The addition of consolidative RT for advanced stage DLBCL patients who demonstrate complete response after chemotherapy is associated with a statistically significant improvement in overall survival and progression free survival when pooling data from two academic medical centers. This analysis provides additional basis for the utilization of consolidative RT in these patients and contributes to a growing body of prospective and retrospective work. The combined data corroborate earlier retrospective studies, which collectively document improvements in overall survival, event free survival, progression-free survival, and in-field control with consolidative RT. Forthcoming randomized data are expected to further elucidate the role of RT for this group. Historically, stage III-IV patients have often been aggregated with stage I-II patients, potentially diluting any measurable benefit associated with consolidative RT. Thus, the true role for consolidative RT in this population is evolving.
Phan et al. previously reported statistically significant improvements in both OS and PFS at 5 years for a large cohort of DLBCL patients, consistent with the results presented here. The majority of patients in the Phan cohort presented with stage III-IV disease (279 from a total of 469), however, RT was delivered to only 39 (14%) advanced stage patients in their analysis as compared with 53 (28%) of patients in our cohort. OS and PFS were both significantly improved for advanced stage patients. They report 5-year OS and PFS after RT of 89% and 76%, respectively, as compared with our figures of 87% and 86%, respectively, at the same time point. Dorth et al. report excellent in-field control and event free survival for stage III-IV patients at 5 years with consolidative RT (92% and 85%, respectively). They reported a 5-year OS survival of 85% for RT patients and demonstrated a trend towards improvement but fell short of achieving statistical significance (78% for the no-RT subset, p = 0.15). The addition of advanced stage Emory patients bolstered the case for an OS advantage, though failed to meet criteria for significance on multivariable analysis likely due to confounding factors related to patient selection, tumor size, and sample size. Nonetheless, our combined Kaplan-Meier analysis demonstrated a clinically meaningful 14.5% improvement at 5 years with the pooled results. While the combined analysis failed to achieve statistical significance with respect to local control, it is reassuring that the overall number of local failures is low (15 events in total) and numerically improved with the addition of consolidative RT (3 vs. 12).
The definition of bulky disease from the standpoint of consolidation is a topic of ongoing debate, but the existing body of literature points to tumor size affecting treatment outcomes. Post-hoc analysis of the MInT study found cut-off sizes of 6 cm and 10 cm corresponded to statistically significant differences in OS and EFS, respectively, among patient treated with R-CHOP and consolidative RT(14, 15). Additionally, prior work by our group has demonstrated a 5 cm cut-point to be meaningful with respect to local control(11). This analysis builds upon these prior findings by demonstrating a PFS benefit that not only persists with a 5 cm cut-off point, but also applies to smaller tumors, a question that has not been addressed by prior work. The results presented here support extending consolidative RT to a larger subset of advanced stage DLBCL patients.
The advent of rituximab substantially improved survival outcomes for lymphoma patients, firmly establishing R-CHOP as the standard of care for DLBCL(2, 15-17). This has led some to question the utility of consolidative RT, particularly as it exposes patients to a separate array of potential toxicities. Thus, the therapeutic effect of consolidative RT, specifically for patients receiving R-CHOP, was of particular interest in our analysis as well as in the other studies described here. Fully 84% percent of MDACC patients and 88% of patients in the Emory/Duke combined cohort received a complete or partial course of rituximab-based systemic therapy. Results from these three centers demonstrate a substantial and clinically meaningful benefit for consolidative RT in patients treated with rituximab, corroborating results of a recent NCCN outcomes database analysis for DLBCL patients across all stages(18).
The results presented here complement prior retrospective studies and support further investigations of consolidative RT in stage III-IV DLBCL. The RICOVER-60 trial included DLBCL patients across all stages and primarily evaluated the addition of rituximab to six versus eight cycles of CHOP-14, with consolidative RT delivered to sites of initially bulky (³7.5 cm) or extralymphatic disease. The impact of RT was later examined by means of a protocol amendment that created a no-RT arm that was not included in the initial randomization. The intent-to-treat analysis showed improved event-free survival and a trend towards improved PFS and OS (the latter two were significantly improved in the per-protocol analysis). The ongoing UNFOLDER trial, published in abstract form, randomized a similar group of patients to two R-CHOP regimens with a second randomization to consolidative RT to bulky and extra-nodal disease(19). Of note, the no-RT arm was closed at interim analysis. Additionally, results from the OPTIMAL>60 trial were published in abstract form and suggest that for elderly patients RT can be spared in bulky disease that is PET-negative after chemotherapy, though patients with persistent disease received RT per protocol(20).
Aside from radiation therapy, other consolidation strategies have been explored for advanced stage DLCL. The randomized, phase II GELA/LYSA trial sought to evaluate the efficacy of two rituximab-based induction regimens along with a PET-driven consolidation regimen(21). According to this protocol, patients with negative PET-CT after completion of 4 cycles of induction therapy (PET4) were treated based upon an interim PET-CT completed after 2 induction cycles (PET2); patients with negative PET2 received consolidative immunochemotherapy while those with positive PET2 received chemotherapy and autologous stem cell transplant (ASCT). For patients treated with induction R-CHOP who exhibited cCR, 4-year results show a PFS of 82.2% and OS of 87.2%. These figures compare favorably with results presented here for consolidative RT (85.9% and 87.4%, respectively at 5 years), suggesting that radiation therapy may be a reasonable alternative to additional immunochemotherapy or ASCT.
One shortcoming of our work is that RT dosing varied significantly across patients in our study cohort with a median dose of 30 Gy, which is lower than doses utilized in many previous large randomized trials of consolidative IFRT in stage I-II non-Hodgkin lymphoma; SWOG 8736 used 40-55 Gy, while LHN 93-1 and 93-4 used 40 Gy, though ECOG 1484 did use 30 Gy(22-25). These trials were conducted in the pre-rituximab era and a subsequent meta-analysis incorporating selected data from these four trials revealed a PFS improvement with IFRT but failed to demonstrate improvement in OS. This meta-analysis was limited by heterogeneity in the pooled dataset, however(26). For patients with advanced disease who respond well to systemic therapy it has been suggested that lower RT doses may, in fact, be preferred since these patients are likely to have received longer courses of chemotherapy and RT will be delivered to larger volumes(27). As such, 30 Gy is regarded as a standard dose at this time.
Other shortcomings include the non-standardization of imaging in response assessment and the age of our dataset. It is now established that PET-CT is the gold standard in the staging of aggressive NHL and has higher specificity, accuracy and positive predictive value in detecting residual disease after systemic therapy than CT alone, though this practice was not established when some of our earlier patients were diagnosed(28, 29). While PET-CT was utilized in many cases, this was not a strict requirement for patients in this combined dataset. As such, were not able to derive meaningful associations related to specific imaging findings, such as Deauville (5 point) score or SUV values as demonstrated previously(11). Finally, concerning our timeline, advances in radiation oncology, including image guidance and volumetric dose planning, have improved the quality of care since the time that our earliest patients received treatment.