The incidence of esophageal cancer, especially esophageal adenocarcinoma, is increasing and its high mortality rate is a notable fact. Improving survival rates of this disease depend on earlier detection through screening and surveillance; however, standard diagnostic modalities, such as endoscopy with biopsy, have several limitations as screening tools, including low negative predictive value and relatively high cost. Recently developed biomarkers such as FISH and improved imaging techniques, may help overcome current problems and provide improved screening and surveillance for esophageal cancer.
Strategies to screen, survey, and prevent esophageal cancer must take into account the changing epidemiology of the disease. Esophageal cancer is relatively uncommon; however, it is the second leading cancer in terms of its ever-increasing incidence. When it occurs, esophageal cancer has one of the highest mortality rates in the United States. In 2008, an estimated 16,470 new patients were diagnosed with esophageal cancer and 14,280 individuals died of this disease. Recent reports show esophageal adenocarcinoma incidence rates rose from 1975 through 2004 among white men and women in all stages and age groups. The incidence of adenocarcinoma among white men increased 463%, from 1.01 per 100,000 person-years in 1975–1979 to 5.69 per 100,000 person-years in 2000–2004. A similar rapid increase was also apparent among white women, with an increased incidence of 335% from 0.17 per 100,000 person-years to 0.74 per 100,000 person-years. This increase was not found in earlier reports because of the rarity of EAC among women. Squamous cell carcinoma of the esophagus is now less common than adenocarcinoma in the United States and is thought to be due to decreased consumption of alcohol and tobacco. It is clear that to affect the increasing rate of esophageal cancer and to improve its prognosis, it is necessary to develop new strategies of screening, surveillance, and prevention.
Squamous cell carcinoma
In contrast to the situation in the United States, squamous cell carcinoma is the most common type of esophageal cancer worldwide, where it remains a significant health care burden, particularly in Asian countries, which have a high incidence of squamous cell carcinoma. In the United States, squamous cell carcinoma is the most common type of esophageal cancer in African American and Hispanic males. Alcohol consumption, smoking, consumption of food and water rich in nitrates and nitrosamines, and specific genetic factors have all been reported as risk factors for cancer development. In addition, Plummer-Vinson syndrome, achalasia, and chronic strictures resulting from acid or lye ingestion have all been implicated as predisposing factors. Although some case-control studies to evaluate familial aggregation were conducted in China, a region where the incidence of squamous cell carcinoma is high, it remains unclear whether squamous cell carcinoma is heritable. Tylosis, a genetic defect in the 17q25 region characterized by hyperkeratosis of the palms and soles is associated with high risk of squamous cell carcinoma by the age of 65 years. Based on experimental evidence, squamous cell carcinoma is believed to develop via progression through a premalignant dysplastic phase before development of carcinoma. Although there are many studies regarding molecular pathogenesis (p53, Rb, and p16 are common alterations) of squamous dysplasia and squamous cell carcinoma, it remains unclear if unique signaling pathways exist in these conditions. Known biologic processes such as methylation have been shown to play a role in the pathogenesis of both squamous and adenocarcinoma.
Barrett’s esophagus and esophageal adenocarcinoma
The known risk factors for esophageal adenocarcinoma (EAC) are chronic gastroesophageal reflux disease (GERD) and Barrett’s esophagus (BE). Persons with recurring symptoms of reflux have an eightfold increase in the risk of EAC. Among patients with BE, the annual rate of neoplastic transformation is reported as approximately 0.5%, although this estimate is based on compiled series with relatively short follow-up times. It seems unlikely that someone diagnosed with BE at age 50 would have a 10% cumulative risk of cancer by age 70 based on current information. Barrett’s is a condition in which normal squamous epithelium of the esophagus is replaced by metaplastic columnar mucosa. This phenomenon is a complication of esophageal mucosal damage caused by GERD. It is thought that histologic evidence of intestinal metaplasia is required before BE can progress sequentially from no dysplasia to low-grade dysplasia, then to high-grade dysplasia, and eventually to adenocarcinoma. It should be noted that this theory is controversial due to several reports of progression to adenocarcinoma without precursor lesions of intestinalized epithelium with goblet cells. However, other markers of intestinal metaplasia, such as cytokeratin profiles or expression of intestinal transcription factors such as CDX2, are usually found in non-goblet cells containing intestinalized mucosa.
Progression of BE to cancer is felt to be a relatively slow process, although there is an increase in cancer rates due to the difficulty in detecting cancers early with random biopsies. Most cancers within BE are found early after diagnosis and the rates of progression after the first year of diagnosis are significantly decreased. Although BE is recognized as a premalignant condition, it does not produce any symptoms itself. Some have thought that the incidence of clinically diagnosed BE has increased as a result of the increased use of endoscopy, others have found that this is not the case. It is generally accepted that Barrett’s esophagus is increasing.
Early diagnosis of BE is critical because metastasis is common with the evolution to carcinoma. The lymphatic supply of the esophagus reaches the lamina propria, lymph node metastasis can occur even in early-stage disease. The prognosis of adenocarcinoma is highly dependent on the stage of the disease. Early neoplastic lesions generally have an excellent prognosis, but the prognosis of most advanced lesions is dismal. The goal of screening is to detect patients with BE, the precursor lesion. The goal of surveillance is to diagnose early stages of EAC in patients with known BE and to intervene so as to prevent progression to fatal cancer.
Barrett’s esophagus and esophageal adenocarcinoma
The known risk factors for esophageal adenocarcinoma (EAC) are chronic gastroesophageal reflux disease (GERD) and Barrett’s esophagus (BE). Persons with recurring symptoms of reflux have an eightfold increase in the risk of EAC. Among patients with BE, the annual rate of neoplastic transformation is reported as approximately 0.5%, although this estimate is based on compiled series with relatively short follow-up times. It seems unlikely that someone diagnosed with BE at age 50 would have a 10% cumulative risk of cancer by age 70 based on current information. Barrett’s is a condition in which normal squamous epithelium of the esophagus is replaced by metaplastic columnar mucosa. This phenomenon is a complication of esophageal mucosal damage caused by GERD. It is thought that histologic evidence of intestinal metaplasia is required before BE can progress sequentially from no dysplasia to low-grade dysplasia, then to high-grade dysplasia, and eventually to adenocarcinoma. It should be noted that this theory is controversial due to several reports of progression to adenocarcinoma without precursor lesions of intestinalized epithelium with goblet cells. However, other markers of intestinal metaplasia, such as cytokeratin profiles or expression of intestinal transcription factors such as CDX2, are usually found in non-goblet cells containing intestinalized mucosa.
Progression of BE to cancer is felt to be a relatively slow process, although there is an increase in cancer rates due to the difficulty in detecting cancers early with random biopsies. Most cancers within BE are found early after diagnosis and the rates of progression after the first year of diagnosis are significantly decreased. Although BE is recognized as a premalignant condition, it does not produce any symptoms itself. Some have thought that the incidence of clinically diagnosed BE has increased as a result of the increased use of endoscopy, others have found that this is not the case. It is generally accepted that Barrett’s esophagus is increasing.
Early diagnosis of BE is critical because metastasis is common with the evolution to carcinoma. The lymphatic supply of the esophagus reaches the lamina propria, lymph node metastasis can occur even in early-stage disease. The prognosis of adenocarcinoma is highly dependent on the stage of the disease. Early neoplastic lesions generally have an excellent prognosis, but the prognosis of most advanced lesions is dismal. The goal of screening is to detect patients with BE, the precursor lesion. The goal of surveillance is to diagnose early stages of EAC in patients with known BE and to intervene so as to prevent progression to fatal cancer.
Screening
Screening for squamous cell carcinoma in the general population in the United States is not recommended due to the low incidence of this type of esophageal cancer and should only be considered for specific subgroups with known risk factors. Patients with oral and oropharyngeal cancer are well known to have high risk of concomitant esophageal squamous cell carcinoma and should be screened for this disease. Screening should also be considered for patients with tylosis. A screening program is promoted by the concept that detection of early neoplasia can lead to an improved outcome. Recent guidelines state that screening for BE still remains controversial because of the lack of documented impact on mortality from adenocarcinoma of the esophagus. Who, when, how and how often to screen for the presence of BE and adenocarcinoma are frequently raised controversies. Because patients are often asymptomatic, it is difficult to identify the entire at-risk group. It is left to the discretion of the physician to determine if screening is necessary given the health of the patient and the likelihood of early disease. Patients with the higher likelihood of BE are generally older white males with chronic reflux symptoms. It would therefore seem rational to screen those high-risk patients; however, such a strategy works only if those high-risk patients comprise a majority of patients who develop cancer. Although several studies showed reflux symptoms are strongly associated with cancer risk, one report demonstrated that about 40% of subjects with adenocarcinoma reported no antecedent symptoms of reflux. These data indicate a substantial number of patients develop adenocarcinoma without identifiable symptoms of GERD. Presumably such a large number of people would be missed by a screening strategy that emphasized symptoms. However, to screen those patients who lack reflux symptoms but have BE would entail screening the entire population over a specific age. Multiple other risk factors for BE have been identified and several studies attempted to predict the presence of BE. Age greater than 40, long duration of GERD symptoms, male gender and hiatus hernia have been suggested as significant risk factors, and at this time the only symptom consistently identified in different studies is heartburn.
Standard endoscopy with biopsy is the most reliable means of establishing a diagnosis of BE. However, it has several limitations as a screening tool, including its low negative predictive value, risk, unpleasant experience, and cost. High negative predictive value is necessary for endoscopy to be a valid screening tool. The negative predictive value of an endoscopy seems to be low and does not meet the level needed for a valid screening method. In one large population-based study which analyzed a total of 589 patients with esophageal or gastric cardia adenocarcinoma, BE was diagnosed in 135 of 589 adenocarcinoma patients and only 23 out of 64 patients who had endoscopy performed more than 6 months before their diagnosis of cancer were identified to have BE. This indicates far fewer patients with adenocarcinoma have an endoscopically recognizable Barrett’s segment than we traditionally expect. A systematic review of the prevalence of BE in adenocarcinoma undergoing resection showed only 4.7% of patients undergoing resection reported a prior diagnosis of BE.
Complications of endoscopy related to diagnostic evaluations are rare but can occur especially for elderly patients. It is estimated cardiopulmonary complications related to sedation and analgesia are the most common type of problems seen with diagnostic endoscopy. Though these complications are generally minor changes in vital signs in most of the cases, myocardial infarction, respiratory depression, and shock can occur, albeit rarely.
Unsedated small caliber upper endoscopy is feasible, tolerable, and accurate compared with conventional sedated endoscopy. Unsedated endoscopy has the potential advantage of decreasing sedation-related complications and can be performed as an outpatient procedure in any setting, which could reduce health care costs. However, in this study only five cases of BE were evaluated and limited information is available on unsedated endoscopy in the diagnosis of BE as a screening test. It is yet unknown whether an unsedated procedure will meet with patient acceptance in the United States, given the cultural preference for sedation. Another study in which a smaller caliber scope was used for unsedated patients showed that the diagnostic accuracy of unsedated small-caliber endoscopy was below the acceptable range for esophageal disease, especially for detecting BE. It seems the thinner the caliber is, the more acceptant the modality is for patients; however, as the caliber of the endoscope decreases, the diagnostic accuracy also decreases. Transnasal endoscopy is also attractive as a screening modality because it can be applied as an unsedated technique. This diagnostic modality was assessed its feasibility for surveillance of patients with BE and its relatively high accuracy proved its potential as a screening modality. However, acceptance among referring physicians and the public is also a limiting factor. Despite transnasal endoscopy actually being better tolerated, the public perception is that of greater discomfort. Esophageal capsule endoscopy is a new technique that has the potential to provide a noninvasive diagnosis of suspected BE. The Pillcam ESO video capsule (Given, Yoqneam, Israel) is a dual-camera wireless capsule endoscope developed specifically for esophageal visualization. A pilot feasibility study with 17 subjects showed its safety and capability of imaging the esophagus clearly. A recently published study assessed the accuracy of capsule endoscopy for the diagnosis of BE. In a prospective blinded study of 90 subjects with BE who underwent capsule endoscopy followed by upper endoscopy, capsule endoscopy had only moderate sensitivity and specificity (67% and 84%, respectively) for identification of BE, and positive and negative predictive value were 22% and 98%, respectively. Although this screening method is more acceptable for patients due to convenience and safety, improvement in its diagnostic accuracy is needed to replace upper endoscopy as a standard screening modality.
Barrett’s esophagus is frequently suspected at endoscopy, especially for patients with GERD symptoms; however, correctly diagnosing BE can be problematic. One study was conducted to evaluate the ability of the endoscopist to predict the presence of BE at index endoscopy. In the study, endoscopists’ interpretations were analyzed with histologic specimens from areas thought to contain BE and positive predictive value was only 34%. Endoscopic detection is difficult because it is based on subtle mucosal changes, and for this reason accurate detection is more difficult in a short segment lesion. New imaging modalities have been developed as adjuncts to white light endoscopy for better visualization of the mucosal changes. Narrow-band imaging is easy to use and allows clear visualization of the mucosal pit patterns and capillary patterns. It showed that the combined recognition of mucosal patterns and capillary patterns improved the diagnostic value for detecting intestinal metaplasia. The same phenomena were reported in other studies and most groups created their own classification of mucosal or vascular patterns correlating to histologic findings. Standardization of the patterns and randomized studies comparing narrow-band imaging with standard endoscopy with biopsy would help to determine an actual clinical impact. Autofluorescence imaging is another promising imaging modality that can differentiate tissue types based on differences in autofluorescence emission. A new endoscopic system has been developed that incorporates high-resolution endoscopy, autofluorescence imaging, and narrow-band imaging in one system. Another recently published international multicenter feasibility study showed that the addition of high-resolution autofluorescence imaging increased the detection of both the number of patients and the number of lesions in patients with BE and the false-positive rate was reduced after detailed inspection with narrow-band imaging. To evaluate its true value as an imaging modality, randomized comparison studies are needed. Although sensitivity and specificity can be assessed from cohort studies, the actual clinical efficacy of these models depends on their acceptability in the community. Currently, there are no validated alternative techniques to screen for BE and consensus is that standard endoscopy with biopsy is the most reliable means of establishing a diagnosis of BE.
In an effort to determine when and how often to screen for BE, several cost-effectiveness models have been produced. One study assessed the cost-effectiveness of performing an initial endoscopic screening of patients with GERD to rule out high-grade dysplasia of BE. Under favorable conditions, this one-time screening endoscopy of all patients with reflux symptoms to prevent death from EAC could be cost-effective compared with no screening, resulting in a cost of $24,700 per quality-adjusted life-year saved. However, under conditions in which there is a relatively high prevalence of BE, high-grade dysplasia, or adenocarcinoma, low false-positive rates of endoscopy with biopsy and low reduction in quality of life with esophagectomy may not reflect the real situation and any variation of these factors make this strategy cost-ineffective. In another study, a single screening examination of subjects at 50 years of age with GERD symptoms resulted in an associated incremental cost-effectiveness ratio of $10,440 compared with no screening. However, because the analysis was sensitive to the prevalence of BE and adenocarcinoma of the esophagus at the time of screening, the incidence of cancer among patients with BE and a change in parameters could make such a screening cost-ineffective. Because there are no definite data regarding how to identify subgroups of patients with BE who are more likely to develop cancer, we continue to rely on unproven mathematical models to evaluate cost-effectiveness of screening.
The purpose of screening is to detect the disease during the preclinical phase and alter its natural course before symptoms appear. Some studies indicate that subjects whose cancers are detected in screening have an improved prognosis compared with those presenting symptomatically. Because survival in EAC is strongly correlated with stage at diagnosis, finding a lesion in the early stage may confer improved prognosis. There are no randomized trials of screening programs to substantiate that endoscopic screening confers a superior life expectancy. One study was conducted to determine the prevalence of BE and possible associated risk factors in a Swedish population in which a random sample of the adult population was surveyed. The study showed that the prevalence of BE was 1.6% in a general population and 43.7% of subjects with BE had no reflux symptoms. This data suggested that a screening program based on reflux symptoms is inadequate to detect BE. For screening to be performed effectively, we need to identify asymptomatic patients with BE, so that we can detect the majority of patients with BE. This major challenge in developing an effective screening strategy for BE led to consensus that screening for BE in the general population cannot be recommended at the current time.
Surveillance
Endoscopic surveillance for patients with BE is recommended due to several crucial assumptions: the establishment of the association of BE with adenocarcinoma, the slow progression of BE through dysplasia to adenocarcinoma, the rapidly rising rate of incidence of adenocarcinoma, the dismal prognosis of adenocarcinoma when detected symptomatically, and the improved prognosis with treatment of dysplasia/neoplasia detected by surveillance. It is thought that BE sequentially progresses from no dysplasia through to low-grade dysplasia, to high-grade dysplasia and to EAC. The time-course of the progression is thought to require years and therefore a surveillance program seems to be a suitable option to detect the disease in a treatable stage. It now appears that not all BE cases progress to adenocarcinoma. Initial reports estimated the risk of developing adenocarcinoma from BE was 1% or more per year. However, several recent studies suggest that this risk was overestimated and the actual rate is approximately half that amount. It is unclear whether BE decreases survival of patients compared with the patients without BE. Previous studies did not show adenocarcinoma arising from BE as having a significant impact on prognosis and one long-term prospective observational study showed no significant overall survival between subjects with BE compared with appropriately matched individuals in the general population. A major limitation of these studies was the predominant composition of elderly people who may have died of other comorbidities. Long-term studies of younger patients with BE is needed to evaluate its impact on survival and that kind of study would likely show a reduction in life expectancy. At the current time, histologic evidence of dysplasia is used as the primary means to discriminate high-risk patients.
There is still controversy on the effectiveness of surveillance endoscopy for patient-outcome. Previous studies showed that esophageal cancers detected in subjects with BE on a surveillance program were associated with longer survival than those diagnosed during evaluation of cancer symptoms. However, the outcome of these studies may have been influenced by lead-time bias, length bias, and selection bias due to characteristics of BE and EAC. Lead-time bias means that longer survival from EAC in patients undergoing surveillance endoscopy may reflect an earlier detection of incurable EAC at a very early stage without any true improvement in the overall survival. Length bias means patients with slow-growing cancers tend to have longer survival that is independent of any benefit from a surveillance program. Elderly patients or patients with significant comorbidity are less likely to be enrolled in a surveillance program and patients enrolled in such a program might have healthier lifestyles or might pay more attention to their health, which may cause selection bias. Those previous small cohort studies also suggested that only a minority of subjects with BE actually died of EAC and that the majority of people with BE died of causes other than EAC. However, a retrospective population-based cohort study of endoscopic surveillance in 23 subjects with BE among 589 subjects with adenocarcinoma showed improved survival benefit among subjects with surveillance-detected cancer. Eleven of 15 subjects who were diagnosed with cancer while in a surveillance program were alive, compared with none of eight subjects who were not under surveillance. None of the deaths in the surveillance-detected patients were from cancer, compared with four deaths from cancer in the non-surveillance-detected patients. There are no prospective randomized studies that have evaluated the efficacy of surveillance in preventing EAC and its related mortality in patients with BE.
Nearly all surveillance in the United States is performed with endoscopy and biopsies. Although there are no definite criteria as to which patients with BE should begin a surveillance program, the current recommendations are based on age, likelihood of survival over the next 5 years, the patient’s understanding of the process and its limitations for detection of cancer, and the willingness of the patient to adhere to recommendations. Because initial endoscopy is performed on symptomatic patients, there is a possibility of erosive esophagitis and with it the potential to reduce the detection rate by interfering with the visual recognition of BE. Therefore, patients undergoing endoscopy for detection of BE should be treated with acid-suppressive therapy before endoscopy.
There is also no standardized technique for endoscopic mucosal biopsy. An assessment of the use of jumbo biopsy forceps in a surveillance program to detect unsuspected carcinoma at esophagectomy in BE patients with high-grade dysplasia showed no statistical differences in the rate of unsuspected cancers found at esophagectomy compared with standard biopsy forceps. All mucosal irregularities in BE such as erosions, nodules, and strictures are sampled due to their association with an increased rate of malignancy. Based on the fact that the distribution of the dysplasia and cancer is usually patchy in BE, four quadrant biopsies every 2 centimeters from the Barrett’s segment, which starts at the gastroesophageal junction and stops at the squamocolumnar junction, are commonly used. Further, in the presence of high-grade dysplasia, four quadrant biopsies every centimeter are recommended to reduce the likelihood of missing coexisting cancers. Surveillance intervals are recommended based on grades of dysplasia ( Table 1 ). One study was conducted to assess whether endoscopic mucosal resection can be used in the diagnosis of lesions within BE whose endoscopic appearance raise suspicion of carcinoma or high-grade dysplasia. The study of 25 subjects showed endoscopic mucosal resection of suspicious lesions diagnosed 13 subjects (52%) with superficial EAC and four subjects (16%) with high-grade dysplasia. It is currently accepted that patients with mucosal abnormalities that raise suspicion of advanced lesions should undergo endoscopic mucosal resection to increase detection of advanced lesions. In a retrospective study of 76 subjects with high-grade dysplasia who had no evidence of cancer on an initial evaluation, the 5-year cumulative incidence of EAC among them was 59%. Another retrospective study found EAC in 14% of subjects with focal high-grade dysplasia, defined as the detection of high-grade dysplasia in fewer than five mucosal crypts in a single biopsy specimen, and in 56% of subjects with diffuse high-grade dysplasia in a 3-year observational period. Subjects with focal high-grade dysplasia demonstrated longer survival than subjects with diffuse high-grade dysplasia. This study also showed that nodularity on endoscopy was associated with a 2.5-fold increased risk of EAC. A noteworthy fact for evaluating an appropriate surveillance program is that even if the initial two endoscopies lack evidence of dysplasia, there is no guarantee that the patient will not develop EAC. One multicenter cohort study showed more than half of the subjects who developed EAC had no dysplasia on their first endoscopies. This phenomenon might partially be explained by the limitation of the biopsy protocols; however, it still has an impact on patient follow-up.