DENVER—Two studies presented at the 2024 annual meeting of the American Foregut Society support use of the tissue systems pathology test for prediction of neoplastic progression in patients with Barrett’s esophagus.
In one analysis, the addition of clinical and pathologic factors to the risk score generated by the test, TSP-9 (TissueCypher, Castle), did not improve its predictive performance.
“For years, we’ve been using clinical pathologic factors to predict the risk of neoplastic progression in patients with Barrett’s esophagus,” said lead study author Rhonda Souza, MD, the co-director of the Center for Esophageal Research at Baylor Scott & White Research Institute and co-director of the Center for Esophageal Diseases at Baylor University Medical Center, in Dallas. In contrast, “TSP-9 offers a systems biology approach that eliminates subjectivity in the interpretation.” The test generates a risk score ranging from 0 to 10 and categorizes patients into low, intermediate and high risk for neoplastic progression within five years.
Dr. Souza and her co-investigators analyzed the TSP-9 risk results and clinical factors from five previous validation studies that included a total of 683 patients (abstract 42). The researchers evaluated whether adding routinely available clinical pathologic factors to the TSP-9 risk score could improve the accuracy of predicting progression to high-grade dysplasia or cancer within five years.
Patients were randomized to a training set to build predictive models based on combinations of TSP-9 and clinical pathologic factors. The researchers then used a validation set to evaluate the predictive performance of these models.
The TSP-9 score alone had the lowest prediction error compared with models that used clinical factors or a combination of TSP-9 and clinical factors. This finding was true for both the overall patient group (including nondysplastic, indefinite for dysplasia, and low-grade dysplasia) and for patients with nondysplastic BE specifically.
In the validation set, the TSP-9 score alone demonstrated the highest positive predictive value for neoplastic progression. Adding clinical factors to the TSP-9 score did not improve its predictive performance and, in some cases, even reduced it.
“Our findings suggest that clinicians can rely on the TSP-9 score alone for the most accurate prediction of neoplastic progression in Barrett’s esophagus patients,” Dr. Souza said. “This could streamline the risk assessment process and potentially improve patient management strategies.”
Less May Be More In Risk Assessment
Commenting during the post-presentation discussion, moderator Jocelyn Burke, MD, an assistant professor at the University of Nevada, Las Vegas, said, “We often think of more data as helpful when we’re trying to make decisions for patients, but in your study, it seems to say that more data is just more. What are your theories regarding why the clinical factors actually weakened the predictive values?”
In response, Dr. Souza noted that the systems biology approach of the TSP-9 test evaluates multiple tissue layers, including epithelial cells, fibroblast and immune cells, as well as angiogenesis factors. In contrast, some clinical factors are subjective, such as the grade of dysplasia, and nonspecific, such as age. “If you think about it,” she said, “it’s really not surprising that these clinical factors actually detract from the accuracy of TissueCypher.”
—Chase Doyle
This article is from the March 2025 print issue.