Abstract
Introduction: The discovery of 3 distinct molecular subgroups of diffuse large B-cell lymphoma (DLBCL) according to cell-of-origin (COO): germinal centre B-cell (GCB), activated B-cell (ABC) and type III/unclassifiable by gene expression profiling (GEP) (Alizadeh et al, Nature 2000; Rosenwald et al, NEJM 2002) was a key advance in understanding the disease biology. The ABC subtype is associated with inferior survival which has persisted in the rituximab era (Lenz et al, NEJM 2008). Determination of COO by GEP has not been incorporated into routine practice however due to a lack of accessibility and requirement for fresh frozen tissue. Barrans et al (BJH 2012) recently demonstrated that COO could be accurately classified by the Illumina DASL® platform, using RNA extracted from routinely processed FFPE tissue from a population-based cohort of 172 R-CHOP-treated patients, and predicted clinical outcome. The aim of this analysis was to determine GEP-based DLBCL COO using the Illumina DASL® platform, to correlate results with outcome, and to validate this methodology using FFPE tissue samples from patients enrolled in the prospective phase III R-CHOP 14 v 21 trial.
Methods: The UK NCRI R-CHOP-14 v 21 trial assessed R-CHOP given 2-weekly versus 3-weekly in 1080 previously untreated DLBCL patients aged ≥18 years and enrolled from 2005-2008. We previously reported that R-CHOP-14 was not superior to R-CHOP-21 for overall survival (OS), progression-free survival (PFS), response rate or safety. COO as determined by the Hans classifier (n=560) was not prognostic for OS (Cunningham et al, Lancet 2013).
All patients with sufficient FFPE material remaining were included in this analysis. RNA was extracted and GEP was performed using the Illumina DASL® platform. Cases were classified as ABC, GCB or type III according to the DAC classifier (Care et al, Plos One 2013). Response was assessed using IWG 1999 criteria. PFS and OS were calculated from date of randomisation and analysed using Kaplan-Meier and Cox regression methods.
Results: 369 patients had sufficient FFPE material remaining for analysis. COO classification was as follows: ABC 15.2% (n=56), GCB 46.3% (n=171), type III 38.5% (n=142). Baseline characteristics are shown in Table 1. Patients with GCB subtype had a significantly higher incidence of BCL-2 (p<0.001)and double-hit rearrangement (p=0.01) vs. those classified as ABC or type III. The median follow-up was 77.8 months. COO by GEP identified 3 distinct prognostic groups for OS (p=0.04) (Figure 1). 5-year OS for patients classified as ABC, GCB and type III was 60.5% (95% CI: 47.6-73.4), 81.4% (95% CI: 75.5-87.3) and 69.9% (95% CI: 62.1-77.7) respectively. Strong evidence of differences between these groups remained after adjusting for age as a continuous variable, International Prognostic Index (IPI) risk factors (WHO performance status, stage, presence of >1 extranodal site of disease, elevated LDH), sex, bulky disease, B symptoms and trial arm; patients classified as GCB or type III had superior OS versus ABC subtype (HR 0.53, 95% CI: 0.31-0.89; p=0.02) and (HR 0.56, 95% CI: 0.33-0.96; p=0.03) respectively. GCB subtype was also independently associated with superior PFS versus ABC subtype (HR=0.59, 95% CI: 0.37-0.96; p=0.03). The difference in PFS between type III and ABC subtypes did not reach statistical significance, but followed a similar trend (HR=0.65, 95% CI: 0.40-1.07; p=0.09). Results of univariate and multivariate analyses of individual factors and OS are shown in Table 2.
Conclusion: Our results demonstrate that the ABC subtype of DLBCL as determined by GEP is independently associated with inferior PFS and OS. Our findings confirm those of Barrans et al (BJH 2012) and serve as a validation cohort for this methodology in the setting of a prospective trial where patients were exclusively R-CHOP-treated. Of note our patient cohort included a high proportion of type III/unclassifiable patients (38.5%) which are being further investigated currently and updated results will be presented at the meeting. Our analysis confirms that GEP-based COO is a significant prognostic biomarker for DLBCL in the rituximab era which can be accurately determined using routinely processed FFPE tissue samples.
Overall survival by cell-of-origin determined by gene expression profiling group
Overall survival by cell-of-origin determined by gene expression profiling group
Cunningham:Medimmune: Research Funding; Merrimack: Research Funding; Celgene: Research Funding; Bayer: Research Funding; Astra-Zeneca: Research Funding; Amgen: Research Funding. Hawkes:Merck Serono: Research Funding; Takeda: Other: travel expenses; BMS: Other: travel expenses, Research Funding. Pocock:Gilead Sciences: Other: Sponsorship to attend the EHA 2016 Meeting; Janssen: Speakers Bureau; Takeda: Honoraria. Ardeshna:Roche: Membership on an entity's Board of Directors or advisory committees, Other: Conference Expenses, Research Funding. Radford:Astra Zeneca: Equity Ownership; GlaxoSmithKline: Equity Ownership; Novartis: Honoraria; Seattle Genetics: Honoraria; Takeda: Honoraria, Research Funding.
Author notes
Asterisk with author names denotes non-ASH members.