Past, present and future therapeutic approaches in nodal peripheral T-cell lymphomas

Abstract

Peripheral T-cell lymphomas (PTCL) encompass over 30 different entities and although they share post-thymic T- or NK-cell derivation, the disease biology and genomic landscape are very diverse across subtypes. In Western populations, nodal PTCL are the most frequently encountered entities in clinical practice and although important achievements have been made in deciphering the underlying biology and in therapeutic advances, there are still large gaps in disease understanding and clinical scenarios in which controversy over best practice continues. CHOP (cyclophosphamide, doxorubicin, vincristine, prednisone)- based chemotherapy continues to be the ‘standard’ treatment, with the addition of brentuximab vedotin (BV) in the combination CHP (cyclosphosphamide, doxorubicin, prednisone)-BV representing a new treatment paradigm in CD30+ PTCL although its benefit is less certain in the non-anaplastic large cell lymphoma subtypes. Given the high risk of relapse, consolidative autologous stem cell transplant is considered in nodal PTCL, outside of ALK-positive anaplastic large cell lymphoma; however, in the absence of a randomized controlled trials, practices vary. Beyond CHP-BV, most study activity has focused on adding a novel agent to CHOP (i.e., CHOP + drug X). However, with high complete remission rates observed with some novel therapy combinations, these regimens are being tested in the front-line setting, with a particular rationale in follicular helper T-cell lymphomas which have a clear sensitivity to epigenetic modifying therapies. This is well exemplified in the relapsed/refractory setting in which rational combination therapies are being developed for specific subtypes or guided by underlying biology. Taken together, we have finally moved into an era of a more personalized approach to the management of nodal PTCL.

Introduction

Peripheral T-cell lymphomas (PTCL) represent approximately 10% of all non-Hodgkin lymphomas although geographic variation is notable.^1,2 In 2022, both the World Health Organization (WHO) 5^th edition lymphoma classification update (WHO-HAEM5)¹ and the new International Consensus Classification (ICC)² were published. These two publications contained classification refinements, including for mature T/NK-cell neoplasms. Most of these refinements are concordant in the two classification systems, although there a few exceptions (see Table 1 in the Introduction to the Review Series on Lymphoma published in this issue of Haematologica³). The so-called ‘nodal’ PTCL subtypes, grouped to separate them from predominantly leukemic, extranodal and cutaneous subtypes of PTCL, represent about 60% of all PTCL in Western populations, and include PTCL not otherwise specified (NOS), anaplastic large cell lymphoma (ALCL) and follicular helper T-cell lymphomas. The last group are referred to as nodal T-follicular helper lymphomas in the WHO-HAEM5 and are considered a family of three lymphomas – angioimmunoblastic type (angioimmunoblastic T-cell lymphoma [AITL]), a NOS type and a follicular type – whereas the ICC describes one entity, follicular helper T-cell lymphoma, with three subtypes (angioimmunoblastic type, follicular type and NOS).³ For the purpose of this review, they are collectively referred to as TFHL with specific subtypes referenced where appropriate. In addition, a new rare nodal PTCL subtype is now recognized, primary nodal Epstein-Barr virus (EBV)⁺ T/NK-cell lymphoma in the ICC (provisional entity) or EBV⁺ nodal T/NK-cell lymphoma in WHO-HAEM5 (distinct entity) (Table 1 in the Introduction to the Review Series³) which, in addition to being EBV⁺, has an activated cytotoxic phenotype (TIA⁺ ± granzyme B/perforin) and a poor outcome that is more similar to that of extranodal NK/T-cell lymphoma.⁴ While progress in PTCL has lagged behind that in B-cell lymphomas, the past decade has been a period of great advancement, both in elucidating disease pathogenesis and in the development of new therapies. Recognizing that drug sensitivities may be different, clinical trials have evolved to focus specifically on PTCL and more recently, on specific subtypes.

Table 1.Largest retrospective series including nodal peripheral T-cell lymphomas treated with primarily anthracycline-based chemotherapy.

With the exception of ALCL, in which brentuximab vedotin (BV)-CHP (cyclosphosphamide, doxorubicin, prednisone) is considered the new standard treatment, uncertainty remains regarding the optimal front-line therapeutic regimen and the role of consolidative high-dose chemotherapy and autologous stem cell transplant (auto-SCT). This is particularly apparent in TFHL in which sensitivity to epigenetic therapies has led to numerous studies evaluating chemotherapy-free, novel therapy combination treatment approaches, also in treatment-naïve patients. Furthermore, as the genomic landscape is uncovered, evolving studies are targeting specific pathway vulnerabilities (e.g., Janus kinase [JAK]/signal transducer and activator of transcription [STAT], phosphatidyl inositol 3 kinase [PI3K]) as well as integrating biological correlates in an effort to understand biomarkers of response and resistance. As a follow-up to the nodal PTCL pathobiology paper by Bisig, Savage and de Leval in this issue, we review the history of CHOP (cyclophosphamide, doxorubicin, vincristine, prednisone) and transplant for nodal PTCL, highlight new treatment advances in the front-line and relapsed/refractory (R/R) settings as well as the promise of a more 'personalized' therapy approach.

The origin of CHOP and impact of prognostic factors in PTCL

Almost 30 years ago, the landmark Southwest Oncology Group (SWOG) phase III study established the CHOP regimen as the standard combination chemotherapy for all aggressive lymphomas. The study was conducted when diagnoses were based on the Working Formulation, and prior to the integration of routine immunophenotyping, but CHOP treatment was similarly adopted for PTCL.⁶ Several large, retrospective series have documented the poor outcome in most patients with PTCL treated with primarily anthracycline-based chemotherapy, including subtype-specific results (Table 1).^{7- 9} Notably, there is limited information on the new entity, primary nodal EBV⁺ T/NK-cell lymphoma, but it appears to have an outcome that is inferior to that of PTCL-NOS.⁴ In these series, given their only recent recognition, primary nodal EBV⁺ T/NK-cell lymphoma and TFHL other than AITL, are combined in the PTCL-NOS subgroup. Of note, the International Prognostic Index (IPI) effectively stratifies patients with PTCL-NOS and ALCL into risk groups^7,10-12 but its usefulness in AITL has been more limited given that most patients fall into a high-risk category.^7,13,14 Overall, patients with ALK⁺ ALCL have a more favorable outcome, although this is in part driven by a younger age at diagnosis and, importantly, those with a high IPI score have a poor outcome, not unlike that of patients with other PTCL subtypes (IPI score ≥3; 5-year progression-free survival [PFS] 23-54%; 5-year overall survival [OS] 23-62%).^10,15 Outcomes are also better in ALK^- ALCL,^{7-1 1} a finding that is more evident in larger series and with central pathology review, given the potential for mis-diagnosis as CD30⁺ PTCL-NOS. The outcome of ALK^- ALCL may also be impacted by the proportion of cases with DUSP22 rearrangement and/or P63 rearrangement, as outlined by Bisig, Savage and de Leval.⁵ A recent study suggested that the presence of TP53 mutations correlates with an inferior PFS in patients treated with CHOP or CHOP-like chemotherapy and is found in approximately one-third of cases of ALK^- ALCL and PTCL-NOS.¹⁶ Overall, longer term follow-up is important in PTCL since, among patients alive and event-free at 24 months, the 5-year risk of subsequent relapse is still 23% for nodal PTCL, compared to only 7% in a prior study of diffuse large B-cell lymphoma.^17,18

There has been some debate on whether anthracyclines are essential for cure, particularly in PTCL-NOS and AITL,^{7,1 9} leading to studies exploring alternative combination therapy regimens. The UK National Cancer Research Institute (NCRI) Lymphoma Clinical study group compared CHOP to GEM-P (gemcitabine, solumedrol, cisplatin) in a randomized phase II study (CHEMO-T), but results were similar with a 2-year PFS of 37% versus 38%, respectively (P=0.82) and a 2-year OS of 51% versus 64%, respectively (P=0.31).²⁰ Although exploratory, patients with PTCL-NOS (as well as enteropathy-associated T-cell lymphoma) had a significantly better outcome with CHOP (odds ratio [OR]=0.036; P=0.012). In contrast, outcomes by treatment arm were similar in AITL (OR=0.69; P=0.578). The SWOG PEGS (cisplatin, etoposide, gemcitabine, solumedrol) regimen, built on the premise of using drugs not effluxed by multidrug resistance-1/permeability glycoprotein (MDR-1/P-gp), gave disappointing results with a 2-year PFS of only 12%, which may also reflect the absence of cyclophosphamide in the regimen. Although data are very limited, CHOP appears to be suboptimal in primary nodal EBV⁺ T/NK-cell lymphoma and it may be that regimens used in extranodal NK/T-cell lymphoma are more appropriate in this entity but this requires further study. Taken together, CHOP appears to be most important in ALCL, variably effective in PTCL-NOS especially given disease heterogeneity, and there is less certainty in AITL. Notably, there are limited data on outcomes with CHOP in the other TFHL subtypes.²¹ More intensive regimens, including the integration of etoposide, have also been explored in the front-line treatment of PTCL. This was first evaluated in a retrospective analysis of all patients with a PTCL diagnosis enrolled on prospective German High-Grade Non-Hodgkin Lymphoma Study Group (DSHNHL) studies (now called the German Lymphoma Alliance [GLA]).¹⁸ An improved event-free survival was observed in young (≤60 years), good-risk (normal lactate dehydrogenase level) patients who received etoposide with a front-line CHOP/CHOP-like regimen; however, the benefit was most evident in ALK⁺ ALCL (3-year event-free survival: 91% vs. 57%; P=0.01) with a trend observed in other nodal PTCL subtypes (3-year event-free survival: 61% vs. 48.3%; P=0.057). The findings did not translate into an OS benefit, and it was an unadjusted analysis.²² Other studies have supported improved outcomes with CHOP plus etoposide (CHOEP) over CHOP alone specifically in ALK⁺ ALCL,^9,12 but this is also the group that derives the greatest benefit from upfront BV-CHP as outlined below. Results with CHOEP are more mixed in the other nodal PTCL.^8,9 In the Netherlands Cancer Registry (NCR), there was no OS benefit of CHOEP over CHOP in a multivariate analysis when adjusted for the IPI score and use of auto-SCT.⁹ Taken together, CHOEP may still be considered a chemotherapy alternative in select patients but should be avoided in older patients given additional toxicity and lack of firm evidence of benefit.

Randomized front-line studies in PTCL

The earliest phase III randomized controlled study was conducted by the Groupe Ouest Est d’Etude des Leucemies et Autres Maladies due Sang (GOELAMS) comparing a doseintensive regimen, VIP-reinforced-ABVD (etoposide, ifosfamide, cisplatin; doxorubicin, bleomycin, vinblastine, decarbazine), to CHOP in newly diagnosed PTCL, including patients with ALCL (ALK status not specified), who represented almost half of enrolled patients (42/86). There was no difference in 2-year event-free survival which was 45% for all patients.²³

The most significant advancement to date in the treatment of nodal PTCL has been the impact of BV in ALCL. BV is an anti-CD30 antibody-drug conjugate linked to the anti-tubulin agent monomethyl auristatin E and was initially developed and approved for classical Hodgkin lymphoma and systemic ALCL in the R/R setting (see below). A phase I study established the safety and promise of CHP-BV in CD30⁺ PTCL (n=26 [ALCL n=19]: 5-year PFS 52%, 5-year OS 80%), with the omission of vincristine due to overlapping peripheral neuropathy with BV.²⁴ The subsequent practice-changing, double-blind, double-dummy, randomized phase III ECHELON-2 trial demonstrated improved PFS (3-year: 57.1% vs. 44.4%; P<0.011) and OS (3 year: 76.8% vs. 69.1%; P=0.024) with BV-CHP compared to CHOP, in 552 newly diagnosed patients with CD30⁺ PTCL (CD30 ≥10%; ALK⁺ ALCL (IPI score ≥2 only).²⁵ Results were maintained in the 5-year update (5-year PFS 51.4% vs. 43%; P=0.0077 and 5-year OS 70.1% vs. 61%; P=0.042).^25,26 The benefit was most striking in ALCL (5-year PFS 60.6% vs. 48.4%; P=0.0009 and 5-year OS 75.8% vs. 68.7%; P=0.053). As a regulatory requirement, the study was powered to evaluate ALCL, and the non-ALCL subtypes were under-represented (AITL n=54, PTCL-NOS n=72). As a result, the statistical comparisons for AITL and PTCL-NOS were unplanned and underpowered but since confidence intervals all crossed 1, the benefit of BV-CHP remains uncertain.²⁶ This also led to differences in regulatory approval of BV-CHP, with the USA Food and Drug Administration including the intent-to-treat population of CD30⁺ PTCL eligible for the study, Health Canada approving it for ALCL and CD30⁺ PTCL-NOS or AITL only, whereas in Europe and the UK, BV-CHP is only approved for ALCL. Several additional comments can be made about ECHELON-2. The study was restricted to cases with CD30 expression of ≥10% but regulatory approval did not specify a CD30 cutoff. A separate phase II study in CD30⁺ (<10%) PTCL is ongoing (NCT04569032) (Online Supplementary Table S1). Previous studies in R/R PTCL have not shown a good correlation between CD30 expression and response to BV, which may reflect the insensitivity of immunohistochemical detection or a broader mechanism of action, targeting the microenvironment.^27,28 In addition, information on the presence of DUSP22 and P63 rearrangements in ALK^– ALCL is not available, which could have an impact on prognosis. Some have also argued whether CHOP is the appropriate comparator. Although CHOEP is an option in PTCL, in the absence of randomized controlled studies, it has not replaced CHOP as a preferred standard therapy. The strongest retrospective data for the use of CHOEP are those for ALK⁺ ALCL, but this is also the group with excellent outcomes with CHP-BV in ECHELON-2 with a 5-year PFS of 87% versus 67% (P=0.0372), despite restriction to patients with an IPI score ≥2. Outcomes were also more favorable in ALK^– ALCL, although there is still room for improvement (5-year PFS 49% vs. 39%; P=0.0054).²⁶ A separate phase II study explored CHEP-BV (cyclophosphamide, doxorubicin, etoposide, prednisone plus BV) in CD30⁺ PTCL, including ALCL, with encouraging results (objective response rate [ORR] 91%, complete response [CR 80%]), but the rate of febrile neutropenia was 21% with routine granulocyte colony-stimulating factor support.²⁹ This regimen may be taken into consideration in very high-risk, younger ALCL patients especially those with an elevated central nervous system risk (e.g., high lactate dehydrogenase, involvement of multiple extranodal sites), given the additional penetration of etoposide across the blood-brain barrier, which does not appear to occur with BV.²⁹

Romidepsin and alemtuzumab have also been evaluated in combination with CHOP/CHOP-like therapy in phase III trials. Romidepsin is a selective histone deacetylase (HDAC) inhibitor and demonstrated modest efficacy in a phase II study in R/R PTCL (see below), ultimately leading to a phase III Lymphoma Study Association (LYSA) study evaluating romidepsin plus CHOP (Ro-CHOP) versus CHOP in previously untreated PTCL, excluding ALK⁺ ALCL.³⁰ The addition of romidepsin did not translate into an improved PFS in the intention-to-treat population (P=0.096) and was associated with increased toxicity.³⁰ The negative results have unfortunately led to de-listing of romidepsin in the R/R setting in the USA and Canada (see below). Alemtuzumab is an anti-CD52 humanized monoclonal antibody. CD52 antigen is a glycosylphosphatidylinositol-linked glycoprotein that is expressed on lymphocytes and monocytes and is variably expressed in PTCL.³¹ Alemtuzumab (A) and CHOP-14 (or CHOEP-14) versus CHOP-14 (or CHOEP-14) were evaluated in a collaboration between the Nordic and DSHNHL/GLA groups in parallel, phase III studies in younger (CHOEP-14, 18-65 years; ACT-1 trial)³² and older (CHOP-14, >65 years; ACT-2 trial)³³ patients with newly diagnosed PTCL, the former incorporating auto-SCT into both treatment arms. In ACT-2, the 3-year event-free survival was 27% versus 24% in the A-CHOP and CHOP arms, respectively (P=0.248) and the experimental arm was associated with significant toxicity.³³ Similarly, there was no benefit of A-CHOP observed in the ACT-1 trial (3-year event-free survival: A-CHOP 35% vs.26%).³²

Role of consolidative autologous transplant in front-line treatment of nodal PTCL: how strong is the evidence?

With the high relapse rate in PTCL, auto-SCT is often considered in first remission. However, in the absence of randomized controlled studies, there is a lack of consensus and, as a result, guidelines, as well as clinical practice, vary. This is highlighted in the randomized studies above in which use of consolidative auto-SCT was at the investigators’ discretion in the ECHELON-2 (blinded) trial, not allowed in the Ro-CHOP study and was integrated into both treatment arms in the younger patients in the ACT-1 study.^26,30,34 Overall, it is challenging to compare studies with the inclusion of diverse subtypes, variable responses leading into the transplant and analyses either in an ‘intent to transplant’ population or from the point of view of auto-SCT with the comparison group for the latter often including non-responders (Table 2).^35-37 The largest prospective study evaluating consolidative auto-SCT was conducted by the Nordic Lymphoma Group (NLG-T-01),³⁸ which enrolled 115 patients with newly diagnosed PTCL, excluding ALK⁺ ALCL, in 24 centers. Patients received CHOEP or CHOP-14 (>60 years) and 70% proceeded to auto-SCT. With a median follow-up of 5 years, the 5-year PFS and OS were 44% and 51%, respectively. With uncertainty around auto-SCT, the GLA and LYSA study groups conducted a phase III study comparing consolidative auto-SCT to allogeneic (allo)-SCT in newly diagnosed, poor-risk PTCL following four courses of CHOEP and one course of DHAP (dexamethasone, high-dose cytarabine, cisplatin); however, this study was stopped early due to futility.³⁹ Ultimately, with an overall transplant rate of 65%, the 3-year event-free survival was 43% in the allo-SCT group and 38% in the auto-SCT group. Notably, the 3-year cumulative incidence of relapse was 40% in the auto-SCT group and 0% in the allo-SCT group, supporting a graft-versus-lymphoma effect; however, non-relapse mortality was significantly higher in the allo-SCT group (3-year 23% vs. 0%), offsetting any overall benefit.³⁹ Retrospective studies have given conflicting results regarding the benefit of consolidative auto-SCT (Table 2).^8,9,40 The LYSA group performed a propensity score matched analysis of auto-SCT in an intention-to-treat population of nodal PTCL patients (n=269) and did not find a PFS or OS benefit of auto-SCT in multivariate analysis. In contrast, the NCR noted improved OS in patients with nodal PTCL (excluding ALK⁺ ALCL) treated in the more recent era when auto-SCT was more routinely applied. Furthermore, in a multivariate analysis of patients diagnosed between 2014-2018 who received CHOP or CHOEP, omission of auto-SCT with primary therapy was associated with a higher risk of death in patients with non-ALK⁺ ALCL subtypes.⁹ As this analysis was not adjusted for response, a separate 9-month landmark analysis and a sensitivity analysis were performed exclusively in CR patients, which also demonstrated improved OS in the auto-SCT group.⁹ Given the variability in interpretation and definition of a partial response in retrospective studies, some studies have evaluated the role of auto-SCT only in patients in CR at the end of treatment, again with mixed results (Table 2).^41,42 In a similar fashion, a subgroup analysis of patients in CR following BV-CHP enrolled on the ECHELON-2 study was performed and documented an improved PFS with consolidative SCT (auto-SCT n=36; alloSCT n=2) (5-year PFS: 65.3% vs. 46.4%, hazard ratio [HR]=0.36, 95% confidence interval [95% CI]: 0.17-0.77) but a similar benefit of SCT was not observed in the CHOP arm (Table 2).⁴³

Table 2.Selected large studies evaluating consolidative autologous stem cell transplant in nodal peripheral T-cell lymphomas.

The recently activated TRANSCRIPT trial will address the role of auto-SCT in patients with nodal PTCL (excluding ALK⁺ ALCL) in CR following induction therapy (NCT05444712) (Online Supplementary Table S1). Despite data limitations, consolidative auto-SCT should still be a strategy to consider with upfront treatment but guidelines differ about whether it should be performed exclusively in patients in CR.^34,44 Further studies are needed, ideally by subtype, to identify lower-risk patients in whom auto-SCT may be omitted and, conversely, determine whether there are molecular markers, such as P53 or DNMT3A mutations, that identify cases in which auto-SCT is futile.

Breast-implant-associated anaplastic large cell lymphoma

After its original description in 1997, breast implant-associated (BIA) was defined as a provisional entity in the revised 4th edition of the WHO classification (WHO-HAEM4R) and was upgraded to a distinct entity by both the ICC and the WHO-HAEM5^1,2 (Table 1 in the Introduction to the Review Series³). Although not a ‘nodal’ PTCL, given its primary extranodal location, it is described here to distinguish it from systemic ALCL as the work-up, management and prognosis differ.

The risk of BIA-ALCL is exclusively associated with textured implants and the time from implant to development of ALCL is 7-11 years. The overall risk varies in series but is likely between 1:1,000 to 1:10,000.⁴⁵ Peri-implant effusion is the most common presentation with 85% of patients having stage 1 disease limited to the seroma ± capsule. The effusion, preferably as a large volume, should be sent for cytology with cell block preparation and flow cytometry including CD30 in the panel. A positron electron tomography scan should be done before surgery as post-surgical inflammatory changes can complicate interpretation.⁴⁵ As recently reviewed,⁴⁵ the mainstay of treatment is implant removal and en-bloc complete capsulectomy, with bilateral removal if textured implants are used. Complete surgical excision in patients with stage 1 disease yields a 5-year disease-specific survival of 95%. For those presenting with stage 2 disease, there are limited data to guide recommendations. Surgery should include removal of the mass and sampling/removal of suspicious lymph nodes.⁴⁵ With incomplete resection, radiation may be considered and in rare cases adjuvant BV has been administered although data supporting this approach are lacking. Although patients with BIA-ALCL were not included in ECHELON-2, BV-CHP would be reasonable in those with disease outside of the breast and lymph nodes, or with lymph node involvement.⁴⁵

Treatment options in relapsed/ refractory PTCL and the promise of personalized therapy

Unfortunately, despite advances in the front-line setting a large proportion of PTCL patients have lymphoma relapse or have primary refractory disease. The only established curative treatment is SCT, although rare long-term remissions have been observed following systemic therapy alone, which, in some cases, may reflect more indolent disease biology.^46-48

With the emergence of genomic techniques, there is a greater understanding of underlying disease biology which has also helped to inform therapeutics. This is best shown in TFHL, which are typified by recurrent mutations in epigenetic modifiers,⁵ with growing evidence of sensitivity to a broad range of agents of this class (Tables 3-5). In a proportion of ALCL and other rarer PTCL subtypes, there is evidence of JAK/STAT pathway activation, leading to recent trials with JAK inhibitors.⁴⁹ Although studies of PTCL-NOS have elucidated the GATA3 and TBX21 molecular subtypes, how this informs treatment decisions remains unknown. Here, we review the overall management of R/R nodal PTCL, highlighting situations in which biology can guide treatment options.

Transplant or no transplant?

Outcomes are historically poor in patients with R/R PTCL, with a median OS from first relapse/progression typically <6 months in those who are not transplanted.⁴⁶ Thus, the first therapeutic decision is whether or not a patient is a transplant candidate. SCT is limited to fit, often younger patients with chemosensitive disease, a term that should be redefined as ‘systemic therapy-sensitive’ with the expanding compendium of modern, novel therapies that also serve as an effective bridge to SCT.^50-53 The prospective International T-Cell Project collected data on 633 patients with relapsed (n=197) or refractory (n=436) PTCL, including those managed with intent to transplant; the median OS for all patients was still only about 6 months, and the 3-year OS was 23%. Overall, only 99 patients (16%) underwent SCT (type not specified) as part of salvage therapy and, not surprisingly, this group had a superior 3-year OS of 48% compared to 30% in patients in a partial remission or CR and were not transplanted (for any reason).⁵⁴ Data are more limited in patients managed with ‘intent to transplant’, especially with auto-SCT. Two retrospective, single-institution studies evaluating outcomes from the point of relapse/progression with intention to incorporate SCT suggested a cure rate of 20-35% with auto-SCT, with dismal outcomes in patients with refractory disease.^55,56

Outside of ALCL, most evidence supports the use of alloSCT in relapsed and especially refractory PTCL. The Center for International Bone Marrow Transplant Research (CIBMTR)⁵⁷ evaluated outcomes in 241 patients with PTCL undergoing SCT between 1996 to 2006 and 2018. Confining the analysis to those beyond first CR, a superior outcome was observed with auto-SCT over allo-SCT (3-year OS 62% vs. 33%, respectively; P=0.0088) with a lower transplant-related mortality (5% vs. 32%; P=0.0088), but PFS and relapse/progression rates were similar. This study was largely driven by a high proportion of ALCL patients who may derive the greatest benefit from auto-SCT in the relapse setting (3-year PFS 53%, 3-year OS 65%) compared to PTCL-NOS (3-year PFS 29%, 3-year OS 42%), although specific outcomes by ALK status were not reported. Only six patients with AITL underwent auto-SCT, limiting the evaluation of this group. The proportion of patients with refractory disease was not specified by subtype, a factor that also strongly influences the information on the utility of auto-SCT.

Studies have highlighted favorable outcomes with allo-SCT particularly in R/R AITL, including a recent combined retrospective registry study from the European Society for Blood and Marrow Transplantation (EBMT) and the CIBMTR, which evaluated 1,942 PTCL patients (AITL, PTCL-NOS, ALCL) undergoing allo-SCT between 2008 and 2018 primarily with R/R disease (70%).⁵² Overall, the 3-year PFS was 50% and the 3-year OS was 60%, highlighting better outcomes in the more modern treatment era. Furthermore, using PTCL-NOS as the reference group, a reduced risk of lymphoma relapse (P<0.001) was observed for AITL, highlighting a unique sensitivity to the graft-versus-lymphoma effect, which is consistent with other studies.^58-60 In contrast, an increased risk of relapse was observed for ALCL (HR=1.3, 95% CI: 1.1-1.6; P=0.01). Not surprisingly, patients in CR had better outcomes than those with a partial response or resistant disease (3-year PFS 57% vs. 47% vs. 36%, respectively; P<0.0001) with response remaining significant in a multi-variate analysis. Although efforts should be made to achieve a deeper response, given that one-third can still be cured, patients should not be denied an allo-SCT if criteria for a partial response are not met. Interestingly, outcomes after haplo-SCT and matched donor transplants were comparable, and use of intensified conditioning did not appear to be advantageous compared with non-myeloablative approaches.⁵² Collectively, studies suggest that auto-SCT may be considered in relapsed ALCL if not used with front-line therapy, but allo-SCT should be favored in AITL and patients with PTCL-NOS should most likely receive an allo-SCT. Apart from some cases of ALK⁺ ALCL, for those with refractory disease, allo-SCT would be preferable. As outlined by Bisig, Savage and de Leval,⁵ the characteristic landscape of TFHL supports a multi-step process derived from underlying clonal hematopoiesis (see Figure 4 in the review published in this issue of Haematologica⁵). TET2 and DNMT3A mutations, which typify TFHL, are also common in clonal hematopoiesis. A recent study using bone marrow samples as well as flow-sorted bone marrow and peripheral blood samples supported the finding that clonal hematopoiesis is prevalent in AITL and showed that progenitor cells harboring identical TET2 and DNMT3A mutations can divergently evolve to AITL and myeloid neoplasms.⁶¹ With still limited data, it remains unclear whether the presence of clonal hematopoiesis should affect treatment choices. However, if available, testing for clonal hematopoiesis should ideally be performed as it may inform on the use of auto-SCT in older patients and whether an allo-SCT may be preferred in some settings, as well as follow-up surveillance.

Is there a preferred salvage chemotherapy in transplant-eligible relapsed/refractory PTCL?

As for diffuse large B-cell lymphoma, there is no standard first-line salvage therapy for R/R PTCL patients, and very few studies have detailed outcomes specifically in PTCL. A subgroup analysis of the Canadian Cancer Trials Group phase III LY.12 study comparing GDP (gemcitabine, dexa-methasone, cisplatin) and DHAP salvage chemotherapy before auto-SCT in R/R aggressive lymphomas, including PTCL (n=59), demonstrated an ORR of 36% with GDP (compared to 46% in aggressive B-cell lymphomas; P=0.12) which was similar to that achieved by DHAP.^62,63 The population was high risk with most patients either having refractory disease (41%) or had relapsed within 1 year (37%), which may have contributed to overall lower response rates, regardless of the study arm. ICE (ifosfamide, carboplatin, etopsoside) chemotherapy is also frequently used,⁵⁵ however, there are no comparative studies in PTCL. Given the high frequency of chemorefractory disease, novel agents have been increasingly used as a bridge to transplant as outlined below.

Novel agents in the management of relapsed/refractory PTCL

Over the past decade, there has been a pivot to perform novel therapy studies specifically in R/R PTCL and, more recently, even in specific subtypes. The majority are singlearm phase II studies and, apart from BV, drug approval may be country specific (Table 2 and Figure 1 from the Introduction to review Series³). The scope of agents under investigation is wide but very few are approved (Table 3).³ Pralatrexate, a folate analog metabolic inhibitor which competitively inhibits dihydrofolate reductase, was the first novel agent to be studied in PTCL after early studies showed preferential sensitivity in PTCL (including cutaneous T-cell lymphoma) compared to B-cell lymphomas.⁶⁴ The PROPEL study⁵¹ evaluated 115 patients with R/R PTCL and the ORR for all patients was 29% (11% CR), the median PFS was 3.5 months, and the median duration of response was 10.1 months (Table 3). The response rate was notably lower in AITL (8%). Subsequent studies explored a different ramped up dosing schedule and use of leucovorin to mitigate mucositis (‘Columbia regimen’) which improved tolerance and appears to maintain efficacy.⁶⁵ BV was developed for use in Hodgkin lymphoma and ALCL due to the disappointing efficacy of the nascent anti-CD30 antibody in these lymphomas. In the phase II registration study for systemic ALCL, patients were eligible following failure of front-line anthracycline-based therapy. The efficacy was striking with an ORR of 86% and CR rate of 57%. The median PFS and median duration of response were 13.3 months and 12.6 months, respectively⁴⁷ (Table 3). In the 5-year follow-up, the PFS was 39% and, overall, 14% of patients remained in CR following single-agent BV in the absence of transplant, suggesting that cure was possible in a minority of patients.⁵² This led to global approval of BV in R/R ALCL (Table 3). In a separate study, BV was evaluated in R/R non-ALCL CD30⁺ PTCL and, although less impressive, efficacy was demonstrated with an ORR of 41% across the entire cohort (n=34) and 54% in AITL, but duration of response was short (all patients 7.6 months). Although not approved, it remains an option in R/R CD30⁺ PTCL especially as a bridge to SCT if funding is available.⁶⁶ More recently, the ALK inhibitor crizotinib was approved by the Food and Drug Administration for patients up to 21 years old with ALK⁺ ALCL, based on a robust CR rate (81%)⁶⁷ (Table 3). Alisertib is an aurora A kinase inhibitor that produced an ORR of 50% in eight PTCL patients enrolled in a phase I study of hematologic malignancies.⁶⁸ This study was followed by the Lumiere study, which was the first randomized phase III study in R/R PTCL, and compared alisertib to investigators’ choice of therapy (pralatrexate, romidepsin, gemcitabine) (Table 4). It did not show superiority of alisertib (ORR 33% [alisertib] vs. 45% [comparators]; median PFS 3.8 months vs. 3.5 months, respectively). This study demonstrated that phase III trials were possible in this setting but also highlighted the challenges with disease heterogeneity.⁶⁹ CCR4 is expressed in 30-40% of cases of PTCL and is associated with the GATA3 subtype of PTCL-NOS. Mogamulizumab is a CCR4 monoclonal antibody and was explored in R/R CCR4⁺ nodal PTCL in a phase II study in Japan. Among 29 patients, the ORR was 34% with CR in 17%; the median PFS was 3 months (Table 3) and led to the approval of mogamulizumab in 2014 in Japan in CCR4⁺ nodal PTCL.¹¹⁴

Table 3.Approved novel agents for the treatment of relapsed/refractory peripheral T-cell lymphomas: global perspective.

Epigenetic therapies and a personalized treatment approach in T-follicular helper cell lymphomas

Epigenetics reflect changes in gene expression in the absence of DNA sequence changes and include histone modification, DNA methylation, noncoding RNA effects and chromatin reorganization. Abnormal activity of HDAC can affect gene expression with epigenetic silencing of tumor suppressor genes and oncogene activation. Aberrant epigenetic alterations play a role in the pathogenesis of TFHL.

Recurrent mutations in epigenetic modifying genes, including TET2, DNMT3A and IDH2^R17 as well as the disease-specific RhoA^G17V mutation, characterize TFHL. As a group these lymphomas respond better to epigenetic therapies (Figure 1 in the Introduction to the Review Series³), showing much higher response rates than those achieved when considering all PTCL (Tables 3 and 4). Histone modifier gene mutations have been reported in 36% of PTCL-NOS and are associated with inferior PFS but may also define a group with an increased response to the HDAC inhibitor chidamide, suggesting that there may be a larger scope of patients who could benefit from a more personalized approach to therapy.⁷⁰

Romidepsin was the first HDAC inhibitor approved for use in R/R PTCL and was associated with an ORR of 25% (CR 15%) (Table 3).⁵⁰ Although responses were infrequent and overall median PFS was only 3 months, some responses were notably durable with a median duration of 28 months with longer follow-up.^50,71 Furthermore, 4/27 (15%) of patients enrolled with AITL still remained in CR over 3 years after entering the study.⁴⁸ Unfortunately, in 2021, the PTCL indication for romidepsin was withdrawn from the USA market and more recently Canada has followed, due to the negative results of a phase III study evaluating RoCHOP in the first-line setting. Belinostat, a hydroxamic acid-derived pan-HDAC inhibitor, and chidamide, the only oral class I/II HDAC inhibitor, are approved in the USA and China, respectively, and have similar efficacy to romidepsin and also a higher ORR in AITL (ORR 46% with belinostat; 50% with chidamide) (Table 3).^72,73 A retrospective multicenter study that compared the efficacy of HDAC inhibitors in TFHL versus PTCL-NOS confirmed a higher response rate in the former (ORR and CR 56.5% and 28.9%, respectively, in TFHL vs. 9.4% and 19.6%, respectively, in PTCL-NOS; P=0.0035)⁷⁴ and in those PTCL cases with ‘typical’ AITL mutations as described above. Beyond HDAC inhibitors, other epigenetic therapies produce high response rates in TFHL (Table 4; Figure 1 in the Introduction to the Review Series³). Oral 5-azacitidine (CC-486) is a hypomethylating agent that inhibits DNA methyl-transferase and was first evaluated in a retrospective study of 12 patients with R/R AITL, five of whom had a concurrent myeloid neoplasm; the ORR was 75%, the CR rate was 50% and the median PFS was 15 months.⁷⁵ This led to the recently reported ORACLE phase III study (NCT03593018) comparing oral azacitidine (n=42) to investigators’ choice of therapy (n=44) (romidepsin n=4, gemcitabine n=24, bendamustine n=16) in TFHL.⁶¹ The median PFS favored 5-azacitidine (5.6 months vs. 2.8 months; P=0.042) but did not reach the pre-specified significance level of P<0.025, suggesting that the study may have been underpowered. Interestingly, despite the favorable PFS, lower ORR and CR rate were observed with 5-azacitidine (ORR and CR 33% and 12%, respectively, with 5-azacitidine vs. 43% and 23%, respectively, with investigators’ choice of therapy) supporting that a greater proportion of patients may have had stable disease as best response, which is also reflected in a more favorable OS (median OS 18.4 months vs. 10.3 months) (HR=0.56, 95% CI: 0.323-0.961).⁷⁶ A separate, ongoing, phase III study in Japan is comparing oral azacitidine to investigators’ choice of therapy (romidepsin or gemcitabine) in R/R AITL, but the results have not yet been reported (NCT03703375) (Online Supplementary Table S1).

Table 4.Selected novel agent/combination therapy in relapsed/refractory peripheral T-cell lymphomas.

Valemetostat is a potent, selective dual inhibitor of enhancer of zeste homologs (EZH2 and EZH1) and stimulates the expression of pro-apoptotic and tumor suppressor genes (Table 4; Figure 1 in the Introduction to the Review Series³).⁷⁷ A phase I dose-escalation trial reported an ORR of 55.6% (24% CR) and 70.6% in AITL (Table 4).⁷⁸ The phase II VALENTINE-PTCL01 trial (NCT04703192) has completed accrual, but results have not yet been reported. There is also emerging evidence to support the integration of epigenetic therapies into the front-line therapy of TFHL. An unplanned subgroup analysis of patients with TFHL enrolled on the LYSA Ro-CHOP study demonstrated a PFS benefit of Ro-CHOP versus CHOP (P=0.046)³⁰ (Table 5). Belinostat-CHOP was evaluated in a phase I study of 23 patients, with AITL as the predominant subtype (n=10, 43%); the ORR was 86% and the CR rate was 57%.⁸¹ In a phase I study of vorinostat-CHOP, all evaluable patients (n=12) achieved a CR, providing additional support for this approach.⁷⁹ Several additional, ongoing studies are integrating chidamide in the front-line setting (Online Supplementary Table S1).

Similarly, a phase II study evaluated CHOP with oral 5-azacitidine (CC-486) in 20 newly diagnosed patients with PTCL, of whom 81% had TFHL.⁸⁰ The azacitidine was given as ‘priming’ to enhance chemosensitization and deepen the response. The ORR and CR rate was 75%, rising to 88% for both in TFHL. Overall, half of all patients underwent consolidative auto-SCT and, with a median follow-up of 21 months, the 2-year PFS was 65.8% for all patients and 69.2% in patients with TFHL.⁸⁰ The presence of TET2 mutations was associated with a more favorable PFS (P=0.014), whereas a trend towards an inferior PFS was observed in cases with a DNMT3A mutation. Building on this regimen, the Alliance group launched a randomized phase II study comparing CHOP (or CHOEP for patients <60 years old) with either the same backbone with oral azacitidine or duvelisib in a 1:1:1 design (Online Supplementary Table S1).

Not surprisingly, adding a novel therapy to the CHOP backbone is associated with additional toxicity. Febrile neutropenia occurred in 21% and 10% of patients treated with Ro-CHOP and CHOP, respectively (Table 5).⁸¹

Beyond epigenetic therapies: promise in T-follicular helper lymphomas and beyond

Lenalidomide is an immunomodulatory drug with a complex mechanism of action including direct effects on the tumor cell mediated by inhibition of the protein CEREBLON as well as anti-angiogenic and immunogenic effects through its impact on the tumor microenvironment (Figure 1 in the Introduction to the Review Series³). Previous studies have shown modest single-agent activity in R/R PTCL (ORR 22-30%, CR 0-11%) and a median PFS of approximately 3 months across studies^82-85 including patients with AITL (ORR 31%, CR 15%).⁸² A phase II study of CHOP and lenalidomide in patients 60-80 years old with AITL demonstrated a CR rate of 41%, which was below the pre-specified target of 55%, although the 2-year PFS and OS rates of 42% and 59% were better than expected based on results from historical series.⁸⁶

Table 5.Studies evaluating CHOP + epigenetic therapies versus epigenetic therapy combinations in treatment naïve PTCL: focus on T-follicular helper lymphomas.

Duvelisib is a dual inhibitor of PI3K-δ and PI3K-γ and showed encouraging activity in a phase I study in which the ORR was 50%.⁸⁷ The findings of the phase II PRIMO registration trial of duvelisib in R/R PTCL were reported at the 2021 American Society of Hematology (ASH) meeting. The ORR was 50% (32% CR); results were not detailed by histological subtype (Table 4).^88,89 The GATA3 molecular subtype of PTCL-NOS is enriched for PI3K-induced signatures providing a potential rationale for duvelisib (Table 4; Figure 1 in the Introduction to the Review Series³). PI3K inhibitors can cause immune-mediated toxicities reminiscent of those caused by checkpoint inhibitors, and these toxicities have led to discontinuation of drug development in many settings in B-cell lymphomas/leukemias. In the PRIMO study, duvelisib was associated with grade 3/4 transaminitis in 27% of patients and pneumonitis in two patients, both of whom died.⁸⁹ However, immune toxicities may be favorably modified by certain concurrent therapies (see below).

Tipifarnib is an oral inhibitor of farnesyltransferase which reduces CXCL12, a chemokine that is highly expressed in AITL and some PTCL-NOS. In a phase II study of tipifarnib for the treatment of PTCL the ORR was 39.7% but was 56.3% in AITL (CR 28.5%) and wildtype CXCL12 genotype in PTCL-NOS was predictive of response (ORR 40%).⁹⁰ With activation of the JAK/STAT pathway being a feature in many PTCL, including ALK⁺ ALCL and a subset of ALK^– ALCL, a JAK inhibitor (ruxolitinib) and a dual JAK/SYK inhibitor (cerdulatinib) have also been explored (Table 4; Figure 1 in the Introduction to the Review Series³).^49,91 Cerdulatinib produced an ORR of 35% across R/R PTCL, increasing to 52% for those with a TFH phenotype.⁹¹ Ruxolitinib was associated with an ORR of 33% in cases with activating mutations in JAK/STAT (cohort 1) and 29% in those pSTAT3-positive by immunohistochemistry (cohort 2).⁴⁹ A preliminary report of a phase I/II study of golidocitinib (JAKPOT8), a selective JAK1 inhibitor, in R/R PTCL described an ORR of 43% (65% in AITL), a CR of 22% and a median duration of response that had not been reached at that time (Table 4).⁹²

EBV is present in most AITL (70-80%) in the surrounding B-cell immunoblasts, as well as in a proportion of PTCLNOS, although some of the latter cases would be re-classified today as primary nodal EBV⁺ T-cell/NK-cell lymphoma by the WHO-HAEM5 if also positive for cytotoxic markers. Regardless, nanatinostat, which induces EBV kinase genes, and valganciclovir which is subsequently activated (i.e., ‘kick and kill’), are being evaluated in a phase II basket study (NAVAL-1) (Figure 1 in the Introduction to the Review Series³) including a cohort with nodal (non-ALCL) PTCL (Online Supplementary Table S1) following an initial report of an ORR of 40% (CR 19%). In the six patients with PTCL-NOS/AITL from the phase I study, the ORR was 67% (CR 50%).⁹³

Are combination novel therapies ready for prime time in treatment-naïve patients?

In an effort to deepen CR, activity in R/R PTCL trials has centered around combination therapies chosen to capitalize on complementary, additive or synergistic activities. In some cases, these chemotherapy-free combination therapies are also being evaluated in treatment-naïve cohorts, challenging the paradigm of upfront CHOP-based approaches (Table 5). A phase I study of romidepsin and 5-azacitidine noted sensitivity in T-cell lymphomas with an ORR of 73% in five patients in an expansion cohort, four of whom had a CR (AITL n=3).⁹⁴ A phase II study followed in PTCL including both treatment-naïve and R/R PTCL cohorts, with the analysis including the five patients from the earlier study.⁹⁵ Considering the evaluable patients, the ORR was 61% (CR 43%) and was 70% (CR 50%) in treatment-naïve patients (n=10) versus 54% (CR 38%) in R/R PTCL patients. With a median follow-up of 13.5 months, the median PFS for all patients was 8 months.⁹⁵ For the TFHL subgroup, the ORR was 80% and the CR was 67%, with a median PFS of 8.9 months versus 2.3 months in other PTCL subtypes. Considering all patients, grade 3/4 thrombocytopenia occurred in 48%, grade 3/4 neutropenia in 40%, and febrile neutropenia in 12% (Table 5). Targeted mutation information was available for 15 patients: numerically higher ORR and CR rate were demonstrated in those with TET2 mutations (69% and 53%, respectively) compared to those with a wildtype genotype (40% and 20%, respectively); however, the differences did not reach statistical significance. A retrospective series evaluating azacitidine (oral or subcutaneous) and romidepsin demonstrated similar favorable efficacy in 26 patients with R/R PTCL, 23 of whom had TFHL, of which one was a composite with diffuse large B-cell lymphoma. The ORR was 76.9% (CR 53%) in the TFHL subgroup.⁹⁶ The combination of azacitidine and romidepsin is being compared to investigators’ choice of therapy (belinostat, pralatrexate, gemcitabine) in an ongoing phase III trial (Online Supplementary Table S1). Romidepsin and lenalidomide were evaluated in a phase I study in treatment-naïve PTCL patients ≥60 years or those <60 years and not considered chemotherapy candidates by the treating physician. Of 20 evaluable patients, 13 (65%) had AITL. Overall the ORR was 75% (CR 30%), rising to 85% (CR 38.5%) in the patients with AITL⁹⁷ (Table 5). Although not evaluated in a treatment-naïve population, a phase I study of duvelisib with romidepsin in R/R patients demonstrated encouraging efficacy (ORR 58%/42%; TFHL 68%/58%) but interestingly, also less hepatic toxicity than lead-in treatment with duvelisib alone (40% vs. 8%), suggesting that romidepsin may offset the immunotoxicity of duvelisib.

Long-term follow-up is needed to determine the curative potential of these chemotherapy-free approaches and it is important to note that the toxicities are not negligible (Table 5). There are ongoing studies evaluating both combination therapies as well as CHOP + novel therapy approaches in treatment-naïve nodal PTCL.

What is the promise of immunotherapy in PTCL?

The efficacy of therapy targeting the programmed death pathway, either through programmed cell death protein 1 (PD1) or its ligand (PDL1), has been well described in extra-nodal NK/T-cell lymphomas⁸⁵ and with upregulation of PDL1 in primary nodal EBV⁺ T/NK-cell lymphoma, it may also be a potential therapeutic approach in this rare entity.⁴ There are limited data for the remaining nodal PTCL, but in all PTCL, there is a potential concern about hyperprogression given that PD1 on T cells may function as a tumor suppressor.⁹⁸ In a phase I study of 12 patients with R/R PTCL, half of whom had AITL, the ORR was only 33% and median PFS <3 months with hyperprogression reported to occur in four patients.⁹⁹ A combination study of romidepsin and pembrolizumab showed more encouraging efficacy (ORR 47.3%, CR 37%), but hyperprogression was reported in two patients, so further studies are needed.¹⁰⁰ Cellular therapy is still in the development phase in PTCL and a full review is beyond the scope of this article, but some of the ongoing studies are highlighted in Online Supplementary Table S1. The majority of studies in PTCL have focused on CD30 as a target and some have reported CR in ALCL but the numbers of patients and the follow-up time are limited and thus the curative potential is still unknown.^101,102 Third-generation products utilizing the CD28 (CD28z) co-stimulatory domain may persist long term and appear to have a more potent anti-tumor effect.¹⁰² Apart from CD30, chimeric antigen receptor (CAR) T-cell targets in PTCL have been challenging due to three main barriers: (i) T-cell aplasia; (ii) fratricide; and (iii) the potential for contamination of CAR T-cell products with malignant cells.¹⁰³ Strategies to circumvent fratricide include capitalizing on the selection of either TRBC1 or TRBC2 for the β-chain constant region to spare normal T cells¹⁰⁴ as detailed in the phase I/II AUTO4 trial (NCT03590574) evaluating TRBC1⁺ PTCL as well as use of NK-cell CAR products which do not express T-cell antigens.¹⁰⁵ Early results from AUTO4 demonstrated that five of nine patients with PTCL achieved a CR, but lack of CAR T-cell expansion may limit the durability of responses.¹⁰⁶ Use of ‘off the shelf’ products, such as allogeneic CD70 CRISPR-Cas9-engineered T cells from healthy donors which incorporate editing of T-cell receptor α and β2-microglobulin genes (NCT04502446 COBALT-LYM) avoids normal T-cell killing and does not rely on personalized manufacturing. AFM13, an innate bispecific CD16/CD30 antibody, is also under investigation in CD30⁺ PTCL (REDIRECT NCT04101331), including ALCL, demonstrating ORR of 40% in a phase Ib/II study (NCT03192202).¹⁰⁷

Conclusions

Advances in the understanding of the biology and molecular underpinnings of PTCL have refined both classifications and therapeutic approaches. BV-CHP has changed the treatment landscape in ALCL and may be considered in selected cases of CD30⁺ PTCL-NOS/AITL but it is recognized that the data are not definitive in the latter subtypes. Auto-SCT still remains a possible choice in upfront treatment, but definitive recommendations are difficult to make in the absence of randomized data. Recent studies highlight the sensitivity of TFHL to epigenetic therapies and, in the future, genomic information may also inform therapy. Future studies should focus on the evaluation of new treatments in specific PTCL subtypes or molecularly defined subgroups, to further refine personalized therapeutic options across a broader range of PTCL.

Footnotes

• Received May 18, 2023
• Accepted August 14, 2023

Correspondence

^°Current address: Department of Hematology, Auckland City Hospital, Auckland, New Zealand

K.J. Savage ksavage@bccancer.bc.ca

Disclosures

HSN has no conflicts of interest to disclose. KJS has received honoraria from and provided consulting for BMS, Merck, Seagen, and Janssen; has sat on a steering committee for Beigene; has received research funding from BMS; has received institutional research funding from Roche; and sat on a Data Safety and Monitoring Committee for DSMC.

Contributions

KJS and HSN co-wrote the paper.