APRIL is a novel clinical chemo-resistance biomarker in colorectal adenocarcinoma identified by gene expression profiling

Abstract

Background: 5-Fluorouracil(5FU) and oral analogues, such as capecitabine, remain one of the most useful agents for the treatment of colorectal adenocarcinoma. Low toxicity and convenience of administration facilitate use, however clinical resistance is a major limitation. Investigation has failed to fully explain the molecular mechanisms of resistance and no clinically useful predictive biomarkers for 5FU resistance have been identified. We investigated the molecular mechanisms of clinical 5FU resistance in colorectal adenocarcinoma patients in a prospective biomarker discovery project utilising gene expression profiling. The aim was to identify novel 5FU resistance mechanisms and qualify these as candidate biomarkers and therapeutic targets.

Methods: Putative treatment specific gene expression changes were identified in a transcriptomics study of rectal adenocarcinomas, biopsied and profiled before and after pre-operative short-course radiotherapy or 5FU based chemo-radiotherapy, using microarrays. Tumour from untreated controls at diagnosis and resection identified treatment-independent gene expression changes. Candidate 5FU chemo-resistant genes were identified by comparison of gene expression data sets from these clinical specimens with gene expression signatures from our previous studies of colorectal cancer cell lines, where parental and daughter lines resistant to 5FU were compared. A colorectal adenocarcinoma tissue microarray (n = 234, resected tumours) was used as an independent set to qualify candidates thus identified.

Results: APRIL/TNFSF13 mRNA was significantly upregulated following 5FU based concurrent chemo-radiotherapy and in 5FU resistant colorectal adenocarcinoma cell lines but not in radiotherapy alone treated colorectal adenocarcinomas. Consistent with APRIL's known function as an autocrine or paracrine secreted molecule, stromal but not tumour cell protein expression by immunohistochemistry was correlated with poor prognosis (p = 0.019) in the independent set. Stratified analysis revealed that protein expression of APRIL in the tumour stroma is associated with survival in adjuvant 5FU treated patients only (n = 103, p < 0.001), and is independently predictive of lack of clinical benefit from adjuvant 5FU [HR 6.25 (95%CI 1.48-26.32), p = 0.013].

Conclusions: A combined investigative model, analysing the transcriptional response in clinical tumour specimens and cancers cell lines, has identified APRIL, a novel chemo-resistance biomarker with independent predictive impact in 5FU-treated CRC patients, that may represent a target for novel therapeutics.

Figure 1

Hierarchical cluster analysis of chemoradiotherapy or radiotherapy treated tumours. This analysis separates pre- and post-treatment biopsies using (a) 86 genes identified as changed in chemoradiotherapy treated patients and (b) 51 genes identified as changed in short course radiotherapy treated patients. (c) Post-treatment tumour biopsies, cluster according to treatment received with the combined set of 137 genes, but (d) pre-treatment tumour biopsies do not. Columns represent tumour samples and rows represent genes (red: up-regulated and green: down-regulated, radiotherapy [blue] or chemoradiotherapy [pink]).

Figure 2

Immunohistochemistry for APRIL in resected colorectal adenocarcinomas. Staining for APRIL was seen in the tumour cells (membrane and cytosol) and stroma (extracellular matrix and stromal cells) of colorectal adenocarcinomas. All combinations of tumour cell and stromal staining were seen. Tumour cell staining could be scored weak, moderate and strong. Examples show strong tumour cell staining and stromal staining.

Figure 3

APRIL protein expression in tumour stroma and survival of colorectal cancer patients. (a). Kaplan-Meier survival plots for tumour stroma APRIL protein expression analysed by immunohistochemistry of 234 colorectal cancer patients following surgical resection.(b) Stromal staining for APRIL in Stage III patients following surgical resection (n = 102) (c) Combined analysis of stage III patients (n = 102) stratified according to adjuvant therapy and tumour stroma APRIL protein. P value is log rank test.