Patterns of delivery of chemoimmunotherapy to patients with follicular lymphoma in the United States: Results of the National LymphoCare Study
Peter Martin MD,
Division of Hematology-Oncology, Department of Medicine, Weill Cornell Medical College, New York, New York
Corresponding author: Peter Martin, MD, Division of Hematology-Oncology, Department of Medicine, Weill Cornell Medical College, 525 East 68th St, New York, NY 10065; Fax: (646) 962-1605; firstname.lastname@example.org
Drug choice and delivered dose of treatment potentially influence outcome in patients treated for follicular lymphoma (FL). Historically, observational studies have evaluated drug choice. The National LymphoCare Study (NLCS) is a prospective, observational study of patients with FL who were enrolled at academic and community practice sites in the United States between 2004 and 2007. In the current study, the authors report on measures of delivered dose and its impact on outcomes for the most common first-line regimens.
All evaluable patients with FL who were treated with initial rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP); rituximab plus cyclophosphamide, vincristine, and prednisone (R-CVP); or rituximab plus a fludarabine-containing regimen (R-Flu) were included. Associations between baseline factors, choice of treatment, number of cycles received, completion of therapy, and patient outcomes were assessed.
A total of 646 patients received R-CHOP, 297 received R-CVP, and 222 received R-Flu. Characteristics were similar between the 3 groups with the following exceptions. Patients receiving R-CHOP were more often found to have grade 3 FL and patients receiving R-CVP were older and had higher Follicular Lymphoma International Prognostic Index scores. The majority of patients (80%) received ≥ 5 cycles of treatment. Toxicity, but not disease progression, was commonly cited as the reason for the early discontinuation of treatment (51% vs 6%). Time to retreatment was shorter for patients receiving ≤ 4 cycles, regardless of the treatment regimen used. The number of cycles was associated with overall survival, progression-free survival, and lymphoma-related mortality for patients receiving R-CVP.
Follicular lymphoma (FL) is the most common type of indolent non-Hodgkin lymphoma in North America and is characterized by a heterogeneous clinical presentation and outcome. To the best of our knowledge, there is no single accepted standard of care for the first-line treatment of this disease. Observational studies have traditionally provided data regarding drug regimens. However, drug delivery factors, including total dose and intensity, also may influence outcomes. The delivery of chemoimmunotherapy in standard clinical practice may be different from when treatment is administered as part of a clinical trial, in which patients are highly selected and adherence to protocol is required. Factors such as age, stage of disease, and prognostic scores impact treatment choices and delivery in patients with diffuse large B-cell lymphoma (DLBCL),[4, 5] but to the best of our knowledge little is known about the impact of these factors on the delivery of therapy to patients with FL, and little is known about the association between delivered dose and clinical outcomes.
The National LymphoCare Study (NLCS) is a prospective, observational study of > 2700 patients with FL who were enrolled at academic and community practice sites in the United States between 2004 and 2007. We previously reported patterns of initial management strategies, demonstrating that just over one-half of patients are initially managed with chemoimmunotherapy. Of those patients treated with chemoimmunotherapy, the most prevalent regimen was rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP) (50%), followed by rituximab plus cyclophosphamide, vincristine, and prednisone (R-CVP) (23%) and rituximab plus a fludarabine-containing regimen (R-Flu) (17%), as well as R-other combinations (9%). We now report for what to our knowledge is the first time a quantitative assessment of delivered dose and intensity for the 3 most common regimens. We also described patient factors associated with dose delivery as well as associations between delivered dose and patient outcomes.
MATERIALS AND METHODS
Between 2004 and 2007, patients with FL who were diagnosed within the previous 6 months and with no prior history of lymphoma were recruited from US community practices and academic centers. All patients provided informed consent. Collected data included demographics such as date of birth, race, prior malignancies, and clinical trial participation; clinical data such as Eastern Cooperative Oncology Group (ECOG) performance status, Ann Arbor stage of disease (highest known stage and studies used to determine staging), the presence of B symptoms, number of lymph node sites and extranodal sites, and bone marrow or splenic involvement; laboratory data such as complete blood count, lactate dehydrogenase (normal vs higher) and albumin; serial management strategies including drug choices but not doses; response to treatment; and events such as disease recurrence and death. Management decisions were made by the treating physician according to clinical judgment. No treatment regimen or prescribed intervention was specified, and the number of cycles/doses required for treatment completion was not stipulated a priori.
All treatments (drugs and start/stop dates and number of cycles but not doses) that patients received for FL were recorded, including the initial treatment regimen and any subsequent therapy. Growth factor use was not recorded. Treating physicians were asked if therapy was completed and if so, how many cycles were administered. The reason for discontinuation was requested when treatment was not completed, as was reason for the initiation of all subsequent treatments. When initial management was observation, this was recorded as the first strategy. Patients who received therapy within 90 days of their date of diagnosis were not included in the observation cohort but were categorized by the initial treatment received. To avoid bias that might arise from prespecified criteria, patients enrolled in a prospective clinical trial dictating treatment were analyzed as “investigational therapy” irrespective of the regimen delivered on the trial and were not reported in the current analysis.
All evaluable patients receiving treatment with R-CHOP, R-CVP, or R-Flu were included in the current analysis. Individuals with FL plus other lymphoma histology or no FL histology were excluded. Patients who received treatment after an initial management strategy of observation were not included, nor were patients who received initial chemotherapy as part of combined modality therapy. Patients who went on to receive a maintenance course of therapy were analyzed according to their induction chemoimmunotherapy.
Response and disease progression were assessed locally, not centrally, in accordance with published International Working Group criteria. Progression-free survival (PFS) was defined as the time from the date of diagnosis to the date of first disease progression or death from any cause. Time to next treatment (TTNT) was defined as the time from the date of treatment initiation to the date of initiation of a next treatment for any reason. Overall survival (OS) was defined as the time from the date of diagnosis to the date of death from any cause. Lymphoma-related mortality (LRM) was defined as the time from the date of diagnosis to the date of death from lymphoma-related causes or toxicity related to lymphoma treatment. Cycle intensity was defined as the expected cycle length (21 days for R-CHOP and R-CVP and 28 days for R-Flu) divided by the actual cycle length (ie, a cycle intensity > 1.1 indicated greater-than-expected intensity, a cycle intensity of 0.9-1.1 denoted cycles of expected length, and a cycle intensity < 0.9 indicated less-than-expected intensity).
The associations between baseline characteristics and choice of treatment (R-CHOP, R-CVP, or R-Flu), the number of cycles received (≤ 4, 5-6, and > 7), and completion of therapy were assessed using the Pearson chi-square test for categorical variables and analysis of variance for continuous variables. Associations between cycle intensity and treatment were assessed using the Pearson chi-square test. Factors associated with the risk of early discontinuation (treatment, age, stage of disease, and grade) were assessed using multivariate logistic regression. Factors associated with the number of treatment cycles were assessed using a generalized logits model. Age-by-stage, stage-by-grade, and all 2-way interactions with treatment were examined using backward stepwise selection (using P > .10). Kaplan-Meier survival methods were used to generate estimates and curves for OS, PFS, TTNT, and LRM stratified by treatment regimen and by the number of treatment cycles within each regimen. Associations between the number of treatment cycles for each treatment regimen and OS, PFS, TTNT, and LRM were assessed using Cox regression with an interaction term for the number of cycles by treatment regimen and adjusting for Follicular Lymphoma International Prognostic Index (FLIPI) components (with age categorized as ≤ 50 years, 51 years-60 years, 61 years-70 years, and ≥ 71), ECOG performance status, and the use of maintenance rituximab.
The NLCS enrolled 2731 patients at 266 sites; approximately 80% of the patients were enrolled at community sites. A total of 1165 patients were considered to be evaluable for the current analysis (Fig. 1). The demographic and disease characteristics of the evaluable patients who were treated with R-CHOP, R-CVP, or R-Flu are summarized in Table 1. A total of 646 patients were initially managed with R-CHOP chemotherapy, 297 were initially managed with R-CVP chemotherapy, and 222 were treated with R-Flu combinations. Patients in the 3 treatment groups were similar except that those receiving R-CHOP were more likely to have grade 3 FL and those receiving R-CVP were older and had higher FLIPI risk scores.
Abbreviations: ECOG, Eastern Cooperative Oncology Group; FLIPI, Follicular Lymphoma International Prognostic Index; R-CHOP, rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone; R-CVP, rituximab plus cyclophosphamide, vincristine, and prednisone; R-Flu, rituximab plus a fludarabine-containing regimen; SD, standard deviation.
Bold type indicates statistical significance (P < .05).
The treatment characteristics are detailed in Table 2. The distribution of the number of cycles delivered demonstrated a high degree of compliance with published regimens. The majority of patients (80%) received at least 5 cycles of treatment, and cycle intensity was as expected for the majority of patients treated with R-CHOP and R-CVP, but not for those who received R-Flu, among whom a higher percentage of patients experienced both a greater-than-expected and less-than-expected cycle intensity. Generalized logits modeling comparing the odds of receiving ≤ 4 treatment cycles versus 5 to 6 cycles and > 7 treatment cycles versus 5 to 6 cycles (Table 3) showed that compared with patients aged ≤ 50 years, those aged > 71 years had a greater odds of receiving ≤ 4 cycles (odds ratio [OR], 2.37; 95% confidence interval [95% CI], 1.44-3.89) and a lower odds of receiving ≥ 7 cycles (OR, 0.68; 95% CI, 0.39-1.18). Compared with individuals with stage I to II disease, those with stage III to IV disease had a reduced odds of receiving ≤ 4 cycles (OR, 0.55; 95% CI, 0.38-0.80). There was a significant interaction between grade and the effect of the treatment selected on the number of treatment cycles received (P = .036). For patients with grade 1 to 2 disease, R-CVP was associated with a greater odds of receiving ≥ 7 cycles (OR, 1.61 [95% CI, 1.03-2.52] compared with R-CHOP), whereas R-Flu was associated with a greater odds of receiving ≤ 4 cycles (OR, 2.23 [95% CI, 1.40-3.59] compared with R-CHOP). For patients with grade 3 disease, the R-Flu regimen was again associated with a greater odds of receiving ≤ 4 cycles (OR, 9.78 [95% CI, 3.59-26.60] compared with R-CHOP).
R-CHOP, rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone; R-CVP, rituximab plus cyclophosphamide, vincristine, and prednisone; R-Flu, rituximab plus a fludarabine-containing regimen; SD, standard deviation.
Patients who still had ongoing treatments (n = 6) at the time of last follow-up were not included in this table.
Bold type indicates statistical significance (P <.05).
Determined using analysis of variance.
Determined using the Pearson chi-square test.
Cycle intensity was defined as the expected cycle length (21 days for R-CHOP and R-CVP and 28 days for R-Flu) divided by the actual cycle length (a cycle intensity > 1.1 indicated greater-than-expected intensity, a cycle intensity of 0.9-1.1 denoted cycles of expected length, and a cycle intensity < 0.9 indicated less-than-expected intensity).
Abbreviations: 95% CI, 95% confidence interval; OR, odds ratio; R-CHOP, rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone; R-CVP, rituximab plus cyclophosphamide, vincristine, and prednisone; R-Flu, rituximab plus a fludarabine-containing regimen.
ORs were estimated using a generalized logits model. Two-way interactions were evaluated for inclusion in the model using backward stepwise selection (P > .10).
Bold type indicates statistical significance (P < .05).
Treatment by grade
Given grade 1 or 2
Given grade 3
Ann Arbor stage
I or II
III or IV
Analysis of Premature Cessation of Therapy
We sought to understand the reason for patients receiving < 5 cycles of therapy. Figure 2 illustrates the percentage of patients who completed treatment by treatment arm and number of cycles. Physicians reported that therapy was not completed in 10% of patients treated with R-CHOP, 13% of patients treated with R-CVP, and 20% of patients treated with R-Flu. Logistic regression analysis indicated that early discontinuation was associated with patient age ≥ 71 years (OR, 2.78 [95% CI, 1.56-4.96] compared with age < 50 years) and first-line treatment with R-Flu (OR, 2.18 [95% CI, 1.36-3.49] compared with first-line therapy with R-CHOP and OR, 1.12 [95% CI, 0.69-1.84] for R-CVP compared with R-CHOP; data not shown). Reasons for treatment discontinuation are listed in Table 4. Toxicity was the most commonly cited reason for early discontinuation, whereas disease progression was rarely cited.
Table 4. Reasons for Treatment Discontinuation
Reason, No. (%)
n = 63
n = 38
n = 44
Abbreviations: R-CHOP, rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone; R-CVP, rituximab plus cyclophosphamide, vincristine, and prednisone; R-Flu, rituximab plus a fludarabine-containing regimen.
The percentage of patients achieving a complete or partial response was associated with the number of cycles received only for those patients treated with R-CVP (P = .005), with those receiving ≤ 4 treatment cycles (73.8%) having a lower response rate compared with those receiving 5 to 6 cycles (90.7%) or ≥ 7 cycles (91.8%) (data not shown). Although not statistically significant (P > .05), a similar trend was noted for R-CHOP and R-Flu. After adjusting for baseline factors and the use of rituximab maintenance, the number of treatment cycles was associated with OS, PFS, and LRM for patients receiving R-CVP (Table 5). The number of treatment cycles was not associated with OS, PFS, or LRM for patients receiving R-CHOP or R-Flu. TTNT was significantly associated with the number of cycles for R-CHOP, R-CVP, and R-Flu, with those patients who received ≤ 4 cycles having a shorter interval between first-line and second-line treatment than those receiving ≥ 5 cycles (Table 5). A sensitivity analysis was conducted excluding patients who discontinued planned treatment early (eg, patients who may have not been able to tolerate a full course of treatment). A similar relation was noted between the number of treatment cycles and OS, PFS, LRM, and TTNT for R-CVP. However, for R-CHOP and R-Flu, there was no longer a significant association observed between the number of treatment cycles and TTNT.
Table 5. OS, LRM, PFS, and TTNT by Treatment and Number of Cycles of Therapy and Baseline Factorsa
Abbreviations: 95% CI, 95% confidence interval; ECOG, Eastern Cooperative Oncology group; HR, hazards ratio; LDH, lactate dehydrogenase; LRM, lymphoma-related mortality; OS, overall survival; PFS, progression-free survival; PS, performance status; R-CHOP, rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone; R-CVP, rituximab plus cyclophosphamide, vincristine, and prednisone; R-Flu, rituximab plus a fludarabine-containing regimen; Ref, reference; TTNT, time to next treatment; ULN, upper limit of normal.
All multivariate Cox proportional hazard models were adjusted for follow-up treatment (R-maintenance/observation).
Bold type indicates statistical significance (P < .05).
Ann Arbor stage
Lymph node sites
In the current study, we report what to the best of our knowledge is the first large prospective study of patterns of delivery for common first-line chemoimmunotherapy regimens in patients with FL in the United States. The findings suggest that adherence to the recommended total number of delivered cycles, as well as the cycle intensity, is relatively high in the United States. This was true for R-CHOP, which was the most commonly prescribed regimen during the 3 years of this observation study, as well as for R-CVP, and was less apparent for fludarabine-based chemoimmunotherapy regimens. The administration of < 5 cycles or > 6 cycles did not appear to have a significant impact on patient outcomes for patients treated with R-CHOP or R-Flu. Patients treated with < 5 cycles of R-CVP were found to have inferior outcomes across all clinical endpoints in a multivariate Cox hazard model that adjusted for maintenance therapy. If these results can be replicated elsewhere, they should give caution to clinicians considering the early cessation of R-CVP. It is interesting to note that this study does not address the choice of a specific chemotherapy regimen or the use of posttreatment rituximab maintenance.
The association between early discontinuation and the R-Flu regimen is consistent with prospective randomized studies demonstrating the increased toxicity of fludarabine-based regimens. The phase 3 ECOG 1496 trial was designed with a 2 × 2 factorial design comparing frontline treatment with CVP versus fludarabine and cyclophosphamide (FC) and rituximab maintenance versus observation. Due to an increased toxic death rate, the FC arm was closed and the remaining patients were all treated with CVP. The FOLL05 trial from the Fondazione Italiana Linfomi randomized 534 patients with previously untreated follicular lymphoma to treatment with either R-CHOP, R-CVP, or rituximab plus fludarabine and mitoxantrone (R-FM). Patients treated with R-FM experienced significantly more grade 3 to 4 neutropenia than those receiving R-CHOP or R-CVP (64% vs 50% vs 28%). In addition, patients treated with R-FM appeared to experience a higher rate of subsequent cancers. Similar findings were reported in a phase 3 trial in older patients with mantle cell lymphoma. Although the results of the current study indicated that toxicity was cited as the most common reason for not completing planned therapy, it accounted for early discontinuation in only 11% of patients receiving R-Flu regimens, suggesting that attempts to make fludarabine-based treatment more tolerable are unlikely to impact many patients. The TTNT outcome was likely worse in patients treated with R-Flu who received ≤ 4 cycles due to early changes in therapy due to toxicity. This is supported by our observation that TTNT was not affected by the number of cycles administered to patients receiving R-Flu or R-CHOP when only those patients who were considered to have completed therapy were analyzed (data not shown).
Contrary to population studies in patients with DLBCL, the delivery of R-CHOP in patients with FL in the current study was excellent.[4, 11] Predictors of low chemotherapy dose among patients with DLBCL are reported to include older age, advanced stage of disease, poor performance status, and no prophylactic filgrastim. Due to the curative intent of therapy in patients with DLBCL, treating physicians are likely to select R-CHOP despite the probability of early discontinuation in poor-risk patients. The older age and higher FLIPI scores of the patients treated with R-CVP in the current study suggest that clinicians avoided R-CHOP in patients with FL who were less likely to tolerate it at the recommended dose intensity, which is an appropriate strategy given the lack of a proven OS benefit of R-CHOP over R-CVP in patients with FL. The lack of an association between the number of cycles of R-CHOP received and OS, PFS, and LRM suggests that some patients with FL may not require a full 6 cycles of treatment and that early surrogates for duration of benefit (eg, interim 18F-fluorodeoxyglucose–positron emission tomography status) should be evaluated.
Despite their elevated age and higher FLIPI scores, > 80% of patients receiving R-CVP completed at least 5 cycles, and only 13% did not complete planned therapy, including 19 of 297 patients (6%) who withdrew from treatment prematurely because of toxicity. Nonetheless, patients who received ≤ 4 cycles of R-CVP had worse OS, PFS, TTNT, and LRM compared with patients receiving 5 to 6 cycles. It is difficult to reconcile the low rate of early cessation of therapy because of toxicity or disease progression with the poor PFS and LRM. At the very least, it appears that stopping treatment prematurely did not benefit these patients, and it is unknown whether improving the delivery of R-CVP would have made a significant difference. Although the delivery of ≤ 4 cycles of R-CVP was associated with worse survival, this analysis does not implicate causality.
There are several limitations to the current study. Treatment decisions were made by the treating physician, with no guidelines regarding treatment choice, duration, criteria for withdrawing or changing treatment, or criteria for retreatment. Because this was an observational study, treatment decisions were not randomized and thus may be biased by patient selection and other confounding factors. Therefore, replication in other patient populations is needed. Nevertheless, the results of the current study are reflective of clinical practice outside the clinical trial setting. The completion of therapy was defined by the treating physician after therapy was stopped, not a priori. Initially, the number of treatment cycles received by patients who discontinued treatment early was not captured in the database and was instead collected retrospectively. When sensitivity analyses that excluded 128 patients from 46 closed sites were performed, similar results were obtained.
Due to the time period over which patients on the NLCS were enrolled, there are no data regarding the frontline use of bendamustine. More recently, multiple phase 3 trials have demonstrated the safety and efficacy of the combination of bendamustine and rituximab compared with R-CHOP, R-CVP, or R-Flu.[12-14] Despite the increased toxicity of the non-bendamustine treatment arms in those studies, treatment appeared to be delivered as expected. Data from the current study support the concept that differences in efficacy between bendamustine and the other regimens are not due to improved tolerability of the drug.
Unlike patients with DLBCL, to the best of our knowledge very little has been published to date regarding the importance of dose delivery in patients with FL. Recent phase 3 trials have demonstrated that neither dose intensity (high-dose therapy plus autologous stem cell transplantation) nor dose density (R-CHOP–14 vs R-CHOP–21, ie, R-CHOP given every 14 days vs every 21 days) impacts OS, and dose density has no impact on PFS. The findings of the current study suggest that the delivery of R-CHOP, R-CVP, and R-Flu regimens in the United States is generally complete. With the possible exception of R-CVP, strategies to improve dose delivery appear unlikely to affect patient outcomes. Rather, the development of effective, well-tolerated, novel therapies should be pursued.
The National LymphoCare Study (NLCS) is sponsored through a partnership between Genentech Inc (South San Francisco, Calif) and Biogen Idec (Cambridge, Mass). The NLCS Advisory Board members participated in all phases of the study, including initial protocol design, prospective determination of data to be collected, and consideration of participating sites.
CONFLICT OF INTEREST DISCLOSURES
Dr. Martin is a member of the Speakers' Bureau for Genentech. Drs. Byrtek and Dawson are employed by Genentech, and hold stock options in Roche, the parent company of Genentech. Dr. Ziemiecki has received fees for conducting the analysis and for writing or reviewing the article from RTI Health Solutions through a contract with Genentech to RTI Health Solutions, his employer. Dr. Friedberg is a member of the board of Lilly and has received royalties. Drs. Cerhan, Flowers, and Link are members of the Advisory Board of and have received travel support from Genentech.