Gastric cancer: Pathology and molecular pathogenesis

INTRODUCTION — Gastric adenocarcinomas represent a clinically, biologically, genetically, and microscopically heterogeneous group of malignant epithelial tumors resulting from various environmental and genetic causes.

This topic will discuss the pathology and molecular pathogenesis of gastric cancer. The epidemiology, risk factors, clinical features, diagnosis and staging of gastric cancer are discussed elsewhere. (See "Epidemiology of gastric cancer" and "Risk factors for gastric cancer" and "Clinical features, diagnosis, and staging of gastric cancer".)

EPIDEMIOLOGY — Worldwide, gastric cancer is the fourth most common cancer, accounting for 6 percent of total cancer incidence, and is the third leading cause of death, accounting for 8 percent of cancer-related deaths, according to data from the World Health Organization (WHO) GLOBOCAN database. Despite a steady decline in the rates of incidence and mortality observed worldwide for several decades, this trend has been lessened, particularly in the non-Hispanic White population [1,2]. In this group, particularly for females, the risk may be related to autoimmune gastritis [1]. (See "Epidemiology of gastric cancer", section on 'Mortality'.)

The distribution of gastric cancer is markedly varied worldwide, with the areas presenting the highest rates of incidence (>60 per 100,000 males) found in Eastern Asia (Korea), Eastern Europe, and Central and Latin America. Conversely, some geographic zones have low incidence rates (<15 per 100,000 population), such as North America, Northern Europe, and Africa [3]. Morphologic differences also align with this epidemiologic diversity. The intestinal type of adenocarcinoma is predominant in high-risk areas, while the diffuse type is relatively more common in low-risk areas. Also, distal cancers of the antrum and pylorus tend to be most common in high-risk countries, while cancers of the cardia are most common in Western nations of low incidence [4]. (See "Epidemiology of gastric cancer", section on 'Regional variation'.)

Sex and age distribution — Males and females are not equally affected; incidence rates are approximately twofold higher in males than in females. As an example, in data from the GLOBOCAN database, the cumulative risk for gastric cancer to age 74 for males versus females in 2018 was 1.87 versus 0.79 percent. However, in younger individuals, the male:female ratio is approximately equal or even shows a female predominance [5,6].

Gastric carcinoma is extremely rare in persons under the age of 30 years [3]. In younger individuals, poorly cohesive diffuse-type tumors are predominant [7-9]. (See "Epidemiology of gastric cancer", section on 'Incidence'.)

ETIOLOGY AND RISK FACTORS

Helicobacter pylori — Helicobacter pylori infection plays an important role in gastric carcinogenesis. It is estimated that at least 89 percent of all non-cardia (figure 1) gastric cancers are caused by H. pylori [10]. Given the central role that H. pylori has in the development of gastric cancer [4], in 1994, the International Agency for Research on Cancer (IARC) recognized chronic infection with H. pylori, commonly acquired during early childhood, as a primary cause of gastric adenocarcinoma [11]. Left untreated, H. pylori infection initiates a slow carcinogenic process, developing over decades and eventually leading to antral-predominant chronic active gastritis and ultimately to multifocal atrophic gastritis, which is a risk factor for intestinal and, less commonly, diffuse gastric adenocarcinomas [12]. The epidemiologic data linking H. pylori to gastric cancer are reinforced by animal models of H. pylori infection, in particular in Mongolian gerbils [13,14]. (See "Association between Helicobacter pylori infection and gastrointestinal malignancy", section on 'Gastric cancer'.)

Nevertheless, only a small minority of individuals infected with H. pylori develops gastric cancer, and it is thought that modulation of the effects of chronic infection by genetic susceptibility, environmental factors, and H. pylori bacterial strain differences all influence the evolution into a neoplastic or nonneoplastic process [15-17]. H. pylori virulence factors that vary between strains, namely the cytotoxin-associated gene A (CagA) and vacuolating cytotoxin A (VacA), appear to influence the pathogenicity of the bacteria and the risk of gastric precancerous lesions and adenocarcinoma [18]. The variation in virulence factors relates to higher degrees of mucosal inflammation and progression of precancerous lesions [19,20]. (See "Pathophysiology of and immune response to Helicobacter pylori infection", section on 'Bacterial strain differences'.)

Aspects of host genetic susceptibility that affect the inflammatory response also impact the gastric cancer risk associated with H. pylori infection. As an example, polymorphisms in genes encoding interleukin 1 (IL-1) beta, the IL-1 receptor antagonist, tumor necrosis factor (TNF) alpha, and IL-10 have been implicated in the initiation and modulation of the inflammatory response and have been associated with susceptibility to carcinogenesis [17,21-25]. (See "Pathophysiology of and immune response to Helicobacter pylori infection", section on 'Inflammatory response'.)

Gland-forming adenocarcinomas (ie, those of the tubular, papillary, mucinous, and mixed types) are causally related to H. pylori. The poorly cohesive histologic type (corresponding to Lauren's classification of "diffuse gastric cancer") is more likely to result from hereditary characteristics modulated by environmental influences, while H. pylori is probably limited to a subset of sporadic cases in this subgroup [5,26,27]. (See 'Hereditary diffuse gastric cancer' below.)

In contrast to non-cardia gastric cancers, there appears to be a strong inverse relationship of H. pylori with cancers of the esophagogastric junction (EGJ) and cardia, and their precursor lesion, Barrett's esophagus [28-30].

The preneoplastic cascade — While H. pylori infection usually starts in infancy or early childhood, there is a long latency period, and cancers are clinically diagnosed four or more decades later. During this period, a prolonged precancerous process takes place, represented by a "cascade" of sequential histopathologic stages: chronic active nonatrophic gastritis, multifocal atrophic gastritis, intestinal metaplasia (complete, then incomplete), dysplasia, and invasive carcinoma in a high-incidence area [15,31]. The risk of gastric cancer in patients with these various premalignant lesions was addressed in two population-based studies:

●In a nationwide cohort study from the Netherlands, atrophic gastritis, intestinal metaplasia, mild-moderate dysplasia, and severe dysplasia were associated with annual incidences of gastric cancer of 0.1, 0.25, 0.6, and 6 percent, respectively [32].

●In another cohort study from Sweden, among patients who undergo gastroscopy with biopsy for clinical indications, approximately 1 in 85 with gastritis, 1 in 50 with atrophic gastritis, 1 in 39 with intestinal metaplasia, and 1 in 19 with dysplasia will develop gastric cancer within 20 years [33].

Endoscopic surveillance strategies for those with gastric intestinal metaplasia, atrophic gastritis, and dysplasia are discussed in detail elsewhere. (See "Gastric intestinal metaplasia", section on 'Management' and "Metaplastic (chronic) atrophic gastritis", section on 'Endoscopic surveillance in selected patients'.)

Nonatrophic gastritis — The first stage of chronic H. pylori infection is a nonatrophic gastritis, which predominates in the antrum and is characterized by a dense, band-like infiltrate of lymphocytes, macrophages, and plasma cells (picture 1). Occasionally, lymphoid follicles are formed. The term "activity" refers to the presence of focal acute inflammation in a background of chronic gastritis. The presence of polymorphonuclear neutrophils (PMNs) correlates closely with active colonization of the gastric lumen by H. pylori organisms.

It is important to emphasize that nonatrophic antral gastritis, which is the predominant finding in patients with an H. pylori-associated duodenal peptic ulcer, is not related to an increased cancer risk. (See "Association between Helicobacter pylori infection and duodenal ulcer".)

In a subset of patients, atrophy with intestinal metaplasia will develop, beginning a sequence of events that may culminate in adenocarcinoma; the rate of progression to atrophic gastritis/intestinal neoplasia in large cohorts is 0.1 to 0.9 percent [34]. The prevalence of intestinal metaplasia in the community among H. pylori-affected patients varies significantly from region to region, with a possible correlation between intestinal metaplasia prevalence and the availability of tobacco [35].

Atrophic gastritis — Atrophic gastritis is an advanced stage characterized by multifocal loss of the original gastric glands (picture 2). In most patients, active gastritis does not lead to gland loss, but when it does, it is the first histopathologic lesion of the preneoplastic cascade related to H. pylori. (See "Metaplastic (chronic) atrophic gastritis".)

Several classifications of chronic gastritis have been developed in an attempt to combine topographic, morphologic, and etiologic information into a reporting system including both grading and staging of gastritis, while other systems, like the Operative Link for Gastritis Assessment (OLGA), rank the histologic phenotypes of gastritis along a scale of increasing mucosal atrophy, with high-stage disease being associated with a high risk of gastric cancer [36]. There is no one universally accepted classification system that is prognostically useful. This subject is discussed in detail elsewhere. (See "Gastritis: Etiology and diagnosis", section on 'Etiology and classification'.)

Intestinal metaplasia — Multifocal atrophy involving the gastric mucosa may be followed by the development of glands presenting an intestinal phenotype. In intestinal metaplasia, the original gastric mucins, which have a neutral pH, are replaced by acid mucins, which may be sialic or sulfated [37,38]. Using a combination Alcian blue (AB) and high-iron diamine stain (not available anymore because of its toxicity) and immunohistochemistry (IHC), gastric intestinal metaplasia has been classified, based on the types of mucins expressed, into the following types (see "Gastric intestinal metaplasia"):

●Type I (complete) intestinal metaplasia expresses only sialomucins and expresses MUC2 (an intestinal mucin), while gastric mucins are either decreased or absent (MUC1, MUC5AC, and MUC6).

●Type II (incomplete) intestinal metaplasia expresses a mixture of gastric (neutral) mucins and intestinal sialomucins.

●Type III (incomplete) intestinal metaplasia, which is the least common, is characterized by the presence of columnar mucous cells containing sulfomucins. Incomplete metaplasia coexpresses intestinal mucin and gastric mucins.

In addition to the pattern and type of mucin expression, currently used classifications for gastric intestinal metaplasia also take into consideration the presence of Paneth cells (complete metaplasia) or crescent architecture changes, dedifferentiation, and the degree of absence of Paneth cells (incomplete metaplasia). Complete-type intestinal metaplasia displays goblet cells and absorptive cells with a brush border and shows decreased/absent expression of gastric mucins (MUC5AC and MUC6) and expression of MUC2 (an intestinal mucin) [39]. Incomplete intestinal metaplasia displays goblet and columnar nonabsorptive cells without a brush border and coexpression of gastric mucins and MUC2.

Morphologically, the first metaplastic glands seen in chronic H. pylori infection phenotypically resemble those of the small intestine, with eosinophilic absorptive enterocytes with a brush border alternating with mucus-producing goblet cells (ie, type I, complete, or small intestinal-type metaplasia (picture 3)). These changes are thought to be triggered by the loss of the acid-secreting parietal cells.

More advanced stages adopt characteristic changes that are more similar to colonic phenotype, with the glands becoming lined by irregular goblet cells (type III, incomplete types IIA/II, and IIB/III incomplete colonic metaplasia (picture 4)) [40,41]. More advanced stages of metaplasia increase the risk of gastric cancer [42,43]. As an example, a systematic review of 10 observational studies showed that the risk of gastric cancer was 4- to 11-fold higher among those with incomplete metaplasia, compared with either complete metaplasia or the absence of incomplete type metaplasia [42].

Incomplete metaplasia is frequently detected around early gastric adenocarcinomas. Some consider colonic metaplasia to represent an early stage of dysplasia, deserving closer endoscopic surveillance than type I (small intestinal) metaplasia [41,44,45]. The metaplastic changes are first seen at the junction of the antrum and the corpus mucosa (the location of the loss of acid-secreting parietal cells, oxyntic atrophy), especially in the area of the incisura angularis. Over time, the foci of metaplasia become larger and more numerous, extending to the antrum as well as to the corpus of the gastric mucosa. The larger the atrophic and metaplastic area, the greater the cancer risk.

Corpus-predominant gastritis with multifocal gastric atrophy with intestinal metaplasia and hypo- or achlorhydria develops in approximately 1 percent of subjects infected with H. pylori. As a consequence of the elevation in gastric pH, a change in the gastric flora is observed, with colonization by anaerobic bacteria responsible for the formation of carcinogenic nitrosamines [21,46]. Hypochlorhydria and low levels of pepsinogen I (produced by the chief cells of the corpus) and gastrin [21] (produced by antral G cells) can be measured in the serum and can be used as an indicator of gastric atrophy and cancer risk [47].

Multiple studies have indicated a positive correlation between the degree of incomplete intestinal metaplasia, the extent of intestinal metaplasia overall, and the risk of gastric carcinoma [37,44,45,48,49]. However, not all studies have been positive [50-52], and there is little genetic mutational evidence to implicate intestinal metaplasia as a true direct precursor of gastric cancer.

Another pattern of metaplasia is described, termed "spasmolytic polypeptide-expressing metaplasia" (SPEM), which strongly expresses trefoil factor family 2 (TFF2, previously designated spasmolytic polypeptide) and is associated with atrophy of the oxyntic mucosa. SPEM has also been termed "pseudopyloric metaplasia," and it has a strong association with both chronic infection with H. pylori and gastric adenocarcinoma [53-56]. While SPEM may represent another separate pathway to gastric neoplasia, the two processes may be linked. Mouse models have shown that SPEM can develop following oxyntic atrophy through transdifferentiation of zymogen-secreting chief cells into cells that have a mucous secretory profile [57]. Increasing evidence suggests that this process of chief cell reprogramming requires the influence of inflammatory cytokines in addition to oxyntic atrophy [58], and that SPEM may represent a physiologic repair mechanism for recruiting reparative progenitor cells in response to mucosal wounds [59]. More recent research has shown that the presence of SPEM allows H. pylori to access deeper regions along the gastric corpus unit, which may allow it to spread throughout the stomach [60-62]. Evolution of intestinal metaplasia from SPEM may lead to a hyperproliferative state in which the infected and inflamed mucosa may be more susceptible to establishment of deleterious mutations in stem or progenitor cell populations [55].

Gastric dysplasia — Gastric dysplasia, which represents a direct precursor of gastric adenocarcinoma, is a neoplastic lesion defined by cytologic and architectural features that are confined to the basement membrane (staged as Tis or carcinoma in situ according to the American Joint Committee on Cancer [AJCC] tumor, node, metastasis [TNM] staging criteria (table 1)). It can present endoscopically as a flat, polypoid (ie, "adenoma"), or depressed lesion [63,64].

The prevalence of gastric dysplasia mirrors that of gastric adenocarcinoma, with regional variation [65,66]. The mean age of patients is approximately a decade younger than that of those with gastric cancer, with a mean age of 61.35 versus 70 years [67]. Male predominance is seen, similar to gastric cancer (male/female = 1.9/1) [67,68].

Etiology and pathogenesis — The risk factors for gastric dysplasia are the same as those for gastric adenocarcinoma. It commonly develops in the background of chronic atrophic gastritis or intestinal metaplasia, and is strongly associated with H. pylori infection [49,69-71].

Although rare, dysplasia can also develop in nonadenomatous gastric polyps, even in the absence of H. pylori infection:

●Among patients with fundic gland polyps, the incidence of dysplasia is low, particularly in sporadic lesions (<1 percent), while the incidence is reported to be up to 41 percent in fundic gland polyps that arise within the context of an inherited syndrome [72-74].

However, even in the setting of familial adenomatous polyposis (FAP), progression from a polyp to an advanced lesion is uncommon, with only approximately 4 percent of cases developing high-grade dysplasia during follow-up [75,76]. The incidence of H. pylori infection is very low in patients with fundic gland polyps. (See "Gastric polyps", section on 'Fundic gland polyps'.)

●In hyperplastic polyps, which commonly develop in the setting of H. pylori infection or autoimmune gastritis, or secondary to bile reflux, the reported prevalence of dysplasia is low (<2 percent), correlates with the size of the polyps, and is more common in lesions larger than >2 cm [77,78]. (See "Gastric polyps", section on 'Hyperplastic polyps'.)

●Gastric dysplasia may also develop rarely in the setting of Peutz-Jeghers-related polyps and juvenile polyposis syndrome. (See "Juvenile polyposis syndrome" and "Peutz-Jeghers syndrome: Clinical manifestations, diagnosis, and management".)

Classification of dysplasia — According to the World Health Organization (WHO), gastric epithelial dysplasia (also referred as glandular intraepithelial neoplasia) is classified into two grades: low- and high-grade dysplasia [79]. Low-grade dysplastic lesions show minimal architectural disarray and are predominantly defined by cytologic features (picture 5). The nuclei usually retain a basal orientation and present either no or limited loss of polarity. By contrast, high-grade dysplasia presents with more pronounced architectural disarray (picture 6). Prominent cytologic atypia is also commonly detected, with cuboidal neoplastic cells, a high nucleus:cytoplasm ratio, clearing of the chromatin with prominent amphophilic nucleoli, and numerous mitoses, which can be atypical. The nuclei frequently extend into the luminal aspect of the cell, and nuclear polarity is usually lost. A third diagnostic category, indefinite for intraepithelial neoplasia, is reserved for cases for which a definitive diagnosis of dysplasia cannot be established with certainty. It may be used if the sampling submitted for evaluation is suboptimal or when a brisk inflammatory infiltrate may limit the morphologic evaluation.

The nomenclature of advanced premalignant lesions in Japan has differed considerably from that of Western countries. In the West, invasive gastric cancer is typically defined by stromal invasion, while in Japan, severe nuclear and architectural abnormalities, even if confined to glandular structures, are considered sufficient to classify a lesion as "strongly suspicious for invasive carcinoma." Furthermore, Japanese pathologists use the term "borderline" to describe lesions that, in Western countries, would be labeled low-grade dysplasia [80]. The Padova classification attempted to reconcile the two systems by assigning numerical categories based on the descriptive findings (table 2) [81]. A similar attempt was made in the Vienna Classification [82]. The various classifications that have been proposed over the years are outlined in the table (table 3) [79].

These discrepancies are more semantic than substantial. In Western countries, close endoscopic monitoring is considered appropriate for patients with low-grade gastric dysplasia, with intervention only if the lesions progress. On the other hand, the majority of clinicians (although not all) recommend surgical or endoscopic resection for patients with high-grade dysplasia [83]. This is because areas of microinvasion can be found in some patients undergoing resection for high-grade dysplasia, although the proportion is not well established. In Japan, most "adenomas" and "early carcinomas" do not undergo surveillance but are treated with endoscopic resection, a technique that is used much less in the West. (See "Early gastric cancer: Epidemiology, clinical manifestations, diagnosis, and staging".)

Histopathologic subtypes — The majority of gastric epithelial dysplasia displays a histomorphology similar to colonic adenomas, including goblet and Paneth cells, and thus, it is sometimes qualified as adenomatous (or intestinal) dysplasia or type I dysplasia [84-87].

Type II dysplasia (gastric phenotype) is a second morphologic type of gastric dysplasia, composed of foveolar- or pyloric-type epithelium. Microscopically, the lesions are composed of low cuboidal cells with round to oval, nonstratified nuclei with a pale to clear cytoplasm (due to the presence of apical mucin caps, a feature of foveolar differentiation) or oncocytic to ground-glass cytoplasm (a feature of pyloric differentiation) (picture 7A-C). The differentiation of nonadenomatous dysplasia can be confirmed immunohistochemically by expression of MUC5AC (a marker of foveolar differentiation) or MUC6 (a marker of pyloric differentiation) [78]. Architectural characteristics of nonadenomatous dysplasia include glandular cystification and papillary and serrated architecture. While some investigators have found that type II dysplasia is more commonly associated with poorly differentiated (diffuse) gastric cancer, this is not universal, although a more aggressive biology of this subtype is suggested by others [84,88-90].

Some patients have dysplasia-like atypia that is limited to the pit epithelium but without involvement of the surface epithelium. The nature and significance of basal gland atypia/pit dysplasia, usually reported at the periphery of bona fide gastric dysplasia and as "indefinite for dysplasia" on a pathology report, remain to be confirmed. This lesion is characterized by an increased nuclear:cytoplasmic ratio, nuclear size and irregularity, hyperchromasia, and increased mitoses at the base of the pits with a normal mucosal surface [91-94]. Most authors have concluded that this represents an important neoplastic precursor lesion in gastric carcinogenesis.

Progression of gastric dysplasia — The rate of progression of gastric dysplasia is now believed to be much lower than previously thought, as shown in the following fairly contemporary population-based studies involving Western populations:

●In one study of 4331 Kaiser Permanente members who were diagnosed with gastric intestinal metaplasia or gastric dysplasia between 1997 and 2006 and followed through December 2016, 3.5 percent of patients with low-grade dysplasia and 59 percent of those with high-grade dysplasia developed an invasive adenocarcinoma within the first year; thereafter, only 4.3 percent of those with low-grade dysplasia progressed to invasive carcinoma [95].

●In a report of all patients with a first diagnosis of a premalignant gastric lesion between 1991 and 2004 who were identified in the Dutch nationwide histopathology registry, the annual incidence of gastric cancer was 0.6 percent for mild to moderate dysplasia and 6 percent for severe dysplasia within five years after diagnosis [32].

●In another series, 15.8 percent of patients with the initial histologic diagnosis "indefinite for dysplasia" developed a gastric cancer within 12 months, indicating that this diagnosis cannot be discounted [96].

In the context of an adenomatous gastric polyp, the probability of finding foci of invasive adenocarcinoma increases with an increase in the size of adenoma, and the risk is as high as 50 percent for polyps >2 cm. Furthermore, a synchronous adenocarcinoma in another area of the stomach has been found in up to 30 percent of patients detected with an adenomatous gastric polyp [86,96].

Epstein-Barr virus — Infection with Epstein-Barr virus (EBV) is associated with a number of malignancies, especially nasopharyngeal carcinoma. (See "Epidemiology, etiology, and diagnosis of nasopharyngeal carcinoma", section on 'Epstein-Barr virus'.)

A possible role in gastric cancer was initially suggested in a study from Korea in which evidence of EBV was found in the tumor cells of 12 of 89 (13 percent) gastric carcinoma patients compared with none of 27 controls with a benign ulcer or any of the benign tissues from the cases [97]. Some of the tumor cells had a histologic appearance similar to that of nasopharyngeal carcinoma. Since then, it has been estimated that between 5 and 10 percent of gastric cancers worldwide are associated with EBV [98,99], although a role for EBV in gastric carcinogenesis, either directly or as a secondary effect, is debated [100].

EBV-associated gastric cancers (EBVaGCs) have distinct clinicopathologic characteristics, including male predominance, preferential location in the gastric cardia or postsurgical gastric stump, lymphocytic infiltration, a lower frequency of lymph node metastasis, perhaps a more favorable prognosis, and a diffuse type of histology in most [101-105], but not all [106-108], series. (See 'Epstein-Barr virus-associated gastric cancer' below.)

In addition, in part due to the overexpression/amplification of programmed cell death ligand 1 (PD-L1) in EBVaGCs, these tumors are good candidates for therapy with immune checkpoint inhibitors. (See 'Epstein-Barr virus-associated gastric cancer' below and 'PD-L1 overexpression and deficient mismatch repair' below.)

Genetic predisposition and hereditary syndromes — First-degree relatives of patients with gastric cancer are almost three times as likely as the general population to develop gastric cancer [109]. This may be partly attributable to H. pylori infection being common in families and to the potential role of inherited IL-1 gene polymorphisms, which may affect the inflammatory response and gastric cancer risk associated with H. pylori infection [21,22]. (See 'Helicobacter pylori' above and "Risk factors for gastric cancer", section on 'Familial predisposition'.)

However, while familial aggregation of gastric cancer occurs in around 10 to 20 percent of patients with gastric cancer, fewer than 5 percent of cases result from an inherited predisposition to cancer [110-114]. One of these syndromes, hereditary diffuse gastric cancer (HDGC), is an autosomal dominant cancer susceptibility syndrome characterized by signet ring cell (diffuse) gastric cancer and lobular breast cancer; it is caused by a germline defect in the CDH1 (E-cadherin 1) gene. (See 'Hereditary diffuse gastric cancer' below.)

Two other types of familial predisposition to gastric cancer are described: familial intestinal gastric cancer (FIGC) and gastric adenocarcinoma and proximal polyposis of the stomach (GAPPS). The risk of developing gastric cancer is high in these families, but only HDGC is genetically explained. A summary of clinical features, recommendations for genetic screening, and the genetic alterations described for these three syndromes is outlined in the table (table 4) [110]. (See "Risk factors for gastric cancer", section on 'Familial predisposition'.)

Hereditary diffuse gastric cancer — HDGC is an inherited form of diffuse-type gastric cancer, a highly invasive tumor that is characterized by late presentation and a poor prognosis. The lifetime risk of gastric cancer in individuals from these families is very high (cumulative risk of gastric cancer by age 80 years is reported to be 70 percent for males and 56 percent for females), and the median age at diagnosis is only 38 (although it ranges from 14 to 85 years of age). As a result, prophylactic total gastrectomy is usually advised, generally between ages 20 and 30. Almost all prophylactic gastrectomy specimens contain foci of early invasive gastric cancer. (See "Surgical management of hereditary diffuse gastric cancer", section on 'Timing of prophylactic total gastrectomy'.)

Affected individuals also have an increased risk for lobular breast cancer (cumulative risk by age 80 is approximately 42 percent). (See "Hereditary diffuse gastric cancer", section on 'Breast cancer'.)

Many families with HDGC have germline pathogenic and likely pathogenic variants in the CDH1 gene that are inherited in an autosomal dominant pattern. Germline CDH1 pathogenic and likely pathogenic variants have been identified in approximately 15 to 50 percent of affected kindreds that meet the clinical criteria for HDGC, as defined by the International Gastric Cancer Linkage Consortium (IGCLC) [115]. According to the two-hit hypothesis, a second-hit somatic event (commonly promoter hypermethylation) leading to biallelic inactivation is necessary for the development of gastric cancer in mutation carriers.

Unlike the somatic mutations in CDH1 that occur in sporadic gastric cancers and cluster around exons 7 and 8, the germline mutations in CDH1 in HDGC families span the whole length of the gene, and no hot spots have been identified. Other potential candidate pathogenic variants have been identified in those who meet clinical criteria for HDGC but lack germline mutations in CDH1, including those in CTNNA1 (the gene for alpha-catenin), BRCA2 (the gene associated with hereditary breast and ovarian cancer syndrome), STK11 (the gene associated with Peutz-Jeghers syndrome), SDHB, PRSS1, ATM, MSR1, and PALB2. (See "Hereditary diffuse gastric cancer", section on 'Molecular genetics'.)

Familial intestinal gastric cancer — FIGC is characterized by an autosomal dominant pattern of inheritance of the intestinal subtype of gastric cancer in the absence of gastric polyposis. No specific genetic alterations have yet been linked to this syndrome. The incidence and specific risk for gastric cancer are not defined. Clinical criteria for diagnosis are outlined in the table (table 4).

The stomach cancers display the common macroscopic features observed in the sporadic setting of gastric cancer of the intestinal (gland-forming) type (see 'Intestinal versus diffuse types' below). Few recommendations are available for management of patients with suspected FIGC [116].

Gastric adenocarcinoma and proximal polyposis of the stomach — GAPPS is an autosomal dominant cancer predisposition syndrome with a significant risk of gastric adenocarcinoma [117]. Point mutations in promoter 1B of the adenomatous polyposis coli (APC) gene cosegregated with disease in six GAPPS families [118]. To date, few families with GAPPS are reported worldwide [117-120]. The age of onset of gastric cancer is variable, and fundic gland polyposis with multifocal dysplasia has been detected as early as 10 years of age [117,120]. The youngest patient who presented with generalized gastric adenocarcinoma presented at the age of 26 years [120]. Scattered hyperplastic and pure adenomatous polyps and mixed polyps containing areas of dysplasia or gastric adenocarcinoma can be detected amid the fundic gastric polyps [117]. Gastric cancers display the features of intestinal-type or mixed gastric cancer [117,120].

Other inherited cancer syndromes — Gastric cancer has also been described in association with certain other inherited cancer syndromes, including Lynch syndrome (hereditary nonpolyposis colorectal cancer), FAP, Li-Fraumeni syndrome, Peutz-Jeghers syndrome, juvenile polyposis, hereditary breast and ovarian cancer syndrome, and, possibly, phosphatase and tensin homolog (PTEN) hamartoma tumor (Cowden) syndrome, but these are all fairly rare causes of gastric cancer. (See "Lynch syndrome (hereditary nonpolyposis colorectal cancer): Clinical manifestations and diagnosis", section on 'Extracolonic manifestations' and "Clinical manifestations and diagnosis of familial adenomatous polyposis", section on 'Extracolonic manifestations' and "Li-Fraumeni syndrome", section on 'Spectrum of malignancies and age at onset' and "Peutz-Jeghers syndrome: Clinical manifestations, diagnosis, and management", section on 'Gastrointestinal cancers' and "PTEN hamartoma tumor syndromes, including Cowden syndrome", section on 'Gastric and duodenal polyps' and "Cancer risks and management of BRCA1/2 carriers without cancer", section on 'Other solid tumors'.)

Other environmental factors implicated in the development of gastric cancer — Other factors that play an important role in the causation of gastric cancer, particularly in conjunction with H. pylori infection, include the following:

●Smoking – Smoking tobacco increases the risk of gastric cancer, in part by potentiating the carcinogenic effect of infection with CagA-positive H. pylori [121-124]. (See "Risk factors for gastric cancer", section on 'Smoking'.)

●Diet and nutrition – Diets high in salted, pickled, smoked, or poorly preserved foods; those with high meat content; and those with low fruit and vegetable content are most commonly associated with an increased risk for developing intestinal-type gastric cancer, particularly in combination with H. pylori infection [125-131]. (See "Risk factors for gastric cancer", section on 'Diet'.)

●Bile reflux – Bile reflux has also been implicated. Patients with a gastric stump after previous gastric surgery have an increased risk of developing gastric cancer (reported frequencies range from 0.8 to 9 percent) [132,133]. The risk appears to increase 15 to 25 years after the original surgery [134-136]. Risk is higher following Billroth II gastrectomy and in those who have undergone surgery for a gastric ulcer, while those operated on for a duodenal ulcer are at a lower risk [137]. (See "Risk factors for gastric cancer", section on 'Gastric surgery' and "Postgastrectomy complications", section on 'Remnant cancer'.)

Autoimmune gastritis and gastric polyps

●Autoimmune gastritis and pernicious anemia – Pernicious anemia is associated with an increased risk of gastric dysplasia and intestinal-type adenocarcinoma. However, the magnitude of the risk is debated; a two- to sixfold excess risk has been reported, but as with other predisposing conditions, the actual degree of risk varies with the duration of disease and geographic location. (See "Risk factors for gastric cancer", section on 'Pernicious anemia'.)

●Finally, as noted above, gastric epithelial polyps, both neoplastic and nonneoplastic, can be associated with the development of gastric cancer; the key features are presented in the table (table 5). An exhaustive review of the preneoplastic nature of gastric polyps is beyond the scope of this review. Gastric epithelial polyps (hyperplastic polyps, adenomas) are discussed in detail elsewhere. (See "Gastric polyps".)

CLASSIFICATION OF GASTRIC CANCER — Gastric cancers have been classified in a number of ways, including by anatomic location, disease extent, histomorphologic appearance, and molecular subtypes.

Anatomic classification and staging — The tumor, node, metastasis (TNM) staging systems of the American Joint Committee on Cancer (AJCC)/Union for International Cancer Control (UICC) for esophagogastric junction cancers (table 6) and for gastric cancers (table 1) are used universally. In the most recent (2017, eighth edition) revision of the AJCC staging classification [138], tumors involving the esophagogastric junction (EGJ) with the tumor epicenter no more than 2 cm into the proximal stomach are staged as esophageal cancers. By contrast, EGJ tumors with their epicenter located more than 2 cm into the proximal stomach are staged as stomach cancers, as are all cardia cancers not involving the EGJ, even if they are within 2 cm of the EGJ. (See "Clinical manifestations, diagnosis, and staging of esophageal cancer", section on 'TNM staging criteria' and "Clinical features, diagnosis, and staging of gastric cancer", section on 'TNM staging criteria'.)

Patterns of disease spread and implications for staging — Gastric carcinomas can spread either by direct extension to adjacent organs, metastasis, or peritoneal dissemination. Depending on the primary site within the stomach, penetration of the serosa may result in direct spread to the pancreas, liver, spleen, transverse colon, and/or greater omentum, and it often leads to early transperitoneal dissemination [139]. Spread via the bloodstream occurs from invasion of tributaries of the portal vein, and the liver is the organ affected most commonly, followed by the lung and then the peritoneum. In general, gland-forming (ie, tubular, papillary, mucinous) carcinomas are more likely to give rise to liver metastases by hematogenous spread, while poorly cohesive carcinomas are more likely to involve the peritoneum than are gland-forming carcinomas. Peritoneal involvement is also more common in younger patients. (See 'World Health Organization classification' below.)

World Health Organization (WHO) poorly cohesive (Lauren diffuse-type) cancers have a greater tendency to invade the gastric wall, sometimes extending to the lower esophagus or to the duodenum via submucosal or subserosal routes or via the submucosal lymphatics. Occasionally, a wide region of the gastric wall or even the entire stomach is extensively infiltrated, resulting in a rigid, thickened stomach, termed "linitis plastica" (image 1 and image 2). Bilateral massive ovarian involvement (Krukenberg tumor) can result from transperitoneal or hematogenous spread.

Lymphatic and vascular invasion carries a poor prognosis in gastric cancer. The incidence of lymph node metastatic disease increases with the depth of tumor invasion [140] and occurs with equal frequency regardless of histologic type. The distribution varies according to the location of the tumor (figure 2). (See "Clinical features, diagnosis, and staging of gastric cancer", section on 'Staging systems'.)

The accuracy of pathologic staging is proportional to the number of regional lymph nodes examined and their anatomic location in relation to the neoplasm. A discussion on the appropriate extent of lymph node dissection and the prognostic influence of primary tumor (T) and nodal (N) stage are discussed in more detail elsewhere. (See "Surgical management of invasive gastric cancer", section on 'Extent of lymph node dissection' and "Surgical management of invasive gastric cancer", section on 'Prognosis'.)

Early versus advanced

Early gastric cancer — The concept of early gastric cancer (EGC) originated in Japan in 1962; at that time, EGC was defined as a neoplasm that could be successfully treated with surgery. EGC is now defined more specifically as an adenocarcinoma that is restricted to the mucosa or submucosa (picture 8), irrespective of lymph node metastasis (T1, any N) (table 1). These cancers have a significantly better prognosis (approximately 90 percent five-year survival rate) than do more advanced stages of gastric cancer. (See "Early gastric cancer: Epidemiology, clinical manifestations, diagnosis, and staging", section on 'Definition'.)

Importantly, this definition, which has been adopted by the Japanese National Cancer Centre, the Japan Gastric Cancer Association, and other international groups, acknowledges that there is a small percentage of patients with EGC who have lymph node involvement [141]. These cancers should still be considered EGC, even if they have nodal metastases, as is supported by decades of investigation and data from screening programs [78].

However, nodal metastases can affect treatment of EGC in two ways (see "Early gastric cancer: Treatment, natural history, and prognosis"):

●Patients with biopsy-proven nodal metastases, patients whose pretreatment local staging evaluation indicates a high likelihood of nodal metastases, and patients with T1 disease who are at high risk for nodal metastases (because of tumor size, configuration, or depth of invasion) may not be appropriate candidates for endoscopic resection. Gastrectomy with removal of the regional nodes is the appropriate treatment strategy for such patients.

●Patients with node-positive resected EGC are candidates for adjuvant therapy.

Countries with a high incidence of gastric cancer and in which asymptomatic patients are screened for gastric cancer have a high incidence of EGCs, ranging from 30 to 50 percent, with lower figures for Western populations (15 to 21 percent). (See "Gastric cancer screening" and "Early gastric cancer: Epidemiology, clinical manifestations, diagnosis, and staging", section on 'Epidemiology'.)

EGCs are endoscopically divided into three types based on their appearance (the Paris system of classification (table 7)). (See "Early gastric cancer: Epidemiology, clinical manifestations, diagnosis, and staging", section on 'Macroscopic classification'.)

Histologically, small (<2 cm) and minute (<5 mm) EGCs are commonly well differentiated. As tumor size increases and submucosal invasion develops, histologic diversity with mixed or poorly differentiated components is more common.

EGCs are characterized by a low incidence of vessel invasion and lymph node metastasis, and a good prognosis (approximately 90 percent at 10 years). Endoscopic resection is an effective therapeutic choice for EGCs without clinical suspicion of lymph node metastasis that meets the following criteria:

●High probability of en bloc resection

●Tumor histology

•Differentiated adenocarcinoma

•Tumor confined to the mucosa

•Absence of venous or lymphatic invasion

●Tumor size and morphology

•Less than 20 mm in diameter, without ulceration

•Less than 10 mm in diameter if Paris classification IIb or IIc (table 7)

Gastrectomy is appropriate for patients with EGC who have lymph node involvement that is detected or highly suspected during the preoperative staging evaluation and for others who do not meet the standard criteria for endoscopic resection. Standard and expanded criteria for endoscopic resection of EGC are presented in detail elsewhere. (See "Early gastric cancer: Treatment, natural history, and prognosis".)

Advanced carcinoma — Advanced gastric carcinomas can display various gross appearances. Bormann's classification divides gastric carcinomas into four distinct morphologic types: polypoid (type 1), fungating (type 2), ulcerated (type 3), and diffusely infiltrative (type 4) (figure 3) [142-145].

Ulcerated tumors differ from benign ulcers by having an irregular margin with raised borders and thickened, uneven, and indurated surrounding mucosa. The ulcer base is necrotic, shaggy, and often nodular. Mucosal folds radiating from the crater are irregular and frequently show club-like thickening and fusion [146]. Invasive adenocarcinoma, particularly when composed of poorly cohesive neoplastic cells, may spread superficially in the mucosa and submucosa, giving rise to plaque-like lesions with flattening of the rugal folds. In some cases, superficial ulceration supervenes. The infiltration may involve the entire thickness of the wall and produce the so-called linitis plastica appearance (image 1). Some mucinous adenocarcinomas may secrete considerable amounts of mucin, giving the tumor a gelatinous appearance.

Histologic classification — The histologic classification of gastric adenocarcinoma is challenging because of intratumoral variations in architecture and/or differentiation, and several histologic classifications have been proposed over the years.

World Health Organization classification — The 2019 WHO classification, which does not take into account histogenesis and grade of differentiation, recognizes several main histologic types of malignant epithelial tumors (tubular, papillary, poorly cohesive signet ring phenotype, poorly cohesive other cell type, mucinous, mixed), as well as rare variants (table 8) [147].

Intestinal versus diffuse types — The WHO classification does not include the morphologic "intestinal" and "diffuse" types of gastric cancer [147]. At the cellular level, despite being classified as "intestinal," the neoplastic cells may show morphologic or immunophenotypical evidence not only of intestinal differentiation but also of gastric, gastrointestinal, unclassifiable, or null differentiation, underscoring the inadequacies of this histomorphologic classification.

Nevertheless, since the seminal paper by Lauren in 1965, gastric adenocarcinomas have historically been divided into two distinct histomorphologic subtypes, intestinal (ie, gland-forming) and diffuse (composed of discohesive cells), which have a distinct morphologic appearance, epidemiology, pathogenesis, and genetic profile [148-150]. Broadly, the morphologic differences are attributable to different genetic and epigenetic alterations, some related to intercellular adhesion molecules, which are preserved in intestinal-type tumors and defective in diffuse carcinomas:

●In intestinal-type tumors, the tumor cells adhere to each other and tend to arrange themselves in tubular or glandular formations, similar to adenocarcinomas arising elsewhere in the intestinal tract (hence their designation as intestinal type) (picture 9). The intestinal type is more common in males and older age groups, and they have a better prognosis.

●By contrast, a lack of adhesion molecules in poorly cohesive carcinomas allows the individual tumor cells to grow and invade neighboring tissues without gland formation. Poorly cohesive carcinomas, including signet ring cell carcinomas, correspond to the diffuse type of the Lauren classification. They are composed of individual tumor cells invading the surrounding tissues and with no gland formation (picture 10). This subtype can display various cell types that include neoplastic cells, such as signet ring cells, but also histiocytes as well as lymphocytes and others with a deeply eosinophilic cytoplasm. A mixture of these cell types including classical signet ring cells is commonly seen. These tumors show a more equal male:female ratio and are more frequent in younger individuals.

It has long been considered that poorly cohesive histology is an independent predictor of a worse prognosis [151,152]. However, more recent studies have begun to question this notion, with the suggestion that signet ring histology is associated with a more advanced stage of disease and that, when adjusted for stage, poorly cohesive tumors do not portend a worse prognosis [153-158]. In a large-scale study, it was shown that early stage signet ring cell/diffuse carcinomas can be indolent, and more than one-half of the cases are presented as EGC [159]. On the other hand, signet ring cell/diffuse carcinomas in advanced stages bestow a worse prognosis than well or moderately differentiated carcinomas.

A molecular basis for this difference is now apparent. The main carcinogenic event in diffuse carcinomas is loss of expression of E-cadherin, a key cell surface protein for establishing intercellular connections and maintaining the organization of epithelial tissues [160]. Loss of expression of E-cadherin leads to defective intercellular adhesions. Biallelic inactivation of the gene encoding E-cadherin (CDH1) can occur through germline or somatic mutation, allelic imbalance events (eg, loss of heterozygosity), or epigenetic silencing of gene transcription through aberrant methylation of the CDH1 promoter. Individuals who inherit germline CDH1 mutations are considered to have hereditary diffuse gastric cancer (HDGC), and they are predisposed both to EGC (picture 11) and lobular breast cancer. (See 'Hereditary diffuse gastric cancer' above.)

In direct contrast, the molecular basis underlying intestinal-type gastric cancers is less well defined. However, it appears to follow a multistep progression that is usually initiated by H. pylori infection. Like intestinal-type cancers, diffuse-type gastric carcinomas also can be induced by H. pylori infection. However, in contrast with intestinal-type cancers, a defined series of preneoplastic stages is not well defined. (See 'Helicobacter pylori' above.)

Early stage HDGC in CDH1 mutation carriers is characterized by multiple foci of invasive (T1a) signet ring cell (diffuse, poorly cohesive) carcinoma in the superficial gastric mucosa without nodal metastases [161-164]. Precursor lesions (Tis) can be recognized: pagetoid spread of signet ring cells below the preserved epithelium of glands and foveolae but within the basal membrane, or in situ signet ring cell carcinoma, corresponding to the presence of signet ring cells within the basal membrane, substituting the normal epithelial cells [27]. In these lesions, E-cadherin immunoexpression is reduced or absent [27,163,165,166]. Because the EGCs that develop in these individuals are multifocal and located beneath an intact mucosal surface [161], early detection is difficult. Given this fact, and the poor prognosis of these tumors when locoregionally advanced, patients with evidence of a CDH1 germline mutation in the context of a family history of HDGC are candidates for prophylactic gastrectomy. (See "Surgical management of hereditary diffuse gastric cancer", section on 'Indications for prophylactic total gastrectomy'.)

Abnormalities in CDH1 have also been linked to sporadic diffuse (and intestinal) carcinomas. Somatic mutations in the CDH1 gene are identified in 40 to 83 percent of sporadic diffuse-type gastric cancers [167-170]. One analysis of 174 sporadic gastric cancers found that CDH1 alterations (both structural alterations, such as mutations or loss of heterozygosity, and epigenetic alterations) were present in 34 percent of diffuse-type gastric cancers and in 26 percent of intestinal cancers [171]. This analysis found no structural alterations among 19 patients with HDGC, but 53 percent harbored an epigenetic methylation.

Special-type gastric cancer — Several subtypes with characteristic morphology are also recognized in the WHO classification, emphasizing the diversity of the disease (table 8) [147]. Some are related to either specific etiopathogenic or molecular alterations.

Epstein-Barr virus-associated gastric cancer — Gastric carcinoma with lymphoid stroma [147,172] or lymphoepithelioma-like carcinoma are associated with Epstein-Barr virus (EBV) infection and are designated as EBV-associated gastric cancers (EBVaGCs) (picture 12) [173-175]. A similar morphology can be observed in gastric cancers with microsatellite instability [176,177]. Overall, the frequency of finding evidence of EBV infection in gastric carcinomas ranges from 2 to 20 percent, with a worldwide average of nearly 10 percent [100,178,179]. However, as noted above, a role for EBV in gastric carcinogenesis, either directly or as a secondary effect, is debated. (See 'Epstein-Barr virus' above.)

EBVaGCs display certain distinctive macroscopic and histologic characteristics [172,175,180]. Gastric carcinoma with lymphoid stroma is the main variant, and it occurs predominantly in the proximal stomach [105,181] and in the gastric stump of patients with a previous partial gastrectomy [182]. The patients are younger than in conventional carcinoma, and males are affected more often [105,172,173,180]. The tumors usually have a well-demarcated pushing, rather than infiltrating, margin. They are typically composed of irregular sheets; trabeculae; ill-defined tubules of polygonal cells admixed with a prominent lymphocytic infiltrate composed of CD8-positive cytotoxic T lymphocytes and, to a lesser degree, B lymphocytes; plasma cells; neutrophils; and eosinophils. Notably, most, but not all, gastric cancers with lymphoid stroma appear to be EBV related (picture 12) [173].

Two other variants of EBVaGCs are observed: tubular carcinomas with limited desmoplasia, a smaller number of lymphocytes, and prominent lymphoid follicles with active germinal centers (designated as "carcinoma with Crohn disease-like lymphoid reaction") and conventional-type adenocarcinomas with scant lymphocytic infiltrate [181]. The prognosis is reportedly better than that for typical gastric cancers [101,102,105-107,173-175,180,182-186].

In part due to the overexpression/amplification of programmed cell death ligand 1 (PD-L1) in gastric cancers associated with EBV, these tumors are good candidates for therapy with immune checkpoint inhibitors [187-190]. (See 'The Cancer Genome Atlas' below and 'PD-L1 overexpression and deficient mismatch repair' below and "Initial systemic therapy for locally advanced unresectable and metastatic esophageal and gastric cancer", section on 'Immunotherapy-based regimens' and "Second and later-line systemic therapy for advanced unresectable and metastatic esophageal and gastric cancer", section on 'Biomarkers and benefit from PD-1 inhibitors'.)

Adenosquamous and squamous cell carcinoma — In adenosquamous carcinoma, the neoplastic squamous component comprises at least 25 percent of the tumor volume, characterized by keratin pearl formation and intercellular bridges in addition to the glandular element [191]. Metastases usually contain both glandular and squamous components, but in some instances, only one component may be present [192].

Pure squamous cell carcinomas of the stomach [193-195] are usually diagnosed at a late stage and have a poor prognosis [195].

Micropapillary carcinoma — Micropapillary carcinoma (MPC) is an uncommon subtype of gastric cancer (incidence approximately 6 percent) that is characterized by irregular, small, mole-like clusters of tumor cells lying within clear lacunar spaces that simulate lymphatic or vascular channels. Most cases are associated with classic tubular or papillary adenocarcinomas. The proportion of the MPC component ranges from 5 to 80 percent [196]. MPC carcinoma is an aggressive subtype, with a high incidence of nodal metastasis at diagnosis [197-199] and worse survival rates than for those with non-MPC gastric carcinomas [196,197].

Gastric adenocarcinoma of fundic gland type — Gastric adenocarcinoma of fundic gland type (chief cell predominant type) [200] is a rare variant of gastric cancer. Despite the presence of minimal invasion into the submucosal layer, lymphatic or venous invasion is extremely rare [201,202]. The characteristic microscopic appearance is an irregularly anastomosing glandular pattern lined by chief cells and a variable number of parietal cells with mildly enlarged and hyperchromatic nuclei (picture 13). The neoplastic chief cells express MUC6 and pepsinogen-I (gastric phenotype).

The best terminology has been debated, with some supporting the term "oxyntic gland polyp/adenoma" as more appropriate for this type of lesion [203,204].

Molecular subtyping — Advances in high-throughput genomic technologies have led to rapid advances in the understanding of the genomic, epigenomic, transcriptomic, and proteomic changes that underlie gastric cancer pathobiology. Gastric cancers have a high frequency of somatic mutations as well as substantial interindividual variability of mutational load. One study, based on next-generation sequencing (NGS)-based estimates, reported that each case of gastric cancer (with the exception of hypermutated cases) presents 50 to 70 nonsynonymous mutations [205]. The frequency of somatic mutations of genes noted in gastric cancer has allowed a three-tiered classification: genes with a high frequency rate of mutations (>5 to 10 percent) across non-multiple gastric cancers, genes with a low frequency rate (recurrently mutated in the 1 to 10 percent range), and bystander/passenger mutations that arise secondary to underlying mutational processes but do not contribute to gastric carcinogenesis. For example, TP53 is the most commonly mutated gene in gastric cancer, with mutations detected in approximately 50 percent of the cases [189,206-208].

Various studies have sought to incorporate multiple molecular readouts to delineate the mechanisms of disease biology and discern potentially useful therapeutic molecular targets. As an example, The Cancer Genome Atlas (TCGA) performed a comprehensive analysis of multiple "omic" (copy number, mutational, DNA methylation, transcriptome, and proteome) readouts with 295 gastric adenocarcinomas profiled on all five platforms [189]. Others, the Asian Cancer Research Group (ACRG), have analyzed 251 gastric cancers through a combination of gene expression profiling, genome-wide copy marker microarrays, and targeted gene sequencing, generating gene expression profiles by comparing their results to a predefined set of gene expression signatures: epithelial to mesenchymal transition (EMT), microsatellite instability, cytokine signaling, cell proliferation, and DNA methylation [208]. Another study led by the Duke-National University of Singapore compared the gene expression patterns among 248 gastric adenocarcinomas using consensus hierarchical clustering with iterative feature selection [209]. Collectively, these studies report a diverse spectrum of mutational frequencies, with dominant subclones centered around a handful of driver mutations that can form the basis for a molecular classification of gastric cancer.

The three studies described above emphasize that gastric cancer is not a single entity but a complex and heterogeneous group of diseases with distinct biologic and clinical features. A premise of precision medicine is that the appropriate matching of tumor molecular subtypes with targeted therapeutic agents will offer superior efficacy and less toxicity, thereby improving clinical outcomes. On a practical level, robust molecular gastric cancer subtyping may resolve, to some degree, the heterogeneity of existing classification schemes, whether surgical, endoscopic, or histopathologic, which remain insufficient to guide treatment for individual patients. These novel attempts at classifying gastric adenocarcinomas integrate the genomic and transcriptomic profiles to define integrative clusters characterized by distinct clinical and morphologic characteristics as well as prognostic outcomes. Ultimately these findings are expected to have significant implications for patient stratification and tailored therapeutic approaches.

The Cancer Genome Atlas — TCGA confirmed a high frequency (>10 percent overall) of somatic mutations in a handful of genes (TP53, PIK3CA, and ARID1A [AT-rich interaction domain-containing protein 1A]) [189]. A second group of genes (SMAD4, RhoA [RAS homolog gene family, member A], KRAS, and APC [adenomatous polyposis coli]) were mutated in 6 to 8 percent of the cases, whereas the vast majority of mutations were discovered in a small (1 to 3 percent) proportion of cases. Based on molecular features, these investigators identified four major molecular subtypes of gastric cancer, which are summarized in the table (table 9), and described in more detail in the sections below:

●Genetically stable gastric cancers – These cases represent approximately 20 percent of gastric cancers. These tumors are usually aneuploid and diagnosed at an earlier age [142]. There is an enrichment of the diffuse subtype in this group (73 percent), and they tend to present at a distal location within the stomach. In this subgroup, CDH1 somatic mutations are present in 37 percent of the cases, but extensive copy number aberrations are rare. Inactivating mutations of ARID1A are noted within this subset, and mutations of the RhoA gene were extensively noted in this group as well. This finding is not unexpected since most of these tumors have a poorly cohesive phenotype, and there is a demonstrated role of RhoA in diffuse gastric cancer cellular motility [210]. A subset of these tumors also showed a translocation between claudin 18 and the RHO GTPase activating protein 26 genes (the CLDN18-ARHGAP26 fusion gene), an alteration also likely to contribute to the diffuse phenotype since it is enriched in younger-onset-age gastric cancers [211], it is associated with loss of epithelial integrity [212], and claudin proteins are a component of the tight junction adherens structure [213].

●Chromosomally unstable gastric cancers – These neoplasms represent approximately 50 percent of gastric cancers, and the frequency of this subtype is increased in EGJ cancers. Chromosomal instability is characterized by DNA aneuploidy, structural changes of chromosomes (eg, translocations), and mutations in various proto-oncogenes and tumor suppressor genes. These cancers mostly exhibit an intestinal morphology, a high number of TP53 mutations (70 percent), as reflected histologically by P53 overexpression (picture 14), and amplifications of tyrosine kinase receptors. Phosphorylation of the epidermal growth factor receptor (EGFR) is significantly elevated in this group, while aneuploidy is also frequent. These lesions with frequent TP53 mutations fall into the proliferation subtype of the Duke-National University of Singapore [209]. Other genes frequently amplified within this molecularly defined subgroup include CCNE1, MYC, ERBB2 (HER2), and KRAS.

Human epidermal growth factor receptor 2 (HER2), which is a member of the EGFR family of proteins, is a transmembrane tyrosine kinase receptor that regulates cell proliferation, differentiation, and survival [214]. Approximately 10 to 20 percent of gastric adenocarcinomas are HER2 positive, and overexpression of this protein indicates the potential for responsiveness to HER2-targeted treatments (ie, trastuzumab). (See 'HER2 overexpression' below.)

●Microsatellite unstable (mismatch repair deficient) gastric cancers – Microsatellite instability is the biologic footprint of deficiency in DNA mismatch repair (dMMR). Microsatellite unstable gastric cancers represent approximately 22 percent of gastric cancer cases and are characterized by a high mutation rate (with an average of 50 mutations per neoplasm, compared with approximately ≤5 in the other subtypes). In microsatellite unstable sporadic gastric cancers, the mismatch repair defect is most frequently caused by an epigenetic event (hypermethylation in the MLH1 promoter region) [215]. In rare cases, a germline mutation in a mismatch repair gene is inherited and causes Lynch syndrome, which most often leads to colorectal and endometrial cancers, but gastric cancers also arise in up to 10 percent of such families [216]. Microsatellite unstable (dMMR) gastric cancers can be identified immunohistochemically by loss of expression of one of the DNA mismatch repair genes (picture 15).

Hypermutations are frequent in the KRAS, anaplastic lymphoma kinase (ALK), ARID1A, and PI3K-PTEN mechanistic (previously called mammalian) target of rapamycin (mTOR) pathway. These lesions occur at a relatively older age and tend to affect females disproportionately [142]. Most arise in the antrum, have an intestinal phenotype, and are more likely to be diagnosed at an earlier stage.

dMMR also identifies tumors that may respond to immune checkpoint inhibitor immunotherapy. (See 'PD-L1 overexpression and deficient mismatch repair' below.)

●Gastric cancers with Epstein-Barr virus infection – These tumors constitute approximately 9 percent of gastric cancers and tend to affect males preferentially. These lesions are frequently located in the fundus and body have a lower frequency of lymph node metastases, and have a lower mortality rate. EBV-positive tumors show an extreme CpG island methylator phenotype. Characteristics of these tumors include CDKN2A promoter hypermethylation and nonsilent mutation of PIK3CA (in 80 percent of the cases). Other frequently mutated genes include ARID1A, which encodes for an antiapoptotic protein [142]. JAK2 and ERBB2 (HER2) amplifications are also common.

This subgroup also shows, in approximately 10 percent of the cases, recurrent 9p24.1 amplification at the locus containing JAK2, CD274, and PDCD1LG2, which is significant since CD274 and PDCD1LG2 encode PD-L1 and programmed cell death ligand 2 (PD-L2). These two proteins help the neoplastic cells to escape from the body's antitumoral immune response, by binding to programmed cell death receptor 1 (PD-1), which is expressed on cytotoxic T cells. These molecules are of particular relevance as targets for immune checkpoint inhibitor immunotherapy. (See 'PD-L1 overexpression and deficient mismatch repair' below.)

From a prognostic standpoint, the EBV-related subtype is associated with the best prognosis, and the genetically stable subtype is associated with the worst survival (figure 4) [217].

Asian Cancer Research Group — The ACRG also identified four molecularly defined subgroups [208], which differ from those identified by TCGA, although there is at least some overlap:

●A mesenchymal group microsatellite stable (MSS)/EMT (15.3 percent of cases) based on the evidence of EMT; this subtype occurred mainly at an advanced stage, at a younger age, and with a diffuse histology (>80 percent), including a large set of signet ring cell carcinomas seeding in the peritoneum with malignant ascites (64 versus 15 to 24 percent of the other subtypes), and it showed loss of CDHG1 expression. This subtype had the worst prognosis.

●An MSS/TP53-negative subtype (35.7 percent).

●An MSS/TP53-positive subtype (26.3 percent); EBV infection was more frequent in this subgroup.

●A microsatellite unstable subtype (22.7 percent); this occurs predominantly at an early stage in the distal part of the stomach and showed mainly intestinal histology.

Overall, the microsatellite unstable subtype has the best prognosis, while the MSS/EMT subtype has the worst (figure 5).

A comparison of the key characteristics of both the ACRG and TCGA molecular classifications is provided in the table (table 10) [218].

Clinical implications — Although the molecular characterization of gastric cancers as described above has identified gene signatures that are prognostically relevant, they are still inadequate to direct molecularly targeted therapy, with few exceptions.

Currently, there are only three therapeutically relevant, routinely tested molecular biomarkers in gastric carcinoma: overexpression of HER2 (ERBB2), which permits the selection of patients with advanced disease who might benefit from trastuzumab, and overexpression of PD-L1/dMMR, both of which identify patients with advanced disease with the potential to benefit from immune checkpoint inhibitor immunotherapy. (See "Initial systemic therapy for locally advanced unresectable and metastatic esophageal and gastric cancer", section on 'Cytotoxic chemotherapy options' and "Second and later-line systemic therapy for advanced unresectable and metastatic esophageal and gastric cancer", section on 'Biomarkers and benefit from PD-1 inhibitors'.)

HER2 overexpression — Approximately 10 to 20 percent of gastric adenocarcinomas have human epidermal growth factor receptor 2 (HER2) gene amplification, which results in overexpression of HER2, a receptor tyrosine kinase and member of the EGFR family (picture 16). Overexpression of HER2 is most frequent in the TCGA-defined chromosomally unstable subset of gastric cancers. ACRG data documented that HER2 gene amplifications are enriched in the MSS/TP53-negative subtype (present in 17 percent of cases) [208]. In general, HER2 amplifications are more common in adenocarcinomas of the gastroesophageal junction (GEJ) compared with gastric adenocarcinomas, in the Lauren intestinal subtype compared with the diffuse subtype, and in well and moderately differentiated tumors compared with poorly differentiated tumors [219]. Others note that the HER2 amplification rate depends on ethnicity and geography [220]. Other HER2 abnormalities, such as mutations, are also known to occur; however, interactions between mutations and protein expression have not been studied extensively compared with amplifications.

HER2 overexpression identifies patients with advanced disease who might benefit from the addition of trastuzumab to a cytotoxic chemotherapy backbone. Trastuzumab is a recombinant humanized monoclonal antibody that targets the extracellular domain IV of HER2 and acts against overexpressed HER2 protein on the cell membrane, blocking the downstream effects [221]. (See "Initial systemic therapy for locally advanced unresectable and metastatic esophageal and gastric cancer", section on 'HER2-overexpressing adenocarcinomas'.)

Accurate determination of HER2 status is important, using a rigorous testing algorithm and robust definitions of gene amplification and protein overexpression to predict those patients who will most benefit from trastuzumab. In comparison with breast carcinomas, the heterogeneity of immunostaining for HER2 protein overexpression is greater in gastroesophageal adenocarcinomas, and the possibility of false-negative testing is higher [222,223]. Furthermore, HER2 protein expression in gastroesophageal adenocarcinomas tends to spare the digestive luminal membrane, resulting in membrane staining that is not completely circumferential in contrast to breast cancers [224-226]. These differences underscore the importance of utilizing tumor-specific criteria to assess HER2 expression in clinical practice. Most robust testing algorithms combine IHC and ISH techniques to assess HER2 status. In general, bright-field silver in situ hybridization (SISH) is superior to the dark-field ISH (eg, fluorescence in situ hybridization [FISH]) [227,228]. (See "Initial systemic therapy for locally advanced unresectable and metastatic esophageal and gastric cancer".)

Concordance of HER2 status between biopsy and resection specimens, as well as primary and metastatic sites, is very high but not perfect [229]. Heterogeneous amplification of HER2 in primary lesions is thought to be responsible for most of the discordant HER2 statuses of primary and metastatic lesions in gastric carcinoma [229,230]. However, concordance rates as high as 95 to 98 percent have been reported in paired samples [231].

Nevertheless, insufficient tumor representation in endoscopic biopsies may lead to a false-negative result. Analyzing HER2 positivity in <6 biopsy samples leads to a false-negative rate that may be as high as 9 percent [229,232-237]. The optimal number of tumor fragments that is required for testing of endoscopic biopsies is at least five [238,239].

Given the differences in interpretive criteria for determining HER2 status in gastroesophageal as compared with breast cancers [224,225,240-242], an expert panel convened by the College of American Pathologists (CAP), American Society of Clinical Pathology (ASCP), and American Society of Clinical Oncology (ASCO) undertook a systematic review of the published literature to provide an evidence-based joint guideline on HER2 testing and clinical decision making in gastroesophageal adenocarcinomas [222]. Among the panel's recommendations were the following:

●For biopsy or resection specimens, a minimum of five specimens, optimally six to eight, should be tested to account for intratumoral heterogeneity (when possible). If there is documentation of HER2 positivity on any specimen, the treating clinician does not need to request additional HER2 testing on additional specimens.

HER2 testing on fine needle aspirate specimens is an acceptable alternative. However, specimens obtained in cytology specimens may not be truly representative given the limited sampling of the tumor. If there is concern about specimen adequacy and HER2 testing is negative, additional available primary or metastatic tumor tissue should be tested.

●When HER2 status is being evaluated, laboratories/pathologists should perform/order IHC staining first. Pathologists should use the Ruschoff/Hoffmann method in scoring IHC results (table 11) and should select the tissue block with the areas of lowest grade or intestinal morphology for testing. A positive result is IHC 3+. A negative result is IHC 0 to 1+ [222].

●FISH or another ISH method is recommended only when the IHC result is 2+ (equivocal) to determine amplification status. In many studies, ISH-positive results are observed in 30 to 50 percent of IHC 2+ tumors. A variety of in situ visualization techniques to evaluate HER2 amplification, including FISH and brightfield ISH using either a HER2 probe or dual HER2 and centromere (chromosome enumeration probe 17 [CEP17]) probes, are all acceptable strategies. A ratio of HER2 signal to CEP17 signal ≥2 is considered positive, while a HER2:CEP17 ratio <2.0 is considered negative. An algorithmic approach to assessing HER2 status on surgical and biopsy material is presented in the figure (algorithm 1).

The initial management of patients with HER2-positive advanced gastric and gastroesophageal junction adenocarcinoma is discussed separately. (See "Initial systemic therapy for locally advanced unresectable and metastatic esophageal and gastric cancer", section on 'HER2-overexpressing adenocarcinomas'.)

PD-L1 overexpression and deficient mismatch repair — The programmed cell death receptor 1 (PD-1) is a key immune checkpoint receptor that is expressed by activated T cells [243,244]. Tumors use the PD-1 pathway to evade immune surveillance. PD-1 binds to its ligands PD-L1 (B7-H1) and PD-L2 (B7-DC), which are overexpressed on tumor cells, thereby preventing the immune system from rejecting the tumor. (See "Principles of cancer immunotherapy", section on 'Tumor evasion of immune surveillance' and "Principles of cancer immunotherapy", section on 'PD-1 and PD ligand 1/2'.)

Treatments are available that target both the PD-1 and PD-L1 molecules, which remove the "brake" on immune surveillance, thereby permitting the body's immune system to recognize and destroy the tumor cells, despite their overexpression of PD-L1 and PD-L2. Pembrolizumab and nivolumab, therapeutic antibodies targeting PD-1, provide durable antitumor activity in advanced PD-L1-expressing esophagogastric cancer; responses seem to be less frequent (but not absent) in patients with PD-L1-nonexpressing tumors. The issue of using biomarkers such as PD-L1 to predict benefit from PD-1-blocking antibodies is in evolution, and discussed in more detail elsewhere. (See "Initial systemic therapy for locally advanced unresectable and metastatic esophageal and gastric cancer", section on 'Immunotherapy-based regimens' and "Second and later-line systemic therapy for advanced unresectable and metastatic esophageal and gastric cancer", section on 'PD-L1 overexpression'.)

Results are more consistent for using defective mismatch repair (dMMR) as a biomarker to indicate potential efficacy of immune checkpoint inhibitors. It has been hypothesized that across a wide spectrum of tumors, including gastric cancer, those that lack the mismatch repair mechanism (ie, dMMR) harbor many more mutations (ie, they are hypermutated) than do tumors of the same type without such mismatch repair defects, and that the neoantigens generated from mutations such as these have the potential to be recognized as "non-self" immunogenic antigens. The biologic footprint of dMMR tumors is high levels of microsatellite instability (MSI-H). In fact, patients with metastatic gastric cancer whose tumors are MSI-H or dMMR (but not those with proficient mismatch repair [pMMR] tumors) as well as those with a high tumor mutational burden experience clinical benefit from PD-1 inhibitors, and some responses are durable. Pembrolizumab is now approved for treatment of advanced disease in a variety of tumor types, including gastric cancer, with either dMMR or high levels of tumor mutational burden. (See "Tissue-agnostic cancer therapy: DNA mismatch repair deficiency, tumor mutational burden, and response to immune checkpoint blockade in solid tumors", section on 'Clinical efficacy of anti-PD-1 therapy' and "Second and later-line systemic therapy for advanced unresectable and metastatic esophageal and gastric cancer", section on 'Biomarkers and benefit from PD-1 inhibitors'.)

SOCIETY GUIDELINE LINKS — Links to society and government-sponsored guidelines from selected countries and regions around the world are provided separately. (See "Society guideline links: Gastric cancer".)

SUMMARY

●Gastric adenocarcinomas represent a clinically, biologically, genetically, and pathologically heterogeneous group of malignant epithelial tumors resulting from various environmental and genetic causes. (See 'Etiology and risk factors' above.)

●Helicobacter pylori infection plays an important role in gastric carcinogenesis. Gland-forming adenocarcinomas (ie, those of the tubular, papillary, mucinous, and mixed types) are causally related to H. pylori and characterized by a defined series of preneoplastic stages, which are not seen with poorly cohesive-type gastric cancers. Importantly, only a small minority of individuals infected with H. pylori develops gastric cancer, and it is thought that modulation of the effects of chronic infection by genetic susceptibility, environmental factors, and H. pylori bacterial strain differences all influence the evolution into a neoplastic or nonneoplastic process. (See 'Helicobacter pylori' above.)

●Familial aggregation of gastric cancer occurs in around 10 to 20 percent of patients with gastric cancer, fewer than 5 percent of cases result from an inherited predisposition to cancer. One of these syndromes, hereditary diffuse gastric cancer (HDGC), is an autosomal dominant cancer susceptibility syndrome characterized by signet ring cell (diffuse) gastric cancer and lobular breast cancer; it is caused in most cases by a germline defect in the CDH1 (E-cadherin 1) gene. (See 'Hereditary diffuse gastric cancer' above.)

●Gastric adenocarcinomas have historically been divided into two distinct histomorphologic subtypes, intestinal (ie, gland-forming) and diffuse (composed of discohesive cells), which have a distinct morphologic appearance, epidemiology, pathogenesis, and genetic profile. The most recent World Health Organization (WHO) classification of tumors of the digestive tract recognizes several important histologic types of malignant epithelial tumors, which include gland-forming types (tubular, papillary mucinous, mixed) and poorly cohesive types (including the signet ring phenotype) (table 8). (See 'World Health Organization classification' above.)

●Broadly, the morphologic differences are attributable to different genetic and epigenetic alterations, some related to intercellular adhesion molecules, which are preserved in intestinal-type tumors and defective in diffuse carcinomas. A lack of adhesion molecules in poorly cohesive carcinomas allows the individual tumor cells to grow and invade neighboring structures without the formation of tubules or glands. Diffuse-type cancers are highly metastatic and characterized by rapid disease progression and a poor prognosis. The main carcinogenic event is loss of expression of CDH1, a key cell surface protein for establishing intercellular connections. Biallelic inactivation of the gene encoding E-cadherin (CDH1) can occur through germline or somatic mutation, allelic imbalance events (eg, loss of heterozygosity), or epigenetic silencing of gene transcription. In direct contrast, the pathogenesis of intestinal-type gastric cancers is less well defined. However, it appears to follow a multistep progression that is usually initiated by H. pylori infection. (See 'Intestinal versus diffuse types' above.)

●Several studies have provided unprecedented insight into gastric cancer genome diversity, revealing distinct genome-driven subtypes, novel oncogenic drivers, and candidate biomarkers. This novel taxonomy of gastric adenocarcinomas is critically needed to advance understanding of this deadly disease and tailor treatments as more targeted therapies are developed and approved that, hopefully, will either replace or be used in association with cytotoxic chemotherapy. (See 'Molecular subtyping' above.)

Although the molecular characterization of gastric cancers as described above has identified gene signatures that are prognostically relevant, they are still inadequate to direct molecularly targeted therapy, with few exceptions. To date, there are only three therapeutically relevant, routinely tested molecular biomarkers in gastric carcinoma: overexpression of human epidermal growth factor 2 (HER2 [ERBB2]), which permits the selection of patients with advanced disease who might benefit from trastuzumab, and overexpression of programmed cell death ligand 1 (PD-L1)/deficient mismatch repair (dMMR), or high levels of tumor mutational burden, all of which may identify patients with advanced disease with the potential to benefit from immune checkpoint inhibitor immunotherapy. (See 'Clinical implications' above.)