Benchmarking Clinical Definitions for Capillary Leak

Sunday, March 22, 2026

12:15 PM - 1:15 PM Central Time

Location: Connections Central - RST 16

Jonathan Popham, MD (he/him/his)

Children's National Hospital, District of Columbia

Disclosure(s): No relevant financial relationship(s) to disclose.
FY

Farah Yasmin, MBBS

Postdoctoral Associate
Yale University School of Medicine

Disclosure information not submitted.
NR

Niranjana Ramani, MBBS

Resident
Yale New Haven Children's Hospital

Disclosure information not submitted.
KO

Kevin Okifo, N/A

Medical Student
Yale University School of Medicine

Disclosure information not submitted.
VR

Veena Rao, PhD

Research Scientist
Yale University School of Medicine

Disclosure information not submitted.
JT

Jeffrey Testani, MD, MTR

Professor of Medicine (Cardiovascular Medicine)
Yale University School of Medicine

Disclosure information not submitted.
PW

Perry Wilson, MD, MSCE

Associate Professor of Medicine (Nephrology) and Public Health (Chronic Disease Epidemiology)
Yale University School of Medicine

Disclosure information not submitted.
Richard Pierce, MD (he/him/his)

Yale University School of Medicine, Connecticut

Disclosure information not submitted.

Introduction: Capillary leak is thought to be common and to significantly contribute to death and disability of patients with sepsis, systemic inflammation, ischemia/reperfusion injury, trauma, and many other etiologies. Several clinical scores have been proposed to approximate capillary leak, but none have been validated. Transvascular albumin escape rate (TERalb), the percentage of albumin which escapes the capillary bed per unit time, is an innovative tool to quantify capillary leak in-vivo. We hypothesized clinical scores for capillary leak should have significant association with TERalb.

Methods: We measured blood volume and trans-vascular escape of I-131 tagged albumin in 198 patients hospitalized for acute on chronic heart failure using Blood Volume Analyzer (BVA-100, Daxor, NY). We performed manual chart review to calculate clinical scores for capillary leak nearest each BVA measurement. Descriptive statistics are reported in medians and quartiles due to skewness and were calculated in GraphPad Prism v10.4.1. Clinical scores were tested for association with TERalb using Spearman association.

Results: The Hematocrit-albumin difference was calculated for 117 patients, with a median score of 33.6 (IQR: 30.3–37.8). The A/G ratio (n=116) had a median of 1.21 (IQR: 1.01–1.41). The capillary leak index (n=78) had a median of 3.6 (IQR: 1.5–15). The vascular leak index (n=12) had a median of 1.26 (IQR: -2.95 to 6.96). The percent fluid overload (n=53) had a median of -1.08 (IQR: -2.09 to -0.38). The SOFA score (n=174) had a median of 1 (IQR: 0–2). No significant associations were found between TERalb and Hematocrit-albumin difference (ρ= -0.15, p=0.09), A/G ratio (ρ= -0.08, p=0.41), capillary leak index (ρ= 0.04, p=0.72), vascular leak index (ρ= -0.08, p=0.79), or percent fluid overload (ρ= -0.08, p=0.56). SOFA score showed no association with albumin escape rate (ρ= 0.0004, p=0.99) or any capillary leak score analyzed.

Conclusions: No endpoint analyzed exhibited statistically significant association with TERalb, suggesting our current clinical approximations of capillary leak may be poor diagnostic markers. Future work will be aimed at deriving and validating clinical markers which correlate with capillary leak in critically and non-critically ill patients.