Comparison of first-line treatment with bendamustine plus rituximab versus R-CHOP for patients with follicular lymphoma grade 3A: Results of a retrospective study from the Fondazione Italiana Linfomi

In the setting of follicular lymphoma (FL), frontline therapy with rituximab, cyclophosphamide, doxorubicin, and prednisone (R-CHOP) has represented for many years the standard of care for patients with symptomatic advanced disease. More recently, the combination of bendamustine plus rituximab (R-B) has emerged as an alternative therapeutic option. We present a retrospective, multicenter, observational study aimed at comparing outcomes and toxicities observed in 145 patients diagnosed with grade 3A FL treated with a first line therapy in 15 Italian Fondazione Italiana Linfomi centers between the 1st of January 2014 and the 30th of May 2018. Seventy patients were treated with R-B and 75 with R-CHOP. In the R-B group, the median age at the time of diagnosis was 67 years compared with 59 years in the R-CHOP group. Patients in R-B group achieved a similar overall response rate (96% vs. 99%) and a better complete remission rate (87% vs. 80%, p=0.035) compared with patients in R-CHOP group. Progression free survival (PFS) was similar between individual treated with R-CHOP and R-B (48- month PFS 77.7% vs. 76.6% respectively, p=0.745). The overall survival was significantly longer with R-CHOP treatment (HR=0.16; 95% IC, 0.04-0.74; p=0.007); however, no statistical significant difference was observed after adjustment for age. With the limitations of the study design, our results suggest that both R-B and R-CHOP seem to be valid first-line treatment options in FL3A.


Introduction
Follicular lymphoma (FL) is the second most common subtype of non-Hodgkin lymphoma (NHL) and represents the 20-30% of NHL cases (1). In Western Countries, its incidence is 2-3 cases/ 100.000 inhabitants/year (2, 3). FL derives from germinal center Bcells, centrocytes (small cleaved follicular center cells), and centroblasts (large non-cleaved follicular center cells). The 2017 World Health Organization (WHO) classification of FL is based on a histologic grading (1, 2, 3A and 3B) based on the number of centroblasts in routine histologic assessment: 0-5 centroblasts per high-power field (HPF) define grade 1; 6-15 centroblasts per HPF define grade 2; more than 15 centroblasts per HPF is accounted as grade 3, which is again subdivided into grade 3A when residual centrocytes are present and grade 3B when solid sheets of centroblasts are present and centrocytes absent (4,5).
Pathological findings and staging affect the identification of a suitable therapeutic approach (6). For limited stages (I or II Ann Arbor stages) FL, radiotherapy represents a chance of curative treatment (7). For patients with advanced stages, a systemic therapy should be started in presence of high tumor burden, according to the Groupe d'Etude des Lymphomas Folliculaires (GELF) criteria (8). Grade 1/2 FL is treated like an indolent lymphoma (9), whereas grade 3B FL has distinct cytomorphological, immunohistochemical and cytogenetic profiles -with a well-known aggressive nature -that requires a treatment analogous to that of diffuse large B cell lymphoma (DLBCL) (10, 11). The behavior of grade 3A FL is still matter of debate (12)(13)(14). Some studies suggest an indolent course, whereas others indicate that it is an aggressive but potentially curable lymphoma (15)(16)(17)(18). Hence, the definition of an optimal treatment for grade 3A FL is controversial. In this setting, historical frontline therapeutic regimens included the administration of rituximab in combination with an anthracycline-based chemotherapy such as cyclophosphamide, doxorubicin, vincristine and prednisone (CHOP) (19)(20)(21) but more recently also bendamustine plus rituximab has become an increasingly common treatment option (22). One large randomized trial (StiL NHL1) demonstrated the superiority of rituximab and bendamustine (R-B) compared to R-CHOP in terms of outcome and toxicity, leading to improved survival (23). Another international randomized study (BRIGHT) confirmed the non-inferiority of R-B compared to R-CHOP/R-CVP (rituximab, cyclophosphamide, vincristine, and prednisone) (24). However, the main limit of these two large trials is a study population of grade 1/2 FL only, excluding patients with grade 3A FL.
To date, no data from prospective randomized study are available. Moreover, only three retrospective studies have compared R-B and R-CHOP treatment in the setting of grade 3A FL, with different outcomes (25)(26)(27). One retrospective multicenter study (25) suggested the advantages of R-B treatment in grade 3A FL; in a second retrospective analysis (26), similar outcomes in terms of progression-free survival (PFS) and overall survival (OS) were observed in both R-B and R-CHOP regimen; finally, in a third retrospective investigation, patients treated with R-CHOP showed a significant longer OS compared to individuals treated with R-B (27).
Here, we present the results of a retrospective multicenter analysis of patients aged ≥ 18 years diagnosed with grade 3A FL and treated either with R-CHOP or R-B as first-line therapy, with the aim of improving our clinical understanding of their comparative efficacy and safety.

Patients
We retrospectively assessed all consecutive patients affected by FL grade 3A treated with a first line therapy in 15 Italian Fondazione Italiana Linfomi (FIL) centers between the 1 st of January 2014 and the 30 th of May 2018. Histologic diagnosis was performed according to the international diagnostic criteria by an expert lymphoma pathologist of each participating center (4). Patients with evidence of grade 3B or with histologic transformation to DLBCL in the biopsy were excluded. Data collection and analysis were approved by the local ethical committee at each center. Informed consents were collected by all alive patients. The study was conducted in accordance with Good Clinical Practice guidelines and the provisions of the Declaration of Helsinki.

Treatment plan
All patients received a first-line immunochemotherapy, consisting of R-B or R-CHOP according to the center's policy. Dose or cycle adjustments occurred in case of significant comorbidity or toxicity, at the investigator's discretion. Patients could receive maintenance treatment with rituximab according to the treatment guidelines of the participating center. The use of granulocyte-colony stimulating factors (G-CSF) was allowed at the investigator's discretion.

Statistical analysis
All patients were evaluated for response to therapy according to international criteria of the 2014 Lugano classification (6) whereas toxicity was classified according to the National Cancer Institute's Common Toxicity Criteria. The primary endpoint was progression free survival (PFS), defined as the time between first treatment and one of the following events: progressive disease, relapse after response, or death from any cause. Secondary endpoints were overall survival (OS), overall response rate (ORR), complete response rate (CR), acute and late toxicity. OS was calculated from the date of diagnosis to date of last follow-up or death. Continuous data were reported as median (I, III quartiles); categorical data were reported as percentage and absolute frequencies. Wilcoxon test was performed for continuous variables, while Pearson Chi-square or Fischer exact test (whatever appropriate) was performed for categorical variables. The time until relapse and mortality was modeled by using a competitive risk approach with a cumulative incidence function. The final estimates were adjusted by gender and age by using a Fine and Gray approach including a frailty term accounting for correlation within the center. The analyses were performed using R 3.5 and the cmprisk package. The limit of significance for all analyses was defined as p < 0.05.

Clinical characteristics at the time of diagnosis
A total of 145 patients were included in the study. Seventy patients were treated with R-B and 75 with R-CHOP. Baseline

Treatment and response
Patients received a median of 6 cycles of chemoimmunotherapy in both treatment groups (range 2-6 for R-CHOP, 4-6 for R-B). Dose reduction was necessary in 6 and 3 patients who underwent R-B and R-CHOP, respectively. Interruption of therapy was necessary in 1 patient treated with R-B after 5 cycles (caused by infection) and 1 patient treated with R-CHOP after 2 cycles (caused by intestinal perforation). The R-B treatment was associated with a less frequent use of central venous catheter compared to R-CHOP (26% vs. 76%).
Sixty-four patients (85%) received maintenance treatment with rituximab every 2 months for 2 years after R-CHOP, and 58 patients (83%) after R-B. Early discontinuation of maintenance therapy was observed for 8 patients (11%) in the R-CHOP group (5 due to progressive disease, 3 due to severe infections) and 8 patients (11%) in the R-B group (5 due to progressive disease, 2 due to second neoplasms, and 1 due to patient decision).

Follow-up
Median follow-up, defined as the time between initiation of first-line treatment and last patient contact, was 54.6 months (range 48.2-60.9 months) and 50.8 months (range 43.9-57.8 months) for the R-B and R-CHOP groups, respectively. Patients who were refractory to induction therapy or relapsed after achieving a response were 16 in the R-B group (23%) and 17 in the R-CHOP group (23%, p=0.258). Specifically, in the R-B group, 3 patients were refractory to R-B induction, 5 patients relapsed during maintenance treatment and 5 patients post maintenance treatment, while 3 patients did not receive rituximab maintenance. Among patients relapsed after R-CHOP treatment, 1 patient was refractory to induction, 5 patients relapsed during maintenance and 6 post maintenance, 5 patients did not receive maintenance. Progression with histological transformation was observed in 5 patients in the R-B group (of these 4 died) and in 3 patients in the R-CHOP group (of these 1 died). A total of 11 patients (85% of the relapsed patients) from the R-B group and 15 patients (94% of the relapsed patients) from the R-CHOP group received a second-line treatment consisting, in most cases, of chemo-immunotherapy. R-CHOP as second line therapy was administered to 6 patients treated with R-B first (on a total of 16 relapsed/refractory patients), and R-B was used as second line therapy after R-CHOP for 5 patients (on a total of 17 relapsed/refractory patients). Eleven patients (16%) died after R-B treatment, with 8 lymphoma-related deaths (other causes of deaths didn't correlate with lymphoma or toxicities). Two patients of the R-CHOP group (3%) died both due to progressive lymphoma.
Four-year PFS among the two groups was superimposable: 76.6% in the R-B group and 77.7% in the R-CHOP group (Figure 1).
OS was significantly longer with R-CHOP than with R-B (Figure 2), (48 months OS 98.7% vs. 84.9% respectively, p=0.007); Progression-free survival (PFS). however, this difference appeared to be influenced by the higher age of the patients in the R-B group as no statistical difference was observed after a final competitive risk analysis adjusted by age (Table 4).

Discussion
Follicular lymphoma is a pathologically and clinically heterogeneous disease with different possible clinical presentations and relevant therapeutic approaches. While the current therapeutic program is standardized for grades 1 or 2 and for grade 3B FL (9)(10)(11), the choice of the most suitable therapeutic regimen for grade 3A follicular lymphoma in clinical practice is still an open question (12-14), with no yet clear evidence regarding the therapeutic efficacy and safety of R-B compared with R-CHOP.
Two prospective trials compared these two therapies in the setting of indolent lymphoma (23,24). In the StiL NHL1 study, the rate of complete response was significantly higher in the R-B group (39.8%) with respect to R-CHOP (30%), with a median PFS more than doubled in the R-B group (69.5 vs. 31.2 months) and a lower rate of toxic effects (23). The BRIGHT study also compared R-B vs. R-CHOP/R-CVP in indolent lymphoma (70% follicular lymphoma) and demonstrated the non-inferiority of R-B treatment, showing a complete response rate of 31% for R-B and 25% for R-CHOP/R-CVP (24). However, only grade 1 and 2 follicular lymphomas have been considered in these two randomized studies, hence data from prospective studies in grade 3A FL are still missing.
Three retrospective analyses so far have tried to compare the outcome with R-B and R-CHOP treatment in grade 3A FL, providing different observations (25)(26)(27). Mondello et al. performed a multicenter retrospective analysis comparing outcomes of R-CHOP and R-B treatment in 132 patients, obtaining similar CR rates in both cohorts, but a significant longer median PFS and less toxicity for patients treated with R-B compared with R-CHOP, i.e., 15 and 11.7 years (p=0.03), respectively. The analysis demonstrated a 3-year OS rate above 90% with no difference between R-CHOP and R-B (25). On the other hand, Shah et al. evaluated the results observed in 103 patients who received rituximab or ofatumumab in combination with anthracycline-based chemotherapy (65 patients), bendamustine (30 patients) or CVP (8 patients). The results showed lower CR rates among patients who received bendamustine, but similar PFS and OS. Regarding the OS, 24-and 60-month rates were 92% and 82% for R-CHOP compared with 86% and 74% for R-B. The differences, however, did not show statistical significance (26). In another German retrospective analysis, a total of 95 patients with grade 3A FL and patients with coexisting grade 1-2 and 3A FL were treated first-line with either R-CHOP or R-B. The PFS-difference did not show statistical significance, while OS was significantly longer after treatment with R-CHOP compared with R-B, i.e., 3years OS of 89% versus 73%, respectively (27). These results may reflect the heterogeneity of FL3A/FL1-2-3A. Similarly to Shah et al. study (26), our analysis did not found statistical differences in PFS and OS in a population of 145 patients with grade 3A FL; the OS improvement observed in the R-CHOP group in the first analysis was not confirmed in the final competitive risk analysis adjusted by age, demonstrating the absence of statistical significant difference in OS in the two groups. In fact, even in presence of homogeneous distribution of most of clinical characteristics, in our study the two cohorts presented a significantly different median age, with older patients in the group treated with R-B compared to patients treated with R-CHOP and, specifically, more patients > 75 years old in the R-B group (16%) compared to the R-CHOP group (3%). We believe that older age might have influenced the choice of treatment regimen, reserving R-B for patients with clinical aspects of greater fragility, not eligible for anthracycline-based therapy, and that this bias might have influenced overall survival of group treated with R-B.
Although more patients died in the R-B group, the rate of transformation was similar across treatment groups, with 5 patients in the R-B group and 3 patients in the R-CHOP group experiencing lymphoma transformation.
Our analysis included only patients with grade 3A FL. ORR was similar in the two treatment groups (96% in R-B group and 99% in R-CHOP group), similarly to the results of the StiL trial (ORR of 93% versus 91% in patients with FL grades 1 and 2 treated with R-B and R-CHOP, respectively) (23), while in the BRIGHT trial ORR in R-B group was superior to that of R-CHOP group (97% vs. 91%, respectively) (24). The percentages of CR in both groups of our analysis were very high (87% vs. 80% with R-B and R-CHOP, respectively), higher than CR rates in both prospective trials (40% vs. 30% in the StiL trial and 31% vs. 25% in the BRIGHT trial) (23,24), probably for the inclusion in the latter two of different types of indolent non-Hodgkin lymphoma besides grade 3A FL.
Different from the results of the two prospective studies (23, 24) and other retrospective analysis (25)(26)(27), we observed similar rates of hematological and non-hematological toxicities in the two groups, as well as late toxicities in terms of secondary primary malignancies. Specifically, our study indicated a higher rate of grade 3-4 hematological toxicities in the R-B group (50%) compared to previous analysis (23)(24)(25)(26)(27). This discrepancy was likely dependent by the advanced age of patients in the R-B cohort compared to the R-CHOP cohort. However, the higher proportion of hematological toxicities did not translate into a statistically significant increase of the rate of infections.
Several limitations are evident in our study. The retrospective study design can lead to bias in the study population, with difficulties to consider patients and disease variables that play an important role in the choice of chemotherapy regimen. For example, the age difference between the two groups considered in our study represent an important bias, which may have influenced both the choice of treatment regimen and the observed results. Another limitation of the present study is related to the histologic diagnosis, which was performed by the pathology department of each participating center in absence of a centralized review system for the accurate distinction between grade 3A FL and other grades.

Conclusions
The choice of optimal therapeutic regimen for grade 3A follicular lymphoma is still an open question. The present multicenter retrospective analysis aimed at providing further data on this clinical challenge. Our results did not show significant differences in PFS and OS between R-CHOP and R-B for the first line treatment of grade 3A FL; thus, both treatments remain appropriate frontline options for this subset of patients, and the choice of one or the other regimen may be dictated by the presence of comorbidities or patient preference. However, data from prospective clinical trials is still warranted to define the specific role of these two different therapeutic strategies for the treatment of grade 3A FL.

Data availability statement
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Ethics statement
The studies involving human participants were reviewed and approved by the ethics committee of each center involved in the study. The patients/participants provided their written informed consent to participate in this study. Publisher's note All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.