Clinical utility of wide-area transepithelial sampling with three-dimensional computer-assisted analysis (WATS^3D) in identifying Barrett’s esophagus and associated neoplasia

Summary

Sampling error during screening and surveillance endoscopy is a well-recognized problem. Wide-area transepithelial sampling with three-dimensional computer-assisted analysis (WATS^3D), used adjunctively to forceps biopsy (FB), has been shown to increase the detection of Barrett’s esophagus (BE) and BE-associated neoplasia. We evaluated the clinical utility of WATS^3D and its impact on the management of patients with BE and dysplasia. Between 2013 and 2018, 432 consecutive patients who had a WATS^3D positive and an accompanying FB negative result were identified. Physicians were contacted to determine if the WATS^3D result impacted their decision to enroll patients in surveillance or increase the frequency of surveillance, recommend ablation, and/or initiate or increase the dose of proton pump inhibitors (PPIs). WATS^3D directly impacted the management of 97.8% of 317 BE patients; 96.2% were enrolled in surveillance and 60.2% were started on PPIs or their dose was increased. WATS^3D impacted the management of 94.9% and 94.1% of the 98 low-grade dysplasia and 17 high-grade dysplasia patients, respectively. As a result of WATS^3D, 33.7% of low-grade dysplasia and 70.6% of high-grade dysplasia patients underwent endoscopic therapy. More than 37% of all dysplasia patients were enrolled in a surveillance program, and nearly 30% were scheduled to be surveilled more frequently. PPIs were either initiated, or the dose was increased in more than 54% of all dysplasia patients. We demonstrate that WATS^3D has high clinical utility. By prompting physicians to change their clinical management in patients with negative FB results, WATS^3D, used adjunctively to FB, directly impacts patient management, and improves patient outcomes.

INTRODUCTION

Barrett’s esophagus (BE), characterized by the replacement of the normal stratified squamous esophageal epithelium with specialized columnar epithelium (intestinal metaplasia with goblet cells), predisposes to esophageal adenocarcinoma (EAC), a tumor whose incidence has risen six-fold in the last three decades.^1,2 Advanced EAC has a poor prognosis, with a 5-year survival of only 3%. However, if detected early, EAC can be endoscopically resected and potentially cured.³ Furthermore, if dysplasia is detected in BE, endoscopic eradication therapy can prevent progression to cancer.⁴ For this reason, screening and surveillance protocols have been designed to detect BE-associated dysplasia and EAC at its earliest stage.

Screening and surveillance guidelines for BE and esophageal dysplasia (ED) require meticulous endoscopic examination using high-resolution endoscopes combined with random four-quadrant forceps biopsies (FBs) obtained at 1–2-cm intervals of columnar mucosa (the Seattle protocol). In addition, visible suspicious areas, such as nodules, masses, and ulcerations, which are more likely to be associated with dysplasia, are sampled separately. Unfortunately, this protocol is prone to a significant degree of sampling error, because it leaves more than 96% of the endoscopically apparent area of BE unsampled.^5,6 Since dysplasia and early cancers are usually highly focal and sometimes macroscopically indistinguishable from metaplastic tissue, random FB-based tissue sampling alone often misses abnormalities located between sampled areas, which results in false-negative results. There are also significant pathologic limitations and diagnostic variability in assessing the presence and grade of dysplasia among pathologists.⁷ This can lead to improper diagnosis and inappropriate treatment of patients in a significant proportion of cases.

Wide-area transepithelial sampling with three-dimensional computer-assisted analysis ([WATS^3D]; CDx Diagnostics, Suffern, NY), used as an adjunct to the Seattle protocol in patients undergoing screening or surveillance of BE, has been recently endorsed by the Standards of Practice Committee of the American Society for Gastrointestinal Endoscopy in the most current edition of the ‘ASGE Guideline on Screening and Surveillance of Barrett’s Esophagus’.⁸ WATS^3D utilizes an abrasive endoscopic brush, which enables sampling of a larger circumferential mucosal area of the esophagus in order to obtain abundant cells from the deep layers of the epithelium (deep crypts), which is precisely where dysplasia develops initially. In addition to individual cells, WATS^3D specimens consist mainly of tissue fragments characterized by compact sheets of cells, large cohesive clusters of cells, and even entire crypts that resemble ‘microbiopsies’, while preserving the ‘three-dimensionality’ of the tissue. The WATS^3D specimen is then analyzed by pathologists with assistance from a specialized three-dimensional computer analysis system that uses neural networks and artificial intelligence to identify abnormal cells for presentation to the pathologist for analysis. Images identified by the WATS^3D computer are reviewed by the pathologist in conjunction with manual microscopy. Pathologists utilize published and commonly accepted pathologic criteria that are used for evaluating FB specimens acquired by the Seattle protocol.

The clinical validity of WATS^3D as an adjunct to the Seattle protocol has been tested and confirmed in five prospective published studies that have all shown a significantly increased rate of detection of BE, BE-associated dysplasia, and EAC in patients undergoing either gastrointestinal esophageal reflux screening or BE surveillance.^9–13 For example, in the largest study to date using WATS^3D, 58 community-based endoscopists at 21 sites enrolled 12,899 patients undergoing screening and surveillance and demonstrated that WATS^3D markedly improved the detection of dysplasia by 242% (95% CI: 191–315) and BE by 153% (95% CI: 144–162).¹³ In a multicenter prospective study by Vennagalanti et al. conducted at 16 major academic centers nationwide involving 160 patients with BE under surveillance, WATS^3D detected an additional 23 cases of high-grade dysplasia (HGD)/EAC missed by the Seattle protocol.¹² In this study, WATS^3D was four times more effective at detecting HGD and EAC compared with the Seattle protocol, which utilizes only FB. All of these studies suggest that sampling error, an inherent limitation associated with current standard screening and surveillance techniques, can be significantly reduced with WATS^3D sampling.

The analytic validity of WATS^3D was confirmed in a study, which demonstrated that pathologists analyzing these specimens with the aid of the WATS^3D computer system established an accurate diagnosis of dysplasia with low interobserver variability.¹⁴ In that study, substantial agreement was noted among pathologists for the diagnosis of low-grade dysplasia (LGD), HGD, and BE without dysplasia, with an overall kappa value 0.86 (95% CI 0.75–0.97). WATS^3D results were demonstrated to be reproducible, in contrast to previously well-documented poor agreement noted among pathologists when assessing the presence and grade of dysplasia in FB specimens, in which kappa values typically range from 0.2 to 0.5.^7,15

Despite the previously proven high diagnostic accuracy of WATS^3D, current financial constraints imposed on health care providers attempting to deliver quality care by utilizing new diagnostic tests led us to evaluate the usefulness and benefits of WATS^3D in routine clinical practice. We specifically investigated the clinical utility of WATS^3D as an adjunct to the Seattle protocol, and its direct impact on patient management and health care outcomes (such as maintaining health or preventing death) in patients when BE or dysplasia was detected in WATS^3D specimens, but not in the same patient’s concurrent FBs.

METHODS

Between 2013 and 2018, we identified 432 consecutive patients from WATS^3D clinical registries, which included both community-based and academic sites. Men and women ages 18 years and older undergoing screening for suspected BE as well as those with known BE undergoing surveillance for dysplasia were included. Patients under surveillance had no prior history of ED or endoscopic eradication therapy. All patients underwent sampling of the BE mucosa according to local clinical practice with WATS^3D performed either before or after FB (Seattle protocol). Although investigators in the clinical registries were instructed to use both WATS^3D and FB to sample suspected BE only in patients displaying salmon-colored mucosa in the tubular esophagus, they were not monitored, and the number and site of biopsies taken at each endoscopy were left to the discretion of the endoscopist. For our study, we included only patients who had a WATS^3D positive diagnosis of either BE (demonstrating intestinal metaplasia) or dysplasia and a concurrent session negative FB result for those specific diagnoses (Figs 1 and 2).

Fig. 1

Barrett’s esophagus with low-grade dysplasia showing cells with enlarged hyperchromatic overlapping nuclei and disorganization. The nuclear chromatin is dense and clumped.

Fig. 2

Barrett’s esophagus with high-grade dysplasia showing markedly enlarged and variable sized cells with high N/C ratio, nuclear irregularity, hyperchromicity, multiple enlarged and irregular nucleoli, and increased mitoses. In high-grade dysplasia, cells are less cohesive so that at the periphery of cell clusters, cells can be seen disconnected from the main cell aggregate.

Of the 432 patients who met these criteria, 317 were diagnosed with non-dysplastic BE, 98 with LGD, and 17 with HGD by WATS^3D only (i.e. FB negative). After we obtained institutional review board approval, physicians in theWATS^3D clinical registries who performed the initial WATS^3D test were contacted and asked to complete an extensive survey aimed at elucidating what patient management action(s) resulted from these results. Physicians were never provided with any recommendations about how they should manage their patients when WATS^3D identified either BE or dysplasia and the concurrent FB were negative.

Data collected ere used to determine if the positive WATS^3D diagnosis resulted in a change in patient management and/or an improvement in patient outcome. Specifically, physicians were asked whether the positive WATS^3D test resulted in: (i) enrolling that patient in an endoscopic surveillance program (i.e. a screening patient with a new diagnosis of BE) or increasing that patient’s frequency of surveillance visits (i.e. a surveillance patient with a new diagnosis of dysplasia, (ii) having that patient undergo endoscopic eradication therapy (ablation and/or endoscopic mucosal resection [EMR] for patients with a new diagnosis of dysplasia) or any other therapeutic endoscopic intervention, and (iii) initiating therapy with proton pump inhibitors (PPIs) in that patient or increasing that patient’s dose of PPI. The results of all WATS^3D and FBs obtained during follow-up surveillance endoscopies were also collected in order to determine disease progression in these patients. Demographic, endoscopic, and pathology data were aggregated and de-identified before analysis.

RESULTS

There were a total of 432 patients included in the study, 317 with BE, and 115 with dysplasia diagnosed by WATS^3D alone. Their demographic features are summarized in Table 1.

Table 1

Cohort demographics

	BE	LGD	HGD	Total
Total number of patients	317	98	17	432
Mean age (range)	60.5 (18–91)	69.1 (38–90)	66.9 (48–86)	62.7 (18–91)
Mean length of suspected Barrett’s segment (cm)	1.5	4.1	4.5	2.3
Gender
Male (%)	42	67	82	49
Female (%)	58	33	18	51
Ethnicity
White (%)	79	97	82	83
Black/African American (%)	2	0	6	2
Hispanic/Latino (%)	7	1	6	6
Asian (%)	7	1	6	6
Other (%)	5	1	0	4

BE, Barrett’s esophagus; HGD, high-grade dysplasia; LGD, low-grade dysplasia.

The clinical impact of WATS^3D positive-FB negative diagnoses for patients with BE, LGD, or HGD is summarized in Table 2.

Table 2

Direct impact of WATS^3D on the management of patients with be, LGD, and HGD

Action	BE (n = 317)	LGD (n = 98)	HGD (n = 17)	Total all BE and Dysplasia (n = 432)
	n (%)	n (%)	n (%)	n (%)
Begin surveillance program	305 (96.2)	37 (37.8)	6 (35.3)	348 (80.6)
Increased surveillance frequency of dysplasia patients	—	29 (29.6)	5 (29.4)	34 (29.6)
Ablation/Antireflux surgery Ablation/EMR	12 (3.7)	33 (33.7)	12 (70.6)	57 (13.2)
PPIs initiated	170 (53.6)	43 (43.9)	7 (41.2)	220 (50.9)
PPI dose increased	21 (6.65)	10 (10.2)	3 (17.6)	34 (7.9)
No change in patient management	7 (2.2)	5 (5.1)	1 (5.9)	13 (3.0)
Direct impact on patient management	310 (97.8)	93 (94.9)	16 (94.1)	419 (97.0)

BE, Barrett’s esophagus; EMR, endoscopic mucosal resection; HGD, high-grade dysplasia; LGD, low-grade dysplasia; PPI, proton pump inhibitor; WATS^3D, wide-area transepithelial sampling with three-dimensional computer-assisted analysis.

Barrett’s esophagus

Of the 317 patients diagnosed with BE, WATS^3D had a direct impact on the clinical management of 97.8% of patients. For example, 305 (96.2%) were enrolled in a surveillance program after the WATS^3D diagnosis, whereas 12 (3.7%) underwent either ablation or antireflux surgery. Although 94 (29.7%) patients were already on PPIs before the WATS^3D diagnosis, after the BE diagnoses were established by WATS^3D, 21 of those (6.6%) had their dose increased and an additional 170 patients (53.6%) were started on PPIs.

Dysplasia

WATS^3D directly impacted the management of 94.9% and 94.1% of all LGD and HGD patients, respectively. For instance, 33.7% of LGD and 70.6% of HGD patients underwent endoscopic ablation or EMR after the WATS^3D diagnosis. Additionally, more than 37% of all LGD and HGD patients were enrolled in a surveillance program, and nearly 30% of them, who were already enrolled in a surveillance program, were scheduled to undergo surveillance more frequently. PPIs were either prescribed to patients initially after the WATS^3D diagnosis, or the dose was increased in 54.1% of LGD and 58.8% of HGD patients, respectively. WATS^3D impacted patient management with dysplasia equally in a community-based setting versus an academic site.

Follow-up

Only the data from 149 BE patients and 32 dysplasia patients who did not undergo ablation, and who subsequently underwent follow-up endoscopy with FB and WATS^3D, was available at the time the study questionnaire was completed, whereas the remaining patients were scheduled for surveillance at future dates.

Of the 149 BE patients who did not undergo ablation, 6 were subsequently diagnosed with LGD by WATS^3D, all of whom were missed by FB again. Of the 28 LGD patients and the 4 HGD patients who did not undergo ablation and whose follow-up endoscopy data were available, 3 of the former developed HGD diagnosed by WATS^3D alone, and 1 of the latter developed EAC identified by WATS^3D and FB; all of the remaining dysplasia patients were kept under close surveillance.

DISCUSSION

The results of our study demonstrate that WATS^3D has very high clinical utility when used as an adjunct to the Seattle protocol and adds value to the decision-making and outcome of patients with BE and associated dysplasia in the real-world setting. In this study, physicians had a high degree of confidence in the WATS^3D diagnoses even when the concurrent FB was negative. This is substantiated by the fact that physicians changed some aspects of management for their patients in 97% of all cases. Significant management changes included invasive treatments such as ablation, EMR, and even antireflux surgery.

Current US professional gastrointestinal society guidelines dictate that if the initial screening endoscopic evaluation is negative for BE, repeating endoscopic evaluation for the presence of BE is not recommended.^16–18 In our study, 305 patients identified with BE by WATS^3D, but missed on FB, would have not undergone any additional endoscopic evaluation, and their disease would not have been detected and potentially allowed to progress to cancer. This is important since previously published studies have shown that patients with EAC identified during a BE surveillance program have markedly improved survival compared with those who have not undergone routine endoscopic surveillance because the cancers are detected at an earlier stage.^19,20

Furthermore, nodal involvement is far less likely to occur in patients undergoing surveillance compared with those who do not.²¹ For instance, in a study by Bhat and colleagues consisting of patients with EAC and a prior diagnosis of BE, survival was enhanced, and both tumor stage and tumor grade were lower, compared with patients who did not have a prior diagnosis of BE.²² Fortunately for the patients in our study, gastroenterologists opted to enroll greater than 96% of all patients who had a new diagnosis of BE made by WATS^3D alone in an endoscopic surveillance program. In fact, six of these patients ultimately developed LGD upon follow-up endoscopy.

A systematic review and meta-analysis of cohort studies have shown that patients with BE maintained on long-term PPI therapy have a 71% decreased risk of progression to HGD and/or EAC compared with those who did not receive acid-suppressive therapy.²³ The management of BE with PPIs for greater than 2–3 years provided greater benefit than using them for a shorter duration,²⁴ although this remains controversial.²⁵ In our study, greater than 60% of BE patients detected by WATS^3D were either started on PPIs or had their dose adjusted higher, invariably reducing their risk of progression to EAC as evidenced by prior published data. Without WATS^3D, and without a BE diagnosis, it is likely these patients would not be subjected to long-term PPI therapy.

For patients with confirmed LGD, current American College of Gastroenterology guidelines recommend aggressive antisecretory therapy for reflux disease with a PPI in order to decrease the histologic changes associated with regeneration or inflammation, combined with endoscopic therapy as the preferred treatment modality, or endoscopic surveillance every 12 months as an acceptable alternative.¹⁷

The American Gastroenterological Association guidelines recommend endoscopy after 8–12 weeks of twice-daily proton-pump inhibitor treatment, and if LGD is confirmed, patients should either undergo endoscopic eradication therapy or surveillance at 6-month intervals for 1 year and then annually.¹⁶ In our 98 patients with LGD diagnosed by WATS^3D, greater than 54% were either started on PPIs or had their dose increased. Furthermore, approximately 34% underwent endoscopic eradication therapy and greater than 67% were either enrolled in surveillance or had increased surveillance frequency. Since dysplasia remains the best clinically available marker of cancer risk in patients with BE, and the pooled annual incidence of EAC and/or HGD among patients with LGD is 1.73%,²⁶ the importance of WATS^3D directly impacting patient management of almost all of our patients with LGD and ultimately improving their health outcomes cannot be overstated. This is further evidenced by the fact that at follow-up endoscopies, which were scheduled entirely based upon the WATS^3D diagnosis, three patients with LGD developed HGD.

EMR and radiofrequency ablation have become the preferred method of treatment for most patients with HGD. Clinical guidelines recommend that visible lesions should be resected with EMR, whereas flat lesions (without raised/focal abnormalities) should be treated with endoscopic ablation.^17,18 The presence of HGD is also related to an increased risk of synchronous EAC.²⁷ In 17 of our patients with HGD, 16 had an initial diagnosis of non-dysplastic BE, and 1 was negative for BE on FB. As the reported rates of progression to EAC in patients with HGD is high and varies from 16% over 7 years²⁸ to over 50% in 5 years,^29,30 aggressive endoscopic treatment of 12 out of 17 HGD patients diagnosed by WATS^3D in this study, was potentially life-saving. If physicians had relied solely on the FB results of these patients, none of which detected any evidence of dysplasia, ablative treatment would never have been administered, and these patients would remain at high risk for developing esophageal cancer and a negative outcome.

Our study has several strengths including the large size of the cohort of patients with BE and associated dysplasia/neoplasia and the fact that it included both community-based practices and academic centers. The results of the utility of WATS^3D as an adjunct to the Seattle protocol are, therefore, the representative of the general population. Additionally, the patient demographics in our study are representative of the general at-risk population for BE and ED.

Our study also has some limitations. Since WATS^3D is a relatively new procedure, some follow-up surveillance visits, which were scheduled for patients with a new diagnosis of BE or dysplasia as a result of the WATS^3D diagnosis, had not yet been performed by the time physicians completed the study survey. Thus, data regarding the progression of disease from BE to dysplasia, LGD to HGD, and HGD to EAC are limited in our study. As a result, we suspect that the impact of WATS^3D on the clinical outcome of these patients would have been significantly higher than what we report in this study if we were able to collect the data from all follow-up surveillance visits.

In summary, WATS^3D provided physicians with a definitive diagnosis of BE, and/or associated dysplasia/neoplasia, when the accompanying concurrent FBs were negative, and the results had a direct impact on the management of these patients. Most importantly, WATS^3D had a direct impact on improving patient outcome. The utility of WATS^3D to influence treatment decisions was high as evidenced by the fact that physicians altered their management in 97% of all patients with BE, LGD, or HGD. WATS^3D has important diagnostic utility in evaluating patients undergoing gastrointestinal esophageal reflux screening and/or BE surveillance by impacting patient management and thereby potentially improving patient outcomes.

ACKNOWLEDGMENTS

We would like to acknowledge all of the physicians who participated in the study: Reddy V, 7 Hill Gastroenterology, Ocala, FL; Persich N, Alliance Surgery Center, Metairie, LA; Williams J, Athens Gastro Association, Athens, GA; Lister D, Baptist Hospital, Heber Springs, AR; Tran T, Baylor Scott and White Surgical Hospital, Sherman, TX; Rascon-Aguilar I, Bradenton Surgery Center, Bradenton, FL; Shim M, Center for Endoscopy, Oceanside, CA; Hellstern P, Citrus Endoscopy Center, Crystal River, FL; Feuer K, Citrus Surgical Center, Orlando, FL; Bailey D, Gislason G, Khan A, Prasad V, Pugh A, Rao S, and Rusche M, Digestive Care Center, Evansville, IN; Shah K, Digestive Disease Center, Laguna Hills, CA; Shelby S, Executive Surgery Center, Danville, CA; Gupta V, Florida Hospital FISH Memorial, Orange City, FL; Ariza S, Gastrodiagnostics, Orange, CA; Dugan V, Gastroenterology Group—Endocenter, Slidell, LA; Dodd K, Hudson Valley Gastroenterology, Kingston, NY; Edelstein M, La Peer Surgery Center, Beverly Hills, CA; Abousaif A, La Veta Surgical Center, Orange, CA; Noel J, Lafayette General Endoscopy Center, Lafayette, LA; Dierenfeldt W, Manhattan Surgical Hospital, Manhattan, KS; Bemanian S, Memorial Care Newport-Mesa, Costa Mesa, CA; Pothamsetty S, Millenia Surgery Center, Orlando, FL; Judah J, Navicent Health Heartburn Treatment Center, Macon, GA; Shaban A, Orange County Institute of GI and Endoscopy, Mission Viejo, CA; Yang R, Pacific Endoscopy Center, Pearl City, HI; McKinley M, ProHEALTH Care Associates, Lake Success, NY; Patel M, Simi Surgery Center, Simi Valley, NY; Shamsi SR, Specialty Surgery Center, Beverly Hills, CA; Kucera S and Kuperman D, Sun Coast Endoscopy Center of Sarasota, Sarasota, FL; and Nudell J, Tampa Bay Regional Surgery Center, Largo, FL; and Shultz R, University Surgical Center, Winter Park, FL.

Conflicts of interest

Conflicts of interest are as follows: 1. Vivek Kaul: None. 2. F Scott Corbett: Consultant, Speaker bureau. 3. Michael Smith: Consultant, Grant Support. 4. Anthony Infantalino: None. 5. Christina Tofani: None. 6. Seth Gross: None. 7. Zubair Malik: none.

Author contributions: Vivek Kaul, Seth Gross, F. Scott Corbett, Zubair Malik, Michael S. Smith, Christina Tofani, and Anthony Infantolino were involved in the concept, acquisition of data, analysis of data, drafting of the manuscript, and critical revision of the manuscript for important intellectual content.