Independent Blinded Validation of a Tissue Systems Pathology Test to Predict Progression in Patients With Barrett's Esophagus

Abstract

Introduction: A risk prediction test was previously validated to predict progression to high-grade dysplasia (HGD) and esophageal adenocarcinoma (EAC) in patients with Barrett's esophagus (BE). The aim of our study was to independently validate this test to predict the risk of progression to HGD/EAC in BE patients with nondysplastic (ND), indefinite for dysplasia and low-grade dysplasia (LGD).

Methods: A single-blinded, case-control study was conducted to stratify patients with BE as low, intermediate, or high risk for progression to HGD/EAC within 5 years using a previously described risk prediction test. Patients with BE who progressed to HGD/EAC after at least 1 year (n = 58) were matched to patients undergoing surveillance without progression (n = 210, median surveillance 7 years). Baseline biopsies with subspecialist diagnoses of ND, indefinite for dysplasia, or LGD were tested in a blinded manner, and the predictive performance of the test was assessed.

Results: This risk prediction test stratified patients with BE based on progression risk with the high-risk group at 4.7-fold increased risk for HGD/EAC compared with the low-risk group (95% confidence interval 2.5-8.8, P < 0.0001). Prevalence-adjusted positive predictive value at 5 years was 23%. The high-risk class and male sex provided predictive power that was independent of pathologic diagnosis, age, segment length, and hiatal hernia. Patients with ND BE who scored high risk progressed at a higher rate (26%) than patients with subspecialist-confirmed LGD (21.8%) at 5 years.

Discussion: A risk prediction test identifies patients with ND BE who are at high risk for progression to HGD/EAC and may benefit from early endoscopic therapy or increased surveillance.

Figure 1.. Independent validation of the Risk Prediction Test in BE patients from Cleveland Clinic and University of Pittsburgh.

KM analysis of probability of progression to HGD/EAC in BE patients with diagnoses of ND, IND and LGD stratified into low-, intermediate-, and high-risk by the risk prediction test. Panel A: BE patients from the two clinical institutions combined (n=268; 58 incident progressors (IP) and 210 non-progressors (NP)); Panel B: BE patients from Cleveland Clinic (27 IP and 111 NP); Panel C: BE patients from University of Pittsburgh (31 IP and 99 NP); Panel D: BE patients excluding those with visible lesions and with less than 2 years progression time (44 IP and 203 NP). HGD/EAC-free surveillance time: number of years between biopsy tested and last endoscopy with ND, IND or LGD (non-progressors) or endoscopy with HGD or EAC (incident progressors).

Figure 2.. Risk Prediction Test Identifies Non-Dysplastic BE Patients who Progress at a Higher Rate than Patients with Subspecialist-Confirmed LGD.

Panel A: KM analysis of probability of progression to HGD/EAC in BE patients with GI subspecialist pathologist confirmed ND stratified into low-, intermediate-, and high-risk by the risk prediction test (n=227 with diagnosis ND, 43 IP and 184 NP). Panel B: KM analysis of probability of progression to HGD/EAC in BE patients with original diagnosis ND stratified into low-, intermediate-, and high-risk by the risk prediction test (n=175 with diagnosis ND, 37 IP and 138 NP). Panel C: KM analysis of probability of progression to HGD/EAC in BE patients stratified by GI subspecialist pathologist review diagnoses of ND, IND and LGD (n=268; 58 IP and 210 NP). Panel D: KM analysis of probability of progression to HGD/EAC in BE patients stratified by original pathologist diagnoses (i.e. diagnoses recorded in the health records at the time of care) of ND, IND and LGD (n=268; 58 IP and 210 NP). PPV was adjusted for disease prevalence as described in the Materials and Methods.

Figure 3.. Representative Images of Biomarkers Labeled in BE Biopsies.

Panels A-D show a ND biopsy from a male patient with 3cm BE segment, hiatal hernia, and no visible lesions who was diagnosed with HGD 2.7 years later. This specimen was scored 9.6/High-Risk by the risk prediction test and exhibited loss of p16, overexpression of p53 and AMACR, increased infiltration of CD68+ cells and HIF-1α+ cells, as well as elevated levels of HER2. Panels E-F show a ND biopsy from a male patient with 4cm BE segment, hiatal hernia, and no visible lesions, who had HGD/EAC-free surveillance time 6.7 years; this specimen was scored 3.6/Low-Risk by the risk prediction test. A and E: p16-green, AMACR-red, p53-yellow; B and F: CD68-green, COX-2-red; C and G: HIF-1α-green, CD45RO-red; D and H: HER2-green, K20-red. Hoechst labeling of nuclei is shown in blue in all panels.