Association of Gene Expression Profiling, Interim PET, EBV Status and PD-L1 Expression in Classical Hodgkin Lymphoma

Introduction 

A challenge in the management of classical Hodgkin lymphoma (CHL) is selecting treatment which will maximise success while minimising treatment side effects. The international Response Adapted Therapy in Hodgkin Lymphoma (RATHL) trial (NCT00678327) demonstrated that treatment may be successfully adapted based on positron emission tomography (PET-CT) scan results after 2 cycles (PET2) of ABVD chemotherapy in advanced CHL. This approach however delays risk stratification meaning some patients may be over or under treated and therefore a baseline predictor is highly desirable. A gene expression-based model performed on RNA from formalin-fixed paraffin-embedded tissue (FFPET) biopsies on the Nanostring platform (Scott et al, J Clin Oncol 2013; 31:692-700) has been published. In addition Steidl et al used gene expression profiling (GEP) to correlate tumour-associated macrophages with survival in CHL (Steidl et al, NEJM 2010; 10:875-885). In order to explore whether GEP at baseline could be used in combination with PET2 results to predict outcome, these published gene signatures were explored on the DASL platform in a series of CHL cases and correlated with PET2 response. Recently, PD-1 ligand (PD-L1) on the cell surface of Reed-Sternberg cells (the hallmark cells of CHL) has been demonstrated to co-opt the PD-1 pathway allowing immune evasion (Green et al, Blood 2010;116:3268-77). Therefore this was also performed by immunohistochemistry (IHC) and GEP as well as assessment of EBV status.

Methods

50 patients with CHL were identified, with at least 2 years of follow up, diagnosed between 2008 and 2013. Deauville scores at PET2 were determined. IHC consistent with CHL, including: CD3, CD19, CD20, CD30, CD79, BOB-1, OCT-2, MUM1 and TARC was performed on 3mm FFPET lymph node sections, and reported by two independent observers. Latent Epstein-Barr virus (EBV) infection was determined by IHC for LMP-1 expression and in-situ hybridisation using fluorescein-labelled peptide nucleic acid probes (Dako, K5201). PD-L1 expression was determined by IHC using a rabbit monoclonal antibody (Cell Signaling Technology, E1L3N, #13684). Three to five 5mm sections of FFPET were used for RNA extraction using the Ambion RecoverAll™ kit standard protocol. GEP was quantified using Illumina’s whole genome cDNA-mediated annealing, selection, extension, and ligation Assay (WG-DASL). 61 genes of interest were analysed for significance in the difference in gene expression between groups and included genes from the two recently reported GEP predictor tools in CHL (Steidl et al, 2010, Scott et al 2013) as well as PD-1 and PD-L1. The Mann-Whitney test was used to assess the significance in the difference between means (significant if p<0.05).

Results

A statistically significant difference in EBV status was found between PET negative (Deauville scores 1-3) and PET positive (Deauville scores 4-5) groups (Fisher’s exact test: P-Value = 0.028). All patients who were EBV positive had a negative PET2 scan. GEP using WG-DASL revealed that only 2 genes of those reported in recent predictor models (Steidl et al 2010, Scott et al 2013) were expressed at significantly different levels between PET negative and PET positive patients (GLUL, RNF144B), both with increased expression in the PET2 positive group. There was no significant difference between PD-L1 expression and PET scores, CD274 gene expression and outcome. Assessment as to whether samples would cluster into groups according to PET positivity or PD-L1 expression status by unsupervised cluster analysis using four recently reported B cell lymphoma gene signatures (Monti et al, Blood 2005; 5:1851-61, Care et al, PloS one 2013; 2: e55895) was performed, but no significant grouping was found.

Conclusion

GEP on the DASL platform was not able to predict PET response. The original models were trained on outcome and therefore retraining of the GEP model based on PET response is likely to be required. EBV status was found to be predictive of PET response but this could not be correlated with other biomarkers to predict outcome. Although PD-1 and PD-L1 targeted therapy has shown exciting results in patients with CHL, PD-L1 IHC expression and CD274 gene expression, did not correlate with PET2 response or outcome. Baseline biomarkers capable of identifying patients likely to benefit from targeted treatment needs to be further investigated in proposed clinical trials.