We assessed the tumor cellularity and tumor-infiltrating lymphocyte (TIL) on hematoxylin and eosin (H&E) slides and observed that these differed according to the HGP. Firstly, we found that in 8/10 patients, r-HGP exhibited higher tumor cellularity (median: 50%; range: 15%-85%) compared to its matched d-HGP samples (median: 17%; range: 1%-90%) (Fig. 1b-c). Additionally, and as expected, we observed a negative correlation between pathologically assessed tumor cellularity and computationally inferred MES (Spearman’s Correlation Coefficient= -0.38; p-value = 0.102) (Supplementary Fig. 2). Regarding the TIL, higher levels were observed in the d-HGP (median: 10%; range: 3%-53%) as compared to the r-HGP (median: 3%; range: 1%-39%) BCLM although not significant (p-value = 0.85; Supplementary Fig. 4a). We observed the same trend for the ImmuneScore inferred by deconvolution (p-value = 0.97, Supplementary Fig. 4b). Lastly, a positive moderate correlation between the TIL and ImmuneScore was seen (Spearman’s Correlation Coefficient = 0.464, p-value = 0.039) Supplementary Fig. 4c).
We next assessed the association between HGP and the quantitative proportions of cell derived from deconvolution analysis (Fig. 2a, Supplementary Table S5-S6, Supplementary Fig. 5). Focusing on immune cells, we reported a trend showing that the immune cells are more abundant in r-HGP except the CD4 + central memory T cells which were more enriched in d-HGP. However, given the limitations of deconvolution these results need to be interpreted cautiously.
Differential Gene Expression Analysis (DGEA) of the 20 matched samples identified 10 up-regulated (thus down-regulated in d-HGP) genes in r-HGP BCLM that were mainly associated with cell cycle and proliferation, DNA repair and nervous system (ASCL1, GFRA1, ZNF8, ZNF586, LINC01087, CCDC146, ASPH, CXCL4, SCTR and NPTX1) (Fig. 2b; Supplementary Fig. 3; Supplementary Table S1-S2). In contrast, six genes known to inhibit cell proliferation and tumor growth were enriched in d-HGP: SIRT6, MPC2, MIR4458HG, ST20-AS1, SREBF2, MRPL40 (Fig. 2b; Supplementary Fig. 3; Supplementary Table S1-S2). A gene clustering analysis of all genes up-regulated in r-HGP revealed that these genes were mainly involved in regulation of the cell cycle, cell division, extracellular matrix organization, nervous system process and actin filament based-process (Fig. 1d-f, Supplementary Table S3-S4). On the other hand, genes up-regulated in d-HGP BCLM were more associated with response to stress, immune activities and wound healing (Fig. 1c-e; Supplementary Table S1-S2).
Gene Set Enrichment Analysis (GSEA) revealed that pathways related to DNA double-strand break (DSB) repair, histone modification, nervous system, mitosis and meiosis were enriched in r-HGP BCLM (Fig. 2g, Supplementary Table S3-S4). Distinct gene sets related to immune activities were found to be enriched in both HGP.