Use of Cytosponge as a triaging tool to upper gastrointestinal endoscopy during the COVID-19 pandemic

During the COVID-19 pandemic, endoscopy services have been severely curtailed—eg, in England, UK, a 30% reduction of diagnostic endoscopies has been reported for the period between January and April, 2020, compared with the same period in 2019, with an estimated 750 oesophagogastric cancers going undiagnosed.¹ A delay in oesophageal cancer diagnosis could adversely affect outcomes, such as has previously been seen with low endoscopy referral rates being linked with poor outcomes from oesophageal cancer.²

The Cytosponge is a non-endoscopic diagnostic tool that was developed to detect Barrett's oesophagus in patients with reflux symptoms. Cytosponge consists of a tethered capsule that is swallowed in a primary or secondary care office setting and collects oesophageal cells, which can be assessed for morphology and immunohistochemical biomarkers of intestinal metaplasia (TFF3) and dysplasia (atypia and p53).3, 4 The safety, acceptability, and diagnostic accuracy of this approach has been assessed in three clinical trials, including the recent BEST3 trial.5, 6, 7 In light of COVID-19 restrictions, we assessed whether Cytosponge could triage patients referred for urgent investigation of alarm oesophageal symptoms.

Between April 8 and May 26, 2020, 123 patients were referred to our department at Cambridge University Hospital (Cambridge, UK) for urgent endoscopy, of whom 14 with dysphagia Mellow score of 3 or more received fast-track endoscopy, while 72 with mild symptoms and no dysphagia were managed via telephone. The remaining 37 patients were deemed eligible for Cytosponge, 21 of whom denied COVID-19 symptoms and accepted the test. Mean patient age was 59·9 years (SD 11·4; range 41–80). Most had dysphagia, with a median Mellow score of 1 (range 0–2; appendix pp 1–2). The nurse administering the Cytosponge reported an incomplete swallow on the basis of string tension on withdrawal (which was considered a high-risk feature) in four patients. Two of these four patients were found to have glandular atypia and p53 positive cells suggestive of dysplasia or cancer (appendix p 3). In the other two patients, an absence of glandular cells confirmed incomplete passage of the Cytosponge. Another four patients had evidence of cellular atypia suggesting a possible neoplastic process. All eight patients were referred for an urgent endoscopy and cancer was diagnosed in four of them. In the other four patients, endoscopy did not identify any concerning findings. Three patients had TFF3 positive cells suggestive of intestinal metaplasia (appendix p 3) and had evidence of peptic disease on endoscopy, with intestinal metaplasia on biopsies in two patients. The remaining ten patients had a normal Cytosponge result (appendix p 3) and were managed via telephone follow-up. Five of these ten patients received an endoscopy subsequently during the study period, which showed reassuring findings.

One patient had particularly interesting findings. A man, aged 54 years, with moderate dysphagia and weight loss had evidence of glandular atypia and aberrant p53 on the Cytosponge specimen. Endoscopy revealed a malignant stricture with tongues of dysplastic Barrett's oesophagus above it (appendix p 3). Thus, although the Cytosponge did not traverse the tumour, it was able to sample sufficiently to provide a diagnosis.

The Cytosponge is an example of an innovation that has undergone rigorous evaluation for diagnosing Barrett's oesophagus. The COVID-19 pandemic has offered the opportunity for more rapid adoption of this tool but the different patient group than originally intended merits careful audit. Although dysphagia was an exclusion criterion in previous trials for Cytosponge, it is safe in patients with eosinophilic oesophagitis.⁸ Our data suggest that Cytosponge is a potentially useful test to triage patients with mild-to-moderate dysphagia and other oesophageal symptoms, among whom the conversion rate to cancer at endoscopy is less than 4%.⁹ Atypical cells, aberrant p53 expression, and evidence of incomplete swallow were high-risk criteria for cancer. In our small series, all patients with cancers showed at least one high-risk feature.

Using the Cytosponge for this indication has several advantages when endoscopy provision is restricted. Cytosponge is an office-based procedure, which can be administered by a single non-medically qualified operator. Additionally, a low rate of aerosol generation is likely given that device removal only takes 3–5 s. Finally, if results are normal and subsequent symptom resolution occurs, testing can offer reassurance and enable prioritisation of endoscopy for those with a more urgent need. However, our findings have limitations. For instance, half the patients with negative Cytosponge results did not receive an endoscopy; therefore, we cannot exclude missed cancers. The device is not designed to detect gastric pathology, so attention is required on screening to elucidate the nature of the symptoms. Nonetheless, this study allowed us to develop an alternative clinical pathway to triage patients referred for urgent endoscopy (appendix p 4).