An objective risk prediction system based on biomarkers and imaging has the potential to fundamentally alter surveillance of Barrett’s esophagus, according to a roundup of clinical studies. The promise is individualized surveillance that is more reliable and less costly, experts said.

In one blinded cohort investigation, the predictive accuracy of the system was found to be “as good as the best-performing expert pathologists we could find for our study,” said Nicola F. Frei, MD, a gastroenterologist affiliated with Academic Medical Center, in Amsterdam.

In a separate pooled analysis of four multinational studies, “utility was established not only in those with low-grade dysplasia but also in those with nondysplastic Barrett’s,” reported Prasad G. Iyer, MD, a professor of gastroenterology and hepatology at Mayo Clinic in Rochester, Minn.

The studies of this emerging tool, called Tissue-Cypher (Cernostics), were presented at the 2021 virtual Digestive Disease Week. Each of the investigators said the fully automated system, unlike histologic analysis by pathologists, is standardized, objective and highly reproducible.

TissueCypher uses fluorescent biomarkers to interrogate Barrett’s tissue embedded in paraffin. When the software integrates information from nine biomarkers in the context of morphologic features and clinical data, it produces risk scores ranging from 0 to 10, which, for clinical use, are categorized as low, medium or high.

In her study comparing the predictive accuracy of TissueCypher and expert and community pathologists (abstract 43), Dr. Frei’s group evaluated tissue samples from 155 patients who had participated in a previous trial comparing radiofrequency ablation and endoscopic surveillance in people with low-grade dysplasia (JAMA 2014;311[12]:1209-1217).

For distinguishing the 22% of patients who progressed from the remainder who did not, the sensitivity of the automated system and the pathologists was about the same (68% vs. 67%, respectively), according to the researchers. TissueCypher outperformed on specificity (78% vs. 65%), but a receiver operating curve analysis showed substantial variability among pathologists not seen with TissueCypher. As a result, “only the best-performing pathologists came close” to the consistency of the automated system, Dr. Frei said.

TissueCypher detected a subgroup of patients who progressed yet had been downstaged to nondysplastic Barrett’s by the pathologists, according to Dr. Frei, who noted a substantial proportion of histologic specimens that pathologists label as indeterminate. TissueCypher has the advantage of using parameters other than dysplasia to draw a conclusion about progression risk.

With histologic analysis, “dysplasia is the sole determinant of management recommendations in Barrett’s,” Dr. Iyer said. But this approach ignores the fact that many other clinical factors, including patient age, sex and smoking have been identified as risk factors, he said.

In addition to the variability in the skills of pathologists for histologic assessments, Dr. Iyer said gastroenterologists also vary in their compliance with surveillance recommendations. Objective systems analysis like TissueCypher, and others in development, have the potential to standardize risk assessment that includes factors other than histology, he said.

In his pooled analysis of relatively large multicenter studies with TissueCypher, Dr. Iyer and his colleagues found that 152 of the 475 patients with Barrett’s who were evaluated progressed, and 323 did not. Using conditional logistical regression, multivariable models were constructed to identify factors that predict the incidence of progression with and without the TissueCypher risk score. A separate analysis of predictors was performed on nondysplastic Barrett’s.

In the model with TissueCypher, age and sex were not independent predictors of progression. Although expert diagnosis of low-grade dysplasia versus nondysplastic Barrett’s was associated with an almost threefold increased likelihood of progression (odds ratio [OR], 2.90; P=0.011), a high- versus low-risk score on TissueCypher was associated with a more than sevenfold increase in the likelihood of progression (OR, 7.09; P<0.001).

In patients with nondysplastic Barrett’s, the relative performance of TissueCypher for predicting progression found an even higher likelihood (OR, 18.07; P<0.00001).

“The data establish the clinical utility of Tissue-Cypher not only for those with low-grade dysplasia but also for those with nondysplastic Barrett’s, where oftentimes the risk prediction is much more challenging,” Dr. Iyer said.

The implication of these data, Dr. Iyer added, is that TissueCypher has the potential to individualize a surveillance schedule—an urgent need given the low rates of progression among patients with low-grade dysplasia and even lower rates among those with nondysplastic Barrett’s. “The surveillance yield with the current approach is low, and this is making current practice inefficient,” Dr. Iyer said.

Nicholas J. Shaheen, MD, the chief of the Division of Gastroenterology and Hepatology at the University of North Carolina at Chapel Hill, said the studies are part of “a growing literature suggesting that Tissue-Cypher is effective” for risk stratification.

TissueCypher “appears to perform somewhat better than a panel of expert pathologists,” said Dr. Shaheen in a state-of-the-art address on Barrett’s surveillance when he cited Dr. Frei’s data specifically. In particular, he was impressed with the evidence that TissueCypher appears to be accurate for downgrading a substantial proportion of patients with indeterminate risk to low risk.

Although Dr. Shaheen noted that this method and several others for improving risk assessment have yet to achieve widespread adoption in Barrett’s management, he agreed that there is a need for objective and reproducible methods for assessing risk for Barrett’s progression.

—Ted Bosworth


Dr. Frei reported no relevant financial disclosures. Dr. Iyer reported financial relationships with Exact Sciences, Medtronic and Pentax Medical. Dr. Shaheen reported financial relationships with CDx Medical, Cernostics, Cook Medical, Interpace Biosciences, Medtronic, Pentax and Steris.