Use of Blood Biomarkers to Predict Gastric Cancer Risk

Objectives:

This study aims to

1. Validate the predictive value of various blood biomarkers, including the panel of antibodies against H.pylori and pepsinogen levels, micro RNAs (miRNAs) and blood-based protein markers in a prospective cohort study involving 8000 participants who have been referred for upper for predicting gastric cancer risk.
2. Record the number of participants who are diagnosed with gastric cancer, including high grade dysplasia, carcinoma in-situ and adenocarcinoma.
3. Develop a cost-effective screening strategy based on 1 and 2.

Study Design This is a prospective cohort study involving 8000 subjects who are referred for upper GI endoscopy for standard clinical indications. Blood will be drawn from subjects for blood tests, including antibodies against H. pylori, pepsinogen levels, specific miRNA levels, blood-based protein markers, before undergoing an upper GI endoscopy for clinical reasons. Subjects will be monitored for a period of 10 years via verification with the National Registry of Diseases Office (NRDO) of Singapore. The correlation between the blood test results and findings from the clinical endoscopy or NRDO database verification will be analyzed to determine the predictive value.

Enrolment Patients undergoing upper gastrointestinal endoscopy will be considered for enrolment provided they meet the inclusion and exclusion criteria. 8000 subjects will be enrolled. The Patient Informed Consent Form will be discussed with the patient. If acceptable, the form will be signed by the patient, Principal Investigator and witness if applicable. The subject will be assigned initials and a number in the order in which they were enrolled. Subjects will also be asked to provide information on demographics, family history of gastric cancer, diet, medical history, and lifestyle habits. The data will be captured in a comprehensive database.

Blood Collection 15 mls of blood will be drawn from each subject. Serum, plasma and white blood cells will be extracted.

Analyses

1. Sera will be assayed for antibodies against H. pylori, Pepsinogen I and II levels to derive H. pylori status and pepsinogen index. Participants will be classified into 4 groups, A, B, C and D, according to H. pylori status and pepsinogen Index. Group A includes subjects negative for both H. pylori antibodies and pepsinogen atrophic gastritis (PGA). Group B subjects will be negative for PGA and positive for H.pylori antibodies. Group C subjects will be positive for both tests. Finally, Group D subjects will be positive for PGA and negative for H. pylori antibodies. The incidence rate and relative risk of gastric cancer as determined by the findings of subjects' clinical gastroscopies or verification with NRDO will be calculated.
2. Total RNA from serum or plasma will be isolated and the expression levels of miRNAs will be determined. The final miRNA expression levels after normalization of both technical and biological variations are analyzed with multiple statistical methods to identify panels of miRNAs with the highest discriminatory power between healthy and disease states. Each participant will be assigned a score based on miRNA expression profile, indicating the possibility of having GC. The result will be compared with endoscope and histology result which is the gold standard for diagnosis of GC. We will compare the performance of our miRNA panel with other serum markers including H. pylori status and pepsinogen levels.
3. Levels of 10 blood-based protein markers (determined from systematic review) will be analyzed from sera through ELISAs (enzyme-linked immunosorbent assay). Subsequent statistical multivariate analysis may be applied to generate a biomarker panel with a superior diagnostic performance compared to the conventional blood-based biomarkers for gastric cancer.

Sample Size Calculation In our previous case-control study, 10% of the control group had pepsinogen (PG) determined atrophic gastritis, and the odds ratio (95%CI) for the risk of developing gastric cancer was 4.02 (2.56-6.30) compared with subjects who had no PG atrophic gastritis. A minimum of 47 gastric cancer cases are required to evaluate the PG-H. Pylori panel in the prediction of gastric cancer risk (with odds ratio 3.5) at power of 90% and p-value <0.05. Based on the estimated gastric cancer prevalence rate among NUH patients of the gastroenterology and upper GI surgery clinics of 0.6%, a cohort size of 8000 subjects will be needed.

Statistical Analysis The chi-square test and the t-test will be used to compare the distributions of selected demographic, lifestyle and other risk factors between cases and controls. The logistic regression method will be used to examine the associations between blood markers measured and risk of gastric cancer. Statistical computing will be carried out using the SPSS 19. The statistical significance level is set at two-sided P value of 0.05.