Screening for Esophageal Squamous Cell Carcinoma and Its Precursor Lesions

Published on 13/02/2015 by admin

Filed under Gastroenterology and Hepatology

Last modified 22/04/2025

Print this page

This article have been viewed 5739 times

Chapter 30 Screening for Esophageal Squamous Cell Carcinoma and Its Precursor Lesions

Sanford M. Dawsey, David E. Fleischer

Introduction

Esophageal cancer is the sixth leading cause of cancer death worldwide.¹ In 2002, it was estimated that there were 462,000 new esophageal cancer cases and 386,000 deaths caused by esophageal cancer, only 25,000 fewer deaths than were caused by breast cancer.¹ About 80% of esophageal cancer cases occur in developing countries.¹ In the United States, esophageal cancer is the ninth leading cause of cancer death; an estimated 16,470 new cases and 14,530 deaths occurred as a result of esophageal cancer in 2009.²

One striking characteristic of esophageal cancer throughout the world is its great geographic variation in incidence, with 10-fold differences reported over distances of a few hundred kilometers.³ Worldwide, the highest risk populations are found in two geographic belts: one in central Asia from the Caspian Sea to north central China and one from eastern to southern Africa (Fig. 30.1).⁴ Age-adjusted incidence rates over 100 cases per 100,000 inhabitants per year have been reported from some areas in these regions.^3,^5,⁶ Populations with intermediate risk are found in southern South America (in southern Brazil, Uruguay, Paraguay, and northern Argentina) and in northwestern France.⁵ People throughout most of the world are considered to be at low risk, with incidence rates less than 10 per 100,000 inhabitants per year.⁵ In the Surveillance Epidemiology and End Results (SEER) cancer registries in the United States, the age-adjusted incidence of EC per 100,000 population per year in 2002–2006 was 7.9 in white men, 1.9 in white women, 9.3 in black men, and 3.0 in black women.⁷ In low-risk countries such as the United States, the male-to-female ratio of cases is usually about 3 : 1 to 4 : 1,⁷ but in the highest risk populations, this ratio approaches or even falls below 1 : 1.^3,⁶

Fig. 30.1 Worldwide incidence of esophageal cancer in men, 2002.

(Data from Ferlay J, Bray F, Pisani P, et al: GLOBOCAN 2002: Cancer incidence, mortality and prevalence worldwide: IARC cancer base No. 5. version 2.0, Lyon, 2004, IARC Press.)

Throughout most of the world, most esophageal cancer cases are esophageal squamous cell carcinomas (ESCCs).⁵ In Western countries, the incidence of ESCC has been gradually declining, and the incidence of esophageal adenocarcinoma has been rapidly increasing over the last 30 years,^5,⁸ so that now greater than 50% of esophageal cancer cases in the United States are esophageal adenocarcinoma.^5,⁷

In most low-risk countries, cigarette smoking and alcohol consumption are the dominant risk factors for ESCC.⁹^–¹¹ In the United States, greater than 90% of ESCC cases can be attributed to these two exposures alone.¹² Additional contributing risk factors include a low dietary intake of fruits and vegetables and factors related to low socioeconomic status.¹³^–¹⁶ A few host medical conditions have also been associated with increased risk of ESCC in low-risk populations, including previous or concurrent squamous cell carcinoma of the head and neck region, achalasia, tylosis, caustic esophageal strictures, and Plummer-Vinson syndrome.¹⁷

In most high-risk populations, tobacco and alcohol are not major risk factors for ESCC. Tobacco consumption in these groups is typically low, both in terms of the prevalence of smoking and in the amount of tobacco consumed by smokers, and alcohol consumption is even lower.¹⁸^–²⁰ In addition, in the highest risk areas, where there are nearly as many ESCC cases in women as in men, virtually none of the women smoke or drink.¹⁸^–²⁰ These high-risk groups may be exposed to some of the major tobacco carcinogens, however, such as polycyclic aromatic hydrocarbons (PAHs), nitrosamines, and acetaldehyde, in other ways. Studies have documented high levels of PAH exposure in Linxian, a county in the high-risk region in north central China²¹; in northeastern Iran²²; and in southern Brazil.²³ The source of these exposures is unknown; in Linxian, it may be related to ingestion of ambient soot particles released from heating and cooking with soft coal in unvented stoves,^24,²⁵ and in Brazil, it may be caused by drinking the beverage maté.^23,²⁶ Nontobacco exposure to nitrosamines and acetaldehyde has also been suggested in Linxian^27,²⁸ and in northeastern Iran²⁹ possibly secondary to changes in oral bacteria that accompany poor oral hygiene and tooth loss. Other risk factors reported in high-risk areas include diets low in fruits and vegetables^20,^30,³¹; low levels of certain micronutrients, especially selenium and zinc^32,³³; low socioeconomic status³⁴; exposure to fungal toxins such as fumonisins³⁵; and drinking hot liquids.^30,^36,³⁷

One of the most consistent risk factors for ESCC in high-risk populations is family history,^20,^38,³⁹ and preliminary molecular studies support a role for genetic susceptibility in the etiology of ESCC in these areas. Studies have shown high frequencies of loss of heterozygosity (LOH),⁴⁰ characteristic patterns of gene expression,⁴¹ and significant differences in both LOH and gene expression by family history^41,⁴² in tumors from north central China, but no major susceptibility gene for ESCC has yet been identified.

Both cell types of esophageal cancer have a very poor prognosis. In SEER data for 1999–2005, the overall 5-year relative (disease-specific) survival for EC patients was 16.4%.⁷ This survival rate is improved from 4.7% in 1975–1979,⁴³ but it is still the third lowest survival rate (after pancreas and liver) among major cancers. In developing countries, the 5-year survival rates are usually less than 10%.⁵

The main reason for poor survival in esophageal cancer is that most tumors are asymptomatic and go undetected until they have spread beyond the esophageal wall. The esophagus is a distensible organ—it distends to let food pass—so most patients do not complain of dysphagia or other symptoms until the tumor significantly obstructs the lumen, and by that time it has usually invaded through the wall or metastasized or both. Significant reduction in esophageal cancer mortality both in low-risk and in high-risk populations requires the development of successful new strategies for screening asymptomatic high-risk individuals that can lead to diagnosis and treatment of more cases at earlier, more curable stages of the disease.

Five components needed for a successful early detection and treatment program for esophageal cancer are as follows:

1 Identification of clinically important precursor lesions, which would be the targets for screening and treatment

2 Accurate, cost-effective primary screening tests that can detect precursor and early invasive lesions and are acceptable to asymptomatic high-risk individuals

3 Reliable techniques for endoscopic localization of precursor and early invasive lesions so that diagnostic biopsies and focal therapy can be accurately targeted

4 Reliable techniques for accurate staging of early invasive lesions so that patients can be triaged to the most appropriate treatment

5 A spectrum of curative therapies for precursor and early invasive lesions that is acceptable to asymptomatic people

Table 30.1 summarizes these components for ESCC and our understanding of the current techniques available for each component. Techniques that are possible but are not yet established are followed by a question mark. Because the subject of this chapter is screening, we discuss only the first three of these components. Staging and endoscopic therapy are discussed in other chapters of this text.

Table 30.1 Components Needed for Successful Early Detection and Treatment Program for Esophageal Squamous Cell Carcinoma

Component	Current State of the Art
Identification of precursor lesion	Squamous dysplasia
Primary screen	Endoscopy
	Cytology?
	Molecular?
Endoscopic localization	Iodine staining
	Narrow band imaging?
	Magnification endoscopy?
Staging	Endoscopic morphology?
	Endoscopic ultrasound?
Therapy
High-grade lesions	Endoscopic mucosal resection
	Focal ablation
	Esophagectomy
Low-grade lesions	Chemoprevention?

Precursor Lesions of Esophageal Squamous Cell Carcinoma

In low-risk countries, squamous dysplasia (including carcinoma in situ) is accepted as the histologic precursor of ESCC because it is the established precursor in other organs lined by squamous epithelium, such as the cervix, and because it is often found adjacent to invasive cancer in esophagectomy specimens.⁴⁴ In high-risk populations, squamous dysplasia is also accepted as a precursor of ESCC,⁴⁵^–⁴⁹ but other histologic lesions, such as chronic esophagitis, atrophy, and basal cell hyperplasia, have also been proposed as precursors, based primarily on differences in the prevalence of these lesions in high-risk and low-risk groups.^46,^49,⁵⁰ These ecologic comparisons have not always been consistent, however.⁵¹ In the only two prospective studies, in which patients who underwent biopsy were followed over time, only squamous dysplasia was significantly associated with the later development of ESCC,^45,^47,⁴⁸ and increasing grades of dysplasia were associated with increasing risk.^45,⁴⁸ In the larger of these prospective studies, members of our group followed 682 adults from Linxian who underwent endoscopy without initial evidence of invasive cancer for up to 13.5 years and compared the cumulative incidence and relative risk of developing ESCC among patients with different initial biopsy diagnoses (Fig. 30.2 and Table 30.2).

Fig. 30.2 Histologic categories used in endoscopic studies in Linxian, China. A, Normal. B, Esophagitis. The epithelium is infiltrated by polymorphonuclear leukocytes. C, Basal cell hyperplasia. The basal zone is greater than 15% of the epithelial thickness, without cellular atypia. D, Mild squamous dysplasia. There is cellular atypia, confined to the lower third of the epithelium. E, Moderate squamous dysplasia. There is cellular atypia involving the lower two-thirds of the epithelium. F, Severe squamous dysplasia. The cellular atypia involves all thirds of the epithelium, without invasion of the lamina propria.

Table 30.2 Incidence and Relative Risk of Esophageal Squamous Cell Carcinoma (ESCC) during 13.5 Years of Follow-up, by Initial Histology, in an Endoscopy Cohort from Linxian, China

We believe that squamous dysplasia is the only confirmed, clinically relevant precursor lesion of ESCC in high-risk and low-risk populations. More recent consensus conferences and World Health Organization guidelines prefer to use the term intraepithelial neoplasia rather than dysplasia throughout the gastrointestinal (GI) tract, and they prefer to subdivide these lesions into two grades: low-grade intraepithelial neoplasia and high-grade intraepithelial neoplasia.^52,⁵³ In the squamous lesions of the esophagus, the guidelines suggest that high-grade intraepithelial neoplasia should include severe dysplasia and carcinoma in situ.⁵² The above-described data suggest, however, that there are three quite different levels of risk that are identifiable by histologic examination, and both moderate and severe dysplasia need to be targeted for screening and treatment.

Nonendoscopic Screening Techniques

Cytologic Techniques

The most common nonendoscopic screening technique for early detection of esophageal cancer is esophageal balloon cytology screening in high-risk populations. Two principal types of cytologic samplers have been used in these screenings: an inflatable balloon sampler first developed in China ⁵⁴^–⁵⁷ and an encapsulated sponge sampler first developed in Japan (Fig. 30.3).⁵⁸^–⁶¹ In the balloon technique, a deflated balloon covered by a cloth net or rubber ribbing is swallowed into the stomach, inflated, and withdrawn, collecting exfoliated cells and scraping the mucosal surface of the esophagus. At the upper esophageal sphincter, the balloon is deflated and removed. In the sponge technique, a polyurethane mesh is compressed inside a gelatin capsule and attached to a string or a thin plastic stylet. The capsule is swallowed into the stomach, where the gelatin dissolves, and the mesh expands. The mesh is pulled up the esophagus by the string, collecting exfoliated and scraped mucosal cells. In both methods, the collected cells are processed and stained for cytology and read for cellular abnormalities. Several studies of both of these methods have reported high sensitivities for detecting ESCC in symptomatic patients, but there are few data on their accuracy for detecting squamous dysplasia or for detecting ESCC in asymptomatic individuals, who would be the target group for any population screening effort.

Fig. 30.3 Sponge and balloon cytologic samplers used in Japan and China.

To investigate further the potential of nonendoscopic esophageal cytology for screening asymptomatic high-risk individuals, we performed two studies in Linxian to evaluate the screening characteristics of two commonly used Chinese balloons, a new American balloon, and an American-made encapsulated sponge. In this study, asymptomatic adults were examined by the cytology samplers, followed by Lugol’s iodine chromoendoscopy of all participants. The cytology slides were read using the Bethesda System, the standard Western cytologic criteria for diagnosing cervical and vaginal smears, and the cytologic diagnoses were compared with the “gold standard” biopsy diagnoses. In the first study,⁶² 439 patients were examined by both a Chinese balloon and the encapsulated sponge, in random order, and in the second study,⁶³ 359 patients were examined by the second Chinese balloon, and 381 patients were examined by the American balloon. The results of these studies were not encouraging: None of the samplers achieved a sensitivity of 50% for identifying biopsy-proven squamous dysplasia or cancer (Table 30.3). The most important problem with this screening technique appears to be two kinds of sampling error: blind sampling of a large organ and missing small lesions and morphologic evaluation of only a small percentage of the collected cells.

Table 30.3 Screening Characteristics of Esophageal Balloon Cytology Samplers for Identifying Patients with Biopsy-Proven Esophageal Squamous Dysplasia in Linxian, China

Molecular Techniques

Molecular markers have several attributes that make them potentially useful for primary screening in an early detection and treatment program for ESCC and its precursor lesions. Molecular changes occur early in the neoplastic process (earlier than morphologic changes). Changes sometimes can be detected in clinical samples such as blood or stool that can be collected noninvasively. Molecular changes in DNA (e.g., hypermethylation, LOH, and mutations) can be amplified by polymerase chain reaction, so rare events can be found in complex clinical samples. Some molecular changes undergo clonal expansion, so they are present over a much larger field of tissue than the discrete foci of morphologic dysplasia, which should make them easier to detect when mucosal sampling is incomplete. Finally, measurement of molecular changes is more objective and less variable than cytologic or histologic identification of dysplasia.

Blood or one of its components would be an ideal clinical sample for primary screening purposes. Possible serologic approaches to screening include looking for tumor-specific DNA, RNA, or proteins that are secreted into the circulation or are released during tumor cell death and looking for autoantibodies produced by the host in response to tumor antigens. A few authors have looked for tumor-specific hypermethylated genes in the serum or plasma of patients with ESCC and have found them to be present in a few cases.^64,⁶⁵ Hibi and colleagues⁶⁴ found hypermethylated p16 in tumor tissue in 31 (82%) of 38 ESCCs and found this same marker in the serum of 7 (23%) of the 31 patients with positive tumors. Kawakami and associates⁶⁵ found hypermethylated APC in tumor tissue from 16 (50%) of 32 ESCCs and in the corresponding serum from 2 (12%) of the 16 tumor-positive patients. In the latter study, detection of hypermethylated APC in the plasma of esophageal adenocarcinomas was significantly associated with tumor stage (1 of 26 [4%] positive in stage I-II tumors; 12 of 26 [46%] positive in stage III-IV tumors); this should probably also be true of ESCCs. Although the proportion of tumors showing some hypermethylation increases when multiple genes are evaluated,⁶⁶ it still seems unlikely that many intraepithelial precursor lesions or stage I ESCCs would shed altered DNA into the serum or plasma that can be detected by such evaluations. Preliminary serologic studies looking for tumor-associated RNA⁶⁷ and proteins⁶⁸ have also been performed. Another possible serologic screening approach that has shown promise for detection of early squamous cell carcinomas of the lung and the head and neck region is identification of host autoantibodies generated against tumor-specific antigens.⁶⁹^–⁷¹

Stool is another clinical sample that could potentially contain information about the esophagus and could be collected noninvasively. Several authors have shown that neoplasia-specific DNA mutations can be detected in stool from patients with colorectal adenomas and adenocarcinomas,^72,⁷³ and the sensitivity of these assays can be increased by testing panels of markers^73a and by employing more sensitive new technologies.⁷⁴ There are also two reports of detection of other, more proximal aerodigestive malignancies, including esophageal carcinomas, by measuring high-molecular-weight or “long” DNA (DNA from nonapoptotic cells, which are more commonly sloughed from neoplasms)⁷⁵ or tumor-specific DNA mutations⁷⁶ in stool. Detection of esophageal precursor lesions or early invasive ESCCs by these methods may be unlikely, but it should still be evaluated.

At least for the near future, nonendoscopic molecular screening for early ESCC and its precursor lesions may still need to depend on evaluation of esophageal cell samples. As discussed previously, current morphology-based cytologic techniques are not sufficiently sensitive to find patients with focal squamous dysplasia, but molecular techniques may improve this sensitivity, especially DNA changes that can be amplified. One study reported promoter methylation findings for eight genes in esophageal balloon cytology samples from 147 patients with endoscopic biopsy diagnoses ranging from normal to severe squamous dysplasia and showed a 50% sensitivity and 68% specificity for identifying patients with high-grade dysplasia.⁷⁷ The most promising possibility may be the detection of molecular changes that have undergone clonal expansion and affect large areas of the squamous mucosa, similar to the “field effects” previously documented for p16 and p53 lesions in Barrett’s esophagus.^78,⁷⁹ If such field-wide molecular abnormalities are identified that reliably precede or accompany squamous dysplasia, a simple, imperfect sampler that is acceptable to patients, such as an encapsulated sponge, may be able to identify accurately or rule out the presence of an abnormal field, which may be sufficient to triage patients appropriately to endoscopy.

Endoscopic Screening Techniques

Various different endoscopes have been used to screen for early ESCC and its precursor lesions. These endoscopes may differ in size, resolution, and magnification characteristics. Ultrathin endoscopes measuring less than 6 mm (with a biopsy channel) may be used. A trial with a 3.1-mm stand-alone battery-powered fiberoptic esophagoscope (which did not have a biopsy channel) showed that it was feasible and accurate in detecting esophageal pathology.⁸⁰ Typically, instruments with diameters of 9 mm are employed. Standard resolution and magnification instruments suffice, but advantages of higher resolution and magnification have been described.⁸¹ Conventional videoendoscopes are equipped with CCD chips of 100k to 300k pixels, meaning that each image is built up from 100,000 to 300,000 individual pixels. This technical feature, pixel density, determines the resolution. Newer instruments with CCDs having 600,000 to 1 million pixels are now commercially available and are referred to as high-resolution instruments. The feature of high resolution is distinct from high magnification, which can be accomplished by either optical magnification or electronic magnification. With optical magnification, a mechanical system that can move a lens at the tip of the endoscope along the longitudinal axis of the endoscope creates the ability for the endoscopist to adjust the focal distance of the device to allow for close viewing of mucosal details. Depending on which endoscope and which optical magnification system is used, the image can be enlarged 75 to 115 times. With optical magnification, image resolution can be preserved during the magnification process. With an electronic magnification technique, the actual magnification is performed by the videoprocessor. An area in the center of the endoscopic images magnifies only the pixels in the target area, which are used to generate the enlarged image. It is often more difficult to preserve image resolution in this method. Features of high resolution and high magnification can be combined. Most of the published literature regarding endoscopic screening describes the use of standard endoscopes.

When screening is performed endoscopically, visual inspection to identify pathology is the first step. In the absence of staining, dysplasia may appear as normal mucosa, irregular mucosa, a small white patch, a focal red area, an erosion, or a plaque. Early ESCC is usually seen as an erosion, a plaque, or a nodule (Fig. 30.4).⁸²

Fig. 30.4 Endoscopic categories used in the studies in Linxian, China. A, Normal mucosa. B, Irregular mucosa, seen focally at 3 o’clock position. C, Small white patch, at 6 o’clock position. D, Focal red area, at 3 o’clock position. E, Small punched-out erosion, at 3 o’clock position. F, Large broad-based erosion, at 11 to 4 o’clock position. G, Plaque, at 2 to 6 o’clock position. H, Nodules, at 4 o’clock position and 5 to 7 o’clock positions.

Chromoendoscopy

Chromoendoscopy has been shown to assist with definition of early precancerous and cancerous lesions in numerous patients. When chromoendoscopy is used, Lugol’s iodine is generally employed for staining. The first description of iodine staining was by Schiller in 1933,⁸³ and it was used in the uterine cervix. The first reports using iodine as a supplement to esophagoscopy were by Brodmerkel,⁸⁴ Nothmann and colleagues,⁸⁵ and Voegeli.⁸⁶ Iodine reversibly binds to glycogen, which is abundant in normal nonkeratinized squamous epithelial cells. Iodine staining causes a brown color change in normal esophageal mucosa. Inflamed, dysplastic, or cancerous cells are relatively depleted of glycogen and do not take up as much of the dye. Foci of glycogenic acanthosis have more than the usual amount of glycogen and appear overstained. Lugol’s iodine can be used to identify abnormal mucosal areas (unstained lesions), including esophagitis, dysplasia, and ESCC (Fig. 30.5). It also can be valuable in delineating the margins of these lesions, which is particularly relevant if endoscopic mucosal resection or an ablative treatment is contemplated.

Fig. 30.5 Detection of squamous dysplasia by iodine chromoendoscopy. Before staining, all of the mucosa appears normal. After staining, the normal mucosa stains brown, but an abnormal area at the 6 o’clock position remains unstained. Biopsy specimens of this unstained lesion showed mild dysplasia.

In a typical procedure, the patient is placed in the left lateral decubitus position. Sedation may or may not be used. Often a “Cetacaine-type” medication is sprayed onto the throat, or a “Dicaine-type” slurry is drunk. If sedation is used, typically a benzodiazepine with or without a narcotic is given. The esophageal mucosa must be assessed in its entirety, so evaluation must begin as soon as the instrument passes the upper esophageal sphincter. The endoscopist searches for mucosal abnormalities as described previously (see Fig. 30.4).

If the initial endoscopy without staining is to be followed by chromoendoscopy, biopsies are deferred until after the staining. Various concentrations of Lugol’s iodine have been employed, but in most cases the concentration is between 1% and 2%. The formula of the iodine stain that we use (12 g iodine + 24 g potassium iodide dissolved in 1000 mL of water) was taken from the work of Endo and Ide.⁸⁷ This formula is slightly stronger than Lugol’s original solution (1 g iodine + 2 g potassium iodide in water to 100 g). The formula we use has been called 1.2% Lugol’s by some authors, referring to its elemental iodine content, and 3% Lugol’s by other authors, referring to the total (iodine + potassium iodide) iodine content. For this reason, it is important to specify the formula that is used.

Some investigators advocate pretreatment with a water rinse or a mucolytic agent to remove mucus from esophageal mucosa.⁸⁸ It has not been our practice to do so.

The dye is delivered through a spray catheter passed via the biopsy channel of the endoscope. Specifically designed spray catheters can be used (e.g., Olympus PW-5L spray catheter; Olympus America, Inc., Melville, NJ) or endoscopic retrograde cholangiopancreatography catheters. Typically, spraying begins distally, at the squamocolumnar junction, and as the endoscope is withdrawn proximally, it is rotated so that the dye covers the entire mucosal surface. One should be careful not to withdraw proximally all the way to the upper sphincter, to avoid inadvertently spraying proximal to the sphincter. Generally, 20 to 25 mL of Lugol’s solution is sprayed. If the entire surface is not covered, further spraying is necessary. The normal mucosa becomes brown in appearance. The dye begins to fade in 5 to 8 minutes, so biopsy specimens should be taken as quickly as possible. If unstained lesions are seen, directed biopsy specimens are taken. If there are no unstained areas, biopsy specimens should be taken in a systematic way. At the end of the examination, the stomach should be carefully suctioned to remove as much of the iodine solution as possible.

Our group evaluated the utility of mucosal iodine staining to improve endoscopic visualization of dysplasia and ESCC in the high-risk population of Linxian, China.⁸⁹ In our mucosal staining study, 225 patients with evidence of dysplasia or carcinoma found on previous balloon cytology were evaluated with endoscopy before and after staining with 1.2% Lugol’s iodine solution. In these patients, 253 unstained lesions, 94 foci of high-grade (moderate or severe) dysplasia and 20 invasive ESCCs were found. Before staining, the sensitivity of visible lesions for identifying high-grade dysplasia or ESCC was 62%, and the specificity was 79%. After staining, the sensitivity of unstained lesions for identifying high-grade dysplasia or ESCC was 96%, and the specificity was 63%. Mucosal iodine staining significantly improved endoscopic detection of high-grade dysplasia and ESCC.

In this study, Lugol’s staining also greatly improved delineation of the significant mucosal abnormalities. Examples of dysplastic lesions before and after iodine staining are shown in Fig. 30.6. The outlines of the lesions are much clearer after staining. Clarity of the lesion outlines is very important when endoscopic treatment is delivered (Fig. 30.7).⁹⁰

Fig. 30.6 Improved delineation of squamous dysplasia by iodine chromoendoscopy. A, Before staining, irregular mucosa is seen at the 3 o’clock position. After staining, the outline of the lesion is much more distinct. Biopsy specimens showed moderate dysplasia. B, Before staining, a large broad-based erosion is seen at the 7 to 10 o’clock position. After staining, the borders of this lesion are clearer, and a second lesion is identified at the 5 o’clock position. Biopsy specimens of both lesions showed severe dysplasia. C, Before staining, the mucosa appears normal. After staining, a large unstained lesion with sharp borders is seen. Biopsy specimens showed severe dysplasia.

Fig. 30.7 Use of chromoendoscopy for endoscopic mucosal resection. A, Dysplastic lesion before staining. B, Immediately after staining with Lugol’s iodine. C, Stain is fading 5 minutes later; two cautery marks are in place. D, Injection with saline. E, Lesion seen through the cap on the tip of the endoscope for snare technique removal. F, Snare now closed around lesion. G, Esophageal defect seen after endomucosal resection. H, Repeat stain with iodine to assess for residual pathology. I, Resected specimen.

Misumi and colleagues ⁸⁸ evaluated the role of Lugol’s dye endoscopy in diagnosing ESCC in 17 lesions in 10 patients. In 14 instances, there was an abnormality before staining—either a color change (redness) or elevation or depression. The lesions ranged in size from 0.7 to 4.0 cm. There was good correlation in all cases between the endoscopic detection of the unstained area and the lesion margins in the final surgical specimen. The investigators concluded that Lugol’s staining was useful in diagnosing ESCC because it found some abnormalities not detected by routine endoscopy, and it provided more accurate information about the extent of the cancer.

Freitag and coworkers ⁹¹ studied patients at risk for squamous cell carcinoma in southern Brazil, where esophageal cancer is the fourth most frequent neoplasm in men and the eighth most common neoplasm in women. When Lugol’s solution was applied in this high-risk group, if unstained lesions were found, there was an 80% sensitivity and 63% specificity for the detection of dysplasia.

Shiozaki and associates ⁹² used Lugol’s iodine to screen asymptomatic patients with head and neck cancer for esophageal cancer. Of 178 screened patients, 9 (5%) had one or more esophageal cancers identified. Of the 13 cancers found, 8 were early stage disease, with no lymph node metastases. Only 4 of the 13 lesions were seen at routine endoscopy. The investigators concluded that Lugol’s chromoendoscopy should be considered in asymptomatic patients with a history of head and neck cancer and in any other patient populations at high risk for ESCC.

Ina and colleagues ⁹³ also evaluated the utility of Lugol’s staining in male patients with oral or oropharyngeal cancer who had no symptoms of esophageal disease. Lugol’s dye was used to screen 101 patients with oral cancer and 26 patients with oropharyngeal cancer. Of these 127 patients, 8 (6%) had esophageal cancer; in 5 of the 8 patients, no abnormality was seen on routine endoscopy or barium swallow.

Muto and associates ⁹⁴ performed chromoendoscopy with Lugol’s solution in 389 patients with head and neck cancer and found 54 (14%) who had a synchronous primary ESCC. They characterized the staining pattern in the 389 patients into four groups. Type I patients had normal brown staining; type II patients had 10 or fewer unstained areas; type III patients had more than 10 unstained areas; and type IV patients had irregular-shaped, multiform unstained lesions. In the latter group, 55% of the patients had ESCC. Muto and associates⁹⁴ also followed their patients for more than 1 year, and patients who had the type IV pattern were more likely to develop a metachronous lesion during that time. They concluded that the irregular-shaped, multiform lesions likely represented a “field cancerization.”

Using iodine staining, Hashimoto and coworkers ⁹⁵ found 8 cases of dysplasia (7%) and 10 cases of ESCC (8%) in 118 Brazilian patients with head and neck cancer. Because most reports using Lugol’s staining in high-risk patients are from Asia, this Brazilian study lends support to its utility in any patient population at risk throughout the world.

In addition to patients with head and neck cancer, alcoholics are another group at high risk for developing esophageal squamous cancer. Yokoyama and colleagues ⁹⁶ studied a cohort of 629 male alcoholics by endoscopy with iodine staining. Unstained lesions were observed in 162 patients (26%), and 36 unstained lesions in 21 patients (3%) were superficial ESCC. Because these lesions were superficial, endoscopic mucosal resection could be performed in 17 patients. In a similar group of 255 alcoholic Japanese patients, Ban and associates⁹⁷ found unstained lesions in 22% and cancer in 4%.

In southern Brazil, Fagundes and colleagues ⁹⁸ studied 190 patients who were at risk for esophageal cancer because of increased consumption of alcohol or the beverage maté (a hot infusion of herbs). Unstained lesions were present in 23 of these asymptomatic patients, of which 6 showed dysplasia. In some patients with a normally stained midesophagus, there was also evidence of dysplasia, but the likelihood of finding dysplasia was eightfold higher in areas that were unstained. The investigators concluded that iodine chromoendoscopy improves the detection of dysplasia and should be added to conventional upper GI endoscopies in patients in selected populations at high risk for squamous cell carcinoma.

Strader and colleagues (D. Strader, personal communication, January 1, 2004)⁹⁹ screened 98 alcoholic American veterans with endoscopy and iodine staining; 28 (29%) had unstained lesions, but no foci of dysplasia or cancer were found. Screening in this study began at age 40, which may have been too young in this population. Meyer and associates¹⁰⁰ prospectively compared the diagnostic accuracy of videoendoscopy with and without Lugol’s staining for the detection of esophageal cancer in 158 alcoholic and smoking patients in France. Before staining, 12 patients had 14 endoscopically identifiable dysplastic or cancerous lesions. After staining, these numbers increased to 13 patients with 17 lesions. The unstained areas were significantly larger than the mucosal abnormalities seen before staining.

In an attempt to understand staining patterns with Lugol’s iodine better, Mori and colleagues ¹⁰¹ applied iodine staining to 24 ESCC specimens resected by subtotal esophagectomy. They divided the staining patterns into four groups: type I, hyperstaining; type II, normal brown staining; type III, less intense staining; and type IV, unstained. Most type IV lesions were invasive carcinomas, carcinomas in situ, or severe dysplasia. Moderate to mild dysplasia or atrophy appeared as type III staining. The investigators also showed that there was good correlation between the histologic tumor margin and the absence of staining.

Table 30.4 summarizes the detection of significant lesions before and after iodine staining in the above-described studies. In these patients, who had various degrees of risk for developing ESCC, 39% of the lesions were detected only after staining with iodine. This finding underscores our belief that mucosal iodine staining improves endoscopic detection and delineation of high-grade squamous dysplasia and ESCC.

Table 30.4 Detection of Squamous Dysplasia and Esophageal Squamous Cell Carcinoma (ESCC) in Studies Using Iodine Chromoendoscopy

Although Lugol’s iodine is generally regarded as safe when sprayed onto esophageal mucosa to facilitate early detection of squamous cell carcinoma, some side effects have been reported.^102,¹⁰³ Mucosal irritation can be manifested by retrosternal pain, discomfort, or nausea. Free iodine, one of the components of Lugol’s, causes these side effects, and the risk of complication is greater with increasing concentrations of iodine.

Sodium thiosulfate (STS), a water-soluble neutral chemical, can be used as an intravenous antidote for cyanide, iodine, arsenic, or heavy metal intoxication. STS spray was first reported as being effective for reducing symptoms caused by staining with iodine solutions in 1993.^103,¹⁰⁴ Kondo and coworkers¹⁰⁵ compared the utility of 5% STS with no treatment and with aluminum magnesium hydroxide gel (Maalox) in 120 patients after spraying the esophagus with 3% Lugol’s iodine. STS was most effective in reducing adverse symptoms, and the authors recommended its routine use after staining with iodine.

Although the largest experience using dyes to assist with screening for ESCC and its precursor lesions is with Lugol’s iodine, there are also some reports using toluidine blue (also called tolonium chloride). Toluidine blue stains cellular nuclei, and this property makes it useful for identifying malignant tissue, which has an increased DNA content and a higher nuclear-to-cytoplasmic ratio. The mucosa of a normal esophagus does not stain with toluidine blue (Fig. 30.8).

Fig. 30.8 Toluidine blue chromoendoscopy.

When toluidine blue is used, a proteolytic enzyme (e.g., proteinase) in water is frequently used.⁸⁷ It may be ingested by mouth or sprayed on the mucosal surface. Toluidine blue is delivered as a 2% solution through a spray catheter in a manner identical to Lugol’s iodine.

Seitz and colleagues ¹⁰⁶ screened 100 patients with a history of alcohol and tobacco abuse using toluidine blue. Two carcinomas and 15 cases of dysplasia were detected. Two studies have looked at the use of toluidine blue dye spraying in patients with head and neck cancers. Hix and Wilson¹⁰⁷ found a 17% prevalence of ESCC in 18 asymptomatic patients with a history of head and neck malignancy. Contini and colleagues¹⁰⁸ found three esophageal neoplasms, two of which were not seen on routine endoscopy. Additional cases of leukoplakia and esophagitis were also identified. To date, no complications have been reported with toluidine blue. Because the dye is absorbed through the GI tract and excreted in the urine, caution should be observed in cases with decreased renal function.

New Techniques for Screening without Chromoendoscopy

Various new adjunct techniques have been combined with standard endoscopy to determine if they can improve the diagnosis of neoplastic esophageal lesions and assist with staging and therapy. These “high-tech” approaches have attempted to improve on the “low-tech” method of chromoendoscopy, which has been extremely valuable and has withstood the test of time. Some more promising newer methods are (1) narrow band imaging (NBI) and multiband imaging, (2) autofluorescence, (3) endoscopic ultrasound (EUS), (4) optical coherence tomography, (5) endocytoscopy, and (6) confocal laser microscopy.¹⁰⁹

NBI is a technique that uses optical filters instead of dyes. In NBI, the depth of light penetration in tissue depends on its wavelength. The shorter the wavelength, the more superficial the penetration. In the visible light spectrum, blue penetrates most superficially and is the most valuable for mucosal imaging. Red light penetrates more deeply and is more appropriate for submucosal imaging. With NBI, the intensity of the blue light improves, which gives better definition to mucosal patterns and to blood vessels, particularly because blue light is preferentially absorbed by hemoglobin. Several endoscopic images with NBI are shown in Fig. 30.9. There has been considerable investigation with the use of NBI for Barrett’s esophagus. Hamamoto and associates¹¹⁰ were the first to use NBI for Barrett’s esophagus. Kuraoka and colleagues¹¹¹ reported on the use of NBI and screening for early esophageal cancer in a group of patients who were at high risk because of heavy alcohol intake and previous head and neck cancers. They found a sensitivity, specificity, and positive predictive value of 100%, 59%, and 10%. Katada and coworkers¹¹² studied NBI for detecting metachronous superficial pharyngeal squamous cell carcinomas.

Fig. 30.9 Detection of early squamous cell carcinoma with narrow band imaging (NBI). A, Small flat erosion is observed at the 5 o’clock position. With white light it is difficult to see. B, With Lugol’s iodine, it shows the lesion more distinctly. Biopsy specimen showed squamous cell carcinoma. C, Using NBI, the lesion is seen as a brown spot, and the contour is more easily appreciated. D, Intraepithelial papillary capillary loops (IPCLs) are noted.

(Photographs courtesy of Dr. Jonathan Cohen.)

Multiband imaging, initially known as Fujinon Intelligent Chromoendoscopy (FICE), uses certain dedicated wavelengths similar to NBI. However, in contrast to NBI, which uses optical filters to narrow the bandwidth, FICE takes an ordinary endoscopic image from the videoprocessor and processes the photons in a way that allows virtual images to be constructed with a choice of different wavelengths (Fig. 30.10). Pasha and coworkers¹¹³ reported on the use of FICE for studying the colonic mucosa. Many exquisite photographs showing high-resolution endoscopy plus NBI are contained in a text edited by Cohen.¹¹⁴

Fig. 30.10 Fujinon Intelligent Chromoendoscopy (FICE). A, Esophageal cancer is seen with white light endoscopy and with FICE. B, Magnified view shows intraepithelial papillary capillary loops (IPCLs) with white light and with FICE.

(Photographs courtesy of Jason Ylizarde, Fujinon, Inc.)

In standard endoscopy, incident light reacts with the tissue, and the reflected light is captured by the endoscope and processor. In addition to the standard imaging that occurs, there are other tissue interactions, one of which is autofluorescence. With autofluorescence, an incoherent violet-blue light is used as the source of excitation. The optical signals that result depend on the structure and molecular composition of the tissues being excited. Because neoplasia changes the composition of the tissue, it is theoretically possible to distinguish between normal tissue, dysplasia, and neoplasia (Fig. 30.11). The Erlangen group has contributed considerably in this area. Mayinger and colleagues¹¹⁵ used 129 endogenous fluorescence spectra in evaluating nine patients with squamous cell carcinoma and four patients with adenocarcinoma. Malignant and benign spectra were differentiated with the aid of a mathematical algorithm; by using this algorithm, the investigators had a sensitivity of 97% and a specificity of 95% for the diagnosis of esophageal carcinoma. A commercially available endoscope by Olympus allows for the detection of autofluorescence.

Fig. 30.11 Autofluorescence imaging. The purple area of autofluorescence seen at the 2 to 6 o’clock position highlights an area of esophageal cancer. Normal nonneoplastic areas are seen in green.

EUS has made a significant impact on staging cancers of the digestive tract including esophageal cancer. To date, however, it has had limited utility for detecting dysplasia or defining the exact level of mucosal or submucosal penetration by early esophageal cancer. A study by May and colleagues ¹¹⁶ looked at the accuracy of staging early esophageal cancer using high-resolution endoscopy and high-resolution EUS. The study was carried out in 100 patients using a 20-MHz miniprobe. Overall, rates of accuracy were 79% using EUS staging for mucosal lesions, but for submucosal tumors the overall accuracy for EUS was only 48%.

Optical coherence tomography is a method analogous to high-resolution EUS, but it employs light waves instead of sound waves. It has a high resolution, but the sampling depth is limited to 1 to 2 mm. It has been studied in assessing Barrett’s esophagus, but it has not been formally studied in ESSC.

The “Holy Grail” for endoscopy has been defined as an imaging system that would allow for an optical biopsy, which is a real-time image that would be equivalent to histopathologic sections. Advances have been made in this area using the principles of light contact microscopy. Endocytoscopy (Olympus, Tokyo, Japan) uses a system by which a 3.2-mm probe is passed through the accessory channel of an endoscope. Kumagai and coworkers ¹¹⁷ studied 12 patients with superficial esophageal cancer. They sprayed the esophageal mucosal surface with 1% methylene blue and studied the targeted mucosa. In normal esophageal mucosa, the nuclei showed regular staining and a low nucleus-to-cytoplasm ratio. In contrast, in esophageal cancer, there were irregularities in cell distribution and a high nucleus-to-cytoplasm ratio. Inoue and associates¹¹⁸ studied endocytoscopy in patients who had already undergone chromoendoscopy and NBI. They reported an overall accuracy for differentiating nonmalignant from malignant pathology in 82% of patients.

Confocal laser microscopy is a technology in which laser illumination is combined with an optical technique that allows different illumination detection systems to be focused in the same plane; it is referred to as confocal. The tissue can be scanned at various depths. Currently two companies have products in this area (EC383FK; Pentax, Tokyo, Japan, and Cellvizio; Mauna Kea Technologies, Paris, France). Both systems permit simultaneous endoscopic and endomicroscopic examination, permitting in vivo histologic examination of the mucosa with high resolution. With confocal laser microscopy, a fluorescein dye is injected intravenously; about 30 seconds later, there is a color change in the GI mucosa. Magnification is up to 1000 times. The observation range is 475 to 600 µm, the lateral resolution is approximately 1 µm, and the vertical resolution is 3 to 7 µm.

Several authors have reported on their experience using confocal laser and endomicroscopy for imaging of early squamous cell cancer.¹¹⁹^–¹²¹ The authors were able to show differences in both the cellular structure and the vascular structure, including a demonstration of intrapapillary capillary loops (IPCLs). The confocal images were compared with standard histology. The cancer tissues exhibited more irregular cellular arrangements, increased diameter of IPCLs, and irregularly shaped IPCLs. This technology is promising but is still in its early stages of development. Confocal images are shown in Fig. 30.12. It is hoped that these “high-tech” developments will contribute to meaningful advances in the management of a large number of patients.

Fig. 30.12 Confocal laser microscopy. A, Confocal image of normal squamous epithelium shows normal cellular structure. B, Image of esophageal cancer shows disorganized cells with enlarged and hyperchromatic nuclei.

(Photographs courtesy of Dr. Cuong Nguyen and Dr. Ananya Das.)

Endoscopic Morphology

Several classification systems have been used for defining the morphologic appearance of early GI cancer. The lack of standardization has led to confusion, and two separate literatures have evolved in the “East” and the “West” concerning the macroscopic appearance of superficial neoplastic lesions. This development discouraged international collaboration. A meeting was held in Paris in 2002 with leaders from the East and the West and terms were agreed on; this is referred to as the Paris classification of superficial neoplastic lesions of the GI tract.¹²² A follow-up meeting was held in Osaka in 2003. A description of this classification follows.

Neoplastic lesions of the esophagus are called superficial when their endoscopic appearance suggests that invasion is limited to the mucosa or submucosa, and these types of lesions are considered to be type 0 (for increasingly deeper lesions, the classification goes to higher numbers I, II, III, and IV). Type 0 is divided into three categories: 0-I, protruding; 0-II, nonprotruding and nonexcavated; and 0-III, excavated. Type 0-II is subdivided further into slightly elevated (IIa), completely flat (IIb), and slightly depressed (IIc). Table 30.5 and Fig. 30.13 summarize this classification.

Table 30.5 Morphologic Classification of Type 0 Esophageal Lesions with a Superficial Appearance at Endoscopy

Morphologic Appearance	Type
Protruding
Pedunculated	0-Ip
Sessile	0-Is
Nonprotruding and nonexcavated
Slightly elevated	0-IIa
Completely flat	0-IIb
Depressed	0-IIc
Excavated	0-III

Fig. 30.13 Paris classification of superficial neoplastic lesions in the digestive tract.

Clinical Application of Screening and Treatment for Early Esophageal Squamous Cell Carcinoma and Its Precursor Lesions

When either squamous cell carcinoma or adenocarcinoma of the esophagus is discovered after symptoms have begun, the prognosis is very poor, with a 5-year survival of less than 15%. In addition, the quality of life of patients who are diagnosed with esophageal cancer at a late stage is poor because in addition to the primary disease and the treatments for it, the social stigma of difficulty eating adds an additional burden to the last months of a patient’s life. Both with screening programs for adenocarcinoma that focus on patients with Barrett’s esophagus and with screening programs for patients with squamous cell carcinoma in high-risk categories, the goal is early diagnosis so that treatment can be carried out at a curable stage.

In the United States and Europe, where adenocarcinoma is common and Barrett’s esophagus has been identified as the precursor lesion, screening for Barrett’s esophagus and surveillance for progression in patients who have Barrett’s esophagus have led to the application of several new therapies. When localized elevated endoscopic abnormalities are found in patients with Barrett’s esophagus, endoscopic mucosal resection is a valuable treatment. If the lesion is superficial and fully removed, consideration can be given to other endoscopic therapies that eradicate larger segments of Barrett’s tissues. A study by Shaheen and colleagues ¹²³ showed in a randomized sham-controlled trial that radiofrequency ablation was beneficial for patients with low-grade or high-grade glandular dysplasia. There has also been interest in pursuing similar endoscopic therapy for patients with early esophageal cancer.

Pouw and colleagues ¹²⁴ more recently described successful radiofrequency ablation for widespread early squamous cell carcinoma in a single case. A collaborative international effort has begun in China to study radiofrequency ablation for high-grade squamous intraepithelial neoplasia and early ESCC (G. Wang, personal communication, January 15, 2009). Chinese investigators from the Cancer Institute, Chinese Academy of Medical Sciences and Feicheng Medical Center, are collaborating with American and Dutch physicians to evaluate this treatment technique in patients with moderate or severe dysplasia or invasive squamous cell carcinoma limited to the lamina propria (T1m2). To be eligible, the esophageal mucosa must be flat, without nodules or ulceration (type 0-IIb), and patients can have no signs of metastasis on EUS or CT. The primary endpoint will be the percentage of subjects who are completely free of dysplasia or cancer at 12 months after the first ablation procedure. If such endoscopic techniques are shown to be effective for treatment of early neoplastic lesions in such high-risk populations, primary screening for early ESCC and its precursor lesions would be even more valuable.