Circulating Endothelial Transcriptomic Signatures Predict Worse Outcomes in COVID-19, Respiratory Failure and ARDS

Highlight

– Elevated circulating endothelial signatures (ECS%) derived by unsupervised bulk-transcriptome deconvolution are higher in non-survivors of respiratory failure in a pediatric cohort (CAF-PINT) and in adults hospitalized with COVID-19 (IMPACC).

– In IMPACC, each 1% increment in baseline ECS% was independently associated with 28‑day mortality (adjusted OR 1.36, 95% CI 1.03–1.79).

– Higher baseline ECS% tracked with worse 28‑day respiratory trajectories, supporting a link between circulating endothelial signal and progression to severe respiratory failure/ARDS.

Background: clinical context and unmet need

Endothelial injury is a central pathological feature of acute respiratory distress syndrome (ARDS) and severe COVID‑19. Histopathologic studies in COVID‑19 have documented endotheliitis, microvascular thrombosis, and abnormal angiogenesis in the lung, implicating the pulmonary endothelium in disease progression and adverse outcomes. Accurate, non‑invasive markers that detect endothelial damage early and predict clinical trajectories could refine risk stratification, guide monitoring, and identify patients for endothelial‑targeted interventions. Traditional approaches to enumerate circulating endothelial cells (CECs) have been limited by technical heterogeneity and lack of a consensus proteomic ‘signature’ that robustly defines endothelial identity in peripheral blood.

Study design and methods

This study applied an unsupervised bulk-transcriptome deconvolution approach to quantify an endothelial cell transcriptomic signature (ECS%) in peripheral blood across two complementary cohorts. The pilot cohort (CAF‑PINT, NCT01892969) included pediatric patients requiring invasive mechanical ventilation. The validation cohort was the NIH‑supported IMPACC study (NCT04378777), an adult multicenter inpatient cohort of patients hospitalized with COVID‑19. Key features:

• Approach: Unsupervised deconvolution of bulk blood transcriptomes to estimate the proportion of endothelial-related transcripts (ECS%).
• Timing: CAF‑PINT day 0 (presumably at enrollment/first sample); IMPACC baseline defined as within 72 hours of hospitalization.
• Primary endpoint: 28‑day mortality. Secondary analyses included associations with predefined respiratory trajectories (from no oxygen requirement through fatal outcome by day 28).
• Statistical methods: Nonparametric comparisons (Wilcoxon rank‑sum) for group differences; one‑way ANOVA for comparisons across respiratory trajectory groups; multivariable logistic regression to adjust for potential confounders and estimate odds ratios per 1% ECS% increment.

Key findings and detailed results

Overall, the deconvolution‑derived ECS% was reproducibly higher among patients who experienced fatal outcomes in both cohorts, supporting a biologically consistent association between circulating endothelial transcriptomic signal and poor prognosis.

Pediatric cohort (CAF‑PINT)

At day 0, ECS% was modestly but significantly higher in non‑survivors compared with survivors of respiratory failure: median 2.8% (IQR 2.4–3.4%) versus 2.6% (IQR 2.2–3.0%), n = 244; p < 0.05 by Wilcoxon rank‑sum. The finding suggests that an increased endothelial transcriptomic footprint in blood at the time of invasive ventilation is associated with mortality in a critically ill pediatric population.

Adult COVID‑19 cohort (IMPACC)

IMPACC provided an independent, larger validation sample (n = 932). Baseline ECS% within 72 hours of hospitalization was higher in non‑survivors than survivors: median 2.9% (IQR 2.6–3.4%) versus 2.7% (IQR 2.3–3.1%); p < 0.001. In multivariable logistic regression controlling for relevant covariates, each 1% increase in baseline ECS% was associated with higher odds of 28‑day mortality (adjusted OR 1.36, 95% CI 1.03–1.79). This effect size implies a clinically relevant gradient of risk for relatively small absolute increases in the estimated endothelial signal.

Association with respiratory trajectories

Baseline ECS% also tracked with predefined respiratory outcome trajectories: patients who required no oxygen had the lowest ECS% (median 2.5%, IQR 2.2–2.8%), whereas the trajectory culminating in death by day 28 had the highest ECS% (median 2.9%, IQR 2.6–3.4%); n = 932, p < 0.001 by one‑way ANOVA. This graded relationship supports a dose–response pattern between endothelial transcriptomic burden and worsening respiratory course.

Interpretation of effect sizes

Absolute differences in median ECS% between survivors and non‑survivors are small in percentage terms (≈0.2–0.3%), but the adjusted odds ratio per 1% increment (1.36) suggests that even modest relative increases reflect meaningful biological and clinical risk in aggregate. Given the narrow range of ECS% values observed, standardized measurement and careful consideration of analytical variability will be critical for clinical translation.

Expert commentary and mechanistic considerations

Biological plausibility is strong: endothelial disruption in the pulmonary microvasculature contributes to alveolar flooding, loss of barrier integrity, coagulopathy, and impaired oxygenation that characterize ARDS. Pathology studies in COVID‑19 have shown endothelial inflammation, thrombosis, and vascular remodeling (e.g., Ackermann et al., NEJM 2020; Varga et al., Lancet 2020), establishing a mechanistic backdrop for bloodborne endothelial signals.

What might ECS% represent biologically? The deconvolution signal could arise from multiple processes: intact circulating endothelial cells detached from injured endothelium, endothelial cell fragments or microparticles containing RNA, or endothelial-like transcriptional programs induced in other circulating cells (transcriptional mimicry). Bulk transcriptomic deconvolution estimates relative transcript contributions rather than cell counts and therefore integrates signal from intact cells and extracellular RNA. Consequently, ECS% should be regarded as a circulating endothelial transcriptomic signature rather than a direct count of viable CECs.

Strengths of the study include replication across pediatric and adult populations, use of an unsupervised transcriptomic approach that avoids reliance on a single proteomic marker, and clinically meaningful endpoints. The observed independent association across a large multicenter COVID‑19 cohort strengthens external validity within hospitalized populations.

Limitations and cautions

Key limitations temper immediate clinical application:

• Deconvolution inference: Bulk deconvolution methods infer relative contributions of cell‑type specific transcripts but cannot definitively distinguish intact circulating endothelial cells from endothelial RNA carried in extracellular vesicles or platelets.
• Effect magnitude and analytical variability: The absolute differences in ECS% are small; assay precision, preanalytic factors (collection, processing), and batch effects could impact measurement reliability. Standardization will be essential.
• Confounding and causality: Although adjusted models were used, residual confounding remains possible; the association is not proof of causality. Elevated ECS% may be a marker of overall illness severity, endothelial perturbation, or other systemic processes that mediate worse outcomes.
• Generalizability: IMPACC focused on hospitalized COVID‑19 patients; results may not generalize to non‑COVID ARDS subtypes, milder illness, or outpatient settings without further study.
• Absence of orthogonal validation: Correlation with independent endothelial biomarkers (e.g., circulating endothelial cell counts by immunophenotyping, soluble thrombomodulin, von Willebrand factor, endothelial microparticles) and tissue correlates would strengthen mechanistic inference.

Clinical and research implications

These results establish circulating endothelial transcriptomic signatures as a promising non‑invasive biomarker that correlates with mortality and respiratory deterioration in severe respiratory illnesses, including COVID‑19. Potential applications include:

• Prognostic enrichment: Integrating ECS% into multivariable risk models could improve early risk stratification for patients at risk of progression to ARDS or death.
• Patient selection and monitoring in trials: ECS% may identify patients more likely to benefit from endothelial‑directed therapies (anticoagulation strategies, anti‑inflammatory or endothelial‑stabilizing agents) and serve as a pharmacodynamic marker.
• Pathophysiology research: Longitudinal profiling of ECS% alongside proteomic and cellular endothelial markers could clarify the mechanisms linking endothelial injury to respiratory failure and identify therapeutic targets.

Next steps and recommendations

Priorities for translation and further research include:

• Technical validation: Assess intra‑ and inter‑assay reproducibility, preanalytic stability, and normalization strategies for ECS% quantification.
• Orthogonal correlation: Correlate ECS% with flow cytometric CEC counts, endothelial microparticles, soluble biomarkers (e.g., thrombomodulin, angiopoietin‑2), and pulmonary histopathology when available.
• Longitudinal studies: Characterize temporal dynamics of ECS% from early infection through recovery to determine windows of maximal prognostic utility and to study response to therapies.
• Broader cohorts: Test generalizability across non‑COVID ARDS, sepsis, and other causes of endothelial injury in both adult and pediatric populations.
• Clinical utility studies: Evaluate whether ECS% measurement alters clinical decision‑making or outcomes when embedded in prospective management algorithms or trials.

Conclusion

Unsupervised deconvolution of blood transcriptomes revealed a reproducible circulating endothelial signature that correlates with 28‑day mortality and adverse respiratory trajectories in pediatric respiratory failure and adults hospitalized with COVID‑19. While the approach provides a promising, non‑invasive window into endothelial perturbation, further technical validation, mechanistic correlation, and prospective clinical evaluation are required before routine clinical deployment. This work represents an important first step toward leveraging circulating transcriptomics to quantify endothelial involvement in ARDS and severe viral respiratory disease.

Funding and clinicaltrials.gov

The primary manuscript lists funding sources and the IMPACC Network; the cohorts are registered as CAF‑PINT (NCT01892969) and IMPACC (NCT04378777). See the cited article for detailed funding disclosures and investigator lists.

References

1. Costa Monteiro AC, Pickering H, Sarma A, Taylor CS, Jenkins MM, Hsu FM, Nadel B, Levy O, Baden LR, Melamed E, Ehrlich LIR, McComsey GA, Sekaly RP, Cairns CB, Haddad EK, Shaw AC, Hafler DA, Montgomery RR, Corry DB, Kheradmand F, Atkinson MA, Brakenridge SC, Higuita NIA, Metcalf JP, Hough CL, Messer WB, Pulendran B, Nadeau KC, Davis MM, Geng LN, Fernandez‑Sesma A, Simon V, Krammer F, Kraft M, Bime C, Calfee CS, Erle DJ, Bosinger S, Eckalbar W, Maecker H, Rahman A, Guan L, Peters B, Kleinstein SH, Augustine AD, Diray‑Arce J, Becker PM, Rouphael N; IMPACC Network; CAFPINT Site investigators of the PALISI Network; Agus M, Kulkarni H, Schaenmann JM, Salehi‑Rad R, Matthay MA, Reed EF, Sapru A. Circulating endothelial signatures correlate with worse outcomes in COVID‑19, respiratory failure and ARDS. Crit Care. 2025 Oct 14;29(1):432. doi: 10.1186/s13054-025-05596-0. PMID: 41088445; PMCID: PMC12522733.

2. Ranieri VM, Rubenfeld GD, Thompson BT, et al.; ARDS Definition Task Force. Acute respiratory distress syndrome: the Berlin Definition. JAMA. 2012;307(23):2526–2533.

3. Ackermann M, Verleden SE, Kuehnel M, et al. Pulmonary vascular endothelialitis, thrombosis, and angiogenesis in Covid‑19. N Engl J Med. 2020;383:120–128.

4. Varga Z, Flammer AJ, Steiger P, et al. Endothelial cell infection and endotheliitis in COVID‑19. Lancet. 2020;395(10234):1417–1418.

Thumbnail image prompt

A clinician in full PPE standing at a bedside, holding a tablet that displays a colorful transcriptomic heatmap and endothelial cell schematic overlay; in the background a semi‑transparent chest X‑ray with bilateral opacities and stylized microvascular network, color palette cool blues with red highlights to suggest vascular injury. High realism, editorial medical illustration style.