Immune-vascular profiles in primary tumors and paired liver metastases of colorectal cancer

Background: The immune and vascular features of the tumor microenvironment influence outcomes of patients with colorectal cancer liver metastases (CRLMs). However, comparative analyses of these features, and their interactions, between primary colorectal tumors and their matched CRLMs, as well as their treatment-driven remodeling are limited. Methods: We analyzed 1-3 TMA cores per tumor (1 mm) from 50 matched primary tumors and CRLM (15 untreated, 35 neoadjuvant-treated with cytotoxic ±VEGF-/EGFR-targeted agents before liver resection). We used multiplex immunofluorescence to quantify T-cell densities (CD3, CD4, CD8, PD1, FOXP3, TIM3, Ki67), vessels (claudin-5) and perivascular (PV) subset (alphaSMA, PDGFRbeta) fractions. Statistical analyses included non-parametric tests (Wilcoxon, Mann-Whitney, Spearman) and Intraclass Correlation Coefficient (ICC) for concordance. Results: Comparisons of untreated paired primary tumors and CRLMs identified significantly lower densities of CD3+CD8+Ki67+, CD3+CD4+Ki67+, CD3+CD4+FOXP3+, and CD3+CD8+PD1+ cells (all p ≤ 0.006) in CRLMs. Regarding vessels, CRLMs showed a lower fraction of alphaSMA+PDGFRbeta+ PV subset (p=0.001), higher fraction of alphaSMA+PDGFRbeta- PV subset (p<0.001), and increased vessel density (p=0.01). Analyses contrasting untreated and treated CRLM showed reduced vessel size (p=0.002) in treated CRLM. Analysis of immune-vascular co-occurrence patterns of potential functional significance was also performed. In untreated primaries, the alphaSMA-PDGFRbeta+ PV subset correlated positively with densities of CD3+CD8+FOXP3+ (r=0.66, p=0.008), CD3+CD8+Ki67+ (r=0.61, p=0.016), and CD3+CD4+PD1+ (r=0.59, p=0.002) cells. In contrast, in untreated CRLMs, the alphaSMA+PDGFRbeta+ PV subset correlated with CD3+CD4+Ki67+ (r=0.56, p=0.03) and CD3+CD4+PD1+ (r=0.62, p=0.013) cell densities. Furthermore, in CRLMs, larger vessel size was uniquely associated with higher CD3+CD4+ cell density (r=0.54, p=0.04). Treatment-exposed CRLMs lacked the associations of untreated CRLMs and instead displayed new associations such as alphaSMA-PDGFRbeta+ PV subset and CD3+CD8+Ki67+ cells (r=0.40, p=0.018). Finally, ICC analyses of inside-case concordance regarding marker status in untreated primary, and CRLM showed that vascular metrics were more stable within patients (alphaSMA-PDGFRbeta+ PV subset ICC=0.60; vessel size ICC=0.36), than T-cell densities (mean ICC=0.02). Conclusion: These exploratory analyses indicate that CRLM is characterized by lower immune cell densities and a different PV stroma compared to primary tumors. Treatment appears to have a stronger impact on CRLM vessel features than on immune features. Furthermore, immune-vascular co-occurrence patterns are site specific and modulated by treatment. Collectively, these preliminary results identify immune-vascular features for further exploration regarding biomarker and drug target potential.