Document resource
Background Lymphoid aggregates (LAs) and tertiary lymphoid structures (TLS) have not been extensively studied in melanoma brain metastases (BM) or lung adenocarcinoma (LUAD) BM despite their prognostic value in solid tumors and immunotherapeutic responses. BM occur in ~30% of patients with solid tumors, most commonly melanoma, lung, and breast cancers, and are among the deadliest forms of cancer metastasis. Effective therapeutic strategies for the prevention and treatment of BM remain elusive, resulting in dismal overall survival. We hypothesized that B-cells and TLS would associate with increased immune infiltration in BM.Methods We assessed the heterogeneity of B-cell infiltration and LA formation in >200 melanoma-BM or LUAD-BM, linking these metrics to survival and molecular alterations. We initially screened BM samples for LAs using immunohistochemistry, followed by multispectral imaging and machine learning to analyze LA and non-LA cellular neighborhoods in matched primary tumors and BM. We applied GeoMx® DSP for proteomic profiling of BM and extracranial metastases (ECM), with validation through gene set enrichment and pathway analysis in independent RNA-seq BM cohorts.Results We identified LAs, but not bona fide TLSs, in ~30% of melanoma and LUAD-BM tumors. LAs associated with improved survival in melanoma-BM but not LUAD-BM ( figure 1). Despite similar LA frequency, the divergence in survival outcome led us to explore differences in LA function and composition. Compared to ECM of melanoma and LUAD, LAs were less frequent in BM. Moreover, GeoMx® proteomic and machine-learning analysis of multispectral images revealed that LAs in melanoma-BM but not LUAD-BM were enriched for active (HLADR+, ICOS+, GZMB+) and proliferating B and T-cells, suggesting potential effector function. Melanoma-BM LAs also had increased intratumoral CD8+T-cells, unlike LUAD-BM, where CD8+Tcells localized peritumorally, indicating immune exclusion. Moreover, LAs in melanoma-BM predominately consisted of germinal center (GC)-derived memory B-cells (MBCs), whereas LAs in LUAD-BM were enriched for non-GC derived MBCs, often dysfunctional in solid tumors. Finally, assessment of TLS/LA derived gene sets in independent melanoma-BM and LUAD-BM RNAseq cohorts confirmed consistent trends in immune activity and patient survival.Conclusions Among tumor metastases, patients with BM are considered to have the greatest mortality and very limited therapeutic options. This study uncovers immunologically active LAs as potential orchestrators of CD8+ T cell mediated response in melanoma-BM, emphasizing their potential as a therapeutic target within the brain’s unique tumor microenvironment. Our findings contribute to an extremely understudied area in cancer metastases, emphasizing B-cells and LAs as promising therapeutic targets, specifically in BM and primary brain tumors.Acknowledgements We thank all the members of the Bruno lab for their support of this work. We are grateful for Marion Joy and Grace Ge at the Translational Pathology Imaging Laboratory (TPIL) for immunofluorescence panel development and multispectral imaging of tissue sections; Laura P. Stabile PhD and Timothy F. Burns MD PhD for providing us with LUAD-BM FFPE tissue, RNAseq data, and their constructive comments and feedback; Thomas Pearce MD PhD and Pitt Biospecimen Core for the melanoma-BM tissues. Thomas Pearce MD PhD and Gabriel L. Sica MD PhD for reviewing tissue slides, tumor annotations, and feedback; Micheal A. Davis MD PhD for Melanoma-BM RNAseq data and their constructive comments and feedback; Rami B Yanes and Xiancheng Wu for their LUAD-BM clinical data collection; UPMC Registry Information Services for providing us with the melanoma-BM clinical data; Riyue Bao, PhD, and Rajesh Acharya PhD at The Hillman Cancer Bioinformatics Services for GeoMX and RNAseq data analysis; Caroline Layding and Curtis Mamoru Tatsuoka PhD at the UPMC Hillman Cancer Center Biostatistics Facility for statistical analysis services; UPMC Shadyside Tissue and Research Pathology Services (TARPS) for pathology services; the University of Colorado Human Immune Monitoring Shared Resource Facility for assisting in multispectral panel optimization.Abstract 181 Figure 1Kaplan-Meier plot of Melanoma-BM (n=56) and LUAD-BM (n=107) patients who underwent craniotomy based on the presence and absence of LA. Survival is calculated from the day of resection to the last contact day. Patients without an exact resection date were