Initial Clinical Trials

Initial reports of clinical trials investigating RFA for the treatment of BE appeared in 2007, when Sharma and colleagues reported their findings from a prospective multicenter study titled Ablation of Intestinal Metaplasia (AIM-I) Trial. Seventy patients with NDBE measuring 2 to 6 cm were enrolled for circumferential balloon-based RFA. Initial studies like this were often limited to patients with BE measuring less than 8 cm because of concerns regarding pain control. RFA was applied at an energy density of 10 J/cm ² (as established during a prior dosimetry phase). A second treatment was applied at 4 months for persistent IM. At 12 months, and after a mean 1.5 treatment sessions per patient, 70% of patients achieved CE-IM, whereas another 25% had persistent, but partial (≥50%) improvement in length of BE.

The investigators hypothesized that treatment efficacy may be enhanced with the incorporation of a forthcoming focal RFA device developed to target anatomically challenging regions, including the flaring gastroesophageal junction in the setting of a hiatal hernia. These devices would also allow treatment of 25% or less of the circumference of the esophagus and up to 3 cm in length with a single application. Subjects were therefore later invited for a follow-up endoscopy and ablation with a focal device if endoscopically or histologically indicated. It should be noted that a higher energy level for focal device was used (and is currently recommended) as compared with the circumferential probe (12 J/cm ² vs 10 J/cm ² ), although the improved efficacy for higher dosimetry has not been conclusively demonstrated. A total of 62 of 70 (89%) participated in the study extension and underwent an additional mean 1.9 treatment sessions focal RFA. At 30 months, CE-IM was achieved in a remarkable 98% of available patients.

Subsequent studies aimed to evaluate the efficacy of RFA in BE with increasing neoplasia, where the opportunity for reducing risk of malignant progression remains greatest. In a prospective multicenter study by Ganz and colleagues, 142 patients with BE and HGD underwent circumferential balloon-based ablation, of which 92 returned for at least one follow-up biopsy. After a median 12 months and median one ablation session, 90% of patients did not have any remaining histologic evidence of HGD (80% CE-D), but only 54% achieved CE-IM. As in the AIM-I Trial, efficacy may have been improved with inclusion of focal ablation.

The efficacy of RFA in neoplastic BE was firmly established in a landmark study by Shaheen and colleagues known as the AIM Dysplasia Trial and published in 2009. This was a multicenter, sham-controlled trial in which 127 patients with neoplastic BE (LGD or HGD) measuring up to 8 cm were randomized to receive either RFA or sham procedure. After initial circumferential balloon-based ablation, patients returned in preset intervals for endoscopic evaluation with biopsies and up to three adjunctive focal ablations as indicated. The primary outcomes were eradication of dysplasia and IM at 12 months. A total 121 patients completed the study protocol. Based on intention-to-treat analysis, 91% of patients with LGD and 81% of patients with HGD in the ablation group achieved CE-D compared with 23% and 19% of those in their respective control arms. Overall, 78% of patients in the ablation group achieved CE-IM as compared with only 2.3% in the control group ( P <.001). Moreover, subjects receiving ablation had a lower risk of disease progression (3.6% vs 16.3%; P = .03). This included risk of progression to esophageal cancer (1.2% vs 9.3%; P = .045), which notably was based on few occurrences, all of which were in patients with baseline HGD.

Patients with a history of early esophageal cancer were largely excluded from early RFA study protocols. Although endoscopic mucosal resection (EMR) can be used to curatively treat IMC, it is recognized that the risk of metachronous HGD or cancer arising from residual BE remains substantial. Therefore, RFA therapy has naturally been extended to this high-risk cohort of patients. In a cohort of 54 patients with HGD and/or early stage cancer treated with EMR followed by RFA, 98% achieved CE-IM, and only three had recurrent dysplasia or carcinoma during a median 60-month follow-up, and all events were managed endoscopically. Several studies have since reported findings supporting the efficacy of RFA in this high-risk population. Thus, RFA is a first-line option for treating background Barrett’s epithelium after nodular IMC has been successfully resected.

The role of RFA in the management of patients with LGD was not immediately apparent from early studies including the AIM Dysplasia Trial, which was not powered to detect a difference in the risk of progression from LGD to HGD or cancer. This issue was recently addressed by Phoa and colleagues in a multicenter, randomized, clinical trial (SURF Study) in which 136 patients with BE and confirmed LGD were randomized to receive ablation or endoscopic surveillance (every 6 months the first year, and annually thereafter). A total of 88% of patients in the ablation group achieved CE-IM compared with 0% in the surveillance group. After 3-year follow-up, one patient in the ablation group (1.5%) progressed to HGD or cancer, whereas 18 patients in the surveillance group (26.5%) did, for an absolute risk reduction of 25%. The study had a higher rate of progression to HGD or cancer than previously published rates, which may limit its generalizability to other populations. This high rate of progression may have been caused by the use of centralized experienced pathologists, a process previously described to select for higher risk of progression in patients with BE with LGD. Therefore, consideration of RFA therapy is warranted in patients with BE with LGD, particularly if an expert pathologist confirms the diagnosis.

Large Registries and Meta-analyses

A growing body of data outside clinical trials continues to demonstrate the efficacy of RFA for treatment of neoplastic BE. However, the real-world success of RFA treatment may be mitigated because of various clinical factors. In a multicenter United Kingdom registry, 335 patients with neoplastic BE (72% HGD, 24% IMC, 4% LGD) were treated with RFA, and 86% achieved CE-D, whereas 62% achieved CE-IM. These registry results reflect more advanced baseline histology, inclusion of complex disease that failed prior photodynamic therapy (8%), and more extensive BE (mean baseline length, 5.8 cm) compared with clinical trial data. In an updated publication from the same registry, however, the rates of CE-D and CE-IM improved in the latter of two time periods (2008–2010 and 2011–2013) from 56% to 83% and 77% to 92%, respectively. This was perhaps related to better lesion recognition and more frequent endoscopic resection (48% to 60%) before initiating RFA, emphasizing the importance of combination therapy in the treatment of BE.

Abstract data from a large United States RFA Patient Registry comprised of 148 institutions (113 community-based, 35 academic-affiliated) also support the efficacy of RFA. Of 1027 patients with BE and LGD, HGD, or IMC who completed RFA treatment, the rates of CE-IM were 67%, 61%, and 73%, and the rates of CE-D were 90%, 82%, and 85%, respectively. RFA efficacy was independent of treatment center (community-based vs academic-affiliated institution).

In a meta-analysis of 18 studies including 3802 patients with BE and progressive neoplasia, the pooled percentage of patients achieving CE-IM and CE-D was 78% (95% confidence interval [CI], 70%–86%) and 91% (95% CI, 87%–95%). However, substantial heterogeneity was noted at 96% and 78%, respectively. Patients with HGD were significantly less likely to achieve CE-IM (relative risk, 0.92; 95% CI, 0.87–0.98) or CE-D (relative risk, 0.94; 95% CI, 0.91–0.97). Progression to EAC occurred in 0.2% of patients during treatment and 0.7% during 1.5-years follow-up. These data overwhelmingly support the efficacy of RFA therapy in eradicating IM and neoplasia.

Predictors of Response

Several other studies have identified statistically significant predictors of poor response to RFA therapy. In one study, active reflux esophagitis, endoscopic resections scar, narrow esophagus, and duration of neoplasia were associated with poor initial response to circumferential RFA. In other studies, increasing BE length, age, and incomplete healing between treatment sessions were associated with longer times to achieving CE-IM. A history of fundoplication does not seem to impact the efficacy of RFA. Data do suggest a correlation between endoscopist RFA volume and rate of CE-IM. The number of treatment sessions required to achieve CE-IM seems to attenuate after treatment of approximately 30 patients.