Ibrutinib as first-line therapy for mantle cell lymphoma: a multicenter, real-world UK study

Advanced stage mantle cell lymphoma (MCL) is an incurable B-cell malignancy with a wide spectrum of clinical presentations and management approaches spanning active surveillance for asymptomatic and clinically indolent disease to urgent immunochemotherapy for aggressive variants.¹ Selected fitter patients can achieve long remissions after intensive cytarabine- and rituximab-containing immunochemotherapy followed by autologous stem cell transplant (SCT) and maintenance rituximab, but most newly diagnosed patients are older and not suitable for intensive therapy. Nonintensive immunochemotherapy is the mainstay of treatment for these patients but associated with inferior outcomes and significant toxicity.

Ibrutinib is an oral covalent Bruton tyrosine kinase inhibitor (BTKi) widely used for treatment of relapsed and refractory MCL (R/R MCL) after a phase 2 trial leading to the US Food and Drug Administration approval in 2013. In this pivotal study, single-agent ibrutinib demonstrated high efficacy and favorable safety in heavily pretreated patients.^2,3 A pooled analysis of several clinical trials showed that ibrutinib was more effective when delivered earlier in the treatment course,⁴ raising the question of whether first-line application could offer safer and potentially more effective treatment than standard immunochemotherapy, especially for older, transplant-ineligible patients.

Several randomized phase 3 trials are investigating this question, including the ongoing UK NCRI ENRICH trial (rituximab-ibrutinib vs rituximab-chemotherapy in patients aged >60 years)⁵ and the SHINE trial (bendamustine-rituximab plus ibrutinib/placebo in patients aged ≥65 years), which reported a longer progression-free survival (PFS) with the addition of ibrutinib.⁶ Two small single-arm phase 2 trials showed high activity for ibrutinib-rituximab without chemotherapy as first-line treatment of MCL, but both enrolled predominantly patients at low risk^7,8 and noted excess toxicity in older patients. To our knowledge, no studies have yet reported outcomes of first-line chemotherapy-free ibrutinib-based therapy in high-risk MCL.

Ibrutinib (with or without rituximab) was approved in England under an NHS (National Health Service) COVID-19 interim agreement as a potentially safer oral alternative to IV chemotherapy for the first-line treatment of MCL. We conducted a multicenter observational cohort study to evaluate clinical outcomes and safety of patients treated under this scheme. Outcomes in high-risk MCL were of special interest.

The study was conducted as an approved NHS England evaluation and qualified for consent exemption per NHS Health Research Authority guidance. The NHS England IBR5CV⁹ scheme, operational from April 2020 to September 2022, was an interim COVID-19 approval for ibrutinib with or without rituximab for adult patients aged ≥ 18 years with histopathologically confirmed MCL, who had received no prior systemic therapy for MCL (excluding prior steroids and local radiotherapy), and Eastern Cooperative Oncology Group (ECOG) performance status 0 to 2. Enrollment in the scheme vs other treatment approaches including active surveillance and immunochemotherapy was at clinician’s discretion. NHS centers with ≥1 registered patients were flagged to the study team, who then contacted centers to identify potential participants. Eligible patients were treated with at least 1 dose of ibrutinib with or without rituximab. Ibrutinib target dose was 560 mg once daily; treatment was continued until disease progression, treatment toxicity, death, or withdrawal for other reasons. IV or subcutaneous rituximab was optionally delivered at clinician’s discretion at 375 mg/m² or 1400 mg for up to six 28-day cycles. Maintenance rituximab treatment was permitted with alternative ibrutinib cycles for up to 2 years. All patients were managed in accordance with institutional standards of care. Anonymized data linked to a unique study identifier were collected from hospital records and entered into a web-based RedCap database curated for this study. Patients could be registered to the database either prospectively or retrospectively. When available, anonymized histopathology reports were reviewed centrally by an experienced lymphoma pathologist. Extranodal spread included confirmed bone marrow involvement, but bone marrow aspirates were not routinely performed. Radiological response and survival data were collected along with all grade ≥3 toxicities; any grade toxicity leading to dose delay, reduction, or permanent discontinuation of ibrutinib; new onset atrial fibrillation (AF) of any grade; and cause of death. Post-ibrutinib treatment details were collected when applicable. Treatment response and disease progression was investigator-assessed by CT and/or positron emission tomography CT according to the Lugano classification, with bone marrow assessment at the discretion of the treating clinician.¹⁰ Toxicity was graded by local investigators using CTCAE version 5 criteria. High-risk MCL was defined as the presence of at least 1 high-risk feature: TP53-positive status, blastoid/pleomorphic variant, or Ki67%/MiB-1 ≥30%¹¹; because TP53 screening was not routine practice at all centers at the time of this study, the low-risk category included patients for whom the aforementioned risk factors were either absent or unknown. TP53-positive status was defined as at least one of: p53 overexpression by immunohistochemistry, TP53 mutation by next-generation sequencing, or TP53 deletion by fluoresence in-situ hybridization. The Simplified MCL International Prognostic Index (sMIPI) score was calculated.¹² The study was conducted according to the principles of the Declaration of Helsinki and the UK Data Protection Act (2018).

Primary objectives were overall response rate (ORR) and overall survival (OS). Key secondary objectives were PFS, incidence of ibrutinib discontinuation due to toxicity, incidence of ibrutinib dose reduction due to toxicity, and OS after ibrutinib discontinuation for patients with progressive disease. Response rates and factors influencing rituximab use were analyzed using χ² test and Fisher exact test. Kaplan-Meier survival analyses were used for time- to- event outcomes. Univariable and multivariable analysis was performed using the Cox proportional hazards model. Statistical analyses were performed using SPSS V28.0.1.1/Rstudio V2022.12.0+353. The database was locked on 31 December 2022 for analysis.

PFS was defined as the time from day 1 of ibrutinib therapy until investigator assessed progression or death from any cause (event) or censored at the last date of clinical review with no evidence of progression (no event). OS was defined as the time from day 1 of ibrutinib therapy until death from any cause (event) or censored at the date of last clinical review (no event). Survival after ibrutinib was defined as the date of cessation of ibrutinib until death from any cause.

Data were collected for 149 patients from 43 NHS centers in England. Baseline demographic and clinical features are summarized in Table 1. The majority (74.1%) were male, and the median age was 75 years (range, 41-94 years). Most patients (92.8%) had stage III to IV disease, with approximately half presenting with extranodal involvement and/or B symptoms (55.7% and 42.0%, respectively). The majority (75.2%) had ECOG performance status 0 to 1. At baseline, 7.8% were considered suitable candidates for high-dose chemotherapy and autologous transplant in a hypothetical prepandemic setting. TP53 mutations and deletions were reported in 6.7% and 2.7%, respectively. At least 1 high-risk disease feature (TP53-positive status, blastoid/pleomorphic variant, and/or Ki67/MiB1 ≥ 30%) was present for 54 patients (36.2%). Breakdown of overlapping high-risk features is described in Table 1. The sMIPI risk category was low risk (0-3) for 13.3%, intermediate risk (4-5) for 36.3%, and high risk (≥6) for 50.4%. Details of available/missing data for each domain are presented in supplemental Table 2.

At a median follow-up of 15.6 months (range, 0-31.0), participants received a median of 8 cycles of ibrutinib (range, 1-33). Ninety-two percent started at full dose ibrutinib. Forty-six patients (33.8%) required a dose reduction or dose delay of ibrutinib, and 62 of 149 patients (41.6%) discontinued ibrutinib. This was because of progressive disease (n = 33 [22.1%]), toxicity (n = 12 [8.1%]), death of any cause (n = 8 [5.4%]), patient choice (n = 1 [0.7%]), and other/unknown reasons (n = 8 [5.4%]; Figure 1). For those who discontinued treatment, median time to discontinuation was 136 days (range, 5-918). Discontinuation due to toxicity occurred at a median of 67 days (range, 5-437).

Rituximab was coadministered with ibrutinib in 55 patients (39.0%), with a median of 6 (range, 1-17) cycles delivered. Twenty-three patients (42.6%) had received >6 cycles of rituximab at the time of analysis (data missing for 1 patient). Disease features significantly associated with rituximab use (supplemental Table 1) were ECOG performance statuses 0 to 1 vs 2 to 4 (OR, 2.493; 95% confidence interval [CI], 1.158-5.444; P = .0260). There was a trend toward an association between the presence of bulk disease and rituximab use (odds ratio [OR], 2.127; 95% CI, 1.015-4.589; P = .055) and absence of high-risk features (OR, 2.014; 95% CI, 0.9738-4.299; P = .0739).

Grade ≥3 all-cause toxicity was reported in 27 of 133 patients (20.3%), summarized in Table 2. Grade 3 to 4 bleeding was reported in 4.0%, comprising 2 cases of intracranial hemorrhage, 2 cases of gastrointestinal bleeding, and 2 cases of epistaxis (1 of which was associated with grade 4 thrombocytopenia). Six patients (4.0%) experienced grade 3-4 myelosuppression (neutropenia, n = 2; thrombocytopenia n = 2; anemia, n = 1; and not stated, n = 1). Grade 3 to 5 nonneutropenic infection occurred in 11 patients (7.4%). Two patients had cerebrovascular accidents (stroke), including 1 grade 5 ischemic stroke occurring 1 week after starting ibrutinib, not associated with AF. New onset AF was reported in 9 of 137 patients (6.6%), of whom 6 had a history of hypertension or coronary artery stenosis. One case of new AF was grade 3 to 4, and 8 were grade 1 to 2. Patients receiving rituximab were more likely to experience grade 3-5 toxicity than those receiving ibrutinib monotherapy (OR, 3.221; 95% CI, 1.202-8.201; P = .019), with a numerically increased number of grade 3-5 infections in patients receiving rituximab (8/54 [14.8%]) vs monotherapy (3/75 [4%]; OR, 4.174; 95% CI, 1.169-14.99; P = .051).

Forty-four patients (32.1%) died during the study period. The cause of death was MCL (n = 31 [70.5%]), COVID-19 (n = 3 [6.8%]), heart failure (n = 2 [4.5%]), or nonneutropenic infection (n = 2 [4.5%]). Death from stroke, ischemic heart disease, subarachnoid hemorrhage, neutropenic sepsis, suicide, and unknown occurred in 1 patient each.

One hundred and four patients underwent radiological response evaluation during the study period and were evaluable for efficacy analysis, summarized in Table 3. For all patients, ORR was 71.2% (radiological CR, 20.2%; partial response [PR], 51.0%). For patients receiving ibrutinib with vs without rituximab, ORR was 78.7% vs 64.9% (OR, 2.0; 95% CI, 0.84-5.00; P = .135) and CR was 27.7% vs 14.0% (OR, 2.3; 95% CI, 0.89-6.35; P = .093). Forty-five patients were not evaluable for efficacy analysis because response was assessed using clinical criteria only (n = 11) or data were missing (n = 34). The ORR in those with ≥1 high-risk feature (59.0%) was significantly lower (P = .047) than in those without or with missing data (77.3%), although CR rates were very similar: 20.5% vs 19.7% (P = .92). Comparing high-risk sMIPI (≥6) vs low- or intermediate-risk (0-5), ORR was 66.7% vs 75.0%, respectively (P = .50) and CR was 23.5% vs 17.3% (P = .23). When including patients with and without radiological response assessment, 93 patients had ≥6 months follow-up, of whom 81 (87.1%) remained on treatment 6 months after commencing ibrutinib.

At the time of analysis, 16 of 21 patients (76.2%) who achieved CR were in ongoing treatment at a median of 17.9 months (range, 5.9-28.7 months). Among 8 patients with high-risk features who attained CR, 1 died of COVID-19 infection, and 2 were alive with progressive disease.

In total, 39 patients (26.2%) progressed on ibrutinib, and 44 patients (32.1%) died. Estimated median PFS for the whole cohort was 26.0 months (95% CI, 14.4-NR) and 12-month PFS was 61.8% (95% CI, 53.5-71.3). Estimated median OS for the whole cohort was not reached (95% CI, 19.9-NR), and 12-month OS was 69.4% (95% CI 61.3-78.4). Of those with progressive disease at any time, the median time to progression was 5.2 months.

Univariable analysis was carried out to identify baseline features predictive of PFS and OS (summarized in Table 4). The presence of ≥1 high-risk feature, but not sMIPI, was significantly associated with shorter PFS and OS. Median PFS in patients with vs without high-risk features was 13.7 months (range, 5.49 to NR) vs NR (range, NR to NR) (hazard ratio [HR], 2.19; 95% CI, 1.28-3.73; P = .004; Figure 2). Median OS for patients with vs without high-risk features was 14.8 months (range, 11.3 to NR) vs NR (range, NR to NR) (HR, 2.36; 95% CI, 1.35-4.27; P = .005; Figure 2). Patients with blastoid or pleomorphic histology had shorter PFS and OS, with median PFS 4.4 months (range, 3.0 to NR) vs 28.5 (range, 21.6 to NR; HR, 4.31; 95% CI, 2.32-8.01; P ≤ .001) and median OS 10.9 months (range, 4.9-NR) vs NR (range, 21.6-NR; HR, 3.15; 95% CI, 1.58-6.26; P = .001; Figure 3). PFS and OS based on the sMIPI group were not significantly different (median PFS for low, 16.8; intermediate, 28.5; high, 21.6 months; median OS for low, 16.8; intermediate, NR; high, 21.6 months). Among 17 patients with a TP53-positive status and available survival data compared with 29 with TP53-negative status, the median PFS was 6.71 months (range, 4.41 to NR), vs 21.57 months (13.7 to NR) (HR, 2.05; 95% CI, 0.89-4.71; P = .091) and median OS was 11.3 months (range, 6.71 to NR) vs 21.6 (range, 14.9-NR; HR, 2.26; 95% CI, 0.86-5.91; P = .097; supplemental Figure 1).

Multivariable analysis was performed using the following variables in 3 models: lactate dehydrogenase (LDH) to LDH upper limit of normal ratio ≥1 vs <1; WCC ≥ 10 vs WCC < 10; blastoid/pleomorphic vs nonblastoid/pleomorphic histology; Ki67 ≥ 30% vs <30%; age; performance status 0-1 vs ≥2; and presence vs absence of high-risk features. TP53 status was not included because of inconsistency of testing between centers nationally. The same variables previously demonstrated to be associated with inferior outcomes after ibrutinib in the second-line setting¹³ were associated with inferior outcomes in this first-line population, with ECOG performance status ≥ 2 and blastoid/pleomorphic histology, in particular, being strongly predictive (Table 5, model 1). Another model showed patients with ≥1 high-risk feature and poorer ECOG status to be strongly independently predictive of adverse outcomes, having adjusted for other features (Table 5, model 2). Model 3 (age, ECOG performance status, LDH ratio, and blastoid/pleomorphic histology) showed these factors to be consistently predictive of adverse outcome. Because of missing TP53 data, the analyses were repeated with the high-risk category defined only by presence of either blastoid/pleomorphic histology or Ki67 ≥ 30%; these data are available in supplemental Tables 3 and 4 and show similar results.

Of the patients who discontinued ibrutinib, 26 of 62 (41.9%) went on to receive second-line therapy, including a Nordic MCL-2 approach (n = 2), cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP)–based (n = 7), bendamustine-based R-BAC or R-bendamustine (n = 12), VR-CAP (n = 2), R-cytarabine/dexamethasone (n = 1), pirtobrutinib (n = 1), and intrathecal methotrexate (n = 1). The best response to second-line therapy was CR in 6 patients (23.1%), partial response or stable disease in 1 patient each (3.8% each), progressive disease in 9 patients (34.6%), and unknown in 9 patients (34.6%). No patient underwent autologous SCT consolidation.

Both patients treated intensively with the Nordic MCL-2 approach died of progressive disease (1 progressed after cycle 1, and another progressed and died after achieving complete response after 2 cycles). Five patients received third-line treatment (rituximab-CHOP, n = 2; pirtobrutinib, high-dose methotrexate plus rituximab, and R-AraC, n = 1 each). Median post-ibrutinib survival was 1.4 months (95% CI, 0.92-5.16); this was significantly longer in those who received further lines of treatment (median, 8.6 months vs 0.6 months; HR, 0.34; P = .002). Median time to progression on ibrutinib treatment in those who received second-line therapy was 5.5 months.

Management of lymphoma during the COVID-19 pandemic was modified to increase therapeutic flexibility, reduce clinical visits, and minimize immunosuppressive complications that could increase the risk and severity of COVID-19 infection and death.¹⁴ In March 2020, NHS England recommended ibrutinib alone or with rituximab as an option instead of chemotherapy for the treatment of MCL in patients who had not previously received systemic therapy.⁹ In this observational cohort study of 149 patients treated under this scheme and followed up for a median of 15.6 months, ibrutinib with or without rituximab was shown to be active as demonstrated by an ORR of 71.2%, CR 20.2%, and median PFS 26.0 months.

These results are inferior to outcomes of 2 recent phase 2 trials of ibrutinib-rituximab, which reported ORR of 84% to 96%, CR 71% to 80%, and 3-year PFS 87% to 93% in the first-line setting.^7,8 It is possible that responses in this study were underestimated because not all sites used regular radiological assessments of response, in part driven by a desire to minimize exposure to COVID-19 infection in health care environments. This is suggested by 87.1% of patients continuing ibrutinib treatment at 6-month follow-up in this study. Importantly, the aforementioned trial populations were also younger, had significantly fewer patients with high MIPI and ECOG status > 2, and included very few patients with any high-risk features as defined in this study (TP53 mutation/deletion/p53 overexpression, blastoid/pleomorphic variant, or proliferation index [Ki67%/MiB-1] ≥ 30%). These are known prognostic factors that adversely affect outcomes¹¹; in this study, high-risk features were present in 35% and were associated with significantly shorter PFS (median, 9.7 months vs NR; P = .002) and OS (14.4 months vs NR; P = .001).

There is preliminary evidence that outcomes in MCL may be improved by adding rituximab to ibrutinib.^15,16 All patients in the aforementioned first-line phase 2 trials received rituximab in combination with ibrutinib, whereas the use of rituximab in this study varied between centers and was surprisingly low at 39% overall. Low usage likely reflects clinician concern during the pandemic based on reports associating rituximab with more severe COVID-19 outcomes^17,18 and reduced vaccine efficacy.¹⁹ In this study, we observed a trend toward improved ORR and CR rates in patients receiving rituximab, in keeping with evidence to date; however, findings were limited by the small sample size and a bias toward the use of rituximab in more favorable prognostic groups, including those with a good performance status (ECOG status, 0-1) and low risk disease.

The majority of ibrutinib and other BTKi studies in the R/R setting consistently report inferior PFS for high-risk MCL subsets.^4,13,20,TP53-positive cases, the strongest predictor of poor treatment response, early disease progression, and death,^21-23 were likely to have been underrepresented in our population because testing was not routinely or consistently performed in many centers during the study period. In multivariable analyses, which excluded TP53 because of inconsistency of data, blastoid histology, the presence of at least 1 high-risk feature, and ECOG status > 2 emerged as the strongest predictors of adverse outcomes. Although we were unable to specifically address the value of ibrutinib in treating TP53-positive MCL, other real-world studies suggest that BTK inhibitors may mitigate the negative influence of TP53-mutated/17p deleted MCL.²⁴ Adding lenalidomide to rituximab-ibrutinib may also abrogate the negative prognostic impact of TP53 mutation on outcomes, as shown by results of a phase 2 trial in R/R MCL.²⁵ 

At a median follow-up of 15.6 months, 42% of patients in our study discontinued ibrutinib mostly because of progressive disease (22.1%) or treatment toxicity (8.1%). The majority stopping ibrutinib for any reason (58.1%) did not receive further therapy for MCL, despite a good baseline ECOG performance score of 0 to 1 for 75% of the population. The observation that few patients were candidates for second-line therapy reflects a biased selection of older (median age 75 years) and transplant-ineligible patients (92.2%) for upfront ibrutinib therapy within the NHS England scheme, consistent with guidelines at the time advocating the ongoing, risk-balanced use of intensive treatments with known long-term efficacy benefits, such as high dose chemotherapy and autologous transplantation, in patients fit for this approach.^14,26 In line with this, just 2 patients in this study underwent intensive second-line therapy using the Nordic MCL2 approach,²⁷ both of whom died early because of progressive disease.

The poor post-ibrutinib outcomes in this study, even in those who received second-line therapy (median OS, 8.6 months; ORR to second-line therapy, 41%) are not dissimilar to the dismal outcomes reported after ibrutinib discontinuation in the relapsed setting in which failure is characterized by acquired mutations and therapy resistance.^28,29 This observation suggests that ibrutinib-driven selective pressures on disease evolution are independent of the line of treatment and raises the question of whether outcomes could be improved by continuing ibrutinib beyond progression until second-line therapy commences. It also highlights the well-recognized unmet need for more effective therapy in patients who fail BTKi therapy, particularly in older or less fit patients represented in this cohort. Bendamustine-based therapy (R-bendamustine or R-BAC) was the most common second-line therapy in this study, used in 46% of patients, likely influenced by results of a UK retrospective cohort study demonstrating impressive efficacy of R-BAC in patients progressing on BTKi (ORR, 83%; CR, 60%; median PFS, 10.1 months) as well as utility as an effective bridge to allogenic SCT (31%).³⁰ This approach may, however, be too intensive for most patients aged ≥70 years and nontransplant candidates.

Collection of specific toxicity data in this study was limited by the retrospective collection of data restricted to grade 3 or 4 events and relatively short follow-up. Rates of treatment discontinuation due to treatment toxicity (8.1%) were similar to the pivotal phase 2 trial and a real world UK study of single-agent ibrutinib in the relapsed setting,^3,13 although considerably lower toxicity than that reported in the phase 2 study of first-line rituximab and ibrutinib, which noted high rates of atrial fibrillation (22%) and treatment discontinuation (42%) in patients aged >65 years followed for a median of 45 months.⁸ The UK NCRI ENRICH trial (ISRCTN11038174), which exclusively enrolled previously untreated transplant-ineligible patients aged >60 years and did not exclude patients with high-risk biological features, will provide important safety and efficacy data for rituximab-ibrutinib in older, nontransplant eligible patients to inform clinical practice; trial results are expected in 2024.

Second-generation BTKis zanubrutinib and acalabrutinib have also recently been licensed in MCL, offering alternatives to ibrutinib with high efficacy and improved tolerability in the second- and subsequent-line settings^31,32; a randomized trial comparing rituximab-zanubrutinib and rituximab-bendamustine is enrolling a similar older (age >60 years), nontransplant eligible population (NCT04002297).³³ 

There are some important limitations to this study. The study’s observational/retrospective nature and missing data may contribute to underreporting of toxicities, and the relatively small sample size and short follow-up limit survival analysis. Variable use of rituximab and choice of response assessment modality are likely driven by institutional preference, as is approach to TP53 testing. These decisions may also have been influenced by the wider context of the COVID-19 pandemic. The generalizability of our findings is limited by the confounding factors of interinstitutional and interphysician preferences, as well as the diverse patient population included.

In summary, results of this study support a role for ibrutinib with or without rituximab in previously untreated MCL, especially in patients at low risk. Notably, poor outcomes in patients with high-risk MCL, in particular blastoid morphology, and very poor post-ibrutinib survival indicate that ibrutinib is unlikely to be the right approach for these patient groups. These patients may benefit from more intensive chemotherapy-based treatments, but choice is limited in older patients and those unfit for treatment intensification. Randomized trials are ongoing and will provide important evidence to clarify the role of BTKi treatment in previously untreated MCL.

Contribution: A.T. and R.S. analyzed the data and wrote the manuscript; A.T., R.S., T.A.E., D.L., and K.L. conceived and designed the study and assisted with manuscript preparation; L.S. provided statistical support for data analysis; A.C. performed central review of pathology reports, provided input for analysis, and reviewed the manuscript; A.T., R.S., T.E., D.L., L.S., R.A., H.W., F.M., R.Z., A.S., R.M., M.B., A.B., and K.L. reviewed and contributed to the manuscript; and the remaining authors coordinated and contributed data collection from their respective centers and reviewed the manuscript.

Conflict-of-interest declaration: T.A.E. reports education honorarium, advisory board honorarium, and travel support from Roche; honorarium, research support, and travel support to scientific conferences from Gilead; advisory board honorarium from Kite; honorarium from Janssen; honorarium and travel support to scientific conferences from AbbVie; honorarium and research funding from AstraZeneca; advisory board honorarium and trial steering committee fee from Loxo Oncology; advisory board honorarium and research funding from BeiGene; and advisory board honorarium from Incyte and Secura Bio. D.L. reports advisory board and sponsored meeting attendance fee from Janssen. R.M. reports honorarium from Janssen; and research funding from Sanofi. M.B. reports honorarium from Janssen. S.P. reports honorarium and/or travel support to scientific conferences from Gilead, Janssen, AbbVie, AstraZeneca, and Takeda; and advisory board honorarium from Incyte. The remaining authors declare no competing financial interests.

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