Background: Atrophic gastritis (AG) is the single most important precursor condition for gastric cancer (GC) known so far. H. pylori infection is the most important causative agent of gastritis, and subsequent AG. The GastroPanel test (Biohit HealthCare, Helsinki, Finland), a blood test evaluating the four biomarkers specific for the gastric mucosa pepsinogen I (P-PGI), pepsinogen II (P-PGII), gastrin-17 (P-G-17) and H. pylori antibody (P-HpAb), is the first non-invasive diagnostic tool providing possibilities for detecting the patients at risk for GC and peptic ulcer as well as malabsorption of vitamin B12, iron, magnesium, calcium and some drugs. A well designed clinical study is warranted to fully assess the performance of GastroPanel examination in detecting the gastric lesions which can lead to GC. The investigators aim to perform a clinical study in an adult population in United Kingdom in order to determine the diagnostic accuracy of the GastroPanel test in evaluating AG and other specific gastric conditions associated with an increased risk for GC. Methods: Two hundred and fifty patients (45 years and older, both genders) will be enrolled among the patients with dyspepsia referred for gastroscopy at Homerton University Hospital (London, United Kingdom). During the same visit, all patients are subjected to gastroscopy examination, with directed biopsies from the antrum and corpus, following the protocol of the operative link on gastritis assessment (OLGA) classification for chronic gastritis and Sydney Classification. Biopsies are examined at the Pathology laboratory of Homerton University Hospital and interpreted using the OLGA staging system as well as the Sydney system for classification of gastritis. Specific aims: The principal goal of this clinical trial is to establish the performance of the GastroPanel examination in detecting AG and other specific gastric conditions associated with an increased risk for GC. In particular, the investigators will evaluate AG in the antrum, AG in the corpus, AG in both antrum and corpus (=atrophic pangastritis), biopsy-confirmed dysplasia (intestinal metaplasia) of the gastric mucosa. For all these conditions, the investigators will calculate the diagnostic accuracy of the GastroPanel test.

At present, the diagnosis of most gastric and oesophageal diseases, requires an endoscopic examination which is an invasive, time-consuming and expensive procedure. At present, there are few non-invasive methods (e.g. tests for Helicobacter pylori) available for the diagnosis of the upper gastrointestinal tract diseases. Any of these tests do not, however, give possibilities for a comprehensive diagnosis of the different phenotypes of gastritis, i.e., whether superficial or atrophic, and located in the antrum or corpus. Importantly, these tests do not give any clues about the severity (grade) of these lesions, as defined by the Sydney and OLGA classification .

To obviate the excessive use of this invasive and expensive procedure (endoscopy), there is an urgent need to develop non-invasive diagnostic tools capable of accurately detecting the patients at high risk for GC, i.e. the different phenotypes of gastritis as well as their related H. pylori infections . After ELISA-testing for P-PG I, p-PG II, P-G-17 and P-Hp-Ab in a plasma sample, an endoscopic examination can be preserved only for those patients whose GastroPanel test results suggest AG, whereas an endoscopic examination can be avoided in subjects with negative GastroPanel result, or in whom the test biomarkers indicate a non-atrophic gastritis or a healthy stomach (18). Gastroscopy is also recommended if the GastroPanel examination reveals high acid output (P-G-17 below 1,0 pmol/l) or chronic H. pylori infection with symptoms.

This clinical trial is conducted as collaboration between Biohit HealthCare (Helsinki, Finland) and Homerton University Hospital (London, UK) (hereafter called "the Partners"). The study is performed in Homerton Hospital, supervised by a steering committee consisting of members from both research Partners.

Enrolment of the patients in the study will take place at Homerton Hospital including consecutive patients over 45 years of age, referred for gastroscopy at the Outpatient Department of Endoscopy. The estimated cohort to be screened is at least 250 subjects (both genders), to reach a cohort of 100 patents enriched with equal numbers (n=25) of all conditions (see: Section 2, above) classified as study endpoints.

Patient enrolment is taking place in a single step. In brief, the potentially eligible patients are identified among the gastroscopy-referral outpatients by the members of the research team. At this stage, every patient will be asked to consent the study and sign a written consent to participate. Because all patients are enrolled among the subjects attending the 11 Endoscopy clinic due to an appointment to gastroscopy, their preparation will be compliant with the preparatory steps needed for the GastroPanel examination (details below).

Eligible patients are all adult females and males, with dyspeptic symptoms (epigastric pain, bloating and epigastric discomfort). However, the following patients should be considered non-eligible: 1) the patients whose treatment requires surgery, or immediate follow-up treatment for major symptoms, as well as 2) those that refuse to participate.

In this study, all patients examined with the GastroPanel test will be subjected to gastroscopy, providing the histological confirmation to be used as the gold standard in calculating the performance indicators for the test.

All patients participating in this study shall undergo a routine gastroscopy examination, which will be complemented by biopsy sampling from the antrum and corpus, according to the principles of the Sydney and OLGA classification sampling. In endoscopy, all observed abnormal mucosal lesions are noted and photographed, and if necessary (e.g. suspicion of malignancy) subjected to additional biopsy.

All statistical analyses will be performed using the SPSS 21.0.0 for Windows (IBM, NY, USA) and STATA/SE 13.0 software (STATA Corp., Texas, USA). The descriptive statistics will be done according to routine procedures. Performance indicators (sensitivity, specificity, positive predictive value, PPV, negative predictive value, NPV and their 95%CI) of individual markers and whole GastroPanel test will be calculated separately for each study endpoint, using the STATA/SE software and the diagti algorithm introduced by Seed et al. (2001). This algorithm also calculates the area under ROC (Receiver Operating Characteristics) called AUC, for each biomarker at each endpoint. Because GastroPanel is a quantitative ELISA test, these ROC curves can be used to identify the optimal sensitivity/specificity balance that gives each biomarker an optimal threshold for detection of each study endpoint. Significance of the difference between AUC values can be estimated using STATA's roccomb test with 95%CI.

Inclusion Criteria:

• Adult females and males over 45 years of age with dyspeptic symptoms (epigastric pain, bloating and epigastric discomfort)

Exclusion Criteria:

• Patients who require surgery or immediate follow-up treatment for major symptoms, including hematemesis, melena, acute epigastric pain
• Patients who previously underwent upper gastrointestinal surgery
• Patients with diabetes
• Pregnant women
• Subjects who refuse to participate or are unable to give consent

Enrollment of the patients in the study will take place at Homerton Hospital including consecutive patients over 45 years of age, referred for gastroscopy at the Outpatient Department of Endoscopy. The estimated cohort to be screened is 250 subjects (both genders), to reach a cohort of 100 patents enriched with equal numbers (n=25) of all conditions classified as study endpoints.

Patient enrollment is taking place in a single step. In brief, the potentially eligible patients are identified among the gastroscopy-referral outpatients by the members of the research team. At this stage, every patient will be asked to consent the study and sign a written consent to participate.