Emerging strategies in peripheral T-cell lymphoma

Peripheral T-cell lymphomas (PTCLs) are a rare heterogenous subset of non-Hodgkin lymphomas that comprise 7% of all new non-Hodgkin lymphoma diagnoses and 15% of aggressive lymphomas in Western countries.¹  In the United States, this constitutes an incidence of 6 per 1 000 000 people.¹  The term PTCLs is commonly used to refer to the nodal or systemic T-cell lymphomas and comprises 19 different entities with varying clinical and pathologic presentation. However, the most common subtypes include PTCL, not otherwise specified (PTCL-NOS), angioimmunoblastic T-cell lymphomas (AITLs), anaplastic large cell lymphoma (ALCL) positive for the anaplastic lymphoma kinase (ALK), and ALCL ALK⁻, which in total constitute more than half of the cases.¹  With improved diagnostic techniques, it is becoming clear that PTCL-NOS is a heterogenous group of diseases. In the most recent World Health Organization classification, some patients with peripheral T-cell lymphoma with a T-follicular helper phenotype were reclassified as follicular T-cell lymphoma (FTCL). Similar to AITL, which likely derives from the same cell of origin, patients with follicular T-cell lymphoma express at least 2 of the characteristic follicular helper T-cell markers (PD1, CD10, BCL6, CXCL13, ICOS, SAP, and CCR5). Research to advance the treatment of PTCL has been challenging because of the rare nature of these disorders, the heterogeneity in the disease, and its aggressive clinical course. Recently, efforts to better understand the biology of these diseases and develop treatments to improve outcomes in both the upfront and relapsed setting are leading to advances in therapies. This review will focus on the treatment of the more common forms of PTCL, such as PTCL-NOS, FTCL, AITL, ALK⁺ ALCL, and ALK⁻ ALCL.

The patient is a 60-year-old woman with no significant past medical history who presented with fevers, a rash, and left axillary lymphadenopathy. She lost 17 pounds in the last 3 months and has had significant fatigue. Biopsy of the rash showed a lymphocytic infiltrate with a predominance of T-cells expressing CD4, BCL6, and CD10. Fluorodeoxyglucose positron emission tomography (PET)/computed tomography (CT) splenomegaly and axillary, retroperitoneal, and inguinal lymphadenopathy measuring between 2 and 4 cm and with maximum standard uptake value of 12. Excisional biopsy of the axillary lymph node shows small to medium atypical lymphocytes with clear cytoplasm that cluster around the endothelial venules in an arborizing pattern. The atypical cells are positive for CD4, CD5, CD2, CD10, BCL6, and PD1 and negative for CD8, CD56, CD20, and CD30 by immunohistochemistry consistent with AITL. The bone marrow biopsy is moderately hypercellular (80%) and shows trilineage hematopoiesis with a moderately increased number of megakaryocytes and 5% involvement by AITL. Laboratory evaluation is significant for an elevated lactate dehydrogenase (340 U/L) and low hemoglobin (8.0 g/dL).

Although there is no standard-of-care front-line combination chemotherapeutic regimen for the more common forms of PTCL, patients who are eligible for anthracycline-based treatment are most commonly treated for curative intent with cyclophosphamide, doxorubicin, vincristine, prednisone (CHOP)-based chemotherapy with or without etoposide. With this approach, the 5-year overall survival (OS) and progression-free survival (PFS) remain approximately 30% to 40% and 20% to 30%, respectively, in PTCL-NOS, AITL, and ALCL ALK⁻.^1,2  Many advocate for autologous transplant in first remission in these subtypes, which is associated with a 45% 4-year PFS.^2,3  ALCL ALK⁺ is more sensitive to CHOP-based therapy, and the 5-year OS and PFS in this disease are 70% to 80% and 60% to 65%, respectively.^1,2 

The majority of data to guide treatment of PTCL in the upfront setting has come from phase 2 studies, retrospective analyses, or registry studies. In a retrospective pooled analysis of 343 patients with PTCL-NOS, AITL, and ALCL treated on several German protocols, CHOEP improved 3-year event-free survival from 51% to 75% (P = .003) in patients 60 years of age or younger with normal LDH.⁴  Similarly, with the addition of etoposide, those with ALK⁺ ALCL showed an improvement in 3-year event-free survival from 57.1% to 91.2% (P = .012), whereas the 3 other subtypes had an improvement in 3-year event-free survival from 48.3% to 60.7% (P = .057). These findings have also been observed in prospective registry studies.² 

Until recently, other strategies to improve initial outcomes of PTCL had not been successful, resulting in increased toxicity without significant improvement in efficacy in largely unselected patients. These include strategies of dose intensification, use of gemcitabine-based therapy, and the addition of alemtuzumab, lenalidomide, or pralatrexate to CHOP.^5-8 

Brentuximab vedotin (BV) is an antibody-drug conjugate consisting of an anti-CD30 monoclonal antibody linked to monomethyl auristatin E, which inhibits microtubules. A 452-patient international, randomized, double-blind, phase 3 study compared BV combined with cyclophosphamide, doxorubicin, prednisone (CHP) compared with CHOP in those with PTCL expressing CD30 in at least 10% of malignant cells by immunohistochemistry.⁹  Of note, 70% of the patients in this study had ALCLs, which universally express CD30. Patients treated with BV-CHP had an improvement in median PFS from 20.8 months to 48.2 months compared with CHOP. Treatment with BV-CHP also reduced the risk for death by 34% compared with CHOP. The improvement in outcomes with BV-CHP was most pronounced in ALCL. Although the study was not powered to evaluate the efficacy in AITL or PTCL separately, in the 54 patients with AITL and 72 with PTCL-NOS, BV-CHP did not improve outcomes compared with CHOP. The toxicity profile of BV-CHP was similar to CHOP, but was associated with an increased risk for diarrhea and persistent neuropathy. This was the first randomized study to show an OS benefit in PTCL, and demonstrated that the addition of effective therapy to a CHOP backbone in a selected patient population can lead to clinical benefit. On this basis, BV-CHP was approved by the US Food and Drug Administration (FDA) for the front-line treatment of CD30-expressing PTCL.

Other methods being investigated in an effort to give more personalized therapy in PTCL include more dynamic markers of response or resistance such as interim PET/CT imaging parameters and evaluation of minimal residual disease. Lack of a complete remission by PET/CT using the Deauville 5-point score after 4 cycles of CHOP-based chemotherapy is associated with very poor outcomes, and studies in the relapsed/refractory setting are beginning to enroll patients with these early suboptimal outcomes.¹⁰  A small series in patients with AITL showed that the mutational profile in the primary tumor of AITL could be detected by cell-free DNA mutational analysis in the majority of cases.¹¹  This suggests that cell-free DNA can be used as a technique for minimal residual disease. Therefore, we are also investigating the use of minimal residual disease assessments in the upfront therapy setting to evaluate whether there are more sensitive markers of response compared with standard PET/CT imaging (NCT03297697). By evaluating minimal residual disease both by the T-cell gene rearrangement and cell-free DNA mutational analysis, we hope to identify populations at higher risk for chemoresistance and evaluate which patients benefit most from autologous transplant in first remission. MRD assessment may also inform rational combination strategies in the salvage therapy setting and serve as an early endpoint to evaluating response to salvage therapy. Similarly, these dynamic strategies can inform who is more likely cured after frontline treatment.

The patient was treated with CHOEP chemotherapy for 6 cycles. Both an interim PET/CT and end-of-treatment PET/CT showed a complete remission. She underwent consolidation with an autologous transplant and remained disease free until 1.5 years after her transplant, when she noted cervical lymphadenopathy. PET/CT shows cervical, retroperitoneal, inguinal fluorodeoxyglucose-avid lymphadenopathy. Core needle biopsy of the cervical lymph node shows recurrent AITL; immunohistochemistry for CD30 is negative.

Outcomes of relapsed or refractory PTCL remain poor, with median survival of 6 to 10 months.¹²  Although allogeneic transplant can be curative in patients with relapsed PTCL in up to 60% of patients, advanced age, medical comorbidities, lack of suitable donor, and most important, ability to adequately control the lymphoma hinder our ability to pursue transplant for many patients with PTCL.¹³  Nevertheless, our understanding of the biology of these heterogeneous diseases is improving. Simultaneously, options for treatment in the relapsed/refractory setting have expanded beyond standard chemotherapy with modest activity in PTCL (such as gemcitabine, platinum-based therapy, etoposide, alkylating agents). Since 2009, there have been 4 drugs that have been approved by the FDA for the treatment of relapsed PTCL: pralatrexate, romidepsin, belinostat, and BV. In the relapsed/refractory setting, the responses rate of these single agents ranges from 20% to 35%, with the exception of BV for ALCL (Table 1).^14-17  Histone deacetylase inhibitors (HDACis) and pralatrexate were serendipitously found to have activity in PTCL, as there were outstanding responders to these agents in their original phase 1 studies. Therefore, as the biologic understanding of their action has lagged behind their clinical development, we are still trying to better understand the complete mechanism of action of HDACis in PTCL. In the current era, however, we are beginning to chip away at our understanding of the molecular drivers of these diseases and subsequently develop rational therapies and combination therapies in PTCL.

It has become clear that epigenetics plays a critical role in cancer pathophysiology and therapeutics. Histones help to maintain DNA configuration and regulate gene expression via histone deacetylases and histone acetyltransferases. HDACis are agents that result in increased histone acetylation, which is believed to result in decreased expression of tumor suppressor genes and oncogenes.

Romidepsin is a class I selective HDACi that was FDA approved for PTCL in 2011 on the basis of a single-group, phase 2 study in 131 patients with relapsed/refractory PTCL. In this study, romidepsin demonstrated a 25% overall response rate (ORR) and a 15% complete remission (CR) rate with a median duration of response of 17 months.¹⁷  Those who achieved a CR had remissions that were maintained for up to 58 months.¹⁷  Of note, those with partial remissions or stable disease for at least 90 days also demonstrated durable responses with similar PFS and OS.

Belinostat is an HDACi that inhibits class I-IV HDACis and was FDA approved for relapsed/refractory PTCL in 2014 on the basis of the phase 2 study of 129 patients with relapsed or refractory PTCL, which showed an ORR of 25.8% and CR of 10.8%.¹⁵  The median PFS and OS were 1.6 and 7.9 months, respectively, and the median duration of response was 13.6 months. Those who achieved a CR had a median duration of response of 29 months.

The mechanism of action of HDACis is poorly understood, and it is likely that HDACis have more global effects that result in cell death. In PTCL and AITL, up to 75% of patients have mutations in genes involved in chromatin modification.^18,19  It has been hypothesized that mutations in chromatin modification including DNA methylation may confer higher sensitivity to HDACi, and the registration trials of belinostat and romidepsin showed that patients with AITL can have sustained remissions with these agents.^15,17,20 

Romidepsin-based combination strategies have led to higher response rates in smaller studies including PTCLs. The combination of romidepsin with lenalidomide has shown a response in 6 of 10 patients with PTCLs, including responses in 3 of 3 patients with AITL.²¹  The combination of romidepsin and lenalidomide with carfilzomib was studied in 13 patients with PTCL and showed responses in 7 of 13 patients with PTCLs, including 5 of 5 patients with AITL.²²  The combination of romidepsin with 5-azacitidine has shown responses in 7 of 9 patients with PTCLs, including 1 patient with AITL with a CR.²³  Similarly, the combination of romidepsin with pralatrexate has shown responses in 10 of 14 patients with PTCLs.²⁴  Although the latter 2 studies did not have a high proportion of patients with AITL, these small experiences suggest that romidepsin-based combinations could be more effective in AITL, or possibly FTCL.

CD30 is expressed in a subset of patients with T-cell lymphoma and is currently approved by the FDA for patients with previously untreated PTCL with CD30 expression, relapsed ALCL, relapsed primary cutaneous ALCL, and relapsed CD30 expressing mycosis fungoides, and is listed in the NCCN compendium for those who have relapsed systemic and cutaneous T-cell lymphomas. In ALCL, CD30 is universally expressed; however, in other forms of PTCL, 20% to 25% of cases have more than 50% expression of CD30, and up to 70% cases have at least come degree of CD30 expression.²⁵  BV is most effective in ALCL, with an ORR of 86% with a median PFS 20 months. Although more than half the responding patients were consolidated with transplant, it is notable that up to two-thirds of patients experience complete remission, and the median duration of response of these patients has still not been reached after a median follow-up of 5 years.¹⁶  Although patients who underwent subsequent consolidative transplant had improved PFS, the median PFS in those who did not receive a consolidative transplant was 39.4 months. In those with CD30 expressing PTCL and AITL, the ORR is 41% with a median PFS of 2.6 months.²⁶  Nevertheless, the degree of CD30 expression within the non-ALCL subtypes does not predict response. Evaluation of patients with AITL and PTCL-NOS treated with BV in a phase 2 study of 35 patients showed that there was no correlation between centrally reviewed CD30 expression and response. In fact, patients with as little as 1% expression demonstrated complete responses.²⁶  Given the data for BV in the relapsed/refractory setting, it was logical to consider a BV-based combination strategy in the upfront setting, which was evaluated in ECHELON 2, as discussed earlier.⁹  The story of BV in PTCL demonstrates that effective drugs in the relapsed/refractory setting can be successfully incorporated into front-line approaches and impact survival of those with PTCL when there is a biomarker-driven strategy.

Phosphoinositide-3-kinase (PI3K) is a membrane-associated lipid kinase that acts upstream of the AKT/mTOR and Raf/MEK signaling cascades. The catalytic subunit of PI3K (p100) exists in various isoforms, with the γ and δ isoforms being predominantly expressed in leukocytes. PI3K signaling has been shown to have important roles in leukocyte development, activation, and migration. In preclinical models, inhibition of PI3K-δ has been shown to decrease regulatory T-cell-mediated immune tolerance to cancer cells.²⁷  Similarly, in a patient-derived xenograft model of AITL, treatment with the PI3K-δ,γ inhibitor duvelisib resulted in repolarization of splenic macrophages to the classically activated M1 phenotype.²⁸ 

In patients with PTCL, the PI3K-δ,-γ inhibitor duvelisib has shown promising single-agent activity. In a phase 1 dose-escalation study of duvelisib in patients with hematologic malignancies, 35 patients with T-cell lymphoma, including 16 with PTCL, were enrolled in either the dose-escalation phase or as part of a T-cell lymphoma-specific expansion cohort. Among those with PTCL, the ORR was 50%, with 3 patients (19%) achieving CR.²⁹  The median duration of response was 8 months. Therefore, a PTCL and cutaneous T-cell lymphoma-specific combination study of duvelisib with either bortezomib or romidepsin was developed and included a group in which subjects received single-agent duvelisib for 1 month as a lead in. Those who were not in a CR at the end of the lead in continued on to their prespecified combination. The ORR to single-agent duvelisib on the lead in was 46%, with a 27% CR rate after 1 month of treatment.²⁸  Interestingly, intrapatient cytokine levels pretreatment and on-treatment were predictive of response to duvelisib in T-cell lymphomas: responding patients had increased CCL1, IL-17α, and soluble CD40L and decreased levels of the IL-12 inhibitory subunit (p40) and CXCL13. In further correlative studies, T-cell lymphoma lines susceptible to growth inhibition by duvelisib were positive for pAKT at baseline. These initial promising results have led to a multiinstitutional phase 2 study of duvelisib in which the optimal dose and efficacy are being confirmed (NCT03372057). Copanlisib, a PI3K-α,-δ inhibitor, has also shown a 21% response rate in PTCL.³⁰  Similarly, there are ongoing single-agent or combination trials with PI3K-δ,-γ inhibitors, such as tenalisib, in PTCL. (NCT03770000)

Increased understanding regarding the cell of origin in FTCL and AITL has led to important developmental therapeutic advances. Azacitidine (5-AZA) is an analog of the pyrimidine nucleoside cytidine that has effects on cell differentiation, gene expression, and DNA synthesis and metabolism. At low doses, it serves as an epigenetic modifier of DNA methylation, and at higher doses, it carries a direct cytotoxic effect. Intravenous and subcutaneous 5-AZA are approved for myelodysplastic syndrome, acute myelogenous leukemia, and chronic myelomonocytic leukemia. A retrospective French cohort study described the outcomes of 19 patients with relapsed/refractory PTCL who were treated with single-agent 5-AZA.³¹  Of note, 10 of these patients had concurrent myelodysplastic syndrome or chronic myelomonocytic leukemia. The ORR was 53% (10/19), but in AITL it was 75% (9/12). Of the 12 patients with AITL, 5 achieved a CR. Notably, all patients with AITL who responded to 5-AZA demonstrated mutations in TET2, which has been associated with response to this agent in other diseases.³¹  This has led to an ongoing international study of 5-AZA in AITL, FTCL, or nodal PTCL with a TFH phenotype (NCT03593018), as well as a study combining 5-AZA with CHOP in untreated PTCL (NCT03542266).

Inducible costimulator (ICOS) is a member of the CD28 (B7) family, which is an important signaling mechanism that maintains a balance between immune response and inhibition of autoimmunity. Given the universal expression of ICOS on follicular helper T cells, therapy directed against ICOS is of interest in this subypte of PTCL. There is an ongoing international study of a humanized monoclonal antibody to ICOS (MEDI-570) in T-cell lymphomas with a special interest in in AITL, FTCL, or nodal PTCL with a TFH phenotype (NCT02520791).

Just as the biology of FTCL and AITL seems to be unique, it also appears that those with ALCL have unique biology. ALK⁺ ALCL is a form of ALCL that carries the canonical translocation t(2,5) (p23;q35) leading to the expression of the nucleophosmin-ALK chimeric protein and constitutes 16% of PTCL in North America.¹  ALK⁺ ALCL is more chemosensitive than other forms of PTCL. Curative options in relapse include treatment with chemotherapy followed by an autologous transplant and/or treatment with brentuximab vedotin.¹⁶  In patients who are either not candidates for these approaches or who relapse after these approaches, crizotinib has been explored. A study of 26 pediatric patients with ALK⁺ ALCL treated with crizotinib at 165 or 280 mg/m² showed an ORR of 88% with CR 80%, with responses being durable or allowing for consolidation with transplantation.³²  Within this study, quantity of circulating tumor DNA for the nucleophosmin-ALK transcript by quantitative real-time polymerase chain reaction correlated with response. Crizotinib is NCCN compendium listed for the treatment of ALK⁺ ALCL for salvage therapy.

Multiple studies have shown that ALK fusion proteins lead to activation of various downstream sequencing pathways including Janus kinase (JAK)/signal transducer and activator of transcription (STAT), PI3K, the extracellular signal regulated kinase.³³  More specifically, the ALK chimeric protein is known to activate STAT3 and gain-of-function mutations in both JAK2 and STAT3 have been reported in approximately 20% of ALK⁻ ALCL. It is now believed that interferon regulatory factor 4 may be involved downstream of STAT3 to potentiate its activation.³⁴  The JAK/STAT pathway has also been found to be activated in other forms of PTCL, such as AITL, enteropathy associated T-cell lymphoma, hepatosplenic T-cell lymphoma, and extranodal NK/T-cell lymphoma.³⁵ 

An ongoing study of ruxolitinib, a JAK 1/2 inhibitor, in a variety of T-cell lymphomas including PTCL has demonstrated an ORR of 27% and a clinical benefit rate (as defined by ORR or stable disease for 6 or more moths) of 38%. However, in the 12 patients who had known activating mutations in the JAK/STAT pathway, the clinical benefit rate was 58%. In the 11 patients who had functional evidence of JAK/STAT activation, the clinical benefit rate was 36%.³⁶ 

Cerdulatinib, an oral inhibitor of spleen-associated tyrosine kinase (SYK), JAK1 and JAK3, also has shown promising activity in PTCL. SYK is known to be overexpressed selectively in PTCL in more than 90% of cases and in some cases this is through expression of SYK in its native form or through a fusion of SYK with interleukin-2-inducible T-cell kinase.³⁷  Mice with a conditional expression of the SYK/interleukin-2-inducible T-cell kinase fusion develop a lymphoproliferative disease.³⁸  In an ongoing phase 2 study of 45 patients with PTCL and 25 with cutaneous T-cell lymphoma, the ORR was 35% with 31% CR in the PTCL cohort.³⁹ 

These early studies demonstrate that targeting of the JAK/STAT pathway as well as SYK can be clinically beneficial in these patient populations. In both studies, we are hopeful that correlative studies will inform whether there are predictors of response to these classes of therapy.

The patient initiated treatment with romidepsin and continued receiving therapy with a partial remission for 6 months. On relapse, she opted to enroll in a clinical trial of a PI3K inhibitor, which led to a complete remission. She is being considered for an allogeneic transplant.

As discussed earlier, histology and expression of CD30 have been the most reliable methods to personalize therapy, particularly for patients with ALCL. As patients with relapsed or refractory PTCL can develop rapid clinical progression and, with the exception of BV in CD30 expressing PTCL, the ORRs to the FDA approved agents is 20% to 30%, the window to achieve disease control is often narrow. Therefore, there are increasing efforts to identify predictive markers of response or resistance to single agents and combination therapies.

Heavican and colleagues have recently published a novel gene expression profile in PTCL that helps further subclassify forms of PTCL into 4 distinct groups according to the predominance of GATA3 expression, TBX21 expression, FTCL phenotype, and those not able to be classified.⁴⁰  Those with PTCL-GATA3 subgroup had higher mutational complexity, including deletion of tumor suppressors such as p53, CDKN2A, PTEN, FAS, and gain of MYC and STAT3. These were associated with inferior survival. In contrast, those with the PTCL-TBX21 subgroup had less aberrant genomes, but were more likely to have gains in cell cycle regulators and immune regulatory genes. Those who had a FTCL phenotype had frequent mutations of genes regulating the epigenome (TET2, IDH2, DNMT3A), as well as TCR and costimulatory signaling pathway mutations. We hope that better understanding of the drivers of different forms of PTCL can help us personalize therapy for patients, akin to how cell of origin has affected therapeutic strategies and drug development in diffuse large B-cell lymphoma.

Until recently, our preclinical models of PTCL were not robust. The development of patient derived xenografts and transgenic mouse models has quickly led to the advancement of understanding the pathophysiology of subtypes of PTCL, especially AITL, and better preclinical drug development.^29,41  In a short period of time, these have helped us better understand the role of MCL1/BCL2/BCLX, MDM2/MDMX, and RHOA in T-cell lymphomas.^41-43 

We anticipate that our understanding of how to tailor therapy based on histology, molecular status, and dynamic measures of response, such as imaging and minimal residual disease, will expand tremendously. We are uncovering that PTCL is a more biologically complex group of diseases, well beyond its known histologic heterogeneity. The oncogenesis of PTCL is likely dependent on multiple factors including modulators of cell signaling, the effect of epigenetics, the relationship of the lymphoma to the microenvironment, and the underlying relationship of the lymphoma with the immune system. Although it is highly unlikely that there is a uniform oncogenic driver for all forms of T-cell lymphoma, rational combination therapies as well as therapies directed toward subtype specific therapy can yield meaningful advances to the field. Although previously, many felt that research in PTCL seemed like searching in the dark, we have now integrated effective therapies into front-line therapies that improve survival for a subset of patients. Similarly, we hope that these recent advances will shed light on methods to improve treatment of patients in both the front-line and relapse settings.

The author thanks Steven Horwitz and Nancy Bartlett for their careful review of the manuscript.

Neha Mehta-Shah, Washington University in St. Louis, 660 S Euclid, Box 8056, St. Louis, MO 63132; e-mail: mehta-n@wustl.edu.