Immune Contexture of Colorectal Adenomas and Serrated Polyps
This study is a cross-sectional analysis of adults with a recent history of colorectal neoplasia, designed to characterize immune-related gene expression patterns in preinvasive colorectal lesions. The study population includes participants with at least one histologically confirmed colorectal adenoma or sessile serrated lesion identified at baseline. Eligible participants were adults aged 35–80 years with tubular adenomas, tubulovillous or villous adenomas, or sessile serrated lesions. Archived formalin-fixed, paraffin-embedded (FFPE) tissue specimens were obtained from participants enrolled in the Vitamin D/Calcium Polyp Prevention Study, the Tennessee Colorectal Polyp Study, and the Diet and Health Studies. RNA was extracted from FFPE tissue samples and profiled using a custom NanoString nCounter gene expression panel comprising approximately 275 genes related to immune function, inflammation, and the tumor microenvironment. Gene expression data underwent standard quality control procedures and normalization using established NanoString workflows.
Gene expression profiling was conducted in multiple NanoString nCounter assay batches. To minimize confounding technical variation, samples were balanced across batches by sample size and lesion histology, with random allocation by source study site. Nanostring cartridge variables to account for batch are provided in the data submission. Quality control began with discarding any sample that had 0 expression across all genes as these samples were believed to have no RNA present. Additional quality control was conducted using a 10-gene colon-specific control panel to identify missingness. Further outliers were identified via principal component analysis. Technical replicates were averaged prior to downstream modeling.
Differential gene expression analyses were conducted across histologic subtypes and anatomic location at both the individual gene and gene signature levels. Individual gene analyses were performed using linear modeling approaches implemented in the LIMMA framework. Gene set-level differences were evaluated using non-parametric, distance-based multivariate analysis of variance. Multiple normalization approaches were evaluated, including nSolver, RUVg, and NanoStringDiff, to assess robustness to technical variation. The dataset includes raw gene expression count data contained within the submitted RCC files, lesion-level characteristics (including histologic subtype and colonic location, size), and participant-level phenotype data. Phenotype data include demographic characteristics (age, sex, race, ethnicity), lifestyle factors, medication use, and source study. Lesion-level and participant-level data are linked using de-identified participant and sample identifiers, enabling downstream users to account for batch structure and other technical and biological factors in secondary analyses.
- Type: Genotype/Expression
- Archive: The database of Genotypes and Phenotypes (dbGaP)
