Although the majority of patients with follicular lymphoma have a life expectancy similar to that of the general population, predicting which patients will relapse and finding therapies that provide durable responses in the relapsed or refractory setting remains an ongoing challenge.
John Pagel, MD, PhD
Although the majority of patients with follicular lymphoma (FL) have a life expectancy similar to that of the general population, predicting which patients will relapse and finding therapies that provide durable responses in the relapsed or refractory (R/R) setting remains an ongoing challenge, according to oncologists specializing in hematologic malignancies who participated in a recent OncLive® workshop held on July 29, 2020, led by John Pagel, MD, PhD, of the Swedish Cancer Institute in Seattle.
FL is the second-most-common subtype of non-Hodgkin lymphoma (NHL) and accounts for 20% to 25% of new diagnoses in Western countries.1 The age-adjusted rates of new cases and deaths were 2.7 and 0.5 per 100,000 individuals per year, respectively, based on data from the Surveillance, Epidemiology, and End Results (SEER) program from 2013 to 2017.2 FL tends to be more common in older adults, with a median age of 60 to 65 years at diagnosis.1 The overall survival (OS) has increased since the 1990s, with a 5-year relative survival of 89% based on data from the SEER 18 registry spanning 2010 to 2016.2
The World Health Organization grading system categorizes FL based on cytologic features.1 Pagel noted that patients with grades 3A and 3B FL often have more aggressive disease and may require a different management strategy than do patients with grade 1 or 2 FL. Grades 1 and 2 FL have a similar Ki67 index (<20%) and phenotypic and cytogenic features but differ in the number of centroblasts per high-powered field (0-5 and 6-15 centroblasts for grades 1 and 2 FL, respectively).1 Grades 3A and 3B FL have a Ki67 index greater than 20% and more than 15 centroblasts per high-powered field, but 3A FL is characterized by centrocytes with expression of CD10 and BCL2 and translocation of BCL2 in about 75% of cases, whereas grade 3B FL is characterized by diffuse areas of centroblasts and usually a lack of BCL2 translocation or lack of expression of CD10 and BCL2.1
Prognostic Scores and Predicting Risk for Relapse
The Follicular Lymphoma International Prognostic Index (FLI- PI) was published in 2004 and is based on results from an international cooperative study that analyzed 4167 cases of FL diagnosed initially between 1985 and 1992.3 Based on the study team’s retrospective analyses, the FLIPI model was built based on age at diagnosis, Ann Arbor stage, hemoglobin level, number of nodal areas involved, and serum lactate dehydrogenase level.3 Classification of patients as low (0-1 factor), intermediate (2 factors), or high risk (3 or more factors) predicted OS at 5 and 10 years.3 The FLIPI index has also been shown to predict OS in the chemoimmunotherapy era, with hazard ratios (HRs) for OS of 2.37 and 6.17 in the intermediate- and high-risk groups, respectively, compared with the low-risk group in an analysis of 2192 patients enrolled in the National LymphoCare Study (NCT00097565).4
However, the FLIPI index may have limited clinical utility because the initial cohort may not reflect the current course of the disease and because the index was based on a retrospective analysis of archive data. To improve on this predictive model, investigators prospectively collected data of 942 patients from 2003 to 2005 to build the FLIPI2 index, a model that includes age, body mass index, hemoglobin level, greatest diameter of the largest involved node, and serum β2 -microglobulin level.5 The 3-year progression-free survival 2 (PFS) was 91%, 69%, and 51% in the patients classified as low (0 factors), intermediate (1-2 risk factors), and high risk (3-5 risk factors), respectively, by the model.5
The PRIMA-prognostic index scoring system was recently introduced to simplify the prognostic scoring system and included bone marrow involvement and β2 -microglobulin level. In the 2 prospective PRIMA trial cohort of patients treated with chemoimmunotherapy, the 5-year PFS was 69%, 55%, and 37% in the low-, intermediate-, and high-risk groups, respectively.6
Introduced in the late 2010s was the m7-FLIPI score, calculated as the sum of predictor values weighted by Lasso coefficients of the FLIPI index score, Eastern Cooperative Oncology Group performance status, and nonsilent mutations in EZH2, ARID1A, EP300, FOXO1, MEF2B, CREBBP, and CARD11 genes. In the training cohort of 151 patients with FL, high- and low-risk groups identified by the m7-FLIPI score had 5-year failure-free survivals of 38.29% and 77.21%, respectively. The m7-FLIPI also outperformed a prognostic model that included only gene mutation status, as well as the FLIPI-2 index.7 However, results of a recent study did not show a difference in time to treatment failure or OS between m7-FLIPI high- and low-risk groups in patients with FL who received rituximab with or without interferon alfa-2a.8
Although Peter Martin, MD, of Weill Cornell Medicine in New York, New York, said that he calculates FLIPI scores for his patients, the scores do not directly affect his management decisions. “It’s the sort of thing you report and then move on with life. You manage the person according to the way that you think that person needs to be managed,” said Martin.
Unmet Needs in POD24 and Aggressive Disease
Despite the good survival outcomes in most cases of FL, progression of disease within 24 months of diagnosis (commonly known as POD24) occurred in 19% of patients with stage II to IV FL who received first-line rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP) and was associated with poorer OS from a risk-defining event at 2 and 5 years compared with patients who did not have POD24 (68% vs 97% at 2 years; 50% vs 90% at 5 years).9 Additionally, Mitchell R. Smith, MD, PhD, of George Washington Cancer Center and George Washington University in Washington, DC, said that the ability to predict the approximately 20% of patients who relapse with aggressive disease remains an unmet need and that analyzing samples collected at initial diagnosis to identify predictors of relapse should be a priority moving forward.
The panelists also discussed the role of PET-CT scans at initial diagnosis for predicting which patients are most likely to relapse with aggressive disease or have disease transformation. Alan Skarbnik, MD, of Novant Health Care Institute in Charlotte, North Carolina, said that he performs a biopsy of the “hottest spot” identified by PET-CT scan and performs repeat biopsies in patients with POD24 to obtain additional in- formation about disease behavior, as he pointed out that these patients also have an increased risk for disease transformation.
Nathan Fowler, MD, of The University of Texas MD Ander- son Cancer Center in Houston, added that he performs PET-CT scans on all his patients at initial diagnosis to establish a baseline, identify extranodal disease, and possibly upstage some patients. “If you’re thinking about radiation or limited-stage radiation, the PET sometimes will move you away from that when you find the sites of distal disease,” he said.
Approximately 10% to 30% of patients with FL present with stage I or II disease,1 and radiation therapy yields 10-year OS rates of 60% to 80% and a median OS of about 19 years.10 With advanced disease, Pagel pointed out that treatment, although not a cure, can be instrumental in prolonging survival. Front- line regimens generally involve chemoimmunotherapy combinations, although nonchemotherapy approaches used in the R/R setting have recently been introduced in this setting and may provide a less toxic option for select patients.
Two phase 3 trials have introduced bendamustine as a more tolerable and efficacious chemotherapy component than R-CHOP or rituximab, cyclophosphamide, vincristine, and prednisone (R-CVP). The phase 3 StiL trial (NCT00991211) showed superior PFS and fewer toxic events with rituximab- bendamustine than with R-CHOP in patients with previously untreated stage III or IV indolent NHL or mantle cell lymphoma.11 Similarly, the phase 3 BRIGHT trial (NCT00877006) showed a higher PFS rate at 5 years with rituximab-bendamustine than with R-CHOP or R-CVP in patients with indolent NHL or mantle cell lymphoma.12
During the workshop, Pagel noted that the newer anti-CD20 antibody obinutuzumab provides an option besides rituximab for anti-CD20 therapy, although the superiority of one agent over the other is unclear. The phase 3 GALLIUM trial (NCT01332968) showed superior PFS with obinutuzumab plus chemotherapy than with rituximab plus chemotherapy (estimated rate at 3 years, 80.0% vs 73.3%, respectively) in the intention-to-treat population of patients with FL, although OS was not different and the rates of grade 3 or higher adverse events (AEs) and of severe AEs were higher in the obinutuzumab arm.13 A subsequent analysis showed that the obinutuzumab arm had a lower cumulative incidence rate of progressive disease or death due to progressive disease at 24 months, corresponding to a 46% average HR-based reduction in risk for a POD24 event.14 Another analysis of data from the GALLIUM trial showed that obinutuzumab was consistently associated with superior investigator- and independent review committee (IRC)–assessed PFS across the different chemotherapy backbones (CHOP, CVP, or bendamustine) at a median follow-up of 41 months (although the chemotherapy regimens were not randomized).15
Although lenalidomide plus rituximab (R2) is commonly considered in the R/R setting, data from the phase 3 RELEVANCE trial (NCT01476787) suggest that the regimen also has efficacy in the frontline setting. In the trial, PFS and OS were similar between R2 and rituximab plus chemotherapy in patients with previously untreated FL.16
“This has been well described as the most positive negative study,” said Pagel. “I like thinking of it that way because there are patients who are certainly going to benefit from [R2], even in the frontline [setting].”
The synergistic effects of lenalidomide and rituximab and superior efficacy of obinutuzumab over rituximab prompted investigators to evaluate the therapeutic benefits of combining obinutuzumab with lenalidomide. Results of a single-center phase 2 trial (NCT02871219) showed an overall response rate (ORR) of 98% and an estimated 2-year PFS of 96% at a median follow-up of 22 months in patients with previously untreated stage II to IV FL with a high tumor burden, although Fowler said in the workshop that he would not consider obinutuzumab- lenalidomide to be a standard-of-care option in the frontline setting yet.17
During the workshop, Kieron Dunleavy, MD, of the Microbial Oncology Program at the George Washington Cancer Center, said that he discusses treatment options with his patients for “an hour and a half” and considers age, findings on PET, and toxicity profile when choosing a regimen. For a younger patient, he said that he is more inclined to treat them, even if they are asymptomatic, than adopt a watch-and-wait approach, and he emphasized the importance of further research on the prognostic significance of gene expression profiling and immune response gene signatures in the current era of therapies.
“There were some studies that showed that the immune response signature wasn’t important, but those studies are pretty historic now in terms of the treatments that we use,” Dunleavy said. “Hopefully it can inform us on who we should be treating and who we shouldn’t.”
The experts also discussed the use of maintenance therapy after frontline treatment was also discussed in the workshop. Results of the phase 3 PRIMA trial (NCT00140582) showed that, compared with observation, 2 years of maintenance rituximab significantly improved PFS and increased rates of conversion from partial to complete or unconfirmed complete response (CR) in patients with FL who had achieved a CR or partial response to induction immunochemotherapy; however, OS was not improved.18
Pagel cautioned that careful selection of patients for maintenance therapy is warranted because of the potential for AEs over time. Lori A. Leslie, MD, of the John Theurer Cancer Center/Hackensack Meridian Health in New Jersey, said that she opts for maintenance therapy after rituximab-chemotherapy (for younger patients) or rituximab-bendamustine (for older patients) as long as the patient is responding to treatment. However, Kami Maddocks, MD, of The James Cancer Hospital at The Ohio State University in Columbus, said that she typically reserves maintenance therapy for patients who are not in complete remission after induction therapy because it has minimal benefit and potential risk for infectious toxicity over time in older patients who have achieved a CR.
Nilanjan Ghosh, MD, of the Levine Cancer Institute in Charlotte, North Carolina, said that he considers single-agent rituximab for patients with low-volume disease, and Martin added that single-agent rituximab may help avoid the need for chemotherapy in some patients with low tumor burden. “If it doesn’t get you everything you wished for, you can still move on from there to chemotherapy or R2, which is now approved in that setting, or to clinical trials,” explained Martin.
In the workshop, Pagel noted that the increasingly shorter remission periods following second and higher lines of therapy high- light the unmet need for therapies with durable responses in the R/R setting. Retrospective data from the observational National LymphoCare Study (NCT00097565) showed that the median PFS in patients with FL was 7 years following frontline therapy and decreased incrementally to 1.50, 0.83, 0.69, and 0.68 years following second-, third-, fourth-, and fifth-line therapy, respectively.19
R2 is a common choice for second-line therapy based on results from the phase 3 AUGMENT trial (NCT01938001), which showed a higher IRC-assessed median PFS (39 months vs 14 months), ORR (78% vs 53%), and CR rate (CRR; 34% vs 18%) with R2 than with rituximab plus placebo in patients with R/R FL or marginal zone lymphoma.20 The overall rate of grade 3 or 4 AEs was higher with R2 (69% vs 32%) and was primarily driven by higher rates of grade 3 or 4 neutropenia (50% vs 13%) and leukopenia (7% vs 2%).20
For patients with FL that is R/R to rituximab, obinutuzumab plus bendamustine followed by maintenance obinutuzumab was approved by the FDA in February 201621 based in part on results from the phase 3 GADOLIN trial (NCT01059630). In the trial, bendamustine-obinutuzumab followed by maintenance obinutuzumab in nonprogressing patients led to significantly longer PFS (not reached vs 15 months with bendamustine monotherapy) in patients with rituximab-refractory indolent NHL.22 An updated analysis showed that the obinutuzumab-bendamustine group had a longer median PFS (26 months vs 14 months) and fewer mortalities (52 patients [25.5%] vs 73 patients [35.0%]) after a median follow-up of 32 months.23
During the workshop, the panelists said that their choice of therapy in the R/R setting depends on patient age and comorbidities, treatment that the patient received in the first-line setting, and timing of relapse. Jonah Shulman, MD, of the Icahn School of Medicine at Mount Sinai in New York, New York, said that he often considers a rituximab-chemotherapy approach for patients who relapse within 2 years after single-agent rituximab, although he said that he has been increasingly giving R2 to patients who are older or may not tolerate chemotherapy. Bita Fakhri, MD, MPH, of the University of California, San Francisco Medical Center, added that transformation may be a concern for patients with early relapse and said that she usually gives obinutuzumab or obinutuzumab-CHOP followed by maintenance obinutuzumab (if the transformation occurred while on maintenance rituximab) or R2 (if the patient received bendamustine-rituximab in the first-line setting).
Several of the panelists said that they do not frequently perform autologous stem cell transplants for FL, although Joshua Brody, MD, of the Icahn School of Medicine at Mount Sinai, said that he considers it for younger patients with POD24 who “burn through” multiple regimens quickly and have a good match for stem cell transplant. Similarly, Smith said that he considers transplant only for a young patient who was refractory to or had a short first remission with chemoimmunotherapy.
Therapies targeting PI3K have also been of interest in the R/R setting based on the important role of the p110δ isoform of PI3K in B-cell and T-cell development, B-cell receptor signaling, and the decreased B-cell number and function observed in mouse embryonic knockout of p110δ.24
Idelalisib is a PI3K inhibitor that targets the p110δ isoform with greater selectivity than the α, β, or γ isoforms,25 and a phase 2 trial (NCT02044822) of idelalisib showed an ORR of 57%, median PFS of 11 months, and median OS of 20 months in patients with R/R indolent B-cell NHL who were previously treated with rituximab and an alkylating agent. Diarrhea and/ or colitis was the most common grade 3 or higher AE, occur- ring in 16% of patients, and AEs led to discontinuation of idelalisib in 25 of 125 patients enrolled.26
Results of a phase 2 trial of copanlisib (NCT01660451), which primarily targets p110α and δ isoforms,27 showed that patients with FL who had received prior rituximab and alkylating agents had an ORR and CRR of 59% and 14%, respectively, and a median duration of response of 12 months.28 However, dose delays were required for 74% of patients, and 91% of these were due to AEs such as transient hyperglycemia, transient hypertension, and neutropenia.28 The most common treatment-emergent AEs reported in an updated analysis included transient hyperglycemia (50%), diarrhea (35%), transient hypertension (30%), neutropenia (29%), pyrexia (27%), and fatigue (26%).29
A phase 2 trial of duvelisib (NCT01882803), which targets the p110δ and γ isoforms, showed that the IRC-assessed ORR was 42% (although this included only 1 CR) in patients with FL who were refractory to rituximab and chemotherapy or radioimmunotherapy, and the median PFS was 9.5 months.30 However, several grade 3 or higher AEs were reported, including neutropenia (25%), diarrhea (15%), anemia (15%), and thrombocytopenia (12%).30
Newer-generation PI3K inhibitors are being developed with the goal of reducing toxicities associated with the current agents. Umbralisib, a dual inhibitor of PI3Kδ and CK1ε, is currently be- ing studied in the phase 2b UNITY-NHL trial (NCT02793583) and is awaiting accelerated approval by the FDA. ME-401, an inhibitor of PI3Kδ, received FDA fast track designation and is currently being studied in the phase 2 TIDAL trial (NCT03768505), which is evaluating intermittent dosing to reduce immune-related toxicities associated with continuous daily dosing of PI3K inhibitors.31 Results of a phase 1b study (NCT03985189) showed an ORR of 78% in R/R FL with low (< 10%) rates of grade 3 AEs of special interest related to ME-401 exposure in patients dosed on an intermittent schedule.32
Ghosh said that in his experience, umbralisib seems to be associated with less liver toxicity and colitis than are idelalisib and duvelisib, and an umbralisib regimen has not required treatment interruptions or discontinuations in most patients. “We’ve had some dose reductions, but [it is] overall much better tolerated than what I would have expected with so many other PI3 kinase inhibitors,” said Ghosh.
Similarly, Dunleavy said that the current PI3K inhibitors would not be his first choice for third-line therapy, but he would like to be able to identify patients with tumor biology that is likely to respond to PI3K inhibition. Dunleavy concluded that the high heterogeneity of FL, particularly after multiple lines of therapy, makes selection of therapy difficult in the R/R setting.
EZH2 is an enzymatic subunit that catalyzes methylation of Lys27 of histone H3, which represses genes involved in cell cycle checkpoints and differentiation.33 EZH2 mutations are present in approximately 15% to 20% of FLs, and most of these are point mutations that result in substitution of tyrosine 641 within the histone methyltransferase domain, leading to a gain-of-function effect.33 Expression of the mutant allele synergizes with deregulation of BCL2, which accelerates the development of lymphomas and provides a rationale for EZH2 inhibition in FL.33
Tazemetostat is an EZH2 inhibitor that received accelerated approval from the FDA on June 18, 2020, for patients with R/R FL harboring an EZH2 mutation who have received 2 or more prior systemic therapies or who have no satisfactory alternative treatment options.34 The FDA also approved the cobas EZH2 Mutation Test (Roche Molecular Systems) as a companion diagnostic test for tazemetostat on the same day.34
The approval was based in part on results from a phase 2, open-label, multicenter study (NCT01897571) of tazemetostat in patients with R/R FL after 2 or more standard prior systemic therapies, including at least 1 anti–CD20-based regimen; the results of those harboring an EZH2 mutation and those who were EZH2 wild-type were analyzed separately. Tazemetostat was generally well tolerated, with 7 grade 3 or higher AEs (3 cases of thrombocytopenia, 2 cases of anemia, and 1 case each of fatigue and asthenia) reported among 99 patients. The investigator- and IRC-assessed ORRs in the EZH2-mutated cohort were 78% and 69%, respectively, and only 1 case of IRC-assessed progressive disease was reported. Responses were also observed in the EZH2 wild-type cohort, with investigator- and IRC-assessed ORRs of 33% and 35%, respectively, and IRC-assessed evidence of tumor reduction in 71%. IRC- assessed duration of response was similar between the EZH2 mutant and wild-type cohorts (11 months and 13 months, respectively), and the IRC-assessed median PFS values were 14 and 11 months for the EZH2 mutant and wild-type cohorts, respectively.35
During the workshop, the faculty members said that although their personal experience with tazemetostat was limited, the data appear promising. “I’ve never personally used it, but I think the trial is interesting, especially in patients with an EZH2 mutation, and [it] is going to encourage us to check for the mutation,” said Fakhri.
Smith added that although the efficacy data with tazemetostat do not appear to be better numerically than with PI3K inhibitors in the EZH2 wild-type setting, the low toxicity of tazemetostat may encourage physicians to select it over a PI3K inhibitor. “People are going to be much more comfortable giving this than starting a PI3 kinase [inhibitor] and worrying about monitoring liver function tests and what happens if the patient has diarrhea,” said Smith. “While I’m not convinced [tazemetostat is] better than those efficacy-wise, in terms of toxicity, I think this might replace [PI3K inhibitors] when you get down to that later-line therapy.”
Leslie added that the low toxicity profile of tazemetostat may make it favorable for combination therapy with other agents. One example of a trial evaluating combination therapy is a recently opened phase 1b/3 trial (NCT04224493) whose investigators plan to add either tazemetostat or placebo to R2 treatment in patients with R/R FL, then compare the results.36
Panelists were somewhat divided on whether they would consider tazemetostat in patients with wild-type or unknown EZH2 status. Shulman said that he would consider tazemetostat in the third-line setting for patients with EZH2-mutated FL but would likely use it further down the line in patients with EZH2 wild-type disease. In contrast, Skarbnik said that he may consider tazemetostat over a PI3K inhibitor if the patient prefers a lower toxicity profile, and he pointed out that the dose interruptions often required with PI3K inhibitors may increase risk for progression. Nevertheless, the panelists stated that community oncologists may choose tazemetostat without diagnostic testing because of its low toxicity and ease of use; its role in that setting remains to be determined.
According to Pagel, key advantages of bispecific antibodies, which target 2 or more antigens simultaneously to exert antitumor effects, include their ability to be given off the shelf and their adaptability in dosing. Mosunetuzumab and REGN1979 are bispecific antibodies that target CD3 (on the surface of T cells) and CD20 (on the surface of tumor cells), and they are in early-stage trials for indolent NHL.
The results of an ongoing phase 1/1b study of mosunetuzumab (GO29781; NCT02500407) showed an ORR and a CRR of 64% and 42%, respectively, in efficacy-evaluable patients with indolent NHL.37 Although CRs appeared durable and Pagel noted an exceptional response in a patient who had received 8 prior lines of therapy, he added that longer follow-up is needed.
The results of a phase 1 study of REGN1979 (NCT02651662) in patients with R/R B-cell NHL showed an ORR of 93% and CRR of 71% among the 14 patients with grade 1 to 3a FL.38 Pagel noted that cytokine release syndrome (CRS) may be a concern with REGN1979, with all-grade CRS reported in 55 of 96 patients and grade 3 CRS reported in 7 patients.38 Maddocks and Brody added that although more data are needed on bispecific antibodies, they may be used more widely than chimeric antigen receptor (CAR) T-cell therapy because of their greater tolerability and availability.
CAR T-cell therapy has also shown promising activity in R/R FL. An interim analysis of the phase 2 ZUMA-5 trial (NCT03105336), which is evaluating axicabtagene ciloleucel in patients with advanced-stage indolent NHL, showed an ORR of 95% and CRR of 80% in the subgroup with FL at a median follow-up of 11.5 months. However, 83% of patients experienced grade 3 or higher AEs, including CRS in 11% and neurologic events in 19%.39
At the conclusion of the workshop, the faculty members pointed out the promising trends in shifting toward nonchemotherapy options for FL and said that further research in effective therapeutic combinations with low toxicity will be important moving forward. “It’s exciting to see that there are potentially curative therapies other than [allogeneic] transplantation out there,” said Ghosh.