Objective Description of the design and preliminary results of baseline recruitment and screening in the endoscopic screening for esophageal cancer in China (ESECC), the first randomised controlled trial (RCT) assessing efficacy and cost-effectiveness of endoscopic screening for esophageal squamous cell carcinoma (ESCC).
Design ESECC trial is a cluster RCT, and 668 villages in rural Hua County, Henan Province, a high-incidence area of ESCC in China, were randomised into two arms at a ratio of 1:1. Screening arm participants were screened by Lugol chromoendoscopy; no screening was performed in the control arm. ESCC-specific and all-cause mortality, incidence of advanced ESCC and cost-effectiveness of screening will be evaluated in the next 10-year follow-up. Here, we report the performance of baseline recruitment and randomisation, prevalence of upper GI lesions and risk factors for ESCC.
Results A total of 17 151 and 16 797 participants were enrolled in screening and control arms from January 2012 to September 2016. The truncated prevalence (aged 45–69 years) of oesophageal and overall upper GI high-grade lesions was 744.0/100 000 and 902.0/100 000. 69.9% of the 113 patients with high-grade oesophageal lesions were of early stage. Risk factors for severe oesophageal dysplasia and more severe lesions in this population included higher age, family history of ESCC, lower body mass index, eating rapidly and frequent ingestion of leftovers.
Conclusion This ESECC trial met the predesigned recruitment and randomisation requirements. Age, family history, undernutrition and unhealthy dietary habits increased the risk for high-grade oesophageal lesions in this high-risk population.
Trail registration number NCT01688908; Pre-results.
- esophageal squamous cell carcinoma
- population-based screening
- randomized control trial
Statistics from Altmetric.com
Significance of this study
What is already known on this subject?
Oesophageal cancer is one of the most common cancers worldwide. Nearly half of new cases in the world occur in China and esophageal squamous cell carcinoma (ESCC) is the predominant histological type.
The main risk factors of ESCC remain unclear. Early detection and treatment may have potential to reduce ESCC incidence and mortality.
The efficacy and cost-effectiveness of endoscopy screening for ESCC has not been evaluated with randomised controlled trials.
What are the new findings?
The first population-based randomised controlled trial evaluating the efficacy and cost-effectiveness of endoscopy screening for ESCC was initiated in January 2012 in Hua County, Henan Province, which is a high-risk region in China, and the recruitment and baseline screening were completed by September 2016.
A total of 33 948 individuals from 668 villages were enrolled, and 15 188 in the screening arm received screening endoscopy with pathologic diagnosis. One hundred thirteen oesophageal high-grade oesophageal lesions (severe dysplasia, carcinoma in situ and ESCC) were detected in baseline screening and early stage lesions (severe dysplasia and carcinoma in situ) accounted for 69.9%.
Risk factors associated with severe oesophageal dysplasia and above lesions in this population include age, family history of ESCC, low body mass index, eating rapidly and frequent ingestion of leftovers.
Significance of this study
How might it impact on clinical practice in the foreseeable future?
The results of this trial will provide direct evidence on the effect of endoscopy screening, and on the most cost-effective screening strategy for use in national screening programmes.
Esophageal cancer (EC) is one of the most common cancers in the world, and is a leading cause of cancer death, with >450 000 new cases and 400 000 deaths in 2012.1 Nearly half of the new cases of EC worldwide are found in China, and esophageal squamous cell carcinoma (ESCC) is the predominant histological type. A series of aetiologic factors for ESCC have been proposed in previous studies conducted in high prevalence regions of China, including age,2 family history of ESCC,3 poor nutritional status,4 smoking and drinking,5 chemical carcinogen exposure6 7 and human papillomavirus infection.8 9 But the main aetiologic factors have not been definitively identified.
Because there are typically no symptoms in the early stages of ESCC, the vast majority of cases are clinically diagnosed at an advanced stage. The overall 5-year survival is about 20% in China,10 and is even lower in low-income countries. However, if the disease is found at an early stage, the 5-year survival may be 80% or greater.11 12 This profound improvement in survival indicates there is clearly a need for effective early detection strategies to enable earlier diagnosis and curative treatment. Esophageal squamous dysplasia (ESD) is considered to be the premalignant precursor lesion for ESCC, and harbours a high risk for progression into invasive cancer.13 ESD as well as early stage malignant lesions (carcinoma in situ (CIS)) are therefore screening targets for ESCC. Various screening methods have been tested, and endoscopy with iodine staining is the gold standard technique for the diagnosis of ESCC and its precursor lesions.14 15 Endoscopic screening has therefore been widely accepted as an optimal strategy in the secondary prevention of ESCC. However, taking the high cost and invasive nature of endoscopy into consideration, the efficacy and cost-effectiveness of such screening, must be evaluated prior to introduction of a population-wide screening programme. Evidence regarding the efficacy of endoscopic screening for ESCC has been based predominantly on observational studies in high-risk regions.11 16 There has been only one non-randomised controlled trial, which reported that endoscopy plays a positive role in reducing mortality.17 No randomised controlled trials (RCTs) evaluating endoscopic screening have been reported to date. Observational studies and trials which are not randomised are limited in determining the true benefits of screening on reducing mortality due to potential lead-time bias, length-time bias and confounding bias. Hence, one-step large-scale population-based RCTs are needed to determine the efficacy of endoscopic screening for ESCC.12 18 19
In January 2012, we initiated the endoscopic screening for esophageal cancer in China (ESECC) trial in rural Hua County of Henan Province, which is a high-risk region in northern central China. This is the first population-based RCT aiming to evaluate the efficacy and cost-effectiveness of endoscopic screening for ESCC worldwide.
Here, we report the trial design and recruitment performance in this ESECC trial, as well as the preliminary findings from this baseline endoscopic screening.
Setting and participants
This ESECC trial was undertaken in rural Hua County, which is an agricultural region in the northern part of Henan Province, Peoples Republic of China with a rural population of 1.1 million. The mortality for ESCC in this area is among the highest in the world.20
Participants were eligible for the study if they meet the following criteria: (1) permanent residency in a target village; (2) age 45–69 years (>5 years of life expectancy) and no history of endoscopic examination within 5 years prior to the initial interview; (3) no history of cancer or mental disorder; (4) negative for HBV, HCV and HIV; (5) voluntary participation and agreement to complete all phases of the examination.
This ESECC trial was designed as a cluster RCT. There are a total of 968 villages in rural Hua County, and 668 target villages were randomly selected from the 846 villages with population sizes ranging from 500 to 3000 in Hua County. These 668 target villages were randomly allocated into the screening arm of the study or the control arm at a ratio of 1:1 (334 villages in each arm), using a blocked randomisation procedure (block size=2) based on the total population size of each village for balancing the sample sizes between the two study arms.
According to the New Rural Cooperative Medical Scheme (NCMS) registration system of Hua County, the combined incidence rate of advanced EC and cancer of the gastric cardia in the targeted population (aged 45–69 years) was estimated to be 184.07/100 000 in 2011. Calculation of required sample size was based on the following assumptions: the average period of progression from severe dysplasia to EC is 5 years, and the accrual time is 5 years; the study period was set at 10 years, and 5% of the participants would be lost to follow-up per year; 5% of the participants in the control arm would seek endoscopic examination independently during the study period. Finally, with a total of 32 337 participants enrolled (1:1 between arms, ~20% of all eligible residents in target villages), statistical power of 86% at a one-sided 0.025 significance level can be achieved, even if only 30% of advanced EC cases were protected by screening. The number of participants enrolled in each target village was determined by the weight of the population size of the very village in the total population of the whole arm.
An informed consent was obtained from each study participant. Basic information including name, gender, date of birth, address and phone number was then collected and managed using a custom-designed database system.21 All participants received a physical examination which included measurement of height, weight and blood pressure. Blood samples were collected to screen for HBV, HCV and HIV, and participants with any of these infections were excluded. A computer-aided one-on-one questionnaire investigation was conducted for all participants by trained interviewers to collect data on potential risk factors for EC. 3-level version of European Quality of Life 5-Dimension (EQ-5D-3L), a standardised generic instrument, was used in this ESECC trial to assess health-related quality of life (HRQOL).22 Standardised explanations regarding the items in the questionnaire were provided if the participant was unable to understand the questions or respond appropriately.
In the screening arm, standard upper GI endoscopy (UGE) with iodine staining was performed by physicians experienced in endoscopic examination. The entire oesophagus and stomach were visually examined and biopsies were taken from all focal lesions.23 Standard sites in the oesophagus (28 and 33 cm distal to the incisors in the 6 o’clock position) were biopsied if no visually identifiable abnormalities were found elsewhere. Biopsy specimens were fixed in 10% formaldehyde, embedded in paraffin, sectioned at 5 µm and stained with haematoxylin and eosin. The biopsy slides were reviewed by pathologists at Anyang Cancer Hospital without knowledge of the endoscopic findings. Diagnoses of ESD (mild, moderate and severe), CIS and squamous cell carcinoma were independently confirmed by two pathologists and discrepancies in their histological diagnoses were adjudicated by consultation. To reflect the situation in real population-level screening as much as possible, participants who were diagnosed with severe squamous dysplasia, CIS or squamous cell carcinoma in the oesophagus or malignant lesions in other sites were only informed of the diagnosis and provided with appropriate medical advice, rather than arranging directly for their further clinical treatment.
In the control arm, no endoscopic screening was performed and an abdominal ultrasound scan which had no association with the diagnosis of EC was used.
Follow-up and outcomes
The primary end point of the trial was EC-specific mortality, and the secondary end points included mortality from all causes, incidence of advanced EC and cost per quality-adjusted life-year (QALY) gained. Two sources of follow-up data were used to identify outcome events in this study, namely door-to-door interviews and electronic registry data. A record of vital events, including the experience of UGE, onset of cancer and death, will continue to be collected through annual door-to-door interviews with all cohort members. Data regarding cancer occurrence and death were also collected from the NCMS of Hua County, which is a government-run health insurance programme in rural China with a nearly 100% participation rate which has been proved to be an ideal data source regarding cancer occurrence, diagnosis and treatment,24 and from the Death Registry of National Centers for Disease Control and Prevention, respectively.
In this study, characteristics of participants in the two arms were compared using the Χ2 test and the rank-sum test for category and continuous variables, respectively. Univariate and multivariate unconditional logistic regression was applied in the analysis of risk factors. All variables were first evaluated with univariate logistic regression. Variables with P<0.25 were subjected to a multivariate logistic regression model and the backward-selection method was used to select variables at a significance level of 0.05. Potential risk factors were investigated in parallel under two conditions where ‘severe dysplasia and above’ (including severe dysplasia, CIS and ESCC) and ‘dysplasia and above’ (including all grades of dysplasia, CIS and ESCC) were defined as separate outcome events. All statistical analysis was completed using STATA V.13.1 (STATA, College Station, Texas, USA). All tests were two-sided and had a significance level of 0.05 unless otherwise specified.
Baseline enrolment, randomisation and screening
Recruitment and screening was completed by September 2016. A total of 18 113 subjects in the screening arm and 17 659 in the control arm consented to participate, and of these individuals, 962 and 862, respectively were excluded due to prior history of cancer, history of endoscopy examination within 5 years or HBV/HCV/HIV infection. The remaining 17 151 individuals in the screening arm and 16 797 individuals in the control arm were enrolled in the study. In the screening arm, all candidate participants were invited for UGE screening, among whom 1544 dropped out before the UGE examination and 303 failed to complete the examination. An additional 5 participants from the screening arm and 33 from the control arm did not complete the questionnaire investigation. Thus, the number of participants who completed all phases of examination and investigation in the screening and the control arms were 15 299 and 16 764, respectively. Figure 1 shows the number of participants in both arms and the reasons for exclusion.
A series of characteristic variables and potential risk factors for ESCC were compared in the screening and control arms. Variables which were associated with higher risk for oesophageal precursor lesions in univariate logistic regression analysis as well as crucial indicators of cancer including age, gender, cigarette smoking and alcohol drinking are shown in table 1. The two arms were well balanced in terms of age (P=0.115), gender (P=0.419), cigarette smoking (P=0.135), alcohol drinking (P=0.562) and type of fuel used for cooking (P=0.916). The median age was 57 years in both arms and 48.9% and 48.5% of the participants were male in the screening and control arms, respectively. Compared with the control arm, participants in the screening arm were more likely to report a family history of EC, were more likely to have a lower body mass index (BMI), were more likely to be unmarried, were likely to have more household income, were more likely to be exposed to kitchen fume, were more likely to habitually eat rapidly, were more likely to prefer high temperature food and were more likely to ingest preserved food. In contrast, participants in the control arm were more likely to receive more education, were more likely to have a larger family, were more likely to use a coal stove for heating and were more likely to ingest leftover food.
Prevalence of upper GI lesions (screening group)
Of all eligible participants in the screening arm, 15 188 had at least one biopsy which was technically adequate for pathologic diagnosis. Among these participants, 113 (0.74%) were diagnosed as high-grade oesophageal lesions, including 63 (0.41%) severe dysplasia, 16 (0.11%) CIS and 34 (0.22%) ESCC, and among these lesions, severe dysplasia and CIS, the early stage lesions, accounted for 69.9% (79/113). Another 473 (3.11%) participants were diagnosed with mild dysplasia and 87 (0.57%) had moderate dysplasia. A diagnosis of acanthosis, oesophagitis or basal cell hyperplasia (BCH) was rendered in 14.00%, 14.97% and 18.93% of participants, respectively. These lesions appeared alone or were accompanied by other lesions. High-grade lesions in other sites in the upper GI tract were also found, including 12 (0.08%) from the gastric cardia, 11 (0.07%) from non-cardial gastric mucosa and 1 (0.01%) from the duodenum (table 2). The 45-year to 69-year truncated prevalence of high-grade oesophageal lesions and overall upper GI lesions was as high as 744.0/100 000 and 902.0/100 000, respectively. Prevalence showed a trend of increase with age and there was a striking increase in the 55–59 years age group (figure 2). More high-grade upper GI lesions were found in males, but no statistically significant gender difference was found in ESCC or its precancerous lesions.
Risk factors for oesophageal precursor lesions (screening group)
A series of potential risk factors for oesophageal precursor lesions were evaluated by logistic regression analysis. In the final multivariate model, older age (Ptrend<0.001), a family history of EC (Ptrend<0.001), frequent ingestion of leftover food (OR 1.63, 95% CI 1.13 to 2.37), low BMI (OR 1.64, 95% CI 1.08 to 2.48) and eating rapidly (OR 2.27, 95% CI 1.18 to 4.37) were associated with a higher risk of severe dysplasia or more severe lesions. Older age (Ptrend<0.001), a family history of EC (Ptrend<0.001), use of a coal stove for heating (OR 1.33, 95% CI 1.14 to 1.56) and heavy fume during cooking (OR 1.25, 95% CI 1.03 to 1.52) were found to correlate with dysplasia and above lesions (table 3).
It will be difficult to make significant breakthroughs in primary prevention of ESCC until critical risk factors are completely understood. Secondary prevention aiming at early detection of precancerous lesions and intervention to prevent progression has therefore been considered to be an appropriate strategy, and has become a focus of ESCC control. Characteristics of ESCC provided great potential for initiation of a screening programme. First, there are clinically significant precancerous mucosal lesions which are detectable during the gradual progression from initiation of tumourigenesis to more advanced stages. Second, these precancerous lesions can be eliminated by therapeutic methods such as endoscopic mucosal resection and ablation procedures. These methods typically involve relatively low trauma and are relatively inexpensive. Moreover, endoscopy with iodine staining has sufficient sensitivity and specificity for detecting EC and precursor lesions.15 However, endoscopy is expensive and may cause severe side effects such as perforation, cardiopulmonary events, iodine allergy, bleeding and potential psychological harm.25 Without evidence from large population-based RCTs, endoscopic screening cannot be recommended for general clinical practice. In western countries with a relatively low prevalence of EC, endoscopic screening is not recommended because the effectiveness of this method remains controversial.26 However, nearly half of the incident cases worldwide occur in China, and ESCC imposes a significant burden on the healthcare system, especially in rural areas of high ESCC prevalence. Population-level screening programme have been initiated in parts of high-risk areas without evaluation of the effect of screening by RCTs. In this context, we launched the first population-based RCT worldwide to evaluate the efficacy and cost-effectiveness of endoscopic screening on EC in 2012. This study will go on for 10 years, and will provide direct evidence of the capacity of endoscopic screening for changing the natural history of ESCC and providing benefit to patients with early screening. This study will also ultimately provide recommendations about the most cost-effective means of screening.
Although age, gender, cigarette smoking and alcohol drinking were well balanced in the two arms of this study, we found that some behavioural and exposure factors had statistically significant differences at the baseline in these two groups. Possible reasons for these differences are as follows. The ESECC trial was conducted in a less developed area of north central China and cluster randomisation was adopted since individual level randomisation would not be practical in this population. Compared with individual randomisation, cluster randomisation and self-preference of study participants due to open-label recruitment might contribute to the differences in the two arms of the study. In addition, some of the significant differences between the groups may be identified simply because of the large sample size. For example, the difference between groups in median BMI appears trivial (24.9 vs 25.4), but nevertheless achieved a P value of <0.001. In fact, an obvious absolute difference was detected only in the variable fume exposure in kitchen. Although differences existed between the study arms at baseline for some risk factors, it seemed that the screening arm had higher risk for development of ESCC and precursor lesions, which would bias the screening effect towards the null. These potential confounders will be further evaluated and adjusted in multivariate models in future analysis.
Baseline screening demonstrated that EC is the most common upper GI cancer in this region of China, and the truncated prevalence of high-grade oesophageal lesions is estimated to be 744.0/100 000, which is consistent with our previous study.23 Early stage lesions accounted for 69.9% and 66.4% of oesophageal high-grade lesions and overall upper GI high-grade lesions, respectively, which confirmed the ability of endoscopy for early detection of upper GI cancers. Both the prevalence of high-grade oesophageal lesions and upper GI high-grade lesions overall showed a trend of increase with age and a remarkable increase in the 55–59 years age group. No ESCC and only two severe dysplasias were found in <50-year age group. This suggests that a tailored screening strategy may be established by setting the starting age for screening higher, for example at 50 years, to improve the efficiency of screening if resources are limited. Although studies based on cancer registries often report a higher incidence of ESCC in males,27 gender difference in detection of ESCC and its precursor lesions was not statistically significant in this study. This inconsistency in results from screening programmes and registries calls for further investigation, but raises the possibility that precursor lesions in females have a lower probability of progressing to advanced stage lesions. We evaluated the association between a series of potential risk factors and the onset of ESCC and precursor lesions in this study. In keeping with previous studies, we identified several factors associated with the high risk of oesophageal lesions in this population, including age,2 family history of ESCC,2 3 low BMI2 4 and chemical carcinogen exposure.6 7 A risk prediction model can be constructed based on these characteristics to identify individuals at high risk in the general population. An accurate and cost-effective prediction model based on this large-scale population-based screening study will have potential for application in real ESCC prevention by making screening more efficient, especially in rural areas of limited resources.28 People in high prevalence areas of China often have relatively low socioeconomic status, and low awareness of their health with limited ability to seek medical advice. Therefore, in addition to assorting individuals at high risk prior to screening, personalised assistance and arrangement for subsequent re-examination and further medical services among individuals who already have high-grade lesions found at screening can also help improve the cost-effectiveness of a screening project.
Like other cancers, EC is a rare disease even in Hua County, which is an extremely high-risk area for ESCC worldwide, and only 113 patients with high-grade lesions in oesophagus were identified in the entire screening group. The cost-effectiveness of the screening can be carefully evaluated by comparing the extra cost per QALY gained from the screening to the willingness-to-pay in the future. At the same time, intensive surveillance and individualised management of the less severe lesions, for instance, moderate dysplasia (0.57%) and mild dysplasia (3.11%) may have the potential to dramatically increase the effectiveness of screening with relatively low cost. According to the consensus of clinical specialists, patients with mild dysplasia should have endoscopic re-examination once every 3 years, and patients with moderate dysplasia should be re-examined once a year.29 30 This standard protocol has been initiated in our ESECC trial and results from the first round of re-examination are to be expected in the near future.
In summary, this ESECC trial successfully recruited over 33 000 participants in a rural area of China with high prevalence of ESCC, and over 15 000 standard endoscopies were performed in the past 5 years to detect early oesophageal lesions. An early detection rate of ~70% was achieved in the baseline screening. This ESECC trial is expected to provide highest-quality evidence regarding the efficacy and cost-effectiveness of population-level endoscopic screening for EC.
The authors thank all the following team members and collaborators for their contributions to the field work done for this study, endoscopic examinations and pathologic diagnosis (in alphabetical order by last name and first name): Changqi Cao, Qiuju Deng, Dong Hang, Jingjing Li, Shijie Li, Yan Li, Fangfang Liu, Zhihao Lu, Na Shen, Haixing Wang, Hui Wang, Jing Wang, Xicheng Wang, Yan Yan, Wenqing Yuan, Chanyuan Zhang, Xiaotian Zhang and Jun Zhou from Peking University Cancer Hospital & Institute; Kun Wang, Ye Wang and Li Zhang from Peking University Third Hospital; Wanju Gao, Mei Guo, Qianqian Meng, Jun Yang, Liheng Zhang, Lixin Zhang and Sanshen Zhang from Anyang Cancer Hospital, Henan Province; Yujie He, Shaojiang Lu and Xiangqin Song from the People’s Hospital of Hua County, Henan Province; Xin Yang and Weiguo Xu from the North China University of Science and Technology Affiliated Hospital, Hebei Province; Zengchao Chen from Shandong Qianfoshan Hospital, Shandong Province. The authors would also like to thank the Government of Anyang City and Hua County, the Health and Family Planning Commission of Anyang City and Hua County, Henan Province and all the participants in the ESECC trial.
ZH, ZL and ML contributed equally.
Contributors Study concept and design: YK, HC and ZH; acquisition of data: ZH, ZL, ML, CG, RX, FL, AL, HY, LS, QW, LD, XL, CZ and YP; analysis and interpretation of data: ZL, ZH, ML, HC and YK; drafting of the manuscript: ZH, ZL, ML, HC and YK; statistical analysis: ZH, ZL and ML; study supervision: ZH, HC and YK.
Funding This work was supported by the charity project of the national ministry of health (grant number 201202014), the natural science foundation of China (grant number 30930102, 81473033, 81773501), the natural science foundation of Beijing (grant number 7100001), the '973' project of national ministry of science and technology (grant number 2011CB504300), the UMHS-PUHSC joint institute for translational and clinical research (grant number BMU20140483), the national key R&D programme of China (grant number 2016YFC0901404), the science foundation of Peking University Cancer Hospital (grant number 2017-4) and the open project funded by the Key Laboratory of Carcinogenesis and Translational Research, Ministry of Education/Beijing (grant number 2017-10).
Competing interests None declared.
Patient consent Obtained.
Ethics approval This trial was approved by the Institutional Review Board of the Peking University School of Oncology, China.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.