Radiation Therapy After R-CHOP for Diffuse Large B-Cell Lymphoma: The Gain Remains

      JCO September 20, 2010 vol. 28 no. 27 4105-4107

It is hard to imagine that, less than two decades ago, the acronyms IPI (International Prognostic Index), FDG-PET ([¹⁸F]fluorodeoxyglucose positron emission tomography), WHO classification, and even R-CHOP (rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone) were not part of the lymphoma language. But in contrast to these advances, we are still employing cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP) after almost four decades (now enhanced with the anti-CD20 antibody rituximab) as our basic chemotherapy for diffuse large B-cell lymphoma (DLBCL), and we keep debating the need for consolidative involved-field radiotherapy (IFRT) for patients with stage I and II disease.Clearly, the issue of treatment consolidation after R-CHOP with IFRT, or alternatively with more chemotherapy, has not been resolved. In an attempt to satisfy all opinions, the National Comprehensive Cancer Network (NCCN) guidelines recommend three cycles of R-CHOP + IFRT for early-stage, nonbulky disease but also allow the administration of six to eight cycles of R-CHOP with or without IFRT.The latter is also the NCCN recommendation for bulky disease. This variety of options in the NCCN guidelines may make everybody happy, but it could be confusing to the nonexpert. In reality, many oncologists simply extend the chemotherapy course and omit radiotherapy (RT).

The shorter chemotherapy followed by RT approach for early-stage disease is solidly based on the results of the benchmark, randomized trial published in 1998 by the Southwest Oncology Group (SWOG) that clearly showed that three cycles of CHOP + IFRT was a safer and significantly more effective treatment than eight cycles of CHOP alone. The SWOG study did not secure the use of combined modality in localized disease for long. For some chemotherapists, the better safety profile of short CHOP + IFRT and its survival advantage, which lasts for almost a decade in patients with a median age of 60 years, still did not justify keeping RT in the program. Moreover, given that all randomized trials in early-stage DLBCL were done in the pre-R, pre-IPI, pre–positron emission tomography (PET) era, it was suggested that the magic of R or a PET conversion will make RT irrelevant. This hypothesis has not been tested yet. Unfortunately, randomized studies in early-stage DLBCL that will incorporate R, IPI, and PET and will also examine the likely advantage of modern, better targeted, safer, and lower-dosage, consolidative RT are not coming to a theater near you anytime soon.

The retrospective work published in this issue from the M. D. Anderson Cancer Center (MDACC) on the benefit of consolidative RT in patients with DLBCL treated with R-CHOP is important. Still, it is not the ideal study to fully settle the role of RT in early-stage disease simply because it is not a prospective, randomized trial. Yet it reflects a common reality: different experts in a large cancer center exercising different approaches to the use of RT in early-stage DLBCL.

At MDACC, over the last decade, 103 (54%) of 190 stage I and II patients received IFRT. In stages III and IV, where IFRT is not commonly used, 39 (14%) of 279 MDACC patients still received RT; 59% of those had bulky disease. In each of the groups, patients who received RT had a significantly better outcome. Several features of the study helped to avoid some of the concerns that tarnish the previous randomized studies that tested the role of RT. First, all patients in the MDACC study were treated with R-CHOP and not with CHOP alone. Second, they all received six to eight cycles of chemotherapy, averting the concern that three cycles of CHOP are not sufficient as systemic treatment for patients with stage-modified IPI more than 1. The study also incorporated IPI, and often FDG-PET, data into the analysis.

Quite impressively, the five-year overall survival (OS) and progression-free survival (PFS) for stage I and II disease treated with RT were 92% and 82%, respectively; whereas without RT, OS and PFS were 73%  and 68%, respectively.

For stage III and IV disease, the minority of patients who received RT had five-year OS and PFS of 89% and 76% compared with 66% and 55%, respectively, for those not receiving RT.

It remains unclear what criteria were used to select those patients with advanced-stage disease to receive RT and how the sites for RT were determined. The data are strong in the early-stage category in which more than half of patients received RT, and there is no apparent selection bias in favor of RT patients. Although this is not a randomized study, the data are solid and are well in line with the randomized studies in the United States. Now, the burden of proof for not using RT in early-stage disease is on those who still do not believe that RT should be used in DLBCL after complete response (CR) to chemotherapy.

Perhaps before designing a new prospective study to determine the role of RT in early-stage disease, we should examine the strengths and flaws of the available studies. They have not completely resolved the controversy concerning using RT but may provide us with some clues and principles. It is not disputed that DLBCL is almost always a systemic disease, even when this is not apparent clinically at presentation. In almost one third of patients, DLBCL presents in only one or two sites, and in some, it persists locally after chemotherapy or (without RT) tends to relapse in the same original site. Furthermore, the responsiveness of DLBCL to RT was well established before the advent of effective chemotherapy, when RT was used alone in localized aggressive lymphomas. For example, a study of early-stage non-Hodgkin's lymphoma from Princess Margaret Hospital (Toronto, Canada) showed that, in a cohort of clinically staged patients younger than age 60 years with no bulky disease, the 10-year, relapse-free survival with RT alone was 77%.

The notion that longer chemotherapy may allow the omission of RT was tested in the SWOG 8736 trial. The results were clearly in favor of keeping RT on board. Furthermore, adding RT appeared to allow a shorter regimen of only three cycles of CHOP, avoid dosage-dependent chemotherapy toxicity, and still control DLBCL in almost 80% of the patients.

The SWOG study also showed that substituting RT with five cycles of CHOP (eight in total) resulted in an inferior disease control and lower OS and was clearly more cardiotoxic and myelotoxic. The issue of toxicity with more doxorubicin-based regimens should not be ignored, particularly as more seniors present with DLBCL. The enthusiasm for short-course CHOP + IFRT decreased when a 2001 update was reported in an abstract (but still is not published). It showed that the PFS and OS curves of the two groups begin to overlap at 7 years and at 9 years after treatment, respectively. These curves overlapped because of the excess number of late lymphoma relapses in the group that received three cycles of CHOP + IFRT. Further analysis of the three cycles of CHOP + IFRT arm using a stage-modified IPI (scoring stage II as an adverse factor) suggested that higher stage-modified IPI patients on this arm could potentially benefit from adding one or more courses of CHOP. Yet the benefit of IFRT prevails, given that the irradiated, originally involved site remained fully controlled. As confirmed by a series from Vancouver, patients with no risk factors have done extremely well on short CHOP + IFRT (5-year OS of 95%). Surprisingly, some physicians rushed to minimize the advantages of the less-toxic, combined-modality arm because of these late, and likely avoidable, events. The principles established by this landmark study should not be dismissed.

The cardiac toxicity risk with longer CHOP was demonstrated not only in the SWOG study. An increasing body of data on cardiac morbidity and mortality in patients with doxorubicin-treated non-Hodgkin's lymphoma is troublesome.In the Surveillance, Epidemiology, and End Results database of 15,454 patients with stages I and II DLBCL, 61% did not receive RT.Those patients probably received higher doses of doxorubicin-based chemotherapy as an alternative to shorter chemotherapy with RT. Interestingly, those not treated with RT had a significantly higher rate of cardiac mortality (P < .001) than those who received RT. A twin study showed that patients with early-stage DLBCL in the Surveillance, Epidemiology, and End-Results database who received RT also had a significantly better disease-specific survival and OS compared with those who did not receive RT (P < .001 with multivariate analysis). This advantage lasted throughout the 15 years of follow-up.

Another randomized trial that showed a significant advantage for adding RT (only 30 Gy) to patients even after they have obtained a CR with eight cycles of CHOP is Eastern Cooperative Oncology Group (ECOG) study 1484. Patients with only a partial response to eight cycles of CHOP were not randomly assigned, but many achieved a lasting CR with the addition of RT. The complete responders after eight cycles of CHOP were randomly assigned to either observation (93 patients) or IFRT (only 79 patients). Six-year disease-free survival (DFS) was 73% for IFRT and 56% for observation (P = .05), and failure-free survival was 75% and 56%, respectively (P = .06). The additional value of IFRT (only 30 Gy) after eight courses of CHOP is quite impressive considering that the study did not have an optimal statistical power due to a high drop-out rate.When the study was dusted off and eventually published 12 years after its closure, it lacked the details for the causes of death. Obviously, attrition of the survival curves in both arms in this elderly population, which received eight cycles of CHOP, could be attributed to multiple nonlymphoma (and non-RT) –related causes. Indeed, although 15-year OS in the as-treated analysis was 16% higher in the RT arm, it did not, in this underpowered study, reach statistical significance (P = .24). The study clearly showed that RT significantly improves disease control rates, even after taking CHOP to the limits of safety.

The approach of substituting RT for highly intensive chemotherapy was taken by the Groupe d'Etude des Lymphomes de l'Adulte (GELA) LNH 93-1 study.This study compared three cycles of CHOP followed by RT versus an aggressive, dosage-intense regimen of doxorubicin, cyclophosphamide, vindesine, bleomycin, and prednisone (ACVBP) followed by sequential consolidation with methotrexate, etoposide, ifosfamide, and cytarabine, in patients with low-risk, early-stage disease at a median age of only 47 years. Five-year event-free survival for the intensive nine-drug regimen arm was better than GELA's three cycles of CHOP + IFRT arm (82% v 74%); OS was only slightly better (90% v 81%). However, when the GELA study was analyzed by the stage-modified IPI and compared with the SWOG study, 5-year OS for patients with no risk factors who were treated with ACVBP was slightly worse, and GELA's three cycles of CHOP + IFRT was significantly inferior to SWOG's three cycles of CHOP + IFRT (91% and 83% for GELA v 95% for SWOG, respectively).ACVBP has never been adopted in North America due to its significant acute toxicity (20% hospitalization rate with each cycle) and higher risk of second tumors, particularly myelodysplastic syndrome/acute myelogenous leukemia and lung cancer.

Another GELA trial (LNH 93-4) compared four cycles of CHOP to four cycles of CHOP + IFRT in patients older than 60 years. Five-year event-free survival were 61% and 64% (P = .6), respectively and OS was 72% and 68%, respectively. Patients in both groups did significantly worse than patients who received three cycles of CHOP + IFRT in the SWOG study (5-year OS of 82%); for patients with two risk factors, the difference in favor of SWOG's combined-modality arm was even greater (58% v 77%). Administration of RT was markedly delayed (median of 7 weeks after chemotherapy) in the GELA study, and 12% of the patients on the RT arm did not receive RT. Notably, in the MDACC study, none of the patients experienced treatment failure in the RT field,whereas in the GELA IFRT arms, relapses occurred in irradiated sites alone at rates of 23% and 21% (LNH 93-1 and 93-4, respectively). The lack of IFRT contribution in this study and the inferior results compared with other studies raise the question of whether delays, quality, and failure to receive IFRT affected the results.

Indeed, the above studies were all done in the era before anti-CD20 treatment. Although R had a more modest effect on patients in early stages of disease compared with advanced stages, it is legitimate to suggest that the optimal regimen for early-stage disease needs to be re-examined with regard to both the extent of chemotherapy and the use of RT. Some have even speculated that R could substitute for RT, but the logic and evidence for that needs to be demonstrated. The recently published SWOG 0014 trial integrated R into three cycles of CHOP + IFRT, the better arm of the original SWOG 8736 trial; the results are encouraging. A carefully matched analysis with the SWOG historical trial showed that, at a median of 5.3 years, adding R to three cycles of CHOP + IFRT resulted in additional gains in 4-year PFS (from 78% to 88%) and OS (from 88% to 92%). Longer follow-up will indicate if these gains will increase or decrease with time.

The saga of RT in early-stage DLBCL has not ended. The MDACC data are likely to strengthen the position of those among us who trust the original analysis and trends in favor of RT in the SWOG and ECOG experiences; others may berate it as just another retrospective study. We all may still hope for the ultimate, modern randomized study. Sadly, that is unlikely to come soon.

The information on the value of consolidation RT in advanced-stage disease in the MDACC study is also interesting. Although little solid information to support or refute the selective incorporation of RT into advanced-stage options is available, the use of RT in advanced-stage DLBCL remained almost a taboo for several decades. We should design a smart program to test RT even in advanced-stage disease using the new tools that the fields of lymphoma and radiation oncology have acquired in the last two decades.