Safety of bendamustine for the treatment of indolent non-Hodgkin lymphoma: a UK real-world experience

Bendamustine is a bifunctional chemotherapeutic agent with broad clinical activity in the treatment of indolent B-cell non-Hodgkin lymphomas (iNHLs), including follicular lymphoma (FL), lymphoplasmacytic lymphoma (LPL), marginal zone (MZL), and mantle cell lymphomas (MCL).¹ 

Early trials in iNHL demonstrated superior clinical outcomes for bendamustine compared with other chemotherapy, including superior progression-free survival (PFS) for bendamustine and rituximab compared with rituximab plus cyclophosphamide, doxorubicin hydrochloride (hydroxydaunomycin), vincristine sulfate (Oncovin), and prednisone (R-CHOP)/cyclophosphamide, vincristine, and prednisone (CVP),² better tolerance than R-CHOP,³ and higher quality-of-life scores than R-CHOP/R-CVP.⁴ Although highly effective, recent large randomized trials flagged a high rate of infectious complications for patients treated with bendamustine compared with other chemotherapies,⁵ especially during an anti-CD20 antibody maintenance phase that was notably not part of treatment in earlier trials. In the GALLIUM trial comparing any chemotherapy in combination with rituximab vs obinutuzumab followed by antibody maintenance,⁵ a post hoc analysis reported a twofold increase in fatal adverse events (AEs) for patients treated with bendamustine who had not commenced a new anticancer treatment compared with in patients treated with CHOP or CVP (4% vs 2%), with a remarkably higher AE rate (13%) in patients aged ≥70 years. Long-term safety data for this trial reported a fatal AE event rate of ∼6% for bendamustine-treated patients at a median follow-up of 7.9 years.⁶ Another trial comparing obinutuzumab-bendamustine and single-agent bendamustine for rituximab-refractory relapsed iNHL reported high overall rates of grade 3 to 5 AEs (73% and 66% in the combined and monotherapy arms, respectively).⁷ The overall rate of treatment-related deaths was similar (∼2%) in both arms, suggesting that most toxicity was due to bendamustine.⁸ 

The published evidence raised concerns among clinicians about the safety of bendamustine for treating iNHL in routine practice. We performed a retrospective, multicenter, observational study to evaluate this question, and identify potential risk factors for toxicity.

Eligible patients received at least 1 dose of bendamustine with/without rituximab induction with/without rituximab maintenance for untreated or relapsed/refractory iNHL (FL, MCL, LPL, and MZL). Patients treated between 1 January 2013 and 31 December 2016 were identified from 9 National Health Service (NHS) centers in the United Kingdom. Every effort was made to recruit consecutively treated patients to avoid selection bias. Patients with chronic lymphocytic leukemia/small lymphocytic lymphoma or transformed lymphoma, or those enrolled on a clinical trial were excluded. Data for patients treated with obinutuzumab-bendamustine were not collected because this option was not funded by the NHS during the study period. This was a fully anonymized, nonconsent, retrospective research study approved by The Christie NHS Foundation Trust and performed in accordance with the Declaration of Helsinki.

Data collection included patient demographics, European Cooperative Oncology Group performance status (ECOG), histological diagnosis, past medical history, previous lymphoma treatment, current lymphoma induction and maintenance treatment, dose reductions and delays, prophylactic antimicrobial and supportive medication, blood results before induction and maintenance therapy, and worse grade during grade 3 to 5 AEs; data included full blood count, serum biochemistry, immunoglobulins, treatment response data, and grade 3 to 5 AEs. Data were collected from the first dose of bendamustine until the start of the next systemic antilymphoma treatment, death, or date of last follow-up at the time of analysis. Baseline comorbidity was assessed according to the adult comorbidity evaluation 27 (ACE-27) index.⁹ 

Bendamustine was administered according to local institutional standards at a recommended full dose of 90 mg/m² in combination with rituximab (375 mg/m²), or at 120 mg/m² as monotherapy, for a total of 6 to 8 3-to-4-weekly cycles, followed by maintenance rituximab, consolidation treatment, or no further therapy at the discretion of the treating physician. Rituximab maintenance was delivered by IV or subcutaneous injection once every 2 or 3 months for up to 2 years for patients with iNHL and for up to 3 years after transplantation for patients with MCL. Dose reductions and the use of primary granulocyte colony stimulating factor (G-CSF) prophylaxis, antimicrobial prophylaxis, and other supportive medications were discretionary.

The primary end point of the study was the rate of treatment-related grade 3 to 5 serious AEs (SAEs). AE causality was assessed by investigators and graded according to common terminology criteria for AEs version 4.3. SAEs were defined as fatal or life-threatening, causing or prolonging hospital admission, or leading to significant disability. Other outcomes of interest included grade 3 to 5 AE frequency by treatment phase (induction, maintenance, and follow-up), grade 3 to 5 infections, opportunistic infections, second cancers, impact of AEs on dose reductions, delays and treatment discontinuation, and deaths related to bendamustine. The induction period was measured from the start of bendamustine with/without rituximab to 3 months after completion of the final induction cycle; the maintenance phase was measured from the start of the first cycle of rituximab maintenance until 3 months after the last maintenance cycle; and the follow-up period was measured from the end of the induction or maintenance period whichever occurred later, until the date of death, last follow-up, or start of the next antilymphoma treatment. Patients were followed up from the start date of bendamustine to the date of death or last hospital visit.

Potential risk factors were examined for an association with the number (1-6) of treatment-related SAEs and the proportion of patients experiencing ≥1 treatment-related SAE, evaluated based on patient age (≤65 vs >65, ≤70 vs >70, or ≤80 or >80 years), sex, ACE-27 score (0-1 vs 2-3), ECOG performance status score before induction and before maintenance, histology (MCL vs FL vs other iNHL), disease stage, Follicular Lymphoma International Prognostic Index score, simplified Mantle Cell Lymphoma International Prognostic Index score, prior fludarabine treatment, prolonged steroid use (defined as ≥20 mg prednisolone for >2 weeks), antibiotic prophylaxis, G-CSF prophylaxis, line of treatment, starting dose of bendamustine (100% vs 75%-99% vs 50%-74%), lymphocyte count (normal vs abnormal), and total globulin count (normal vs abnormal). Outcomes were compared for patients receiving firstline therapy vs later lines of treatment.

This was a descriptive analysis with no formal power calculations. Frequency tables were provided for categorical demographic variables. Summary statistics together with box plots and histograms were provided for continuous demographic variables. Descriptive analyses were applied to summarize AE data. Fisher exact tests were applied to assess the association between the number of AEs that patients experienced and a series of factors. Proportion tests were applied to assess the difference in proportions of patients with ≥1 AE in subgroups of relevant factors. All presented P values are 2-sided. Statistical analyses were performed with R version 3.5.3 (2019 The R Foundation for Statistical Computing).

The study enrolled 323 patients treated with bendamustine from 9 participating UK centers between 1 January 2013 and 31 December 2016. The median age of patients at iNHL diagnosis was 65 years (range, 20-92 years). FL was the most common histology (54%).

Overall, 150 patients (46%) had no comorbidities (ACE-27 score of 0), and 86 (27%) had moderate to severe comorbidities (ACE-27 score of 2-3). Cardiovascular disease, respiratory disease, and diabetes mellitus were the most common comorbidities reported in 24%, 9%, and 6% of patients, respectively. Baseline characteristics are summarized in Table 1.

The vast majority of participants (96%) received bendamustine in combination with rituximab; the remainder received bendamustine monotherapy. Sixty percent were treated in the firstline setting for iNHL. In patients receiving bendamustine for relapsed/refractory iNHL (40%), the median number of prior treatment lines was 2 (1 to >5) and the most common therapy before bendamustine was R-CHOP or R-CVP (64%).

Most patients (86%) initiated bendamustine at full dose, and 79% of patients completed planned induction treatment, with a median of 6 bendamustine cycles delivered (range, 1-8 cycles; 3 patients received 7 cycles of bendamustine, and 1 received 8 cycles). After induction treatment, 147 patients (46%) commenced maintenance rituximab treatment and 88 (60%) completed planned maintenance. The median number of maintenance cycles delivered was 8 (range, 1-12).

Primary G-CSF prophylaxis was administered to 65 patients (20.1%) during induction; 72% received primary antiviral prophylaxis, 57% against Pneumocystis jirovecii pneumonia (PJP; cotrimoxazole, 55% and nebulized pentamidine, 2%) and 21% against fungal infections. Patients receiving primary G-CSF prophylaxis were significantly more heavily pretreated (mean, 1.806 prior lines of therapy [G-CSF group] vs 1.648 (no G-CSF); P = .0130; 95% confidence interval, −0.4528 to −0.1363) but not significantly older (mean age, 63.0 vs 63.9 years) than those without G-CSF prophylaxis. Very few patients (6%) received high-dose steroids (prednisolone of ≥20 mg within 2 weeks) before starting induction therapy.

The median follow-up was 38.9 months (range, 36.3-40.7 months). Median PFS for patients treated in the firstline or relapsed/refractory settings was 181 or 114 months, respectively. Median overall survival and PFS across all histologies were 153 and 133 months, respectively. Kaplan-Meier curves for overall survival and PFS are shown in the supplemental Appendix.

All 323 patients in the study were included in the safety analysis. In total, 156 patients (48%) experienced 248 grade 3 to 5 AEs of any cause, of which 163 (66%) occurred during induction, 33 (13%) during maintenance, and 52 (21%) during follow-up. Of the AEs occurring during induction and maintenance, 25 of 163 and 19 of 33, respectively, led to treatment discontinuation. There was no difference in the rate of grade 3 to 5 AEs between patients treated in firstline vs subsequent-line settings (45.9% vs 51.9%, respectively; P = .31).

A total of 70 of 323 patients (21.7%) experienced 89 SAEs related to treatment, the primary end point of the study, accounting for the majority (87%) of the 102 reported SAEs. Most events occurred during induction (63%; 56 of 89), followed by maintenance (20%; 18 of 89) and follow-up (17%; 15 of 89). When considered in relation to the number of patients per treatment phase, the highest proportion of SAEs occurred during maintenance (18 events per 33 patients; 54%) followed by induction (56 events per 163 patients; 34%) and follow-up (15 events per 53 patients; 28%). There was no difference in incidence of grade 3 to 5 treatment-related SAEs between patients receiving bendamustine as firstline treatment vs those receiving it as a subsequent line of treatment (20.1% vs 21.7%, respectively; P = .78). AEs and SAEs stratified based on treatment phase are summarized in Table 2, and AEs according to category are shown in Table 3.

In total, 77 patients (24%) experienced 91 grade 3 to 5 infections, of which 44% were treatment-related SAEs, affecting 39 patients (12%). The majority of infections occurred during induction (49% of all infections and 53% of treatment-related SAEs), followed by maintenance (21% of all infections and 30% of treatment-related SAEs) and follow-up (30% of all infections and 18% of treatment-related SAEs). Most infections were nonneutropenic, with neutropenic sepsis accounting for only 4.8% and 9.0% of all grade 3 to 5 AEs and all grade 3 to 5 SAEs, respectively. Fifty patients (15%) experienced at least 1 episode of grade 3 to 5 neutropenia (40 patients; 12%) or febrile neutropenia (12 patients; 4%). Grade 3 to 5 neutropenic episodes mostly occurred during induction (induction: 42 events [83%]; maintenance: 6 events [12%]; and follow-up: 3 events [6%]).

Infections are summarized in Table 4 based on anatomical site and microbiological cause. The most common sites were respiratory (n = 54) and urinary tract (n = 11). The clinical source of infection was not identifiable in 14 cases. There was no significant difference in rates of grade 3 to 5 AE and treatment-related grade 3 to 5 SAE infections between firstline treatment and subsequent-line treatments (21.2% vs 24.8% [P = .50] and 11.3% vs 12.4% [P = .86], respectively).

Ten patients (3%) had grade 3 to 5 opportunistic infections during the study, including 8 nonfatal infections (5 PJP, 3 despite prophylaxis; 2 varicella-zoster virus [VZV] and 1 aspergillosis) and 2 fatal infections (a case of JC virus leading to progressive multifocal leukoencephalopathy during follow-up after firstline bendamustine and maintenance and a case of metapneumovirus infection during secondline bendamustine induction). Median lowest lymphocyte and neutrophil counts at the time of these infections were 0.21 and 1.6, respectively. Three of the opportunistic infections were considered related to bendamustine, and the remainder related to rituximab. The 3 opportunistic infections related to bendamustine were metapneumovirus grade 5 after cycle 2 full-dose bendamustine with rituximab as secondline therapy for FL; VZV grade 3, 19 days after cycle 6 full-dose bendamustine with rituximab as firstline therapy for marginal zone lymphoma; and neutropenic aspergillosis grade 4 in follow-up, 13 months after cycle 6 rituximab with bendamustine as fourthline therapy for FL. Half of the 8 nonfatal opportunistic infections led to treatment discontinuation. There were 3 other grade 3 to 5 nonopportunistic viral infections that did not affect treatment (2 influenza A and 1 rhinovirus infections). There were no cases of cytomegalovirus (CMV) or hepatitis B reactivation.

There were 10 second cancers in the cohort (3% of all patients): a nonfatal case of nonmelanocytic skin cancer, 20 months after treatment initiation, and 5 cases of myelodysplastic syndrome (MDS; 3 fatal) occurring at 15, 16, 44, and 51 months, respectively, after starting treatment; the time to onset was not available for the remaining case. Bendamustine was firstline therapy for 1 of these patients, and thirdline and fourthline for 2 patients each. Of the 5 cases of MDS, 4 had received prior alkylating agents or autologous stem cell transplantation, and there was no risk factor other than age (aged ≥80 years) in the remaining case. A further 4 patients died of second cancers (1 case each of metastatic squamous cell carcinoma, myeloma, transitional cell carcinoma of the kidney, and metastatic non–small cell lung cancer).

At the time of analysis, 91 of 323 patients (28%) had died, most commonly in the context of active lymphoma (n = 54; 59%), including progressive disease (n = 43; 47%) and any treatment-related death (n = 11; 12%). Within these groups, 12 patients (13%) died of infections related to treatment or underlying lymphoma. Seven deaths (8%) occurred due to unrelated causes at the time of progressive disease, and 31 deaths (34%) were unrelated to lymphoma or treatment. Most deaths occurred during follow-up (78%), followed by induction (14%) and maintenance (8%). Patients receiving bendamustine at secondline or later line were more likely to die from any cause than those receiving it as firstline treatment (19.6% vs 41.1%; P < .0001).

Treatment-related events (n = 11) included 6 definite and 3 possible deaths related to bendamustine (2.8% of all patients) and 2 related to rituximab (0.6%). The bendamustine-related deaths were 6 deaths during induction (4 nonneutropenic infections, including an opportunistic metapneumovirus, 1 neutropenic sepsis, and 1 cardiac disease) and 3 deaths from MDS, all during follow-up, considered at least possibly related to bendamustine; however 1 patient also had a prior autologous stem cell transplantation. Two deaths related to rituximab (0.6% of all patients) included 1 case of progressive multifocal encephalopathy due to JC virus during follow-up after firstline bendamustine and maintenance and 1 case of PJP in a patient with neutropenia after 1 dose of maintenance rituximab in the firstline setting. Additionally, 6 patients died of toxicity relating to subsequent lymphoma therapy after bendamustine (excluding maintenance rituximab), including 3 infections complicating allogeneic stem cell transplantation. A full breakdown of deaths is shown in Table 2.

Of the 196 AEs reported during induction and maintenance, 25 (13% of AEs) resulted in a dose reduction, including 17 (9%) leading to dose delay. A further 43 (22% of all AEs) led to permanent discontinuation of treatment in 13% of patients. Infection, neutropenia, and infusion- related reactions were the most common AEs leading to treatment discontinuation (n = 18, n = 8, and n = 5, respectively), followed by gastrointestinal symptoms (n = 3); thrombocytopenia (n = 3); respiratory symptoms (n = 2); and cardiac event, rash, MDS, and hypomagnesemia (n = 1 each). According to investigator assessment, 21 of 43 (49%) events leading to treatment discontinuation were related to bendamustine treatment.

Results of univariate analysis for risk factors associated with the incidence of ≥1 treatment-related SAE are shown in Table 5. The following risk factors were significantly associated with increased risk: mantle cell vs follicular histology (P = .015), MCL vs other nonfollicular iNHL (P = .0036), preinduction ECOG performance status score of 2 vs ECOG performance status score of 0 (P = .0154), premaintenance ECOG performance status score of 1 or 2 vs ECOG performance status score of 0 (P = .0021 and P = .0053, respectively), receipt of primary G-CSF prophylaxis (P = .02847), and abnormal total globulins before induction (P = .0274). The paradoxical association between G-CSF and treatment-related SAEs likely reflects the preferential use of G-CSF in patients at heightened risk of any serious toxicity. Because the rate of neutropenic sepsis was very low, overall SAE rates were not affected by G-CSF use, and it was not possible to specifically examine the association between G-CSF use and neutropenic sepsis.

None of the other analyzed risk factors, including age, sex, ACE-27 score, disease stage, FL International Prognostic Index/simplified Mantle Cell Lymphoma International Prognostic Index score, prior fludarabine treatment, prolonged steroid use, antibiotic prophylaxis, line of treatment, starting dose of bendamustine, lymphocyte count, and total globulin count, were associated with adverse safety outcomes.

This multicenter, retrospective, observational study is 1 of the largest studies, to our knowledge, to date, evaluating the safety and toxicity profile of bendamustine for patients with iNHL treated outside of a clinical trial setting. We evaluated 323 patients, previously treated or untreated, receiving bendamustine-rituximab (96%) or bendamustine monotherapy in routine practice for common iNHL lymphomas: FL (54%), LPL (17%), MCL (10%), and MZL (10%). Patients were followed up during induction, maintenance, and after treatment for a median of 34 months.

Grade 3 to 5 SAEs related to bendamustine were reported in 21.7% of participants, with ∼50% due to infections, most commonly of respiratory and urinary tract origin. Thirteen percent of patients stopped treatment because of bendamustine-related toxicity, most commonly infection, and 2.8% of patients died of causes deemed by investigators to be related to bendamustine. The most common causes of bendamustine-related deaths were nonneutropenic infections during induction and MDS during follow-up. Neutropenic sepsis events were rare.

AEs occurred during all treatment phases; however the relative risk of experiencing a treatment-related grade 3 to 5 SAE was highest during the maintenance phase, with events reported in 54% of patients compared with in 34% of patients during induction and 28% of patients during follow-up before initiating next antilymphoma treatment.

A high rate of infections during maintenance and follow-up (21% and 30% of all grade 3-5 infections, respectively; and 23% of all deaths) is 1 of the most important observations of this study, mirroring data from the GALLIUM trial of previously untreated FL, in which a higher frequency of grade 3 to 5 infections in bendamustine-treated patients was shown to be driven by events during the maintenance phase.⁵ The added risk of maintenance is further inferred by findings from a meta-analysis of 9 randomized, controlled trials in which no association between bendamustine and higher rates of any grade infection was reported, presumably because the study included very few maintenance-treated patients.¹⁰ 

In this study we did not observe a significant effect of increasing age on the risk of infectious complications or death; however, our analysis was limited by a small number of patients in a heterogenous population. Very large population-based studies, including 2 Surveillance, Epidemiology, and End Results database studies, have shown a clear association with increasing age. The first study, involving 9395 patients aged ≥65 years receiving chemotherapy for iNHL (FL, MZL, or LPL) from 2006 to 2013, showed higher rates of bacterial pneumonia, other unspecified bacterial infections, viral infections, and opportunistic infections, reaching statistical significance for CMV, VZV, and histoplasmosis. The hazard ratio for PJP infection was 3.26, but this was not statistically significant.¹¹ The second study reported outcomes for 1791 patients with previously untreated iNHL aged >65 years receiving bendamustine.¹² Compared with R-CHOP/R-CVP, patients treated with bendamustine had significantly higher rates of hospitalization, infection, and pneumonia, extending into the second year of follow-up.¹³ Persisting risk fits with results of this study in which almost one-third of all grade 3 to 5 infections occurred during follow-up, supported by studies showing delayed T-cell reconstitution for up to 25 months after bendamustine^14,15 and sustained reductions in CD3⁺CD4⁺ cells.⁵ Lymphopenia is commonly seen during treatment with rituximab and CD4⁺ lymphopenia has been reported to continue throughout maintenance rituximab, with gradual recovery after stopping.¹⁶ Thus, the widely reported and prolonged risk of infection is most likely related to the T-cell–depleting effect of bendamustine. Most patients in this study became lymphopenic during induction treatment; however, CD4/CD8 levels were not routinely or consistently recorded in our cohort, making it impossible to draw correlations in this study.

Interestingly, although neutropenia was relatively common, the overall rate of febrile neutropenia for bendamustine-treated patients was low in this study (4.8% and 7.8% of all grade 3-5 AEs and SAEs, respectively) and most infections related to bendamustine, including fatal events, were nonneutropenic in nature. This is consistent with other studies showing lower rates of neutropenia and febrile neutropenia for bendamustine than for CHOP.^2,3,5 

The rate of opportunistic infection of any grade was 3.1% in this study. A Medicines and Healthcare products Regulatory Agency Drug Safety Update published in 2017 recommended prophylaxis and monitoring for opportunistic infections and hepatitis B reactivation in bendamustine-treated patients based on observations of higher rates of PJP, VZV, CMV, and hepatitis B reactivation.¹⁷ This report was published after the recruitment period for this study, which may explain why almost half of the patients in our study did not receive primary prophylaxis against PJP and viral infections. It is also pertinent to note that 3 of 5 patients in this study developed PJP infections despite prophylaxis, underscoring the importance of clinical suspicion and surveillance for opportunistic infections even when prophylaxis is given. Instances of PJP can lead to treatment discontinuation, as observed in our study, or death, as reported in the GADOLIN study, in which 1% of patients treated with bendamustine monotherapy died of treatment-related PJP.⁷ 

In terms of noninfectious AEs such as gastrointestinal and dermatological toxicities and infusion reactions, the rates recorded in our study broadly reflect the literature. Second cancers, including MDS,¹⁸ have been reported but registry study data suggest that patients treated with bendamustine are no more likely to develop a second cancer than those treated with R-CHOP or R-CVP (rates, 4% vs 6%; P = .6).¹² The rate of MDS in this study (1.5%) is lower than that reported by Martin et al¹⁸ (4%), but that follow-up time was shorter.

Mantle cell histology, abnormal total globulins, and poor performance status were associated with an increased risk of treatment-related grade 3 to 5 AEs in this study. A larger sample size might have resulted in statistically significant associations for other patient- and treatment-related factors reported in the literature.

The findings of this study are limited by the retrospective nature of data collection; thus, AEs might have been underreported, and there might have been investigator bias in assigning causality. In addition, the study population did not include patients treated with bendamustine combined with obinutuzumab because this was not funded during the study period. Although obinutuzumab and rituximab have similar mechanisms of action and toxicity profiles, safety data from this study cannot be extrapolated to patients receiving obinutuzumab and bendamustine.

This multicenter, retrospective, observational study of bendamustine treatment for iNHL in routine practice demonstrates rates of bendamustine-related treatment discontinuation, dose delays and reductions, hematological toxicity, and grade 5 events that are comparable with trial population outcomes despite including previously treated and untreated patients as well as patients who were older, frailer, and with more comorbidities. Notably, the rate of fatal AEs related to bendamustine (2.8%) is similar to the rituximab-chemotherapy arm of the GALLIUM trial (3.4%), which included bendamustine, CVP, and CHOP in an exclusive firstline FL population.

This study highlights important safety considerations when administering bendamustine, including vigilant monitoring and long-term surveillance for infection especially in patients receiving maintenance rituximab, and for opportunistic infection despite the use of antimicrobial prophylaxis. Patients with MCL, poor baseline performance status, and weakened immunity (evidenced by low total globulins), as demonstrated in this study, and older age, as demonstrated elsewhere, are at heightened risk of treatment-related AEs. These patients should, therefore, be considered for treatment modifications and increased supportive care, in line with published practice guidelines.^13,19,20 

Contribution: R.S., R.B., A.A., and K.M.L. prepared the manuscript; R.S., A.A., X.W., and K.M.L. analyzed the data; A.A. coordinated study data collection; K.M.L. conceived and designed the project; and all other authors submitted data and contributed to manuscript preparation.

Conflict-of-interest disclosure: N.M.-C. received honoraria from Janssen; received travel support and honoraria from and served on the advisory board of AbbVie; received travel support and honoraria from AstraZeneca; and served on the advisory board of and received honoraria from Takeda. C.P.F. consults and performs educational activities for AbbVie, AstraZeneca, Atara Biotherapeutics, Celgene/Bristol Myers Squibb, Genmab, Gilead/Kite, Incyte, Janssen, Lilly, MorphoSys, Ono, Roche, and Takeda, and received research funding from BeiGene. A.D. serves on the advisory board of and received research funding, honoraria, and travel fees for scientific conferences from Celgene and Roche; serves on the advisory board of and received research support and honoraria from Kite; serves on the advisory board of and received research support from Karyopharm Therapeutics; received research support and honoraria from Acerta Pharma/AstraZeneca and ADC Therapeutics; serves on the advisory board of Incyte; received research support from Merck Sharp & Dohme (MSD); and serves on the advisory board of AbbVie. T.A.E. reports education and advisory board honoraria, and scientific conference travel support from Roche; reports honoraria and research and scientific conference travel support from Gilead; reports education and advisory board honoraria from Kite; reports honoraria from Janssen; reports honoraria and scientific conference travel support from AbbVie; received honoraria, research funding, and scientific conference travel support from AstraZeneca; received advisory board honoraria from, and served on the steering committee of, Loxo Oncology; received advisory board honoraria and research funding from BeiGene; and received advisory board honoraria from Incyte and Secura Bio. W.O. reports speaker honoraria from Roche, Takeda, Pfizer, Kite Gilead, AstraZeneca, Novartis, Kyowa Kirin, Incyte, and Janssen; serving on advisory boards for Roche, Servier, Takeda, MSD, Kite Gilead, Novartis, Beigine, Autolus, Kyowa Kirin, Incyte, Sobi, Janssen, and Synoes; and support for medical education from Novartis, Takeda, and Roche. W.T. has received honoraria, speaker's fees, and travel support to international conferences from Roche, Takeda, Incyte, and Gilead. The remaining authors declare no competing financial interests.

Correspondence: Kim M. Linton, Division of Cancer Sciences, School of Medical Sciences, The University of Manchester, 555 Wilmslow Rd, Manchester M20 4GJ, United Kingdom; email: kim.linton@manchester.ac.uk.