Risk Prediction For The Prevalence And Progression Of Precancerous Lesions Of Esophageal Squamous Cell Carcinoma In A Prospective Cohort Study In High-Risk Areas In China

AimsBased on the early detection and early treatment program in China,this study aims to describe the distribution and progression pattern of esophageal squamous cell carcinoma(ESCC)and precancerous lesions,to establish predictive models to predict the prevalence and progression of esophageal cancer and precancerous lesions,and to screen and validate protein biomarkers to specifically identify esophageal cancer and precancerous lesions.Materials and MethodsBetween 2005 and 2009,subjects aged 40-69 years old were enrolled in the endoscopic screening with iodine staining program in three high-risk areas in China,namely Linzhou in Henan,Cixian in Hebei and Feicheng in Shandong.Positive endoscopic re-examination and negative cancer registry based surveillance were carried out until December 31th,2016.According to the results of initial screening and pathological diagnosis,the detection rates of ESCC and precancerous lesions were described,and the ranges were displayed by 95%confidence interval(95%CI).Unconditional logistic regression model was used to explore the risk factors associated with the prevalence of ESCC and precancerous lesions.ROC curves and AUCs were applied to test the predictive value of the models.The best cutoff of the predictive value was chosen by Youden index.We then calculated the sensitivity,specificity,the proportion of high-risk population of each predictive value and made convenience-check tables for patients to check their risks to have ESCC or precancerous lesions conveniently.The progression was defined as severe dysplasia/carcinoma in situ(SD/CIS)or higher diagnosed in the positive re-examination.The incidence was defined as ESCC occurred in the cancer registry based surveillance.Cumulative progression and incidence rates of precancerous lesions were explored.Kaplan-Meier survival curve and Log-Rank test were used to calculate the progression and ESCC onset time,and to compare the survival time between the treated and untreated SD/CIS patients.Unconditional logistic regression model was applied to explore the risk factors related to the progression of precancerous lesions to ESCC or onset of ESCC.ROC curves and AUCs were applied to test the predictive values of the models.Convenience-check tables were also made for patients to check their risks to progress and ESCC occur.Tandem Mass Tag(TMT)test was carried out to screen the protein biomarkers for ESCC and precancerous lesions.Candidate proteins were chosen according to the following criteria:1)the protein expression ESCC/LGIN>2;2)HGIN/LGIN>1;3)the proteins were ordered by the ESCC/LGIN sequence from high to low.The top 15 proteins were selected.The Parallel Reaction Monitoring(PRM)test was used to validate the candidate proteins internally with the same samples with the TMT test.Positive candidate proteins by PRM were externally validated by immunohistochemistry(IHC)Results1)21,652 subjects were recruited in three high-risk areas from 2005 to 2009.5,329(24.61%)were detected as precancerous lesions.Among them,1,714 cases were basal cell hyperplasia(BCH,7.92%,95%CI 7.56%-8.28%);2S422 cases were mild dysplasia(mD,11.19%,95%CI 10.77%-11.61%);727 cases were moderate dysplasia(MD,3.36%,95%CI 3.13%-3.61%);332 cases were SD/CIS(1.53%,95%CI 1.38%-1.71%);134 cases were ESCC(0,620,95%CI 0.52%-0.73%).Multivariate Logistic regression analysis showed that,using SD/CIS or above as outcome variables:gender(male,OR 1.43,95%CI 1.11-1.72),age(50-59 years,OR 4.02,95%CI 3.04-5.32;60-69 years old,OR 7.30,95%CI 5.46-9.78)and family history of cancer(yes,OR 1.33,95%CI 1.10-1.61)were independent risk factors for SD/CIS or above.AUC was 0.704 for this model.The sensitivity was 72.10%and the specificity was 58.43%.The subjects at high risk for SD/CIS and needed to be screened was 43.08%of the whole population.2)Between 2005 and 2016,3,114 subjects were enrolled in the endoscopic re-examination and 21,111 were recruited in the cancer registry based surveillance.The cumulative progression and incidence rates of precancerous lesions increased with grades.After 11 years of follow-up,less than 5%of the normal and BCH cases progressed.More than 5%of mD,MD and SD/CIS progressed after 8 years,1 year and 2 years,respectively.Less than 5%of the normal,BCH,mD and MD cases became cancer during the follow-up time and more than 5%of SD/CIS became cancer after 4 years.Multivariate logistic regression analysis with progression as outcome variables,showed that:(1)age(50-59 years,OR 4.37,95%CI 1.28-14.95;60-69 years,OR 6.31,95%CI 1.60-24.80)and the number of family members(>3,OR 2.33,95%CI 1.04-5.24)are independent risk factors for the progression of mD;(2)gender(male,OR 2.99,95%CI 1.03-8.65),water sources(well water,OR 5.01,95%CI 1.43-17.60)and alcohol(ever,OR 3.10,95%CI 1.15-8.37)were independent risk factors for the progression of MD.The AUC,sensitivity,specificity,and percentage of high-risk population for mD and MD model were 0.674 and 0.743,80.65%and 69.57%,50.69%and 70.18%,50.08%and 32.80%,respectively.High risk populations in the two predictive models were different.3)The cumulative incidence of patients in the endoscopic treatment group was significantly lower than that in the untreated group(3.20%(95%CI 1.44-6.97)vs.10.65%(95%CI 6.90-16.10)p=0.006).The ESCC-free survival time in treated group was significantly higher than that in the untreated group(p=0.043).4)5,577 proteins were identified in the TMT test.15 specific candidate protein biomarkers capable of distinguishing ESCC and precancerous lesions were screened,and then internally validated by PRM test,Nine of them were detected(UCHL1,PFN2,TNC,ZNF428,THY1,PRG3,IFIT3,THBS1 and DEFA1),which were highly expressed in ESCC tissues specifically.External validation by IHC showed UCHL1 specifically highly expressed in ESCC tissues.ConclusionsThe quality control recommendation for ESCC screening in China was as follows,BCH 7.92%(95%CI 7.56%-8.28%),mD 1.19%(95%CI 10.77%-11.61%),MD 3.36%(95%CI 3.13%-3.61%),SD/CIS 1.53%(95%CI 1.38%-1.71%),ESCC 0.62%(95%CI 0.52%-0.73%).We established predictive models for prevalence and progression risks.The subjects were suggested be assessed for the risk before endoscopic screening.Men,the elder and residents with family history of cancer were prior to be screened.Men,the elder,drinkers and residents with high number of family members were re-examined with high priority.Normal and BCH were suggested endoscopically screened for one time without re-examination.The follow-up intervals for mD and MD were proposed to be extended to 8 years and 5 to 6 years,respectively.The SD/CIS patients were recommended to receive endoscopic treatment immediately as before,but high-risk populations were supposed to be further selected to avoid overtreatment.Specifically expressed protein biomarkers have significant potential application prospect to predict the prevalence and progression of precancerous leisons.UCHL1 was highly expressed in ESCC tissues and the candidate protein biomarkers are to be further validated in the prospective studies.