Researchers have developed a mathematical model that predicts the risk for visible recurrent dysplasia after complete eradication of Barrett esophagus (CE-BE), as reported in Gastroenterology.
The model incorporates data from the Dutch Barrett Expert Center (BEC) registry, which includes outcomes for all patients with early Barrett neoplasia who had endoscopic treatment in the Netherlands since 2008. The model was validated in 2 external, separate databases from Switzerland and Belgium.
Investigators included patient and treatment characteristics that were known to the physician at the time of CE-BE and with clinically or biologically plausible effects on the risk for recurrent disease. The prognostic model was developed with use of a Fine and Gray survival model.
Eligible participants had at least 1 radiofrequency ablation (RFA) treatment and achieved CE-BE before December 31, 2018. The primary endpoint was recurrent disease, which was defined as a histologic finding of low-grade dysplasia (LGD), high-grade dysplasia (HGD), or esophageal adenocarcinoma (EAC) in the esophagus or cardia during follow-up.
A total of 1154 patients (mean age, 64±9 years; 82% men) achieved complete endoscopic and histological remission of BE after RFA with or without endoscopic resection (ER) and were included in the analysis. The mean follow-up was 4±2 years.
Visible recurrent LGD, HGD, or EAC occurred in 38 patients. The worst histologic recurrence grade was LGD (n = 14), HGD (n = 7), or EAC (n = 17). The annual recurrence risk was 1.0% (95% CI, 0.8-1.4) for patients with recurrent LGD or worse and 0.7% (95% CI, 0.4-1.0) for those with recurrent HGD or worse.
According to the multivariable LASSO model, younger age was associated with an increased risk for recurrence (hazard ratio [HR], 1.01), as was male sex (HR, 1.37), HGD or EAC at baseline compared with LGD (HR, 1.02), increasing length of BE (HR, 1.16), greater number of ER treatments (HR, 1.18), and an incident lesion (HR, 2.88).
In the combined external datasets, the Brier score was 0.38 (95% CI, 0.10-0.74), with lower scores indicating better overall performance (range, 0-1). The C-index was 0.92 (95% CI, 0.86-0.94), with higher scores indicating better discrimination (range, 0-1).
The researchers noted that the low number of recurrences could limit the performance of the model with a risk for overfitting. Other limitations included the treatment of patients who developed nondysplastic BE during follow-up, which might underestimate the true dysplastic recurrence risk, as well as the lack of data on p53 staining.
“We developed and externally validated a model to predict visible dysplastic recurrence after initial successful endoscopic treatment of BE-related neoplasia in a setting of centralized care,” the investigators commented. “Based on 6 clinical features, our model showed excellent model performance in external validation. This model may help to determine personalized surveillance intervals.”
Disclosure: Some of the study authors declared affiliations with biotech, pharmaceutical, and/or device companies. Please see the original reference for a full list of authors’ disclosures.
van Munster S, Nieuwenhuis E, Bisschops R, et al. Dysplastic recurrence after successful treatment for early Barrett’s neoplasia: development and validation of a prediction model. Gastroenterology. Published online March 16, 2022. doi:10.1053/j.gastro.2022.03.020