Modulation of the Tumor Microenvironment By CCL19 in Primary CNS Lymphomas

Introduction:

Refractory or relapsed central nervous system lymphoma (r/r CNSL) remains a therapeutic challenge with poor overall survival. The biological mechanisms underlying clinical heterogeneity in response to standard immunochemotherapy remain largely unknown. Previous publications suggest a critical role of CCL19 for parenchymal retention of malignant B-cells and formation of CNSL in the brain of mice (O'Connor et al., Cancer Cell, 2019). Here, we characterized CCL19 in primary CNSL tissue, explored its prognostic role, the impact on the tumor microenvironment (TME), and its spatial/cellular origins in patients with CNSL using a multi-omics approach.

Methods:

We comprehensively profiled 82 tumor specimens obtained through stereotactic biopsies (n=74) or brain tumor resections (n=8) from primary CNSL (PCNSL) patients by bulk RNA sequencing (RNA-Seq), targeted capture next-generation sequencing (CAPP-Seq), and shallow whole-genome sequencing. CIBERSORTx was used to define immunological and CNS-specific cell subsets. The immune cell landscape and genetic/transcriptional features were further correlated with clinical and radiological parameters. To investigate the cellular and spatial origin of CCL19, 10x single nucleus RNA-Seq (snRNA-Seq, n=7, 150,300 cells) and spatial transcriptomics (n=7) of primary CNSL samples were performed.

Results:

CLL19 expression was significantly correlated with tumor volumes measured by magnetic resonance imaging (MRI, p=0.02, r=0.4), supporting the results observed in mouse models. Furthermore, CCL19 was a strong prognostic marker in patients undergoing curative-intent high-dose methotrexate-based immunochemotherapy, with high CCL19 expression levels being associated with significantly unfavorable progression-free survival (p=0.009, HR=2.4, CI: 1.2–4.7) and overall survival (p=0.02, HR=2.5, CI: 1.1–5.8). We further observed that CNSL tumor biopsies revealing high CCL19 expression levels were significantly enriched for CD4+ memory T cells and NK cells in a resting state and tended to be associated with a higher number of macrophages in the TME. CCL19-high tumors did not present with a specific mutational or copy number profile; yet, all EBV-positive cases (n=5) showed high CCL19 levels. We further delineated the unique cellular components of the brain from seven CNSL tissue samples by snRNA-Seq and found that the stromal cell compartment displayed the highest expression of CCL19 in the CNSL TME. Finally, spatial transcriptomics from seven CNSL tumor slices showed particularly high CCL19 expression in vessel-rich areas and decreasing CCL19 levels in less-vascularized regions.

Conclusion:

Collectively, we identified CCL19 as a prognostic molecular marker in CNSL, predicting unfavorable clinical outcomes. Our results indicate that CCL19, derived from stromal cells in vessel-rich areas, might cause retention of malignant B-cells in the brain parenchyma of CNSL patients and may be associated with modifications of the immune landscape towards a cold microenvironment.