Dysplasia-Stratified Surveillance Identifies Optimal Strategies for Preventing Esophageal Adenocarcinoma in Barrett's Esophagus: An Incidence-Based Cost-Effectiveness Model

This cost-effectiveness analysis of Barrett's esophagus surveillance in the Japanese population concludes that annual endoscopy is optimal for low-grade dysplasia and annual breath testing for high-grade dysplasia, while routine surveillance offers limited value for non-dysplastic or short-segment cases.

The Gist

The Big Picture: A "Barrett's Esophagus" Weather Forecast

Imagine your esophagus (the tube connecting your throat to your stomach) is a garden. In some people, the soil changes type due to acid reflux. This change is called Barrett's Esophagus (BE).

Most of the time, this garden is just a little weird-looking but safe. However, in rare cases, weeds can grow that turn into a dangerous plant called Esophageal Adenocarcinoma (EAC), a type of cancer.

The problem is: How often should we check the garden?

• If we check too often, we waste money and bother the gardener (the patient).
• If we check too rarely, we might miss the weeds until they take over.

This paper is a "cost-benefit calculator" for Japan. It asks: What is the smartest way to check these gardens based on how "weird" they already look?

---

The Five Types of Gardens (The Subgroups)

The researchers realized that not all gardens are the same. They split the patients into five groups based on two things: how big the weird patch is (segment length) and how angry the cells look (dysplasia grade).

Think of it like grading the risk of a storm:

1. USSBE & SSBE (Ultra/Short Segment): A tiny patch of weird grass. (Low Risk)
2. LSBE-NDBE (Long Segment, No Dysplasia): A big patch of weird grass, but it's still just grass. (Low-Medium Risk)
3. LSBE-LGD (Long Segment, Low-Grade Dysplasia): The grass is starting to look a bit like a weed. (Medium-High Risk)
4. LSBE-HGD (Long Segment, High-Grade Dysplasia): The weeds are almost fully grown and ready to bloom into cancer. (Very High Risk)

---

The Tools in the Toolbox

The study tested different ways to inspect these gardens:

• The "Gold Standard" (Endoscopy): A camera on a long stick goes down the throat. It's accurate but invasive, expensive, and a bit uncomfortable.
• The "Sponge Test": You swallow a tiny sponge on a string, pull it back up, and check the cells. Less invasive.
• The "Breath Test": You blow into a machine that smells for cancer chemicals. Very easy, no swallowing.
• The "miRNA Test": A blood test looking for specific genetic markers.
• The "AI Assistant": A computer program that helps the doctor looking through the camera spot tiny problems the human eye might miss.

---

The Results: What Did the Calculator Say?

The researchers ran a simulation over a lifetime for 50-year-olds. Here is the "Verdict" for each garden type:

1. The Tiny & Big "Just Weird" Gardens (USSBE, SSBE, LSBE-NDBE)

• The Verdict: Don't bother with a special surveillance plan.
• The Analogy: Imagine checking your front door for a burglar every single day when you live in a safe neighborhood with no crime. You spend a fortune on locks and alarms, but you almost never catch a thief. The cost of checking is higher than the tiny chance of a problem.
• The Strategy: Just keep an eye on them during regular checkups. Don't schedule special, expensive tests.

2. The "Weedy" Garden (LSBE-LGD)

• The Verdict: Check once a year with the Camera (Endoscopy).
• The Analogy: The weeds are starting to sprout. If you ignore them, they will choke the garden. You need a professional gardener (the doctor with the camera) to come in every year to pull them out before they become a forest.
• The Strategy: Annual Endoscopy is the most cost-effective way to save lives here.

3. The "Ready-to-Bloom" Garden (LSBE-HGD)

• The Verdict: Check once a year with the Breath Test.
• The Analogy: The weeds are huge and about to turn into a monster. You need to check every year, but you don't need the heavy, expensive camera equipment every single time. A simple "sniff test" (Breath Test) is actually the best deal. It catches the problem early enough to stop it, but it's cheaper and easier for the patient than the camera.
• The Strategy: Annual Breath Test is the winner here.

---

The "Magic Glasses" (AI)

The study also looked at what happens if doctors wear "Magic Glasses" (AI) while using the camera.

• The Result: The AI makes the camera much better at spotting tiny weeds.
• The Impact: If the camera becomes super-smart, it becomes even better for the "Weedy" garden (LGD). It helps catch problems earlier, making the camera even more worth the cost.

---

Why Does This Matter? (The "So What?")

1. Stop Wasting Money on Safe Gardens:
In Japan, many people have tiny patches of Barrett's Esophagus. The old way was to check everyone the same way. This paper says: "Stop checking the safe gardens so often." It saves the healthcare system millions of dollars that can be used elsewhere.

2. Focus on the Real Danger:
By saving money on the low-risk groups, we can afford to be very aggressive with the high-risk groups (the ones with dysplasia). We can use the best tools (like breath tests or AI cameras) to make sure we catch the cancer early.

3. New Tech is a Game Changer:
The study shows that non-invasive tools (like breathing into a machine) might be better than the uncomfortable camera for the highest-risk patients. This could make patients much happier and more willing to get checked.

The Bottom Line

This paper is a roadmap for doctors and policymakers. It says: "Treat the garden based on how bad the weeds look."

• No weeds? Relax.
• Small weeds? Check yearly with a camera.
• Big weeds? Check yearly with a breath test.

It's about using the right tool for the right job to save lives without breaking the bank.

Technical Summary

1. Problem Statement

Barrett's Esophagus (BE) is the primary precursor to Esophageal Adenocarcinoma (EAC), a malignancy with rising incidence and poor prognosis. While Western guidelines recommend structured endoscopic surveillance based on BE segment length and dysplasia grade, Japan lacks a standardized national surveillance framework.

• Context: In Japan, EAC detection has historically relied on incidental findings during gastric cancer screening. However, as Helicobacter pylori eradication reduces gastric cancer rates, incidental EAC detection is expected to decline.
• Challenge: The risk of EAC varies significantly by BE subtype (ultra-short, short, and long segments) and dysplasia status (nondysplastic, low-grade, high-grade). Current practices are inconsistent, and it is unclear which surveillance strategies (endoscopy vs. emerging non-invasive tests) are cost-effective for specific Japanese subpopulations.
• Goal: To evaluate the cost-effectiveness of dysplasia-stratified surveillance strategies for the Japanese BE population to inform evidence-based policy.

2. Methodology

The study employed a state-transition (Markov) model from a healthcare payer perspective, simulating a hypothetical cohort of 50-year-old individuals over their lifetime.

• Model Structure:

  • Five independent models were constructed for distinct BE subgroups: Ultra-short segment BE (USSBE), Short-segment BE (SSBE), Long-segment BE without dysplasia (LSBE-NDBE), LSBE with Low-Grade Dysplasia (LSBE-LGD), and LSBE with High-Grade Dysplasia (LSBE-HGD).
  • Incidence-Based Approach: Unlike traditional models that simulate natural history progression (e.g., BE $\to$ Dysplasia $\to$ Cancer), this model relied on empirically observed annual EAC incidence rates for each subgroup. It did not assume unobserved transitions between health states, reducing structural uncertainty.
  • Cycle: 1-year cycles with half-cycle correction; costs and Quality-Adjusted Life-Years (QALYs) discounted at 3%.

• Surveillance Strategies Evaluated:

  • Modalities: Conventional endoscopy, Sponge test (Cytosponge), Breath test, and miRNA test.
  • Intervals: 1, 2, 3, 4, 5, or 10 years.
  • Logic: Non-invasive tests served as primary screening; positive results triggered confirmatory endoscopy. A "No Surveillance" strategy served as the reference.
  • Scenario Analysis: AI-assisted endoscopy was evaluated as an enhanced version of conventional endoscopy.

• Parameters & Inputs:

  • Incidence: Derived from Japanese data for USSBE/SSBE and Western meta-analyses for LSBE subgroups (due to lack of Japanese dysplasia-specific data).
  • Performance: Sensitivity, specificity, and adherence rates were sourced from literature and assumptions.
  • Costs: Based on the Japanese national fee schedule (converted to USD) and assumptions for emerging technologies.
  • Outcome Measures: Net Monetary Benefit (NMB), Incremental Cost-Effectiveness Ratio (ICER), QALYs, and EAC deaths averted.
  • Threshold: Willingness-to-pay (WTP) of US$50,000 per QALY.

• Analysis Types: Base-case analysis, deterministic one-way sensitivity analyses, probabilistic sensitivity analysis (10,000 Monte Carlo simulations), and scenario analyses.

3. Key Contributions

• First Integrated Risk-Stratified Analysis: This is the first study to provide a fully integrated cost-effectiveness analysis of EAC surveillance across five distinct BE subgroups globally, specifically tailored to the Japanese epidemiological context.
• Incidence-Based Framework: The study introduces a transparent modeling approach that anchors cost-effectiveness directly to observed cancer incidence rather than hypothetical progression pathways, avoiding assumptions about preclinical disease dynamics.
• Evaluation of Emerging Technologies: It is the first to model the cost-effectiveness of non-invasive triage tools (breath test, miRNA) and AI-assisted endoscopy specifically within the context of BE surveillance in Japan.
• Policy Relevance: It addresses the critical gap in Japanese guidelines by providing data-driven recommendations for a population where BE is frequently ultrashort-segment and EAC incidence is rising.

4. Key Results

A. Cost-Effectiveness by Subgroup

• Low-Risk Groups (USSBE, SSBE, LSBE-NDBE):

  • Finding: Surveillance was not cost-effective for any strategy compared to "No Surveillance."
  • Reasoning: The annual EAC incidence is too low to justify the costs of repeated testing. For example, in LSBE-NDBE, annual endoscopy increased costs significantly with negligible QALY gains.
  • Recommendation: Routine surveillance is not justified for these groups.

• Intermediate Risk (LSBE-LGD):

  • Finding: Annual Endoscopy was the most cost-effective strategy.
  • Metrics: Produced the highest Net Monetary Benefit (NMB: US$841,993). It averted 83 EAC deaths per 10,000 individuals over a lifetime.
  • Comparison: Non-invasive tests (breath/miRNA) offered slightly higher QALYs but were not cost-effective due to higher costs relative to the marginal benefit.

• High Risk (LSBE-HGD):

  • Finding: Annual Breath Test was the most cost-effective strategy.
  • Metrics: Highest NMB (US$598,066) with an ICER of US$16,318 per QALY. It averted 295 EAC deaths per 10,000 individuals over a lifetime.
  • Comparison: While miRNA tests prevented the most deaths (311), they were not cost-effective due to high costs. Annual endoscopy was less efficient than the breath test in this high-risk group.

B. Impact of AI-Assisted Endoscopy

• In scenario analyses, AI-assisted endoscopy (with improved sensitivity/specificity) maintained cost-effectiveness for LSBE-LGD and LSBE-HGD, often outperforming conventional endoscopy by reducing false positives and increasing early detection.

C. Sensitivity Analysis

• Drivers: Cost-effectiveness was primarily driven by subgroup-specific EAC incidence and test adherence rates.
• Thresholds: In LSBE-LGD, surveillance became cost-effective only when annual incidence exceeded 0.009. In LSBE-HGD, the breath test was favored over endoscopy when endoscopy adherence was high (>83.5%) but breath test adherence remained slightly lower.

5. Significance and Implications

• Stratified Surveillance Framework: The study supports a shift from uniform surveillance intervals to a dysplasia-specific framework. It validates that "one-size-fits-all" surveillance is inefficient, particularly in Japan where low-risk (USSBE/SSBE) cases dominate.
• Resource Optimization: Policymakers can avoid unnecessary procedures and costs in low-risk populations (USSBE, SSBE, NDBE) while prioritizing intensive surveillance for high-risk groups (LGD, HGD).
• Role of Non-Invasive Tools: The findings suggest that non-invasive technologies like the breath test are highly valuable for high-risk (HGD) populations, potentially reducing the burden of frequent endoscopies and improving patient compliance.
• Global Applicability: Although modeled for Japan, the incidence-based framework is generalizable. By substituting local incidence and cost data, this model can be applied to Western healthcare systems to refine their own surveillance guidelines.
• Future Outlook: As gastric cancer screening patterns change in Japan, integrating these risk-stratified strategies into existing endoscopic pathways will be crucial for maintaining early EAC detection without overburdening the healthcare system.

In conclusion, the paper provides a robust economic justification for tailoring BE surveillance to specific risk profiles, advocating for annual endoscopy for LGD and annual breath testing for HGD, while recommending against routine surveillance for nondysplastic and short-segment BE.