Field defects in progression to gastrointestinal tract cancers

Abstract

A field of defective tissue may represent a pre-malignant stage in progression to many cancers. However, field defects are often overlooked in studies of cancer progression through assuming tissue at some distance from the cancer is normal. We indicate, however, the generality of field defects in gastrointestinal cancers, including cancers of the oropharynx, esophagus, stomach, bile duct, pancreas, small intestine and colon/rectum. Common features of these field defects are reduced apoptosis competence, aberrant proliferation and genomic instability. These features are often associated with high bile acid exposure and may explain the association of dietary-related factors with cancer progression.

1. Introduction

A “field defect” or “field of cancerization” is a region of tissue that precedes and predisposes to the development of cancer [1]. As we show here, field defects have been identified in most of the major areas subject to tumorigenesis in the gastrointestinal tract. The generality of GI field defects has only recently become apparent, and thus is not yet widely recognized. Field defects are of interest because they give insight into the early stages of progression to cancer, and may, in the future, provide biomarkers of cancer risk. We will focus here on cancers of the GI tract, as there is substantial evidence for field defects preceding these types of cancer, as well as evidence for significant common features among the particular field defects leading to the different types of GI cancer.

Initiation of a solid tumor is widely considered to occur by a series of somatically inherited changes (i.e. mutations or transmissible epigenetic events (epimutations) such as methylation of CpG islands [2]). These changes are thought to contribute to early progression, usually by producing an inherited proliferative advantage relative to surrounding cells. In addition, genetic instability or a mutator phenotype, may accelerate this process of natural selection of mutant cells [3]. If, in a normal population of dividing cells within a tissue, a cell acquires a proliferative advantage through a mutation or an epimutation, it will tend to expand clonally and replace neighboring cells. Thus, a patch of abnormal cells will arise. Within this patch, a second such mutation or epimutation may occur so that a particular cell acquires a proliferative advantage compared to other cells within the patch, and this cell then expands clonally forming a secondary patch within the first patch. Within this new patch, the process may be repeated several more times over an extended period, perhaps decades, until a malignant cell arises which clonally expands into a cancer (Figure 1). If this is the general process by which solid tumors arise, then tumors generally should be associated with a field defect in neighboring tissue that may, or may not, appear superficially “normal”, but in actuality reflects a succession of pre-malignant events. Below, we describe the substantial and varied evidence for field defects in the different cancers of the GI tract starting with oropharyngeal cancer and then proceeding distally to colon cancer.

Fig. 1.

Colonic epithelial patches of field defects arising and expanding within pre-existing areas with field defects, in progression to adenocarcinoma. The dark area represents an adenocarcinoma.

To orient further discussion of the various field defects, we show in Figure 2 a probable general pathway for how a field defect may develop. However, development of a particular field defect may not proceed through all of the steps shown, may have additional steps, or information relevant to some of the steps may be lacking.

Fig. 2.

Probable general pathway for how a field defect may develop.

2. HNSCC (head & neck squamous cell carcinoma) (oropharyngeal/laryngeal cancer)

Patients with HNSCC have a considerable risk of a second primary malignancy, after a first malignancy has been surgically removed. Second primary tumors are a major cause of treatment failure and death in patients with HNSCC. The overall incidence rate of a second primary tumor is about 13–14% [4,5], although this may include instances of tumors arising because of incomplete surgical resection. Thus for individuals with a HNSCC, regions of the upper aerodigestive tract are at increased risk for developing additional malignant lesions, by a process referred to as “field cancerization”. The premalignant field is characterized by allelic imbalance [6,7], as well as aneuploidy and chromosomal aberrations [8]. The genetically altered field of cells shows a high proliferative capacity [7]. The field appears to be principally of monoclonal origin with multiple subclones that arise by clonal divergence [9]. Some oral cancers arise in a predominantly white lesion of the oral mucosa referred to as leukoplakia. These particular lesions represent a field defect that can develop into carcinoma at a rate of about 2–3% per year [10].

3. Esophageal adenocarcinoma

Barrett’s esophagus (BE) is a pre-malignant lesion resulting from chronic gastroesophageal reflux disease (GERD) in which the squamous epithelium of the distal esophagus is replaced by metaplastic columnar epithelium. About 10% of individuals with chronic GERD develop BE, and patients with BE are at least 30-fold more likely than GERD patients without BE to develop adenocarcinoma, one of the most lethal of all cancers, BE is one of the clearest examples of a field defect, as it is readily distinguishable from the usual normal squamous epithelium. At the gross level, BE usually appears as a well-defined area, with irregular margins, consisting of salmon-pink, velvety mucosa similar to the adjacent gastric mucosa. At the histological level, its defining feature is the presence of goblet cells, similar to those of the epithelium of the small intestine and colon. Barrett’s metaplasia is characterized by active as well as chronic inflammation [11] suggesting that inflammation contributes to the BE field defect.

Gastric acid and bile acids, present in reflux, are two major risk factors for the development of BE [12]. Clinical studies have identified several primary bile acids as well as the secondary bile acid, deoxycholate, in the refluxate of BE patients[12]. Bile acids cause oxidative stress, DNA damage, and apoptosis [13]. Normal squamous epithelium is sensitive to induction of apoptosis by deoxycholate, whereas BE metaplasia is apoptosis-resistant [14]. Anti-apoptotic proteins Bcl-x_L, and Mcl-1 are expressed at higher levels in BE metaplastic tissue than in nearby squamous epithelial tissue, indicating that these proteins may be contributing to the observed apoptosis resistance of BE [15]. These findings suggest that exposure of the normal squamous epithelium of the esophagus to bile acids at low pH may cause apoptosis of the squamous cells, and repeated exposure may select for apoptosis resistant BE cells [14].

Progression to adenocarcinoma is a process of clonal evolution within BE. The most common known genetic (and epigenetic) alteration in BE is inactivation of the tumor suppressor gene CDKN2A [also known as p16, or p16(INK4)][16]. Loss of CDKN2A function leads to loss of normal cell cycle arrest in late G1. CDKN2A alterations are found in about 85% of BE patients, and inactivation of both alleles of CDKN2A appears to be an early event causing clonal expansion [16]. The same mutation, or pattern of loss of heterozygosity of CDKN2A, can be observed in hundreds of thousands of crypts across centimeters of BE tissue, suggesting that CDKN2A inactivation is associated with selective clonal expansion.

Mutations in the p53 gene are not ordinarily seen in non-displastic BE, but occur in about 57% of adenocarcinomas [17]. The specific p53 mutation found in an adenocarcinoma is frequently identical to the p53 mutation found in the associated region of dysplasia [17], suggesting that the adenocarcinoma arose in a dysplastic BE field. Clonal proliferation of karyotypically abnormal cells is frequent in BE [18]. There is increasing genomic instability during premalignant neoplastic progression as detected by high resolution array comparative genomic hybridization [19]. Tetraploidy and aneuploidy are typically seen after p53 mutant cells are apparent, suggesting that p53 loss may lead to chromosome instability [16].

There is an increase in proliferating cells and an expansion of the proliferative compartment in BE metaplasia [20]. Expression of E-cadherin declines as the BE→dysplasia →adenocarcinoma sequence progresses, and this reduction may stimulate invasion and metastasis [21]. Other genetic events occurring early in the neoplastic progression of BE metaplasia include loss of the tumor suppressor genes Adenomatosis Polyposis Coli (APC), Retinoblastoma (Rb), and Deleted in Colorectal Cancer (DCC)[20].

The normal appearing squamous epithelium of the esophagus of patients with BE and BE-associated adenocarcinoma has numerous alterations in gene expression compared to squamous epithelium of normal individuals without metaplasia or neoplasia [22]. These findings indicate that BE metaplasia may arise in a pre-existing field defect of the squamous epithelium.

Thus, progression to esophageal adenocarcinoma appears to be stimulated by bile acid and gastric acid exposure and may involve an initially altered field of squamous cells, which then gives rise to a field of BE metaplasia. In the BE field, selective proliferation of CDKN2A defective cells, followed by further selective proliferation of p53 mutant cells, may lead to increased genetic instability and alterations in gene expression that promote increased cell division, apoptosis resistance, invasion and metastasis.

4. Esophageal squamous cell carcinoma

Increased risk of esophageal squamous cell carcinoma in the U.S. is associated with smoking and alcohol intake [23]. A systematic characterization of the esophageal mucosa nearby squamous cell carcinomas indicated that mutational defects in the p53 protein are multifocal, and occur in early precursor lesions [24]. These observations were considered to support the field cancerization concept for the origin of esophageal squamous-cell carcinoma.

5. Gastric cancer

While there has been a marked decline in distal gastric cancer in the US and elsewhere, the incidence of adenocarcinoma of the proximal gastric cardia has been increasing, particularly in Western countries [25]. Gastroesophageal reflux appears to play an important role in the development of adenocarcinoma of the gastric cardia [25,26], and to have an etiology similar to esophageal adenocarcinoma [25]. Most non-cardia (distal) gastric adenocarcinomas arise as a long term complication of Helicobacter pylori infection of the stomach [25]. H. pylori infection triggers the progressive sequence of chronic gastritis, gastric atrophy, intestinal metaplasia, dysplasia and finally gastric adenocarcinoma. H. pylori infection potently induces methylation of CpG islands in the DNA of the gastric mucosa and this accumulation of aberrant methylation constitutes a field defect for gastric cancers [27]. Aberrant DNA methylation of CpG islands in promoter regions can permanently inactivate tumor suppressor genes, as can mutations. Even in H. pylori–negative individuals (who may have had past exposure), levels of CpG methylation of the gastric mucosa were significantly increased in individuals with a single gastric cancer compared to individuals without gastric cancer. Methylation in the gastric mucosa was even further increased in individuals with multiple gastric cancers [28]. These findings indicate that development of a field defect characterized by epigenetic CpG methylations is an early event in progression to gastric cancer. The Akt (protein kinase B) oncogenic signaling pathway is increased in both primary stomach cancers and adjacent normal appearing gastric tissues [29]. This finding further indicates the presence of a pre-neoplastic field defect in morphologically normal gastric mucosa that has, in addition to increased CpG methylation, increased expression of the Akt signaling pathway.

After surgery for gastric cancer, the remnant stomach has a 1.8% to 2.4% chance of developing a further metachronous gastric cancer [28]. After endoscopic resection of gastric cancer, which preserves a larger portion of the stomach, the cumulative incidence of further metachronous gastric cancer(s) is as high as 8.5% to 14.0% [28]. Although some of these cancers may reflect incomplete surgical resection, these observations indicate the presence of a field defect remaining after the removal of the initial cancer. In the remnant stomach of rats after gastrectomy, bile acids are implicated in gastric cancer due to duodenogastric reflux [30]. In humans, duodenogastric reflux is also implicated in gastric stump carcinoma [31].

The ability of H. pylori to establish a chronic infection and to stimulate a chronic and active inflammation with the production of a wide array of inflammatory mediators is probably important in the development of gastric malignancy [32]. Areas of chronic gastritis caused by infection, and in which gastric cancers arise, are early field defects. Many of the mediators and byproducts of inflammation are mitogenic and mutagenic [33]. The release of pro-inflammatory cytokines, reactive oxygen species, and up-regulation of Cox-2 all may contribute to an environment conducive to neoplastic transformation. The mechanisms likely involve direct DNA damage [34], inhibition of apoptosis, subversion of immunity, and stimulation of angiogenesis. Interleukin 1β (IL-1β) is an important pro-inflammatory cytokine in the context of H. pylori infection. Particular polymorphisms of IL-1β are associated with increased risk of non-cardia gastric cancer in the presence of H. pylori [35].

6. Bile duct cancer (cholangiocarcinoma)

Cholangiocarcinoma (CC) is an adenocarcinoma arising from the bile duct epithelium. Those CC’s that arise within the liver are referred to as intrahepatic CCs. Those that arise at the confluence of the right and left hepatic duct are hilar CCs, and those that arise between the duodenal papilla (or ampula of Vater) and the hepatic hilum are extrahepatic CCs [36].

Primary sclerosing cholangitis (PSC) is a chronic inflammatory disease of the intrahepatic and extrahepatic bile ducts that is progressive and ultimately leads to liver cirrhosis in a large proportion of patients. PSC predisposes to the development of CC in about 12% of patients [37]. In PSC the anti-apoptotic protein Mcl-1 is strongly expressed in bile duct epithelial cells through the action of the IL6/AKT survival signaling pathway [38], suggesting the presence of an apoptosis resistant field. Most adenomas and carcinomas of the extrahepatic bile ducts arise in the region of the duodenal papilla (or papilla of Vater) [39], suggesting the possibility of a field defect caused by elevated exposure to bile acids.

7. Pancreatic cancer

The most common type of pancreatic tumor in humans is ductal adenocarcinoma. In a transgenic mouse model of pancreatic ductal adenocarcinoma (in which transforming growth factor alpha is overexpressed and p53 is deficient), the development of the carcinoma is preceded by pre-malignant transdifferentiation of acinar cells to ductal-like cells [40]. Bcl-x_L, an anti-apoptotic protein, is highly expressed in pre-malignant tubular complexes formed of ductal-like cells compared to acinar cells in normal appearing areas. This increased expression is due to induction by the transcription factors Stat3 and NF-κB. Bax, a pro-apoptotic protein, shows the opposite pattern of expression [40]. These observations indicate a shift in the balance of pro-apoptotic and anti-apoptotic proteins in a pre-malignant field (field defect) of ductal-like cells in the direction of apoptosis resistance. In epidemiological studies, ingestion of a Western style high fat diet is correlated with incidence of pancreatic cancer, and dietary fat stimulates bile acid secretion [41]. Since bile acids are known to activate NF-kappaB through multiple mechanisms [42], the transgenic model of pancreatic ductal adenocarcinoma is consistent with this anti-apoptotic pathway contributing to cancer progression. Most adenocarcinomas of the pancreas occur in the head of the gland, which is in close proximity to bile (Figure 3), leading to the suggestion that bile acids have a role in the pathogenesis of pancreatic cancer [41].

Fig. 3.

The diagram on the left shows the location of the hepatic ducts, the common bile duct, the stomach, duodenum, pancreas and how the bile duct and pancreatic duct form the duodenal papilla where the bile and pancreatic enzymes enter the duodenum. The diagram on the right shows an enlarged area containing the duodenal papilla, with the duct mucosa indicated and red irregular forms indicating where some tumors may form.

Another type of pancreatic cancer is intraductal papillary-mucinous tumor (IPMT). Multiple, distinct K-ras mutations of different ductal hyperplasias arise in a given pancreas (and are likely due to an underlying field defect in IPMTs [43].

8. Cancer of the small intestine

In Los Angeles County (California, U.S.A.), 53% of adenocarcinomas of the small intestine arose in the duodenum even though the duodenum comprises only 4% of the entire length of the small intestine [44]. Furthermore 57% of these duodenal cancers occurred in the 6–7cm segment containing the duodenal papilla (or papilla of Vater) that accounts for only about 1% of the entire length of the small intestine. This is the place where the duodenal papilla, formed by the joining of the common bile duct and pancreatic duct, empties bile and pancreatic secretions into the duodenum (see Figure 3)[39]. Small intestinal cancers have been particularly well studied in persons with familial adenomatous polyposis who have a germ-line mutation in the APC gene. In this condition, there is an increased risk for adenomas and cancers of both the small and large intestines. In the small intestine, these lesions occur mainly in the duodenum near the duadenal papilla (or papilla of Vater) [39] where their distribution parallels mucosal exposure to bile [45,46]. These findings suggest that, in addition to the increased susceptibility to cancer caused by the germline APC mutation, elevated bile acid exposure near the duodenal papilla may give rise to clones of aberrant cells derived from this local field of increased susceptibility.

9. Colon Cancer

Sporadic adenocarcinoma, the most common type of colorectal cancer, appears to develop by a pathway involving the following stages: normal flat mucosa, formation of a field of defective flat mucosa, aberrant crypt foci, adenoma with low grade dysplasia, adenoma with high grade dysplasia, and adenocarcinoma. However, it is currently uncertain whether all sporadic adenocarcinomas progress through each of these stages. Proteomic analysis has revealed numerous changes in protein expression in a field of morphologically normal appearing mucosa of patients with colorectal neoplasia [47]. Gene expression is also altered in macroscopically normal colonic mucosa from individuals with a family history of sporadic colon cancer [48]. In addition, alterations of gene expression occur in the normal appearing colonic mucosa of human colon cancer patients that corresponded to alterations found in the normal appearing colonic mucosa of cancer prone APC^min mice [49]. In 46% of colonic tumors, there is methylation of the promotor of the DNA repair gene O⁶-methylguanine-DNA methyltransferase (MGMT). Among the tumors that had MGMT methylation, 94% of samples from apparently normal mucosa associated with the tumors also had MGMT promoter methylation. Often MGMT methylation was detected in samples taken as far as 10 cm away from the tumor, indicating a field defect [50].

Patients with resected large adenomas (adenomas ≥ 1.5 cm) and thus at increased risk of colorectal cancer, had imbalance of proliferation and apoptosis in their left colon and sigmoid rectum compared to patients who never had a tumor [51]. Patients with resected adenomas of the colon or rectum had different expression of sialic acids in their “uninvolved” colon and rectal mucosa, compared to sialic acids of colonic or rectal mucosa of patients without colorectal tumors [52], indicating a field defect with respect to sialic acids in the patients with colorectal cancer.

Bcl-x_L is an anti-apoptotic protein that inhibits apoptosis by preventing release of cytochrome c from the mitochondria. Bcl-x_L expression increased at 1 cm and 10 cm away from colorectal adenocarcinomas in non-neoplastic colorectal mucosa suggesting the presence of an apoptosis resistant field [53]. A similar finding was also reported for the anti-apoptotic protein Bcl-2 [54]. Expression of the anti-apoptotic messenger RNA Survivin in colon tumors predicts poor patient survival due to subsequent recurrent colorectal carcinomas [55]. About half of patients with Survivin positive tumors had normal appearing mucosa that also expressed this messenger RNA. This suggests that Survivin positive tumors often arise in a Survivin positive field of normal appearing, but apoptosis resistant, mucosa.

A review of numerous studies on cell proliferation pattern in individuals with colon adenomas or cancer indicates that expansion of the proliferative compartment (a lumenward displacement of the zone of active cell proliferation in the crypts) is present along the entire colon of individuals at increased risk, no matter where the lesion is [56].

A high fat, low fiber Western style diet, as well as elevated fecal bile acid concentration are important risk factors for colon cancer (reviewed in [13]). Bile acids induce oxidative/nitrosative stress, DNA damage and apoptosis in the colonic epithelium [13]. Epithelial cells of the flat non-neoplastic colonic mucosa of individuals with colon cancer often have reduced capacity to undergo induction of apoptosis by bile acid compared to epithelial cells of individuals without neoplasia [42,57–59]. In rats, feeding a diet containing 0.2% cholic acid resulted in the development of increased resistance to apoptosis of colon crypt cells [60]. Apoptosis resistant crypts may spread through the flat mucosa by a process of crypt fission [61]. These findings suggest that repeated exposure to high levels of bile acids as a consequence of a Western style diet may select for cells resistant to induction of apoptosis giving rise to a field of apoptosis-resistant epithelium that may further evolve to malignancy.

A recent publication [62], in fact, indicates the crucial role a Western style diet plays in allowing a field defect to progress to colon cancer. In this study, 1009 patients had resected stage III colon cancer, and likely had a field defect, with colon cancer recurring in 324 patients in a median of 5.3 years, or 32% recurrence (although incomplete surgical resection may also have been partly responsible). These patients all had initiated fluorouracil-based adjuvant chemotherapy after surgery, and were evaluated for diet during and 6 months after adjuvant chemotherapy. Compared with patients in the lowest quintile of Western dietary pattern, those in the highest quintile experienced an adjusted hazard ratio for disease free survival of 3.25. On the other hand, eating a “prudent” dietary pattern (low intake of milk fat, refined grains, red meat, sweets, French fries, etc.) was not significantly associated with cancer recurrence or mortality.

10. Colon cancer in the setting of chronic ulcerative colitis

Chronic ulcerative colitis (UC), an inflammatory bowel disease (IBD), is characterized by a field defect of abnormal mucosa at increased risk for colon cancer. Exposure of the colonic mucosa to elevated levels of the bile acid deoxycholic acid (DOC) may be a key etiologic factor in IBD, since mice fed a diet supplemented with DOC develop IBD which mimics the human disease [63].

Microsatellite instability (MIN) is indicative of defective DNA mismatch repair. Whereas colonic tissue from individuals without UC shows no evidence of MIN, MIN occurs in about 40–50% of long term UC patients [64]. In some of the patients with cancer or high grade dysplasia, MIN occurred in tissue remote from the site of the cancer or dysplasia [64]. Also UC patients with high grade dysplasia or cancer show chromosome instability (CIN) throughout the colon [65]. These findings indicate that widespread genetic instability precedes dysplasia and cancer in UC. Both the nondysplastic and dysplastic mucosa of UC patients with neoplasia have fields of significantly elevated CpG island methylation compared with UC patients without dysplasia and non-UC controls [66]. Thus, UC patients with high grade dysplasia or neoplasia have pan-colonic genetic instability accompanied by widespread abnormalities in age-related methylation.

A study of p53 alterations in individual crypt cells from UC patients led to the conclusion that DNA alterations are initiated in colonic crypts and expand to adjacent crypts through crypt fission [67]. A continuous process of DNA mutation, clonal expansion through crypt fission and clonal succession appears to generate a defective field that initiates the development of inflammatory-associated colon cancer.

11. Synergistic effects of bile acids and other risk factors in GI cancer

The etiology of field defects is most probably multifactorial. As indicated above, diet plays an important role in GI cancer. However, other factors such as nicotine from smoking, alcohol consumption, low pH, viruses, bacterial and carcinogens can also increase cancer risk. It is known that bile acids plus low pH, and bile acids plus nicotine act synergistically in increasing oxidative stress and DNA damage [68,69]. Since there is cross-talk among various stress-response pathways, it is probable that other synergistic effects will be ascertained in the future.

12. Conclusions

Evidence indicates that cancers throughout the GI tract often arise in a field of tissue characterized by genetic instability and gene expression changes that provide a proliferative advantage. These fields appear to evolve by the processes of mutation (or epimutation) and natural selection, which may occur gradually over decades. Bile acids appear to play a prominant etiologic role in the evolution of field defects leading to GI cancers. Field defects have also been reported in other organs, such as lung, vulva, cervix, breast, bladder and skin (see [9] for references). Thus, the field defect may be a broadly general precursor to malignancy. If this supposition is valid, then the general practice of performing colonoscopies without obtaining biopsies for the evaluation of hypothesis-driven biomarkers of colon cancer risk may have to be changed in the future as biomarker research matures. In addition, the common practice of identifying markers of malignant progression by comparing gene expression in a tumor to that of neighboring non-involved “normal” tissue will often miss early changes already present in the surrounding field.