Efficacy Of Biomarkers For Screening Of Gastric Cancer And Premalignant Lesions In High-risk Area: A Cross-sectional Population-based Cohort Study

OBJECTIVE A cross-sectional study was conducted based on a general population in Wuwei,Gansu province,China,gastric cancer high incidence area.In the first stage,we have carried out epidemiological investigation,gastroscop,and histopathological examination for gastric cancer(GC)and premalignant lesions(PML),based on which,we have further carried out biomarker's tests.We aimed to analyze the association of biomarkers with GC and PML,and construct a scoring system for preliminary screening of high-risk population of GC and provide a scientific basis for improving the screening model of GC.METHODSU sing cluster sampling method,“Wuwei Cohort”was constructed comprising total25,000 subjects from March 2013 to April 2016,subjects aged 35 to 70 years in 9towns and villages in Wuwei municipality,Gansu province,where GC is most prevalent.Of the remaining 23,346 participants,2,001 either declined participation or were ineligible for gastroscopy,and 21,345 underwent gastroscopy.We randomly selected 9,396 participants to undergo serological testing for anti-Helicobacter pylori immunoglobulin G(Hp-IgG),pepsinogen(PG)I,PGII and high-sensitivity C-reactive protein(hs-CRP)using an automated biochemical analyzer.Among them,4,069subjects with pathology reports of gastric biopsy were extracted for TFFs,and 83subjects with a medical history of malignancy were excluded.Therefore,3,986subjects who completed the survey and had complete baseline data were evaluated in the final analysis.Since there was no loss of glands in the cascade stage of non-atrophic gastritis,the patients with non-atrophic gastritis were combined with the normal patients served as control group(CG).The participants were classified into five groups based on Correa's cascade of GC development:CG(n=773);CAG(n=746),IM with CAG(n=1,002),LGD accompanied by CAG and/or IM(n=1,334),and high-grade intraepithelial neoplasia/dysplasia(HGD)or GC(n=131).The PML groups include CAG,IM,and LGD groups.HGD was the defined as equivalent to carcinoma in situ.The association of GC risk biomarkers with the progression of GC,and the prediction capacities of these biomarkers in high-risk population of GC and PML were analyzed and identified by using odds ratio-adjusted models and receiver operating characteristic(ROC)curve analyses.the participant characteristics were presented as mean±standard deviation(SD),median(25^th–75^th percentiles[interquartile ranges]),or numbers and percentages as appropriate.The groups were compared using parametric or nonparametric test for continuous variables and a chi-square test for categorical variables.A two-sided p value of p less than 0.05 was considered statistically significant.RESULTS 1.Among 3,986 subjects for the final analysis,the average age of the participants was 51.8 years,and 54.7%of the participants were males,91.1%of the participants were farmers.In total,65.4%of the participants had H.pylori infection as confirmed by serum Hp-IgG testing.Risk factors for GC and PML include age,gender,education level,occupation,income,body mass index,smoking,eating of fried food,H.pylori infection,and the medical history of gastritis,peptic ulcer,polyps(P<0.05).2.The serum levels of TFF1 and TFF2 showed an incremental trend from the CG CAG,IM,and LGD groups to the GC group(P<0.001).However,the distribution of serum TFF3 level differs across the five groups,a significant up regulation can be observed in the IM group.With the intensification of gastric mucosal lesions,the serum level of Hp-IgG gradually increases,the highest Hp-IgG is seen in the LGD group,but it significantly reduces in the GC group(P<0.001).Different PGI circulation levels are observed in each group(P<0.010),and PGI/II ratio change shows a decremental trend from the groups CG to GC,while PGII serum level change is shown in a reversed trend(P<0.001).There is no statistically significant relevance between serum hs-CRP and the histological grades(P=0.84).3.In three odds ratio(OR)-adjusted models of CAG,IM,LGD and GC groups,TFF1,TFF2 and TFF3 had a better association with GC and PML due to their high ORs and stable ORs of TFFs after adjusting for risk factors and other serum biomarkers.ROC curve analysis of these biomarkers also showed that TFF1–3combination had the largest area under the ROC curve in the groups of GC and PML,showing a high possibility of prediction in the high-risk population.4.In the screening scoring model for high-risk populations of gastric cancer,the higher of the serum lever of TFF1,TFF2 and TFF3,the higher the risk of GC.Males have a higher risk of GC than females.The total score of the system establishment is14 points,with 8 points as the optimized cut-off value,its sensitivity and specificity are the best.CONCLUSION This study demonstrates for the first time that TFF1,TFF2 and TFF3 were significant association with GC and PML compared with pepsinogen and Hp-IgG.TFF family were more robust and independent in assessing the risk of GC and PML in the population-based cohort study.Through the establishment of a preliminary screening model for the high-risk population of GC,it was found that TFF1,TFF2,TFF3 and male are statistically significant in the scoring system.TFF family may have the potential non-invasive biomarkers for GC and PML screening in a large-scale general population,as well as for opportunistic screening in high-risk areas of GC.Serum detection of TFFs not only convenient for the early identification of GC and PML,but also provides a preliminary screening population for gastroscopy,and provides a scientific basis for improving the screening of GC.