Temporal Dissociation Between Intravascular Albumin Mass and Transcapillary Escape Dynamics in Sepsis: A Longitudinal Characterization of Albumin Homeostasis Using Routine Laboratory Parameters

Background: In sepsis, albumin homeostasis is altered by capillary leak and recovery mechanisms (synthesis and interstitial–lymphatic return), which are difficult to capture clinically. A joint evaluation of ratio-based escape dynamics and mass-based balance may clarify their temporal interplay in sepsis. Methods: In a prospective longitudinal cohort of 389 sepsis patients admitted to an intermediate medical care unit, serial daily sampling (up to five reassessments; 1897 observations) was used to derive: a transcapillary escape rate (TER)-like index from the hourly percent change in the albumin/hemoglobin ratio and net albumin leakage (NAL) from changes in intravascular albumin mass normalized to plasma volume. Indices were analyzed as continuous measures and by sign-defined combined states. Generalized estimating equations (GEE), patient-level slope analyses, and state-transition analyses were performed. Baseline SOFA and 30-day mortality were explored as effect modifiers. Results: The TER-like index peaked early (median +0.34%/h at day 1) and declined to negative values by day 5 (median −0.07%/h), with the TER-like index > 0 decreasing from 65.3% to 45.8%. NAL was frequently negative and heterogeneous (NAL ≥ 0 in 37.5% of observations). TER-like index and NAL were inversely correlated (ρ = −0.54; p < 0.001). In GEE, the TER-like index was associated with NAL (β = −7.46 g/L per 1%/h; 95% CI −8.69 to −6.22; p < 0.001) with time-varying effects (interaction p < 0.01). The dissociation state (TER-like index ≤ 0 with NAL < 0) increased from 2.3% at day 1 to 22.6% at day 5 (p < 0.001). Conclusions: TER-like index escape dynamics attenuate early; however, net intravascular albumin balance often remains negative, supporting temporal uncoupling between “leak” and recovery. An integrated TER-like/NAL assessment offers a pragmatic framework for the clinical phenotyping of albumin homeostasis using routine laboratory data.