Analysis of patient samples from the phase III LNH03-6B trial: expression of BCL2, but not co-expression of BCL2 and MYC, predicts survival in elderly Diffuse Large B-Cell Lymphoma patients

Earlier this year, on 24^th January 2017, T. Petrella from CHU Dijon, France, and colleagues published findings in the Annals of Oncology which aimed to investigate the prognostic value of Cell of Origin (COO) by IHC after evaluating concordance with Gene Expression Profiling (GEP) and MYC and BCL2 co-expression.¹

In total, 602 patients with a median age of 70 years (range, 59–80 years) were enrolled in the Lymphoma Study Association (LYSA) phase III, randomized, LNH03-6B trial (NCT00144755), which aimed to compare the efficacy of R-CHOP given every 14 or 21 days; no difference was found.²

In the current analysis, paraffin-embedded samples from 285/602 patients enrolled in the LNH03-6B trial were evaluated for COO using the Hans algorithm. Moreover, using frozen tissue samples from a subset of 62 patients, COO defined by the transcriptome according to the Wright algorithm was correlated with that defined by the Hans algorithm.¹

Key Highlights:

• 3-yr PFS = 58.9%; 3-yr OS = 68.3%
• Higher IPI (3­–5 vs 0–2) was significantly associated with low PFS and OS
• By Hans algorithm (285 pts): samples classified as GCB = 118 (41%); non-GCB = 167 (59%)
• BCL2 (70% threshold) expressed in 61% tumors; MYC (40% threshold) expressed in 34% tumors
• MYC and BCL2 co-expression reported in 25% of cases; double-expressor phenotype marginally associated with non-GCB phenotype (P = 0.09)
• By transcriptome analysis on frozen tissue (62 pts): classified as GCB = 30 pts (39%); ABC = 24 pts (46%); unclassified = 8 pts (15%)
• Overall concordance of Hans algorithm score to transcriptome method = 91% (49/54)
• Patients also evaluable by Reverse Transcription-Multiplex Ligation-Dependent Probe Amplification (RT-MLPA) = 56/62
• Overall concordance of Hans algorithm score to RT-MLPA method = 94%
• In univariate analysis:
  • COO by Hans algorithm and BCL2 (50% or 70% threshold) significantly impacted PFS (P = 0.002, 0.007, and 0.002, respectively) and OS (P = 0.003, 0.03, and 0.01, respectively)
  • MYC expression and MYC/BCL2 co-expression (40%/70% or 40%/50%) did not significantly impact on PFS (P = 0.71, 0.99, and 0.92, respectively) or OS (P = 0.75, 0.96, and 0.99, respectively)
  • Pts with non-GCB DLBCL had significantly worse PFS (P = 0.01) and OS (P = 0.03) when their tumors were BCL2 positive
• In multivariate analysis, both GCB and non-GCB phenotypes according to Hans algorithm and BCL2 (70% threshold) had a significant prognostic impact on PFS (HR = 1.78, P = 0.003 and HR = 1.79, P = 0.003, respectively) and OS (HR = 1.85, P = 0.005 and HR = 1.67, P = 0.002, respectively) independently of IPI

Overall, it was found that GCB/non-GCB according to the Hans algorithm and BCL2 expression, but not MYC expression, predicted OS and PFS in this sample of patients treated with R-CHOP. They went on to state that “The prognostic impact of BCL2 is particularly important in the era of targeted therapy […] as this expression may predict a better response to anti-BCL2 targeted agents.”

Moreover, it was found that the Hans algorithm shows a good concordance with GEP (91%) and targeted RT-MLPA (94%) and so “in the era of targeted agents […] IHC evaluating COO using the Hans algorithm may remain a useful tool”. Although, they caution that their concordance was determined in a small subpopulation and therefore requires validation in larger populations.

Abstract:

BACKGROUND: Our aim was to evaluate whether the cell of origin (COO) as defined by the Hans algorithm and MYC/BCL2 coexpression, which are the two main biological risk factors in elderly patients treated with rituximab, cyclophosphamide, doxorubicin hydrochloride, vincristine, and prednisolone (R-CHOP), maintain their prognostic value in a large prospective clinical trial.

PATIENTS AND METHODS: We evaluated 285 paraffin-embedded samples from patients (60-80 years of age) enrolled in the Lymphoma Study Association (LYSA) trial LNH03-6B who were treated with R-CHOP. We correlated the COO defined by the transcriptome according to the Wright algorithm with that defined by the Hans algorithm in a subset of 62 tumors with available frozen tissue samples.

RESULTS: The non-germinal center B-cell-like (non-GCB) phenotype according to the Hans algorithm and BCL2 expression (but not MYC and BCL2 coexpression) predicted worse progression-free survival (PFS; hazard ratio (HR)=1.78, p = 0.003 and HR = 1.79, p = 0.003, respectively) and overall survival (OS; HR = 1.85, p = 0.005 and HR = 1.67, p = 0.02, respectively) independently of the International Prognostic Index (IPI). The correlation between the Hans algorithm and the Wright algorithm was 91%, with an almost perfect concordance according to a kappa test (0.81).

CONCLUSIONS: Our results suggest that immunohistochemically defined COO remains a useful tool for predicting prognosis in diffuse large B-cell lymphoma (DLBCL) when performed under optimized standardized conditions and that BCL2 expression may help to identify elderly patients at risk for relapse and who could potentially respond to anti-BCL2 targeted agents. In this prospective phase III trial, the coexpression of MYC and BCL2 does not appear to predict worse survival.