Controversies of cardiac glands in the proximal stomach: A critical review


  • Qin Huang

    1. Department of Pathology and Laboratory Medicine, the Veterans Affairs Boston Healthcare System and Harvard Medical School, West Roxbury, Massachusetts, USA
    Search for more papers by this author

Dr Qin Huang, Department of Pathology and Laboratory Medicine, Veterans Affairs Boston Healthcare System and Harvard Medical School, 1400 VFW Parkway, West Roxbury, MA 02132, USA. Email:


Cardiac glands (CG), along with oxyntocardiac glands, in a normal human constitute cardiac mucosa (CM) that is positioned in the proximal stomach with a length of 10–30 mm, according to traditional teaching. This doctrine has been recently challenged. On the basis of studies on autopsy and biopsy materials in the esophagogastric junction region, some investigators have reported the presence of CG in only 50% of the general US population. They believed that CG were an acquired, metaplastic lesion as a result of gastroesophageal reflux disease. Subsequent recent study results from other research groups showed the presence of CG in the proximal stomach in embryos, fetuses, pediatric, and adult patients in most Europeans and Americans, and almost all Japanese and Chinese patients. These new data showed the following important findings: (i) CG are confirmed to be congenital in the proximal stomach; (ii) the length of CM is much shorter, approximately 5 mm in Caucasians in Europe and North America, and approximately 13 mm in Japanese and probably also in Chinese; (iii) CG are also present in the distal superficial esophagus underneath squamous mucosa in almost all Japanese and Chinese patients, but not so common in Caucasians in Europe, and not clear in Caucasians in North America. The recent data indicate a clear difference in the distribution of CG in the proximal stomach among different ethnic populations, and might explain different disease pathogenesis mechanisms among various ethnic patient groups.


Cardiac glands (CG) in a healthy human consist of primarily mucus cells, scattered parietal cells, a few undifferentiated cells in the neck, and many endocrine cells in the base, but no chief cells.1–3 When the number of parietal cells increases, often in the basal half of gastric mucosa, parietal cells admix with mucus cells to form oxyntocardiac glands. CG show two growth patterns: (i) tubular, similar to gastric pyloric glands; and (ii) compound acinar or racemose, mimicking the duodenal Brunner glands. CG secrete mucus that forms a protective blanket on the gastric surface. At the subcellular level, these mucus cells are equipped with short microvilli at the apical surface and secretory granules in the apical cytoplasm, which can be highlighted with periodic acid–Schiff (PAS) reaction for carbohydrates and negative on the alcian blue stain at pH 2.5 or lower, which is similar to the staining pattern of mucus cells elsewhere in the stomach.1,2 Anatomically, CG and oxyntocardiac glands are mainly concentrated around the esophagogastric junction (EGJ) in a narrow zone and also clustered in small numbers at the upper esophagus, and rarely in other parts of the esophagus.2–4 In the distal esophagus, CG might be present underneath the squamous mucosa and above the muscularis mucosae; these are also known as superficial esophageal CG.4–7

Traditional teaching holds that the CG, along with oxyntocardiac glands, are congenital and form the cardiac mucosa (CM) in the most proximal part of the stomach, a transition zone of 10–30 mm in length, which abuts proximally with the esophageal squamous mucosa and distally with gastric fundic oxyntic glands.2,4 Recent research results, mainly from the Chandrasoma groups, challenge this doctrine.8–10 They reported that in the EGJ region of unselected adult autopsies, CG were present in only 29% of cases and oxyntocardiac glands in 44%; even in selected autopsies with the entire EGJ examined microscopically, CG were detected in only 44%. Recent autopsy studies further found that the length of the CM was in fact not 10–30 mm, but varied between 1 mm and 4 mm in pediatric patients,8,11 and approximately 5 mm in most adults.8,12 Therefore, CG, regardless of whether or not present in the proximal stomach or in the distal esophagus, are believed to be an acquired metaplastic lesion.10 As such, the CM is no longer considered to be part of the proximal stomach, but the distal esophagus.8–10,13 In supporting this notion, the study results from Europe and North America show similar clinical, molecular, and pathological features of adenocarcinomas arising in the EGJ region.14–16 In the most recently-updated cancer staging manual by the American Joint Committee on Cancer and International Union against Cancer, the entire section of gastric cardiac cancer involving the distal esophagus has been removed from the chapter on the stomach into that of the esophagus.17 The manual requires using the esophageal cancer-staging criteria for pathological staging of cancers arising in the proximal stomach with esophageal involvement.17 This dramatic change in paradigm has prompted us to critically review the histological evidence on CG in the proximal stomach published in the recent English literature.

Embryos and fetuses

In 1961, Salenius published a histological study on gastric mucosal development at different gestational ages.18 The formation of gastric pits was found at week 8 in all portions of the stomach, except for the pylorus and cardia. Parietal cells were the first differentiated, glandular cells, while pyloric glands and CG started to develop at week 13.18 These findings were confirmed by the data from a recent, similar study in 2003.19 Using the PAS–alcian blue stain, the investigators showed the presence of CG in all embryonic specimens.19 In their report, the CG formed a single layer of the epithelium lined with tall columnar cells, with a mucus-filled, apical cytoplasm containing both neutral and sialylated mucins, which differed from parietal cells conspicuously on hematoxylin–eosin (HE)-stained sections. Only neutral mucin was present in most superficial foveolar CG. From gestational age 15 weeks onwards, these mucus cells started to form the CG that opened into pits. At week 23, the squamous mucosa with remnant ciliated columnar cells replaced the primitive esophageal mucosa, and was positioned proximally to the CG. In the last trimester, the CG were further differentiated. At week 41, all CG secreted neutral mucin. The authors emphasized that both the CG and the CM were present in all sections of all cases19 (Table 1). This conclusion was similar to, but differs to some extent from, that of an earlier fetal study in 2001.20 In that report with routine, histological sampling, the CG and oxyntocardiac glands in the transition zone were found in 6% and 52% of cases, respectively.20 Both studies found that the number of CG increased with increasing gestational age.19,20 In 2003, Park et al.21 studied the same transition zone with either HE or PAS staining (Table 1). They found that this transition zone measured <0.4 mm, always contained oxyntic cells, but lacked CG in 20% of cases. This observation is different from that reported by De Hartogh et al.,19 who stated that the “CM was distal to, or straddled, the angle of His in all cases”.

Table 1.  Ontogeny of cardiac mucosa in humans
Ethnic populationnGestational age (weeks)Absence of cardiac mucosa (%)Presence of cardiac mucosa (%)Length (mm)Relation to ageReference
  1. NA, not available.

Korean1518–342278NANot related21

In summary, all but one21 study reported the presence of CG and the CM in the proximal stomach between esophageal squamous and gastric fundic glands in all embryos and fetuses (Table 1). The length of the CM was variable. The data establish the congenital nature of CG at this stage of human development.

Pediatric population

Several groups of investigators carried out the search mainly in pediatric autopsy cases (Table 2). In 2000, Chandrasoma et al.,8 in a prospective study with the entire EGJ examined microscopically, reported the absence of the CM in approximately half the cases, and the presence of the CM in the other half, but in a very short length (mean length: 0.3 mm). This study was criticized for poor specimen preparation and obvious autolysis, demonstrated in the images the authors published.8,22 In a well-designed and executed study,11 Kilgore et al. systematically investigated the histology of the entire EGJ on PAS-stained sections in consecutive autopsy cases without reflux esophagitis, Barrett esophagus, anti-acid medications, or Helicobacter pylori infection. They reported the findings of a normal squamocolumnar junction (SCJ), as the landmark of the EGJ, and the presence of CG in all cases, exclusively on the gastric side of the SCJ/EGJ. In their study, the length of the CM was 1.8 mm on average (Table 2). The shortcoming of their study was the lack of controls and clinicopathological information.23 In similar studies of newborns, infants, and young children–adolescents, the authors reported the presence of the CM distal to the SCJ in almost all cases, and concluded that the CM was congenital20,24 In a retrospective study of endoscopic biopsies at the place 1 mm below the SCJ, Glickman et al. reported the presence of the CM in all cases, including CG in 81%, and mixed oxyntocardiac glands in 19%.23 In no case did the authors identify pure oxyntic glands immediately below the SCJ. They concluded that the CM was congenital in the pediatric population. However, because of the overwhelming presence of inflammation in 88% of the cases and the absence of proper controls without inflammation, the authors could not determine with certainty the nature of the CG with regard to the physiologic gastric components or the pathological, metaplastic changes in response to gastric acid insults or Helicobacter pylori infection.23

Table 2.  Existence of cardiac mucosa in pediatric populations (≤20 years old)
Ethnic populationTissue typeEGJnAge (years)Absence of cardiac mucosa (%)Presence of cardiac mucosa (%)Length of cardiac mucosa (mm)Relation of cardiac mucosa to ageReference
  1. NA, not available; EGJ, esophagogastric junction.

USAutopsyEntire73–1843570.3 (0.07–0.7)?8
USAutopsyEntire30Neonate–1801001.8 (1–4)?11
USAutopsySelected100Neonate–1801001 (0.1–3)?24
KoreanAutopsyEntire8Neonate–1522780.17(0.1–0.27)Not related21

In summary, all but two8,21 studies in pediatric patients showed histological evidence of the presence of CG and the CM on the gastric side of the EGJ as a congenital, not acquired, gastric structure, but the length of the CM was short, less than 2 mm on average (Table 2). All studies used the distal end of squamous mucosa, i.e., the SCJ as the mucosal EGJ, which might be potentially problematic.25

Because the length of the CM increases with age, some investigators believe the CM to be metaplastic.26 It remains unknown as to: (i) the relationship between the length of CG, the CM, and developmental age; (ii) the status of superficial esophageal CG as a normal developmental structure; and (iii) whether or not there are the differences in the CG and the CM among different ethnic populations.

Adults in North America and Europe

In adults, the status of CG and the CM in the proximal stomach remains controversial (Table 3). A series of studies by the Chandrasoma groups showed morphological evidence of the absence of gastric CG and the CM in a substantial number of adult, US patients. For example, in an endoscopic histological study of the tissues biopsied above and below the EGJ in adults with 24-h pH monitoring and measurement of lower esophageal sphincter pressure, they reported the absence of CG and the CM in 26% of cases, and a statistically-significant association of the presence of CG with reflux esophagitis, as evidenced with an esophageal luminal pH <4, lower esophageal sphincter pressure, the presence of hiatal hernia, and active esophagitis.27 In a subsequent endoscopic biopsy study within 40 mm of the EGJ, they further showed a strong correlation between the length of CG and CM, and the amount of acid exposure in the esophagus.9 The results of this study were disputed with regard to the biopsy site, because it was not clear whether or not their biopsies included the SCJ, and the possibility of sampling errors in the proximal gastric fundic region was obvious.22,23 In addition, the absence of controls without inflammation makes their arguments weak. In a retrospective autopsy study with one selected EGJ section examined microscopically, the same investigators reported a complete absence of CG in 67% of cases and similar results (64%) from 11 prospective autopsies with the entire EGJ examined microscopically.8 The authors concluded that the CG were acquired as an early metaplastic response to inflammation related to gastric acid insult. This study was also criticized for poor preparation of autopsy EGJ specimens and obvious autolysis, which were present in the images that the authors published.22,28 In 2003, the Chandrasoma et al. published their histological study results on consecutive endoscopic biopsies at the EGJ.10 In that study, they defined CG and oxyntocardiac glands as ‘abnormal’ columnar mucosa that had a length of 1–40 mm, while the columnar mucosa with pure oxyntic glands was defined as ‘normal’. They reported cases with pure oxyntic glands, CG, and oxyntocardiac glands in 39%, 43%, and 18%, respectively, and the prevalence of intestinal metaplasia increased with the increasing length of the CM. They concluded that ‘cardiac mucosa is absent in over 50% of the general population. When present, its extent is in the 1–9 mm range in over 95% of the general population and approximately 85% of a population undergoing endoscopy’.10,29 These investigators advocated defining the proximal end of gastric fundic oxyntic mucosa as the true mucosal EGJ.13

Table 3.  Existence of cardiac mucosa in adult populations (≥20 years old)
Ethnic populationTissue typeEGJnAbsence of cardiac mucosa (%)Presence of cardiac mucosa (%)Length of cardiac mucosa (mm)Reference
EuropeanResectedEntire363975 (1–15)12

Several groups of investigators in Europe and North America conducted a series of studies in an attempt to confirm or refute the findings by the Chandrasoma groups. In 2002, German pathologists Sarbia et al.12 systemically investigated the histology of the entire EGJ in surgically-resected specimens for squamous cell carcinoma in patients with no gross evidence of the columnar-lined esophagus. They defined the EGJ as the macroscopic junction between brown–red gastric mucosa and gray esophageal mucosa. The authors reported the detection of CG and oxyntocardiac glands in 97% of cases, with a mean length of 5 mm (range: 1–15 mm). They did not identify any case that showed a direct transition of gastric fundic oxyntic glands to the esophageal squamous mucosa. In addition, they also reported the findings of the intra-esophageal presence of CG and oxyntocardiac glands above the deep esophageal glands and ducts in 25% of cases. However, in as much as 61% of cases in their report, there existed squamous islands among the CG or oxyntocardiac glands, which indicates that their analysis was actually in the tissues taken from the distal esophagus, not in the proximal stomach, since squamous islands are indicative of the esophagus.25,30

To contribute to the debate on the existence of gastric CG and CM, Marsman et al.22 conducted an endoscopic biopsy study in 63 of 198 unselected patients with biopsies at or immediately below the endoscopically-normal-appearing SCJ that was defined as the EGJ. They reported a uniform presence of the CM in the proximal stomach, including CG in 62% of cases and oxyntocardiac glands in 38%. Therefore, they concluded that the CG and the CM were congenital, not acquired.22 Their conclusion was confirmed in a similar study of volunteer health-care workers in the USA.31

In summary, all studies showed a consistent presence of CG and the CM on the gastric side of the EGJ in most, but not all, patients, which is slightly different from that shown in fetuses and pediatric populations. The length of the CM in this transitional zone is short; approximately 5 mm on average. The absence of the CM in over 39% of adult Americans, as reported by the Chandrasoma groups, has not been confirmed. It should be noted that most studies used the SCJ as the landmark of the EGJ, which might potentially be the source of errors.25

Adults in Japan and China

In contrast, recent studies on the distribution of CG in the EGJ region in adults from Japan and China show different results from those reported in Europe and North America. In Japan, Misumi et al. studied the relationship between the mucosal EGJ,32 which was defined as the distal end of esophageal longitudinal vessels, and CG in endoscopic biopsy patients without reflux esophagitis, ulcers, hiatal hernia, or tumors in the esophagus and stomach. They also systemically mapped CG in the entire EGJ region in additional, resected specimens for esophageal carcinoma and another cohort of resected specimens with cancers in either the lower esophagus or the proximal stomach. They reported the presence of CG in an area between 7.5 mm proximal and 13 mm distal to the EGJ. Histologically, the CM straddled the EGJ approximately 10 mm proximally and 20 mm distally.32 However, in the Results section of their paper, the description and the detailed mapping diagram for the distribution CG showed the actual distance between the CG and the EGJ in the scale of millimeters, not centimeters, as the authors used in the Abstract and other parts of the paper. It remains unknown whether or not this discrepancy was due to typographic errors. In a recent histological study in Japan, Nakanishi et al.28 investigated the entire EGJ macroscopically and microscopically in surgically-resected specimens for upper and middle esophageal squamous cell carcinomas. They reported the existence of CG in the proximal stomach in all cases, and superficial esophageal CG in 95% of cases, with mean lengths of 13 mm and 4 mm, and ranges of 2–64 mm and 1–26 mm, respectively. The results clearly indicate the presence of both gastric and superficial esophageal CG and CM in almost all Japanese patients. The superficial esophageal CG are believed to protect the squamous mucosa from acidic injury.7 Indeed, the Japanese Research Society of Gastric Cancer defines the gastric cardia as the region where the CG and the CM are located.33 However, the EGJ landmark used in that study was the angle of His. With this EGJ landmark, the authors could not confidently differentiate the mucosal EGJ from the columnar-lined esophagus that is not uncommon in the Japanese population.3

Recent endoscopic and histological study results in Chinese patients are similar to those reported in Japanese patients. For example, Law et al. performed endoscopic biopsies at or immediately below the SCJ, which showed a normal appearance in 94% of cases, and none with the SCJ shifted proximally towards the esophagus.34 The authors reported the presence of CG in 73% of cases in the proximal stomach below the SCJ/EGJ line, but did not describe the status of oxyntocardiac glands.34 In another histological study of the EGJ in 44 resected specimens for gastric cardiac cancer in Chinese patients, with 31 cases having the entire EGJ examined microscopically, Fan et al.5 used the most distal end of squamous mucosa, along with deep esophageal glands and ducts, as the landmarks of the EGJ to investigate the distribution of the CG.5,25 They found that the CG were distributed not only distally in the proximal stomach, with a mean length of 7 mm (range: 3–20 mm), but also proximally underneath the squamous mucosa into the distal superficial esophagus, with a mean length of 7 mm (range: 3–18 mm). In their report, chronic inflammation was present in 95% of cases, and 64% with Helicobacter pylori infection.5 The major limitations of their study included a small sample size and the potentially-confounding factor of cancer involvement in the tissues they studied.

In summary, the results from recent studies in Japanese and Chinese populations show a universal presence of CG and the CM in the proximal stomach, and also in the distal superficial esophagus, with approximate lengths of 13 mm distally and 7 mm proximally from the EGJ, which differs substantially from the data reported in Europeans and Americans. The major limitations of these studies include: (i) the lack of universally-accepted, histological gold landmarks to define the EGJ;25 (ii) the absence of controls without chronic inflammation or cancer involvement in the EGJ region; and (iii) the use of biopsy tissues in many studies.


The results of the most recent studies in fetal, pediatric, and adult populations confirm CG and CM in the proximal stomach as a congenital structure with a much shorter length than previously believed, and a various distribution pattern among different ethnic populations. In general, the length of CG and the CM in Europeans and Americans is mainly influenced by reflux esophagitis. In contrast, in Japanese and Chinese populations, in whom reflux esophagitis is not as common as in Caucasians, CG and the CM are almost always present, not only in the proximal stomach, but also in the distal superficial esophagus underneath the squamous epithelium in most cases. These differences between Caucasians and Asians might result in different clinicopathological characteristics of carcinomas occurring in this region between these two different ethnic patient populations.35 It appears that CG and the CM straddle the EGJ and encase the distal end of squamous mucosa, probably providing the squamous mucosa with protective mucin against insults of various toxic chemicals. This speculation remains to be investigated.

In summary, the following facts appear to emerge from the extensive studies in the recent English literature:

  • 1CG and the CM are present in the embryo–fetal stomach at term delivery, and thus, qualified as a congenital tissue of the proximal stomach.
  • 2The CM consists of both CG and oxyntocardiac glands and is concentrated in a narrow zone of the proximal stomach in a healthy person. The length of the gastric CM is approximately 5 mm on average, not 10–30 mm as believed, in most adult Europeans and Americans, and at least 13 mm in Japanese and probably also in Chinese patients.
  • 3CG are also present in the distal superficial esophagus underneath the squamous mucosa, which is common in Japanese and Chinese populations, uncommon in Europeans, and largely unknown in Americans. The average length of the superficial esophageal CM is approximately 7 mm in Japanese and Chinese patients.

The following issues remain to be clarified in upper gastroenterology: (i) the relationship between the length of the CG, the CM, and developmental ages; (ii) the status of superficial esophageal CG in Caucasians and other ethnic populations; (iii) the differences in the distribution patterns of CG among different ethnic fetal and pediatric populations; (iv) the use of universally-accepted, histological gold landmarks to define the EGJ;25 (v) comparison of CG and the CM in resected specimens among different ethnic populations; (vi) the genetic differences among proximal gastric, superficial distal esophageal, esophageal metaplastic CG, and the buried esophageal Barrett glands;36 (vii) functional significance of the CG in the EGJ region in the protection of the squamous mucosa; and (viii) malignant transformation of CG in the EGJ region.6


The author thanks Professor Raj K. Goyal, of the Veterans Affairs Boston Healthcare System and Harvard Medical School, for helpful discussion on the project and consistent support.