Helicobacter pylori

A meta-analysis (a study that pools the results of earlier studies) of 184 studies from 62 countries estimated that about 4.4 billion people were infected in 2015. Pooled prevalence was highest in Africa (70.1%) and lowest in Oceania (24.4%); by country it ranged from 18.9% in Switzerland to 87.7% in Nigeria [3].

A later meta-analysis of 1,748 studies from 111 countries found that adult prevalence fell from 52.6% before 1990 to 43.9% in 2015 to 2022, a decline of 15.9% over three decades. Prevalence in children and adolescents was 35.1% in 2015 to 2022 and did not fall significantly in any World Health Organization region. Stomach cancer incidence fell globally and in the countries where infection became less common [4].

The Maastricht VI/Florence consensus of 41 experts from 29 countries describes H. pylori gastritis as an infectious disease, listed in the International Classification of Diseases, 11th Revision, and concludes that in principle every infected person should be treated [9].

IARC estimated the share of cancers caused by 12 infectious agents classified as Group 1 carcinogens, the category for agents with sufficient evidence of causing cancer in humans, using GLOBOCAN incidence data for 2024. Infections accounted for 2.3 million new cancers (12% of the total). H. pylori accounted for 760,000, followed by human papillomavirus (750,000), hepatitis B virus (360,000), Epstein-Barr virus (260,000) and hepatitis C virus (160,000). Eastern Asia had 42% of all infection-attributable cancers [1].

The attributable fraction depends on how infection is measured, because infection and the antibodies that blood tests detect can be lost as cancer develops [7]. Using the more sensitive immunoblot blood test instead of the older ELISA test, IARC raised its estimate of non-cardia stomach cancers attributable to H. pylori from 74.7% to 89.0% [2]. A projection for people born between 2008 and 2017 in 185 countries expects 15.6 million lifetime stomach cancers in that generation, 76% attributable to H. pylori, with two-thirds in Asia and 42% in areas where the disease is now uncommon, owing to demographic change [10].

The strongest estimate of risk comes from studies that stored blood before anyone was diagnosed. A pooled analysis of 12 case-control studies nested in prospective cohorts, with 1,228 stomach cancer cases, found that infection was associated with an odds ratio (a measure of how much more common the exposure is among cases) of 3.0 for non-cardia stomach cancer (95% confidence interval 2.3 to 3.8). The odds ratio was 5.9 (3.4 to 10.3) when blood had been drawn 10 or more years before diagnosis, which the authors considered the best estimate because infection and antibodies can be lost as cancer develops. Infection was not associated with cardia cancer (odds ratio 1.0, 0.7 to 1.4) [7].

Absolute risk is low for most carriers. In a Japanese cohort of 1,526 patients with ulcers, stomach polyps or indigestion followed with endoscopy for a mean 7.8 years, stomach cancer developed in 36 of 1,246 infected patients (2.9%) and in none of 280 uninfected patients. Patients with severe atrophy, inflammation concentrated in the body of the stomach (corpus-predominant gastritis) or intestinal metaplasia were at significantly higher risk, and none of the 275 patients with duodenal ulcers developed stomach cancer [5].

Strains of H. pylori differ in their cancer risk. CagA (cytotoxin-associated gene A) is a protein produced by some strains. In a meta-analysis of 16 age- and sex-matched case-control studies (2,284 cases and 2,770 controls), antibodies to H. pylori were associated with 2.28 times the risk of stomach cancer and antibodies to CagA with 2.87 times. Among infected people, CagA-positive strains carried 1.64 times the risk of other strains (95% confidence interval 1.21 to 2.24), and 2.01 times (1.21 to 3.32) for non-cardia cancer. Neither infection nor CagA status was associated with cardia cancer [11].

Endoscopy with biopsy and microscopic examination is the confirming test. Non-invasive tests include the urea breath test, in which the bacterium's urease enzyme breaks down a swallowed labelled urea that is then measured in the breath, the stool antigen test and blood antibody tests (serology). A Cochrane review of 101 studies with 11,003 participants compared these tests with biopsy. At a fixed specificity of 90%, the carbon-13 urea breath test detected 94% of infections, the carbon-14 breath test 92%, serology 84% and the stool antigen test 83%. In a hypothetical 1,000 people with a 53.7% infection rate, the carbon-13 breath test would miss about 30 infections, serology 86 and the stool antigen test 89. The authors concluded that breath tests had high accuracy in people without a gastrectomy who had not recently taken antibiotics or proton pump inhibitors (a class of acid-suppressing drugs), and that the comparison was mostly indirect [8].

Treatment combines antibiotics with acid suppression. The American College of Gastroenterology 2024 guideline prefers 14 days of bismuth quadruple therapy for a first treatment when antibiotic susceptibility is unknown. Rifabutin triple therapy or potassium-competitive acid blocker dual therapy for 14 days are alternatives for people without penicillin allergy. Regimens containing clarithromycin or levofloxacin are reserved for infections with confirmed susceptibility, and the guideline recommends a test of cure after every treatment [12]. The Maastricht VI/Florence consensus identifies rising antibiotic resistance as the main challenge and supports susceptibility testing, including newer molecular methods [9].

A 2025 meta-analysis of 11 randomised trials and 13 observational studies found that eradication therapy lowered stomach cancer incidence in healthy H. pylori-positive adults (8 trials, relative risk 0.64, 95% confidence interval 0.48 to 0.84) and stomach cancer death (5 trials, 0.78, 0.62 to 0.98). In patients whose early stomach cancer had been removed at endoscopy, it lowered new stomach cancers (3 trials, 0.52). Observational studies gave consistent results [6].

In a Korean trial of 1,838 infected first-degree relatives of stomach cancer patients, a 7-day course of lansoprazole, amoxicillin and clarithromycin was compared with placebo. Over a median 9.2 years, stomach cancer developed in 1.2% of the treated group and 2.7% of the placebo group (hazard ratio 0.45). Among participants whose infection was cleared, 0.8% developed cancer, against 2.9% of those with persistent infection (hazard ratio 0.27). Adverse events were mild and more common with treatment (53.0% against 19.1%) [13]. In a second Korean trial of patients who had early stomach cancer removed, new stomach cancer developed in 7.2% of the treated group and 13.4% of the placebo group over a median 5.9 years (hazard ratio 0.50), and atrophy of the stomach lining improved in 48.4% against 15.0% [14].

The Shandong Intervention Trial in Linqu County, China, a high-risk area, randomised 2,258 adults with antibodies to H. pylori to two weeks of amoxicillin and omeprazole or placebo. Over 22 years the treatment was associated with lower stomach cancer incidence (odds ratio 0.48) and lower stomach cancer mortality (hazard ratio 0.62) [15].

Full deep dive: H. pylori and stomach cancer.

H. pylori is the largest single risk factor for stomach cancer but acts alongside others. The highest salt intake was associated with 68% more stomach cancer than the lowest across 10 prospective cohorts (relative risk 1.68) [16]; salt is described under added salt. Current smokers had 53% more stomach cancer than never smokers across 92 studies, and about twice as much cardia cancer (relative risk 2.08) [17]; see smoking. The highest processed meat intake was associated with 45% more stomach cancer (relative risk 1.45) [18]. Having a parent, sibling or child with stomach cancer was associated with 84% higher odds (odds ratio 1.84) [19], described under family history and genetics. The stomach cancer chart places each of these factors on one scale.

Prevalence estimates depend on the test used and on which populations were studied; many countries have no recent data [3, 4]. Attributable-fraction estimates are models that combine incidence registries with relative risks from other studies, and they change with the test used to define infection [2, 1]. The cohort that showed cancer only in infected people enrolled hospital patients with stomach symptoms in Japan, not a general population [5].

The randomised trials were run mainly in East Asia, where stomach cancer is common, and the pooled estimate for death rests on 5 trials [6]. Several trials used regimens, such as 7-day clarithromycin triple therapy, that current North American guidance no longer prefers as a first choice [13, 12]. The diagnostic-accuracy comparison was mostly indirect, and every included study had at least an unclear risk of bias [8].