Burden of disease from Helicobacter pylori infection in western Canadian Arctic communities

The CANHelp research program arose from the confluence of three constituencies: residents of western Canadian Arctic communities worried about Hp infection and its link to stomach cancer; health care practitioners frustrated by poor effectiveness of available clinical management strategies for this frequently encountered infection; and public health officials seeking evidence to inform infection control strategies. In the early 2000s, NT health care officials sought input from University of Alberta researchers to respond to concerns voiced by community leaders. In 2006, a meeting between academic scientists and NT medical directors generated support for community-driven research aimed at describing the burden of disease from Hp infection in concerned communities. NT health care partners recommended the Hamlet of Aklavik for the initial project because Aklavik community leaders had advocated for research to reduce health risks from Hp infection. The Aklavik H. pylori Project launched in 2007. Word of its success generated interest in neighboring communities. Invited by community leaders, the research team launched community H. pylori Projects in Old Crow YT in 2010, Tuktoyaktuk NT in 2011 and Fort McPherson NT in 2012. Projects launched in 2016–2017 are not included in this report. The University of Alberta Health Research Ethics Board approved this research, and as required by law, we obtained annual research licenses in both NT and YT before collecting data.

Population estimates (from census nearest project launch) were 594 (2006) for Aklavik, 245 (2011) for Old Crow, 854 (2011) for Tuktoyaktuk, and 792 (2011) for Fort McPherson (Fig. 1) [18‐21]. Most residents of participating communities identify as Indigenous: by census counts, in Aklavik, 92% were Indigenous (mainly Inuvialuit (western Canadian Inuit) or Gwich’in (Athabaskan) First Nation); in Old Crow, 90% were Vuntut Gwitchin (Athabaskan) First Nation; in Tuktoyaktuk, 92% were Inuvialuit; and in Fort McPherson, 94% were Indigenous (mainly Gwich’in) [18, 22]. Aklavik, 113 km south of the Arctic Coast, is accessible by air, ice road in winter and water in summer [23]. Old Crow, north of the Arctic Circle on the Porcupine River, is accessible only by air [24]. Tuktoyaktuk, on the Arctic Ocean coast, is accessible by air, highway year-round since November 2017 (before that by ice road in winter). Fort McPherson, on the Peel River, is accessible by highway year-round [23].

Each project launch required the formation of a planning committee of community residents; the process for recruiting committee members was devised by community partners. An exception was the Tuktoyaktuk project, launched at the request of the Inuvialuit Regional Corporation [25], with planning occurring in partnership with regional Inuvialuit leadership and a Tuktoyaktuk community health representative. Planning committees and other community partners provided input on design and implementation of all research activities to ensure they adhered to community priorities and cultural values. Planning committees and other community partners reviewed all research reports before they were made public.

Project recruitment activities in each community encouraged all residents to participate in baseline research activities: interviewer-administered questionnaires; Hp screening; and endoscopy with gastric biopsies for culture and pathological examination. After endoscopy activities concluded, randomized trials offered treatment to eliminate Hp infection to those testing positive. Participants gave written informed consent for overall project participation with additional consents for endoscopy and treatment. Participants < 17 years old required written parental consent, with children deemed sufficiently mature providing written assent. Details can be found on the CANHelp Working Group website [9].

Beginning in January 2008, we offered screening for Hp infection using the 13C-urea breath test (UBT), the most accurate non-invasive test for detecting Hp infection in children and adults [26]; most well-designed validation studies estimate both sensitivity and specificity greater than 95% against biopsy-based diagnosis [27], though it should be noted that biopsy-based diagnosis is not an optimal gold standard because tissue sampling can miss Hp in the stomach due to its patchy distribution and diagnostic accuracy also depends on tissue preparation techniques and observer expertise [26, 28, 29]. The UBT avoids these limitations, but its accuracy depends on optimal breath sample collection [30]. We used nondispersive isotope-selective infrared spectroscopy to measure the 13C/12C ratio in breath samples collected before and after consumption of 13C-labeled urea [31, 32]; a positive result reflects presence of urease in the stomach, a highly specific marker for Hp infection. We asked participants to avoid acid-suppressing medications for 24 h before the test and to fast except for water for 4 h before the test. We obtained a baseline breath sample, asked participants to swallow 50–75 mg of 13C-labeled urea dissolved in citric acid, and collected a second breath sample 30 min after the urea solution was swallowed [17].

Using ultra-slim Olympus gastroscopes, gastroenterologists performed unsedated upper gastrointestinal endoscopy for consenting participants, regardless of Hp status, in temporary endoscopy clinics organized in community health centers (2008 in Aklavik, 2012 in Old Crow, 2013 in Fort McPherson and at the Inuvik Regional Hospital for Tuktoyaktuk participants) [33]. The regional planning approach used for the Tuktoyaktuk project created logistic constraints that reduced time periods, relative to the other community projects, for recruiting participants and carrying out project activities; as a result, participation was not sufficient to warrant resources required for an endoscopy team to visit this community. Instead, we offered to pay travel expenses for Tuktoyaktuk residents willing to undergo endoscopy in Inuvik (a 2-h drive). Gastric biopsies collected during endoscopy were transported to the University of Alberta for histopathologic examination (2–3 from antrum, 2–3 from corpus) and tissue culture (1 from antrum, 1 from corpus) [14]. A single pathologist (SG) evaluated antral and corpus biopsies separately, grading Hp density, active and chronic gastritis, gastric atrophy, and other neoplastic lesions using the updated Sydney system [34]. We differentiate active gastritis, characterized by polymorphonuclear neutrophil infiltration occurring in the context of chronic inflammation, from chronic gastritis, characterized by the presence of mononuclear cells, chiefly lymphocytes, plasma cells and macrophages [34]. For culture, antral and corpus biopsies were pooled and mechanically homogenized; resulting suspensions were plated on brain heart infusion/yeast extract/5% horse serum agar plates and incubated at 37 °C in microaerophilic conditions. Cultures were checked for growth every 48 h, for up to a month. Colonies consistent with Hp growth were sub-plated and expanded for generation of glycerol stocks. Hp classification was confirmed by urease, catalase, oxidase and 16S ribosomal RNA PCR testing.

We classified Hp infection status using all available results (UBT, histopathologic examination, and/or bacterial culture) for each participant. When results were discordant, we used an algorithm based on the probability that the status was negative or positive (for example, UBT positive + histopathology positive + culture negative was classified as positive; UBT negative + histopathology positive + culture negative was classified as negative). Among 254 participants with classifiable Hp status from 3 tests, 216 (85%) were concordant on all 3 tests; among 53 participants with classifiable Hp status from just 2 tests, 47 (89%) had concordant status. Of the 3 tests, UBT and histopathology agreed most frequently, for 263 of 272 (97%) participants with results from both, while the results of culture diverged more frequently, agreeing with UBT for 218 of 254 (86%) participants with both of these tests and with histopathology for 252 of 289 (87%) participants with both of these tests. The more frequent discordance of culture is not surprising given the technical challenges inherent in tissue culture of Hp in the laboratory [30, 35]. The high agreement of UBT and histopathology indicates comparable and excellent accuracy of these 2 diagnostic tests. Comparing UBT against histopathology as the gold standard, estimated sensitivity is 96% (95% confidence interval (CI), 94–99%) and estimated specificity is 97% (95% CI, 94–100%). Conversely, comparing histopathology against UBT as the gold standard, estimated sensitivity is 99% (95% CI, 98–100%) and estimated specificity is 91% (95% CI, 85–97%).

The cross-sectional design is appropriate for investigations of the burden of disease from Hp in a community setting, given that the infection onset generally goes undetected, the infection often persists indefinitely without symptoms, and gastric disease caused by the infection is often asymptomatic. Screening for prevalent cases is, therefore, the starting point for describing the occurrence of Hp infection and related disease in a population. To describe the burden of infection in participating communities, we present the estimated prevalence of Hp infection and 95% confidence intervals by age, sex, ethnicity and community. To describe the burden of gastric disease related to Hp infection and demonstrate relatedness of specific conditions to Hp infection, we present the estimated prevalence of endoscopic diagnoses (esophagitis/esophageal erosions, Barrett’s esophagus, gastritis, gastric erosions, gastric ulcer, duodenitis, duodenal erosions, duodenal ulcer) and histopathologic diagnoses (active gastritis, chronic gastritis, gastric atrophy, benign MALT hyperplasia or lymphoid aggregates, lymphoepithelial lesions, intestinal metaplasia, dysplasia, carcinoma) and 95% CIs by Hp status. In analyses of endoscopy or histopathology results, we excluded the small number of participants who underwent treatment to eliminate Hp after the screening UBT and before endoscopy.

The four community projects enrolled 934 participants, representing 38% of the combined populations of these communities (Table 1). The first two community projects had the highest participation, with 64 and 85% of residents of Aklavik and Old Crow, respectively. Hp status was available for 878 (94%) of the 934 participants. Table 2 shows participation in UBT screening and endoscopy by community, with UBT results available for 90% (841/934) and histopathology results available for 33% (308/934). Due to the logistic challenges, only 5 Tuktoyaktuk residents had eligible histopathology results: each fell in a different age group; 4 were women; 5 were Indigenous (4 were Inuvialuit, the dominant ethnicity in Tuktoyaktuk); and 4 were Hp-positive with a distribution of endoscopic and histopathologic outcomes reflective of the larger study population.

Table 3 shows the distribution of age, sex, community and ethnicity in all 934 participants and in two study population subsets: the 878 with Hp status and the 308 with histopathology results. Aside from the exclusion of young children from endoscopy, the age distribution of the two subpopulations approximates that of the total study population. The sex distribution is similar across study population subsets, with women accounting for 54–56% of each. The proportion identifying as Indigenous is 90–92% across study subpopulations. The representation of communities is nearly identical in the total study population and among those with Hp status, but among those with histopathology results Aklavik is overrepresented due to a much higher proportion of participants undergoing endoscopy in the first project, and Tuktoyaktuk is underrepresented due to only 13 participants traveling to Inuvik for endoscopy. Just 56 (6%) participants lacked data on Hp status; these participants were disproportionately younger and male.

Among 878 participants with Hp status, 62% were Hp-positive (Table 4). The prevalence was 45% (95% CI, 36–55%) in children under 15, 69% (95% CI, 63–74%) in the 15–34-year age group, and 61% (95% CI, 57–66%) in the 35–96-year age group. The prevalence was somewhat higher in men (65%; 95% CI, 60–70%) than women (59%; 95% CI, 55–63%). Across communities, the prevalence ranged from 57% in Tuktoyaktuk to 68% in Old Crow. The largest variation in prevalence was by ethnicity, with non-Indigenous participants having a much lower prevalence (22%; 95% CI, 13–31%) than Indigenous participants (66%; 95% CI, 62–69%).

Table 5 shows the prevalence of abnormalities observed during endoscopy by Hp status. Of note, 79% of Hp-positive participants had gastric mucosa that appeared normal. Visible esophagitis and Barrett’s esophagus were more prevalent in Hp-negative participants (11 and 7%, respectively) than Hp-positive participants (8 and 4%, respectively). Paradoxically, the prevalence of visible gastritis, gastric erosions and gastric ulcers was not markedly different in groups defined by Hp status: 12–15% had visible gastritis, 8–11% had gastric erosions, and 3–4% had gastric ulcers. The prevalence of visible duodenal lesions was lower: duodenal lesions were observed in 9% of Hp-positive participants and 4% of Hp-negative participants. No Hp-negative participants had visible duodenal erosions or ulcers and 4% had duodenitis, while 1% of Hp-positive participants had duodenal erosions, 1% had duodenal ulcers, and 7% had duodenitis. In all 308 participants with complete endoscopic assessment, the gastric to duodenal ratio was 2 for inflammation, 8 for erosions, and 3 for ulcers. In 271 Indigenous participants, the gastric to duodenal ratio was 2 for inflammation, 6 for erosions, and 8 for ulcers.

We examined whether the similar frequencies of endoscopically observed gastric abnormalities in groups defined by Hp status were due to previous Hp infection eliminated by antimicrobial therapy among Hp-negatives or recent use of proton pump inhibitors (PPI) or H2-receptor antagonists (H2RA), which decrease the density of Hp organisms, thereby reducing the sensitivity of diagnostic tests. In the 84 Hp-negative participants with treatment history data, 30 (36%) had previous treatment; the proportion classified as having normal gastric mucosa was 80% among those treated previously and 77% among those not treated previously (for duodenal mucosa, these proportions were 97 and 96%, respectively). Among 302 participants with medication data, recent PPI/H2RA use was reported by 11% of Hp-positives and 27% of Hp-negatives. To assess the hypothesis that Hp-negatives with abnormal gastric mucosa were false negatives due to medication use, we compared the proportion classified as having normal gastric mucosa by PPI/H2RA use: in 85 Hp-negatives, 74% among users and 79% among non-users (for duodenal mucosa, 91 and 98%, respectively); in 218 Hp-positives, 68% among users and 80% among non-users (for duodenal mucosa, 96 and 91%, respectively). Thus, this paradox does not appear to be due to previously treated Hp infection or medication use among current Hp-negatives.

Groups defined by Hp status had strikingly different frequencies of abnormal histopathology (Table 6). The proportion with normal gastric mucosa was 77% among Hp-negative participants evaluated and just 1 of 224 Hp-positive participants evaluated. Compared to Hp-negative participants, Hp-positive participants not only had a much higher prevalence of histologic abnormalities, but they also had a much higher severity gradient, although the frequency of intestinal metaplasia was low in both groups and the difference between groups less striking. Of the 4 Hp-negatives with intestinal metaplasia, 2 had a record of previous treatment to eliminate Hp.

The results show good statistical precision for describing the severity distributions of the histopathological outcomes and differentiating these distributions in groups defined by Hp status. The study population yields even greater precision for differentiating the presence or absence of these outcomes: active gastritis occurred in 96% (95% CI, 92–98%) of positives and 2% (95% CI, 0–8%) of negatives; chronic gastritis occurred in 99% (95% CI, 98–100%) of positives and 13% (95% CI, 6–20%) of negatives; gastric atrophy occurred in 43% (95% CI, 36–49%) of positives and 0% (95% CI, 0–4%) of negatives; intestinal metaplasia occurred in 17% (95% CI, 12–22%) of positives and 5% (95% CI, 1–12%) of negatives; MALT hyperplasia or lymphoid aggregates occurred in 74% (95% CI, 69–80%) of positives and 12% (95% CI, 5–19%) of negatives; lymphoepithelial lesions occurred in 3% (95% CI, 1–6%) of positives and 0% (95% CI, 0–4%) of negatives. Thus, each of these histopathological outcomes was strongly associated with prevalent Hp infection in this population.