The endoprevention of esophageal cancer

This introductory article summarizes decades of research and clinical care from many dedicated gastrointestinal endoscopists. Pioneers such as Bergein Overholt, Charles Lightdale, and Kenneth Wang provided the scientific basis and clinical context for the development and use of esophageal endosonography and endoscopic resection and ablation. Not satisfied with diagnosing and staging esophageal cancer, they developed the endoscopic technology to detect and treat esophageal disease, to prevent the development of invasive cancer (endoprevention). These developments paralleled my own career; one that began with Richard Kozarek at the Virginia Mason Clinic (Seattle), who had trained with Robert Sanowski, at the Maricopa County Veterans Administration Medical Center in Phoenix, Arizona. Dr Kozarek instilled in me an appreciation for gastrointestinal endoscopy, the importance of clinical research, and the need to support our professional societies, especially the American Society for Gastrointestinal Endoscopy (ASGE). My fellowship training with Kenneth Wang introduced me to cutting edge basic science and, clinical ablation research using photodynamic therapy, to understand disease mechanisms and investigate new treatments.

In 2003, the ASGE had established 3 new groups of members with specific interests, called SIGs (special interest groups). Soon thereafter, several endoscopic devices and technologies were developed for the treatment of Barrett disease. To create a forum to bring innovators and clinicians together, I asked more than 100 ASGE members to support a new SIG called Endoluminal Therapy for Esophageal Disease (ETED) in November, 2004. This broad and somewhat unwieldy term, suggested by Michael Kimmey and Michael Wallace at the 14th International Symposium on Endoscopic Ultrasonography (EUS 2004; Tokyo, Japan), was meant to include all gastrointestinal and surgical endoscopists, who used every type of resection or ablation technology and device in the treatment of esophageal diseases, malignant and premalignant.

The ETED SIG met for the first time on May 16, 2005, at Digestive Disease Week (DDW) in Chicago, Illinois, bringing together leading clinical researchers and industry innovators. This group, subsequently under the leadership of Bruce Greenwald and Drew Schembre, has continued to provide ASGE members with information and access to clinical experts, including sponsoring hands-on demonstration programs at DDW. Further, the ETED SIG has provided a platform to stimulate interaction and collaboration among ASGE members that are interested in new technologies for the diagnosis and treatment of Barrett disease, including the use of endoscopic therapy to prevent the development of esophageal cancer (endoprevention).

Barrett esophagus

Barrett esophagus (BE) develops as a result of chronic, pathologic reflux of gastro-duodenal contents into the esophagus. In North America, the diagnosis is considered based on the endoscopic finding of salmon-colored epithelium in the distal esophagus, followed by histologic confirmation of specialized intestinal columnar epithelium (specialized intestinal metaplasia [SIM]). Barrett esophagus is estimated to be present in 1% to 2% of the US adult population, with recent reports suggesting an increasing prevalence. Rex and colleagues reported a 6.8% prevalence in a general population of patients undergoing colonoscopy. In this study, as might be expected, the prevalence of SIM, was even higher (8.6%) among patients who reported gastroesophageal reflux disease (GERD) symptoms. In a Department of Veterans Affairs (VA) clinic study, Gerson and colleagues reported a 25% prevalence of SIM in a predominantly white, male, non-GERD population (>50 years of age), undergoing sigmoidoscopy. The cause of this observed increase in the number of Barrett cases is unclear, but it may be related to the increase in the prevalence of patients reporting GERD, and increased awareness among gastroenterologists. The clinical importance of Barrett disease depends on the risks of progression to dysplasia and neoplasia in comparison with similar premalignant conditions.

Barrett esophagus

Barrett esophagus (BE) develops as a result of chronic, pathologic reflux of gastro-duodenal contents into the esophagus. In North America, the diagnosis is considered based on the endoscopic finding of salmon-colored epithelium in the distal esophagus, followed by histologic confirmation of specialized intestinal columnar epithelium (specialized intestinal metaplasia [SIM]). Barrett esophagus is estimated to be present in 1% to 2% of the US adult population, with recent reports suggesting an increasing prevalence. Rex and colleagues reported a 6.8% prevalence in a general population of patients undergoing colonoscopy. In this study, as might be expected, the prevalence of SIM, was even higher (8.6%) among patients who reported gastroesophageal reflux disease (GERD) symptoms. In a Department of Veterans Affairs (VA) clinic study, Gerson and colleagues reported a 25% prevalence of SIM in a predominantly white, male, non-GERD population (>50 years of age), undergoing sigmoidoscopy. The cause of this observed increase in the number of Barrett cases is unclear, but it may be related to the increase in the prevalence of patients reporting GERD, and increased awareness among gastroenterologists. The clinical importance of Barrett disease depends on the risks of progression to dysplasia and neoplasia in comparison with similar premalignant conditions.

Risk of progression to dysplasia and esophageal adenocarcinoma

The risk of a patient with nondysplastic Barrett esophagus progressing to esophageal adenocarcinoma has been reported to be 0.4% to 1% per patient per year, a risk 30 to 125 times higher than the general population. In 2005, according to the American Cancer Society, there were 14,520 new cases of esophageal cancer in the United States; most were cases of adenocarcinoma, and 13,570 deaths were associated with this disease. This represents a 300% to 500% increase in US esophageal cancer incidence over the last 30 years.

Sharma, and colleagues reported that on the initial diagnosis of Barrett esophagus in 1376 patients ( Table 1 ), a large number of patients had already developed low-grade dysplasia (LGD; 7.3%), high-grade dysplasia (HGD; 3.0%), or adenocarcinoma (6.0%). Subsequently, 618 of the nondysplastic SIM cases from this series underwent endoscopic surveillance for an average of 4 additional years. Over this time interval, a significant number of these previously nondysplastic patients progressed to LGD (16.1%), HGD (3.6%), or adenocarcinoma (2.0%) ( Table 2 ). In this study, the annual risk of a patient with nondysplastic BE progressing to diagnosis of HGD or adenocarcinoma, for which the standard of care is surgical esophagectomy, was 1.4% per patient (1 in 71 patients).

Colon adenoma/colon and rectal carcinoma versus BE/esophageal adenocarcinoma: a comparison of premalignant conditions

According to the National Cancer Institute Surveillance, Epidemiology, and End Results data for 2005, the lifetime risk of developing colon and rectal carcinoma (CRC) is 5.7%, whereas that of esophageal cancer is 0.5%. In 2005, there were 145,290 new cases of CRC in the United States, whereas there were 14,520 new cases of esophageal cancer. Although CRC has a much higher incidence than esophageal cancer (an approximately 10-fold difference), the age-adjusted death rate for CRC is 20.5 per 100,000 (population) versus 4.4 per 100,000 for esophageal cancer (less than a 5-fold difference). Furthermore, the CRC death rate for men of all races is 24.8 per 100,000 versus 7.7 per 100,000 for esophageal cancer (an approximately 3-fold difference). This difference in the incidence and death rate for CRC and esophageal cancer is due to the difference in 5-year survival for the 2 disease states: 64.1% for CRC versus 14.9% for esophageal cancer, with the latter being 1 of the lowest 5-year survival rates of any cancer diagnoses.

The risk of progression to CRC for a patient with polyps of the colon and to esophageal adenocarcinoma for one with nondysplastic BE are identical (0.5% per patient per year). The patient with BE may be found to progress to HGD (0.9% per patient per year), resulting in an aggregate risk (1.4% per patient per year) of developing a disease state for which the standard of care is esophagectomy.

The surgical intervention for most CRC stages is segmental or hemicolectomy and for most esophageal cancer stages and HGD, is esophagectomy. The morbidity and mortality associated with removal of a segment of colon is low. However, esophagectomy carries a much higher risk for longer-term complications and death. Although mortality rates reported for esophagectomy in referral centers are typically 4% to 6%, several recent reports show that patients undergoing this operation in small, low-volume community hospitals incur a much higher mortality risk. A recent US study examined 8657 cases that were treated between 1988 and 2000, and it evaluated the mortality associated with esophagectomy. A random sample of 20% of these cases showed that the overall in-hospital mortality rate was 11.3%, but it was lower in high-volume surgical centers, decreasing to 7.5%. Additionally, several large studies have found that 30% to 50% of patients experienced at least 1 serious postoperative complication, such as pneumonia, myocardial infarction, heart failure, or wound infection, and that the average length of hospital stay was at least 2 weeks. Late surgical complications, such as anastomotic strictures, are common, occurring in 10% to 56% of patients and require follow-up endoscopic dilation. Respiratory function may remain depressed for 6 months after esophagectomy. Removal of the gastroesophageal junction and relocation of the stomach remnant into the chest may be associated with severe, refractory, gastroesophageal reflux and long-term pulmonary complications; these procedures may even put a patient at risk of BE recurrence and possible development of recurrent Barrett dysplasia or carcinoma.

Prevention of CRC and HGD/adenocarcinoma

After the advent of barium radiography of the colon, it was hypothesized that the development of CRC was preceded by malignant transformation of adenomatous polyps, and the subsequent development of the metaplasia-dysplasia-carcinoma sequence. Other investigators, however, have cited the absence of residual adenomatous tissue in excised malignancies, and they suggest that colon cancers only rarely arise from adenomatous polyps. Surveillance studies with matched cohorts, such as the National Polyp Study, found that patients undergoing endoscopic removal of adenomatous polyps were significantly less likely to develop colon carcinoma, a risk reduction of approximated 80% to 90%. Subsequently, the use of screening colonoscopy and removal of colon polyps has been recognized as the most effective method for diagnosing and ultimately reducing the risk of developing CRC. Therefore, the paradigm related to colon polyps and CRC prevention is: (1) screen candidate patients for colon polyps (ie, detect the precursor lesion for CRC); (2) remove the precursor lesion that has a 0.5% risk per patient per year of progression to CRC (ie, prevent progression to CRC).

Historically, the paradigm for BE and esophageal adenocarcinoma prevention has been strikingly different: (1) do not screen patients, but detect BE (precursor lesion for adenocarcinoma) incidentally, on endoscopy indicated for GERD symptoms; (2) once detected, do not remove the precursor lesion, even though the lesion incurs a 1.4% per patient per year risk of progressing to HGD or adenocarcinoma; (3) survey the nondysplastic BE patient every 3 years to detect progression to HGD or adenocarcinoma; (4) remove the esophagus when HGD or adenocarcinoma is detected.

Rationale for endoscopic treatment of Barrett disease

Given the elevated risk, inherent in nondysplastic BE, of progression to HGD or adenocarcinoma, the conservative “watch and wait” approach being the standard of care for these patients needs explaining. There may be several reasons, such as (1) the limited ability to detect dysplasia in BE patients who undergo standard-resolution video or fiberoptic endoscopy, (2) the subsequent dependence on random biopsy protocols, and (3) the highly variable histopathology interpretations. Another explanation may be the absence of a technique for the safe, effective (complete), and reproducible removal of all SIM tissue in a patient.

There are multiple challenges inherent in achieving safe, effective, and reproducible removal of BE, and each of these factors must be considered when evaluating a technique for managing this disease: (1) access (the targeted portion of the esophagus is approximately 30 to 40 cm from the incisors); (2) irregular nature of the esophageal lumen (an uneven epithelial target); (3) mucous and gastric contents affecting delivery of the ablative energy; (4) esophageal motility; and (5) very tight margin between the ablation being “deep enough” (to the muscularis mucosae) and “too deep” (beyond the submucosa).

The ideal means to achieve safe, effective and reproducible ablation of BE requires a skilled endoscopist performing the interventional techniques to remove the Barrett epithelium entirely, without residual disease, especially subsquamous disease (buried glands). Endoscopic ablation methods should have a low rate of complication, such as stricture formation, bleeding, perforation, and must be well-tolerated by the patient. The endoluminal techniques that have been developed for removing BE include circumferential balloon-based radio-frequency ablation, aminolevulinic acid and porfimer sodium photodynamic therapy, endoscopic mucosal resection and submucosal dissection, laser ablation, argon plasma coagulation, multipolar electrocoagulation, and liquid nitrogen and carbon dioxide cryotherapy.

However, these devices probably differ in their method of ablation, including in treatment depth. What depth of tissue ablation is required to effectively eliminate BE? In a study of resection specimens that were measured after tissue fixation and mounting, Ackroyd and colleagues reported that the thickness of nondysplastic BE (500±4 μm, range 390–590 μm) is similar to that of normal squamous epithelium (490±3 μm, range 420–580 μm). These ex vivo data are useful information; if an ablation technique can be shown to repeatedly and uniformly penetrate at a minimum to the muscularis mucosae (approximately 700 μm) and at a maximum to the top of the submucosa (approximately 1000–1500 μm), then the Barrett epithelium can be reliably and safely removed. However, a recent study examined the use of endosonography to measure the esophageal wall thickness in BE patients with dysplasia and compared the results with a group of control patients. In this study of 76 patients (most with LGD or HGD), the mean esophageal thickness was significantly greater for BE patients, compared with controls (3.36mm versus 2.4 mm, P <.05). There was also a trend toward increasing esophageal wall thickness with a greater degree of dysplasia, in BE patients, although these differences were not statistically significant. This raises important questions regarding differences among BE patients, the methods used to measure mucosal disease, and how these differences may alter treatment outcomes. These variability factors include thickness of Barrett esophageal wall layers, blood flow, and mucosal oxygenation; these factors are important for developing dosimetry systems to monitor the effect of esophageal ablation therapy of all types, for delivering optimal results and for avoiding complications, such as incomplete ablation, stricture, or perforation. Some of this ex vivo disease-depth information was subsequently used in the development of radiofrequency energy ablation (RFA). The use of RFA in a prospective multi-center randomized controlled trial of patients with Barrett LGD and HGD was recently reported in the New England Journal of Medicine . At a follow-up of 12 months, among patients with LGD, complete eradication of dysplasia occurred in 90.5% in the ablation group compared with 22.7% in the control group ( P <.001). Among patients with HGD, complete eradication occurred in 81.0% of those in the ablation group compared with 19.0% of those in the control group ( P <.001). RFA-treated patients had less disease progression (36% versus 16.3%, P = .03) and fewer cancers (1.2% versus 9.3%, P = .045). Complications in RFA-treated patients were limited to bleeding in 1 patient and stricture in 5 patients (6.0%). These early results represent important progress toward the goal of an ideal form of endoscopic therapy for Barrett disease. Yet, a small number of patients did experience dysplasia progression, including cancer despite the use of RFA; questions remain about the durability of these results and whether they can be achieved outside of elite endoscopy centers. How to select patients for treatment (risk stratification) and whether nondysplastic BE patients should be treated remain to be investigated in long-term controlled studies. Despite these limitations, studies such as these move us steadily closer to ideal endoscopic treatment modalities for the safe, effective, reproducible, stable, and durable treatment of Barrett disease, to prevent the development of dysplasia or carcinoma.