Comparative analysis of T-cell subsets and vessel features in matched primary colorectal tumors and corresponding resected liver metastases.

3139 Background: Outcome after liver resection for colorectal cancer metastases (CRLM) is partly determined by factors such as the number, size, and vitality of the metastases, as well as the T- and N-stage of the primary tumor. The tumor microenvironment—particularly immune cell infiltration and vascular features—also influences outcome. Data on how these factors compare between paired primary colorectal tumors and matched CRLM are limited. Methods: We used TMAs from matched primary tumors and CRLM samples of 50 patients, of which 15 were untreated and 35 had received neoadjuvant therapy (cytotoxic ± VEGF-/EGFR-targeted agents) before liver resection. Each tumor consisted of 1–3 tissue cores (1 mm) from both the tumor center and the invasive margin. Multiplex immunofluorescence was performed to assess T-cell (e.g., CD3, CD4, CD8, PD1, FOXP3, TIM3, Ki67), and vessel (claudin-5, αSMA, PDGFRβ) markers. Nonparametric Wilcoxon signed rank and Spearman correlations were used for cell density comparisons. Cox regression for continuous variables was used for disease-free survival (DFS) associations. Results: We observed significantly lower densities of CD3+CD8+Ki67+, CD3+CD4+Ki67+, and CD3+CD4+FOXP3+ cells in CRLM than in paired primary (all p < .006) in both untreated and pretreated cohorts. The αSMA+PDGFRβ– vessel subset was more prevalent in CRLM compared with the primary tumor in the untreated cohort (p < .001). In the untreated cohort, larger vessel size in CRLM (but not in the primary tumor) showed a positive correlation with CD3+CD8+PD1+, CD3+CD4+PD1+, and CD3+CD4+FOXP3+ densities (Spearman r = .54–.60, p = .02–.04). In the pretreated cohort, higher tumor vitality and/or CDX2+ expression in CRLM (indicative of poor treatment response) were each negatively correlated with cytotoxic (CD3+CD8+) and helper T-cell (CD3+CD4+) subsets (r = -.42 to -.63, p < .01). DFS after metastasectomy was associated with vessel and T-cell features. Regarding vessel metrics in the small untreated cohort, αSMA–PDGFRβ– vessel subset in primary showed a negative trend (p = .06) as did smaller vessel size in CRLM (p = .06). CD3–CD4+TIM3+ in CRLM was negatively associated with DFS in pretreated (p = .04), with a trend also in untreated (p = .10). Conclusions: Densities of certain T-cell subsets are significantly lower in matched CRLM than in primary tumors indicating immune desert phenotype. Vessel subset profiling suggests differences between primary tumors and CRLM, possibly relevant for treatment response. Poor pretreatment effect, i.e., high vitality and CDX2+ density in CRLM, was negatively correlated with several T-cell subsets, a correlation not seen in untreated. The poor prognosis association of CD3–CD4+TIM3+ cells in CRLM merits further investigation.