Screening for Esophageal Squamous Cell Carcinoma: Recent Advances

Abstract

Esophageal squamous cell carcinoma (ESCC) is the most common type of esophageal cancer worldwide, with a high mortality due to advanced stage at diagnosis. Although most common in an area known as the Asian Esophageal Cancer Belt, which extends from the Caspian Sea to northern China, and in parts of Africa, high-risk populations also exist elsewhere in the world. Screening for ESCC has been practiced in a few geographic areas and high-risk populations, with varying levels of success. Esophageal squamous dysplasia is recognized as the precursor lesion for ESCC. Endoscopic screening for ESCC/esophageal squamous dysplasia is expensive and not sufficiently available in many high-risk regions. Recent advances in non-endoscopic screening enhanced by biomarker-based disease detection have raised the prospect of improved accuracy and availability of screening for esophageal squamous dysplasia and early stage ESCC. Development of a cost-effective, accurate, and well-tolerated screening test, if applied in endemic areas and high-risk populations, has the potential to reduce mortality from this deadly disease worldwide. In this review, we summarize recent developments in endoscopic and non-endoscopic screening modalities.

Introduction:

Esophageal cancer ranks as the 6^th most common cause of cancer death worldwide, with 456,000 new cases and 400,000 deaths annually^{1, 2}. Nearly 90% of esophageal cancer cases worldwide are esophageal squamous cell carcinoma (ESCC)^2–4. The highest risk areas for ESCC are found in two geographic belts, the Asian Esophageal Cancer Belt across central Asia, from the Caspian Sea to northern China, and a belt on the eastern coast of Africa, from Ethiopia to South Africa (Figure 1).

Figure 1:

The Asian and African esophageal cancer belts (in dark blue), where over 90% of cases are esophageal squamous cell carcinoma.

(Reprinted with permission from Ferlay J SI, et al. GLOBOCAN 2012 v1.0, [Internet]. Lyon, France: International Agency for Research on Cancer, 2013.)

Unfortunately, most patients with ESCC present with advanced disease: 5-year survival rates are less than 20% in developed countries and less than 5% in many developing countries where most of the disease occurs^{3, 5, 6}. Additionally, the outcomes of patients diagnosed after the onset of symptoms have not changed in the last several decades despite advances in therapy⁵. Early stage carcinoma, however, is associated with a substantially improved survival of 80–90% at 5 years when treated endoscopically or surgically^{3, 7}. Therefore, early detection strategies are needed to improve outcomes of patients with ESCC.

Analogous to the association of Barrett’s esophagus (BE) and its related dysplasia to esophageal adenocarcinoma (EAC), esophageal squamous dysplasia has been shown to be a precursor to ESCC based on natural history studies conducted in China^{8, 9}. These studies have shown that the risk of progression to cancer increases depending on the degree of dysplasia⁸. This has led to endoscopic treatment of high grade dysplasia (HGD). Given the considerable time needed in the transformation from dysplastic epithelium to malignancy¹⁰, a window exists in which detection and treatment of esophageal squamous dysplasia can reduce ESCC incidence.

In this review, we summarize and evaluate both conventional and emerging modalities to screen for this lethal disease and its precursor.

Rationale and Outcomes of Screening and Surveillance Programs

The main rationale for ESCC screening is to detect asymptomatic esophageal squamous dysplasia and early ESCC, to enable curative treatment. Several studies have demonstrated increases in early detection, mortality reduction, and cost effectiveness by endoscopic screening programs in high-risk populations.

In endemic areas in China, a community assignment trial with 10 years of follow up demonstrated a 33% reduction in the cumulative ESCC related mortality in intervention communities where 40–69 year-old adults were screened once by Lugol’s chromoendoscopy, compared to communities without screening.¹¹ Similar results were found in other large retrospective studies of high-risk Chinese populations.^{12, 13} Modeling studies have shown that population-based screening of high risk populations in China can be cost effective.^{14, 15}

In non-endemic areas, evidence supports endoscopic screening in patients with a history of head-and-neck cancers. In a case-control study, Su et al. found a higher prevalence of secondary ESCC in head-and-neck cancer patients receiving routine endoscopies compared to the non-screening group ( 4.5% vs. 3.0%, p=0.04), with earlier stage at diagnosis (p=0.03).¹⁶ Screening and early diagnosis correlated with improved survival, as Murakami et al. reported significantly superior 5-year survival in patients whose secondary ESCC was diagnosed on endoscopic screening versus symptomatically (60.4% versus 0.0%, p <0.01).¹⁷

For another high-risk population-tylosis, limited data from a small case series has shown that annual screening endoscopy detects early ESCC, with improved outcomes compared to those diagnosed after onset of symptoms.¹⁸

Given these findings, screening programs have been proposed for high-risk populations around the world. However, effective screening programs must be accurate, safe, cost-effective, and facilitate curative therapeutic intervention.¹⁹ With the advent of endoscopic treatment of esophageal squamous dysplasia and early stage ESCC, screening programs have gained additional impetus.

Current recommendations on ESCC Screening (Table 1)

Table 1:

Recommended Screening Guidelines for High Risk Populations

Risk factor	Screening method and duration	Outcome	Level of evidence^20
Head and neck cancer	Endoscopy with Lugol’s or NBI every 6 months to 1 year after completion of therapy for HNSCC, for 10 years	• Detects earlier stage disease • Improved survival • No evidence for cost-effectiveness,	II-III (moderate)
Tylosis	4 quadrant biopsies from proximal, middle, and distal esophagus starting at age 30; repeat every 1–3 years	• Effective for early diagnosis • Only beneficial for Type A (late onset) Tylosis	III-IV (low)
Achalasia	Yearly EGD 10–15 years after disease onset +/− Lugol’s solution	• No evidence for cost effectiveness • Need to screen many patients to detect one cancer	III (low)
Asian or African high-risk populations	One time Lugol’s chromoendoscopy beginning at the age of 40	• Screened groups have lower ESCC incidence and mortality rates	II-III (moderate)
History of caustic esophageal injury	Endoscopy every 2–3 years 10–20 years following the injury	• No evidence for effectiveness	IV (low)

^{^}Levels of evidence: Level I evidence: presence of at least one prospective, randomized, controlled trial, level II evidence: well-designed cohort or case-controlled studies; level III evidence: case series or flawed clinical trials; level IV evidence: opinions of respected authorities or expert committees; level V evidence: insufficient evidence to form any opinions

Given the low burden of disease in the general population of the United States, guidelines from the American Gastroenterology Association do not advocate population based screening for ESCC.²⁰ However, screening may be worthwhile in endemic areas for individuals over a certain age, and screening may also be indicated in certain high-risk groups such as patients with head-and-neck cancers, tylosis, or history of caustic ingestions. Although smoking and alcohol abuse have a dose-dependent association with ESCC, screening for ESCC is not routinely recommended unless other risk factors are present. A recently published risk-prediction model from an endemic population in China incorporating more than 10 risk variables was able to predict severe squamous dysplasia with an AUC between 0.62–0.85, with age being the single most significant risk factor.²¹ Further refinement of similar risk-prediction models is crucial to identify the target individuals for screening in most populations. Currently proposed screening guidelines are summarized in Table 1.

Technologies currently utilized and in development for ESCC/Squamous Dysplasia screening

Endoscopic screening methods (Table 2)

Table 2:

Pros and Cons of Various Endoscopic Screening Methods

Endoscopic screening modalities	Pros	Cons	PIVI criteria for Squamous Dysplasia Reached?
All modalities	•Visualizes the mucosa	• Uncomfortable if not done with sedation (which is frequently not available in resource poor areas) • Requires operator training and experience • Expensive equipment	•
Conventional white light endoscopy	• Readily available in developed areas	• Lacks sensitivity for precursor lesions	• No
Chromoendoscopy	• Inexpensive • Improves sensitivity for precursor lesions/dysplasia • Short learning curve • Clarity of lesion borders	• Irritant/allergic reactions to iodine • Lower specificity for precursor lesions (before biopsy diagnosis)	• Yes
Endoscopy with Narrow Band Imaging	• Improves sensitivity and specificity for precursor lesions/dysplasia • Does not require iodine	• Increased cost of equipment • Longer learning curve • Requires more operator expertise	• Yes
Transnasal endoscopy	• No need for sedation • Improves cost effectiveness	• Currently not standard of care, not widely available • Lack of trained operators • Cannot be combined with Lugol’s or NBI to detect precursor lesions. • Not yet tested on precursor lesions • Smaller biopsies • No therapeutic capabilities • Increased risk for patients with ENT pathology	• Yes (only if combined with FICE)
Endocytoscopy	• Cellular level resolution may obviate need for some or all biopsies	• A contact probe technology • Small field of view • Can only be used with other lesion localization technologies (Lugol’s or NBI) • Increased cost of equipment • Additional training needed • Not yet tested on precursor lesions	• Yes
Microendoscopy	• Inexpensive and portable • Cellular level resolution may obviate need for some or all biopsies • Can be portable and added to standard endoscope • High sensitivity and specificity for dysplasia and early stage cancer.	• A contact probe technology • Small field of view • Can only be used with other lesion localization technologies (Lugol’s or NBI) • Additional training needed • Variable cost-effectiveness when compared to standard endoscopic screening techniques	• Yes

Traditionally, endoscopy has been the modality of choice for ESCC screening. However, endoscopy encompasses a heterogeneous spectrum of technology: their applications in ESCC screening have been summarized in Table 2.

Lugol’s Chromoendoscopy

Although conventional white light endoscopy (WLE) can reliably detect invasive cancer of the esophagus, it lacks sensitivity for detecting esophageal squamous dysplasia, the precursor to ESCC and the main target of screening, which often appears similar to normal squamous mucosa.^{22, 23} While high-resolution WLE has been used in the detection of dysplasia in colonic polyps and Barrett’s esophagus, its role has not been formally evaluated in the detection of squamous dysplasia. Rather, Lugol’s chromoendoscopy is the current standard to highlight areas of abnormality, significantly increasing the ability to detect esophageal squamous dysplasia.^24–27

In Lugol’s chromoendoscopy, iodine binds reversibly to glycogen, which is less abundant in immature and rapidly dividing cells in the squamous epithelium, such as those found in esophageal squamous dysplasia or esophagitis. Dysplastic and inflammatory esophageal mucosa remain unstained compared to normal epithelium, allowing for targeted biopsies.²⁸ Figure 3 shows a dysplastic lesion on WLE, NBI, and Lugol’s chromoendoscopy.

Figure 3:

Endoscopic images of the same lesion of esophageal squamous dysplasia, with WLE, NBI, Magnifying NBI, and Lugol’s chromoendoscopy.

a. | Endoscopic finding with conventional white light imaging. A flat reddish lesion was seen in the upper thoracic esophagus (between the arrows). At the edge of the lesion, the vascular pattern in the surrounding mucosa was interrupted.

b. | Endoscopic finding with narrow band imaging of the lesion from panel a. The lesion was contrasted as darker area (arrow) and the boundary of the lesion is clearer than in white light images.

c. | Endoscopic finding with magnifying narrow-band imaging. Dilated irregular microvascular pattern was clearly visualized in the lesion. A distinct demarcation line between the background mucosa and the lesion is visible (arrow). Note the color of the epithelium in the lesion is browner than the surrounding mucosa. Accordingly, endoscopic diagnosis of intramucosal squamous cell carcinoma or high-grade dysplasia can be made with high confidence.

d. | Endoscopic finding using Lugol’s chromoendoscopy for the lesion depicted in panels a and b. The lesion is clearly visualized as a lighter (yellow) unstained area within the brown stained area. The boundary of the lesion can be clearly delineated with chromoendoscopy. (Reprinted with permission from Veitch AM, et al. Nat Rev Gastroenterol & Hepatol. 2015)

In a study of patients with head and neck cancer undergoing WLE followed by Lugol’s, the detection of advanced cancer was similar between WLE and Lugol’s, but WLE was only able to detect 55% of esophageal squamous dysplasia identified with chromoendoscopy.²⁵ Chromoendoscopy has been found to have a sensitivity ranging from 92%–100% and specificity of 37%–82% for detecting esophageal squamous dysplasia.^{27, 29–32} The main reason for the wide range in specificity is that esophagitis can appear as an unstained lesion, but this is readily discernable on histology.

Lugol’s chromoendoscopy is inexpensive and relatively easy to perform and interpret. Lugol’s solution contains iodine, potassium iodide and water; it can be made on site in developing countries. Therefore, Lugol’s chromoendoscopy is currently the standard of care for ESCC screening. Adverse effects of Lugol’s solution are infrequent but include irritant esophagitis/gastritis, hypersensitivity reactions, aspiration pneumonia, and thyrotoxicosis in patients with endemic goiters.^{33, 34} Furthermore, the interpretation of abnormal findings depends on the clinician administering the test, and interobserver variability has not been adequately described for ESCC screening.

Optical enhancement techniques: Narrow Band Imaging

By utilizing specific “narrow band” wavelengths (415 nm and 540 nm) that match the absorption peaks of hemoglobin, narrow band imaging (NBI) allows for clearer depiction of capillary networks and superficial mucosal patterns.³⁵ esophageal squamous dysplasia or early ESCC is characterized by a “brownish” appearance with abnormal vascular patterns of intracapillary papillary loops (IPCLs) (Figure 2). Detailed classifications of these abnormalities have been proposed and validated in the identification and staging of dysplasia and early carcinoma.³⁶

Figure 2:

Original intrapapillary capillary loop (IPCL) pattern classification.

The IPCL pattern classification includes two sets of diagnostic criteria. IPCL pattern classification from IPCL type I to type V-1 is used for the tissue characterization of flat lesions (red outline). IPCL pattern classification from IPCL type V-1 to type VN reflects cancer infiltration depth (blue outline). IPCL type III corresponds to borderline lesions which potentially include esophagitis or low-grade intraepithelial neoplasia. IPCL type III should be considered for endoscopic follow up. In IPCL type IV, high-grade intraepithelial neoplasia appears, and then further treatment with endoscopic mucosal resection (EMR) / endoscopic submucosal dissection is recommended. EMR/esophageal squamous dysplasia for IPCL types V-1 and V-2 should be also considered as they are definite M1 or M2 lesion with no risk of lymph node metastasis. IPCL type V3 corresponds to an M3 lesion and diagnostic EMR/esophageal squamous dysplasia should be applied as a “complete biopsy” to decide on a final treatment strategy. IPCL type VN corresponds to a “new tumor vessel” often associated with sm2 invasion with significantly increased risk of lymph node metastasis. Surgical treatment should be recommended.

(Reprinted with permission from Inoue H, et al. Ann Gastroenterol. 2015 Jan-Mar; 28(1): 41–48.)

In a prospective study with 202 subjects with high-risk factors for ESCC, where each subject sequentially underwent WLE, NBI, and chromoendoscopy, Nagami, et al., found that the accuracy, sensitivity, and specificity of NBI for detecting invasive ESCC and high grade intraepithelial neoplasia were 77.0, 88.3, and 75.2 %, respectively; compared to chromoendoscopy with 68.0, 94.2, and 64.0%, respectively.³¹ NBI was not statistically inferior to chromoendoscopy in sensitivity (p=0.67); but significantly superior in specificity and overall accuracy (p=0.01).³¹ A recent meta-analysis of 18 studies including more than 1,900 patients also found that NBI had a significantly improved specificity compared to chromoendoscopy, with no significant difference in sensitivity.³²

Fujifilm Blue Laser Imaging (BLI) (LASEREO; Fujifilm Co., Tokyo, Japan) utilizes two lasers in addition to white light to sharpen image quality.³⁷ A small retrospective study showed that endoscopists preferred BLI for detection of ESCC³⁸ but no objective evidence exists in its application as a screening tool for ESCC. Furthermore, the interobserver variability of narrow band imaging for esophageal pathology is not clear, but promising results have been reported for upper aerodigestive tract lesions.³⁹

Post Processing Imaging technology (i-SCAN, FICE)

Several software driven advances have allowed for increased mucosal sharpness, clarity, and image processing in real-time. The i-SCAN (Pentax, Tokyo, Japan) technology utilizes surface, contrast, and tone enhancement.⁴⁰ FICE (Fuji Intelligent Chromoendoscopy) (Fujinon, Fujifilm Medical Co, [Saitama, Japan]) enhances image quality through the use of spectral estimation technology.⁴¹

Using Lugol’s and FICE, Li et al ⁴⁴ examined 257 patients with suspicious esophageal lesions (27 ESCC, 22 HGIN, 40 LGIN, and 12 normal esophagus). Compared to Lugol’s, which detected ESCC, HGIN, LGIN with a sensitivity of 88.9%, 77.3%, and 77.5% respectively, FICE’s sensitivity was 92.6%, 86.4%, 87.5%, respectively. However, these improvements were not statistically significant (p=0.64).

When combined with transnasal endoscopy (TNE) in a study of 99 patients with history of head and neck cancer, FICE was found to be more sensitive than TNE with WLE alone for detection of superficial ESCC (100% vs. 25%, respectively), with comparable specificities (96.8% vs. 97.8%).⁴²

Transnasal endoscopy

Ultrathin transnasal endoscopy (TNE) allows for visualization of the esophagus without sedation, abrogating the risks of sedation and reducing costs. Huang, et al., found TNE to be well-tolerated in over 98% of patients.⁴³ Wang, et al., demonstrated a reasonable safety profile with only 2 of 441 patients undergoing TNE having complications treated conservatively.⁴⁴ When combined with FICE, Arantes, et al., observed sensitivities and specificities of over 90% for detecting ESCC.⁴⁵ Further study with this modality in the identification of esophageal squamous dysplasia is required, but without the risks and costs of sedation, TNE could become highly utilized in endemic areas.

Despite extensive data on patient acceptability, TNE is not widely used. This may reflect lack of physician training or experience, and perhaps lack of patient and physician preference for unsedated transnasal procedures.

Endocytoscopy

The ability to combine a microscope with a flexible endoscope was introduced in the early 2000s.⁴⁶ Endocytoscopy can diagnose dysplastic lesions without biopsy, reducing the wait for pathology and the costs for another procedure. The endocytoscope (Olympus GIF-1T240, Olympus Medical Systems Corp. Tokyo, Japan) can be passed through the instrument channel of a standard endoscope, where a soft hood at the end of the endoscope allows for continuous contact with the mucosa (Figure 4).⁴¹ Methylene or toluene blue spraying is required for staining to enhance visualization of dysplastic cellular elements.

Figure 4:

Microendoscopic imaging of squamous dysplasia.

a. | Endocytoscope probe which can be passed through the instrument channel of an endoscope.

b. | The border between esophageal squamous cancer (upper) and normal squamous epithelium (lower). The density and nucleus:cytoplasm ratio is much higher in cancer compared to normal epithelium.

(Reprinted with permission from Kumagai Y, et al. Endoscopy 2004;36:590–594.)

An early study of 75 patients by Inoue, et al., found a diagnostic accuracy of 82% for distinguishing between malignant and non-malignant lesions of the esophagus.⁴⁷ Kumagai, et al., found that the need for histologic biopsy could be obviated in up to 93.5% of cases.⁴⁸ More recently, Kumagai reported a sensitivity and specificity of 100% and 80%, respectively, for the detection of ESCC using endocytoscopy.⁴⁹ Further research in this field includes the application of higher magnification endoscopes, in which preliminary results appear promising.^{50, 51}

However, while endocytoscopy may obviate the need for biopsies, this approach is more operator dependent, requiring experience to interpret the imaging and initiate appropriate treatment. No studies on ESCC screening have reported the interobserver variability of this technique. Furthermore, in order to highlight abnormalities, Lugol’s staining or NBI may be required, again increasing the costs. Further study is required to elucidate the role of endocystoscopy in the detection of esophageal squamous dysplasia.

High Resolution Microendoscopy

High resolution microendoscopy (HRME) utilizes a contact probe containing a fiber optic microendoscope to visualize the esophageal epithelium (Figure 5). Before insertion of the probe through the instrument channel of an endoscope, the mucosal area of interest (usually identified by Lugol’s or NBI) is sprayed with proflavine dye, which enables visualization of the superficial mucosal nuclei.⁵² Image analysis software is then used to interpret the images and identify dysplastic cellular features. This approach has been shown to be feasible for detection of ESCC, with sensitivities ranging from 84% to 93% and specificities ranging from 92% to 97%.^{53, 54}

Figure 5:

High-resolution microendoscope⁹²

Lugol’s unstained lesions (left) are imaged with HRME (“optical” biopsy) and compared with corresponding tissue biopsy (histopathologic biopsy) (original magnification, 100x). Only 1 of the 2 Lugol’s abnormal areas was neoplastic (upper panel) as determined by the imaging software, based on mean nuclear area, nuclear-to cytoplasmic ratio, nearest inter-nuclear distance, nuclear eccentricity, nuclear solidity, and the major axis of the ellipse best approximating each nucleus.

(Reprinted with permission from Protano et al. Gastroenterology 2015; 149: 321–329.)

HRME systems cost considerably less than other advanced screening platforms, making this modality more attractive for developing nations. A screening trial of 147 patients in the United States and China showed that the addition of HRME to Lugol’s significantly improved the specificity (88% vs 48%, p<0.001) and positive predictive value (45% vs 22%, p<0.0001) for detecting severe esophageal squamous dysplasia and ESCC, compared to Lugol’s alone. Fifty-five patients would have been spared from biopsy with HRME.⁵¹ Markov models incorporating both average and high risk populations in China have shown that HRME is more cost-effective compared to no screening, but results were mixed when compared to traditional screening methods.¹⁵

Image analysis advances that highlight abnormal nuclei or automatically calculate a score for the degree of dysplasia present based on an algorithm of microscopic features may improve accuracy by decreasing interobserver variation.⁵⁵ Furthermore, portable tablet systems have now been developed, potentially allowing for the use of HRME in remote regions.⁵⁶

Video Capsule Endoscopy

Video capsule endoscopy has emerged as a non-invasive method to image the gastrointestinal tract. Accordingly, there has been interest in using this as a screening method for ESCC. However, an initial study comparing the use of video capsule endoscopy to Lugol’s chromoendoscopy in a high-risk population of patients with head and neck cancer found poor sensitivity and specificity in detecting neoplastic lesions in the esophagus.⁵⁷

Non-endoscopic Screening Methods

The community-assignment study by Wei et al shows that endoscopic screening with appropriate treatment of esophageal squamous dysplasia and early ESCC can reduce ESCC mortality in a very high-risk population in China.¹¹ However, widespread endoscopic screening is not possible or cost-effective in most high-risk areas in Asia and Africa. Table 3 shows the esophageal cancer incidence rates in 2012 from high-quality population-based cancer registries in ESCC endemic areas along with proportions of esophageal squamous dysplasia and early ESCC diagnosed on screening endoscopies of asymptomatic adults in three of these sites. The Wei study was done in Cixian, China, where ESCC incidence was ~150cases/100,000 population/year, and found HGD in 11 of every 100 endoscopies.⁵⁸ In such a site, primary endoscopic screening of the population may well be indicated and cost-effective. But in most high-risk populations, with ESCC incidence rates of ~20–40/100,000/year, only 1%–3% prevalence of HGD, and less developed health infrastructures, primary endoscopic screening of the population will not be possible or cost-effective. To reduce ESCC mortality in these populations, simple, inexpensive, and minimally-invasive tests need to be developed that are acceptable to unsedated asymptomatic adults and can specifically triage those with moderate to HGD to endoscopy for definitive diagnosis and treatment. Table 4 and Table 5 summarize the characteristics and current status of non-endoscopic screening methods for ESCC.

Table 3:

Esophageal Cancer Incidence and the Prevalence of High-Grade Dysplasia in Endoscopic Screening Studies from Selected Sites in Asia and Africa^{11, 28, 64, 93}

	EC incidence		HGD Prevalence
	Males	Females
Cixian, China	193	109	11%
Yanting, China	101	68
Golestan, Iran	23	19	1%
Nairobi, Kenya	21	15	3%
Blantyre, Malawi	38	23
Eastern Cape Province, RSA	32	20

^*Esophageal cancer incidence/100,000 population/year from population-based cancer registries in Cancer Incidence in Five Continents, Vol X, IARC;

^§Wei et al¹¹;

^ɣRoshandel et al⁶⁸;

^ƩMwachiro et al.²⁸

Table 4:

Non-endoscopic screening methods and characteristics

Screening Method	Author (year)	Study Size (N)	Dysplasia and ESCC (N)	Assay method	Sensitivity	Specificity
Direct Sampling Techniques
Inflatable Balloon	Roth (1997)63	439	123 (28%) esophageal squamous dysplasia 16 (4%) ESCC	Cytology only	47% esophageal squamous dysplasia+	81% esophageal squamous dysplasia+
Inflatable Balloon	Pan (2008) 62	725	232 (32%) esophageal squamous dysplasia +	Cytology only	46% esophageal squamous dysplasia+	84% esophageal squamous dysplasia+
Mechanical Balloon	Pan (2008)62	725	232 (32%) esophageal squamous dysplasia+	Cytology only	39% esophageal squamous dysplasia+	85% esophageal squamous dysplasia+
Sponge	Roth (1997)63	439	123 (28%) esophageal squamous dysplasia 16 (4%) ESCC	Cytology only	24% esophageal squamous dysplasia+	92% esophageal squamous dysplasia+
Inflatable Balloon	Adams (2008)58	147	72 esophageal squamous dysplasia (25 mild, 26 moderate, 21 severe)	Cytology + Panel of Methylated DNA markers*	50% HGD+	28% HGD+
Sponge	Roshandel (2014)64	301	18 esophageal squamous dysplasia, (4 with HGD)	Cytology + p53	22% esophageal squamous dysplasia+_ 100% HGD+	97% HGD+
Breath Markers
Volatile Organic Compounds	Zou (2016)68	86	29 esophageal cancers (not specified)	7 ions analyzed on a home-made breath mass spectrometer—PTR-MS apparatus (Ion Sniffer 2020Q).	86.2% for Esophageal cancer	89.5% for Esophageal cancer
Blood Markers
Autoantibodies	Zhou (2014)94	567	65 esophageal squamous dysplasia 88 ESCC	6 marker panel^	64% (for ESCC) 38% for esophageal squamous dysplasia+	94% (for ESCC)
Autoantibodies	Xu (2014)71	513	237 ESCC	6 marker panel~	45% for ESCC	95% for ESCC
miRNA	Zhang 201095	430	290 ESCC	7 miRNA panel#	80% for ESCC	90% for ESCC
Circulating tumor cells and mRNA	Cao (2009)96	182	108 ESCC	Survivin mRNA expressing circulating tumor cells	47.2% for ESCC	100% for ESCC

^*AHRR, p16INK4a, MT1G, and CLDN3

^{^}P53, IMP1, P16, cyclin B1, P62, and C-myc

^~p53, NY-ESO-1, MMP-7, Hsp70, Prx VI, and BMI-1

^#miR-10a, miR-22, miR-100, miR-148b, miR-223, miR-133a, and miR-127–3p

Table 5:

Pros and Cons of Non-endoscopic Screening Methods

	Pros	Cons
Esophageal mucosal sampling devices (brushes, balloons, sponges)	• Minimally invasive direct tissue sampling •Can be combined with any molecular test • Low operating cost • High specificity	• Cytology-only yields low sensitivity • Devices are hard to swallow and may lead to poor compliance
Auto-antibodies (blood)	• Non-invasive (serum) • Stable assay	• Low sensitivity
Circulating tumor cells (blood)	• Non-invasive • High specificity • Prognostic value	• Low sensitivity • Not found in early disease
Circulating miRNA (blood	• Non-invasive (detected from plasma or serum) • Stable and consistently expressed • High sensitivity and specificity	•Limited number of studies
Methylated DNA (cell sample or blood)	• Can be non-invasively detected in plasma • Stable and consistently expressed	• Limited number of studies
Volatile Organic Compounds (breath)	• Non-invasive	• Limited number of studies

Esophageal cytology samples

Non-endoscopic esophageal sampling devices have been studied extensively as inexpensive, tools for ESCC screening. The unsedated patient swallows a brush, a deflated balloon, or a sponge in a gelatin capsule attached to a cannula or string that is tethered outside the mouth. After it reaches the stomach, the balloon is inflated or the gelatin capsule dissolves and the sponge expands, and then the cell collection device is pulled up the esophagus and out, collecting superficial squamous cells from the esophagus. Traditionally, the recovered material is smeared on slides for Papanicolaou-stained cytologic examination. Figure 6 illustrates some of the common balloons and sponges used in China and Japan.

Figure 6:

Balloons and sponges used in prior studies for screening of esophageal squamous cell carcinoma and/or dysplasia.

(Reprinted with permission from Pan QJ, et al. Acta Cytol 2008;52:14–23.)

These methods are attractive as they can be performed without endoscopy or sedation, and can be applied at scale in limited resource settings.^59–61 Unfortunately, the results have been less than optimal. A study of more than 700 patients comparing balloon cytology to endoscopic biopsy reported sensitivities close to 40%.⁶² Similarly, cytology obtained with sponges from a study of 439 Chinese patients also produced a disappointing sensitivity of 24% for esophageal squamous dysplasia and ESCC.⁶³

Esophageal cytology specimens combined with biomarkers

Combining molecular biomarkers to esophageal cytology obtained non-endoscopically may improve the sensitivity of the test. In a study of 147 Chinese patients with esophageal biopsies ranging from normal to severe esophageal squamous dysplasia,⁸⁹ a 4-marker panel of differentially methylated promoter regions in AHRR, p16INK4a, MT1G, and CLDN3, assayed from cells collected with an inflatable balloon, was able to detect HGD with a sensitivity and specificity of 50% and 68%, respectively.

In another study, 301 Iranian adults were examined by a swallowed sponge (Cytosponge, Medtronic, [Minneapolis, USA]), followed by Lugol’s chromoendoscopy. Combined analyses of cytology and p53 staining showed a sensitivity of 100% at 97% specificity for detecting four patients with HGD, although sensitivity was much lower for any grade of dysplasia (64% by morphology, 11% by p53).⁶⁴ Despite the small number of cases, this study shows promise for the Cytosponge as a safe and feasible tool for diagnosing esophageal squamous dysplasia, and should be evaluated with other biomarkers in larger studies.

Breath markers

Volatile organic compounds

The measurement of volatile organic compounds (VOCs) from breath is completely non-invasive. Human breath is a complex biological sample containing more than 250 VOCs.⁶⁵ The potential role of VOCs in detecting cancer was creatively demonstrated by Sonoda et al., who showed that canine olfaction of VOCs found in exhaled breath and stool samples was accurate in discriminating patients with colorectal cancer from normal controls (sensitivity 0.91 and specificity 0.99).⁶⁶ Since this study, many breath mass spectrometry devices have been developed to replace the canine nose.⁶⁷ Although no studies specific to ESCC have been conducted, a pilot study from China on 29 patients with esophageal cancer and 57 healthy relatives reported an AUC of 0.943 with 7 ions analyzed on a home-made breath mass spectrometer—Ion Sniffer 2020Q (Hefei, China).⁶⁸ Since the study population was from China, where ESCC comprises over 90% of esophageal cancer cases, one could assume that most of the esophageal cancer cases were ESCC. However, these results should be interpreted with caution given the small sample size and lack of blinding.

Blood Markers

Autoantibodies

Auto-antibodies to tumor associated antigens (TAAs) have been proposed as biomarkers for cancer due to their stability in serum. The most comprehensively studied TAA is P53, in which a mutation results in a non-functioning protein that has a longer half-life than the native protein, and the subsequent anti-P53 can be detected non-invasively in serum. However, a meta-analysis of 15 studies reported a disappointing sensitivity range of 15%–60% at specificity 91–100% for anti-P53 as a single marker.⁶⁹ A 6-marker panel of auto-antibodies in 237 late-stage ESCCs versus 134 normal did not do much better, with a sensitivity of 45% at a specificity of 95%.⁷⁰ Evidence is even more limited for esophageal squamous dysplasia. Therefore, autoantibodies for screening purposes of esophageal cancer are still far from clinical application.

Circulating tumor cells

Circulating tumor cells (CTCs) assayed from blood have been investigated for their prognostic use in esophageal cancer. The presence of CTCs has been found to correlate strongly with nodal metastases, and poor progression-free survival in patients with esophageal cancer.^71–73 Since the presence of CTCs portends advanced disease, CTCs would not be an appropriate screening test for early cancer.

Circulating miRNA

MicroRNAs (miRNAs) are small noncoding RNAs that bind to target messenger RNAs (mRNAs) resulting in RNA degradation and/or post-translational inhibition. MiRNAs are stable, abundantly expressed, and can be consistently detected in blood. In a meta-analysis of 27 studies, the overall sensitivity and specificity of miRNAs for detecting ESCC were 79.9% (95% CI: 76.2%–83.1%) and81.3% (95% CI: 75.7–85.9), respectively, with an AUC of 0.87 (95% CI: 0.84–0.90).⁷⁴ However, only four of the 19 miRNAs were studied in more than one of the publications. Other meta-analyses were limited by small numbers of studies and lack of validation.⁷⁵ Nevertheless, the concept of miRNAs does hold potential as a non-invasive screening tool for ESCC, and further studies are underway.^76–78

Methylated DNA markers

Oncogenesis by way of modifications in DNA methylation occurs via two fundamental changes: (1) hypermethylation of CpG islands in gene promoters, which can silence tumor suppressor genes; and (2) hypomethylation of repetitive genetic elements, which may lead to genomic instability or oncogene activation.⁷⁹ Methylated DNA markers (MDMs) have been shown to be broadly informative markers of neoplasia⁸² and are a critical component of a multi-target stool DNA test, FDA-approved for average risk colorectal cancer screening.⁸³

MDMs have been utilized to develop minimally invasive diagnostic tools for BE and BE related dysplasia. The identification, validation and pilot testing of MDMs for BE diagnosis has been recently described. Following agnostic discovery by reduced representation bisulfite sampling (RRBS) and validation in independent biopsy and whole esophageal brushing cohorts⁸⁴, the utility of these markers was confirmed in a pilot randomized trial. In 19 BE cases and 20 controls, esophageal cytology specimens obtained via a 25 mm capsule sponge device (Sponge On a String (SOS) (EsophaCap) (Capnostics, New Jersey, USA), coupled with a 2 marker MDM panel detected all 19 cases of BE (100% sensitivity at 100% specificity).⁸⁵ In a subsequent ongoing validation study (SOS 2 trial) including 90 BE and 58 controls, a two-marker MDM panel assayed from SOS samples accurately distinguished BE cases from controls with an AUC of 0.97 (95% CI 0.94–0.99).⁸⁶ A three marker panel of MDMs for the detection of BE related dysplasia has also been previously described.⁸⁷ MDMs have also been applied to the Cytosponge as mentioned previously, with AUCs as high as 0.877 (95% CI 0.84–0.91) in a pilot study with 20 BE cases and 10 normal controls, and a validation cohort with 149 BE cases and 129 normal controls.⁸⁸ In samples from non-endoscopic balloons from 86 individuals, tests of CCNA1 and VIM DNA methylation detected BE metaplasia with 90.3% sensitivity and 91.7% specificity in another recently published study.⁸⁴

Along these same lines, MDMs for ESCC and squamous dysplasia, partnered with a device such as SOS, can lead to the development of a cost-effective, minimally-invasive, and highly accurate way to screen for this deadly disease. In a prior discovery study, we reported MDMs with high discrimination for the detection of ESCC.⁸⁹ The 15 best-performing MDMs were then chosen for a multi-national validation study with tissue from 86 ESCC cases from the US, 114 from Iran, and 133 from China, with similar numbers of country-specific normal controls. Although the subjects came from three distinct populations with varying risks of ESCC, our candidate MDMs performed well at discriminating ESCC tissue from normal esophageal tissue across all sites. Using a marker panel of 2–4 MDMs, the AUCs were 0.997 (95% CI 0.99–1.0), 0.99 (95% CI 0.98–1.0), and 0.97 (95% CI 0.93–0.99), for US, Iran, and China respectively, corresponding to sensitivities and specificities >95%.⁹⁰ These markers need to be tested on esophageal cytology specimens obtained via a capsule sponge sampling device. A report from China applied MDMs to esophageal balloon cytology samples from 147 patients with endoscopic biopsy diagnoses ranging from normal to severe squamous dysplasia in an endemic area, and a 4-marker panel (AHRR, p16INK4a, MT1G, and CLDN3) was able to detect severe squamous dysplasia with a sensitivity and specificity of 50% and 68%.⁵⁸

Not only can ESCC-specific MDMs be detected in esophageal tissue, we have also demonstrated its feasibility as a plasma-based biomarker for esophageal cancer in a recent pilot study (ref DDW abstract).⁹¹ Using MDMs discovered through the aforementioned whole methylome discovery effort, 12 candidate MDMs were chosen for a Phase I study with 85 cases (76 EAC and 9 ESCC) and 98 controls. At a specificity of 91%, a five-marker panel assayed from plasma detected 74% of esophageal cancer overall (74% of EAC, and 78% of ESCC), with an overall AUC of 0.93 (95% CI 0.89–0.96). The test was more sensitive for higher-stage disease, but was still able to detect Stage I and Stage II disease at sensitivities of 56% and 83%, respectively. Although only 9 cases of ESCC were included in this study, this preliminary research has established a potential for using MDMs for detection of ESCC in an entirely non-invasive way. With ongoing larger-studies and optimized assays, active research in this area may transform the future of screening for ESCC especially in resource-poor endemic areas.

Future Directions

ESCC is a lethal disease largely due to its usually advanced stage at diagnosis. Screening programs targeting high-risk populations are needed to detect curable dysplasia and early stage ESCC. Critical advances have been made in high quality endoscopic screening modalities. However, the majority of patients afflicted by ESCC are found in resource-poor regions. For screening to occur successfully at a population level in these regions, minimally-invasive tools that are accurate and cost effective need to be developed.

The use of molecular biomarkers in esophageal cell samples acquired non-endoscopically (via a sampling device such as a sponge) or in blood holds great potential in the future for screening in populations at high risk for ESCC. However, the performance of such biomarkers is yet to be optimized, and results need to be validated in large studies. Additionally, further risk-stratification of individuals needs to be achieved in areas with lower incidence of ESCC and esophageal squamous dysplasia. Such risk-stratification models combined with a minimally-invasive test could transform screening and reduce mortality from ESCC.

Acronyms:

ESCC

esophageal squamous cell carcinoma

HGD

high grade dysplasia

EAC

esophageal adenocarcinoma

BE

Barrett’s esophagus

NBI

narrow band imaging

TNE

transnasal endoscopy

WLE

white light endoscopy

FICE

Fuji Intelligent Chromoendoscopy

BLI

Blue Laser Imaging

HRME

High Resolution Microendoscopy

HGIN

High Grade Intraepithelial Neoplasia

LGIN

Low Grade Intraepithelial Neoplasia