Favourable outcomes for high‐risk Burkitt lymphoma patients (IPI 3‐5) treated with rituximab plus CODOX‐M/IVAC: Results of a phase 2 UK NCRI trial

Abstract Introduction Outcomes after frontline treatment of Burkitt lymphoma (BL) have improved with the introduction of dose‐intense chemotherapy regimens, such as CODOX‐M/IVAC. While rituximab has increased survival rates for most forms of high‐grade B‐cell lymphoma, there has previously been hesitancy about incorporating it into BL treatment, partly due to concerns about increased toxicity. Prospective data using the standard dose CODOX‐M/IVAC regimen in combination with rituximab are lacking. We conducted a single‐arm phase 2 trial to assess the efficacy and toxicity of R‐CODOX‐M/R‐IVAC. Methods Eligible patients were aged 18–65 years, with newly diagnosed BL with MYC rearrangement as the sole cytogenetic abnormality, and high‐risk disease, defined by an International Prognostic Index (IPI) score of 3‐5. Patients received two cycles of R‐CODOX‐M chemotherapy alternating with two cycles of R‐IVAC, followed by two further cycles of rituximab alone. The primary endpoint was 2‐year progression‐free survival. Results Thirty‐eight patients were registered but after central pathology review, 27 patients had confirmed BL and commenced study treatment. Median age was 35 years, 14.8% patients had central nervous system involvement and 18.5% were HIV positive. Twenty‐two (81.4%) patients completed four cycles of chemotherapy. There were two treatment‐related deaths (7.4%). Two‐year progression‐free and overall survival rates were 77.2% (90% confidence interval [CI]: 56.0‐89.0) and 80.7% (90% CI: 59.6‐91.5), respectively. Conclusions This prospective trial demonstrates excellent survival rates with R‐CODOX‐M/R‐IVAC in a high‐risk BL cohort. It provides reassuring evidence regarding the feasibility of this regimen and also provides a benchmark for future studies.


INTRODUCTION
Burkitt lymphoma (BL) is a rare and very aggressive form of B-cell lymphoma, accounting for around 2% of all non-Hodgkin lymphomas (NHL) in adults. BL is characterised by the presence of a germinal centre Bcell phenotype, proliferation fraction approaching 100%, and t (8;14) or variant MYC rearrangement (MYC-R) as the sole cytogenetic abnormality [1][2][3]. Atypical features are often present, making the differential diagnosis between BL and other high-grade B-cell lymphomas (HGBL) challenging. Clinically, BL is a rapidly progressive tumour with high rates of extranodal involvement and a propensity to spread to the central nervous system (CNS) [4,5]. Although high cure rates can be achieved, frontline treatment usually offers the only opportunity for disease cure, with very poor outcomes for patients with relapsed or refractory disease [6,7].
Survival rates have improved markedly over recent decades [8,9], largely through the introduction of dose-intense chemotherapy regimens incorporating both CNS penetrating drugs and hyperfractionated alkylating agents [1,[10][11][12]. Most protocols include rapid cycling of multiple non-cross resistant and cell cycle-specific agents, which are administered as soon as haematological recovery occurs to prevent re-emergence of resistant disease clones. The Magrath regimen, consisting of alternating courses of CODOX-M and IVAC chemotherapy, encompasses these principles and has resulted in encouraging survival rates in a number of BL studies [1,[13][14][15].
Several modifications to the original regimen, including a reduction in methotrexate and vincristine doses, have improved tolerability without compromising efficacy [1]. Despite these advances, around one-third of adult patients will progress after frontline CODOX-M/IVAC or similar dose-intense chemotherapy regimens [1,9] and improvements are still needed to reduce failure rates further.
The addition of the anti-CD20 antibody rituximab to standard chemotherapy regimens has considerably improved outcomes and is considered standard of care for most forms of B-cell NHL [16,17]. BL has high levels of CD20 expression and rituximab can induce direct cell death in BL cells in vitro [18]. However, concerns about potential myelotoxic and immunosuppressive effects of rituximab have delayed the adoption of immunochemotherapy into clinical practice for BL.
Indeed, it is only recently that randomised trials have demonstrated that rituximab improves outcomes in BL, when combined with the intensive Lymphome Malin B (LMB) regimen [19]. A preliminary study of rituximab in combination with the intensive hyper-CVAD regimen in BL suggested that the addition of rituximab can improve eventfree survival by as much as 28%, when compared with historical controls [20]. In light of these early encouraging results, we conducted a prospective, multi-centre trial to assess the efficacy and toxicity of rituximab in combination with CODOX-M/IVAC in a high-risk cohort of BL patients.

METHODS
The phase 2 R-CODOX-M trial included two parallel treatment cohorts assessing the same treatment regimen in different types of HGBL. Outcomes for patients in the BL cohort are reported here; the diffuse large B-cell lymphoma (DLBCL) cohort will be reported separately. This design permitted crossover of patients between the two trial arms following central pathology review, which facilitated the inclusion of a uniform diagnostic cohort of confirmed BL patients (see Figure 1).

Eligibility
Eligible patients were aged 18-65 years with stage II-IV, previously untreated CD20 + HGBL and an International Prognostic Index (IPI) score of 3-5 [21]. BL was defined as HGBL with a germinal centre phenotype, absent BCL-2 expression, high proliferation rate (>90%) and the presence of a MYC-R, without BCL2 or BCL6 rearrangement, consistent with current WHO guidelines [3]. Original diagnostic material was centrally reviewed by the Leeds Haematological Malignancy Diagnostic Service.
Inclusion criteria included adequate liver, renal, cardiac and bone marrow function, unless directly attributable to disease infiltration.
Performance status (PS) was permissive. In light of emergent data

Study treatment
There was no pre-phase chemotherapy, although prior administration of corticosteroids for up to 10 days was permitted. All patients received either allopurinol or rasburicase before treatment, according End-of-treatment response was assessed 4 weeks after completion of chemotherapy using contrast-enhanced CT according to international guidelines [22]. The study design predated widespread availability of 18 fluorodeoxyglucose positron emission tomography (PET) for response assessment but was encouraged to assess residual masses.
Radiotherapy was not included in this trial but was permitted for those with initial disease bulk or CNS disease, and for residual PET-positive disease.

Endpoints and statistical methods
The primary endpoint was the progression-free survival (PFS) rate at 2 years. Secondary endpoints included complete response (CR) rate, overall survival (OS) and toxicity.
Using a Fleming design, recruiting at least 30 BL patients would give 80% power to show an improvement of 20% at 2 years (from 65% to 85%) with a 1-sided 5% alpha. The trial was also open to high-risk DLBCL patients (95 required) and, given that the LY10 trial had noted that as many as 50% of patients with highly proliferative HGBL were not true BL [1], we therefore assumed that a reasonable number may switch cohorts after central review, and a total target of 150 was fixed.
Recruitment was to be stopped once this target was reached, irrespective of the number of BL patients treated.
PFS was calculated as the time from registration until either disease progression or death, on an intention-to-treat basis, however those found to be ineligible after registration or those who withdrew before starting treatment (for reasons unrelated to their disease) were excluded. Patients who were alive and progression-free were censored at the date last seen. All analyses were performed using Stata v15.1 (Stata Corp, TX). In total, 27 BL patients were included in this study. Baseline patient characteristics are shown in Table 1. CNS disease was present in four patients (14.8%), although only CSF cytology, but not immunophenotyping, was mandated, therefore low-level subclinical leptomeningeal involvement could not be excluded.

Toxicity
Details of all grade 3-5 toxicities are presented in Table2

Patient outcomes
Overall response rate was 85.2%, with 21 patients (77.8%) achieving CR or unconfirmed CR (CRu) and two partial responses (7.4%). One patient had progressive disease during treatment and three patients were not evaluable due to study withdrawal (n = 2) or death (n = 1).
After a median follow-up of 56.9 months (range 2.

DISCUSSION
Although the addition of rituximab to CODOX-M/IVAC is considered standard of care for high-risk BL in a number of countries worldwide [23,24], it was adopted into routine care without any direct clinical trial evidence in this patient group. This is the first trial to prospectively assess the efficacy of standard CODOX-M/IVAC with rituximab. Two-year PFS with this regimen was 77.2% (90% CI: 60.1-87.6), which compares favourably to a PFS of 60-65% for high-risk BL  comparisons between studies, an apparent improvement in PFS was observed in our trial despite recruitment of higher risk patients (IPI score 3-5) than in LY10 (IPI score ≥ 2). Other studies have also shown improved outcomes with rituximab in BL [24][25][26]. A randomised phase 3 trial demonstrated a 13% improvement in PFS with rituximab for BL patients treated with the dose-intense LMB regimen [19]. We considered using a randomised trial design, including an arm without rituximab, but felt this would lack equipoise in light of the clear survival benefit with rituximab in DLBCL [16,27] and excellent preliminary results in BL [20]. Our results add to the growing body of evidence that immunochemotherapy should be considered standard of care in BL.
One limitation of this study is the sample size. We did not meet the 2-year PFS rate of ≥85% we hoped for, with the 90% confidence interval (60.2-87.6%) unable to exclude the lower limit of 65%. We had aimed for 30 patients, but recruited 27 with confirmed BL, due to high rates of crossover after central pathology review. However, the main reason that we did not meet our endpoint is that our initial prediction of a 20% improvement in PFS, based on a historical comparison, was F I G U R E 3 Progression-free survival (A) and overall survival (B) according to IPI score over-optimistic [20]. Although the trial was not powered to demonstrate smaller improvements, our observed PFS rate (77.2%) still represents a clinically meaningful improvement and is consistent with the magnitude of benefit seen with rituximab in other studies [19].
Evidence for use of R-CODOX-M/R-IVAC has largely been based on retrospective studies, which report PFS rates of 74-81% and OS rates of 72-77% [24,25,28,29]. These studies were heterogeneous inate prognosis for the 80-90% of BL patients identified as 'high risk' [30,32]. The IPI has consistently been identified as a prognostic measure in BL, along with age and PS [4,26,33,34]. Indeed, a large, recent BL real-world data study showed that other high-risk features, such as CNS involvement, were associated with IPI and failed to retain independent prognostic significance after correction for IPI [35]. Our BL cohort, defined by an IPI score ≥3 and with 48% PS ≥2, is therefore higher risk than most 'high-risk' BL studies, even though patients were relatively young (median age of 36 years).
Our outcomes are broadly in line with IPI 3-5 patients receiving dose-intense immunochemotherapy regimens in other trials, all of which include hyperfractionated cyclophosphamide and CNSpenetrating agents. The GMALL B-NHL2002 trial reported a 4-year OS of 75% for patients with an IPI score of 3-5 [4]. The CALGB10002 study reported 4-year OS of 72% for IPI score 3 and 55% for IPI score 4-5 [34]. Very good survival rates have been achieved in trials of lower intensity dose-adjusted EPOCH-R therapy, although these include patients with a higher median age but otherwise lower clinical risk profile than our trial, therefore comparison between regimens is not possible [32]. One advantage of R-CODOX-M/R-IVAC is the shorter duration of chemotherapy: 14 weeks, compared with 18-26 weeks for other high-risk BL regimens [4,19,34,36]. Notably, a recent retrospective study that demonstrated a favourable profile with respect to cost and treatment duration for R-CODOX-M/R-IVAC compared with other intensive BL regimens [37].
Toxicity in our study was largely manageable, demonstrating that CODOX-M/IVAC with rituximab is safe and feasible in high-risk BL.
Even though most patients had widespread, advanced-stage disease, full dose intensity and concurrent administration of rituximab with R-CODOX-M was deliverable from day 1 in most patients, with only one instance of tumour lysis. Only two patients (7.4%) discontinued treatment for toxicity-related reasons, which support real-world data showing high completion rates with R-CODOX-M/IVAC [37]. Many regimens include a short chemotherapy pre-phase to reduce tumour burden, although this trial and other studies suggest that this may not be required in most cases [24]. We found no evidence to suggest that toxicity with rituximab in this trial was higher than with CODOX-M/IVAC alone. In particular, haematological recovery, reflected in the length of each treatment cycle, was very similar in our trial to the previous LY10 trial [1]. G-CSF was administered in both studies, although in pegylated form in this study. It is important to note, however, that both LY10 and the DLBCL arm of this trial identified higher toxicity rates in older patients, therefore this intensive regimen is only applicable to fit older patients with a relatively good PS [38].
A strength of this trial was the use of central pathology review to determine allocation to either the BL or DLBCL/HGBL trial arms, thus identifying a uniform cohort with confirmed BL. Diagnostic discrepancy rates for BL were high: 11 of 36 patients (30.6%) registered as BL switched to DLBCL/HGBL after central review. Larger BL trials have identified non-BL diagnoses in 10-21% of patients on central review [15,19,34]. Even specialist haematopathologists may be unable to reach diagnostic consensus in up to 35% of BL cases [39]. This highlights the challenge in differentiating BL from other forms of HGBL in clinical practice, and the importance of comprehensive central pathology review in BL studies. Nevertheless, R-CODOX-M/R-IVAC was also effective in the BL-like patients that were reclassified to DLBCL/HGBL in this study and therefore may be effective in borderline or atypical BL cases.
In summary, this prospective trial confirms that the rituximab with CODOX-M/IVAC is deliverable, safe, and effective, with encouraging survival rates in a high-risk BL cohort. Our findings provide a benchmark for future studies and confirm that R-CODOX-M/R-IVAC should remain a recommended regimen for the treatment of BL. Randomised trials are needed to determine the optimum immunochemotherapy approach in BL. An international randomised trial comparing R-CODOX-M/R-IVAC with lower intensity dose-adjusted EPOCH-R therapy is currently underway