A multi-institutional analysis of diffuse large B-cell lymphoma (DLBCL) treated with consolidative radiotherapy and the impact of cell-of-origin on outcomes

Diffuse large B-cell lymphoma (DLBCL) is the most common non-Hodgkin’s lymphoma in adults, accounting for about 30–60% of all cases.1 It has an aggressive natural history, with a prognosis of less than a year without treatment.2 Radiotherapy (RT), historically, has been an integral part of DLBCL treatment.3 However, with the implementation of new systemic therapy agents, the use of RT has declined.4,5 The discovery of the chimeric monoclonal antibody rituximab against CD20 receptors has greatly improved the control and cure of DLBCL.6 Data from the MINT studies suggest that rituximab reduced the risk posed by bulky disease, but did not eliminate it.7 In line with that, many studies have shown the improved outcome with the addition of RT, especially in the context of bulky disease.8,9

However, clinicians have noticed that the behaviour of DLBCL can be varied, and attempts have been made to better classify DLBCL.10 Based on gene expression profiling studies, DLBCL can be divided into 2 distinct subtypes: Germinal Centre B cell (GCB) and non-Germinal Centre B- cell subtype (non-GCB).11 Immunohistochemistry based algorithms have been shown to have good concordance with gene expression profiling for cell-of-origin classification.12 Studies based on Western populations have suggested that GCB-subtypes are associated with improved outcomes.13 However, these findings could not be replicated in the Asian population.14 It is important to note that these patients were treated primarily with chemotherapy, and the impact of cell-of-origin for patients undergoing consolidative RT is unclear.

The aim of this study is to report the outcome of patients with DLBCL treated with

rituximab, cyclophosphamide, hydroxydaunorubicin, oncovin, prednisone (R-CHOP) or R-CHOP like chemotherapy and consolidative RT. In addition, we classified patients according to cell-of-origin (where information available) and determined the impact on the outcomes.

Seventy-four patients were included in the study and the demographic characteristics are shown in Table 1. The median age was 61 years ranging from 14–88 years. Fifty-eight percent were males and the same proportion had early-stage disease (Stage I–II). Bulky disease was seen in 43% of patients. About a third of the included patients had IPI scores of 0–1. Sixty-one percent had at least 6 cycles of chemotherapy. About two-thirds of the patients were treated to a dose of 36 Gy or less. Information on cell-of-origin was available in 60% of patients and was equally distributed between GCB and non-GCB origin. Median follow up of the cohort was 3 (0.7–16) years.

Survival and patterns of relapse

The 5-year OS was 92% (median survival not reached) (Figure 2) and 5-year DFS was 73% (median survival not reached) (Figure 3).

Figure 2

Figure 3

Fifty-three patients (72%) were in complete remission post-RT. Patients who achieved complete remission had a significantly better DFS (HR 11.05, 95% CI 4.11–29.69, P < 0.01). (Figure 4). The presence of initial bulky disease was associated with an inferior DFS (HR 3.16, 95% CI 1.02–9.78, P = 0.04) (Figure 5).

Figure 4

Figure 5

Uni-variable and multi-variable analysis for DFS and OS are presented in Table 2. Response to RT (P < 0.001) and tumour bulk (P < 0.02) were significant predictors of DFS. Only response to RT (P = 0.011) was a significant predictors of OS. There was no association between cell-of-origin and DFS or OS (P = 0.16, P = 0.61 respectively).

Table 2

UNI- and multivariable analysis for disease-free survival (DFS) and overall survival (OS)

Variable	DFS						OS
	Univariable analysis			Multivariable analysis			Univariable analysis
	HR	95% CI	P	HR	95% CI	P	HR	95% CI	P
Age (continuous)	1.03	1–1.06	0.081	1.02	0.98–1.06	0.37	1	0.96–1.05	0.86
Gender (ref: male)	1.78	0.70–4.51	0.23				1.38	0.34–5.51	0.65
ECOG (continuous)	1.64	0.92–2.95	0.096	0.42	0.18–1.00	0.05	1.05	0.35–3.15	0.93
Stage 1–2 (ref) vs. 3–4	1.19	0.46–3.01	0.71				1.84	0.44–.64	0.4
B symptoms (ref: yes)	0.51	0.8–1.48	0.22				0.25	0.5–1.23	0.088
Elevated LDH yes (ref) vs. no	1.06	0.34–3.34	0.92				2.38	0.28–20.18	0.43
IPI (ref) vs. 2–5	1.48	0.55–3.99	0.43				2.32	0.46–11.69	0.31
Bulk < 7.5 cm(ref) vs. ≥ 7.5 cm	3.16	1.02–9.78	0.045	6.10	1.3427.87 –	< 0.02	3.19	0.46–22.15	0.24
Baseline PET SUV ≤ 20 (ref) vs. > 20	0.25	0.45–1.35	0.11				-	-	-
Chemotherapy	2.24	0.78–6.34	0.13				1.73	0.39–7.69	0.47
< 6 cycles (ref) vs. ≥ 6 cycles
Dose < 36 Gy (ref) vs. ≥ 36 Gy	0.89	0.34–2.30	0.81				2.21	0.52–9.37	0.28
RT response CR(ref) vs. non CR	11.05	4.11–29.70	<0.001	5.64	2.7811.45 –	<0.001	6.26	1.53–25.7	0.011
Cell of origin (ref GC)
NGC	3.72	0.74–18.54	0.11				2.57	0.23–28.94	0.44
unknown	3.60	0.79–16.50	0.10				3.77	0.44–32.49	0.23

ECOG = Eastern Cooperative Oncology Group performance status; GC = germinal centre B cell (GCB); IPI = international prognostic index; NGC = non-GC; RT = radiotherapy; SUV = standardized uptake value

DLBCL is an aggressive condition, which can behave variably.10 The decision on whether to use consolidative RT remains controversial, especially in advanced stages where complete response has been achieved.19 In this study, we report the outcomes of our patients treated with consolidative RT.

We report encouraging survival and disease control rates in this cohort–5-year OS of 92% and 5-year DFS of 73%. Our results are congruent with other contemporary series9,20, 21, 22, 23, 24, 25, which are summarised in Table 3. While assessing for predictors for improved DFS, we found that patients who were in complete response (post-RT) had improved DFS on multivariable analysis (HR 5.64, 95% CI 2.78–11.45, P < 0.001). This was not an unexpected finding as it is likely suggestive of better tumour biology. In addition, patients who had bulky disease (> 7.5cm) had an increased risk of relapse (HR 6.1 95% CI 1.34–27.87, P < 0.02). Bulky disease (at initial presentation) is considered to be an indication for consolidative RT, although the definition of bulk has varied across studies, from 5 cm(8, 26) to 10 cm.20 In our institution, we use 7.5 cm as a definition of bulk, in line with the MINT studies.7 As such, it is likely that consolidative RT reduces the risk of disease recurrence, but does not prevent it. As for OS, only complete response (post-RT) was predictive of improved OS.

Secondly, we were able to classify about two-thirds of our patients by cell-of-origin (into GCB vs. non-GCB). IHC markers of CD10, BCL-6 and MUM-1 were routinely performed from 2013. As such, 27% were classified to have GCB, 29% non-GCB and remainder were unclassified. Based on univariate analysis, non-GCB was not deemed to be a significant predictor of worse DFS or OS. (HR 3.72 95% CI 0.74–18.54, P = 0.11; HR 2.57 95% CI 0.23–28.94, P = 0.44), compared to GC. This should only be considered as hypothesis-generating as the number of events from our cohort is relatively small. In addition, we would like to qualify that there were no statistically significant differences between the GCB and non-GCB groups, in terms of initial bulky disease (P = 0.09) or response to RT (P = 0.27)

Thirdly, with regards to patterns of recurrence, some previous studies have analysed the patterns of failure in patients with DLBCL. Shi et al analysed patients with DLBCL, who achieved complete remission after R-CHOP.26 Almost half of the patients with advanced-stage DLBCL failed at the initial presenting sites even after achieving complete remission with R-CHOP. In addition, around half of such local failures occurred at initial bulky or bony sites. The local failure rate was 44% in the R-CHOP alone group compared to 7% with R-CHOP plus consolidative RT. Our series echoes the findings of Shi et al., where only 7% of patients failed inside the treatment field with a local control rate of 93%. However, it is important to note that these were not isolated local failures. As such, RT continues to provide excellent local control for bulky and/or residual disease.

The UNFOLDER study examining the role of consolidative RT (for bulky and/or extranodal sites) in patients who had achieved complete response to chemotherapy underwent early termination of the no-RT arm, due to increased number of recurrences.27 The full results are eagerly awaited.

Our study has several strengths. Our data is well-curated, as all the patients were treated at two institutions which rely on electronic medical records, electronic PACS (picture archiving and communication system), and where the management of majority of the cases are discussed at the weekly lymphoma tumour board meetings. Moreover, the patients were regularly followed up by haematologists and radiation oncologists. Secondly, we are the first to examine the clinical relevance of cell-of-origin on RT outcomes. However, we acknowledge the limitations of our study. Despite close follow-up, there are patients with missing data, as with any retrospective study. In addition, we captured all patients who received consolidative RT - and this included patients with both limited and advanced disease, where the outcomes can be different. Our data spans over 15 years, where staging methodology, chemotherapy choices and response assessment modalities have evolved. Moreover, there is evidence to show that the survival of DLBCL patients has improved over the years.28 Lastly, the overall number of events (recurrence or death) in our analysis was small, so it is possible that we had insufficient power to detect prognostic factors (type II error). It would also have been useful to have a control group of patients who were treated with chemotherapy alone.

Although DLBCL is considered to be an aggressive form of non-Hodgkin’s Lymphoma, it has an excellent outcome with modern treatment. RT contributes significantly towards local control and survival in patients with bulky disease or residual disease following first-line chemotherapy. The cell-of-origin, by Hans algorithm, may not be a relevant prognostic factor in patients undergoing consolidative RT. A well-designed randomised controlled trial, comparing patients treated with chemotherapy alone, would be useful to determine the additional benefit of RT.