Gene expression analysis

RNA was extracted from snap-frozen HCC explant tumor specimens and purified using RNeasy Mini Kit (Qiagen; Hilden, Germany). RNA quality was verified by Nanodrop spectrophotometer (VWR; Radnor, PA) and Bioanalyzer (Agilent; Santa Clara, CA). Gene expression microarray analysis was performed using the Affymetrix Human Gene 2.0 ST platform (Thermo Fisher, Waltham, MA). The initial cohort included eleven samples, but two of them did not pass the quality control filter required for microarray processing, therefore the remaining nine were used for further analysis. We compared gene expression of the explant HCC specimens between those who developed recurrence (n = 6) versus those who did not develop recurrence (n = 3).

Microarray data were pre-processed (background subtraction and quantile normalization) by Robust Multi-array Average (RMA) utilizing the oligo package in R version 3.6 [26]. Gene annotation was done using pd.hugene.2.0.st annotation files [27]. Statistical analysis was performed with limma package [28]. After having fit the model with lmFit function (linear model), the differential gene expression was calculated using eBayes function (moderated t-test, p-value, B stats). A gene was considered differentially expressed between the two groups if p-value < 0.05. In addition, filtering criteria by fold change (FC) was selected: FC ≥ 2 (upregulation) or FC ≤ 0.5 (downregulation).

For identification of protein–protein interactions and pathways in explant HCC that were predictive of HCC recurrence, we analyzed the common significant genes/proteins separately for the two groups using STRING software, version 10.5 (https://string-db.org). The transcriptomic data have been deposited to Gene Expression Ominibus (GEO) repository with the dataset identifier {"type":"entrez-geo","attrs":{"text":"GSE164368","term_id":"164368"}}GSE164368.