Unveiling the Tumor Microenvironment in Extranodal NK/T-Cell Lymphoma (ENKTCL) Patients with in Situ Spatial Transcriptomics

Background: ENKTCL is characterized by heterogeneity in genomic mutations, EBV-encoded gene expression, and host HLA-variants conferring differing susceptibility to viral genes. However, the cellular architecture of the ENKTCL tumor microenvironment (TME) and its potential association with sensitivity to chemotherapy across diverse racial and ethnic groups globally remain poorly understood.

Aim: We sought to decipher the heterogeneous landscape of ENKTCL TME in diagnostic patient samples at the cellular level and in situ to identify key cell populations, their spatial distribution within the TME, and cell-to-cell interactions that correlated with chemotherapy response.

Methods: We conducted high-resolution, in situ spatial transcriptomic profiling using the 10x Genomics Xenium platform on 12 formalin-fixed, paraffin-embedded (FFPE) ENKTCL tumor samples. We designed a custom ENKTCL-specific 480 target gene panel through systematic analyses of publicly available clinical, genomic, and transcriptomic data. Specific probes for ~20 genes related to EBV infection and viral replication were also incorporated to capture viral dynamics and effects on the TME. Importantly, the cohort incorporated patients from distinct ethnic backgrounds across the United States (n=7) and South America (n=5) to enable a more heterogeneous representation of this disease. Within the cohort, 5 cases were male and 7 female, and the median age of diagnosis was 52 (range, 25-71). The median Ann Arbor stage at diagnosis was I (range, I-IV). Of the 12 cases, 11 had nasal, and 1 case had cutaneous involvement as the primary site of disease. Overall, 6 patients had advanced stage, and 6 patients had localized disease. Based on the PINK score, patients were identified as low risk (n=3), intermediate risk (n=2), and high risk (n=7). The cohort received distinct but heterogeneous frontline treatments that included SMILE (n=4), CHOEP (n=1), DeVIC (n=3), MTX/Asp/Dex (n=1), cisplatinum with radiotherapy (n=2), and palliative radiation therapy (n=1). After frontline treatment, patients exhibited CR (n=5), PR (n=4), or PD (n=3). Following frontline therapy, 1 patient underwent allogeneic stem cell transplant. The median time for relapse was 5 m (range, 4-6). Overall, 3 patients underwent second-line treatment, with GemOx (n=1), Brentuximab Vedotin (n=1), and SMILE (n=1), achieving PR (n=2) and PD (n=1). Third and fourth-line treatments included romidepsin (n=1), palliative chemotherapy (n=1), cytotoxic T-lymphocytes on a clinical trial (n=1) and pralatrexate (n=1).

Results: In situ transcriptomics enabled an in-depth, unbiased interrogation of the cellular and transcriptional landscape of the ENKTCL TME. After quality control, we identified a total of 2,122,592 cells from all samples in the cohort, spanning 9 major lineages, including tumor-infiltrating lymphocytes such as NK, T, B, plasma, and myeloid cells, as well as fibroblasts, endothelial, epithelial, and mast cells, characterized based on canonical marker expression. Intensive lymphocyte infiltration was observed, with over 700,000 T and NK cells and approximately 650,000 myeloid cells composing the largest compartments. More than 100,000 cells were identified as EBV-infected, including macrophages, NK, NKT, and transformed B cells. Notably, EBV-infected NK, B, and mast cells were more abundant in CR patients, while changes were also observed in macrophages, with higher abundances in the PD and relapsed patient groups. Only the PR group presented significantly elevated abundances of EBV-infected NKT cells, while CD8+ T cells were less prevalent in CR patients. We uncovered numerous differences between groups at the organizational, spatial, cellular, and molecular levels, pointing to specific ENKTCL TME endophenotypes associated with outcomes.

Conclusion: This study represents the first use of spatial transcriptomics to interrogate the ENKTCL TME in situ. The extensive insights gained from cell co-localization and cell-cell communication signals will prioritize novel biomarkers that underlie responsiveness to chemotherapy. Correlative multi-omic analyses with genomic alterations and HLA specificity in a first-of-its-kind multiregional and highly multiplexed cohort are underway and will be reported at the meeting.