A gene expression signature to predict disease progression for Hodgkin Lymphoma patients who achieve a complete metabolic response after 2 ABVD courses

Background: Classical Hodgkin Lymphoma (cHL) is one of the most manageable human cancers. The early identification of patients who experience relapse after completion of front-line therapy, regardless of their initial stage currently represent an unsolved need. It is likely that disease progression reflects some innate features that escape the current prognostic criteria but that can be revealed by a deep analysis of the molecular assets of the lesions. We employed a deep gene expression analysis searching for molecular determinants that could anticipate the risk of relapse among patients who achieve a complete metabolic response after 2 ABVD courses (iPET-). Patients and methods: We conducted a retrospective search of our local clinical archives to include patients with the following characteristics: confirmed cHL histology, age >18, iPET negative after 2 ABVD course (Deauville score 1–3). We retrieved the baseline diagnostic biopsy to conduct a gene expression analysis by nCounter Nanostring Technology using the PanCancer Immune-panel. Genomic data were correlated with clinical and laboratory data and with patients’ outcomes. Primary endpoint of this analysis was Progression Free Survival (PFS). Results: Out of 215 cHL patients seen in our institution from 2004 to 2019, 148 achieved a negative iPET after 2 ABVD course. FFPE material was available for 120 iPET- cases which constitute the study population. Forty percent of patients were older than 45 years, 38% had stage III-IV, and 15% had Bulky disease. With a median follow up of 63 months (range, 7–139 months) we recorded 31 events for PFS. The resulting 4 year PFS rate was 77% (95% CI: 71.1–84.9). Analysis performed by Cox Proportional Hazard model identified 54 genes whose expression was significantly associated with PFS (p ≤ 0.05). Of these, 43 were positively associated with improved PFS indicating a potential protective role of these factors. Vice versa, only 11 genes were significantly associated with reduced survival probability. Gene Ontology analysis showed that protective genes were enriched in B-cells related pathways and response to cytokines, pointing to a shielding function of the microenvironment with respect to disease aggressiveness. Unsupervised clustering analysis using the 43 genes protective signature identify two separate clusters of patients (Figure 1A). Kaplan Meyer curve analysis showed that Cluster 2 patients had a significative reduced PFS as compared to Cluster 1 (p = 0.00043), supporting the prognostic relevance of these genes (Figure 1B). The research was funded by: AIRC-IG2021-25802 Keywords: Genomics, Epigenomics, and Other -Omics, PET-CT, Tumor Biology and Heterogeneity No conflicts of interests pertinent to the abstract.