Title : Profiling the response to neoadjuvant chemoradiotherapy in Locally Advanced Rectal Cancer
Abstract:
Background: Neoadjuvant chemoradiotherapy (CRT) is the golden standard in locally advanced rectal cancer (LARC), although the response is not uniform among patients. Profiling the response to CRT in patients with LARC would enable more adequate selection of patients who would benefit most from CRT while minimizing adverse effects and toxicity. The aim of this research was to identify a set of genes with predictive potential in LARC patients treated with neoadjuvant CRT.
Methods: Gene Set Enrichment Analysis (GSEA) was performed on selected data sets, and Hallmark, KEGG, and Reactome gene sets were used to compare expression levels in LARC patients who responded well to therapy versus those who did not, according to patohistological tumor regression grading (TRG) categories (responders TRG 1-2, non-responders TRG 3-5). Gene expression patterns were analyzed using the public database NCBI Gene Expression Omnibus (GEO). The interactive publicly available databases the Human Protein Atlas and UALCAN were used to analyze the Cancer Genome Atlas (TCGA) transcriptome data and confirm the expression of selected gene candidates in rectal cancer. The analyzed patient cohort consisted of 102 patients (68 males and 34 females, age range 38-76 years) with clinically and pathologically confirmed locally advanced rectal adenocarcinoma (stage II/III, ECOG PS≤2). Patients were treated with preoperative chemoradiotherapy for 5 weeks (5-fluorouracil/leucovorin/50.4 Gy in 28 fractions) and the pathological response was evaluated using the Mandard tumor regression grading (TRG) system. Gene expression was determined by qRT-PCR from FFPE biopsy samples. Receiver operating characteristics (ROC) analysis and area under the curve (AUC) with 95% confidence interval (CI) was applied for the investigation of the discriminatory potential of gene expression with significance set at p < 0.05.
Results: GSEA highlighted the significance of gene expression sets of E2F targets, G2M checkpoint, DNA repair, fatty acid metabolism, glycolysis and gluconeogenesis, cholesterol homeostasis and inflammatory response in this setting. As an example, we evaluated one gene candidate from these sets, the thymic stromal lymphopoietin (TSLP), which is involved in the maintenance of immune homeostasis within the gut by modulating the Th1(Th17)/Th2 balance. In silico analysis showed that high TSLP expression is an unfavorable prognostic factor in colorectal cancer, with a 5-year survival rate for high expression of 54%, and 64% for patients with low TSLP-expressing tumors (p=0.035, median follow up time: 1.92 years). In our patient cohort, patients who achieved a good pathological response (TRG1-2) had significantly lower pre-treatment levels of TSLP compared to patients with a partial/no response (TRG3-5), and ROC analysis showed that TSLP expression might predict an unfavorable response to preoperative chemoradiotherapy (p<0.001).
Conclusions: Using publicly available in silico data, successful prediction of gene candidates for optimized treatment in various oncological scenarios is becoming increasingly important. In this study, it was determined that increased levels of TSLP might be correlated with a poor response to preoperative chemoradiotherapy in locally advanced rectal cancer patients, by inducing a switch to Th2-mediated immunity or other mechanisms that are currently being explored.