Update on the Classification of and Diagnostic Approaches to Mature T-Cell Lymphomas

T follicular helper cell lymphoma (T-FHCL) is identified by the antigen markers expressed by the neoplastic cells, including CD10, BCL6, PD1, CXCL13, CXCR5, ICOS, and SAP. To define the TFH phenotype, expression of minimally 2 and ideally 3 or more of the above-listed markers is needed, in addition to CD4 and certain T-cell lineage–specific antigens.³ 

The current WHO classification divides T-FHCL into 3 categories² : angioimmunoblastic T-cell lymphoma (AITL), follicular T-cell lymphoma, and nodal PTCL with TFH phenotype. All 3 categories of T-FHCL have similar clinical manifestations.^4,5  The distinction among the 3 categories is based on histopathologic features, including the distinction of morphologic variants within each category. Of note, the 3 categories of T-FHCL may have overlapping histopathologic features in concurrent biopsies of different lymph nodes or sequential biopsies in the same patients, and thus may represent morphologic variations of the same disease.^5,6 Table 1 shows the diagnostic features distinguished among 3 categories of T-FHCL and their distinction from PTCL, not otherwise specified (NOS).

Angioimmunoblastic T-cell lymphoma is characterized by polymorphic cell infiltrate, increased high endothelial venules, and prominent follicular dendritic meshworks that can be highlighted by CD21/CD23 stains and are often expanded beyond lymphoid follicles; eosinophils are commonly increased, and neoplastic T cells are typically medium in size with clear cytoplasm and frequently form aggregates. Three morphologic variants, including perifollicular, interfollicular, and diffuse growth patterns, can be recognized⁷  (see Supplemental Figure S1 in the supplemental digital content containing 4 figures at https://meridian.allenpress.com/aplm in the August 2022 table of contents). The 3 patterns may represent chronologic manifestations or consecutive progression of the same disease, with perifollicular and interfollicular variants representing relatively early-phase AITL.⁸ 

In follicular T-cell lymphoma, nodal architecture is completely or partially effaced by nodular (follicular) proliferation of medium-sized lymphocytes with a moderate amount of clear cytoplasm. Immunohistochemical analysis often shows an inverted pattern of B- and T-cell distribution, with T cells in the center of lymphoid follicles surrounded by B cells.^5,6  CD21 often highlights well-formed follicular dendritic meshworks underneath the lymphoid follicles, as seen in follicular lymphoma. Two morphologic variants are recognized: follicular growth pattern and progressive transformation of germinal center (PTGC)–like growth pattern (see Supplemental Figure S2).⁵  The PTGC–like growth pattern, particularly when a low number of neoplastic T cells are present, is easy to miss without careful examination to confirm the morphologic atypia and phenotypic aberrancy of the neoplastic T cells. Cases with TFH phenotype that do not fit into either AITL or follicular T-cell lymphoma would be classified as nodal PTCL with TFH phenotype as a provisional entity.² 

All 3 categories of T-FHCL are frequently associated with latent EBV infection. This bystander infection could lead to clonal expansion of an EBV-positive B-cell population, resulting in EBV-positive B-cell lymphoproliferative disorder.⁹  Presence of Hodgkin/Reed-Sternberg–like cells, either EBV positive or EBV negative, frequently raises the differential diagnosis of classic Hodgkin lymphoma.^10,11 

The diagnosis of T-FHCL often needs to be confirmed by demonstrating T-cell clonality. Ninety percent of AITLs have their T-cell clone detected by Biomed TCRG gene rearrangement and/or TCRB gene rearrangement analysis.^12,13  Of note, in the perifollicular pattern of AITL and PTGC-like growth pattern of follicular T-cell lymphoma, T-cell clones may be missed because of low levels of neoplastic T cells. Angioimmunoblastic T-cell lymphoma often shares a signature mutation landscape with the other 2 categories of T-FHCL, including mutations in TET2, DNMT3A, and RHOA genes^14,15 ; IDH2 mutation is relatively more frequent in AITL. This signature mutation profile can be used not only to confirm T-cell clonality, but also to distinguish from PTCL, NOS, which shows a more heterogeneous and different mutation landscape.¹⁶ TET2 and DNMT3A mutations are associated with advanced stage and short progression-free survival in AITL.¹⁴  In addition, specific mutation may become a useful marker for detecting minimal residual disease.

ALK-negative ALCL, defined as a CD30⁺ PTCL lacking expression of ALK protein but otherwise morphologically indistinguishable from ALCL, ALK positive, has been recognized as a distinct entity.²  Generally affecting older adults, with a median age of 58 years and the highest incidence in the sixth decade, ALK-negative ALCL is a genetically heterogeneous disease with widely disparate clinical outcomes. The 5-year overall survival varies significantly from 17% to 90% depending on the specific gene rearrangement status.¹⁷ 

Most ALK-negative ALCLs show at least some suggestion of T-cell phenotype and partial to complete loss of one or more T-cell antigens, most commonly with absent expression of CD3, CD5, or TCR. Many cases express the cytotoxic markers TIA1, granzyme B, and/or perforin. Most cases show clonal rearrangement of TCR genes, whether or not they express T-cell antigens.¹⁷ 

The 2017 revised WHO classification²  included the distinct gene rearrangements associated with a fraction of ALK-negative ALCLs. Approximately 30% of the cases harbor DUSP22 rearrangements and are associated with an up to 90% 5-year survival rate and favorable prognosis, similar to that seen in ALK-positive ALCL. Cases with DUSP22 rearrangements characteristically show a sheetlike growth pattern, “doughnut” cells by morphology, and less pleomorphism of tumor cells.¹⁸  The neoplastic cells typically lack cytotoxic marker expression.¹⁸  In addition, approximately 8% of the cases have TP63 gene rearrangements, which are associated with the worst prognosis and a 17% 5-year survival rate. These rearrangements are essentially mutually exclusive but rarely can occur concurrently.¹⁹  For prognostic stratification, it is recommended that all ALK-negative ALCLs undergo fluorescence in situ hybridization testing for rearrangements involving DUSP22 (DUSP22-IRF4 locus on 6p25.3) or TP63 (most commonly occurring as TBL1XR1/TP63 fusion [inv(3)(q26q28)]). The so-called triple-negative ALCL lacking rearrangements of ALK, DUSP22, and TP63 has an intermediate prognosis and 5-year survival rate of 42%.¹⁷ 

Many GEP studies have been performed to elucidate ALK-negative ALCL biology, define the borders with other PTCL subtypes, and provide new genomic classifiers for stratification of patients. Agnelli et al²⁰  found TNFRSF8, BATF3, and TMOD1 genes exhibited significantly higher expression in ALK-negative ALCL compared with PTCL, NOS. A 3-gene model was validated to separate ALK-negative ALCL from PTCL, NOS, with 97% accuracy.²⁰  Cases with DUSP22 rearrangement lacked expression of genes associated with JAK1/STAT3 pathway activation but showed DNA hypomethylation, high expression of cancer-testis antigens, CD58 and HLA class II, and minimally expressed PD-L1.²¹  A microRNA expression profiling study²²  demonstrated an 11-microRNA signature that distinguished ALK-negative ALCL from PTCL, NOS, with 90% probability. Inactivation of TP53 and/or PRDM1²³  and activating mutations of JAK1 and/or STAT3 genes are frequently seen in ALK-negative ALCL. Recurrent chimeras combining a transcription factor (NFkB2 or NCOR2) with a tyrosine kinase (ROS1 or TYK2) were also discovered in wild-type JAK1/STAT3 ALK-negative ALCL.²⁴  Interestingly, constitutive STAT3 phosphorylation was observed in 43% of ALK-negative ALCLs, 76% of which had nonmutated JAK1/STAT3, suggesting alternative mechanisms of pathway activation.²⁴  In addition, oncogenic truncated Erb-B2 receptor tyrosine kinase 4 (ERBB4) was reported to be ectopically expressed in 24% of ALK-negative ALCL, but not in ALK-positive ALCL or PTCL, NOS. Preliminary data suggest ERBB4 expression may be mutually exclusive with rearrangements of other genes such as TP63, DUSP22/IRF4, ROS, and TYK2, indicative of the presence of an ERBB4-positive subgroup.²⁵  The diagnostic utility and clinical significance of these genomic abnormalities have to be further studied in a large panel of patients.

Breast implant–associated ALCL (BIA-ALCL) is a rare form of slow-growing PTCL that develops around implants used for breast augmentation or reconstruction. It is recognized as a provisional entity by the 2017 WHO classification²  with morphologic and immunophenotypic features indistinguishable from those of ALK-negative ALCL (see Supplemental Figure S3). All reported patients with BIA-ALCL to date are female, with a median age of 52 years (range, 28–87 years). Most patients present with a unilateral peri-implant effusion (seroma) and less frequently with a mass or hardening adjacent to the implant that involves breast parenchyma or soft tissue and/or regional lymph node. The median time from implant placement to diagnosis is 8 years, with a range of 1 to 30 years. Several studies reported a higher risk of disease development with textured implants compared with smooth, nontextured-surface implants.²⁶  In most cases, explantation of the implant with a complete capsulectomy removing all disease, without chemotherapy, is considered to be curative and confers an excellent event-free and overall survival.²⁷ 

Fine-needle aspiration of peri-implant effusion can be submitted for cytology, cell block, flow cytometry study, and molecular testing. If a mass lesion is present, biopsy of the mass is the preferred diagnostic approach. Fixation and mapping of the capsulectomy specimen to select multiple representative sections are advised to assess for microscopic tumor involvement and capsular invasion.^28,29  Based on the status of capsular involvement by tumor cells, the pathologic tumor stages are defined as T0 to T4.²⁹  Tumors extending beyond the capsule and regional lymph nodes and/or incomplete capsulectomy are associated with worse prognosis.³⁰ 

Although these cases are rarely tested, the majority of tested cases show a complex karyotype and carry monoclonal TCRG and/or TCRB rearrangements. No ALK, DUSP22, or TP63 gene rearrangements are identified in BIA-ALCL.^31,32  Recent studies^32,33  have reported frequent activating mutations in the JAK1 and STAT3 pathways and loss of function alterations of epigenetic modifier. Gene expression profiling and RNA-seq studies provide further genetic justification to recognize BIA-ALCL as a separate disease entity.^34–36 

Enteropathy-associated T-cell lymphoma (EATL), formerly known as type I EATL, is strongly associated with celiac disease, occurs in populations of northern European origin, and shows polymorphic cellular composition, whereas monomorphic epitheliotropic intestinal T-cell lymphoma (MEITL), formerly known as type II EATL, shows no association with celiac disease, occurs predominantly in Asian and Hispanic populations, and is monomorphic in cytology.²  EATL and MEITL differ in terms of epidemiologic, clinical, histologic, immunophenotypic, and molecular genetic findings (see Table 2; Supplemental Figure S4).^37,38  Transcriptome studies³⁹  have demonstrated distinct expression signatures that separate EATL and MEITL, supporting the concept that the 2 lymphoma types are different entities.

EATLs commonly lack surface CD3 and T-cell receptor expression.⁴⁰  T-cell receptor αβ can be expressed in approximately 25% of EATL cases, whereas T-cell receptor γδ is seen only in rare cases.⁴¹  In contrast, the tumor cells in MEITL more commonly express T-cell receptor γδ.⁴²  Immunophenotypic variation is not unusual in both lymphomas, and deviation in 1 or 2 markers is acceptable if other features are compatible. Of note, immunostaining of CD30 may be strong and uniform in the lymphoma cells in some EATL cases, making differentiation from ALCL challenging.⁴³ 

EATL and MEITL show some overlapping molecular genetic features. However, there are differences, including the gains of 1q and 5q more commonly found in EATL and the gains in 8q (MYC locus) more commonly found in MEITL. The mutational landscape is largely similar between EATL and MEITL. SETD2 is the most frequently mutated gene in EATL and MEITL, with different occurrence rates (approximately 15% versus up to 90%).^44,45  In addition, the JAK-STAT signaling pathways are the most frequently mutated pathways, with frequent mutations in STAT5B as well as JAK1, JAK3, STAT3, SOCS1, GNAI2, SH2B3, and other genes.^44–46  Mutations in JAK3, STAT5B, and SH2B3 appear to be more frequent in MEITL compared with EATL. Furthermore, mutation in G protein alpha inhibiting activity polypeptide 2 (GNAI2) has been reported in up to 24% of MEITL cases, but not in EATL.⁴⁶ 

Indolent T-cell lymphoproliferative disorder of the gastrointestinal tract is a clonal T lymphoid cell proliferation infiltrating the lamina propria that can involve all gastrointestinal sites, more commonly small intestine and colon. The disease has an indolent clinical course, is persistent for years to decades, and lacks response to conventional chemotherapy. Disease progression and transformation are rare. The most appropriate management approach is watchful waiting.

Published cases show heterogeneous pathologic and molecular features.^47,48  Generally, the disease is characterized by nondestructive expansion of small to medium-sized lymphoid cells with minimal cytologic atypia within the lamina propria of the mucosa. Focal infiltration in muscularis mucosae and submucosa can be seen. Immunophenotypically, the T cells express pan–T-cell markers (CD2, CD3, CD5, and CD7), but a subset of cases can show loss of CD5 or CD7. Most of the cases are CD4⁺ (approximately two-thirds) or CD8⁺ (nearly one-third). Rare cases show CD4 and CD8 dual positivity or negativity.⁴⁷  CD56 is negative and T-cell receptor αβ is positive. The Ki-67 proliferation index is low (<5%–10%).^38,41,47 

All cases show clonal T cell gene rearrangement, which is very helpful to support the diagnosis. Recurrent STAT3-JAK2 fusions have been reported in some CD4⁺ cases,^47,49  whereas activating STAT3 mutations were not detected.⁴⁸  Single-nucleotide polymorphism analyses have not demonstrated recurrent abnormalities, but rather some diverse chromosomal abnormalities.⁵⁰ 

Peripheral T-cell lymphoma, not otherwise specified (PTCL, NOS), is the most common mature T-cell lymphoma in North America and Europe, accounting for approximately one-third of all T-cell lymphoma cases. The category of PTCL, NOS, is considered a diagnostic wastebasket because it is a biologically and clinically heterogeneous disease and does not fit any specific WHO²  entities within T-cell lymphoproliferative disorders. Correspondingly, PTCL, NOS, demonstrates the most diverse morphologic and immunophenotypic characteristics with nodal and/or extranodal involvement. The lymphoma cells usually express CD3 and CD2 with down-regulation of CD5 and/or CD7. The majority of PTCL, NOS cases are CD4⁺/CD8⁻, although cases with CD4⁻/CD8⁺, CD4⁺/CD8⁺, and CD4⁻/CD8⁻ are sometimes seen. In addition to CD4, expression of a single TFH marker is allowed.

Gene expression profiling studies reveal that GATA3 expression is associated with inferior outcome of PTCL, NOS,⁵¹  and 2 mutually exclusive PTCL, NOS subgroups with significant difference in cell of origin and prognosis have recently been identified⁵² : one group demonstrates high expression of TBX21 (PTCL-TBX21), whereas the other group shows overexpression of GATA3 (PTCL-GATA3). TBX21 is a transcriptional master regulator of Th1 differentiation and GATA3 is a transcriptional master regulator of Th2 differentiation. The PTCL-TBX21 group constitutes approximately 50% PTCL, NOS, and shows Th1 phenotype with a gene signature enriched with interferon γ and NF-kB pathways. A subset of PTCL-TBX21 is positive for CD8 with overexpression of cytotoxic T-cell–related genes. The majority of PTCL-TBX21 cases show polymorphous appearance with abundant inflammatory cell infiltrates admixed with fewer tumor cells. The PTCL-GATA3 group accounts for approximately one-third of PTCL, NOS, and demonstrates a Th2 phenotype with a gene signature associated with MYC and PI3K pathways. Most PTCL-GATA3 cases comprise numerous large tumor cells with fewer background inflammatory cells. Prognostically, PTCL-TBX21 has a 5-year overall survival of 40%, whereas the 5-year overall survival of PTCL-GATA3 is close to 20%. Recently Amador et al⁵³  proposed a simple immunohistochemical algorithm using 4 antibodies against TBX21, CXCR3, GATA3, and CCR4 to predict these 2 GEP-defined molecular subgroups (Figure). TBX21 and CXCR3 are for the PTCL-TBX21 subtype whereas GATA3 and CCR4 are for the PTCL-GATA3 subtype. Addition of CXCR3 and CCR4, chemokine receptors associated with the Th1 and Th2 phenotype, respectively, improved the accuracy of classification. This immunohistochemistry panel accurately reproduced the GEP results in the majority of cases (approximately 85%) in the original study, potentially providing a robust and practical tool in PTCL, NOS subclassification. Verification of this immunohistochemistry algorithm by a large study may be helpful to promote its application in practice and improve clinical decisions according to the underlying biology.

Lennert lymphoma, currently named lymphoepithelioid lymphoma, was initially identified based on its morphologic characteristics. It is characterized by diffuse and sometimes interfollicular infiltrates composed of a variable number of neoplastic T cells admixed with a rich background of reactive epithelioid histiocytes. It has less extranodal involvement and portends a better prognosis compared with other PTCL, NOS. The WHO 2017 classification²  points out that the neoplastic cells are cytotoxic CD8⁺ T cells; however, the immunophenotypic features of the neoplastic cells is a subject of debate. Hartmann et al⁵⁴  proposed that Lennert lymphoma cells have a nonactivated cytotoxic phenotype with TIA-1⁺ and granzyme B⁻ as well as a lack of TFH markers. However, Kurita et al⁵⁵  reported that a significant subset of Lennert lymphomas comprise CD4⁺ neoplastic T cells with expression of one TFH marker. The TFH marker–positive Lennert lymphomas demonstrate a worse overall survival compared with TFH marker–negative counterparts despite the lack of a significant difference in histologic presentation. Limited analyses of genetic abnormalities and GEP also suggest enrichment of helper T cells in Lennert lymphoma.⁵⁶ 

Primary EBV⁺ nodal T/NK-cell lymphoma is recommended as a variant of PTCL, NOS, according to the WHO 2017 classification.²  Compared with EBV⁺ extranodal NK/T-cell lymphoma, this disease entity is associated with older age, advanced stage, and poorer survival. A recent copy number aberration study and GEP revealed that primary EBV⁺ nodal T/NK-cell lymphoma is different from the extranodal counterparts, with frequent loss of TCRA loci (on chromosome 14q11.2); upregulation of CD2, CD8, and PD-L1; and depressed expression of CD56.⁵⁷  Therefore, it is crucial to include nodal involvement as one of the defining features for this entity.

The diagnosis and classification of mature T-cell neoplasms are frequently challenging. Integrating clinical, morphologic, and immunophenotypic features usually will provide evidence to reach a preliminary decision. Many cases need molecular genetic studies to finalize the diagnosis and refine the classification. There will be some cases for which experts may have difficulty determining or different opinions about the classification. Molecular genetic studies and mutation profiling can be very helpful for these challenging cases. Even when not required for the final diagnosis and classification, molecular and genomic tests may provide clinically significant information for management decisions.