A prospective study to evaluate the role of gene expression profiles (GEP) in predicting clinical outcome of patients (pts) receiving preoperative chemotherapy for oesophagogastric (OG) cancer
4501 Background: Whilst preoperative chemotherapy has demonstrated survival benefit for pts with potentially resectable OG cancer it is not possible to predict the benefit for an individual pt. This study was designed to prospectively correlate GEP with clinical outcome. Methods: Eligible pts were deemed to have resectable disease after staging CT, EUS, and laparoscopy as indicated & following discussion at the multidisciplinary team meeting. All pts received neoadjuvant platinum & fluoropyrimidine based chemotherapy & clinical data were entered prospectively onto a study specific database. GEP were produced from total RNA isolated from snap frozen pre treatment tumour biopsies obtained at baseline endoscopy. Labelled cDNA was hybridised versus a universal human reference using an in house c DNA array of 22,000 clones. Results: Of the pts with adequate follow up accrued between 2002–2005, 35 met the quality control measures for the arrays. Median age=66 yrs (47–83); male=32, female=3; tumour subsites: oesophagus=23, oesophago-gastric junction (OGJ)=12; adenocarcinoma=35; T stage: T 2=3, T3=30, T4=2; N stage: N0=12, N1=23; performance status 0=7, 1=28. Median follow up=938 days. Median overall survival (OS) = 570 days. Prognostic groups were designated according to the median OS (days) of the group: good > median and poor < median. Supervised hierarchical clustering of normalised data revealed significantly differentially expressed genes based on OS (p<0.01) with 2 distinct clusters: a poor outcome group: N= 17 (2yr OS 17.6%) [95% CI: 4.3–38.3], a good outcome group: N=18 (2 yr OS 55%) [95% CI: 30.5–74.8]. Of the differentially expressed genes, those involved in receptor tyrosine kinase signalling & cell growth were amongst the most significantly affected pathways. Conclusions: This novel technique using GEP in tumour biopsies has successfully identified groups of tumours with distinct gene expression profiles that correlate with survival. The approach warrants further validation in a larger cohort. It could facilitate the development of tailored treatment according to individual tumour biology in OG cancer. No significant financial relationships to disclose.