Diagnostic Performance Indicators in Upper GI Endoscopy:PROSPERO Study

BACKGROUND Esophagogastroduodenoscopy (EGD) is the gold-standard test for investigating upper gastrointestinal (GI) tract pathology. The learning curve for EGD is generally steep and it is often performed by trainees at the mid stages of core training as well as nurse endoscopists. Currently the Joint Advisory Group for GI endoscopy (JAG) recommends a minimum number of 200 procedures to acquire the skills and perform endoscopy independently. However, there has been very little research done to set the standard required to measure EGD performance. The only validated quality indicators in EGD include the successful J-manoeuvre rate and D2 intubation rate, which reflect very basic skills and do not measure diagnostic ability. This contrasts with key performance indicators (KPI) for colonoscopy, for which extensive research has led to development of formal quality performance indicators. In colonoscopy formal quality indicators include (i) cecal intubation rate, (ii) adenoma detection rate, (iii) rectal retroversion rate, (iv) colonoscopy withdrawal time, (v) polyp retrieval rate, (vi) post-colonoscopy colorectal cancer rate and (vii) perforation rate. This has allowed researchers to demonstrate that poor quality colonoscopy leads to an increase in the incidence of interval cancers.

Similar benchmarks have not been set for the upper GI tract. A recent UK study revealed that 7.7% cases of upper GI cancers were missed on previous EGD and that there was a correlation between missed cancer and high number of procedures per lists and hence, arguably, those performed under user time pressure. This is in line with studies on other Western populations indicating that between 5.3% and 13.9% of patients with upper GI cancers have had normal EGD reported within the previous 3 years. Given that gastroscopy is the most commonly performed procedure in endoscopy units in the UK and its demand is increasing, there is an urge to identify diagnostic KPI to measure diagnostic skills, to improve quality of the endoscopies and to reduce the rate of missed diagnoses. Furthermore, such minimum quality standards are extremely important to counteract the potential negative effects on the performance of the pressure faced by endoscopy departments to deliver diagnostic and therapeutic procedures in a timely fashion. The importance of adequate procedure time for diagnostic accuracy has been demonstrated both in the detection of oesophageal and gastric neoplastic lesions. For detection of upper GI neoplasia, it has been demonstrated that a minimum of 7 minutes per procedure should be spent in order to ensure adequate visualisation and allow detection of high-risk lesion . In a Korean study, that compared faster with slower endoscopists, all of whom had adequate training and experience, the slower endoscopists (withdrawal time from duodenum >3min) had a significantly higher detection rate of upper GI neoplasia. Finally, in the context of Barrett's oesophagus (BO), longer inspection times during EGD correlated with increase detection rate.

Pre-malignant lesions in the upper GI tract are generally understudied when compared to those in the lower GI tract. Furthermore, their epidemiology seems to vary with geography, and this has not been inadequately investigated in the UK. Barrett's esophagus (BE) is the only known precursor to esophageal adenocarcinoma. The population prevalence is difficult to estimate given that the condition can be asymptomatic, and the diagnosis is only made by endoscopy and biopsy. It is estimated that the prevalence in the general population in Western Countries is 1-2% with significant geographic variation, but increases in patients undergoing endoscopic procedures (4.0% in Dutch population, 5.6% USA). However, the majority of studies looking at BE prevalence are confounded by the fact that most procedures were performed for GERD symptoms generating selection bias. In studies performed in patients referred for colonoscopy (n=1500 approximately) the prevalence of BE varied between 6.8% and 16.7%.

Similar to oesophageal cancer, gastric cancer has its own pre-malignant precursors including gastric intestinal metaplasia (GIM) and gastric atrophy (GA) . The endoscopic diagnosis of GA and GIM however is not straightforward as the colorimetric shift from normal to pathological mucosa is subtle and additional endoscopic signs such as disappearance of gastric folds can be affected by the degree of air insufflation. The prevalence of GA can reach up to 33.3% in Western Europeans and is higher in East Asia. For GIM, the reported population prevalence in Western Countries ranges from 7 to 28.6%. These rates are however probably affected by selection bias, such as enrichment of individuals affected by H. pylori infection. In a Dutch study on patient referred for colonoscopy 0.8% had GA, 7.1% had both GA and GIM and 1.4% had gastric dysplasia.

For esophageal squamous dysplasia, the precursor lesion of oesophageal squamous cell carcinoma, prevalence estimates are between 3 and 38%. This reflects the variation in endoscopic methods, histopathological scoring and the studied populations from high-risk geographic regions such has Korea. The prevalence of squamous dysplasia in Western Countries in not well known, since it is generally very difficult to diagnose on white light endoscopy. It can be argued that squamous dysplasia is vastly underdiagnosed in clinical practice as the ratio between oesophageal adenocarcinoma and squamous cell carcinoma in the UK is just over 2:1, however squamous dysplasia is rarely diagnosed.

Duodenal adenoma is another understudied pre-malignant lesion that can in fact undergo malignant transformation in 30-85% of cases and as such, needs to be recognised and treated early and definitively . These occur very often in the context of familial adenomatous polyposis (FAP) syndrome, whereas sporadic duodenal adenomas are much rarer with a prevalence of 0.03-0.4% (32, 33).

The majority of studies looking at prevalence of pre-malignant upper GI conditions in the general population referred for EGD are limited by the lack of standardised endoscopy and biopsy protocols. In addition, none of the studies looked at variation of prevalence based on experience. Furthermore, no expert review of endoscopic images has been performed to ensure quality of the diagnosis. Although quality standards in upper GI endoscopy have been proposed by the British Society of Gastroenterology and European Society of Gastrointestinal Endoscopy, the majority of the statements in these guidelines have a low level of evidence and weak to moderate strength of the recommendation. Most importantly, none of the statements refer to diagnostic KPI, given that the expected detection rate of the upper GI pre-neoplastic and early neoplastic lesions at endoscopy is unknown. This represents a gap that our study aims to tackle.

HYPOTHESIS The hypothesis is that the prevalence of upper Gi pre-malignant condition is underestimated based on available literature data and that the detection rate of pre-cancerous upper GI lesions improves with a more rigorous and systematic approach to performing EGD. In the future, using diagnostic KPI to monitor performance of upper GI endoscopists will result in improved outcomes of oesophago-gastric cancer.

STUDY OBJECTIVES The aim of this study is to assess endoscopic and pathological outcomes during routine EGD using standard photo-documentation and biopsy protocol.

The primary objective is to define the prevalence of pre-malignant or early malignant upper GI tract lesions as measured by a standardised endoscopy and biopsy protocol. This is further broken down into categories of pre-malignant pathology, namely: 1) Barrett's esophagus with and without dysplasia; 2) squamous dysplasia;, 3) gastric atrophy/gastric intestinal metaplasia; , 4) gastric adenoma; and 5) duodenal adenoma.

The secondary objectives are:

1. Missed rate of endoscopic diagnosis compared to pathological diagnosis
2. Missed rate of endoscopic diagnosis compared to expert review of endoscopic photos
3. Correlation between duration of the endoscopy and detection of pre-neoplastic and neoplastic lesions
4. Quality of the endoscopic pictures taken in routine endoscopy
5. Correlation between endoscopist experience and rate of detection
6. Correlation between use of image enhanced endoscopy and rate of detection
7. Interobserver agreement among expert and not expert endoscopists
8. Correlation between life-style factors and pathological outcomes

STUDY DESIGN This will be a multicenter population study. Recruiting centers will be a range of Teaching and District General Hospitals in England. 1000 patients referred for urgent or routine endoscopy to evaluate upper GI symptoms will be recruited. The detection rates of upper GI tract pre-malignant lesions in our study procedures will be compared to standard historical data from each participating institution. An equivalent population from the year prior to the start of recruitment, matched for age, sex, indication for endoscopy and referral pathway, will be retrieved from the archives of each participating NHS Trust and directly contrasted to this study population to evaluate whether the detection rate of pre-malignant upper GI tract lesions improves with a more rigorous and systematic approach to performing EGD using a standardized photo documentation and biopsy protocol. In this way, the investigators can evaluate the immediate impact of our study and the utility of the proposed diagnostic KPI. As an example, the investigators already have date to show that in 2019 over 8,500 EGD were performed in Cambridge alone, of which 53% had biopsies taken. Based on matched pathology record, only 29 new diagnoses of GA or GIM were made during the same time period, suggesting the possibility of large under-detection of premalignant stomach. The endoscopic pictures will be evaluated by two expert endoscopists with interest in detection and treatment of upper GI pre-malignant and early malignant conditions. The quality of the pictures will be scored on a visual analogue scale based on two parameters: optical quality and representation of the intended section of the upper GI tract. The endoscopists will note the presence of premalignant conditions. In case of disagreement, a third investigator will review the picture to achieve a gold standard endoscopic diagnosis.