Construction Of A Hematological Index-based Scoring System For The Screening Of Gastric Precancerous Lesions

Objectives:The aim of this study is to explore the independent risk factors for gastric precancerous lesions,so as to construct a relevant prediction model and a screening scoring system,and to verify its application value.Methods:We included 2376 subjects who participated in the early detection and treatment screening project for upper gastrointestinal cancer in Yangzhou area from March 2019 to June 2021;the subjects were divided into two cohorts,training cohort(1664 cases)and validation cohort(712 cases)in a ratio of seven to three.We collected the subject demographics(gender,age,and body mass index),previous history(smoking history,drinking history,tea-consuming history,family history of cancer,gastropathy history and H.pylori infection history)and daily diet preference;the blood routine,gastric mucosal serology test(PGR and G-17),14C-urea breath test,gastroduodenoscopy and gastric mucosal biopsy and pathology were performed.According to the severity of the gastric mucosal lesion,the training cohort was grouped,and the comparison between groups were made to analyze the correlation between each parameter and the gastric precancerous lesion and screen out the predictors of gastric precancerous lesion.According to the Framinghan Heart Study methods,a gastric precancerous lesion screening scoring system was established.The discrimination and calibration degree of model was evaluated,and internal validation was performed in the training cohort.Comparison between the gastric precancerous lesion screening scoring system and Li’s new gastric cancer screening scoring system was made to confirm the advantage and value of gastric precancerous lesion screening scoring system.Statistical methods:according to the data type,the comparison between groups was made with independent sample t test,univariate analysis of variance,Mann-Whitney U test,KruskalWallis test and χ2 test;the independent risk factors of gastric precancerous lesion were tested by multivariate logistic regression analysis;predictive efficacy and optimal cut-off value of each parameter,model and scoring system were evaluated by ROC curve;the calibration of model was evaluated by Hosmer-Lemeshow test;agreement assessment based on kappa coefficients was made.All statistical tests were bilateral,when P<0.05,the difference was considered statistically significant.Results:1.The comparison between the training cohort and the validation cohort was made and there was no statistically significant difference between two groups.2.The gastric precancerous lesion group was classified into three groups,atrophic gastritis group,intestinal metaplasia group and atypical hyperplasia group and a comparison was made between the three groups.There was statistical difference in the gender,age,smoking history,drinking history,frequency of fried and fire-cured food intake,gastropathy history,H.pylori infection history,PGR,red blood cell,hemoglobin and mean corpuscular volume among three groups.The comparison between groups showed that the proportion of female patients was lower and the proportion of smoking and drinking patients was higher in the intestinal metaplasia group and atypical hyperplasia group than that in the atrophic gastritis group;the age of the atrophic gastritis group was younger than that of intestinal metaplasia group;the number of patients who preferred fried and fire-cured food in the atypical hyperplasia group was more than that in the atrophic gastritis group and the number of patients with previous gastropathy was more than that in the other two groups;the patients with previous H.pylori infection were observed more commonly in the atrophic gastritis group and the red cell,hemoglobin and PGR of the group was also lower than that of the other two groups.3.Comparison between the gastric precancerous lesion group and the control group was made,the proportion of male patients,the number of patients with the smoking drinking history and the previous H.pylori,G-17,white blood cell,monocyte count,monocyte percentage,standard deviation of red blood cell distribution width,monocyte-to-lymphocyte ratio and panimmune-inflammation values of gastric precancerous lesion group were significantly higher than these of control group and the number of the patients with gastropathy history and the PGR of gastric precancerous lesion group were significantly lower than these of control group.4.ROC curve was drawn for the continuous predictors of gastric precancerous lesion with greater significance(P<0.01),which was showed by univariate analysis and the optimal cutoff values of G-17,PGR,monocyte count,monocyte percentage,standard deviation of red blood cell distribution width and pan-immune-inflammation values were respectively 2.91pmol/L,9.81,0.38×109/L,6.15%,45.55fL and 38.51.5.G-17,PGR,monocyte count,monocyte percentage,standard deviation of red blood cell distribution width,pan-immune-inflammation values,gender,smoking history and gastropathy history were considered into logistic regression model analysis and the results suggested that previous H.pylori infection,RDW-SD≥45.55fL,PGR≤9.81,G-17≥2.91pmol/L and PIV≥38.51 were the independent risk factors of gastric precancerous lesion.The risk prediction model constructed based on the regression analysis results had a better discrimination,which was suggested by the result that AUC of the model was 0.754;HosmerLemeshow goodness of fit test indicated that the calibration of the model was higher.6.Gastric precancerous lesion screening scoring system was established including five items,H.pylori infection,RDW-SD,PGR,G-17 and PIV,and the range of total scores were 0 to 16 points;the 8 points was the cut-off value and according to the cut-off value,the target population receiving screening was divided into two groups,high-risk population of gastric precancerous lesion(8～16 points)and low-risk population of gastric precancerous lesion(0～7 points).7.Gastric precancerous lesion screening scoring system was applied in the validation cohort and the AUC=0.739.The comparison between the gastric precancerous lesion screening scoring system and Li’s new gastric cancer screening scoring system was made,the sensitivity,specificity,positive predictive value,negative predictive value and accuracy of the former were no less than that of the latter.Conclusions:Previous H.pylori infection,RDW-SD,PGR,G-17,and PIV were independent risk factors of gastric precancerous lesion.According to the above factors,gastric precancerous lesion screening scoring system was established,and the system has a greater clinical value and can be applied to risk stratification of gastric precancerous lesion before large-scale endoscopic screening to enhance the efficiency and accuracy of the screening.