Stomach cancer

Stomach cancer, also called gastric cancer, is a malignant tumour that arises from the cells of the stomach, most often its inner mucosal lining. It is one of the most common cancers worldwide and a leading cause of cancer death, and its burden is very uneven across regions. Most cases are gastric adenocarcinomas, which develop slowly through a recognised sequence of chronic inflammation, atrophy, intestinal metaplasia and dysplasia before invasive cancer forms. Its clinical course depends heavily on how deep the tumour has grown, whether lymph nodes or distant organs are involved, and where in the stomach it begins. The disease is a useful case study in how an infectious agent, cell-level genetic change, diet and host factors combine to cause a cancer, and in how the same disease can differ sharply in incidence and survival between countries.

This entry covers the disease from its basic definition and anatomy to its causes, pathology, diagnosis, staging, treatment and prognosis, and then turns to prevention and public health. It explains the main histological classifications and the epidemiology of incidence and mortality, with particular attention to the role of bacterial infection and the evidence from eradication studies. It also reviews the research history that shaped current understanding, from early pathology descriptions to the discovery of the bacterium's role, and it closes with the questions that remain unresolved, such as cancers that occur without infection and the design of screening programmes.

Where the wording of a medical topic in this entry describes established findings, it follows major international sources such as the World Health Organization and cancer registries, and it avoids clinical recommendations for individual patients.

Definition, anatomy and sites of origin

Definition

Gastric cancer is a malignant tumour arising from the epithelium that lines the stomach, most often the glandular cells of the mucosa. It is distinct from oesophageal cancer, which develops in the muscular food pipe above the stomach, and from small intestinal cancer, which arises in the duodenum, jejunum or ileum and has its own risk profile and treatment. The boundary with oesophageal cancer is the main source of classification difficulty.

Anatomical subdivisions and the cardia

The stomach is conventionally divided into five regions. The cardia is the small zone immediately surrounding the opening where the oesophagus joins the stomach. Above it lies the fundus, the dome-shaped upper portion. The body forms the large central part, the antrum the lower section, and the pylorus the narrow outlet into the duodenum. For classification, tumours are therefore described either as cardia tumours, including those centred on the gastro-oesophageal junction (GEJ), or as non-cardia tumours, which arise in the fundus, body, antrum or pylorus.

The WHO now groups adenocarcinomas that straddle the oesophagogastric junction under one heading, whichever side holds most of the tumour mass, and this category absorbs many tumours previously labelled cardia cancers. WHO definition includes many tumours formerly called cancers of the gastric cardia

Surgeons Siewert and Stein proposed a three-part system in the 1990s that places junctional tumours according to the position of their epicentre relative to the cardia. Siewert I cancers constitute most upper GEJ carcinomas with an epicenter located 1–5 cm above the anatomic junction. Siewert II cancers, also called true junctional cancers, extend from 1 cm above to 2 cm below the junction. Siewert III cancers are subcardial, extending 2–5 cm below the junction.

Why location matters

Location determines which staging scheme and treatment pathway applies. Under the AJCC system, tumors with an epicenter less than 2 cm into the gastric cardia be treated as esophageal cancers, while tumors with epicenter greater than 2 cm into the cardia be staged and treated as gastric cancers. In practice, Siewert I cancers often approached as esophageal carcinomas through a transthoracic esophagectomy and Siewert III cancers often approached as gastric cancers with gastrectomy and perigastric lymphadenectomy, while Siewert II tumours are managed according to where most of the disease lies.

The proportion of junctional tumours also varies by region. Proximal gastric cancer, cancer of the cardia and distal oesophageal cancer together accounted for about 30%-40% of all gastric carcinoma in Western countries, and China has a far higher share than Japan and Korea. Because of this, the distribution of tumour sites in a population partly reflects the regional balance between cardia and non-cardia disease.

Spread through the stomach wall, lymphatics and blood

Tumours spread in several ways. Growth through the wall can invade neighbouring structures directly; adjacent organs such as liver, pancreas, spleen, and colon are frequently involved by direct extension. Lymphatic dissemination reaches distant nodes, and the stomach's drainage is intricate, since the stomach is supplied with blood by five main vessels, (the right and left gastric, the right and left gastro-epiploic, and short gastric arteries) the venous and lymphatic drainage of the stomach is complicated.

Haematogenous spread most often targets the liver. Hematogenously, liver is frequently involved. Lungs and bone are also recognised sites, and the most common sites of distant metastases from GC are peritoneal metastases, ovarian metastases, hematogenous metastases that spread to the liver, lungs, and bone, as well as metastases to the distant lymph nodes. Blood-borne spread is inefficient: one review cited in a case report estimated that only ~20% of the cells reach an end organ.

Peritoneal involvement follows penetration of the gastric wall. As the cancer penetrates the gastric wall, peritoneal metastasis frequently occurs. Peritoneal dissemination differs mechanistically from the other routes, because the peritoneal metastatic cascade notably differs from the hematogenous and lymphatic cascades, both of which entail directional, intra- and extra-vasation through vessel walls. Tumour location also influences the pattern: proximal gastric cancer localization has been associated with hepatic metastases in one series.

Histological types and classification

Adenocarcinoma and the Lauren classification

Adenocarcinoma accounts for the large majority of gastric malignancies, and most descriptions of gastric cancer histology are built around it. The most enduring framework is the one proposed by the Finnish pathologist Pekka Laurén in 1965, which sorts adenocarcinomas into intestinal, diffuse and mixed (indeterminate) types. The intestinal type is made of cohesive cells that line gland-like structures and tend to resemble the lining of the intestine. The diffuse type is made of poorly cohesive cells that have lost their attachments to one another; they infiltrate the stomach wall as single cells or small clusters, often with little glandular formation and with abundant mucin. Signet-ring cell carcinoma belongs to this group. Because of this architecture, diffuse tumours tend to spread along the layers of the stomach wall and are harder to see on endoscopy, while intestinal tumours form more visible masses. In the original series, intestinal tumours made up roughly half of cases, diffuse tumours about a third, and mixed tumours the remainder, although the proportions vary by population and by study.

Laurén's system remains in wide use because it is simple and reproducible, but it has limitations. Many tumours show both components, and the definition of a "pure" subtype is not standardised, which contributes to variation between observers and to variable prognostic findings. Some studies report that diffuse-type tumours carry a worse prognosis, partly because they are often diagnosed at a more advanced T or N stage, while other analyses question whether the classification adds independent prognostic value once stage is accounted for.

WHO classification and clinical practice

The World Health Organization classification of digestive system tumours provides a more detailed scheme. Its gastric adenocarcinoma categories include papillary, tubular (well, moderately and poorly differentiated), mucinous, and poorly cohesive carcinoma, the last subdivided into signet-ring cell carcinoma and a not-otherwise-specified group. A mixed category is recognised for tumours that combine intestinal-type and diffuse-type components, although no fixed percentage defines it. The WHO scheme also lists several rare variants, such as fundic-gland-type adenocarcinoma, hepatoid adenocarcinoma, and carcinoma with lymphoid stroma. Broadly, the poorly cohesive category corresponds to Laurén's diffuse type, and the tubular and papillary categories largely correspond to the intestinal type, but the correspondence is not exact, particularly for poorly differentiated tumours that Laurén's system tended to place in its indeterminate group.

The 2019 WHO edition, the fifth-edition volume of the series, also stated that histopathology remains the gold standard for diagnosis in most cases, even though some categories are now defined partly by molecular features. For consistency in reporting, international pathology guidance favours the WHO scheme, while the Lauren classification continues to be used alongside it. Clinically, the histological subtype is not only descriptive: HER2 (ERBB2) status is tested in advanced disease to identify candidates for anti-HER2 therapy, and HER2 positivity is more common in intestinal-type tumours.

Rarer malignancies

Non-adenocarcinoma tumours make up a small minority of gastric malignancies. Mucosa-associated lymphoid tissue (MALT) lymphoma is a low-grade B-cell lymphoma arising in the gastric mucosa, and it is closely associated with chronic inflammation and has been linked to bacterial infection; its management differs markedly from that of carcinoma, and it is often treated with eradication of the infection before other options are considered. Gastrointestinal stromal tumours arise from the mesenchymal cells that regulate gut motility rather than from the epithelium, and are usually defined by activating mutations in KIT or PDGFRA. Neuroendocrine tumours derive from enteroendocrine cells; the older term "carcinoid" is still encountered, and the WHO scheme grades these lesions by proliferation index into well-differentiated tumours and neuroendocrine carcinomas. Squamous cell carcinoma and small cell carcinoma of the stomach are extremely rare. Earlier WHO editions grouped small cell carcinoma with the other epithelial carcinomas, and the separate recognition of such tumours reflects their different biology and treatment.

Precancerous lesions and early neoplasia

Gastric carcinoma usually develops through a sequence of mucosal changes rather than appearing abruptly. Dysplasia, sometimes called intraepithelial neoplasia, refers to epithelial cells with atypical nuclear and architectural features that are confined within the basement membrane and are therefore not invasive. Pathologists grade dysplasia as low or high, and high-grade lesions carry a substantially higher risk of progression. Early gastric carcinoma is defined as a tumour that invades no deeper than the mucosa or submucosa, irrespective of lymph node involvement in some schemes; such lesions can often be removed endoscopically, and the depth of invasion is a central determinant of the feasibility of that approach. The recognition of these early forms, in which Japanese pathology and endoscopy have played a leading role, has shaped both classification and treatment decisions.

Epidemiology and geographic patterns

Stomach cancer is one of the most frequently diagnosed and fatal cancers worldwide. In the GLOBOCAN 2022 estimates, stomach cancer ranked fifth for incidence (968,350 new cases) and fifth for mortality (659,853 deaths), accounting for 4.9% of all cancers diagnosed that year and 6.8% of cancer deaths. Other estimates give higher totals: the Global Burden of Diseases study counted 1.23 million new cases and 0.95 million deaths in 2021. The difference reflects different data sources and country coverage, since GBD 2021 includes data from 204 countries and territories, while GLOBOCAN 2022 includes only 185. Age-standardised rates allow comparison across populations with different age structures; globally, the worldwide rates were 9.2 and 6.1 per 100,000 persons for incidence and mortality.

The disease is consistently more common in men. In the GBD 2021 estimates, age-standardised incidence was 8.6 per 100,000 for females and 20.9 per 100,000 for males. Across countries, the male-to-female ASIR ratio exceeds 2:1 in some analyses, and projections of case counts show males almost twice as numerous as females.

Incidence varies widely by region. Age-standardised incidence ranged from about 2.5 to 56.8 per 100,000 across countries in 2018. The highest gastric cancer incidence and mortality were found in Eastern Asia, and East Asia accounted for 59.85% of new cases and 56.23% of deaths in 2020. Within this region, China carries the largest absolute burden. Rates remain high in some Latin American registries as well: in Quito, Ecuador, incidence in men fell from 30.4 to 18.8 cases per 100,000 between 1985 and 2013, but the decline stalled after the late 1990s.

Incidence has fallen in most populations over the past several decades, although not everywhere. Registry data from Sweden show that the incidence rate decreased by 44% for men and 41% for women between 1960 and 1980. In the West Midlands of England, the age-standardised incidence of gastric carcinoma fell by 20% between 1962 and 1981. In urban Shanghai, the rate for men dropped 20% between 1972–74 and 1987–89, while the reduction among women was minor. Overall, the decline over the past century has been attributed to decreases in risk factors such as H. pylori infection, tobacco smoking, and the intake of salt-preserved food. Between 1980 and 2018, incidence fell in 29 countries and mortality fell in 41, yet the decrease is much less marked in some populations, such as US White individuals and individuals in Canada, Colombia, Brazil, Denmark, Germany, India, and Israel. Some countries have also reported rises in younger adults: incidence increased in populations younger than 40 years in a number of countries, and in the United States a notable rise in proximal and noncardia stomach cancer has been described in younger generations.

Projections depend heavily on demographic change. A Lancet modelling study estimated that if current rates stay the same, annual new cases would rise by 62%, from 1.09 million in 2020 to 1.77 million by 2040, whereas if rates decreased by 2% per year, new cases would reach 1.18 million. Another analysis projected that the global population will reach 9.19 billion by 2040, with incidence and mortality rates declining by roughly 0.6% per year. Because ageing and population growth will offset the decline in incidence and mortality rates, resulting in a substantial increase in the number of new cases and deaths. Since the median age at diagnosis is 70 years, population ageing has a strong effect on case counts. For China, ARIMA projections suggested that China will continue to carry the highest gastric cancer burden through 2040.

Risk factors: infection, diet and lifestyle

Helicobacter pylori

The International Agency for Research on Cancer (IARC) placed Helicobacter pylori in Group 1, its category for agents with sufficient evidence of carcinogenicity in humans. That 1994 decision relied mainly on three cohort studies, which together gave an odds ratio of about 3.8 for gastric cancer in infected compared with uninfected people. Infection drives chronic gastric inflammation, and the bacterium also weakens the mucosa's DNA repair capacity and depletes antioxidants such as vitamins C and E.

Attributable fractions depend on the population. For non-cardia tumours, one estimate places the infection-related share at about 78% in China and 87% in low-risk countries such as Australia, Europe and the USA. A 2022 analysis of 27 countries estimated an average of roughly 70% for all gastric cancer, with values from 40.7% in Malaysia to 82.3% in South Africa. Japanese estimates put the ten-year cancer risk at about 5% among infected adults.

Epstein–Barr virus

Epstein–Barr virus (EBV) is present in roughly one in ten gastric cancers, where the viral genome is found in every tumour cell. EBV-positive tumours show the most extreme DNA hypermethylation of any cancer type studied, and they belong to a broad CpG island methylator phenotype. Recurrent PIK3CA mutations occur in 40–80% of these tumours, compared with 3–42% of other gastric cancer subtypes, and PD-L1 amplification and ARID1A alterations are also common. The virus maintains a restricted latency programme in which EBNA1 tethers the viral episome to host chromosomes. After adjustment for stage and prognostic factors, EBV positivity has been associated with lower mortality.

Diet

A dose-response meta-analysis of 76 prospective cohort studies, covering 32,758 cases among more than six million participants, provides the largest systematic assessment of diet and gastric cancer. It found positive associations for high-salt foods and inverse associations for total fruit and white vegetables, with the protective effect linked partly to vitamin C. Pooled estimates indicate that each 5 g/day increase in dietary salt raises risk by about 12%. Smoked and salted foods are consistently associated with higher risk, and pickled vegetables have been identified as a risk factor in several reviews, possibly because of their salt content and nutrient loss during fermentation. Much of the earlier evidence came from case-control designs, which are more vulnerable to recall bias than prospective cohorts.

Tobacco, alcohol and obesity

Pooled analyses show that current smoking is associated with an odds ratio of about 1.6 and former smoking with about 1.4 for stomach cancer. The association is strongest for cardia tumours, and risk declines gradually after cessation. Men show a larger relative increase than women in some meta-analyses.

Alcohol shows a consistent but modest association. A meta-analysis of 81 studies gave a pooled odds ratio of 1.20, with risk rising with daily intake. Beer and spirits, but not wine, were most strongly associated in the prospective cohort analysis. Heterogeneity between studies is substantial and varies by region and tumour subsite, and one review found higher risk among former than current drinkers, a pattern that may reflect people quitting because of illness.

For obesity, the evidence is more subsite-specific. A meta-analysis of 24 prospective studies found no overall association with overweight or obesity, but obesity was linked to a higher risk of cardia cancer, with a relative risk of about 1.8. Another analysis reported roughly a 10% higher risk for every 5-unit increase in body mass index, while a 2020 pooled analysis found no significant overall association for overweight or obesity.

Hereditary and genetic predisposition

Gastric cancer clusters in families, and having a first-degree relative (parent, sibling or child) with the disease raises an individual's risk, typically by about two- to three-fold in population studies. Part of this familial effect is environmental: relatives often share Helicobacter pylori infection acquired in childhood, as well as dietary habits. Inherited mutations of high penetrance explain only a small fraction of cases, and most familial risk is thought to reflect many common variants of small effect together with shared exposures.

Germline cancer syndromes

Hereditary diffuse gastric cancer (HDGC) is the best-characterised inherited form. It results from germline loss-of-function mutations in CDH1, which encodes the cell-adhesion protein E-cadherin. Carriers have a very high lifetime risk of diffuse-type gastric cancer, with some estimates near 70% in men and lower but still substantial risk in women, and tumours can appear at a young age. Diffuse lesions are hard to detect by endoscopy and infiltrate the stomach wall, so surveillance is considered unreliable, and several expert guidelines advise that carriers be considered for prophylactic total gastrectomy. CDH1 carriers also face an increased risk of lobular breast cancer.

Lynch syndrome is caused by germline mutations in mismatch-repair genes such as MLH1, MSH2, MSH6 and PMS2, or by deletions affecting EPCAM. It mainly predisposes to colorectal and endometrial cancer, but it also raises gastric cancer risk, with estimates varying widely between studies. Gastric tumours in these patients usually show mismatch-repair deficiency and microsatellite instability, which is why tumour testing is used to flag possible Lynch syndrome.

Li–Fraumeni syndrome, caused by germline TP53 mutations, produces an excess of sarcomas, breast cancers, brain tumours and adrenocortical tumours, and gastric cancer is among the additional cancers reported in affected families. Its clinical significance lies mainly in identifying families with early-onset, multiple cancers, for whom TP53 testing and broad cancer surveillance are considered. Familial adenomatous polyposis (FAP), caused by germline APC mutations, produces hundreds to thousands of colorectal adenomas. Gastric involvement includes fundic gland polyps and gastric adenomas, with gastric adenocarcinoma reported in a subset of patients, and the reported risk is higher in some Asian populations. In FAP, gastric findings influence endoscopic follow-up rather than cancer prevention alone.

Somatic changes within tumour cells

Most gastric cancers arise without an inherited predisposition, and their cells accumulate somatic mutations and chromosomal or genomic changes over time. TP53 is among the most frequently altered genes, and sporadic diffuse-type tumours often carry CDH1 alterations or RHOA mutations. The Cancer Genome Atlas project (2014) proposed a molecular classification of four groups: Epstein–Barr virus (EBV)-positive tumours, microsatellite unstable (MSI) tumours, genomically stable tumours and chromosomally unstable tumours.

Microsatellite instability reflects defective mismatch repair, which leads to a high burden of errors in repetitive DNA sequences. In sporadic cases it is most often caused by epigenetic silencing of MLH1 through promoter methylation rather than by a germline mutation. MSI tumours make up a minority of gastric cancers, and their many altered sequences produce abnormal peptides that can be recognised by the immune system, which is one reason they respond to immune checkpoint inhibitors and are tested for as a biomarker.

EBV-associated tumours constitute roughly one in ten gastric cancers. They are characterised by PIK3CA mutations, widespread CpG-island hypermethylation of DNA and frequent amplification of genes encoding PD-L1 and PD-L2 and of JAK2, and they often display a dense lymphoid infiltrate. These molecular features distinguish them from the other subtypes and have prompted their separate consideration in treatment research.

Pathogenesis and the stepwise model of carcinogenesis

Chronic gastritis and the inflammatory response

Most intestinal-type gastric adenocarcinomas are thought to begin with chronic inflammation of the gastric mucosa, most often caused by Helicobacter pylori infection and less often by autoimmune gastritis. The bacterium's lipopolysaccharide engages Toll-like receptor 4 on mononuclear cells, and its virulence protein CagA activates the NLRP3 inflammasome. Through activation of the NF-κB pathway, infected epithelial and immune cells release pro-inflammatory cytokines such as IL-1β, TNF-α, IL-6 and IL-8. IL-1β is of particular interest because it is also a potent inhibitor of gastric acid secretion, and at high concentrations TNF-α can injure the mucosa. Infection additionally disturbs the balance between regulatory T cells and Th17 cells, increasing the output of IL-17 and TGF-β, which in turn feed into IL-6 signalling. Inflammation of this kind is persistent rather than self-limiting, and it provides the continuing pressure that drives the later changes in the mucosa. Genetic variation in the IL-1 and TNF genes has been studied as a modifier of risk, but pooled analyses of individual polymorphisms have produced inconsistent results.

The stepwise progression

The Correa model, proposed by Pelayo Correa in 1975, describes intestinal-type cancer as the end point of a sequence of progressive mucosal changes. It was revised in 1988 and 1992. In its later form the sequence runs from normal mucosa to non-atrophic gastritis, then to multifocal atrophic gastritis, complete intestinal metaplasia (small-intestine type), incomplete intestinal metaplasia (colonic type), low-grade dysplasia, and high-grade dysplasia, which is equivalent to carcinoma in situ. Atrophy, the first specific step, is the loss of normal glandular tissue. Metaplasia is the replacement of gastric epithelium by mucin-secreting intestinal-type epithelium. Dysplasia is the stage at which the cells show clear neoplastic features but have not yet invaded beyond the basement membrane.

Estimated annual progression rates from the precursor stages were derived largely from Western cohorts. These estimates are about 0.1% a year for atrophic gastritis, 0.25% for intestinal metaplasia, 0.6% for mild-to-moderate dysplasia, and 6% for severe dysplasia. The cascade is not strictly linear. Lesions can stagnate or regress, and the rate of progression varies widely between populations and individuals. Because of this uncertainty, the risk attached to atrophy alone remains debated, and surveillance strategies differ between countries.

Molecular changes that drive the transition

At the molecular level, the transition from metaplasia to cancer involves successive genetic and epigenetic alterations. Gastric epithelial cells gradually lose their native phenotype and are replaced by cells with an intestinal phenotype, which acquire increasing autonomy and then develop dysplastic and carcinomatous features. Mucin expression changes characteristically: complete intestinal metaplasia is marked by MUC2, whereas incomplete metaplasia shows MUC2, MUC5AC and MUC6. Alterations in TP53 and loss of heterozygosity, along with changes in GATA6 and MUC6, have been associated with outcomes in early gastric cancer.

Epigenetic silencing is another important route. H. pylori infection has been associated with atypical DNA methylation of the gastric mucosa. In one study, IL-6 and IL-10 activated the JAK/STAT pathway, and this was accompanied by hypermethylation of the promoter of the suppressor of cytokine signalling 1 gene, which silenced that gene. Several other pathways, including NF-κB, MAPK, PI3K/AKT, Hippo and JAK/STAT, are activated during infection and together promote cell proliferation, immune evasion and tissue remodelling. Changes in the balance between proliferation and apoptosis are central to this process: the cytokine environment can alter the expression of death receptors on epithelial and tumour cells, which affects whether damaged cells are removed or survive.

Intestinal-type and diffuse-type tumours

Intestinal-type tumours, which form cohesive glandular structures, are the tumours that most often follow the Correa cascade. Their development typically takes decades and occurs mainly in older adults. Diffuse-type tumours, which consist of poorly cohesive cells including signet-ring cells, differ in origin. They arise more often without the visible atrophy and intestinal metaplasia that precede intestinal-type cancer, and their precursor lesions are less commonly recognised. They can occur in younger adults, and their progression is generally faster. Their development is closely linked to loss of E-cadherin function, which is encoded by the CDH1 gene, and the hereditary form of this disorder is discussed elsewhere in this entry. Because the two types follow different routes, the Correa framework explains intestinal-type disease much better than diffuse-type disease.

Clinical presentation and symptoms

Silent early disease and delayed diagnosis

Early gastric cancer, in which the tumour is confined to the mucosa or submucosa, often causes no symptoms at all. When symptoms do appear, they are usually nonspecific, such as persistent indigestion, heartburn or a feeling of upper abdominal discomfort. Patients and physicians readily attribute these complaints to gastritis, peptic ulcer or a minor infection, so the tumour often remains undetected until it has invaded deeper layers or spread. Clinical descriptions from the early twentieth century already made this point, noting that a previously healthy adult with new digestive complaints should arouse suspicion. Because the symptoms of early disease mimic benign conditions, delay in seeking care is common, and the most reliable route to early detection in high-incidence countries is systematic endoscopic screening, discussed elsewhere in this article.

Typical presenting features

Epigastric pain is among the most frequent complaints. In early disease it is typically vague and dull rather than sharp, and it tends to become more frequent and severe as the tumour grows. Some patients first notice pressure or fullness after meals rather than true pain. Early satiety, meaning a sense of fullness after a small meal, and anorexia are also common, and an aversion to food, particularly meat, was described in older case series as an important early sign.

Weight loss is one of the most common presenting features and has several contributing mechanisms. Reduced food intake from anorexia, pain and early satiety lowers calorie intake, while the tumour and the accompanying inflammatory response, mediated in part by cytokines, increase metabolic demand. Nausea, vomiting, bloating and belching are further common complaints.

Bleeding is frequently occult. Fewer than one in five patients present with overt bleeding, either as black, tarry stools (melena) or as vomiting of blood (haematemesis). More often, slow blood loss from the tumour surface produces iron-deficiency anaemia, which may first appear as fatigue, weakness or breathlessness on exertion. In some younger patients the illness can begin abruptly with haematemesis, followed by persistent vomiting and rapid weight loss.

Symptoms that depend on tumour location

Symptoms are shaped by where the tumour sits within the stomach. Tumours of the cardia and gastro-oesophageal junction often cause dysphagia, typically difficulty swallowing solid food that progresses over time, and may resemble oesophageal cancer clinically. Tumours of the antrum and pylorus are more likely to obstruct the gastric outlet, causing vomiting of food eaten earlier in the day and marked early satiety. Pyloric involvement is generally associated with the most constant and copious vomiting. Tumours that involve only one curvature of the stomach may produce little or no vomiting, which can delay recognition.

As disease advances, obstruction becomes more prominent. Persistent vomiting, a palpable abdominal mass and clinical signs of bowel obstruction may appear, and physical examination in advanced disease may reveal cachexia. A palpable mass is not found in most early cases but may be detected in advanced disease.

Paraneoplastic and systemic signs

Several signs arise from effects beyond the local tumour. Acanthosis nigricans, a condition of velvety, hyperpigmented and thickened skin in body folds such as the axillae, neck and groin, is the best-known cutaneous marker, and the malignant form is most often linked to gastric adenocarcinoma. Its skin changes may appear before gastrointestinal symptoms. Related signs include tripe palms, a thickening of the palmar skin with exaggerated creases, and the sign of Leser–Trélat, a sudden eruption of numerous seborrhoeic keratoses. Skin lesions of this kind can regress after successful treatment of the tumour, which is why they are regarded as clinically useful.

Systemic paraneoplastic phenomena include Trousseau syndrome, in which recurrent or migratory venous thrombosis signals a hypercoagulable state, and disseminated intravascular coagulation. Advanced disease can also produce a firm, enlarged left supraclavicular lymph node (Virchow's node), an umbilical nodule, and ascites. Spread to the liver may cause hepatomegaly and, when the liver is extensively involved, jaundice.

Differences between early and advanced presentation

Early tumours are usually discovered incidentally, during endoscopy performed for dyspepsia or during screening, and they seldom present with biopsy-proven symptoms that demand urgent attention. Advanced tumours present with weight loss, anaemia, bleeding, obstruction, or signs of peritoneal and hepatic spread. The gap between these two presentations is the principal reason that clinical symptoms alone are an unreliable guide to early disease, and why persistent or worsening digestive complaints in an adult warrant prompt endoscopic evaluation.

Diagnosis and investigations

Diagnosis depends on direct inspection of the stomach combined with tissue sampling. Upper endoscopy (gastroscopy) lets the endoscopist see the mucosa, judge the size and position of a lesion and take targeted biopsies. A pathologist then establishes the histological type, most often adenocarcinoma, and separates malignancy from benign ulceration or a lymphoma. Because early lesions can look subtle, endoscopists use enhancement techniques. Chromoendoscopy applies dyes such as indigo carmine or acetic acid to highlight irregular or depressed surface patterns. Narrow-band imaging uses filtered light to emphasise mucosal and vascular detail, and magnifying endoscopy shows the microsurface and microvascular architecture at high power. Many guidelines advise sampling several sites within a suspicious area, because a single specimen can miss cancer that is present elsewhere in the lesion.

The barium upper gastrointestinal series, in which the patient swallows a radiopaque suspension and the stomach is examined with fluoroscopy and plain films, was once a standard way to detect ulcers and masses. It shows the outline of the stomach wall and filling defects, but it cannot assess mucosal colour or texture, cannot obtain tissue and misses many small or flat lesions. Endoscopy has therefore replaced it in most settings. Barium studies still have a niche role where endoscopy is unavailable or when a stricture needs to be mapped.

Serum tumour markers, including carcinoembryonic antigen (CEA), carbohydrate antigen 19-9 (CA19-9) and CA72-4, are measured in suspected cases, but they cannot establish a diagnosis on their own. In one series of patients with gastric cancer, each marker alone reached a sensitivity of only about 20 to 28 per cent, rising to roughly 48 per cent when combined. Levels are typically higher in advanced disease and metastasis, and they can also be raised in benign conditions or normal in many patients with cancer. Their main value is therefore supplementary: they help monitor treatment response and recurrence, and they add information to the endoscopic picture rather than replacing it. Some studies report that CEA positivity is more frequent in cardia tumours.

Imaging is used after the diagnosis is confirmed, to define the local and systemic extent of disease. Contrast-enhanced CT of the chest, abdomen and pelvis is the mainstay of preoperative assessment. It shows wall thickening, invasion of adjacent organs, enlarged nodes and metastases in the liver, peritoneum and lungs. Its limitations are notable: in one review, CT's sensitivity for nodal disease was about 78 per cent and its specificity about 62 per cent, and it is weak at detecting early, shallow tumours.

Endoscopic ultrasonography (EUS) passes a high-frequency probe through the endoscope to produce layered images of the wall. It is the most useful method for T staging, particularly for separating mucosal and submucosal lesions from deeper invasion, a distinction that determines whether endoscopic resection is appropriate. Its drawbacks are that a tumour causing stenosis may prevent the probe from passing, and that accuracy depends on the operator. In one clinical series, EUS was slightly more accurate than PET-CT for nodal staging (76.2 per cent versus 72.5 per cent), while PET-CT was more reliable for visceral metastases. PET-CT uses a glucose analogue to show metabolically active tumour and can reveal occult distant disease that changes management. It is not routine for primary staging, however, and it is less sensitive for small nodal deposits. Laparoscopy and MRI are used selectively, for example to check peritoneal spread in potentially resectable disease.

Staging systems

The TNM framework

Gastric cancer is staged with the tumour–node–metastasis (TNM) framework. The current staging standard follows the 8th edition TNM classification, jointly established by the UICC and AJCC, which mainly assesses how deeply the tumour invades, how far lymph node metastasis extends, and whether distant spread is present. Those three parameters are then combined into stage groups. Integrating them places gastric cancer into stages 0 through IV, which guide treatment decisions and prognosis.

The T category depends on the depth of penetration through the wall. In situ carcinoma remains within the epithelium, whereas T1 lesions reach the lamina propria, muscularis mucosae or submucosa, and T2 lesions extend into the muscularis propria, with T3 and T4 reserved for deeper or organ-invading growth. T1 is further split into T1a, for invasion of the lamina propria or muscularis mucosae, and T1b, for invasion of the submucosa. The N category counts involved regional nodes. In the 8th edition, the N3 group is divided into N3a (7–15 nodes) and N3b (16 or more). The Japanese classification followed suit: its greatest change was the separate inclusion of N3a and N3b.

Stage grouping and its consequences

The 8th edition reshuffled several combinations compared with the 7th. T1N3b and T2N3b were upstaged from IIB or IIIA in the 7th edition to IIIB. T3N3b was moved up from IIIB to IIIC. T4bN0 and T4aN2 were moved down from IIIB to IIIA. Because these changes mostly affected stage III, the revision produced clearer prognostic separation. In one Korean single-institution series, 5-year overall survival was 40.8% for stage IIIB and 20.2% for stage IIIC under the 8th edition, whereas the 7th edition showed no significant difference between the same groups. A SEER analysis reported a similar pattern, with stages IIIB and IIIC of the 7th edition being divided in the 8th edition with clearly different survival rates. The same analysis found that 8th-edition stage IIIC and stage IV had similar survival.

Clinical, pathological and surgical staging

Staging is completed in stages. Clinical staging is assigned before treatment, using endoscopy, imaging and other preoperative tests. The Japanese framework distinguishes clinical staging, the staging before definitive surgery, and surgical staging, which is determined during the operation. Pathological staging was introduced later and requires microscopic proof. Pathological staging therefore depends on examination of the resected stomach and its lymph nodes, which is why a substantial part of the final stage is only known after surgery. Molecular or immunohistochemical features are not practical for most gastric cancers, so staging still relies on TNM classification.

After neoadjuvant treatment, a prefix is used. Tumours classified after preoperative chemotherapy or chemoradiotherapy carry the prefix y, with clinical classification written as ycTNM and pathological classification as ypTNM. Only viable tumour cells are counted when calculating ypTNM. Peritoneal involvement is also treated specially in Japan, where free cancer cells found in the peritoneal cavity (CY1) are classified as M1, which generally rules out curative surgery under the guideline algorithm.

Laparoscopic staging addresses a gap in imaging. Computed tomography often misses small peritoneal deposits, so a thorough exploration of the abdominal cavity is carried out during staging laparoscopy to detect occult peritoneal metastases that preoperative imaging may have missed. In a systematic review, staging laparoscopy outperformed computed tomography in sensitivity, specificity, and both predictive values. In a 400-patient cohort, occult peritoneal metastases were detected in 20% of patients. Its benefit is also practical: it can increase the curative resection rate and decrease unnecessary laparotomies.

Staging and treatment selection

Stage determines therapy. In Japan, the patients eligible for the ACTS-GC adjuvant trial had pathological stage II, IIIA or IIIB disease after R0 gastrectomy with at least D2 lymphadenectomy, which shows how stage-based groupings define trial and guideline populations. Because stage groups, prognosis and the use of chemotherapy, chemoradiation or surgery are linked, changes to the classification can alter treatment eligibility as well as survival estimates.

Surgical and endoscopic treatment

Endoscopic resection for very early lesions

Endoscopic removal is used for gastric cancers confined to the mucosa or submucosa, where the risk of lymph node spread is very low. It is estimated that endoscopic treatments account for more than 60% of all treatments for early gastric cancer in Japan. The method grew out of national screening. Snare-based resection for early gastric cancer appeared in Japan in the 1990s as endoscopic mucosal resection (EMR), after screening programmes begun in the 1970s found more early tumours. The endoscopic treatment of these early lesions in Japan evolved as a logical consequence.

EMR had limits. Because a large lesion could not be removed in one piece, pathologists could not reliably judge tumour depth, and local recurrence rose in large lesions. Endoscopic submucosal dissection (ESD), using newly developed endoscopic knives, made it possible to remove a large lesion en bloc. Sources differ on when ESD was introduced: one review places its development in the late 1990s, while another says ESD was pioneered in Japan in the early 2000s as a method for en-bloc excision. ESD's advantages come with costs. ESD has a higher incidence of complications and requires greater endoscopic skill than conventional EMR.

Japanese guidelines set the eligibility criteria. For submucosal invasion, the expanded criteria for curative endoscopic resection include a size of 30 mm or less, differentiated-dominant histology, lack of vessel involvement, and submucosal invasion of less than 500 μm (SM1). Lesions outside these criteria generally require surgery.

Partial and total gastrectomy

When endoscopic treatment is not appropriate, surgery remains the core treatment. Distal tumours are usually treated with a partial (subtotal distal) gastrectomy, while proximal or extensive tumours may require total removal of the stomach. The extent of resection has functional consequences. In one comparative study, more dysphagia and eating restrictions were reported in patients after total gastrectomy when compared to patients after subtotal distal gastrectomy.

Lymph node dissection

Lymphadenectomy is central to curative surgery. For gastric cancer patients, surgical resection with en-bloc lymphadenectomy is the cornerstone of curative treatment. The extent of nodal clearance is classified as D1 (perigastric nodes) or D2 (perigastric plus named regional nodes along the major arteries). D2 dissection is standard in Japan and Korea, where it is combined with surgical training and high case volumes, while Western practice has historically varied more.

Reconstruction after gastrectomy

After distal gastrectomy, the surgeon must reconnect the gut. Common techniques for reconstructive surgery following gastrectomy include the Billroth I, Billroth II, and traditional Roux-en-Y approaches. Each has trade-offs. The Billroth II method is recognized for its lower propensity for anastomotic leaks due to reduced tension at the anastomotic junction, but it alters the flow of gastric juice, bile and pancreatic secretions. Billroth I has its own drawbacks: in a prospective cohort, Billroth I patients developed dumping syndrome more often than Billroth II/Roux-en-Y patients, who, in turn, experienced more bile reflux gastritis. A five-year randomised trial found that Billroth I and Roux-en-Y reconstructions were generally equivalent regarding body weight change, nutritional status, and QOL 5 years after distal gastrectomy, although Roux-en-Y more effectively prevented reflux esophagitis and diarrhea.

Minimally invasive surgery

Laparoscopic gastrectomy has been adopted widely, particularly in Asia, and its benefits have been tested in Western populations. The Dutch LOGICA trial randomised 227 patients to laparoscopic (n = 115) or open gastrectomy (n = 112) between 2015 and 2018. Laparoscopic surgery reduced blood loss (150 v 300 mL, P < .001) but lengthened operating time (216 v 182 minutes). Postoperative complications were 44% versus 42%, in-hospital mortality was 4% versus 7%, and 1-year overall survival was 76% versus 78%, with no significant differences between the groups. Hospital stay was also unchanged. A meta-analysis of randomised trials reached a similar conclusion for locally advanced disease, finding similar risks of short-term mortality and serious adverse events within 30 days after surgery, though the amount of evidence was low for most outcomes. The authors of the LOGICA trial noted that it remains unclear whether the results of these Asian studies can be extrapolated to the Western population.

Operative complications and centre experience

Outcomes after gastric surgery depend on the operating team and the institution. Randomised trials have therefore restricted enrolment to experienced units; LOGICA took place in 10 experienced upper GI centers in the Netherlands, and the Korean KLASS trial required that all the participants performed more than 50 cases of open and laparoscopic gastrectomies, respectively. Complication rates, anastomotic leaks, and perioperative mortality are generally understood to be lower in high-volume centres, although this relationship varies by procedure and setting.

Systemic and adjuvant therapies

Systemic treatment of gastric cancer is given before surgery (neoadjuvant or perioperative), after resection (adjuvant), or as first-line therapy for advanced or metastatic disease. In resectable disease, the evidence rests mainly on cytotoxic chemotherapy and, more recently, on immunotherapy. In advanced disease, drug choice increasingly depends on biomarker testing of the tumour.

Chemotherapy in resectable disease

The MAGIC trial, published in 2006, showed that adding epirubicin, cisplatin and fluorouracil before and after surgery raised five-year survival to 36%, compared with 23% for surgery alone. Oral capecitabine was later shown to be an equivalent substitute for infusional fluorouracil, and anthracycline-based triplets became the European standard for locally advanced disease. The FLOT4 trial, reported in 2019, compared fluorouracil, leucovorin, oxaliplatin and docetaxel (FLOT) with epirubicin-based regimens. It found a median overall survival of 50 months versus 35 months, and FLOT replaced the older triplets as the European standard. Retrospective real-world cohorts from several centres have reported survival broadly in line with the trial results. In East Asia, where the disease is often detected at earlier stages, postoperative adjuvant chemotherapy with oral S-1 has been established in Japan, and capecitabine-based regimens have been studied in Korea, following trials such as ACTS-GC and CLASSIC.

HER2-positive disease

Roughly a fifth to a third of gastric and gastro-oesophageal junction adenocarcinomas overexpress HER2, and testing by immunohistochemistry, with in situ hybridisation for equivocal cases, identifies candidates for anti-HER2 therapy. In the ToGA trial, adding the monoclonal antibody trastuzumab to fluoropyrimidine and platinum chemotherapy improved median overall survival to 13.8 months, compared with 11 months with chemotherapy alone. The KEYNOTE-811 trial then showed benefit from adding the PD-1 antibody pembrolizumab to trastuzumab and chemotherapy. The FDA granted accelerated approval for this combination in May 2021, and the European label was later restricted to tumours with a PD-L1 combined positive score (CPS) of at least 1. For disease that progresses after first-line therapy, trastuzumab deruxtecan, an antibody–drug conjugate, is recommended for HER2-positive tumours.

Immune checkpoint inhibitors and other molecular targets

Immune checkpoint inhibitors are used in both first-line and later settings. Nivolumab combined with fluoropyrimidine and platinum chemotherapy is a standard first-line option for HER2-negative tumours with PD-L1 CPS of at least 5, and pembrolizumab with chemotherapy is recommended for CPS of at least 10. Tumours with mismatch repair deficiency or high microsatellite instability are a separate group: immunotherapy can be given as a single agent because these tumours respond to checkpoint blockade without chemotherapy. In the perioperative setting, the MATTERHORN trial found that adding the PD-L1 antibody durvalumab to FLOT improved event-free survival, with a hazard ratio of 0.71.

Zolbetuximab, a monoclonal antibody against claudin 18.2, a tight-junction protein expressed in gastric epithelium, is the first agent targeting this molecule. In the SPOTLIGHT and GLOW trials, it improved progression-free and overall survival when added to chemotherapy in HER2-negative tumours. Eligibility requires immunohistochemical staining in at least 75% of tumour cells with moderate-to-strong membranous intensity. Regulatory approval followed in Japan and the European Union in 2024, and in the United States later that year.

Radiation therapy and chemoradiation

Radiation is used less often than chemotherapy but has defined roles. In the United States, the INT-0116 trial established postoperative chemoradiation with fluorouracil and leucovorin as a survival-improving option for high-risk resected disease, particularly node-positive or incompletely resected tumours. Its use is more limited in Japan and Korea, where surgery and adjuvant chemotherapy are emphasised. Radiation is also used with palliative intent to control bleeding, pain or obstruction in advanced disease, and to treat selected locally advanced or junctional tumours in combination with chemotherapy.

Prognosis and survival

Survival from gastric cancer depends above all on the stage at diagnosis. In United States registry data from the Surveillance, Epidemiology, and End Results (SEER) program for patients diagnosed between 2015 and 2021, the five-year relative survival rate was about 77% for localized disease, 37% for regional spread to nearby nodes or structures, and 8% for distant metastasis, with 38% across all stages combined. These figures reflect only the stage at first diagnosis and do not account for later progression or recurrence. Registry statistics also lag behind treatment changes, because they describe patients treated in earlier years.

Stage is not the only prognostic variable. Histological subtype, the depth of invasion and the number of involved lymph nodes all shape outcome, and the presence of peritoneal or hepatic spread is among the strongest adverse indicators. In one series of 354 patients treated by laparoscopic D2 gastrectomy, five-year overall survival for pathological stages I, II and III was 93.7%, 78.5% and 42.2%, respectively. In the same cohort, peritoneal recurrence was more frequent with deep (T4) or extensive nodal (N3) disease, tumours of 70 mm or more, vascular invasion and poorly differentiated histology. Molecular features, including HER2 status and the biomarkers used to select systemic therapy, are increasingly used to refine prognosis within a given stage, although their independent prognostic weight varies between studies.

International differences are substantial. Survival in Western countries has been reported at roughly 10–30% at five years, whereas figures from East Asia are much higher. In South Korea, the five-year survival rate for 2017–2021 was 77.9%, a rise of about 34% over the 1993–1995 period, which the national screening programme is credited with helping to produce. Japanese data show similar gains, and in a Tokyo-area cohort of patients found through population screening, five-year survival was 90% in men and 86% in women. These results reflect the high proportion of early tumours detected by endoscopy and radiography, rather than a biological difference alone. Population-level effects are less clear-cut: a synthetic control analysis estimated that Korea's nationwide programme was associated with a relative reduction of about 17% in gastric cancer mortality, rising to about 41% by the fifteenth year, while the corresponding estimate for Japan was not statistically distinguishable from no effect.

Recurrence remains the main cause of treatment failure after curative resection. In a multicentre Chinese cohort of 3,278 patients who underwent curative gastrectomy, about half developed recurrence during a median follow-up of 61 months. Risk was highest in the first year or two after surgery, with a peak near 13 months overall and earlier peaks in stage III disease. Peritoneal dissemination was the most common and most aggressive pattern, while locoregional recurrence tended to appear later. Late recurrence beyond five years is less frequent; one Seoul series recorded late relapse in about 9% of patients, most often locoregional or peritoneal. Current follow-up schedules are based largely on observational data. European and international guidelines generally recommend clinical review, blood tests and imaging at intervals of three to six months in the first year after surgery for advanced disease, and less frequently for early disease, with endoscopy after total gastrectomy performed mainly when symptoms are present. Randomised evidence on the value of intensive surveillance is lacking, and the psychological burden of frequent testing is a recognised concern.

Prevention and screening programmes

Eradication of Helicobacter pylori

The most direct preventive strategy targets the infection that drives most non-cardia cancers. The largest population trial comes from Linqu County in Shandong, China. It enrolled about 180,000 residents across 980 villages and assigned infected people to a four-drug antibiotic regimen, a symptom-only treatment, or no treatment, with uninfected people left untreated. After an 11.8-year follow-up (2011–2022), the trial confirmed a 13% reduction in gastric cancer incidence among all treated participants and a 19% reduction among those with successful H. pylori eradication. The benefit was larger in younger adults: a 35% reduction in incidence and a 43% reduction in mortality among individuals aged 25–45 years.

Pooled analyses give a consistent but not uniform picture. A 2025 meta-analysis in <i>Gastroenterology</i> found that, among healthy infected adults without neoplasia at baseline, eradication was associated with a 36% relative risk reduction for future gastric cancer and a 22% reduction in the relative risk of gastric-cancer mortality, but no effect on all-cause mortality. Another systematic review reported a reduction in incidence but no clear effect on gastric cancer-specific mortality. Long-term follow-up also shows timing and selection effects. In a placebo-controlled trial followed for 26.5 years, participants receiving treatment had a lower incidence of gastric cancer compared with placebo (HR 0.57). However, the favorable effects were only evident among those without precancerous gastric lesions and those without dyspepsia symptoms at baseline. A Japanese cohort found a short-term rise in risk in the first year after eradication, with protection appearing only after several years.

Endoscopic screening in high-incidence countries

Japan and South Korea are the two countries with nationwide gastric cancer screening programmes, and their evidence is largely observational. In a Japanese population-based prospective study of 80,272 participants, compared with the unscreened group, gastric cancer mortality was significantly lower, 37% and 61%, in the radiographic and endoscopic screening groups respectively. A community-based case-control study in Tottori and Niigata reported roughly a 30% lower mortality among people screened endoscopically within 36 months before diagnosis.

Korea launched its programme in 1999 and expanded it in 2002. The programme was extended to all adults aged 40 years or older, and each participant underwent either endoscopy or an upper gastrointestinal series every two years until 2017. A nested case-control study of about 16.6 million people found an overall reduction in gastric cancer mortality of 21%, and a reduction in gastric cancer mortality of 47% for upper endoscopy screening in individuals aged 40–74 years. The effect rose with repeated examinations, and no associations were found for UGI series. Japanese screening, by contrast, has conventionally started at age 50 under its national guidelines, so the age ranges differ between the two countries.

Screening also carries harms. Japanese guidance notes that the complication rates were higher in endoscopic screening than in radiographic screening, and that evidence of the H. pylori antibody and serum pepsinogen tests was insufficient.

Cost-effectiveness, health-system barriers and primary care

Modelling studies have tested which screening schedules offer value. A Japanese microsimulation found that scenarios simulating the current national endoscopic screening guidelines in Japan, namely biennial and triennial endoscopic screening from age 50 with no stopping age, were not on the efficient frontier. Yet the current biennial programme would prevent 80.5% of gastric cancer mortality relative to no screening, at a substantially higher cost per 1000 simulated individuals than some alternatives. In Korea, a 2025 Markov model concluded that endoscopy-only screening was more cost-effective method than UGIs for the NCSP, and that a three-year interval with an upper-age limit of 69 years was the most cost-effective strategy. The lower sensitivity of barium studies explains part of this result, since patients with false-negative results may lose the chance of early detection.

Adding H. pylori testing is an open question. Previous studies have reported that screening for H pylori is a cost-saving intervention that can significantly decrease gastric cancer burden in areas with a high prevalence of H pylori infection. However, no study has examined whether incorporating H pylori screening into national endoscopic screening is cost effective. Health-system capacity is a practical constraint for endoscopic programmes. Korea, for example, began recommending endoscopy in medical institutions through its national endoscopic quality assessment programme in 2009. Primary care clinicians are generally positioned to identify high-risk individuals, such as those with a family history or known infection, for referral, although the evidence for specific selection criteria remains limited.

Dietary and environmental measures, such as reducing salt intake and tobacco use, are discussed in the risk-factor section. Their population effects are harder to quantify than those of eradication, because they are difficult to randomise and are usually evaluated in observational studies.

Research history

Systematic study of gastric cancer began with pathology. Nineteenth- and early twentieth-century pathologists described stomach tumours mainly by their gross appearance, but the modern framework came in 1965, when Per Lauren proposed a two-part histological scheme. Laurén divided the histology of gastric cancer into two groups, i.e., the intestinal type and the diffuse type; later, the indeterminate type was included to describe an uncommon histology. The intestinal type forms cohesive cells which form gland-like structures, whereas the diffuse type consists of tumor cells lack cell-to-cell interactions. The diffuse and intestinal Lauren types account for approximately 85% of gastric carcinomas. The scheme's durability is notable: it is still widely accepted and employed by pathologists and physicians today.

The discovery of Helicobacter pylori rewrote the causal picture of stomach disease. In Perth, Western Australia, the pathologist Robin Warren noted that small curved bacteria colonizing the lower part of the stomach (antrum) in about 50% of patients from which biopsies had been taken, and he saw that signs of inflammation were always present in the gastric mucosa close to where the bacteria were seen. Barry Marshall, a young clinical fellow, joined the work, and the culture of the organism is traditionally dated to Easter 1982, when Marshall was first able to culture H. pylori from biopsy material in the laboratory. Medical opinion at the time held that stress and lifestyle factors were the major causes of peptic ulcer disease, and the findings were initially met with scepticism and criticism. In a 1983 letter to The Lancet, Marshall suggested that the bacteria may have a part to play in other poorly understood, gastritis associated diseases (i.e. peptic ulcer and gastric cancer). Warren and Marshall also developed a rapid urease test and a urea breath test for diagnosis, as a series of important manuscripts from their collaboration. The 2005 Nobel Prize in Physiology or Medicine was awarded jointly to Marshall and Warren, and H. pylori is the first bacterium to be associated with cancer.

Epidemiological work on migration and diet provided the evidence that gastric cancer had environmental rather than purely inherited causes. Studies of Japanese emigrants in the United States and Hawaii, including the series by Haenszel and Kurihara published from 1968, compared incidence across migrant generations and with rates in Japan. Rates among Japanese-Americans were lower than in Japan and declined in successive generations, moving toward those of the surrounding population. These findings, together with the geographic contrasts in incidence, supported the view that diet and other exposures, such as the consumption of salted, smoked and pickled foods, were central to risk. Changes in food preservation, notably the spread of refrigeration, were later proposed as one reason for the long-term decline in some populations.

The stepwise model of carcinogenesis was formalised by Pelayo Correa, who in the early 1990s described gastric cancer as a multistep process running from chronic gastritis through atrophy and intestinal metaplasia to dysplasia and carcinoma. This framework linked the histological categories to a sequence of precursor lesions and gave the Lauren subtypes a developmental context. Subsequent work moved beyond histology. The Lauren classification system comprises two subtypes (intestinal and diffuse) and the WHO classification has four subtypes (papillary, tubular, mucinous and poorly cohesive). Neither system is particularly clinically useful as their prognostic and predictive capabilities cannot adequately guide patient management. Large-scale genomic studies addressed this limitation. The Cancer Genome Atlas project analysed 295 primary gastric adenocarcinomas and proposes a new classification of gastric cancers into 4 subtypes, and these subtypes were defined as Epstein–Barr virus (EBV)-positive, microsatellite instability (MSI), chromosomal instability (CIN), and genomically stable (GS). Later, The Cancer Genome Atlas (TCGA) project and Asian Cancer Research Group (ACRG) recently categorized gastric cancer into molecular subtypes.

Unresolved questions and ongoing debates

Helicobacter pylori is the best-established cause of gastric cancer, but it does not explain every case, and published estimates of the uninfected share differ. One source notes that the attributable fraction of H. pylori for gastric cancer has been estimated at about 70%, while another states that almost all cases of gastric cancer are caused by H. pylori infection. The gap partly reflects how tumours are classified. Some cancers arise in stomachs with no infection history and no inflammation, atrophy or intestinal metaplasia on endoscopy, which does not fit the classic Correa sequence that begins with infection-driven gastritis pylori-negative mucosa without inflammation, atrophy, or intestinal epithelialization is rarely found. We report on two cases of advanced gastric cancer without endoscopic evidence of gastric mucosal atrophy. The cardia is the clearest example: tumours there are most of them not linked to H. pylori infection, unlike those found in the distal part of the stomach, and some are thought to resemble reflux-related oesophageal adenocarcinoma.

Candidate causes for uninfected cancers include diet, smoking, host genetics and, in a small subset, Epstein–Barr virus. Heterogeneity in these tumours is attributed to demographic features, lifestyle, high salt with nitrate intake, race and genetic variables, and Epstein–Barr-associated lymphoepithelioma-like carcinoma accounts for about 5% of gastric cancer. The infection itself is also not sufficient on its own. It is not clear why only some H. pylori-infected individuals develop gastric cancer, and host susceptibility, environmental factors and strain virulence are all thought to influence the outcome. Whether the remaining cases share a single mechanism, or represent several distinct diseases, remains unresolved.

A second debate concerns the rise of gastric cancer in younger adults. In the United States, the incidence of gastric cancer has been rising steadily in younger adults, even while it has been declining among older adults for the past 3 decades. One analysis found that early-onset cases fell by 1.9% a year from 1973 to 1995 and then rose by 1.5% a year from 1995 to 2013, and that their share of all gastric cancers grew from a nadir of 18.4% in 1990 to >30.0% in every year since 2012. Part of this proportional increase is arithmetic, since older-age incidence has fallen, but the absolute rise in younger patients is itself the focus of inquiry. Most early-onset gastrointestinal cancers are sporadic, or without a known cause, and a review in JAMA linked most of them to lifestyle factors such as obesity, lack of exercise, poor diet, cigarette smoking or alcohol consumption. Still, hereditary variants are found in 15% to 30% of people with early-onset gastrointestinal cancer, and this share cannot account for the pace of the rise by itself. Detection is a further confound: early-onset gastric cancer often goes undetected because screening guidelines rarely start before age 40, so apparent increases may partly reflect when and how patients are examined.

Screening raises questions about target populations and starting ages. Early-onset disease is not trivial in scale: one trial registry describes it as comprising 20-30% of new GC diagnoses, mainly among individuals aged 30-40 years. Any age threshold therefore involves trade-offs between detecting these cancers earlier and the cost, procedural burden and false-positive findings that come with examining larger, lower-risk populations. Serological tools have been used in research to define risk groups. For example, one study classed patients as having atrophic gastritis when serum pepsinogen I was below 70 ng/dl and the PG I/II ratio was under 3.0, but the performance of such markers in population screening has yet to be settled. Trials of blood-based tests, such as a liquid biopsy signature to detect early-onset gastric cancer, are still in the research stage.

Artificial intelligence is being developed to improve endoscopic detection, which is a difficult task. Endoscopy is currently the most effective method to detect early gastric cancer, but endoscopy is not infallible and early gastric cancer can be missed, partly because early GC mimics atrophic gastritis in the background mucosa. In one retrospective validation using 500 still images, a computer-aided detection system reported a sensitivity of 0.93 and a specificity of 0.87, and its assistance improved the results of less experienced endoscopists, with the largest gains among beginners accuracy (0.88, 95% CI: 0.85 to 0.91), sensitivity (0.93, 95% CI: 0.88 to 0.98), specificity (0.87, 95% CI: 0.84 to 0.90), and AUC (0.962). Such studies are not yet generalisable. Reviews identify data quality, false positives and negatives, geographical biases, and regulatory barriers as obstacles, and one external multicentre evaluation found accuracy of 74.1%, with no significant difference from real-time diagnoses by institution-based specialists. Validation across diverse populations, equipment and hospital settings is therefore a central requirement before these tools can be widely adopted.