Beyond the Biopsy: How AI is Redefining Esophageal Cancer Prevention

FRIENDSWOOD, TX – December 12, 2025 – For millions of people living with Barrett’s esophagus (BE), a common precancerous condition, the diagnosis initiates a period of anxious uncertainty. The standard approach, known as “watchful waiting,” involves regular endoscopic surveillance to look for signs of progression toward esophageal adenocarcinoma (EAC), a cancer with a grim five-year survival rate. The problem is that this surveillance relies on traditional pathology, a process prone to subjective interpretation and diagnostic variability. A patient’s path—whether it’s continued surveillance or aggressive intervention—can depend entirely on which pathologist reviews their biopsy.

Now, a disruptive technology is poised to replace that uncertainty with data-driven precision. Castle Biosciences, a Texas-based diagnostics firm, has solidified the clinical power of its TissueCypher® Barrett’s Esophagus test with a new landmark study. A systematic review and meta-analysis (SRMA) published in the Journal of Clinical Gastroenterology confirms that the AI-powered test dramatically outperforms traditional methods in predicting which patients will progress to cancer. This innovation isn't just an incremental improvement; it represents a fundamental shift in how we approach cancer prevention, moving from a reactive to a proactive model.

AI-Powered Certainty in a World of Ambiguity

The newly published meta-analysis, which synthesizes data from six prior studies and over 8,000 patient biopsies, provides the most robust evidence to date for TissueCypher's predictive power. The results are striking: patients flagged as high-risk by the test were found to be 6.7 times more likely to progress to high-grade dysplasia (HGD) or esophageal cancer within five years compared to their low-risk counterparts.

More importantly, the test provides actionable numbers that align with clinical decision-making. Patients with high-risk results showed an annual progression rate of 5.6%. This figure is critical because it significantly surpasses the 1.7% annual progression rate typically seen in patients with low-grade dysplasia (LGD), a long-established threshold that triggers therapeutic intervention. By identifying a group whose risk exceeds this benchmark, TissueCypher gives physicians a clear, evidence-based rationale to act decisively, often before dysplasia is even visible under a microscope.

“Our findings provide strong evidence that TissueCypher delivers meaningful risk stratification for patients with Barrett’s esophagus,” said Dr. Caitlin C. Houghton, a foregut surgeon at Keck Medicine of USC and the study's lead author. “By identifying which patients are truly at high risk for progression to esophageal cancer — and which are not — TissueCypher can help physicians personalize care, flagging those who may benefit from earlier intervention and providing confidence in continuing routine surveillance for those at low risk.”

At its core, TissueCypher leverages an AI-driven spatialomics platform. Instead of just relying on a pathologist’s visual assessment of cell structure, the test analyzes a standard biopsy sample at a molecular level. It identifies complex protein expression patterns and cellular structures—subtle signatures of progression that are invisible to the human eye—to generate an objective, personalized risk score. This allows it to see the future of the tissue, predicting progression long before the cellular changes of dysplasia become apparent.

Overcoming the Limits of Traditional Care

The current standard of care for Barrett’s esophagus has long been a source of clinical frustration. Guidelines from major bodies like the American College of Gastroenterology (ACG) recommend endoscopic surveillance, but the interpretation of biopsy results can be notoriously inconsistent. Two highly skilled pathologists can look at the same tissue sample and arrive at different conclusions—one seeing non-dysplastic tissue, the other seeing low-grade dysplasia. This interobserver variability creates a significant dilemma, leading to potential over-treatment of low-risk patients and, more dangerously, under-treatment of those at high risk.

This is the precise gap TissueCypher was designed to fill. By providing an objective, reproducible score, it removes the guesswork. The American Gastroenterological Association (AGA) has already acknowledged this potential, issuing a best practice advice article suggesting the test may be beneficial for risk-stratifying patients with non-dysplastic BE. This signals a growing recognition within the medical community that molecular diagnostics are essential for overcoming the inherent limitations of traditional histopathology.

“Relying solely on a visual review of a biopsy is like trying to predict a storm by just looking at the clouds,” noted one independent gastroenterologist at a major academic medical center. “TissueCypher gives us the barometric pressure, the wind speed, the satellite data. It doesn't just change the diagnosis; it changes the entire conversation with the patient from one of uncertainty to one of clear, actionable strategy.”

The Economic and Clinical Calculus of Precision

The impact of this technology extends beyond the clinic, carrying significant economic implications for the healthcare system. A health economics study, conducted in collaboration with the Mayo Clinic, found that a TissueCypher-guided management strategy could reduce healthcare costs by $17 million for a cohort of 100,000 BE patients. This efficiency is driven by a dual benefit: confidently de-escalating care for the low-risk majority and appropriately escalating it for the high-risk minority.

For the large number of patients who receive a low-risk score, the test provides peace of mind and a data-backed reason to avoid costly and invasive endoscopic eradication therapies (EET) and reduce the frequency of surveillance. This not only saves money but also spares patients from unnecessary procedures and the associated anxiety.

Conversely, by up-staging patients whose risk was underestimated by traditional pathology, the test funnels them toward timely intervention. The same economic study projected that this targeted approach could lead to a 58.4% reduction in esophageal cancer progression among patients who receive EET based on an intermediate or high-risk score. It’s a clear win-win: better patient outcomes achieved with greater cost-efficiency.

Navigating the Path to Widespread Adoption

For Castle Biosciences, the powerful clinical validation from the new meta-analysis is a major strategic milestone. TissueCypher is a key growth driver for the company, which reported a 92% year-over-year increase in test volume in the second quarter of 2025 and an industry-leading gross margin of 77% for the test. This demonstrates not only strong clinical demand but also a viable business model.

However, the path from innovation to universal adoption is rarely smooth. The primary hurdle remains payer coverage. While TissueCypher was granted Advanced Diagnostic Laboratory Test (ADLT) status by the Centers for Medicare & Medicaid Services—a move that facilitates reimbursement—coverage among private insurers is still a patchwork. Some, like Geisinger Health Plan, have established positive coverage policies. Others, including certain Blue Cross Blue Shield affiliates, still deem the test investigational, citing a need for more direct evidence comparing health outcomes against the standard of care.

Despite these hurdles, the momentum appears unstoppable. As more evidence accumulates and clinical guidelines evolve, the case for objective, AI-driven risk stratification becomes increasingly compelling. By empowering physicians to make more confident, personalized decisions, TissueCypher is not just improving the management of a single condition; it is demonstrating the transformative power of precision diagnostics to make healthcare more proactive, effective, and sustainable.