An update on genetic aberrations in T-cell neoplasms

A disease category that includes four different entities: (1) systemic EBV-positive T-cell lymphoma of childhood; (2) chronic active EBV disease; (3) hydroa vacciniforme lymphoproliferative disorder; and (4) severe mosquito bite allergy.^, Additionally, EBV-positive haemophagocytic lymphohistiocytosis, a benign disease which can induce major illness, can be discussed with this category of diseases.

EBV-positive haemophagocytic lymphohistiocytosis (EBV-positive HLH) is not a neoplasm per se, but rather an EBV-driven cytokine storm found in a small subset of children who have immune defects in EBV immunosurveillance.^, Typically, EBV infects B-cells in most people, but in EBV-positive HLH, T-cells are infected by the virus. A number of known inherited genetic mutations can lead to abnormal immune response and EBV-positive HLH (familial HLH), but some cases appear to be without such aberrations (non-familial).^, PCR-based clonality assessment can show T-cell clones in EBV-positive HLH.

Systemic EBV+ T-cell lymphoma of childhood (sEBV-TNHL) is the most deadly of the aformentioned diseases but can be difficult to distinguish from benign EBV-positive HLH as both can be associated with haemophagocytic histiocytosis. T-cells harbouring EBV have virally-expressed LMP1 protein and an over activated NF-KB pathway increasing cell survival and proliferation.^, sEBV-TNHL also often has somatic chromosomal abnormalities seen on karyotype, whereas EBV-positive HLH does not. Although a marker of sEBV+TNHL disease, karyotypic abnormalities are not particularly sensitive as sEBV-TNHL can be without chromosomal abnormality. Although no specific chromosomal abnormality is seen in all, or even most cases of sEBV+TNHL, some commonalities are present: cases with alterations typically have a complex karyotype (three or more chromosomal alterations) and almost 40% of cases in the literature have add(9)(p24); additionally, alterations of chromosome 1, 7, 11, 17, 20, 21 and X were seen in over 20% of reported cases.^{,  ,  ,  ,  ,} 

Chronic active EBV disease (CAEBV) is an uncontrolled replication of lymphoid cells which can be neoplastic or reactive. CAEBV can be polyclonal, oligoclonal or monoclonal with a proportion of monoclonal cases being neoplastic. In this way, CAEBV is the bridge between the always non-neoplastic EBV-positive HLH and the always neoplastic and malignant sEBV+TNHL. CAEBV can be systemic (CAEBV of T and NK-cell type, systemic form; sCAEBV) or cutaneous as in hydroa vacciniforme lymphoproliferative disorder (HV-LPD) and severe mosquito bite allergy (SMBA). Few studies show CAEBV shares similar somatic mutations (e.g., DDX3X and KMT2D) alterations as seen in T-cell and NK-cell lymphomas indicating this as a premalignant condition. sCAEBV patients have infectious mononucleosis symptoms (fever, fatigue, haepatosplenomegaly, skin rashes, and lymphadenopathy) that do not resolve after three months. A proportion of these cases have HLH, which may transform to the more deadly systemic EBV+ T-cell lymphoma. sCAEBV patients do typically produce IgG to EBV; however, this immunoglobulin response is unable to keep the EBV in check, and peripheral blood will find >10^2.5 copies/mg EBV DNA.

sCAEBV is characterised by T-cells or NK-cells expression of viral genes EBNA1, LMP1 and LMP2 with resulting downstream oncogenic pathways activated and tumour suppressors inhibited. Somatic mutations in DDX3X, BCOR/BCORL2 and TET2 were seen in 20% of CAEBV in one study.^, Interestingly, mutations in DDX3X and BCOR are also seen in extra-nodal NK/T-cell lymphoma of nasal type (ENKTL, discussed later). Chromosome alterations associated with CAEBV include deletion of 6q or gain of 6p in HV-LPD. Similarly, chromosome 6q losses are also seen in ENKTL suggesting commonalities in the molecular pathogenesis of these two diseases.^,

A very rare form of peripheral T-cell lymphoma that most often occurs in elderly patients. At present, these neoplasms are currently classified under the rubric peripheral T-cell lymphoma, not otherwise specified. However, data are abundant supporting this idea that these neoplasms will become a separate entity. Primary EBV positive nodal T-/NK-cell lymphoma shares an immunophenotype with ENKTL. Although primary EBV-positive nodal T-/NK-cell lymphoma can extend extranodally, it typically does not involve the skin, gastrointestinal system or nose, unlike ENKTL. One study of primary EBV-positive T/NK-cell lymphoma showed recurrent copy number aberrations in approximately 20% of cases, including loss of chr14q11.2 (100%), chr3q26.1 (67%), and chr22q11.23 (33%).

ENKTL is a lymphoma of predominantly NK-cell origin although some cases can originate from cytotoxic T-cells. This neoplasm was designated previously with the qualifier ‘nasal type’ but this designation is deleted in the new WHO classification. The disease has a variable prognosis when found only near the nasal passages, but if outside of this region, ENKTL is an aggressive lymphoma. ENKTL is associated with EBV in virtually all cases which likely indicates a pathogenic role. The genetics of ENKTL are somewhat similar to some EBV-positive T-cell and NK-cell lymphoproliferative diseases of childhood, although ENKTL occurs almost exclusively in adults. ENKTL is another disease driven by EBV infection which, intriguingly, can show a 30 base-pair deletion in the viral gene LMP1 that seems to be associated with an increased risk of lymphomagenesis. A recent study by Xiong et al. identified at least three distinct genetic subtypes of ENKTL. The TSIM subtype is associated with JAK/STAT pathway activation, NK cell origin, TP53 mutation, genomic instability including deletion of 6q21 and amplifications of 9p24.1 and/or 17q21.2, as well as PD-L1/2 overexpression.^, JAK/STAT activation can occur by a variety of mechanisms, including amplifications of chromosome 9p24 (JAK2) and 17q21 (STAT3, STAT5B), and mutations of STAT3, JAK3, and JAK1.^, The HEA subtype is associated with more epigenetic changes through mutations in HDAC9, EP300 and ARID1A, NF-KB activation, T-cell origin, and T-cell receptor signalling pathway activation. Lastly, The MB subtype is associated with MYC over expression and a poor prognosis with a 3-year overall survival rate of approximately 38.5%. In contrast, the 3-year OS rate for the TSIM and HEA subtypes is 79.1% and 91.7%, respectively.

Cytogenetic information on ENKTL is limited and varied but many cases harbour chromosomal alterations, most commonly 6q deletions (50%), focal 1q gain (50%) and 17p losses (40%); the 6q21–q25 region of loss includes PRDM1, FOXO3, ATG5, AIM1, HACE1 and PTPRK and the loss of 17p11–p13 includes the TP53 gene. Somatic mutations are most commonly seen in MLL2, TP53, DDX3X, BCOR, JAK3, and STAT3. The most common fusion detected in ENKTL is CTLA4::CD28, reported in approximately 29% of cases.

ATLL is a T-cell neoplasm caused by HTLV-1 which can present clinically in four different ways: the acute leukaemic form, the lymphomatous variant, a chronic form, and a smoldering variant. Genetically, all forms have HTLV-1 DNA integrated into human DNA, which is not seen in carriers of HTLV-1 without disease. The viral gene TAX acts as an oncogene by inactivating Rb, P53 and activating NF-KB. The viral gene HBZ is also important in oncogenesis (Fig. 1).

Recently, others using whole genome sequencing have shown that over 30% of ATLL cases have mutations in the gene CIC, most well known for its association with low grade gliomas. CIC has two isoforms: a long form and a short form. In low grade gliomas, CIC mutations tend to be in the shared exons of both isoforms, but in ATLL, mutations occur in the exons of the long isoform only. Copy number alterations or mutations of ATXN1 are also common. Physiologically ATXN1 and CIC interact to create a complex that represses transcription. As 53% of ATLL cases have either ATXN1 or CIC alterations, it is likely that dysregulation of this complex is a critical driver of ATLL in addition to viral infection. Mutations in CCR4 (C-C chemokine receptor 4) are also common, and most patients with ATLL show overexpression of CCR4 which is associated with skin involvement and worse prognosis. However, a study by Ishida et al. showed potential therapeutic application of mogamulizumab, a monoclonal antibody targeting CCR4. Other genes commonly mutated in ATLL not discussed above include: PLCG1 (>30%), PRKCB (>30%), CARD11 (∼30%), and STAT3 (∼25%). Recurrent fusions between CD28 and either ICOS or CTLA4 were also found in a subset of ATLL cases. TP53 can be mutated in ∼16% of cases and tends to be mutated in more aggressive ATLL where there has been loss of TAX expression.

Copy number alterations are not uncommon in ATLL. This includes 9p24 amplification leading to PDL1 amplification, seen in 10–20% of cases, which is more common in the aggressive subtype, but it is associated with poor prognosis independent of subtype. Other copy alterations include 9p21 deletion (CDKN2A, 20–30%), 13q32 deletion (20–30%), 6p22 deletion (ATXN1, 10–20%) and deletion of 6q21 (PRDM1, 10–20%).^, Table 2 summarises the molecular features of virally driven mature T-cell lymphomas or lymphoproliferative disorders.

EATL is a coeliac disease-linked lymphoma that occurs in the small intestine, typically the jejunum or ileum. About 90% of patients with coeliac disease have inherited the HLA isoforms DQ2 or DQ8, which are thought to contribute directly to the development of gluten intolerance. Patients with EATL typically have a preceding diagnosis of coeliac disease and EATL, often preceded by so-called ‘refractory coeliac disease’ where patients continue to have symptoms and histological crypt hyperplasia and villous atrophy with increased lymphocytes despite a gluten-free diet. Refractory coeliac disease is stratified into two types: type I (polyclonal T-cell lymphocytosis) and type II (clonal T-cell lymphocytosis); type II refractory coeliac disease (RCD) is about 5-fold more likely to transform into EATL when compared with type I (Fig. 2).

As type I RCD is not clonal, somatic genetic abnormalities are not expected nor identified. Type II RCD, on the other hand, demonstrates many genetic abnormalities as are seen in EATL; in one study, JAK1/STAT3 pathway mutations were seen in 80% of RCDII and 90% of EATL – with over 50% of all cases harbouring gain-of-function mutations in a single codon (1097) of JAK1. Gains of chromosome 1q are seen in both RCD type II and EATL, with additional gains in chromosomes 5q, 9q (focal), and 7q and losses of chromosomes 8p, 9p, 13q, and 16q (focal) identified in EATL, but not in RCD type II.^, Segmental amplification of 9q31.3-qter which includes NOTCH1 and ABCL1 is the most common abnormality seen in EATL, occurring in 70–80% of cases.^{,  ,}  EATL also shows losses of 9p21 (CKDN2A/2B) and TP53 in approximately 50% and 25% of cases, respectively, which are uncommon in RCD type II. The genetic differences between de novo EATL and RCD type II-progressed EATL are minimal. A study by Nicolae et al. found that genes involved in the JAK/STAT pathway were mutated in approximately 50% of patients with EATL, involving mutations in STAT3 (∼20%), STAT5B (∼10%), JAK1 (∼20%), and JAK3 (∼10%).

MEITL is another type of lymphoma arising in the small intestine, primarily the jejunum and ileum.^, Unlike EATL, MEITL is not associated with coeliac disease or DQ2/DQ8 HLA isoforms. Genetically, MEITL is also different from EATL. Although some chromosomal alterations, such as gains in 9q, are also seen in MEITL, the MYC locus (8q24) is much more likely to be amplified in MEITL than EATL, and chromosome 1q and 5q gains are much less common. Loss-of-function mutations in SETD2 are seen in over 90% of MEITL cases (and are quite rare in EATL).^, Mutations affecting the JAK/STAT pathway are common in MEITL, present in ∼80% of cases and involving STAT5B (∼60%), JAK3 (∼46%), SH2B3 (∼20%) and JAK1 in a small subset.^, Less frequent mutations are seen in STAT3, TYK2, and SOCS1; as in most diseases, STAT5B and STAT3 mutations appear to be mutually exclusive. Mutations affecting the MAPK pathway are also common in MEITL and involve TP53 (∼33%), BRAF (∼26%), KRAS (∼20%) and NRAS (∼10%), which appear to be mutually exclusive. CREBBP mutations have also been described; however, they appear to occur exclusively in tandem with STAT5B and/or JAK3 mutations.^, Overall, SETD2, STAT5B, JAK3, BRAF and KRAS mutations appear to be more common in MEITL than in EATL.

Mycosis fungoides (MF) accounts for over 50% of all cutaneous lymphomas. Classically, patients present with skin patches that progress to plaques or tumour stage over the course of many years. The lymphomatous T-cells show epidermotropism and a proclivity for infiltrating the basal layer of the skin/basement membrane. MF has an ultra-violet light mutational signature, similar to epithelial and melanocytic skin cancers that arise as a result of unprotected sun exposure. In a UV light signature, cytosine to thymine changes are seen throughout the genome and suggest that UV light may play a role in disease pathogenesis; however, UV therapy is sometimes used in the treatment of MF and the temporal relationship between UV therapy and molecular sequencing has been underexplored.^,

Although initially described as the leukaemic presentation or leukaemic transformation of MF, there is some evidence suggesting SS may be a distinct disease entity, at least in some cases.^{,  ,}  A study by Wang et al. showed that Sézary syndrome is characterised by mutations and loss-of-function of TP53, as well as mutations in CCR4, PLCG1, FAS and in CARD11 in over 10% of cases. Homozygous loss of CDKN2A/B is also common (58% cases). Fusions between CTLA4 and CD28 have been described as well in SS.^, Others have described the genetics for mycosis fungoides and Sézary syndrome together, which adds complexity to the interpretation of the data. Larocca and Kupper described the following recurrent alterations in MF/SS: single nucleotide variants in MLL3, TP53, CARD11, ARID1A, STAT5B, ZEB1, FAS, PLCG1, and CDKN2A and copy number variants in TP53, ZEB1, STAT5B, ARID1A, CDKN2A, FAS, DNMT3A, ATM, PRKCQ, and TNFAIP3. Recurrent mutations in STAT3, JAK3, TNFR1F1B, and RHOA have also been noted in SS/MF.^, Genetic sequencing examining mycosis fungoides and Sézary syndrome separately, before and after UV light therapy is needed to better identify the genetic underpinnings and molecular biology of these diseases.

SPTCL is a lymphoma that involves subcutaneous adipose tissue.^, Adipocyte rimming by lymphoma cells is a key characteristic histological finding in SPTCL. Clinically, many patients have autoimmune disorders, especially systemic lupus erythematosus, and haemophagocytic syndrome may be a complication of SPTCL. In 2019, others reported that 11/13 studied patients with SPTCL had germline mutations in the gene HAVCR2. This report confirmed an earlier study that found that 60% of SPTCL had germline HAVCR2 mutations. HAVCR2 encodes TIM-3 a protein expressed on T-cells that regulates immune cell function; loss of TIM-3 is associated with SPTCL, immune over activation and HLH and provides a unifying molecular basis for the clinical features of this disease. Recurring somatic mutations were not impressively identified in SPTCL, although there may be some increase in PI3K/AKT/mTOR pathway alterations (Li, Lu et al. 2018). Due to the association with immune over activation from TIM-3 loss, immune suppression is the first line treatment for all SPTCL patients. SPTCL with HAVCR2 mutations are more likely to have systemic illness including haemophagocytic lymphohistiocytosis, present at a younger age and have worse relapse-free survival than HAVCR2 wild-type SPTCL.

Two primary cutaneous CD30-positive T-cell lymphoproliferative disorders exist which are thought to be two ends of a spectrum of disease: (1) lymphomatoid papulosis (LyP), and (2) primary cutaneous anaplastic large cell lymphoma (C-ALCL, Fig. 3A–D). A third group of borderline lesions with features between LyP and C-ALCL also have been described. LyP lesions are small, often multifocal, and clinically self-resolve but can, and often do, return. C-ALCL may follow LyP in some patients; it is a larger lesion (>2 cm), often solitary, that is less likely to resolve or regress. When these two lesions cannot be distinguished from each other the term borderline lesion may be used. By definition, these neoplasms do not carry ALK fusions/rearrangements. DUSP22::IRF4 is seen in a small proportion of LyP cases and more frequently in C-ALCL (20–57%). Less commonly, (4–15%) NPM1::TYK2 fusions can be seen in primary cutaneous CD30-positive T-cell lymphoproliferative disorders, and a small subset of C-ALCL have TP63 rearrangement.^{,,  ,  ,} 

The other rare primary cutaneous T-cell lymphomas/lymphoproliferative disorders which are CD30 negative include: (1) primary cutaneous gamma delta T-cell lymphoma (PCGD-TCL); (2) primary cutaneous CD8-positive aggressive epidermotropic cytotoxic T-cell lymphoma (PC8AE-TCL); (3) primary cutaneous acral CD8-positive T-cell lymphoproliferative disorder(PCA8-TCL), and (4) primary cutaneous CD4+ small/medium T-cell lymphoproliferative disorder (PC4-LPD). Given their rarity, large studies examining the genetic landscape of each entity are limited. PCGD-TCL cases appear to have KRAS, NRAS, MAPK, MYC, MYCN, JAK3, STAT3 or STAT5B missense mutations as well as loss of ARID1A, CDKN2A, FBXW7, SOCS1, TP53 and FAS. PC8AE-TCL cases have been described with multiple chromosomal gains and losses, suggestive of chromosomal instability, a complex karyotype and TP53 loss.^,

HTCL is an aggressive lymphoma with a unique genetic signature that presents in the liver, bone marrow, and spleen, without lymphadenopathy. Chromosomally HTCL is characterised by isochromosome 7q (25–70%). Other copy number alterations that can occur include trisomy 8 (8–53%), gains in chromosome 7q, 8q, or 1q and 10p loss.^, Mutations in the tumour suppressor gene SETD2 (∼70%) and STAT5B (∼30%) are the most common gene alterations. Although SETD2 and STAT5B mutations can also occur in MEITL, the i(7q) commonly seen HSTCL is absent in MEITL. Other less common mutations found in HTCL include those involving chromatin modifying and DNA methylation genes such as INO80, ARID1B, TET2, TET3, SMARCA2, and DNMT3A. Mutations in STAT3 (9%), PIK3CD (9%), TP53 (9%), IDH2 (6%), and UBR5 (9%) have also been reported (Fig. 4).

BIALCL is an uncommon type of ALCL that represents <1% of all non-Hodgkin lymphomas and approximately 2% of all extranodal lymphomas. BIALCL arises in the fibrous capsule surrounding textured breast implants. There is a higher incidence of germline BRCA1 and BRCA2 mutations in patients who develop BIALCL. Several studies have found somatic mutations in JAK1, STAT3, STAT5, or SOCS1, suggestive of a JAK/STAT pathway driven pathogenesis.^{,,  ,}  Mutations in DNMT3A and TP53 also have been described. Unlike systemic ALCL, rearrangements in ALK, DUSP22, and TP63 have not been identified in BIALCL. However, a STAT3::JAK2 fusion has been reported in a small case series. Cytogenetically, chromosome 20q loss has been reported in up to 66% of cases. Table 4 summarises the molecular features of primarily extra-nodal mature T-cell lymphomas.