Pre-treatment plasma proteomic markers associated with survival in oesophageal cancer

Abstract

Background: The incidence of oesophageal adenocarcinoma is increasing worldwide but survival remains poor. Neoadjuvant chemotherapy can improve survival, but prognostic and predictive biomarkers are required. This study built upon preclinical approaches to identify prognostic plasma proteomic markers in oesophageal cancer.

Methods: Plasma samples collected before and during the treatment of oesophageal cancer and non-cancer controls were analysed by surface-enhanced laser desorption/ionisation time-of-flight (SELDI-TOF) mass spectroscopy (MS). Protein peaks were identified by MS in tryptic digests of purified fractions. Associations between peak intensities obtained in the spectra and clinical endpoints (survival, disease-free survival) were tested by univariate (Fisher's exact test) and multivariate analysis (binary logistic regression).

Results: Plasma protein peaks were identified that differed significantly (P<0.05, ANOVA) between the oesophageal cancer and control groups at baseline. Three peaks, confirmed as apolipoprotein A-I, serum amyloid A and transthyretin, in baseline (pre-treatment) samples were associated by univariate and multivariate analysis with disease-free survival and overall survival.

Conclusion: Plasma proteins can be detected prior to treatment for oesophageal cancer that are associated with outcome and merit testing as prognostic and predictive markers of response to guide chemotherapy in oesophageal cancer.

Figure 1

CONSORT (Schulz et al, 2010) diagram for the oesophageal cancer clinical study protocol. CT, computed tomography; US, ultrasound.

Figure 2

Peak intensity data at baseline for (A) oesophageal cancer patients vs case controls and (B) according to survival. Sample group statistics (mean and s.d.) obtained for (A) plasma peaks (m/z) differing significantly (P<0.05, appropriate parametric or non-parametric test) between case controls (normal, n=36) and oesophageal cancer (tumour, n=21) patients at baseline or (B) according to survival from date of diagnosis, between the greater than 11 months group (good, n=5) and the <7 months group (poor, n=5).

Figure 3

Peak intensity data at baseline according to disease-free survival. Sample group statistics (mean and s.d.) obtained for plasma peaks (m/z) differing significantly (P<0.05 appropriate parametric or non-parametric test), according to disease-free survival from date of diagnosis, between the >11 months group (good, n=4) and the <7 months group (poor, n=6).

Figure 4

Multivariate analysis, associations with survival and disease-free survival. Associations between peaks (m/z) intensities and survival or disease-free survival determined by BLR. Direct lines indicate statistically significant independent associations (P-values shown). Positive (+) and negative (−) associations are depicted.