Abstract
Background:
Identification of predictive markers of response to treatment is a major objective in breast cancer. A major problem in clinical sampling is the variability of RNA templates, requiring accurate management of tumour material and subsequent analyses for future translation in clinical practice. Our aim was to establish the feasibility and reliability of high throughput RNA analysis in a prospective trial.
Methods:
This study was conducted on RNA from initial biopsies, in a prospective trial of neoadjuvant chemotherapy in 327 patients with inoperable breast cancer. Four independent centres included patients and samples. Human U133 GeneChips plus 2.0 arrays for transcriptome analysis and quantitative RT-qPCR of 45 target genes and 6 reference genes were analysed on total RNA.
Results:
Thirty seven samples were excluded because i) they contained less than 30% malignant cells, or ii) they provided RNA Integrity Number (RIN) of poor quality. Among the 290 remaining cases, taking into account strict quality control criteria initially defined to ensure good quality of sampling, 78% and 82% samples were eligible for transcriptome and RT-qPCR analyses, respectively. For RT-qPCR, efficiency was corrected by using standard curves for each gene and each plate. It was greater than 90% for all genes. Clustering analysis highlighted relevant breast cancer phenotypes for both techniques (ER+, PR+, HER2+, triple negative). Interestingly, clustering on trancriptome data also demonstrated a "centre effect", probably due to the sampling or extraction methods used in on of the centres. Conversely, the calibration of RT-qPCR analysis led to the centre effect withdrawing, allowing multicentre analysis of gene transcripts with high accuracy.
Conclusions:
Our data showed that strict quality criteria for RNA integrity assessment and well calibrated and standardized RT-qPCR allows multicentre analysis of genes transcripts with high accuracy in the clinical context. More stringent criteria are needed for transcriptome analysis for clinical applications.