E180: Lymph Node Yield at Pancreatectomy is Influenced by Tumor Biology in Addition to Institutional and Surgical Factors

Higher lymph node (LN) yield in cancer surgery is associated with improved survival in some retrospective studies. This has been attributed to improved regional disease control with extensive lymphadenectomy, stage migration in stage-stratified analyses, and more detailed pathological review or surgical radicality in high-volume institutions, informing recommended minimum LN yield (LNY) for some cancer types. Prior studies suggested a relationship between tumor microenvironment and LNY in colorectal cancer, but this has not been demonstrated in other cancers such as pancreatic ductal adenocarcinoma (PDAC).

Methods:

Bulk RNAseq data were obtained from The Cancer Genome Atlas (TCGA) PDAC study. Cases with non-PDAC histology, metastases at diagnosis, and incomplete staging were excluded. Selected samples included only primary tumors from institutions submitting at least two cases to TCGA. Consortia and commercial biobanks were excluded, allowing adjustment for median institutional LNY. Differential expression (DE) analysis was performed using DESeq2. Associated biological pathways were characterized by gene set enrichment analysis (GSEA). All p values in RNAseq analyses were Benjamini-Hochberg multiplicity-adjusted. Survival was plotted by the Kaplan-Meier method.

Results:

Variation was noted in LNY by submitting center across the TCGA study (range in median 10.5-30, p = 0.001). In the final cohort of 51 cases, 90 genes were significantly associated with differential LNY compared to the site median (p< 0.05, absolute log2(fold-change) >0.5). The most significant associations were for CCL25, FABP2, MS4A10, CYP4F22, SLC26A3, APOB, and APOA4 (p< 0.00001 for each). GSEA demonstrated 26 associated pathways (p< 0.01), including immune response and lipid metabolism pathways. Leukocyte marker gene expression suggested heterogenous contribution of the tumor immune microenvironment to the described effect. Clustering with the high-LNY cases by gene expression profile was associated with numerically but not statistically longer recurrence-free survival (median 17.1 vs 11.9 months, p = 0.29).

Conclusions:

Although institutional factors influence LN yield at pancreatectomy for PDAC, a bioinformatics analysis controlling for this effect demonstrates that yield is also significantly associated with the gene expression profile of the primary tumor, potentially influencing outcome. This relationship between tumor biology and LNY potentially confounds studies demonstrating an apparent survival impact of higher LN counts.