Diffuse large B-cell lymphoma (DLBCL) is very common in adult lymphoma, and the incidence has gradually increased in recent years (1). Most DLBCL cases occur in patients aged 60-70 years old at diagnosis, and approximately 40% are over 70 years old (2, 3). The advanced age among DLBCL patients at presentation demands that we place our focus on the management of DLBCL in elderly and very elderly patients (4).

It is well-known that rituximab + CHOP (cyclophosphamide, doxorubicin, vincristine, and prednisolone plus rituximab) is the fundamental and standard regimen for treatment of B-cell lymphoma, especially for DLBCL (5, 6). R-CHOP regimen has also been proven to be a suitable treatment for elderly patients with DLBCL, mainly in patients with low-risk (7). Elderly patients with high-risk cannot tolerate standard doses due to factors such as age, comorbidities, or performance status (8). Different protocol, such as dose-adjusted CHOP chemotherapy with rituximab (DA-POCH-R), was presented as the alternatively recommended treatment (9). Elderly patients can achieve a long-term remission rate of 50-60% after receiving attenuated R-CHOP therapy (10, 11). The trial (NCT01148446) aiming at compared standard R-CHOP with RminiCEOP (epirubicin, cyclophosphamide, vinblastine and prednisone with rituximab), and demonstrated that DLBCL patients with low-risk disease and over 72 years had a better outcome on R-miniCEOP compared to R-CHOP (12). Additionally, several studies analyzed the impact of the dose-dense regimen and treatment cycles. For instance, the trial RICOVER60 found that no significant differences were identified between 6 and 8 cycles of the R-CHOP14 regimen administered in DLBCL patients older than 60 years (13). As for the analysis of the dose-dense regimen, one trial was conducted by the GELA group and limited to elderly population; another trial was run by the British National Investigation (BNLI) among patients of all age (14, 15). The results of these two studies involving more than one thousand participants both failed to identify significant differences of benefits in the arm of dose-dense regimen. In addition to focusing on improving the survival outcomes, the adverse events of the therapies for the elderly are also important measuring factors. Previous studies reported that adverse events, involving cardiac and hematologic toxicity, were common in older patients during the R-CHOP regimen (13, 14). Substituting doxorubicin (gemcitabine or etoposide) with less cardiotoxic anthracyclines (e.g., epirubicin and pegylated doxorubicin) or other molecules might be a safe and effective approach. For DLBCL patients ≥60 years old, R-COMP (with Myocet ^® instead of conventional doxorubicin) was similar in efficacy to R-CHOP, but unfortunately no significant reduction in early cardio-toxicity was observed (16). The replacement of etoposide for doxorubicin in standard-dose R-CHOP (named as R-CEOP) was recommended to be used in patients who have a contraindication to anthracyclines; however, it is unclear whether this compromises clinical outcomes (17). These data strongly indicated that the first-line treatment options for the elderly need to be further explored and discussed.

To better define and tailoring the best frontline therapeutic strategies for elderly DLBCL patients among current and past available treatment approaches, we carried out this systematic review and network meta-analysis comprising all available randomized clinical trials to date.

The most prevalent NHL histotype, DLBCL, peaked in incidence in the sixth decade (31). The standard regimen for DLBCL patients is Rituximab-CHOP; however, in elderly patients, there is currently no known standard first-line immuno-chemotherapy regimen. Our study, which is the first to thoroughly analyze the efficacy of various treatment regimens based on their clinical advantages and safety profiles, included 13 head-to-head phase 2 and 3 RCTs with 1839 DLBCL patients. Pairwise comparisons and ranking of various treatment options were evaluated as part of the analysis, which was carried out in a Bayesian hierarchical modeling framework using both direct and indirect evidence. When improved clinical outcomes and fewer toxicities are required, the findings of our network-analysis did in fact offer clinicians crucial guidance in order to choose from the first-line regimens that are already on the market.

Our NMA suggested that the adjustment to front-line R-CHOP21 regimen of DLBCL led to some improvement in clinical outcomes and treatment responses in elderly patients, although we only observed improved disease controls with statistically significance in R2CHOP21 compared to RCHOP21. A previous NMA by Pasvolsky et al. reached a similar conclusion, but their study was restricted to analyzing the response to R-CHOP in combination with another drug (R-CHOP + X) for DLBCL patients of all ages (32). The treatment options included in our study were more diverse compared to the NMA by Pasvolsky et al., and our study highlighted the treatment measurement in elderly patients. Undoubtedly, personalized clinical decision-making is a complex process for any individual patient. Prolonged survival time, increased response rates, reduced side effects, patient comorbidities, and patient preferences should be taken into account. In light of this, the objective of this study was to provide clinicians with evidence of efficacy from large randomized controlled trials rather than a design for identifying a regimen that should be most effective for all elderly patients. Clinicians may be better weigh the balance between potential advantages and risks when selecting the optimal treatment choice for any individual patient with the help of pairwise efficacy comparisons and ranking of the various treatment alternatives in this network meta-analysis.

Undeniably, in the current era of frequent emergence of new drugs, rituximab, as the first tumor-targeted drug and the first monoclonal antibody in history, is of epoch-making significance for the treatment of lymphoma, particularly DLBCL (33). The birth of rituximab has improved the 5-year OS of patients with aggressive B-cell lymphoma represented by DLBCL by at least 15%, and the cure rate has been significantly improved (6). This phenomenon has resulted in a huge improvement in efficacy and survival prognosis for B-cell lymphoma around the world. In view of the greater clinical benefits that rituximab brings to DLBCL patients, many researchers have begun to wonder whether RCHOP can be used in combination with other immuno-therapeutic drugs to further improve the clinical benefits of DLBCL patients. Lenalidomide, a new-generation immuno-modulator, may have potential effects on other mechanisms, such as immunomodulation and enhancement of antibody-dependent cytotoxicity in DLBCL patients (34, 35). Because of its novel activated mechanism, synergy with rituximab or chemotherapy, and moderate toxicity profile, lenalidomide has long been a key candidate for chemo-immunotherapy in the frontline treatment of DLBCL, and it was allowed to be safely combined with R-CHOP. Lugtenburg et al. investigated whether R-CHOP with early rituximab intensification (R2CHOP) could improve the clinical outcomes of patients, but it failed to have any effect on this intensification regimen (24). On the contrary, Nowakowski et al. proved that R2CHOP had an association with an obvious clinical benefit for PFS in most groups in 2021 after the results of the study by Lugtenburg et al. were published (24, 25). Our NMA revealed that R2CHOP21 created improved benefits in OS with no statistical significance compared with RCHOP21, but for PFS, this regimen was superior to RCHOP21 with statistical significance; meanwhile, other regimens did not exhibit any improvement in OS or PFS of elderly DLBCL patients. In terms of AEs assessment, only one study by Nowakowski et al. was included in our analysis; compared with the R-CHOP arm, the most common grade 3/4 adverse reactions of the R2CHOP21 arm were increased hematological AEs while all these toxicities did not bring about an increase in rates of therapy associated deaths or bleeding complications (25). This phenomenon implied that the adverse reactions in the R2-CHOP group were tolerated to some degree. In the future, it is still necessary to explore the mechanism of lenalidomide and the molecular biological characteristics of DLBCL in order to achieve more precise treatment.

Considering the main clinical outcomes, the place of R-CHOP21 in treatment options for total DLBCL patients remains important, while there exist some patients with poor efficacy on this regimen, notably in freshly treated DLBCL patients who are susceptible to refractory or early recurrence and have high-risk or other systemic problems (36). Due to the cardiotoxicity caused by doxorubicin, the implement of RCHOP should be restricted in elderly individuals. The drug’s interaction with the ferricion, which results in the generation of free radicals that contribute to lipid peroxidation and progressive myocyte destruction, causes doxorubicin-induced cardiotoxicity (37). Several schemes have been proposed to reduce cardiotoxicity caused by anthracyclines in elderly DLBCL patients, involving the implementation of slow infusions or decreased doses of doxorubicin, application of cardio-protective agents, substitution by other anti-neoplastic drugs, or selection of other anthracyclines with less cardiotoxic, such as epirubicin, liposomal formulations of doxorubicin, etc (38–40). Myocet ^®, a non-polyethylene glycol liposome doxorubicin, has proven that its cardio-toxicity is low and its anti-tumor activity is analogous to that of conventional doxorubicin based on a clinical trial about breast cancer (37). Myocet^® is also linked to decreased mucositis and myelosuppression due to its pharmacokinetics and pharmacodynamic properties (41). In our study performed for elderly DLBCL patients, the use of non-pegylated doxorubicin instead of conventional doxorubicin (R-COMP) exhibited the best clinical benefits in EFS, CR, and OR, although no obvious significant difference was determined between RCOMP and RCHOP21. Similar to a previous study, Sancho et al. pointed out that the efficacy of the R-COMP group was analogous to R-CHOP (16). Moreover, our results demonstrated that RCOMP was associated with less 3-4 grade neutropenia, neuropathy, and cardiac toxic events. In a word, R-COMP is a feasible immuno-chemotherapy schedule for elderly DLBCL patients when RCHOP is not suitable for application. Epirubicin, another epi-isomer of doxorubicin with less cardiotoxicity has been a potential candidate for anthracycline dose intensification (38). Based on our analysis, RminiCEOP was related to less anemia and thrombocytopenia, and infection of grade III/IV. RminiCEOP could serve as an alternative therapy for elderly populations with hematological system complications.

In addition, our analysis assessed the possible superiority of the dose-dense regimen (R-CHOP14) compared with the standard regimen (R-CHOP21), but failed to identify a significant benefit for R-CHOP14 in elderly patients with DLBCL. Meanwhile, our results are consistent with previous work by Delarue et al. (27). Other protocols, such as DA-EPOCH-R, GCHOP, RCPOP, and CHOP have failed to demonstrate improved efficacy in elderly DLBCL patients compared to RCHOP21. It was obvious that RCPOP was inferior to RCHOP21 in OS, and either CHOP14 or CHOP21 was significantly less effective than RCHOP21 in almost all evaluations of clinical outcomes.

Nervelessness, the following issues should be considered carefully. Firstly, the patients included in our study have used granulocyte stimulating factor according to their conditions during the treatment process, so the treatment rankogram in analysis of 3 to 4 grade neutropenia in our results should be treated with caution. Secondly, there are some differences in the age of the population, follow-up time, treatment time arrangement, and so on. These differences might lead to some heterogeneity in our results. Thirdly, in our included studies, the Eastern Cooperative Oncology Group (ECOG) scores of patients ranged from 0 to 4, but mainly focus on 0 to 2, so that our conclusions were more suitable for patients with better performance status (ECOG ranged from 0 to 2), and may be usefulness in patients who with a poor performance status (ECOG ranged from 3 to 4). Further studies should be designed to solve this issues.

In general, our meta analysis only provided a reference value. Owing to the lack of direct comparisons of certain regimens or the insufficiencies of number and quality in the original studies, we made some indirect comparisons in this study based on the transitivity assumption. Further refinement of regimen comparisons lacking direct evidence is what we need to focus on.

Our network meta-analysis compares the widest range of treatment options to date, including R-CHOP, CHOP, R-COMP, R-CPOP, and so on. In general, our findings indicated that R2CHOP21 exhibited better improvement than RCHOP21 in different clinical outcomes. However, we only observed a statistically significant improvement in PFS; this phenomenon suggested that rituximab-based standard therapy upon CHOP has irreplaceable advantages as first-line therapy. In terms of decreasing toxicities from treatments, RCOMP, RminiCEOP and RCHOP21 exhibited lower toxicity in different AEs. Therefore, these findings could provide valuable recommendations for clinicians in making better decisions from multiple promising treatment options for elderly DLBCL patients by fully considering their clinical benefits and toxicity profiles.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding authors.

The literature search and study selection were conducted independently by two authors (YW and XR), with discrepancies reviewed and resolved by consensus (LH and ZZ). The data extraction was retrieved and summarized independently by two authors (YW and XR) following Cochrane Collaboration guidelines. The bias risk of all the enrolled RCTs was assessed by two authors (XL and LP) using the Cochrane Risk of Bias Tool. The data analysis were performed by three authors (YW, XR and KH). All authors contributed to the article and approved the submitted version.