Prognostic impact of RDI of vincristine in patients with DLBCL receiving R-CHOP: a supplementary analysis of JCOG0601

Diffuse large B-cell lymphoma (DLBCL) is the most common subtype of malignant lymphoma.¹ R-CHOP (rituximab, cyclophosphamide [CY], doxorubicin [DXR], vincristine [VCR], and prednisolone) has been the standard of care for DLBCL for >20 years, and it cures approximately two-thirds of patients.^2,3 Maintaining the dose intensity (DI) of R-CHOP is important to achieve the expected treatment outcomes.^4-7 In particular, the DIs of CY and DXR are considered to play an important role in their efficacy; nonetheless, in some cases, they must be reduced due to toxicities, such as myelosuppression, infection, and cardiotoxicity.^4,7 The primary toxicity associated with VCR is neurotoxicity. Some patients experience long-lasting peripheral neuropathies (PNs), such as neuralgia and sensory and motor neuropathies, which severely affect their daily activities and quality of life.^8,9 Paralytic ileus or severe constipation may also occur.⁹ Therefore, the DI of VCR is sometimes reduced based on the severity of VCR-related neurotoxicity. However, inconsistent results have been reported from retrospective data regarding the impact of VCR dose reduction on the prognosis of patients with DLBCL, which remains controversial.^10-12 Recently, VCR dose reduction was reported not to be associated with poor prognosis in older adult patients with aggressive B-cell lymphoma, based on the post hoc analysis of the prospective clinical trial RICOVER-60^13,14; however, in the RICOVER-60 trial, R-CHOP was administered in 14-day cycles rather than in the standard 21-day cycles.

Several risk factors for the development of PN, such as VCR dose, age, sex, and race, have been reported but not clearly established.^9,15,16 By identifying patients with high-risk factors for VCR-related neurotoxicity, physicians can take preemptive measures to prevent the development of VCR-related neurotoxicity, such as early dose reduction and discontinuation of VCR.

Considering the abovementioned background, we conducted a supplementary analysis of JCOG0601, a randomized phase 2/3 trial performed by the Japan Clinical Oncology Group (identifier: jRCTs031180139).¹⁷ This study aimed to evaluate (1) the clinical impact of the relative DI (RDI) of VCR (RDI_O) and (2) the risk factors for developing VCR-related neurotoxicity in patients with DLBCL treated with 21-day cycles of R-CHOP.

JCOG0601 is a randomized phase 2/3 trial to confirm the optimal dosing schedule for rituximab. R-CHOP and CHOP with 8 doses of weekly rituximab (RW-CHOP) were compared for progression-free survival (PFS) in untreated patients with DLBCL.¹⁷ A total of 422 patients meeting the following criteria were enrolled in JCOG0601 between December 2007 and December 2014: previously untreated CD20⁺ DLBCL according to the World Health Organization classification, third edition; Ann Arbor stages I to IV; aged 20 to 79 years; Eastern Cooperative Oncology Group performance status (PS) 0 to 2; and no central nervous system involvement. There was no significant difference in PFS between the 2 arms (hazard ratio [HR], 0.95; 90.6% confidence interval [CI], 0.68-1.31). The dose of each drug was reduced due to adverse events, as outlined in the protocol. Regarding VCR, the dose was initiated at 1.4 mg/m² (maximum 2.0 mg) and reduced based on toxicities, including PN (neuralgia, motor, or sensory), ileus, and hyponatremia due to the syndrome of inappropriate secretion of antidiuretic hormone. No specific supportive treatment for VCR-related toxicities was outlined in the protocol other than VCR dose reduction. The outcomes of VCR-related toxicities after dose reduction were not evaluated within the scope of this study. Additional details of JCOG0601, such as protocol treatments and dose reduction criteria for each drug, are summarized in the supplemental Appendix.

In this study, we used the data from JCOG0601 with a cutoff date of 19 December 2017. Because PFS was not significantly different between the R-CHOP and RW-CHOP arms, this supplementary analysis was performed without distinguishing between the treatment arms. The patient flow diagram for this study is shown in Figure 1. Of the 422 patients enrolled in JCOG0601, 401 who received at least 6 courses of R-CHOP (n = 206) or RW-CHOP (n = 195) were analyzed to investigate the impact of the RDI_O on PFS. To evaluate the risk factors associated with the incidence of VCR-related neurotoxicity, 421 patients who received at least 1 course of the protocol treatment were analyzed.

The RDI represents the ratio of the amount of drug actually administered per body surface area (BSA) during the actual period (actual DI) to the amount of drug planned to be administered per BSA during the planned period (planned DI).¹⁸ In addition to RDI_O, 2 subgroups were classified based on the RDI of CY and DXR (RDI_C+H): the high RDI_C+H subgroup comprised patients showing an RDI_C+H of ≥85% for both drugs, whereas the low RDI_C+H subgroup included those showing an RDI_C+H of <85% for 1 or both drugs.^6,10,19 The details of the RDI calculation method are provided in the supplemental Appendix. VCR-related neurotoxicity is defined as neuralgia, sensory PN, motor PN, and ileus of grade ≥2 that are considered to affect patients’ quality of life.

The impact of reduced RDI_O on PFS was first assessed using several cutoff values of RDI_O (95%, 90%, 85%, 80%, 70%, 60%, and 50%). The cutoff value that best divided PFS into 2 groups using the log-rank test was used in further analyses. PFS, according to the RDI_O, was estimated using the Kaplan-Meier method. In subgroup analysis, the impact of RDI_O on PFS was evaluated in the high and low RDI_C+H subgroups. Univariable and multivariable analyses for PFS were performed using the Cox proportional hazards model, including the following factors as covariates: RDI_O, sex, the presence of B symptoms, the presence of a bulky mass (≥5 cm), International Prognostic Index (IPI) risks, RDI_C+H subgroups, and the cell of origin (COO; germinal center B cell–like [GCB] vs non-GCB) according to the Hans algorithm.²⁰ 

The univariable and multivariable analyses for VCR-related neurotoxicity were performed using the log-binomial model with the following factors as covariates: age, sex, BSA ≥1.43 m² (which means the maximum VCR dose of 2 mg was reached), Eastern Cooperative Oncology Group PS, presence of gastrointestinal involvement, having diabetes mellitus as a comorbidity, serum albumin level, and having an elevated liver enzyme (alanine aminotransferase) level at baseline. All P values were 2 sided. All statistical analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC).

The study protocol of JCOG0601 was approved by the protocol review committee of Japan Clinical Oncology Group and the respective institutional review boards.

The baseline clinical characteristics of the 401 patients are found in Table 1. The median age was 62 years (range, 20-79), 54.6% of the patients were male, and 97.5% had a PS of 0 or 1. The disease was classified as Ann Arbor stage I or II in 53.4% of the patients. In addition, 12.5% of the patients had B symptoms, and 82.5% had a low or low-intermediate IPI risk. The median RDI_O of the 401 patients was 97% (interquartile range, 89%-100%), and only 16 patients (4%) had an RDI_O of <50%. The median RDI_C and RDI_H were both high (96% [interquartile range, 88%-100%]; supplemental Figure 1).

At the data cutoff date, the median follow-up duration was 59.4 months, and 3-year PFS was achieved in 81.5% (95% CI, 77.3-85.0) of the 401 patients (Figure 2A). Regarding PFS, among the cutoff values investigated, 95% of RDI_O (<95% [n = 161] and ≥95% [n = 240]) showed the smallest P value (log-rank 2-sided, P = .0679), and 3-year PFS was achieved in 83.7% (95% CI, 78.4-87.8) of patients with high RDI_O (≥95%) and 78.2% (95% CI, 71.0-83.8) of those with low RDI_O (<95%; Figure 2B). PFS results and HR for PFS of low RDI_O against high RDI_O with other cutoff values of RDI_O are found in the supplemental Appendix and supplemental Figure 2. The number and percentage of patients who experienced VCR dose reductions or discontinuations, the time to progression, and the sites of recurrence in each group (RDI_O < 95% and RDI_O ≥ 95%) are summarized in the supplemental Appendix and supplemental Table 1.

In the high RDI_C+H subgroup (n = 324), patients with high RDI_O (≥95%) tended to have better PFS than those with low RDI_O (<95%; 3-year PFS, 85.2% vs 79.8%; P = .0904), a trend similar to that in the overall population (Figure 3A). In contrast, there was no difference in PFS between patients with high RDI_O (≥95%) and those with low RDI_O (<95%) in the low RDI_C+H subgroup (n = 77; 3-year PFS, 73.3% vs 74.1%; P = .8576; Figure 3B).

According to the univariable analysis, the presence of B symptoms (HR, 1.918; 95% CI, 1.147-3.205), high-intermediate or high IPI risk (HR, 1.732; 95% CI, 1.072-2.798), low RDI_C+H (HR, 1.795; 95% CI, 1.149-2.804), and a COO finding of not available (NA) or deficit (HR, 2.009; 95% CI, 1.138-3.545) were significant risk factors for poor PFS, whereas low RDI_O was not (HR, 1.452; 95% CI, 0.971-2.171). Multivariable analysis revealed that only the presence of B symptoms (HR, 1.859; 95% CI, 1.078-3.207), high-intermediate or high IPI risk (HR, 1.721; 95% CI, 1.031-2.873), and a COO finding of NA or deficit (HR, 2.076; 95% CI, 1.170-3.684) were associated with poor PFS (Table 2).

The baseline clinical characteristics of the 421 patients are summarized in supplemental Table 2. A total of 127 patients (30.8%) experienced ≥1 VCR-related neurotoxicities of grade ≥2, including neuralgia (n = 44 [10.5%]), sensory PN (n = 107 [25.4%]), motor PN (n = 32 [7.6%]), and ileus (n = 7 [1.7%]; supplemental Table 3). In the univariable and multivariable analyses, among the factors investigated, only BSA <1.43 m² was significantly associated with VCR-related neurotoxicity (risk ratio, 1.479; 95% CI, 1.018-2.150; Table 3).

This supplementary analysis is the first to evaluate the impact of RDI_O on PFS in patients with DLBCL treated with rituximab plus 21-day cycles of CHOP, using data from a prospective trial. We used an RDI_O of 95% as the cutoff value because (1) it provided optimal PFS stratification and, (2) from a clinical perspective, an RDI_O < 95% means that the VCR dose was reduced or omitted at least once, which was used in previous reports.^12,14 As a result, among patients who received at least 6 courses of rituximab plus CHOP, VCR dose reduction was not associated with poor PFS as much as the presence of B symptoms and high-intermediate or high IPI risk according to multivariable analysis, although patients with low RDI_O tended to have poor PFS. No marked differences were observed between the groups (RDI_O, <95% and RDI_O, ≥95%) in terms of the time to progression and sites of recurrence.

Previous studies have reported inconsistent results regarding the clinical significance of reduced RDI_O. Utsu et al reported that among patients who had maintained RDI_C+H, those with RDI_O < 85% had a lower 2-year overall survival (OS) than those with RDI_O ≥85% (74.3% and 95.8%, respectively; P = .047).¹⁰ Moreover, Marshall et al revealed that patients with RDI_O < 25% had poor OS.¹¹ In contrast, Morth et al reported that the omission of VCR was not associated with disease-free survival and OS.¹² These results were obtained from retrospective observational studies. In addition, some of these reports included patients who could not complete their planned treatment because of disease progression, making it difficult to interpret the impact of a reduced RDI on the patients’ clinical outcomes.^11,12 Recently, a post hoc analysis of the RICOVER-60 trial evaluated the significance of VCR dose reduction in older adult patients with aggressive B-cell lymphoma treated with 14-day cycles of CHOP (CHOP14), with or without rituximab.¹⁴ As a result, in the entire population (n = 796), there was no difference in PFS between the groups with and without VCR dose reduction (HR, 1.0; 95% CI, 0.8-1.2). Among restricted patients treated with 6 courses of CHOP14 plus 8 doses of rituximab (n = 234), which was the most efficacious treatment regimen tested in the RICOVER-60 trial, the PFS was slightly but not significantly better in patients without VCR dose reduction than in patients with VCR dose reduction (HR, 1.1; 95% CI, 0.7-1.7), corresponding with our findings. However, in the RICOVER-60 trial, patients with aggressive non-Hodgkin lymphoma of the B-cell type (according to the Revised European-American Lymphoma Classification), not only those with DLBCL, were eligible.¹³ In addition, because CHOP14 with or without rituximab was adopted as the protocol treatment, their results could not be extrapolated to patients with DLBCL treated with the standard R-CHOP regimen.¹⁴ Nevertheless, the strengths of this study are as follows: (1) to our knowledge, this study is the first to evaluate the impact of RDI_O in 21-day cycles of R-CHOP, using data from prospective clinical trials with definite dose reduction criteria; and (2) only patients with DLBCL who had received at least 6 courses of protocol treatments were included, allowing us to mitigate the impact of disease aggressiveness and focus on the impact of RDI_O. Furthermore, in this study, the accurate periods required for treatment completion were used to calculate the RDI, the methodology of which is deemed clinically reasonable, although not considered in some previous reports.¹¹ 

Among the several cutoff values investigated, VCR dose reduction was not significantly associated with PFS. Because patients with low RDI_O tended to have low RDI_C+H (Table 1), a subgroup analysis was performed to evaluate the impact of RDI_O in patients with high or low RDI_C+H. The cutoff value of RDI_C+H (85%) was used based on previous reports and clinical relevance.^6,10,19 In the high RDI_C+H subgroup, patients with RDI_O < 95% tended to have poor PFS, similar to the overall population, whereas VCR dose reduction did not affect the PFS in the low RDI_C+H subgroup. This means that when RDI_C+H is low, the prognosis is poor regardless of the RDI_O, indicating that RDI_C+H may have a stronger prognostic impact than RDI_O, as previously reported.^4,7 In contrast, although the multivariable analysis identified the presence of B symptoms and high-intermediate or high IPI risk as risk factors, corresponding with previous reports,^21,22 not only RDI_O but also RDI_C+H were not significant risk factors affecting the PFS. We speculate that this is partly because only patients who completed at least 6 courses of treatment were included in this study, and the RDIs remained relatively high as a whole. A COO finding of NA or deficit was identified as a risk factor for poor PFS. Although the exact reason could not be determined, we considered the following as possible reasons. More patients with stage IV disease were included in the COO findings of NA or deficit (GCB, 21%; non-GCB, 22%; NA or deficit, 29.9%; respectively). In addition, patients with extranodal or aggressive disease might be less likely to provide sufficiently large tumor samples, which could contribute to a COO result of NA or deficit.

Although several factors such as VCR dose, age, sex, and race have been reported as risk factors for developing PN,⁹ the relationship between BSA and PN has not been sufficiently evaluated in adult patients with DLBCL treated with R-CHOP. Among the factors investigated in this study, only a BSA <1.43 m² was significantly associated with VCR-related neurotoxicity of grade ≥2. This may be because patients with a BSA <1.43 m² received a relatively higher VCR dose than those with a BSA ≥1.43 m², whose VCR dose was fixed at 2 mg.

In this study, females had a higher risk of developing VCR-related neurotoxicity (risk ratio, 1.184; 95% CI, 0.833-1.684) than males, although the difference was not significant (Table 3). In addition, more females belonged to the low RDI_O group (Table 1). This likely reflects the fact that females generally have a smaller BSA than males and thus receive a relatively higher dose of VCR.

This study had some limitations. This study was a post hoc supplementary analysis. In addition, because the median RDI_O in this population was very high, the applicability of our results to other cohorts, including patients with lower median RDI_O, may not be robust. Concerning VCR-related neurotoxicity, some confounding factors may have been left out, such as concomitant medications or genetic polymorphisms that can affect the activity of the CYP3A4 enzymes, which mainly metabolize VCR.^9,23,24 

In conclusion, to our knowledge, this study is the first to reveal that VCR dose reduction may not be associated with poor PFS as much as the presence of B symptoms and high-intermediate or high IPI risk, using data from a prospective trial with rituximab plus 21-day cycles of CHOP. Therefore, if patients with DLBCL can complete rituximab plus CHOP treatment, VCR dose reduction due to toxicity may not significantly impair treatment efficacy.

The authors are grateful to the members of the Japan Clinical Oncology Group Data Center and Operations Office for their support in this study. The authors thank the patients, physicians, nurses, and staff members who participated in this multicenter trial for their excellent cooperation.

This study was supported, in part, by the National Cancer Center Research and Development Fund (23-A-16, 23-A-17, 26-A-4, 29-A-3, 2020-J-3, and 2023-J-03), Grants-in-Aid for Cancer Research (20S-1, 20S-6, 23-A-16, and 23-A-17), and Health and Labour Sciences Research Grants for Clinical Cancer Research (19-20 and 22-14) from the Ministry of Health, Labour and Welfare.

Contribution: T.N., T.S., K.S., T.K., and W.M. conceptualized and designed the study; R.M. performed statistical analyses; T.N. and T.S. wrote the original draft of this manuscript; and all authors read and approved the final manuscript.

Conflict-of-interest disclosure: T.S. reports receiving honoraria from Sanofi, Janssen, Chugai, Amgen, Genmab, and AbbVie. K.S. reports receiving consulting fees from BeiGene, Daiichi Sankyo, Gilead Sciences, Janssen, Chugai, Eli Lilly, and Ohara; research funding from Kyowa Kirin; and honoraria from Eisai, Takeda, Janssen, Bristol Myers Squibb, Chugai, Kyowa Kirin, Nippon Shinyaku, Daiichi Sankyo, Meiji Seika Pharma, Ono, AbbVie, Novartis, Gilead Sciences, Genmab, Mundi Pharma, and Sanofi. T.K. reports receiving honoraria from Bristol Myers Squibb, Chugai, Nippon Shinyaku, Sanofi, AstraZeneca, AbbVie, SymBio, and Janssen. D.M. reports receiving research funding from Ono, Janssen, Eisai, Chugai, Kyowa Kirin, Merck Sharp & Dohme (MSD), Zenyaku Kogyo, Sanofi, SymBio, Takeda, AbbVie, AstraZeneca, Bristol Myers Squibb, Genmab, Novartis, Otsuka, Taiho, Pfizer, and Astellas; and honoraria from Ono, Nippon Shinyaku, Janssen, Mundi Pharma, Eisai, Chugai, Kyowa Kirin, MSD, Zenyaku Kogyo, Sanofi, SymBio, Takeda, AbbVie, AstraZeneca, Bristol Myers Squibb, Genmab, and Novartis. W.M. reports receiving research funding from Chugai, Janssen, Ono, Kyowa Kirin, Genmab, and Nippon Shinyaku; and honoraria from Mundi Pharma, Takeda, Ono, Eisai, Chugai, Bristol Myers Squibb, AstraZeneca, Nippon Shinyaku, Gilead Sciences, Nippon Kayaku, MSD, Janssen, AbbVie, and Amgen. S.I. reports receiving consulting fees from Pfizer, Janssen, Bristol Myers Squibb, AbbVie, GlaxoSmithKline (GSK), Otsuka, and Novartis; research funding from Bristol Myers Squibb, Pfizer, Amgen, Janssen, Sanofi, Takeda, Ono, Daiichi Sankyo, AbbVie, Shionogi, Chugai, and Alexion; and honoraria from Bristol Myers Squibb, Janssen, Sanofi, Pfizer, and AstraZeneca. K.O. reports receiving honoraria from Novartis, Meiji Seika Pharma, Chugai, Kyowa Kirin, Genmab, SymBio, Yakuzemi Total Learning, Janssen, and CSL Behring. K.A. was a visiting professor at Tokai University School of Medicine; and reports receiving honoraria from Janssen, Meiji Seika Pharma, Nippon Shinyaku, Bristol Myers Squibb, Eisai, and Ono. H.N. reports receiving research funding from AbbVie, AstraZeneca, BeiGene, Genmab, Janssen, Eli Lilly, Takeda, Kyowa Kirin, MSD, Mitsubishi Tanabe, Chugai, Daiichi Sankyo, Celgene, Zenyaku Kogyo, Solasia, Ono, Nippon Kayaku, and Haihe Biopharma; and honoraria from AbbVie, AstraZeneca, Genmab, Janssen, Eli Lilly, MSD, Eisai, Ono, Sumitomo Pharma, Chugai, Meiji Seika Pharma, Mundi Pharma, GSK, Bristol Myers Squibb, Nippon Kayaku, and BeiGene. The remaining authors declare no competing financial interests.

Correspondence: Tomotaka Suzuki, Department of Hematology and Oncology, Nagoya City University Graduate School of Medical Sciences, 1 Kawasumi, Mizuho-cho, Mizuho-ku, Nagoya 467-8601, Japan; email: [email protected].