Evolving therapeutic landscape in follicular lymphoma: a look at emerging and investigational therapies

Idelalisib, a PI3K δ selective inhibitor, was the first approved PI3K inhibitor (PI3Ki) for relapsed FL based on a phase II study (n = 125) of heavily pretreated patients (median prior lines of therapy = 4) with indolent NHL (58% of which were FL), showing a median PFS of 11 months [36]. Potentially life-threatening autoimmune grade 3–4 AEs including colitis, hepatitis, and pneumonitis can occur at any time during treatment [37] and must be immediately recognized as the patients may require a short course of systemic steroids after exclusion of infectious causes [38]. In addition, opportunistic infections including pneumocystis jiroveci pneumonia (PJP) and cytomegalovirus (CMV) reactivation can occur with regular monitoring for CMV indicated while on therapy. Copanlisib is a PI3Kα/δ inhibitor which shown a median PFS of 12.5 months (n = 142 pts, 73% with FL) in the phase II CHRONOS-1 study leading to its approval [39, 40]. Its overall safety profile is more favorable than idelalisib, although hyperglycemia and hypertension were seen at a higher rate with copanlisib which can be explained in part, at least for the hyperglycemia, by greater PI3Kα inhibition.

Duvelisib is a PI3K δ/γ inhibitor currently approved for FL treatment based on the DYNAMO trial which showed a PFS of 9.5 months (n = 129 pts, 64% with FL), with diarrhea (48.8%), nausea (29.5%), and neutropenia (28.7%) being the most common side effects [41]. It is important to note that a greater proportion of patients included in the study were refractory to prior chemo-immunotherapy (77%).

Umbralisib is a PI3Kδ selective inhibitor that also has activity against casein kinase (CK)1-epsilon proteins. The CK1 proteins are a family of proteins that are involved in various processes including DNA repair, mitotic checkpoint signaling, and the immune response [42]. In particular, CK1ε plays an important role in the translation of lymphoma oncogenes as well as the Wnt-β catenin pathway. Of note, in a mouse model of CLL, treatment with umbralisib led to less intestinal and liver inflammation compared with idelalisib and duvelisib and correlated with a higher number of peripheral Treg cells in umbralisib mice, suggesting there may be a protective effect of CK1ε inhibition from the autoimmune toxicities induced by PI3K inhibition [43]. In a study of rel/ref indolent lymphomas including FL, single-agent umbralisib (n = 117) showed an ORR and CR rate of 45.3% and a 5.1%, respectively with a median PFS of 10.6 months[44] leading to its accelerated approval for rel/ref FL after at least three prior lines of therapy in February of 2021.

Given the ongoing serious risk of autoimmune toxicity and infectious complications with long-term PI3K inhibition, particularly with strong PI3Kδ blockade, there is significant interest in balancing the anti-tumor effects with these toxicities. A possible way of mitigating these risks is with intermittent or induction/maintenance dosing schemes instead of continuous dosing. However, there are no ongoing trials with approved PI3K inhibitors comparing alternate dosing schemes and current approaches have rather focused on developing new PI3K inhibitors with better efficacy and/or safety profiles.

Parsaclisib is a structurally distinct PI3Kδ selective inhibitor with an overall greater potent inhibition of PI3Kδ than the other currently approved PI3K inhibitors. Early results from a phase1/2 of parsaclisib with and without itacitinib, a selective JAK1 inhibitor, showed an ORR and CR rate of 63% and 13% in the monotherapy arm in FL patients (n = 5) with results awaited from an ongoing phase 2 study in rel/ref FL (NCT03126019) [45]. Other PI3Kδ inhibitors currently in phase 2 clinical trials include zandelisib (ME-401, a selective PI3Kδ inhibitor, NCT03768505) and YY-20394 (a selective PI3Kδ inhibitor, NCT04370405) and tenalisib (PI3K δ/γ inhibitor, NCT03711578). Many drugs with dual PI3K/mammalian target of rapamycin (mTOR) inhibition have also been developed but their utility in providing improved efficacy at the expense of toxicity given their safety profiles in clinical trials is unclear [46].

Combinatorial approaches with approved PI3K inhibitors are currently ongoing and have been nicely detailed in a recent review [47]. In the cooperative group trial, when idelalisib was combined with R2, there was significant hepatic toxicity with 2 of 7 enrolled patients passing away from toxicity, one from hepatic failure and another from complications related to colitis leading to premature closure of the study [48]. Given the overall better safety profile of copanlisib, duvelisib and umbralisib compared to idelalisib, these PI3Kis may be better suited to be incorporated into combination treatment strategies either with chemotherapy or immunomodulatory agents. The preliminary results of a phase III study incorporating copanlisib into chemoimmunotherapy with either R-CHOP or BR in the relapsed setting showed an ORR and CR rate of 90% and 50%, respectively for BR + copanlisib (n = 10) and 100% and 30%, respectively for R-CHOP + copanlisib [49]. Four patients within the BR + copanlisib group and five patients within the R-CHOP + copanlisib group required discontinuation of treatment due to AEs. Pneumonitis did occur in one patient but no AEs related to colitis or hepatic failure were reported. Likewise, duvelisib was also studied in combination with BR in patients with relapsed chronic lymphocytic leukemia (CLL) or NHL (n = 29 including both indolent and aggressive lymphomas, 15 of which had FL) in phase I study with a dose-expansion phase [50]. Thirty-seven percent of patients experienced an SAE with 19.6% of patients determined to be related to duvelisib, including cases of CMV colitis, CMV esophagitis, inflammatory colitis, acute lung injury, and generalized rash, with no cases of hepatitis. The limited numbers of FL patients in this study preclude efficacy analysis, but given the serious toxicities when combining duvelisib with bendamustine likely related to the significant lymphosuppression, the role of combination duvelisib + bendamustine versus reserving duvelisib monotherapy for a later line is unclear. In the FRESCO trial (NCT02605694) duvelisib was combined with either rituximab or R-CHOP for patients with progression within the first 24 months after initial therapy but was subsequently withdrawn by the sponsor.

Combination studies with umbralisib are currently under investigation. The combination of obinutuzumab and umbralisib is one of the three arms of the ongoing SWOG1608 phase II trial (see Approach to the POD24 patient). An initial phase I study combining umbralisib with the novel anti-CD20 antibody ublituximab (U2) in rel/ref B-NHL and CLL found this combination to be safe with no new safety signals compared to the umbralisib alone with an ORR and CR rate of 46% and 17%, respectively[51]. As discussed earlier, given the overall reduced potential for autoimmune toxicity of umbralisib compared to the other PI3K inhibitors, combination therapies of umbralisib with lenalidomide/ anti-CD20 combinations may be safer with strong anti-FL activity. A clinical trial combining U2 along with lenalidomide (NCT04635683) is currently ongoing.

Another potential area of interest is combining PI3Ki with Bruton’s tyrosine kinase (BTK) inhibitors given the synergy and potentially non-overlapping toxicity. Acalabrutinib was studied in combination with PI3Kδ inhibitor and was shown to be safe and tolerable in early phase clinical trials in rel/ref B cell malignancies [52]. A phase Ib/II study is underway looking at the combination of acalabrutinib and duvelisib in rel/ref indolent NHLs (NCT04836832).

EZH2 is histone lysine methyltransferase gene, which as a part of the polycomb repressor complex 2 (PRC2), catalyzes the addition of methyl groups to K27 of H3. This leads to transcriptional repression of target genes. EZH2 is important in the formation of GC in mice and it has been shown that mutations in EZH2 in addition to overexpression of BCL2 lead to B cell lymphomas [57]. EZH2 may also have immune modulatory effects wherein it suppresses the immune effector trafficking by repressing Th1 type cytokines [58]. EZH2 is mutated in ~ 20% of FL with the vast majority of mutations resulting in the substitution of tyrosine 641 leading to gain of function of methyltransferase activity. This distinguishes EZH2 mutations from most other epigenetic enzyme mutations which are loss of function mutations and thus provide a more readily available therapeutic target. Tazemetostat is a first-in-class oral inhibitor of EZH2 that has shown an overall favorable safety profile in the first in human phase I study of patients with rel/ref NHL [59]. The phase 2 study of tazemetostat in relapsed FL (who received 2 or more prior therapies, n = 99) demonstrated an ORR of 69% and 35% in EZH2 mutant (n = 45) and wild type (n = 54) patients, respectively, with a median PFS of 13.8 months and 11.1 months [60]. Grade 3 treatment-related AEs were infrequent and included thrombocytopenia (3%), neutropenia (3%), and anemia (2%). Tazemetostat is an attractive therapeutic option, especially in elderly patients or patients with co-morbidities that may preclude other therapies such as PI3Ki or chemotherapy given the clinical activity and favorable safety profile.

Valemetostat is a dual EZH1/2 inhibitor that has shown activity in both B and T cell NHLs in the Japanese population with an ORR of 53% with a particularly high response rate in T cell NHLs (ORR = 80%) [61]. The study is currently enrolling patients with T cell NHLs within the dose-expansion phase. Whether dual inhibition of EZH1/2 offers any benefit compared to selective EZH2 inhibition in FL is an open question that warrants further investigation.

Vorinostat is an oral inhibitor of class I and 2 HDACs and is currently approved for the treatment of T cell NHLs. As discussed previously, given the high rate of acetyltransferase loss of function mutations in FL, another potential therapeutic approach to counteract the loss of histone acetylation marks in these FLs would be to target the histone deacetylases to try to restore epigenetic homeostasis in these tumors. In a phase II study of relapsed indolent lymphomas and mantle cell lymphoma (MCL) (n = 56, 39 with FL), the ORR was 49% in the subset of patients with FL with a median PFS of 20 months. The toxicity was limited to cytopenias which were easily managed [62]. Mutation analysis revealed a mutation rate of 67.7% and 21.4% in CREBBP and EP300, respectively, with no correlation between the presence of a mutation and clinical response to vorinostat, although the limited number of patients precludes the determination of these mutations as predictive markers of response. Preclinical work has shown that inhibition of HDAC6 can lead to upregulation of CD20 with enhanced efficacy with anti-CD20 antibodies [63]. In a phase II study of rituximab in combination with vorinostat in newly diagnosed and rel/ref indolent B-NHLs (22 of 28 were FL), the ORR and median PFS were 41% and 18.8 months, respectively in the previously treated patients [64]. Patients who achieved a CR (n = 10) were allowed to come off treatment after 2 additional cycles of therapy with the option of retreatment upon relapse. Six remained in CR after a median follow-up of 27 months and of the 4 patients who underwent retreatment, 2 achieved a second CR, suggesting that time-limited therapy may be possible with this combination. It is unclear if vorinostat could resensitize rituximab refractory patients given the lack of information on rituximab refractoriness in the study. Studies of combination chemoimmunotherapy with vorinostat have largely been limited to diffuse large B cell lymphoma (DLBCL) with a few FL patients enrolled in these studies. The pan-HDAC inhibitor panobinostat that is currently approved for relapsed multiple myeloma has been studied in B-NHLs in early phase clinical trials in combination with mTOR inhibitor everolimus which inhibits cell proliferation by inhibiting protein translation [65]. Thrombocytopenia was the primary dose-limiting toxicity (DLT) with a median PFS of 4 months in FL patients. Studies on newer HDAC inhibitors, including abexinostat, mocetinostat, belinostat, entinostat, quisinostat, and chidamide have largely focused on T cell NHLs and aggressive lymphomas, except for mocetinostat which showed modest single-agent activity in FL. In a phase II trial of rel/ref DLBCL and FL, the ORR was 11.5% in the FL cohort (n = 31) [66].

The DNMTs have an opposing role to chromatin acetyltransferases by the methylation of DNA leading to transcriptional repression by maintaining or inducing a closed chromatin state thus precluding the binding of transcription factors. The DNMT inhibitors, 5-azacytidine, and decitabine have been established as effective therapeutics in myelodysplastic syndrome (MDS) and Acute Myeloid Leukemia (AML). In a small phase I study of low dose decitabine in patients with rel/ref CLL and DLBCL (n = 20), no responses were seen with 8 patients having a stable disease with correlative studies showing no differences in DNA methylation [67]. In another preliminary analysis of a study incorporating 5-azacytidine with R-CHOP in rel/ref lymphomas (n = 10, 3 with FL), one CR and one PR awere seen in the FL patients [68]. Given that only small DNMT inhibitor trials with limited numbers of FL patients have been reported, the potential of DNMT inhibitors in FL is currently unclear.

The PRMTs catalyze the mono or dimethylation of histones, with PRMT5 being overexpressed in NHLs [69]. In addition, PRMT5, like EZH2, is required for GC formation and lymphoma survival through its interaction with BCL6, and thus, like EZH2, may also be important in the maintenance of epigenetic dysregulation in FL [70]. PRMT5 inhibitors have shown significant activity in pre-clinical models of Epstein Barr virus (EBV +) lymphomas, DLBCL, and MCL [71, 72]. Clinical trials of PRMT5 inhibitors including GSK3326595 (NCT02783300) and PF-06939999 (NCT03854227) are currently in the early stages of development. PRMT5 inhibition holds promise as another epigenetic therapy for FL.

The BET proteins are a group of “histone readers” that act downstream of histone acetylation by binding to histone acetyl marks and recruiting transcription factors to DNA. Birabresib was the first BET to be evaluated in a clinical trial, but with no responses seen in DLBCL patients (n = 2) in the first-in-human study [73]. In another preliminary analysis of a dose escalation phase I study of the BET inhibitor CPI-0610 in B-NHL (n = 44, 8 with FL), one PR was seen in FL patients [74]. Thus far, response rates have been very modest and the high rate of histone acetyltransferase mutations in FL likely may result in resistance to this class of inhibitors. Restoration of acetyl marks by other epigenetic therapies may be required for the improvement of responses to these inhibitors.

At present, the therapeutic potential of many epigenetic therapies may likely lie in novel combinations of already approved agents or rational combinations defined by synergy in the pre-clinical studies. A phase II trial of tazemetostat with rituximab (NCT04762160) and a phase I trial incorporating R2 with tazemetostat (NCT04224493) both in rel/ref FL after one prior line of therapy are currently underway. The interaction of other epigenetic modifiers with EZH2 is an exciting avenue of research. Preclinical work using a combination of acetyltransferase and methyltransferase inhibitors with EZH2 inhibition and profiling of epigenetic and downstream transcriptional changes will likely uncover interesting and novel combinations of epigenetic therapies with more potent clinical activity and will hopefully lead the way to rational clinical studies of epigenetic therapies in FL.

CD19 CART therapy has revolutionized the management of relapsed lymphoid malignancies offering potentially curative therapy to patients who were previously deemed incurable by contemporary therapies. Axicabtagene ciloleucel (axi-cel, Yescarta) received FDA approval in 2017 for rel/ref DLBCL after two prior lines of treatment based on the ZUMA-1 trial that showed an ORR and CR rate of 83% and 58% with a median PFS of 5.9 months [92]. Similarly, tisagenlecleucel (tisa-cel, Kymriah) received approval in 2018 based on results of the JULIET trial showing an ORR and CR rate of 52% and 40% with a relapse-free survival of 65% at 12 months. Both therapies have since been investigated in FL with impressive results.

BiTes are antibodies composed of two distinct antibody chains, one which recognizes an epitope present on T cells and the other of which is directed to an epitope on the target cell of interest resulting in direct cell-mediated toxicity by the cross-linked T cell to the tumor cell [95]. Although BiTes are not currently approved for FL at that time of this writing, the preliminary results with these agents in highly pre-treated FL appear promising in early phase clinical trials (Table 3). Larger efficacy studies are currently ongoing.