What is new in the classification of peripheral T cell lymphomas?

In the group of Epstein–Barr virus (EBV)-driven lymphoproliferative disorders of childhood [21], hydroa vacciniforme lymphoproliferative disorder (LPD) replaces what was previously designated hydroa vacciniforme-like LPD, because essentially all such lesions are associated with EBV infection. The ICC further recognizes two variants: a classic indolent form (limited to the skin) and a systemic aggressive form of the disease, more common in non-Caucasians. Chronic active EBV disease now replaces chronic active EBV infection to denote a pathologic disease condition, in line with the notion that pathogenic mutations indicating a neoplastic process are detected in a subset of patients.

The terminology, definition, and diagnostic criteria of extranodal NK/T cell lymphoma (ENKTCL) nasal type are unchanged in the ICC. Since this lymphoma is known to occur at various extranodal sites besides the nasal area which is involved in typical cases, “nasal type” was dropped in the WHO-HAEM5.

Cases of primary nodal EBV-positive T cell or NK cell lymphoma, formerly considered as a subtype of PTCL, not otherwise specified (NOS), are now categorized as a separate entity, namely primary nodal EBV+ T cell/NK cell lymphoma, provisional in the ICC, or nodal EBV+ T and NK cell lymphoma in the WHO-HAEM5 [21]. This rare disease, most prevalent in East Asia, involves lymph nodes and is frequently disseminated but lacks nasal involvement and tends to occur in elderly adults, in association with HIV infection or immunodeficient conditions [14]. Pathological features distinct from ENKTCL include a monomorphic large cell morphology, less frequent necrosis, negativity for CD56, positivity for CD8, and more frequent derivation from T cells than from NK cells [14, 19]. The tumor is characterized by loss of 14q11.2, upregulation of immune pathways, low genomic instability and recurrent mutations involving the epigenetic modifiers, such as TET2 and DNMT3A, and JAK-STAT pathway genes [27].

In 2017, the developing concept that follicular helper derivation represents a unifying feature of a large group of nodal CD4+ T cell lymphomas was reflected by the creation of an umbrella term “nodal T cell lymphoma of T follicular helper (TFH) origin” to encompass angioimmunoblastic T cell lymphoma, follicular T cell lymphoma, and nodal PTCL with T follicular helper phenotype. Since then, this notion has been reinforced by additional evidence indicating shared molecular and genetic features [10], and, importantly, clinical data suggest that this grouping might be relevant to treatment decisions, as TFH lymphoma appear more sensitive to epigenetic therapies than non-TFH PTCLs (Fig. 2a; [4]). Therefore, the ICC considers one single disease entity, namely follicular helper T cell lymphoma, comprising three subtypes, angioimmunoblastic, follicular, and NOS (Fig. 2b–d; [11]). This entity by definition excludes primary cutaneous CD4+ T cell lymphoproliferations which also feature a TFH phenotype. The WHO-HAEM5 proposal is more conservative, considering a family of three related entities of nodal T follicular helper cell lymphomas. The TFH immunophenotype is defined by the expression of at least two and ideally three TFH markers out of a panel of at least five markers (CD10, BCL6, PD1, ICOS, CXCL13) that it is now recommended to test for routinely and systematically when a diagnosis of TFH lymphoma is considered or must be excluded [1, 5]. TFH lymphomas frequently carry mutations in TET2, DNMT3A, RHOA, and IDH2, which are rarely seen in combination in other PTCL entities; hence, mutational testing may be diagnostically useful [8, 12].

The four entities of anaplastic large cell lymphomas (ALCLs) are identical in both proposals: ALK-positive (ALK+) and ALK-negative (ALK−) ALCL, primary cutaneous ALCL (within the spectrum of CD30-positive cutaneous T cell lymphoproliferative disorders), and breast implant-associated (BIA-)ALCL. Among ALK− ALCLs, those with DUSP22 rearrangement (25–30% of cases; Fig. 3a–d) differ from those devoid of this alteration, as they usually lack JAK-STAT3 activation and EMA expression, less frequently express cytotoxic molecules, harbor MSC mutations in about one third of cases, and have distinctive transcriptomic signature and methylation profiles [17, 18]. The clinical impact of DUSP22 rearrangement remains controversial: the initially reported markedly superior prognosis of these cases was not confirmed in subsequent studies, while data from more recent cohorts still support an intermediate prognosis of DUSP22-rearranged ALK− ALCL, standing between ALK+ ALCL and DUSP22-non rearranged ALK− ALCL [23, 24]. Taking into account its biological and prognostic peculiarities, the ICC recognizes DUSP22-rearranged ALCL as a genetically defined subtype of the disease and recommends systematic FISH testing for DUSP22 in ALK− ALCL [8]. Other structural aberrations are recurrent in ALK− ALCL but less common. These include TP63 rearrangements, associated with an adverse prognosis [23]; as well as fusion genes involving tyrosine kinases such as JAK2, FRK, ROS1, and TYK2, which may represent potential therapeutic targets [8].

BIA-ALCL (Fig. 3e) is recognized as a definitive entity both in the ICC and WHO-HAEM5. While histopathologically it largely overlaps with systemic ALK− ALCL, the pathogenetic association of BIA-ALCL with the microenvironment of textured breast implants is unique. At the genetic level, a highly characteristic 20q13.13 loss has been reported in two thirds of cases [7], and mutations in epigenetic modifiers such as KMT2C, KMT2D, and CREBBP are also frequently detected [16]. Similar to systemic ALCL, activation of the JAK-STAT3 pathway is a constant feature of BIA-ALCL, most commonly through mutations of STAT3 and/or JAK1 [16]. In contrast, rearrangements of ALK, DUSP22, or TP63 associated with systemic ALCLs are not observed. The prognosis of BIA-ALCL is generally excellent after surgical removal of the periprosthetic fibrous capsule, but is less favorable in cases of infiltration of the adjacent breast parenchyma [15].

The three main aggressive types of primary intestinal T cell lymphomas (enteropathy-associated T cell lymphoma [EATL], monomorphic epitheliotropic intestinal T cell lymphoma (MEITL), and intestinal T cell lymphoma, NOS) are unchanged [9]. EATL occurs in populations with a higher prevalence of HLA haplotypes predisposing to celiac disease, as a complication of celiac disease and refractory celiac disease, or de novo in individuals with no history of malabsorption. The tumors may be multiple and present as ulcers or, less commonly, masses, comprise a polymorphous infiltrate with admixed inflammation, and often pleomorphic to anaplastic lymphoma cells. The typical immunophenotype is CD3+ CD4− CD8− CD30+/− TCR-silent EBV-negative, with expression of cytotoxic molecules. MEITL presents as a tumor mass and spans a morphologic spectrum. While typical cases are monomorphic with little necrosis, other tumors exhibit pleomorphic cytology and/or other atypical features like necrosis, brisk mitotic activity, and angiocentricity [26]. In MEITL, the neoplastic cells are CD3+ CD4− CD8+ CD56+ TCR-positive (gamma-delta more commonly than alpha-beta) EBV-negative. Genomic features may be helpful in differentiating between EATL and MEITL: alterations in the JAK/STAT pathway genes target primarily STAT3 and JAK1 in EATL, and STAT5B and JAK3 in MEITL. Deleterious alterations of the SETD2 gene, translating into reduced H3K36 trimethylation, are almost constant and rather specific to MEITL [22]. Type II refractory celiac disease has been added to the list of entities in the ICC, as this represents an “in situ” neoplastic condition precursor to EATL, and recent works have shown that it often already harbors driving mutations in JAK1 and/or STAT3 similar to those present in EATL [6].

The formerly provisional “indolent T cell LPD of the gastrointestinal tract” is confirmed in the ICC with the addition of “clonal” to emphasize its neoplastic nature. Indeed a variety of somatic genetic alterations have been found in these cases, including a recurrent JAK2::STAT3 fusion in a subset of CD4+ cases [25]. In WHO-HAEM5, the name has been modified to “indolent T cell lymphoma,” given the fact that transformation into a high-grade PTCL has been described in some patients. Both proposals have created a new category to classify the indolent gastrointestinal LPD of NK cells (Fig. 4), which also carry a variety of genetic mutations, including a recurrent JAK3 small in-frame deletion [28]. These T and NK LPDs of the gastrointestinal tract are in general restricted to the mucosa and represent a diagnostic challenge and should not be confused, on the one hand with inflammatory conditions, on the other hand with aggressive lymphomas, since their course is usually indolent despite possible relapses, multifocality, and chronicity, and they do not respond to chemotherapy.

The group of PTCLs, not otherwise specified (NOS), remains a diagnosis of exclusion (Fig. 5). Cases with a TFH immunophenotype must be excluded, since lymphomas with no morphologic specification but showing a TFH immunophenotype, defined by the expression of two or ideally three TFH markers, are classified as TFH lymphoma, NOS. Moreover, caution must be applied in this scenario to exclude primary cutaneous T cell lymphomas or human T-lymphotropic virus type 1 (HTLV-1)-associated adult T leukemia/lymphoma, as these entities, which are often CD4+, may show expression of TFH markers [20].

Two biological subtypes of PTCL, NOS, namely PTCL-TBX21 and PTCL-GATA3, have been identified by gene expression profiling, and are characterized by overexpression of transcription factors TBX21 or GATA3 and corresponding target genes, with different prognoses and distinct oncogenic pathways ([12, 13]; Table 1). An immunohistochemical algorithm using four markers applied sequentially (TBX21, CXCR3, GATA3, and CCR4) can provide surrogate information on the molecular subtypes [3], and a digital nanostring-based assay has recently been published [2]. However, it is acknowledged that there is currently too little evidence to recommend molecular subtyping of PTCL, NOS, in routine clinical use [3]. PTCL-GATA3 demonstrates high genomic complexity characterized by biallelic deletion/mutation of TP53, CDKN2A/B, or RB1, and carries a worse prognosis compared to PTCL-TBX21, which shows low genomic complexity and few recurrent specific genetic changes.

In conclusion, the updated classifications of T and NK cell neoplasms confirm the diversity and complexity of these disorders. Nevertheless, the accumulating knowledge of their biology is translated into more meaningful categories, and an increasing importance of molecular testing for precision diagnosis and tailored therapy.