Pathologic determinants of ctDNA-shedding in colorectal cancer metastases.

Abstract

183 Background: In colorectal cancer (CRC), the concordance of genomic alterations between plasma-derived ctDNA and matched tumor tissue is lower in patients with lung-only metastases than in those with liver metastases. This suggests that organ-related and histopathologic characteristics may influence ctDNA-shedding. We therefore aimed to investigate which histopathologic factors are associated with ctDNA-shedding. Methods: This prospective study included patients who had undergone resection of the primary site of CRC and were scheduled for lung or liver metastasectomy. Plasma was collected immediately before surgery for ctDNA analysis of 12 genes commonly altered in CRC, using the Plasma-Safe-SeqS CRC assay (Sysmex, Kobe, Japan). To determine whether detected alterations were tumor-derived, matched tissue samples were also analyzed using an NGS-based panel assay. Cases with mutations detected in both tissue and plasma were defined as ctDNA-shedding, those with mutations in tissue but not in plasma as ctDNA-non-shedding, and cases without tissue mutations were considered non-mutated and excluded from the analysis. Digital pathology quantified metastatic size, necrosis area, total tumor-cell count, and cell density using HALO Image Analysis Platform. ROC curves were used to determine cutoff values for each factor, and factors associated with ctDNA-shedding were evaluated. Results: Among 37 analyzable cases (lung=26, liver=11), ctDNA-shedding was observed in 13 (35.1%). Detection differed by organ: lung 4/26 (15.4%) vs liver 9/11 (81.8%). ROC analyses yielded the following cutoffs (AUC): size 15.0 mm (AUC 0.877), necrosis area 19.822 mm² (AUC 0.869), total tumor-cell count 552,471 (AUC 0.881), density 7,834.418 (AUC 0.671). Multivariate analysis including these factors and organ site identified necrosis area as significantly associated with ctDNA-shedding (P=0.0292). Correlation coefficient analysis with plasma mutant allele fraction (MAF) showed significant associations for necrosis area (P<0.0001) and tumor size (P<0.0001) whereas total tumor-cell count (P=0.3438) and density (P=0.7020) were not significantly correlated. Conclusions: ctDNA detectability in resectable CRC metastases was lower in lung than in liver metastasis, but pathological analysis demonstrated that necrosis area remained a significant factor for ctDNA-shedding in multivariate analysis, and necrosis area was also significantly correlated with MAF. These findings indicate that although ctDNA-shedding varies by organ, necrosis is an important histopathologic feature and suggest that necrosis should be considered a notable factor when interpreting plasma-based analyses. Clinical trial information: UMIN000045411 .