Radiofrequency ablation: time to think of poor responders

Abstract

Since its first description in 2004 by Ganz et al. [1], radiofrequency ablation (RFA) has become the standard treatment of flat dysplastic Barrett’s esophagus (BE) after endoscopic resection of all visible lesions. Compared with alternative eradication methods, such as stepwise endoscopic resection, (hybrid) argon plasma coagulation, and cryoablation, RFA is the best validated and most used ablation method that is both effective and relatively safe, and is unanimously recommended by various guidelines. “Moving forward, the optimal management strategy for patients without, after successful or after failed RFA will require individual clinical risk-stratification tools that balance the risks of the procedure, progression, and death related to esophageal cancer against the patient’s age and life expectancy, comorbidities, and the risk of death from other diseases.” RFA achieved complete esophageal resurfacing to normal squamous epithelium after a median of three sessions in 94 % of patients in the Dutch Barrett’s registry [2]. In the Netherlands, the management of patients with early neoplastic BE is uniquely organized, with care centralized in high-volume Barrett’s Expert Centers that follow an agreed treatment protocol, and a data registry maintained since 2008. Studies from other groups have reported lower complete remission rates of BE after RFA. Hence, consensus on a suggested minimum complete ablation rate of BE as a quality parameter remains difficult. The UK registry demonstrated an improvement in the RFA results over time as experience in the method increases [3]. Following their previous RFA outcome study, van Munster et al. [4] present results of an observational study, again based on data from the Dutch Barrett’s Expert Center registry, in this issue of Endoscopy. The study provides a detailed analysis of the characteristics and outcomes of patients with neoplastic BE who had a poor response to RFA therapy. From this registry, 1386 patients who underwent at least one RFA session for treatment of Barrett’s neoplasia were included (73 % with high grade dysplasia or early cancer). A total of 134 patients (10 %) were poor healers but all achieved complete esophageal healing after additional time and intensified reflux therapy with increased doses of proton pump inhibitors. Among the poor healers, 66 % finally achieved complete eradication of BE (CE-BE); however, stricture development was significantly higher compared with patients with normal healing (34 % vs 14 %; P < 0.01). After complete mucosal healing, half of the initially poor healers had achieved more than 50 % squamous regeneration after the first RFA, and all but two of these patients eventually achieved excellent RFA results (97 % CE-BE). Conversely, 74 of the 1386 patients (5 %) had poor squamous regeneration after the first RFA session and 47 of these patients (64 %) did not achieve CE-BE. The residual Barrett’s burden probably explains the observed higher rate of progression to advanced disease in this group of patients with poor squamous regeneration after RFA. The endoscopic definitions of poor healing (active inflammatory changes with mucosal swelling and exudates and/or presence of ulcerations ≥ 3 months after RFA) and poor squamous regeneration (< 50 % BE regression) present challenges in clinical practice as these features are difficult to estimate when unusually shaped Barrett’s areas and large islands remain after RFA. Regenerative and inflammatory alterations after RFA are often hard to distinguish from neoplastic changes. The endoscopic assessment of inflammation and the quantification of squamous regeneration is very operator dependent and subject to bias. The Prague classification that reports the extent of BE by estimating the circumferential and maximal length is often not helpful or not applicable after RFA. Artificial intelligence-based methods promise to reliably quantify the residual area of BE after RFA in three-dimensional reconstruction of the esophageal mucosa [5]. Smart algorithms that support the endoscopist in recognition of inflammation and neoplasia based on the endoscopic appearance and morphology are under clinical evaluation [6]. Current guidelines remain silent on how to manage patients with residual BE despite several RFA attempts. Patients with poor squamous regeneration are mainly also those with poor mucosal healing (91 % in the van Munster study). A pragmatic approach to a poor healing response would be to postpone RFA for at least 6 weeks in order to check the patient’s compliance and maximize the antireflux medication including proton pump inhibitors and additional H2 antagonists. To avoid poor healing, it seems mandatory to control the reflux before embarking on endoscopic treatment. Fortunately, time is on our side in cases of poor response without dysplasia and without worrying endoscopic features. Visible lesions that emerge/resurface after RFA can be treated by endoscopic resection and should not be ablated. Endoscopic surveillance with the option of endoscopic resection seems a reasonable option, at least in frail patients or in those with significant comorbidities and limited life expectancy. In this and other studies, certain risk factors have been identified to predict poor response to RFA: higher body mass index, longer BE length, reflux esophagitis, genetic biomarkers, and poor squamous regeneration after first endoscopic treatment. Most recently, the thickness of the Barrett’s epithelium measured by volumetric laser endomicroscopy has also emerged as a predictor of RFA response. An increase in the mean Barrett’s epithelium thickness of 100 μm resulted in a 12 % lower response to RFA when this was assessed as percentage reduction in Prague score [7]. Whether adaptation of the ablation energy to the thickness of the Barrett’s mucosa would result in higher rates of complete...