Spatial transcriptomic analysis of tumor tissue from patients with metastatic urothelial carcinoma (mUC) receiving ipilimumab (IPI) plus nivolumab (Nivo) combined with sacituzumab govitecan (SG).

863 Background: We previously reported that the combination of IPI-NIVO plus SG shows high efficacy with 83.3% objective response rate (ORR) and durable responses as first-line treatment for cisplatin-ineligible mUC patients in a Phase I/II trial. However, two grade 5 immune-mediated myocarditis events attributed to IPI-NIVO occurred, leading to early trial termination after accruing 25 patients. Given the high ORR, we performed correlative analysis to identify a predictive biomarker for response. Methods: For Nanostring GeoMx DSP spatial transcriptomics, 19 baseline tumor samples were evaluable. Patients were categorized as responders (CR+ PR, n=15) and non-responders (SD+ PD, n=4). The raw Nanostring GeoMx probeQC count data was quality-controlled for both segment and gene levels using GeomxTools package in R version 4.3. The segments with flags such as ‘Low Percent Aligned Reads’, ‘Low Percent Stitched Reads’, ‘Low Surface Area’, ‘Low Nuclei Count’, ‘Low Negative Mean’, and ‘Low gene detection rate’ were excluded from further analysis. In total, 35 and 24 segments were left for responder and non-responder groups, respectively. Genes with raw count smaller than the ROI-specific limitation of quantification (LOQ) and low detection rate among segments were excluded from downstream analysis. Totally, 8230 genes were left for 59 segments for further analysis. The differential expression between groups was done using a linear mixed-effect model. Results: 90 and 136 genes were significantly up- and down-regulated in responder group and non-responder group (P value < 0.05 for both). The top upregulated genes in responders included: IGHG2, IGKC, ACKR2, KRT80 and SCD, while those upregulated in non-responders included: S100A2, AKR1C3, MYCN, MAGEA3, and OBSCN (P value < 0.05) . Gene set enrichment analysis (GSEA) revealed gene signatures from B cell-mediated immunity pathway significantly enriched in the responder group (FDR = 0.02), while gene signatures from PRC2_EZH2_UP.V1_DN related to polycomb-mediated transcription reprogramming and promoting tumorigenesis were significantly enriched in the non-responder group (FDR = 0.0017). Trop2 and Topoisomerase 1 genes were upregulated in responders but not statistically significant. Conclusions: Spatial transcriptomics analysis suggests that upregulated B cell-mediated immunity pathways may predict response to combination IPI-NIVO + SG in patients with mUC. The PRC2-EZH2 pathway may represent a potential resistance mechanism and therapeutic target. Transcriptomic evaluation of the therapeutic targets of the antibody and the payload in antibody-drug conjugates warrant further study as predictive biomarkers. Further analyses to identify biomarkers for severe immune events may help optimally select patients for the best therapeutic index. Clinical trial information: NCT04863885 .