, TIA1+, granzyme B+ and CD56+. Row 4: Neoplastic cells show nuclear expression of (a) MATK but are negative for EBER, although occasional (b) EBV-infected B-cells may be present. Majority of cases display (c, d) the TCRαβ phenotype and (e, f) the γδ TCR lineage is seen in a smaller number. Row 5: Loss of CD8α is seen in the recurrence of a (a, b) CD8α+ primary tumour while aberrant expression of (c, d) CD20 but not PAX5/BSAP is noted. Row 6: Interphase FISH using dual colour break-apart probe shows (a) gains and (b) translocation of C-MYC while a clonal proliferation can be demonstrated by PCR for (c) TCRγ gene analysis, even in cases that have a TCRβTCRγ phenotype. (d) Immunostaining for MATK shows lack of nuclear staining in neoplastic cells of classical Type I EATL. (e) Scattered EBV-infected cells may be seen in Type II EATL but (f) double staining with CD79a confirms EBER+ cells are B-cells.

Most pan-T markers were expressed, including CD2 (71%), CD3 (100%) and CD7 (96%). CD5 was usually negative, with only 3 cases (6%) showing variable degrees of staining. This tumour was typically positive for CD8 (88%) and CD56 (87%), while EBER expression was limited to scattered non-neoplastic B-cells (usually <10% of cell population). Of the 39 CD8+ cases that were available for further testing, 30 (77%) expressed CD8αα homodimers and 9 cases (23%) expressed CD8αβ heterodimers.

A constant feature was the cytotoxic phenotype with all but one case (98%) showing expression of TIA1. The single negative case expressed granzyme B instead, which was present in 29/46 (63%) of cases. TCRβ was expressed in 22/54 cases (41%), of which two cases also showed focal and weaker staining for TCRγ, probably aberrant in nature. Fourteen of the 54 cases (26%) expressed TCRγ and not TCRβ, in keeping with TCRγδ lineage. Of the remaining 24 cases, 18 lacked both TCR receptors and 6 cases were not evaluable due to lack of material or internal positive control (Figure 2). The reason for TCR-silent cases was unclear but not likely to be technical in nature as there were positive-staining reactive cells as internal control (present in 18 cases). Given that a TCR was expressed in IELs of adjacent mucosa in six of these cases, the loss of TCR might be due to downregulation of TCR expression as part of clonal evolution.

Figure 2
Figure 2

Phenotypic subsets of Type II EATL correspond to the subsets of intra-epithelial T-cells in the intestinal mucosa. Most Type II EATLs express TCRβ with a significant minority showing TCRγδ phenotype, in keeping with the relative proportions of αβ- and γδ-T-cells in the intestinal mucosa. Regardless of T-cell lineage, most Type II EATLs express CD8αα homodimers, corresponding to the relative proportions of CD8αα versus CD8αβ subsets in the normal intestinal mucosa. ND, not done.

Aberrant expression of the pan-B marker CD20 has been reported3 and was seen in 24% of our cases. In keeping with our previous observation,18 extensive (>70%) nuclear expression of the novel marker MATK was present in 47/54 cases (87%).

Phenotypic heterogeneity was a prominent feature and focal loss of staining in a proportion of tumour cells was common for MATK, TCRβ, TCRγ, CD8α and CD56. Two of the five cases in our series that relapsed after treatment displayed a change in phenotype in the relapsed lesions, with loss of CD2, CD45RO and CD8 in one case and loss of MATK and CD56 in the other.

By way of comparison, six cases of Type I EATL in Caucasian patients with coeliac disease were obtained from a collaborator in the United Kingdom and examined with regard to their histology and immunophenotype. They featured medium-to-large lymphoid cells with irregular nuclear contours and a greater degree of anaplasia. There was often an associated inflammatory infiltrate comprising variable numbers of eosinophils, histiocytes, small lymphocytes and plasma cells. Five of the six cases lacked both CD4 and CD8 with variable expression of CD30. They displayed a TIA1+CD3+CD7+CD5 phenotype, and unlike Type II EATL, nuclear staining for MATK was not seen.

Peripheral zone

Adjacent to the main tumour was a peripheral zone characterized by prominent epitheliotropism but with relatively little involvement of the submucosa and muscularis propria. The mucosa showed some degree of villous blunting and goblet cell depletion but not crypt hyperplasia. IELs were morphologically atypical and resembled cells in the central tumour zone, with larger nuclei and more dispersed nuclear chromatin than the IELs in the distant mucosa (Figure 1). This histological appearance had been attributed to lateral spread of neoplastic cells from the central tumour zone,3 somewhat akin to the radial growth phase in melanoma.

The immunophenotype in this zone was generally identical to that of the main tumour mass, although in 4/9 cases, the phenotype was intermediate between that of the central tumour zone and the IELs in the distant mucosa. This was true of CD56, granzyme B and MATK, which were negative in IELs of the distant mucosa, weakly positive in the peripheral zone and positive in the central tumour zone (Figure 3). By contrast, in two CD8 tumours, the expression of CD8 in the peripheral zone was intermediate in intensity between IELs of the distant mucosa (strongly positive) and cells of the central tumour (negative).

Figure 3
Figure 3

Phenotypic variation in the distant mucosa, peripheral zone and central tumour zone. Upregulation of MATK and CD56 with downregulation of CD7, CD8, TCRβ and TCRγ during transformation of IELs in the distant mucosa to lymphomatous cells in the central tumour.

Distant mucosal zone

Beyond the peripheral zone, there was preservation of villous architecture as well as retention of goblet cells (Figure 4). However, increased IELs (>30 per 100 epithelial cells) were noted, featuring morphologically normal lymphocytes with small nuclei, condensed chromatin and scanty cytoplasm. This zone of increased IELs usually extended for a variable length but could be discontinuous and patchy in distribution with stretches of intervening normal mucosa. Increased IELs in distant mucosa were seen in 26 cases, but due to the limited material available for assessment in many instances, it was not possible to determine whether they were always present, as some have suggested.3

Figure 4
Figure 4

Prominent epitheliotropism without enteropathy in Type II EATL. Despite the prominence of intra-epithelial lymphocytes, there is no significant villous atrophy or crypt hyperplasia. Higher magnification shows preservation of goblet cells and the brush border despite florid epitheliopism by CD8+ T-cells.

Depending on which markers were examined, a discordant phenotype might be observed between the tumour and the corresponding IELs, mostly involving CD2, CD7, CD8, CD56, MATK, granzyme B, TCRβ and TCRγ. As illustrated in Figure 3, CD56, granzyme B and MATK were usually weak or negative in IELs of the distant mucosa, whereas positive expression was seen in cells of the central tumour zone. By contrast, CD2, CD8, TCRβ and TCRγ (depending on lineage) were expressed in cytologically normal IELs, whereas they might be lost in the central tumour zone (Figure 5). Of the 18 tumours that lacked both TCRβ and TCRγ expression, the phenotype of the associated IELs could be determined in 6 cases, being positive for TCRβ in 4 and TCRγ in 2 cases.

Figure 5
Figure 5

Phenotypic changes in Type II EATL with respect to anatomical location. This chart illustrates the phenotypic changes in various markers in the distant mucosal zone, peripheral zone and central tumour zone.

Genetic findings

Interphase FISH using a locus-specific probe for C-MYC and a centromeric probe for chromosome 8 was performed in 24 cases. They showed translocation and gains (due to aneusomy) of C-MYC in 1 (4%) and 7 (29%) cases, respectively. PCR for TCRγ gene analysis confirmed a clonal pattern in 27 of 29 cases tested (93%), including 7 cases that lacked expression of both TCRs, thereby excluding NK-cell origin and confirming a T-cell lineage. In four cases with sufficient material for testing, clonality analysis was separately performed on the central tumour and matched distant mucosa containing IELs, all of which showed a clonal relationship between the tumour and the IELs.


Is Type II EATL associated with coeliac disease?

The association with coeliac disease is a subject of much debate. Delabie et al.1 reported an association with coeliac disease in both the forms of EATL, whereas numerous Asian publications on Type II EATL disputed this.3, 4, 10, 11, 12, 13, 14 None of the patients in this current study had a clinical history of pre-existing coeliac disease. Admittedly, more sensitive investigations such as anti-endomysial and anti-transglutaminase antibodies were only performed in a few cases (all of which were negative). Thus, we might have missed a small proportion of patients with subclinical coeliac disease. Nonetheless, given the rarity of coeliac disease in Asia, these tests were not routinely performed in the absence of a suspicious history.

There is also a controversy over the existence of enteropathic changes in the histology. Although Takeshita et al.16 found ‘enteropathy-like changes’ (so defined because the authors noted increased IELs and villous atrophy but not crypt hyperplasia), another Japanese group concluded that no evidence of enteropathy could be found.22 Part of the problem may lie in the fact that the term ‘enteropathy’ has been applied rather loosely and that the histology of Type II EATL varies from one area of the tumour to another.

As noted by other investigators,3 the pathology of this neoplasm can be divided into three zones with differing morphological features. Without an appreciation of these zonal differences, the stunted villi in the peripheral zone may be misconstrued as ‘villous atrophy’ while another observer may recognize that the villous architecture is normal in the distant mucosa. In this series, the histology in the distant mucosa is distinct from the enteropathic changes seen in coeliac disease and ulcerative jejunitis.

What is the typical immunophenotype of Type II EATL?

The expression of CD8 and CD56 was originally presented as defining features of this tumour,2, 15 but a significant number of cases have been reported to lack either or both these markers.3, 4, 12, 16 The IELs were described to share a similar phenotype as the neoplastic cells in the lymphoma,2 but Chuang et al.17 found that the phenotype of the neoplastic cells in the central tumour differed from IELs in the distant mucosa. Finally, Chan et al.3 opined that the phenotypes could be either similar or different.

We have shown that not only is there phenotypic variation between cases, but the phenotype may also differ between histological zones in a given case. The atypical IELs in the peripheral zone are often phenotypically concordant with those in the main tumour zone or intermediate in expression between cells of the central tumour and distant mucosa. For example, CD56 expression is often strong in the central tumour zone, weak in the peripheral zone and negative in the distant mucosal zone. By contrast, IELs in the distant mucosal zone often displayed a discordant phenotype with respect to the main tumour, usually involving CD2, CD8, TCRβ, TCRγ, MATK and CD56. This may explain the conflicting opinions over whether lymphocytes in the intra-epithelial compartment have a similar2 or discordant17 phenotype compared with the main tumour.

In addition, there is also phenotypic variation with respect to chronology. A 48-year old man (case 4) presented with Type II EATL in the small intestine that was resected and was found to be CD8+ initially. At the first recurrence, the neoplasm was found to be weakly CD8+ and at the second recurrence, the tumour became CD8.

Seen in 24% of cases, aberrant expression of CD20 (Figure 1) may potentially create diagnostic confusion if only a limited panel of immunostains is used. However, the expression is usually more heterogeneous and weak, and distinction from B-cell lymphoma can be made by the absence of other pan-B markers. EBV-infected B-cells as a consequence of a reduction in immune surveillance in T-cell lymphomas may be encountered and occasionally may be sufficiently numerous as to create diagnostic confusion with extranodal NK/T-cell lymphoma. As we have previously reported,18 high-level nuclear expression of MATK is a useful additional marker of Type II EATL to resolve diagnostic difficulties.

What is the TCR lineage of Type II EATL?

Type II EATL is considered a neoplasm of IELs with a TCRαβ phenotype,2, 15 but it has been reported that 78% of cases expressed the TCRγδ receptor.3 In this larger series, apart from 18/54 (33%) cases that were TCR-silent, more cases (22/54, 41%) expressed TCRαβ receptor and although a significant number (14/54, 26%) showed γδT-cell lineage, this is a lower figure than what had been reported.

What is the cellular origin of Type II EATL?

The presence of clonal TCR gene rearrangements and the expression of TCRβ or TCRγ proteins confirmed the T-cell origin of this neoplasm. A previous study had shown that clonal TCR gene rearrangements in IELs were identical to those of the lymphomatous component17and we found that to be true in all four cases that were so tested. Taking into consideration the phenotypic changes in the histological zones, it appears that this neoplasm arises from CD8+ cytotoxic T-cells that may downregulate CD2 and CD8 but upregulate MATK, granzyme B and CD56 during tumour progression.

Intestinal T-cells comprise effector memory cells that are TCRαβ+ and CD4+ or CD8αβ+ (type ‘a’) and a subset that may be either TCRαβ+ or TCRγδ+ but lacking in CD4 or CD8αβ type ‘b’). ‘Type a’ cells are mainly located in the lamina propria while ‘type b’ cells often express CD8αα instead of CD8αβ and are located in the intra-epithelial compartment.23, 24 Intestinal IELs therefore belong to a heterogeneous population that is mostly CD8+TCRαβ+ (60–90%), of which two-thirds express CD8αα and a third expresses CD8αβ. The remainder are CD8+TCRγδ T-cells (10–40%), most of which also express CD8αα.25 They are thought to be important in maintaining enterocyte homoeostasis and have an immunoregulatory role in gut immunity.24

Expression of TCR was either absent or not evaluable in 24/60 (40%) cases in this study, but of the remainder, cases of TCRαβ lineage predominated with a TCRαβ:TCRγδ ratio of 1.6:1 and with the majority of CD8+ cases expressing the CD8αα (77%) rather than CD8αβ (23%) phenotype, regardless of the TCR lineage. Although our observations were only limited to the 60% of cases with TCR and CD8 expression, it is of interest that these figures are in line with the relative proportions of IEL subsets in the normal intestinal mucosa (Figure 2). This suggests that the probability of neoplastic evolution from its precursor cell may depend entirely upon the relative frequency of the latter, that is, purely random, or ‘stochastic’, and not skewed by any intrinsic predisposition of any subset to undergo neoplastic transformation.

What is known about the prognosis and preferred treatment?

With an overall median survival of only 7 months, our data confirmed the reportedly poor prognosis of Type II EATL.1 However, treatments used in this multicentre study were heterogeneous, and a significant proportion received non-anthracycline-based chemotherapy, limiting somewhat our ability to conclude definitively on the prognosis. Nonetheless, it should be noted that the International T-cell Lymphoma Project reported that patients did not benefit from the use of an anthracycline-containing regimen over a non-anthracycline-containing regimen.26

In our series, 32 patients underwent surgery and chemotherapy while 10 had resection only at the time of this study, usually because they were not fit for further treatment. As is true for most subtypes of mature T-cell lymphoma, the optimal treatment for Type II EATL is not known. This is reflected by the variations in treatment in this multicentre study. It is interesting to note that two patients who were treated with upfront high-dose chemotherapy and autologous stem cell support (patient 6, 17) appeared to experience prolonged overall survival although this could also be due to selection bias. At present, optimal treatment of Type II EATL remains to be defined and novel strategies are urgently required.

A distinct entity deserving a new name

The term EATL was chosen for a disease that is a sequela of coeliac disease and which arises from IELs that are increased in this intestinal disease. Type II EATL was considered a variant of EATL primarily because increased IELs are also seen in this disorder.

The term ‘epitheliotropism’ refers to the propensity to associate with epithelial cells. In this respect, the increased IELs seen in coeliac disease, Type I and Type II EATL can all be broadly considered ‘epitheliotropic’. However, the increase in IELs in coeliac disease and Type I EATL is present in the context of enteropathy. Apart from increased IELs, there are also clinical (malabsorption, chronic diarrhoea) and pathological (villous atrophy, crypt hyperplasia, elevated anti-glutaminase antibodies) features of enteropathy. Indeed, the WHO classification espoused the term ‘Enteropathy-associated T-cell lymphoma’ rather than ‘Enteropathy-type T-cell lymphoma’ to emphasize the link with coeliac disease.

By contrast, the clinico-pathological features of enteropathy are lacking in Type II EATL. Gastrointestinal symptoms in Type II EATL are often not apparent until fairly late in the disease, with perforation being a common initial presentation. When diarrhoea is present, the history is measured in months rather than years; reflecting the effects of the lymphoma rather than pre-existing enteropathy. Histologically, epitheliotropism is often prominent despite the limited or complete absence of villous atrophy and crypt hyperplasia (Figure 4). The degree of IELs is often not in keeping with the lack of crypt hyperplasia and of a quantity that exceeds that expected of Marsh stage 1 in coeliac disease. In addition, although the monoclonal IELs in coeliac disease, refractory jejunitis and Type I EATL are described as cytologically normal,27 in Type II EATL, the IELs in the peripheral zone are cytologically atypical and appear to be the result of lateral spread.3 So neoplastic lymphocytes not only invade vertically but often there is also a radial growth pattern showing intra-epithelial spread. In other words, tumour epitheliotropism and not enteropathy is the prominent feature in Type II EATL.

In summary, the central tumour in Type II EATL often displays transmural invasion and arises from IEL precursors, which have been shown to be clonally related. Neoplastic lymphocytes with cytological atypia undergo lateral intra-epithelial spread to the peripheral zone. The morphologically normal but increased IELs in the distant zone reflect the remnant precursors from which the tumour has arisen.

It is generally recognized among Asian pathologists that Type II EATL is a separate entity that should not be considered a form of EATL. One term that has been proposed is ‘Monomorphic intestinal T-cell lymphoma’.3 However, there is a spectrum of cytological appearance in both the diseases. Some cases of Type II EATL show a greater degree of nuclear pleomorphism while tumour cells in Type I EATL were in fact described as being ‘relatively monotonous’ in the 2008 WHO classification.2

We therefore propose the alternative name ‘Epitheliotropic intestinal T-cell lymphoma’ in reference to the prominent tumour epitheliotropism in the peripheral zone but the lack of association with coeliac disease. So EATL will refer to a primary intestinal T-cell lymphoma with enteropathy while epitheliotropic intestinal T-cell lymphoma is a disease with prominent epitheliotropism but without pre-existing enteropathy. As epitheliotropism is seen in both the entities, one might argue that Type I EATL should be called ‘Epitheliotropic enteropathic T-cell lymphoma’ while Type II EATL be referred to as ‘Epitheliotropic non-enteropathic T-cell lymphoma’ but that would seem rather unwieldy.

The term ‘primary epitheliotropic intestinal T-cell lymphoma’ has been used to describe an intestinal T-cell lymphoma with prominent intra-epithelial lymphocytosis in horses, cats and dogs.28, 29, 30 It is of interest that a detailed description of a case in a horse did not show features of coeliac disease or jejunitis. In addition, the duration of intestinal symptoms was short and the epitheliotropic lymphocytes displayed greater cytological atypia than the IELs in coeliac disease and classical EATL.29


  1. 1.

    , , , , , et al. Enteropathy-associated T-cell lymphoma: clinical and histological findings from the international peripheral T-cell lymphoma project. Blood 2011; 118: 148–155.

  2. 2.

    , , . Enteropathy-associated T-cell lymphoma. In: Swerdlow SHCE, Harris NL, Jaffe ES, Pileri SA, Stein H, Thiele J, Vardiman JW, (eds). WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues 4th edn. IARC: Lyon, France, 2008; pp 289–291.

  3. 3.

    , , , , , et al. Type II enteropathy-associated T-cell lymphoma: a distinct aggressive lymphoma with frequent gammadelta T-cell receptor expression. Am J Surg Pathol 2011; 35: 1557–1569.

  4. 4.

    , , , . Primary intestinal T-cell and NK-cell lymphomas: a clinicopathological and molecular study from China focused on type II enteropathy-associated T-cell lymphoma and primary intestinal NK-cell lymphoma. Mod Pathol 2011; 24: 983–992.

  5. 5.

    , , , , , et al. Genotype frequency and F ST analysis of polymorphisms in immunoregulatory genes in Chinese and Caucasian populations. Immunogenetics 2007; 59: 839–852.

  6. 6.

    , , , . Allelic and haplotypic diversity of HLA-A, -B, -C, -DRB1, and -DQB1 genes in the Korean population. Tissue Antigens 2005; 65: 437–447.

  7. 7.

    , , . Association of HLA type and Mooren’s Ulcer in Chinese in Taiwan. Br J Ophthalmol 2003; 87: 797–798.

  8. 8.

    , , , , . Allele frequencies and haplotypic associations defined by allelic DNA typing at HLA class I and class II loci in the Japanese population. Tissue Antigens 2000; 56: 522–529.

  9. 9.

    , , , . Distribution of HLA gene and haplotype frequencies in Taiwan: a comparative study among Min-nan, Hakka, Aborigines and Mainland Chinese. Tissue Antigens 1999; 53: 51–64.

  10. 10.

    , , , , , et al. Primary CD56 positive lymphomas of the gastrointestinal tract. Cancer 2001; 91: 525–533.

  11. 11.

    , , , , , et al. NK-cell lineage predicts poor survival in primary intestinal NK-cell and T-cell lymphomas. Am J Surg Pathol 2009; 33: 1230–1240.

  12. 12.

    , , , , , et al. Enteropathy-associated T-cell lymphoma--a clinicopathologic and array comparative genomic hybridization study. Hum Pathol 2010; 41: 1231–1237.

  13. 13.

    , , , , , . Primary intestinal T-cell lymphoma resembling lymphomatous polyposis: report of a case. Virchows Arch 2000; 437: 450–453.

  14. 14.

    , , , , , . Intestinal T-cell and natural killer-cell lymphomas in Taiwan with special emphasis on 2 distinct cellular types: natural killer-like cytotoxic T cell and true natural killer cell. Hum Pathol 2008; 39: 1018–1025.

  15. 15.

    , , , , , et al. Most CD56+ intestinal lymphomas are CD8+CD5-T-cell lymphomas of monomorphic small to medium size histology. Am J Pathol 1998; 153: 1483–1490.

  16. 16.

    , , , , , et al. Pathological and immunohistological findings and genetic aberrations of intestinal enteropathy-associated T cell lymphoma in Japan. Histopathology 2011; 58: 395–407.

  17. 17.

    , , , , , et al. The phenotype of intraepithelial lymphocytes in Taiwanese enteropathy-associated T-cell lymphoma is distinct from that of the West. Histopathology 2008; 53: 234–236.

  18. 18.

    , , , , , et al. Nuclear expression of MATK is a novel marker of type II enteropathy-associated T-cell lymphoma. Leukemia 25: 555–557.

  19. 19.

    , , . Ninety percent of celiac disease is being missed. J Pediatr Gastroenterol Nutr 2007; 45: 497–499.

  20. 20.

    DAKO. Histology FISH Accessory Kit. 2010 [cited 15 May 2012]; Working Procedure]. Available from .

  21. 21.

    , , , , . A simplified method of detection of clonal rearrangements of the T-cell receptor-gamma chain gene. Diagn Mol Pathol 1992; 1: 173–179.

  22. 22.

    , , , , , et al. CD8+, CD56+ (natural killer-like) T-cell lymphoma involving the small intestine with no evidence of enteropathy: clinicopathology and molecular study of five Japanese patients. Pathol Int 2008; 58: 626–634.

  23. 23.

    , , , . Intraepithelial lymphocytes: exploring the third way in immunology. Nat Immunol 2001; 2: 997–1003.

  24. 24.

    , . Intestinal T cells: facing the mucosal immune dilemma with synergy and diversity. Semin Immunol 2009; 21: 130–138.

  25. 25.

    , . The anatomy of the immune response. In: Roitt’s Essential Immunology 10th edn. Blackwell Science: London, UK, 2001; pp 155–157.

  26. 26.

    , , . International peripheral T-cell and natural killer/T-cell lymphoma study: pathology findings and clinical outcomes. J Clin Oncol 2008; 26: 4124–4130.

  27. 27.

    , , , . Mucosal intra-epithelial lymphocytes in enteropathy-associated T-cell lymphoma, ulcerative jejunitis, and refractory celiac disease constitute a neoplastic population. Blood 1999; 94: 260–264.

  28. 28.

    , , , , , . Feline epitheliotropic intestinal malignant lymphoma: 10 cases (1997–2000). J Vet Intern Med 2003; 17: 326–331.

  29. 29.

    , , , , . Primary epitheliotropic intestinal T-cell lymphoma in a horse. J Vet Diagn Invest 2002; 14: 150–152.

  30. 30.

    , , . Primary epitheliotropic intestinal T-cell lymphoma as a cause of diarrhoea in a horse. Can Vet J 2010; 51: 522–524.

Download references


This study was funded by the Programme Project Grant and the CISSP programme (2010/022) from the National Medical Research Council of Singapore, the Singhealth Foundation Project Grant (SHF/FG438P/2010), HSBC Trustee (Singapore) Limited as trustees of the Major John Long Trust Fund and the Chew Woon Poh Trust Fund.

Author information


  1. Department of Pathology, Singapore General Hospital, Singapore, Singapore

    • S-Y Tan
    •  & L Tan
  2. Department of Pathology, University Malaya, Kuala Lumpur, Malaysia

    • S-Y Tan
    •  & P-L Cheah
  3. Singhealth Tissue Repository, Singapore Health Services, Singapore, Singapore

    • S-Y Tan
    •  & M Koh
  4. Department of Pathology, Chi-Mei Medical Centre, Tainan, Taiwan

    • S-S Chuang
  5. Department of Pathology, Taipei Medical University, Taipei, Taiwan

    • S-S Chuang
  6. Department of Pathology, National Taiwan University Hospital, Taipei, Taiwan

    • S-S Chuang
  7. Department of Medical Oncology, National Cancer Centre, Singapore, Singapore

    • T Tang
    • , K Tay
    •  & S-T Lim
  8. Department of Pathology, Samsung Medical Center, Sungkyunkwan University, Seoul, Korea

    • Y-H Ko
  9. Department of Pathology, Tan Tock Seng Hospital, Singapore, Singapore

    • K-L Chuah
  10. Department of Pathology, National University of Singapore, Singapore, Singapore

    • S-B Ng
  11. Department of Haematology-Oncology, National University Cancer Institute of Singapore, National University Health System, Singapore, Singapore

    • W-J Chng
  12. Cancer Science Institute of Singapore, National University of Singapore, Singapore, Singapore

    • W-J Chng
  13. Nuffield Department of Clinical Laboratory Sciences, Oxford University, Oxford, UK

    • K Gatter
  14. Department of Pathology, Queen Mary Hospital, Hong Kong SAR, China

    • F Loong
  15. Department of Pathology, First Guangzhou People’s Hospital, Guangzhou, China

    • Y-H Liu
  16. Eastern Health Pathology, Melbourne, Victoria, Australia

    • P Hosking
  17. NCCS-VARI Translational Research Laboratory, National Cancer Center, Singapore, Singapore

    • B-T Teh
  18. Laboratory of Cancer Therapeutics, Duke-NUS Graduate Medical School, Singapore, Singapore

    • B-T Teh
  19. Institute of Cell and Molecular Biology, A*STAR, Singapore, Singapore

    • S-Y Tan


  1. Search for S-Y Tan in:

  2. Search for S-S Chuang in:

  3. Search for T Tang in:

  4. Search for L Tan in:

  5. Search for Y-H Ko in:

  6. Search for K-L Chuah in:

  7. Search for S-B Ng in:

  8. Search for W-J Chng in:

  9. Search for K Gatter in:

  10. Search for F Loong in:

  11. Search for Y-H Liu in:

  12. Search for P Hosking in:

  13. Search for P-L Cheah in:

  14. Search for B-T Teh in:

  15. Search for K Tay in:

  16. Search for M Koh in:

  17. Search for S-T Lim in:

Competing interests

The authors declare no conflict of interest.

Corresponding author

Correspondence to S-Y Tan.

Supplementary information

About this article

Publication history







Author Contributions

SYT, LT and STL designed research, performed research, contributed bioresources and data, analysed data and wrote the paper. SSC, TT, YHK, KLC, SBN, WJC, KG, FL, PH, YHL, PLC, BTT, KT and MK performed research, contributed bioresources and research data and analysed data.

Supplementary Information accompanies this paper on the Leukemia website (http://www.nature.com/leu)

Further reading