Introduction

The “Positron Emission Tomography-Guided Therapy of Aggressive Non-Hodgkin Lymphomas” (PETAL) trial investigated the ability of interim [18F]-fluorodesoxyglucose (FDG) PET scanning to guide treatment in aggressive non-Hodgkin lymphomas [1]. Patients whose scan remained positive after two cycles of cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP) (plus rituximab (R) in CD20-positive lymphomas) were randomly assigned to receive another six cycles of R-CHOP or an intensive methotrexate- and cytarabine-based protocol yielding excellent results in pediatric non-Hodgkin lymphomas [2] and adult Burkitt’s lymphoma [3]. Using the ΔSUVmax method for PET evaluation [4], interim scanning reliably predicted outcome, but treatment intensification failed to improve survival in the study population as a whole and in all subgroups analyzed [1].

During the first 2 years of recruitment, patients with a negative interim scan uniformly received six cycles of R-CHOP (two cycles before and four after interim scanning). When publications suggested that increasing the cumulative dose of rituximab may improve outcome in CD20-positive lymphomas [5, 6], the protocol was amended, and patients with CD20-positive lymphomas and a negative interim scan were randomly assigned to receive six cycles of R-CHOP or the same treatment with two additional doses of rituximab. Because negative interim scans were more frequent than positive scans, randomization in the PET-negative part was terminated earlier than in the PET-positive part. After the end of randomization, interim PET-negative patients with CD20-positive lymphomas uniformly received eight doses of rituximab, which, by then, was considered the standard of care for aggressive B cell lymphomas [7].

Within the group of 255 randomized patients with interim PET-negative CD20-positive lymphomas, increasing the exposure to rituximab failed to improve outcome [1]. Because the trial included several histological entities, the group of randomized patients was too small for meaningful subgroup analyses. Adding the patients treated before or after the randomization period to the randomized group almost tripled the number of interim PET-negative patients with CD20-positive lymphomas. This allowed us to study rituximab exposure in three major subtypes of aggressive B cell lymphoma: diffuse large B cell lymphoma (DLBCL), primary mediastinal B cell lymphoma (PMBCL), and follicular lymphoma (FL) grade 3 [8].

Methods

Patients

Patients 18 to 80 years of age with newly diagnosed aggressive B cell or T cell lymphomas and an ECOG performance status ≤ 3 were eligible for registration. Lymphoblastic, Burkitt’s, transformed indolent, and primary central nervous system lymphomas were excluded. A reference pathological review was obtained in 98% of cases [1]. All patients gave written informed consent.

Procedures

The PETAL trial was registered under EudraCT 2006-001641-33 and ClinicalTrials.gov NCT00554164 and performed according to the 1964 Declaration of Helsinki and its later amendments in 55 oncological and 23 nuclear medicine sites in Germany. The protocol was approved by the Federal Institute for Drugs and Medical Devices and the ethics committees of the participating sites. The trial design has been described previously [1]. This analysis is limited to patients with CD20-positive lymphomas and a negative interim PET scan. After baseline investigations, patients received a pre-phase [9] consisting of vincristine (1 mg, day 1) and prednisone (100 mg for 3–7 days) followed by two cycles of R-CHOP (rituximab, 375 mg/m2, day 1; cyclophosphamide 750 mg/m2, day 2; doxorubicin 50 mg/m2, day 2; vincristine 2 mg, day 2; prednisone 100 mg, days 2–6; granulocyte colony-stimulating factor, 5 μg/kg, daily from day 5 until recovery of neutrophils > 1/nl). Patients with CD20-positive lymphomas and a negative interim scan received a further four cycles of R-CHOP with or without two additional doses of rituximab in 14-day intervals.

All efficacy and safety assessments were done by the primary investigators using the international response criteria for malignant lymphomas [10] and the Common Terminology Criteria for Adverse Events [11]. The end-of-treatment remission status was assessed by computed tomography (CT). Follow-up visits were scheduled according to international guidelines [10].

Statistical analysis

The primary end-point was event-free survival, measured from the date of interim PET scanning to treatment failure (progression, relapse, change to a treatment inconsistent with the trial protocol, toxicity-related treatment termination, death from any cause). The randomized comparisons in the interim PET-positive and interim PET-negative parts of the trial have been reported previously [1]. This account covers interim PET-negative patients with CD20-positive lymphomas who were randomized or allocated to six cycles of R-CHOP with or without two additional doses of rituximab. The analysis was done in the intention-to-treat population, and included response, event-free, progression-free, overall survival, and safety. Planned subgroup analyses included lymphoma subtype, age, sex, and International Prognostic Index (IPI) risk group [12]. All analyses were exploratory. Time-to-event end-points were analyzed using the Kaplan-Meier estimator and the log-rank test. All tests were two-sided applying an exploratory alpha of 0.05. Cox proportional hazard regression model was employed to adjust for effects of the stratification variables used at randomization (sex, age (18–50, 51–60, 61–70, 71–80 years), IPI risk group) [1]. The same factors were considered for propensity score matching which was carried out with a 1:1 case/control ratio.

Results

Demographic data

Between November 2007 and December 2012, 862 patients with aggressive non-Hodgkin lymphomas were treated in the PETAL trial. Six hundred seven had CD20-positive DLBCL, 42 PMBCL, 25 FL grade 3a, and 17 FL grade 3b. Median baseline maximum standardized uptake values were 20.5 for DLBCL (range, 3.5–75.2), 21.6 for PMBCL (7.9–54.6), 12.8 for FL grade 3a (2.4–39.0), and 12.6 for FL grade 3b (6.5–53.8). The proportion of patients with a negative interim PET scan was 89.7% in DLBCL, 88.1% in PMBCL, and 83.3% in FL grade 3. In the interim PET-negative group, 197 of 544 DLBCL, 14 of 37 PMBCL, and 9 of 35 FL patients underwent randomization between six cycles of R-CHOP and six cycles of R-CHOP with two additional doses of rituximab. Randomization was restricted to the recruitment period between February 2010 and October 2011. Before that time, interim PET-negative patients with CD20-positive lymphomas uniformly received six R-CHOP cycles, and thereafter, they were uniformly treated with six R-CHOP cycles with two additional doses of rituximab. Because baseline data and treatment results (Electronic Supplementary Material, Tables S1 and S2) did not significantly differ between randomized and non-randomized patients, the groups were combined for this analysis.

Table 1 describes patients’ baseline characteristics. PMBCL patients were more often female and tended to be younger than DLBCL and FL patients. The features of patients receiving six or eight doses of rituximab were well balanced.

Table 1 Characteristics of patients with aggressive B cell lymphomas and negative interim positron emission tomography findings treated with six or eight doses of rituximab
Full size table

Treatment adherence was good. Five hundred and seventy-eight of a total of 616 interim PET-negative patients (93.8%) received six cycles of CHOP (range, 2–8), 304 of 335 patients (90.8%) allocated to six, and 255 of 281 patients (90.8%) allocated to eight doses of rituximab received the planned treatment. Among patients allocated to six rituximab doses, eight (2.4%) erroneously received seven or eight. Major reasons for premature treatment termination were toxicity (n = 14) and patient’s preference (n = 5). Progression on therapy with a switch to an alternative regimen was rare (n = 2). The median follow-up time was 52 months (interquartile range, 40–64).

Outcome in diffuse large B cell lymphoma

There were no statistically significant differences in response rate, event-free, progression-free, and overall survival between DLBCL patients receiving six or eight doses of rituximab (Table 2, Fig. 1). Stratification according to the IPI showed a statistically non-significant trend for improved event-free survival in the low-intermediate risk group favoring six and a similar trend in the high-risk group favoring eight doses of rituximab, with a weaker effect on overall survival (Fig. 2).

Table 2 Remission and survival rates in patients with aggressive B cell lymphomas and negative interim positron emission tomography findings treated with six or eight doses of rituximab
Full size table
Fig. 1
figure 1

Event-free survival and overall survival in interim positron emission tomography-negative patients with diffuse large B cell lymphoma. The patients were randomized or allocated to receive six cycles of R-CHOP (rituximab, cyclophosphamide, doxorubicin, vincristine, prednisone) or the same treatment with two additional doses of rituximab. Survival started on the day of interim positron emission tomography (PET) scanning. Shaded areas correspond to 95% Hall-Wellner confidence bands

Full size image
Fig. 2
figure 2

Event-free survival and overall survival in interim positron emission tomography-negative patients with diffuse large B cell lymphoma in relation to the risk group of the International Prognostic Index. The patients were randomized or allocated to receive six cycles of R-CHOP (rituximab, cyclophosphamide, doxorubicin, vincristine, prednisone) or the same treatment with two additional doses of rituximab. Survival started on the day of interim positron emission tomography (PET) scanning. Shaded areas correspond to 95% Hall-Wellner confidence bands

Full size image

In subgroups defined by sex or age, survival differences between six and eight doses of rituximab were not observed (data not shown). However, when sex and age were combined, female patients below the age of 60 years had significantly inferior event-free survival when treated with eight as compared to six doses of rituximab, with a concomitant trend for decreased overall survival (Fig. 3). Fatal cases were restricted to patients 51 to 57 years of age. Causes of death included relapse (n = 3), infection (n = 3), and unknown reason (n = 1) among 66 patients receiving eight doses of rituximab, as compared to relapse (n = 1) and unknown reason (n = 1) among 48 patients treated with six doses. Female patients above the age of 60 years and male patients of any age failed to show statistically significant survival differences between six and eight doses of rituximab (Fig. 3).

Fig. 3
figure 3

Event-free survival and overall survival in interim positron emission tomography-negative patients with diffuse large B cell lymphoma in relation to sex and age. The patients were randomized or allocated to receive six cycles of R-CHOP (rituximab, cyclophosphamide, doxorubicin, vincristine, prednisone) or the same treatment with two additional doses of rituximab. Survival started on the day of interim positron emission tomography (PET) scanning. Shaded areas correspond to 95% Hall-Wellner confidence bands

Full size image

Stratification factors in the randomized subset included age, sex, and IPI risk group. To account for imbalances in these factors, the data were subjected to multivariable Cox regression analysis. The results obtained in the unadjusted population were confirmed after adjustment for confounding factors (Table S3). For further corroboration, the groups receiving six or eight doses of rituximab were harmonized by propensity score matching, yielding a total of 430 patients. The results in the matched population were similar to those obtained in the other analyses, but trends for subgroup-specific survival differences were less pronounced (Table S3, Figs. S1–S3).

Outcome in primary mediastinal B cell lymphoma

Although the rate of complete morphological remission, as assessed by CT, was lower in PMBCL than it was in DLBCL, event-free and overall survival tended to be superior (Table 2, Fig. 4). Two patients received consolidative mediastinal radiotherapy. There were no statistically significant differences between six or eight doses of rituximab (Table 2). This was confirmed in the Cox model and the propensity score matched population, but the validity of the results was limited by small numbers (data not shown).

Fig. 4
figure 4

Event-free survival and overall survival in interim positron emission tomography-negative patients with diffuse large B cell lymphoma (DLBCL), primary mediastinal B cell lymphoma (PMBCL), follicular lymphoma (FL) grade 3a, or follicular lymphoma grade 3b. The patients received six cycles of R-CHOP (rituximab, cyclophosphamide, doxorubicin, vincristine, prednisone) with or without two additional doses of rituximab. Survival started on the day of interim positron emission tomography (PET) scanning. Shaded areas correspond to 95% Hall-Wellner confidence bands

Full size image

Outcome in follicular lymphoma grade 3

As in DLBCL and PMBCL, the number of rituximab doses had no impact on outcome in FL (Table 2). This was confirmed in the Cox model and the propensity score matched population (data not shown). Statistically significant differences in the course of FL grade 3a and FL grade 3b were not observed (Fig. 4) which was confirmed by propensity score matching (Fig. S4).

Safety

The frequency and severity of side effects was similar in patients receiving six or eight doses of rituximab (Table S4). Grade 3 or 4 anemia (P = 0.0001), leukopenia (P < 0.0001), infection (P < 0.0001), diarrhea (P = 0.0081), and creatinine increase (P = 0.0011) were significantly more frequent above than below the age of 60 years. Treatment-related death (1.6%) also tended to occur more often in the older age group (P = 0.0530). Sex-related differences were limited to anemia (P = 0.0086) and leukopenia (P = 0.0863) which were more frequent in female than in male patients.

Discussion

In the PETAL trial, two additional doses of rituximab failed to improve survival in interim PET-negative B cell lymphoma patients treated with six cycles of R-CHOP. A negative interim scan indicates chemotherapy sensitivity and translates into good long-term outcome [4]. Patients with a positive interim scan uniformly received eight doses of rituximab. Their outcome was poor, indicating therapy resistance [1]. Whether changes in rituximab exposure would impact survival in patients with resistant lymphoma was not investigated. Thus, our conclusions are limited to chemotherapy-sensitive lymphomas which comprised almost 90% of cases.

Several prospective trials have tested the impact of rituximab dose on outcome in aggressive B cell lymphomas. In the single-arm DENSE-R-CHOP-14 trial, tripling the number of rituximab doses in the first two of a total of six R-CHOP cycles led to a significant increase in infection-related morbidity and mortality, but had no impact on response rate or survival, as compared to a historical control treated with only one rituximab dose per cycle [13]. In the randomized HOVON84 trial, doubling the dose of rituximab in the first four R-CHOP cycles did not improve progression-free survival in the study population as a whole or in any of the sex- and age-related subgroups analyzed [14]. Likewise, in the GOYA trial, replacing eight standard doses of rituximab by ten more elevated doses of the novel CD20 antibody obinutuzumab failed to improve outcome [15]. In the PETAL trial, only DLBCL high-risk patients had a potential benefit of increased rituximab exposure. This, however, did not reach statistical significance and was less apparent in the propensity score matched population.

Further evidence of the limited value of increasing rituximab exposure in DLBCL is provided by randomized trials investigating maintenance therapy. In the US Intergroup study ECOG4494/CALGB9793, rituximab maintenance had no effect on outcome in newly diagnosed patients above the age of 60 years treated with rituximab and CHOP [16]. Similar observations were made in the LNH98-3 trial in which young high-risk patients received first-line high-dose chemotherapy with or without rituximab maintenance [17]. The AGMT-NHL13 trial included adult patients of any age in first complete remission after a rituximab-containing induction. In the trial population as a whole, rituximab maintenance did not improve outcome, but male patients had a statistically significant benefit with regard to event-free and progression-free survival [18]. Similar observations were made in the HD2002 trial [19]. By contrast, in the CORAL study exploring maintenance in relapse, only female patients appeared to benefit from prolonged rituximab exposure [20].

Male patients of any age and female patients below the age of 60 years eliminate rituximab faster than older female patients do. This has led to the assumption that the antibody may be under-dosed in the first-named patient groups [5, 6]. In the PETAL trial, none of these subgroups benefitted from increased rituximab exposure. Unexpectedly, female patients between 50 and 60 years of age fared worse with eight as compared to six rituximab doses. The increase in mortality was related to relapse and infection. Our conclusion that increased rituximab exposure may be hazardous in female patients is supported by safety data from the rituximab maintenance arm of the AGMT-NHL13 trial in which adverse events in general and infection in particular were more frequent in female patients than they were in male patients [18].

Long-term outcome in PMBCL was excellent, with only one of 37 patients dying within the observation period. Five patients with a positive interim PET scan were not included in the present analysis. Although two of them responded poorly to first-line therapy, none died (data not shown), raising the survival rate in the total PMBCL population to 98%. These results are in line with a subgroup analysis of the UK National Cancer Research Institute R-CHOP-14 versus R-CHOP-21 trial, which suggested that PMBCL patients may benefit from short treatment intervals [21]. In the PETAL trial, R-CHOP was given in 14-day intervals, except for the interval between cycles 2 and 3 which was 3 weeks (to avoid false-positive interim PET findings) [1]. In the UK trial, 58% of patients received additional radiotherapy. By contrast, in the PETAL trial, only two of 37 patients with a negative and one of five patients with a positive interim PET scan were subjected to mediastinal irradiation (7%). Thus, two-weekly R-CHOP is an excellent treatment option for PMBCL, and PET can identify patients who may be spared radiotherapy. Our results are similar to those reported in the dose-adjusted EPOCH-R trial in which the overall survival rate in PMBCL was 97%, and PET monitoring reduced the frequency of radiotherapy to 4% [22]. Administration of EPOCH-R, however, is cumbersome, with dose modifications from cycle to cycle and prolonged infusion of etoposide, vincristine, and doxorubicin, which may necessitate hospitalization. In addition, etoposide is a leukemogenic agent, putting patients with otherwise excellent prognosis at undue risk of developing a life-threatening secondary disease [23].

The PETAL trial included both grade 3a and grade 3b of FL. Whether these grades reflect differences in natural history remains controversial. Some studies suggest that the course of grade 3a is indolent, resembling grades 1 and 2, while grade 3b behaves like an aggressive lymphoma [24]. Others come to the conclusion that, although the biology of grade 3a and grade 3b may differ, their clinical course is largely indistinguishable [25,26,27]. Although our observations are limited by small numbers, with the R-CHOP regimen, clinical differences between grade 3a and grade 3b were not apparent.

In conclusion, increasing the exposure to rituximab did not improve outcome in B cell lymphomas with a fast metabolic response to R-CHOP. Irrespective of lymphoma entity, six doses of rituximab appear to be sufficient.