THE RISK OF CANCER IN THE DIGESTIVE TRACT
Geographic variations in incidence
In 2008, in the IARC GLOBOCAN database[1] the respective worldwide ASR incidence rate/100 000 for esophageal, gastric and colorectal cancer was at 19.2, 19.8 and 20.4 for men and at 4.2, 9.1 and 14.7 for women which are exposed to a lower risk than men, particularly for esophago-gastric cancer. However there are considerable variations of the risk between geographic regions or countries, and in the near future an increased number of incident cases is expected. This temporal trend is in relation to the global aging of the population, particularly in developing countries.
Esophageal squamous cell cancer occurs in relation to nutritional deficiencies and a low socio-economic status, and to alcohol and tobacco consumption. The incidence is relatively low in North America and Europe. Some areas of China, South Africa (Transkei), South America, are exposed to a much higher risk. In Asia, areas of high risk include Iran, Northen India and China, where the ASR incidence in men and women is as high as 183.3 and 123.1/100 000 in Cinxian county in 1993-1997. In Western countries the incidence tends to lower since 20 years in relation to a decreased consumption of alcohol. The prognosis of this cancer is still very poor, with a 5-year relative survival as low as 12.3 % in the European Union in 1995-1999[5,6].
Esophageal adenocarcinoma develops in Barrett esophagus. The risk of malignancy in this columnar lined esophagus is now evaluated as 1 case per 200 years/patient. The incidence of this cancer is higher in North America and Northen Europe than in Asiatic countries. In cancer registries without histology subgroups, the incidence of adenocarcinoma is joined to that of squamous cell cancer. In registries with histology subgroups the proportion of adenocarcinoma varies with the regions[9]; the proportions are 59.5% in the white non hispanic population of the United States, 44.9% in Denmark, 53.0% in the Netherlands contrasting with 3.2% in Japan, 4.1% in France 3.9% in Brazil in 1988-1992. In the United States, during the period 1976-1987, the incidence of esophageal adenocarcinoma still increased[3] with an annual progression between 4% and 10%. Adenocarcinoma in esophagus is linked to the following factors: caucasian, male sex, old age, alcohol consumption, continuous smoking, dietary deficiency in fruit and vegetable, and a long history of reflux symptoms. The 5-year relative survival is low, in the range 25% to 30%, after surgical treatment.
Stomach cancer develops in the background of chronic atrophic gastritis in relation to Helicobacter pylori (H. pylori) infection[10-15]. The incidence is high in countries of Eastern Asia including Japan, in South America, much lower in North America and in Europa. In 1988-1992 the respective ASR incidence rate/100 000 for men and women was at 7.2 and 3.4 in the United States, 7.1 and 3.2 in Denmark, contrasting with high figures in Japan: (51.3 and 19.8 in Osaka registry) or in Chile: (43.1 and 16.0 in Valdivia registry). The recent worldwide decrease in incidence applies to the distal, but not to the proximal stomach. This temporal trend is attributed to changes in life style, diet and control of H. pylori infection. In Japan during period 1963-1989, the incidence decreased by 47% in men and 43% in women. In Western countries most cases are detected at advanced stage and the 5-year relative survival is low: in 2002-2004, its median value is 24.9 % in Europe and 25% in the United States. Detection occurs at an earlier stage in Japan and in 1987-1989, the 5-year survival is as high as 47% in the Osaka registry[4].
Colorectal cancer is more frequent in developed than in developing countries of the world. Favouring factors include a high intake in calories from meat and fat, excess weight and a low physical activity. In 1998-2002 the respective ASR incidence/100 000 for men and women is at 38.6 and 28.3 in the United States, 39.3 and 29.8 in Denmark and 37.4, and 21.7 in Japan. In Japan the incidence increases dramatically since 1975. In contrast in the United States the risk is stable since 1985, with a decline in mortality. The decrease in incidence is linked to the detection (and treatment) of premalignant neoplastic lesions during colonoscopy. The 5-year survival progressed in the United States, from 52.3% to 65.2% in men and from 52.3% to 66.3% in women during period 1975-2002 (3). In Japan it progressed from 37.1% to 71.7% in men and from 32.8 to 72.1% in women during period 1962-1996[4]. In the European Union the 5-year survival for both sexes, progressed from 54% in 1995-1999 to 56.8% in 2000-2002[5,6].
Precursors and premalignant lesions
The term of precursor applies to temporal ancestors of neoplasia like chronic inflammation and to premalignant neoplastic lesions which progress to cancer in successive stages of localized, regional, and distant invasion. In the digestive tract[16-25], cancer is called “superficial” when invasion is limited to mucosa and submucosa. The assessment of a precursor depends on histology and on molecular markers such as inactivation of the suppressor gene TP53 encoding p53 protein, or the CDKN2A gene encoding P16; or amplification of CCND1 encoding cyclin D1 protein or overexpression of proto-oncogenes like epidermal growth factor receptor or c-myc. The Vienna classification of neoplasia in the digestive mucosa[26,27] applies to superficial neoplastic lesions in the stomach and colon: group 1: negative for neoplasia; group 2: indefinite for neoplasia; group 3: low grade intraepithelial neoplasia; group 4: high grade intraepithelial neoplasia and intramucosal cancer; group 5: invasive cancer in the submucosa. The depth of invasion of a superficial cancer in the submucosa is measured by micrometry. The legitimacy of endoscopic resection is ensured up to 200 μm in esophagus, 500 μm in stomach and 1000 μm in colon. In addition, qualitative indices of poor prognosis are found in tumor grade, images of vascular invasion, and tumor budding.
For esophageal squamous cell cancer, chronic non erosive esophagitis which has a high prevalence in Henan, China, Transkei in Africa is classified as a precursor of neoplasia[21]. In contrast reflux esophagitis which is often erosive is not a precursor[19].
For esophageal adenocarcinoma, columnar metaplasia, with or without areas of intestinal metaplasia, is a precursor showing increased risk of development of neoplasia. Columnar metaplasia is linked to gastro-esophageal reflux; the surgical correction of reflux will not suppress the risk of cancer in the metaplastic segment of mucosa. The prevalence of columnar metaplasia in the adult population is estimated at 1.6% in Sweden[24].
For gastric cancer, atrophic gastritis with infection by H. pylori is considered as a precursor with increased risk of development of neoplasia. This risk is linked to achlorhydria and the possible direct oncogenic role of the bacteria, implanted in the epithelial cells[10,12]. The sequence leading from chronic gastritis to cancer has been described by Correa[16]. Pre-malignant neoplastic areas in the gastric mucosa are usually flat and the majority of polypoid lesions like the cystic polyps are not neoplastic.
For colorectal cancer chronic inflammation of the mucosa is a precursor and there is an increased risk of neoplasia in inflammatory bowel disease (IBD). In contrast most cases of cancer occur in absence of inflammation in the mucosa, through the sequence adenoma-carcinoma. Polypoid and non-polypoid adenomas are premalignant neoplastic lesions which develop through genetic alterations like APC inactivation, KRAS mutation and TP53 inactivation. Recently distinct genomic profiles have been established by DNA sequencing in colorectal carcinogenesis[28-30]: serrated lesions including hyperplastic polyps and sessile serrated lesions provide a common pathway between neoplastic and non-neoplastic precursors.
SCREENING FOR CANCER IN THE DIGESTIVE TRACT
Opportunistic and organized screening
When asymptomatic persons, at the individual scale, are the target of early detection of cancer, screening is called opportunistic; in this situation endoscopy, the primary procedure of detection, is included in an individual health contract between patient and doctor. When screening for cancer is adopted as a population based national policy, by the heath authorities of a country, the intervention is called organized or mass screening; then endoscopy is performed only in those persons positive to a relatively simple filter selection test. Actually there is no clear cut distinction between indications of endoscopy in organized or in opportunistic screening. Opportunistic screening by primary endoscopy, with free examination at regular intervals, may be offered by an industrial company, or by health care insurances or hospitals to their employees of affiliated persons. In this situation the screening is organized, but not population based. In the near future cancer screening should undergo a strategic revolution with the development of new selection tests prior to endoscopy. Stool and blood DNA tests based on epigenetics and nanotechnology, will be more easy to collect and have a larger spectrum, with enough sensitivity and specificity.
For squamous cell esophageal cancer early detection in opportunistic screening is based on upper gastrointestinal (GI) endoscopy in asymptomatic persons smoking and drinking alcohol. Organized screening trials are proposed to high risk populations in some areas of China exposed to a high risk and in Southern Brazil. The selection test is a balloon or sponge scrapping for cytology followed by upper GI endoscopy in positive cases.
For esophageal adenocarcinoma in columnar lined esophagus, the risk is too low to justify opportunistic or organized screening in asymptomatic persons, in the absence of reflux symptoms.
For stomach cancer, opportunistic screening with endoscopy is of current practice in persons asymptomatic or complaining from dyspepsia in regions of the world exposed to a high risk, particularly in Asia; Population based screening trials are occurring since 1963 in Japan[8,31-34]; the selection test is photofluorography with 7 small films, or the measure of the Pepsinogen I/II ratio in the blood[32]. The filter test is followed by upper GI endoscopy in positive cases. The prolonged policy of organized and opportunistic screening in this country resulted in a higher proportion of early gastric cancer and an increased 5-year survival.
For colorectal cancer, opportunistic endoscopic screening with flexible sigmoidoscopy or colonoscopy in asymptomatic persons from age 50-year, is of current practice in regions of Western countries exposed to a high risk. Organized and population based screening protocols are also proposed in in Europe, North America and in Japan. The fecal occult blood test (FOBT) is still the current selection test, either as a Guaiac FOBT or as a immunochemical test, like in Japan; this test requires only one stool sample and no specific diet. Various randomized trials on the efficacy of the FOBT test have been conducted[35-40]: (1) the American Minnesota trial, conducted in volunteers, with an annual rehydrated guaiac test, achieves a reduction in mortality in the screened group of 33%[37]. However the rate of positive tests is very high as well as the proportion of screened persons submitted to colonoscopy; and (2) the 2 European population based trials (Nottingham trial, Funen trial) based on a biennial non rehydrated test show a lower reduction is mortality in the screened group (15%-18%) but are more cost/effective because the proportion of colonoscopies is much lower[35,36]. The sensitivity of the FOBT trials to the detection of premalignant colorectal lesions is fairly low; therefore there is no impact on the incidence of cancer.
Evaluation of screening for cancer
Evaluation is not applicable to opportunistic screening at the individual scale. In contrast, evaluation of benefits, drawbacks, and cost/effectiveness, including follow-up of the results, is required for organized population based screening. The intervention is launched by the health authorities of a country in age classes exposed to the risk, from the age 50-year. Priority in evaluation of the benefit is given to the prevalence of the concerned cancer, screening is worthwhile only if there is a sizable impact on the health of the population. Aging persons are exposed to multiple concurrent causes of mortality. A life saved from digestive cancer is not saved from other diseases. Screening trials in age classes prone to cancer have an excessively high cost and a negligible impact on the global mortality of the population.
The benefit of a screening intervention with early detection of premalignant and malignant lesions is shown by the impact on the 5-year survival and mortality from the targeted tumor site. This impact is analyzed on death statistics and, in reference to stage and treatment, in cancer registries.
The harms and drawbacks of screening are as follows: (1) the intervention has an impact on morbidity with the potential complications of endoscopy; and (2) the number of cases detected and their mortality is artificially increased as compared to data in cancer registries because some of the detected cases would not be the final cause of death. This factor is well acknowledged for the screening of prostate cancer with the PSA antigen.
Multiple interfering bias must be accounted for, in the evaluation: (1) the lead time bias applies to cancer detected at an early preclinical stage, then the length of the survival is prolonged by definition; (2) the selection bias applies to cases detected during the first years of the intervention; these cases usually have a slow evolution and better prognosis; and (3) the stage migration bias refers to progress in early detection of cancer occurring along the years and independently from the screening intervention. Similarly the progress in treatment improves the results and increases artificially the impact of the intervention on mortality and 5-year survival, this is why randomized trials are required.
ENDOSCOPY IN DIAGNOSIS OF CANCER
Strategy of endoscopic diagnosis
Nowaday the basic material used in the endoscopy unit fulfills the following requirements: a recent model of electronic video-endoscope in the HDTV standard with an adapted processor and monitor-availability of image magnification either with an optical or electronic zoom or an adaptative focal in the objective-a technique of image processing, of which neutrophil bactericidal index is the more reliable. Chromoscopy is often used during the step of detection, indigo carmine being the most frequent dye. In agreement with the Japanese school, endoscopic diagnosis is based on two successive steps of detection and characterization[41,42] with prediction of histology and staging of depth of invasion in order to adopt treatment decision between 3 categories: (1) abstention in non-neoplastic lesions; (2) endoscopic resection in neoplastic lesions fulfilling the criteria of complete cure; and (3) surgical treatment.when those criteria are not ensured.
Detection of an abnormal area in the mucosa requires a complete cleanliness of the lumen in upper GI endoscopy as well as in colonoscopy, abnormal areas are first detected in standard vision on any of the following criteria: (1) obvious elevation or depression; (2) mucosal discoloration; and (3) interruption in the course of superficial capillaries. Nonpolypoid lesions, even large, can be missed when the operator lacks cognitive knowledge and training for a slight change in the color of the mucosa[41]. Exploration of the proximal colon requires a special attention, because poorly visible and flat lesions are more frequent and often covered with mucus. Chromoscopy is then used with indigocarmine, a non absorbable dye at 0.5% dilution, or absorbable dyes like crystal violet (0.2%) or methylene blue (0.5%). The lesion is classified in size as diminutive (up to 5 mm), small or intermediate (5 to 9 mm), or large (10 mm and more). Lesions with a superficial appearance are classified in the categories of type 0 of the Paris classification[43]: (1) subtype 0-1, polypoid or sessile (0-1p, 0-1s); (2) 0-2 non-polypoid with variants elevated (0-2a) flat (0-2b) or depressed (0-2c); and (3) 0-3 ulcerated. These categories have a distinct prognostic value; depressed lesions, even small have a higher risk of malignancy than elevated lesions. In Japanese series the proportion of non-polypoid (0-1p) lesions is 84% in esophagus, 95% in stomach and 49% in colon.
Characterization follows detection and predicts the risk of malignancy. Prediction is based on morphologic subtype and on the microarchitecture, explored in magnification, with or without chromoscopy, and image processing[44-48]. The microarchitecture of the epithelium (pits grooves and crests) is called “pit pattern”; that of the superficial subepithelial capillaries is called “vascular pattern”. The prediction of massive invasion in the submucosa from the pit pattern and the vascular pattern is a contra-indication to endoscopic treatment.
Morphology of neoplastic lesions in the digestive tract
In the esophagus, most superficial type 0 lesions in squamous cell cancer are flat or depressed. In a metanalysis of 143 centres in Japan the distribution of subtypes was: 2.2% in 0-1, 12.6% in 0-2a, 31.4% in 0-2b, 53% in 0-2c and less 1% in 0-3 (in 43). In magnifying endoscopy the vascular pattern of the normal squamous epithelium is distributed in regular intrapapillary capillary loops (IPCL). At a very early stage of neoplasia the IPCL are elongated; later they disappear with development of irregular vessels with a large diameter. For type 0 adenocarcinoma developed in the columnar lined esophagus the non-polypoid morphology is also predominant. In magnifying endoscopy the pit pattern of neoplastic areas shows an abrupt change in size of epithelial crests, which are irregular and distorted or an amorphous surface suggesting submucosal invasion. The alterations of the vascular pattern also relate to the depth of invasion in the columnar epithelium. Both histological types of cancer in the esophagus vary in morphology and in the time trends of incidence in various countries[49].
In the stomach the usual morphology of superficial neoplasia is non polypoid. Lesions flat or depressed are detected as discoloured areas (pale or red). Flat neoplastic lesions are easily missed during gastroscopy: A recent analysis of the miss rate for early gastric cancer has been conducted in Japan in the Fukui registry, cross referenced with databases on gastroscopy: the sensitivity of this procedure for gastric cancer was 81% with 19% false negative[33]. In the stomach most polypoid lesions correspond to non-neoplastic fundic cystic polyps or hyperplastic polyps. In magnifying endoscopy the normal pattern of regular round pits in the mucosa of the gastric fundus is replaced by a gyrus and villous pattern with irregular epithelial crests. An amorphous surface pattern suggests submucosal invasion. Concerning the vascular pattern in the gastric fundus, the honeycomb network of capillaries around the neck of gastric pits is replaced by abnormal vessels with a mesh, coil or corkscrew appearance. In the antrum the coiled subepithelial capillaries around epithelial crests are replaced by irregular and large vessels.
In the colon and rectum, the upward growth of adenomas results in polypoid type 0-1 neoplastic lesions. Polyps over 10 mm in diameter, are called advanced adenomas, Non-polypoid adenomas are classified as 0-IIa. Depressed adenomas account for only 5% of neoplastic lesions, but invasion of the submucosa is frequent: the proportion reaches 35.9% in 0-2c lesions in a large series (n = 25 862) collected in Akita and Yokohama and 27% in another series[28,29] collected in Hiroshima. Laterally spreading tumors are large lesions, either granular or non-granular with an often depressed morphology. Sessile serrated lesions are non-neoplastic lesions, also large, over 10 mm, with a predominant lateral growth pattern; which have a significant risk of progression to serrated adenomas. Finally the majority of non neoplastic hyperplastic polyps are classified as non-polypoid 0-2a. In magnifying endoscopy the pit pattern of the colonic mucosa is described in categories: (1) type 1, normal, with small and regular pit opening; (2) type 2 with less contrasted pit openings for non-neoplastic hyperplastic lesions; (3) types 3l, 3s, 4, 5i, 5n for neoplasia with progression of irregularity of epithelial rests and branching from low to high risk of malignancy and submucosal massive invasion in type 5n. The vascular pattern is also classified in categories as faint, network; dense, irregular and sparse, the later suggesting submucosal invasion. The sensitivity of colonoscopy in the detection of colorectal cancer is superior to that of barium enema, with a higher proportion of Dukes-A stage among detected cases. Concerning premalignant adenomas the miss rate is high (27%) only for lesions, under 5 mm in diameter.
Endoscopic treatment of neoplastic lesions
After endoscopic diagnosis and prediction of the risk of malignancy of a lesion in the digestive mucosa with a superficial appearance there are 3 options in treatment decision: (1) non neoplastic lesions deserve no treatment, this applies to small hyperplastic polyps; (2) neoplastic lesions with prediction of curability by local treatment deserve endoscopic resection and “en bloc” resection is preferred. Lesions less than 25 mm are resected by endoscopic mucosal resection, lesions over 25 mm should be resected with submucosal dissection, by the endoscopic submucosal dissection technique. The curability of malignant neoplastic lesions developed in the squamous epithelium of the esophagus by endoscopic resection requires more severe criteria. Cure is ensured when the depth of tumor is intraepithelial (m1), micro-invasive (m2) and intramucosal (m3). When there is a superficial invasion of the submucosa (sm1) the risk of lymph node invasion is still acceptable for a local treatment. When invasion is deeper (sm2 and sm3), the risk reaches 40% and endoscopic resection should not be performed. In synthesis, histopathology of the specimen with serial sections is always required to confirm the legitimacy of the endoscopic treatment; and (3) neoplastic lesions without prediction of curability by local treatment deserve a surgical treatment. Decision on surgery is based on the depth of invasion in the submucosa and risk of lymph node invasion.
Endoscopy in screening and surveillance
Endoscopy has a major role in the detection and characterization of neoplastic lesions along the digestive tract in all screening strategies, as a primary procedure, applied to all screenies in opportunistic screening or after the selection of persons positive to a filter test in organized, population based, screening. Surveillance refers to endoscopy scheduled at regular intervals (2 to 5 years), independently of symptoms. It can be proposed to persons who are at increased risk of cancer because of specific environmental factors connected to life style, or to persons having a disease considered as a precursor of cancer. The endoscopic surveillance of the esophageal lumen is proposed to persons at risk of squamous cell cancer because of a high consumption of alcohol and tobacco. Surveillance is also proposed to persons in which a columnar lined esophagus has been detected when they are Caucasians, of male sex and are intensive smokers. The endoscopic surveillance of the gastric mucosa is proposed to patients with pernicious anemia and to those with chronic atrophic gastritis in the post operative stomach. The endoscopic surveillance of the colonic lumen is proposed to persons which were treated for adenomatous polyps or cancer[50,51] because new lesions may occur. It is also proposed to persons with IBD of continuous evolution. First degree relatives of persons who had a colorectal cancer are proposed a screening colonoscopy, followed by surveillance procedures if adenomas were detected.
Cost/effectiveness of endoscopy in screening
Cost/effectiveness (C/E) studies must be conducted in population based screening interventions conducted under the control of Health Authorities. The C/E ratio is estimated on the following criteria-cost per year of life gained in the population submitted to screening. The benchmark of the cost of a valid intervention is estimated at 40 000 $ per year of life gained in the concerned population. The evaluation takes in account the prevalence of the tumor in the population aged 50-year and over, and is based on modelization. The assumptions concern the efficacy of detection, dwelling time of premalignant lesions to malignancy, cost of procedures and treatment. With respect to digestive endoscopy the cost of procedures and treatment is limited to those performed in persons found positive to the selection test.
For esophageal cancer, there is no reliable analysis of the C/E ratio in screening for esophageal squamous cell cancer even in countries with a high risk, and organized screening is not recommended for esophageal adenocarcinoma.
For stomach cancer, in Japan organized population based screening is operational. The traditional filter test is the 7 films radiophotofluorography. Recent studies confirm the comparable validity of the Pepsinogen I/II ratio as a filter test. In Western countries there is no population based screening, but many studies taking in account the frequency of the clinical symptoms of dyspepsia, have compared the costs of 2 distinct strategies of opportunistic screening in patients with dyspepsia: the prompt endoscopy strategy uses no filter test and upper GI endoscopy is the primary exploration. In contrast the so-called empirical strategy uses eradication of H. pylori as a filter test and upper GI endoscopy is proposed only if symptoms persist after the eradication. A randomized trial of the 2 strategies ended in higher cost of the empirical strategy.
For colorectal cancer the C/E ratio of organized and population based screening with FOBT as a filter test is well under the bench mark of a valid intervention, in the range 20 000 to 26 000 $ per life-year gained[40]. The following assumptions are adopted: (1) prevalence of premalignant polyps estimated at 30% in persons aged 50 years and over; (2) dwelling time of premalignant lesions to cancer estimated at 5 to 10 years-sensitivity of FOBT around 60% to cancer and 10% to polyps; (3) sensitivity of colonoscopy for neoplastic lesions 90%. Protocols wit the FOBT filter test before decision of colonoscopy have been compared to protocols based on primary flexible sigmoidoscopy or on primary colonoscopy. Depending on the strategy 3000 to 7000 life-years are gained for 100 000 persons screened at a cost of 250 to 1200 $ for persons submitted to endoscopy.
Primary colonoscopy for direct and opportunistic screening is progressively accepted in Western countries[52]. The C/E ratio depends on the compliance of population; on morbidity associated to test, on age range of the persons screened and on the intervals in surveillance endoscopy. The option-once in a life colonoscopy-is the most effective model and would cost around 750 $ per person screened. The option-colonoscopy every 10 years from age 50-years-is valid and cost/effective if the dwelling time from polyp to cancer is around 10 years. In such protocols the quality control of colonoscopy deserves extreme attention: Interval cancers occurring a few years after a false negative endoscopy have a disastrous impact on cost/effectiveness of the strategy[53-55].
Flexible sigmoidoscopy with a depth of insertion between 48-55 cm, can be used as a filter test in homogenous socio-economic groups rather than in population based mass screening[56]. The procedure is performed as well by GI specialized nurses; the potential of cancer detection is 40% to 60% of that of colonoscopy which is recommended in a second step, if a distal neoplastic lesion. Is found the Kaiser medical program conducted by Selby et al[57] a 50 % reduction in mortality from cancer in the distal colon and rectum was obtained. However in the absence of sentinel lesions in the distal colon, 20% to 30% of advanced neoplasia would be missed by sigmoidoscopy screening because only 30% of subjects with proximal neoplasia have an index distal colorectal lesion.
