B0197 - Clinical Utility of an AI-driven Precision Medicine Test for Barrett’s Esophagus: Results From a Randomized Controlled Trial

Monday, October 24, 2022

10:00 AM – 12:00 PM ET

Location: Crown Ballroom

Presenting Author(s)

Rebecca Critchley-Thorne, PhD

Castle Biosciences
Pittsburgh, PA

John Peabody, MD, PhD¹, Jamielyn Cruz, MD¹, David Paculdo, MPH¹, Divya Ganesan, MS¹, Rebecca Critchley-Thorne, PhD², Nicholas J. Shaheen, MD, MPH, MACG³, Sachin Wani, MD⁴
¹QURE Healthcare, San Francisco, CA; ²Castle Biosciences, Pittsburgh, PA; ³University of North Carolina at Chapel Hill, Chapel Hill, NC; ⁴University of Colorado, Anschutz Medical Center, Aurora, CO

Introduction: Barrett’s esophagus (BE) is a treatable precursor to esophageal adenocarcinoma (EAC). Physician adherence to BE management guidelines is poor, leading to reduced dysplasia detection and inappropriate treatment. We conducted a randomized controlled study to determine the clinical utility of a precision medicine test that predicts risk of progression to high grade dysplasia (HGD) or EAC.

Methods: 259 practicing general and interventional gastroenterologists and gastrointestinal surgeons participated in this IRB-approved study. We measured the clinical care for three simulated BE patient types: non-dysplastic BE, indefinite for dysplasia, and low-grade dysplasia that each had three variants: high-risk clinical profile and high-risk test score [variant A], low-risk clinical profile and high-risk test score [variant B], and high-risk profile and low-risk test score [variant C]. After collecting baseline data on clinical care for the cases, the precision medicine test was introduced to two intervention groups who viewed its educational materials. In a second round, intervention 1 was given the test results while intervention 2 optionally ordered the results. Their care was compared to controls. Quality of care scores were calculated based on the number of physician responses that matched evidence-based criteria for the management of BE.

Results: Intervention 1’s quality of care scores showed improvement compared to control for all case variants (difference-in-difference scores A: +1.8, p=0.600; B: +6.7%, p=0.013; C: +4.3%, p=0.129, see Table). After controlling for physician and practice characteristics, intervention 1 outperformed controls in correctly determining risk of progression to HGD/EAC among all three case variants (p< 0.01 for all) and ordering the correct primary treatment (p=0.02, see Table). Intervention 2 participants ordered the test results in 21.9% of cases, and those who did performed similarly to, or even outperformed, intervention 1. Those who did not order the test performed similarly to controls (p >0.10).

Discussion: This precision medicine test for BE optimized appropriate care for both patients at high-risk and low-risk for disease progression to HGD/EAC, particularly for patients with low-risk clinical profiles but high-risk test results. With increased test use, physicians may treat early disease stages more efficiently and accurately and prevent progression to EAC.

	Difference-in-difference*
	Intervention 1 (received risk prediction test results)			Intervention 2 (given option to order risk prediction test)
	Intervention 1			Test Orderers	Test Non-Orderers
Diagnosis+Management Quality of Care Scores
Variant A		+1.8%	-5.5%			-1.5%
Variant B		+6.7%**	+6.8%			+3.2%
Variant C		+4.3%	+5.7%			+3.5%

Correctly Assessing Risk of Progression to HGD/EAC
Variant A	+60.3%**			+73.0%**	-1.2%
Variant B	+65.7%**			+66.6%**	+10.4%
Variant C	+32.6%**			+59.9%**	-1.9%

Correct Primary BE Management
Eradication therapy	+14.6%			+30.8%***	-1.0%
Endoscopic surveillance	+18.5%**			+5.9%	+2.5%

Variant A: high-risk clinical profile and high-risk test result Variant B: low-risk clinical profile and high-risk test result Variant C: high-risk clinical profile and low-risk test result Compared to controls; p-value < 0.05; **p-value < 0.10

Table: Quality of Care (Diagnosis and Management) Difference-in-Difference Results Across Study Arms.

Disclosures:

John Peabody: QURE Healthcare – Employee.

Jamielyn Cruz: QURE Healthcare – Employee.

David Paculdo: QURE Healthcare – Employee.

Divya Ganesan: QURE Healthcare – Employee.

Rebecca Critchley-Thorne: Castle Biosciences – Employee.

Nicholas Shaheen: Aqua Medical – Consultant. CDx Diagnostics – Grant/Research Support. Cernostics – Consultant. Cook Medical – Consultant. Exact Sciences – Consultant. Interpace Diagnostics – Grant/Research Support. Lucid Medical – Grant/Research Support. Medtronic – Grant/Research Support. Pentax – Grant/Research Support. Phathom Pharmaceuticals – Consultant. Steris – Grant/Research Support.

Sachin Wani: Ambu – Grant/Research Support. CDx Medical – Grant/Research Support. Cernostics – Consultant. Exact Sciences – Consultant. Lucid Medical – Grant/Research Support.

John Peabody, MD, PhD¹, Jamielyn Cruz, MD¹, David Paculdo, MPH¹, Divya Ganesan, MS¹, Rebecca Critchley-Thorne, PhD², Nicholas J. Shaheen, MD, MPH, MACG³, Sachin Wani, MD⁴. B0197 - Clinical Utility of an AI-driven Precision Medicine Test for Barrett’s Esophagus: Results From a Randomized Controlled Trial, ACG 2022 Annual Scientific Meeting Abstracts. Charlotte, NC: American College of Gastroenterology.