Neoplastic Recurrence After Successful Treatment For Early Barrett’s Neoplasia: Development of a Penalized Prediction Model

Aims Endoscopic resection (ER) +/- radiofrequency ablation (RFA) is the treatment of choice for Barrett’s esophagus (BE) with early neoplasia. Since long-term outcomes are limited, patients still undergo regular follow-up endoscopies after successful treatment (ie,complete eradication of BE, CE-BE). We aimed to develop a prediction model for recurrence, which can be used for personalized follow-up.

Methods We collected data from the Dutch Barrett Expert Center Registry, a nationwide registry that captures outcomes from all BE patients that underwent endoscopic treatment for early BE-neoplasia in expert centers. Recurrence was defined as histologic finding of low-grade dysplasia(LGD), high-grade dysplasia(HGD) or esophageal adenocarcinoma(EAC) during endoscopic follow-up. We built a prognostic survival model taking account of competing risks (i.e,Fine and Gray) with LASSO penalization. We assessed: age, gender, baseline pathology, BE-length, new visible lesion(s)(VL) during ablation, poor healing, persisting esophagitis, number of ablation endoscopies, number of ER endoscopies, persisting IM in cardia.

Results

All patients (n = 1154)	No recurrence (n = 1116)	Recurrence (n = 38)	Univariable analysis - Hazard ratio [95 %CI]	LASSO model - Hazard ratio
Max. BE length, cm, mean±SD	5 ± 3	7 ± 3	1,18 [1,12;1,26]	1,16
Baseline HGD/EAC, n (%)	814 (73)	34 (90)	2,52 [0,89;7,09]	1,41
Number of ER endoscopies, median (IQR)	1 (0-1)	1 (0-1)	1,63 [1,17;2,26]	1,18
New visible lesion(s) during ablation phase, n (%)	62 (6)	9 (24)	4,34 [2,05;9,31]	2,85

1,154 patients were included with a mean endoscopic follow-up of 4 years (±2) per patient. Overall, 38 patients developed recurrence (0.8 %/person year[95 %CI 0.6-1.1]; LGD n = 14(37 %); HGD n = 13(34 %); EAC n = 17(45 %)). Following characteristics were independently associated with recurrence (strongest-weakest): new VLs during ablation, higher number of ER-endoscopies, increasing BE-length, HGD/EAC at baseline, younger age, male. The internally validated C-statistic was 0,76[95 %-CI 0,73; 0,79]. For example, a 50y male with 10cm BE with EAC and 3 ER sessions including 1 for a new VL during RFA, had a cumulative risk for recurrence of 48 % during 7y. In contrast, a 70y female with 3cm BE with flat LGD, had 3 % risk.

Conclusions We built the first prediction model for recurrence after successful treatment of early neoplastic BE in a centralized setting, with good discrimination. If external validation confirms its predictive power, this model can help clinicians and patients to manage expectations and determine a personalized follow-up strategy.

Citation: van Munster SN, Nieuwenhuis EA, Weusten BLAM et al. OP74 NEOPLASTIC RECURRENCE AFTER SUCCESSFUL TREATMENT FOR EARLY BARRETT’S NEOPLASIA: DEVELOPMENT OF A PENALIZED PREDICTION MODEL. Endoscopy 2021; 53: S31.

Publication History

Article published online:
19 March 2021