Shunt: pathophysiological mechanisms and clinical implications of ineffective circulating volume

Abstract

The classical concept of shunt confines the phenomenon to the pulmonary circulation, leaving microcirculatory flow maldistribution in distributive shock outside existing terminology. Tissue hypoxia can persist despite normal cardiac output; current hemodynamic stability criteria incorporate no microcirculatory dimension. This review addresses both gaps by proposing a unifying framework. We propose ineffective circulating volume (ICV) as blood flow that fails to achieve effective gas exchange, whether at the alveolar-blood interface (pulmonary shunt) or the blood-cell interface (systemic shunt). Pulmonary shunt and septic microcirculatory heterogeneity are established phenomena, whereas their unification as ICV—particularly the term systemic “functional shunt”—is a hypothesis requiring clinical validation. Anatomical shunt involves fixed structural bypasses correctable surgically; functional abnormalities reflect failure of active distribution mechanisms—hypoxic pulmonary vasoconstriction suppressed by inflammatory mediators, and conducted responses disrupted by endothelial injury. Supportive interventions may improve their physiological consequences, whereas durable reversal depends on resolution or treatment of the underlying disease. The ICV framework subdivides circulatory hypoxia into a macrocirculatory sublayer (global oxygen delivery insufficiency) and a microcirculatory sublayer (flow maldistribution), the latter potentially persisting even when macrocirculatory parameters are normalised. Acute respiratory distress syndrome and distributive shock may share three upstream molecular mechanisms—glycocalyx degradation, spatially heterogeneous inducible nitric oxide synthase overexpression, and neutrophil extracellular traps—that could impair regulatory systems at both interfaces. Changes in sepsis resuscitation targets are compatible with, but do not validate, the ICV framework’s emphasis on macro-microcirculatory uncoupling. No study has simultaneously quantified shunt at both interfaces in the same patient cohort, and no randomised trial targeting the shared upstream mechanisms has used dual-interface ICV as a combined primary endpoint. Translating the ICV framework into prospectively defined clinical endpoints remains an open methodological challenge.

1. Introduction: from venous admixture to ineffective circulating volume

In classical physiology, “shunt” refers to venous blood bypassing functional alveolar-capillary units, entering the left heart without oxygenation and causing hypoxemia—quantified by the shunt fraction (Q_s/Q_t) via the Riley-Cournand model (Radermacher et al., 2017; Albert and Jobe, 2012). This framework has two limitations. First, “shunt” is almost exclusively assigned to the pulmonary circulation, although non-nutritive microcirculatory flow can likewise be ineffective (Elbers and Ince, 2006; Ince and Sinaasappel, 1999). Second, tissue hypoxia may persist when global oxygen delivery is normal or elevated: in distributive shock, cardiac output is often supranormal and mixed venous oxygen saturation (SvO₂) may paradoxically increase (Pries et al., 2010; Ince and Sinaasappel, 1999). We use “systemic shunt” as a functional analogy for this maldistribution, not as accepted nomenclature or a measured shunt fraction equivalent to Q_s/Q_t.

We propose reinterpreting “shunt” through the concept of ineffective circulating volume (ICV)—blood that fails to achieve effective gas exchange, whether at the alveolar-blood interface (pulmonary shunt) or the blood-cell interface (systemic shunt). Cardiac output and effective gas exchange capacity are therefore not interchangeable. ICV is a conceptual framework rather than a directly measured intravascular compartment. It further divides by structural basis into anatomical shunt (fixed abnormalities, potentially correctable by anatomy-specific intervention) and functional shunt (physiological dysregulation whose magnitude or consequences may be modified supportively, while durable reversal generally requires treatment of the underlying disease) (Table 1).

Table 1.

Comparison of pulmonary shunt and systemic shunt.

Category	Pulmonary shunt	Systemic shunt
Interface	Alveolar-blood gas exchange	Blood-cell gas exchange
Mechanism	Anatomical or functional (V/Q=0)	Microcirculatory functional short-circuit
Main consequence	Refractory arterial hypoxemia	Circulatory hypoxia
Typical disease	ARDSa, pulmonary AVMb	Sepsis, distributive shock
Key indices	SaO2↓, PaO2/FiO2↓	Low or inappropriately normal O2ER may suggest extraction impairment; high O2ER suggests global DO2 insufficiency

^aARDS, acute respiratory distress syndrome; ^bAVM, arteriovenous malformation.

2. Pulmonary shunt: pathophysiology, monitoring, and treatment

2.1. Mechanisms and pathophysiology

Pulmonary shunt reduces arterial oxygen content when venous blood enters the left heart without completing O₂–CO₂ exchange. Anatomical shunt results from fixed structural bypasses—patent foramen ovale (PFO), pulmonary arteriovenous malformations (PAVM), and intrapulmonary anastomoses—whose quantitative contributions vary (Homma et al., 2016; Velthuis et al., 2015). Functional shunt is more prevalent in critical care: consolidation, atelectasis, or exudate eliminate alveolar ventilation while capillary perfusion is maintained, forming V/Q=0 zones that are physiologically equivalent to anatomical shunt (Albert and Jobe, 2012; Gierhardt et al., 2021). The hallmark is refractory hypoxemia unresponsive to increased FiO₂.

In normal lungs, hypoxic pulmonary vasoconstriction (HPV) is the key endogenous defence against functional shunt (Figure 1): alveolar hypoxia triggers mitochondrial O₂ sensing, ROS-mediated K_v channel inhibition, membrane depolarization, Ca²⁺ influx, and smooth muscle cell (SMC) contraction, redirecting blood toward ventilated zones (Gierhardt et al., 2021). In ARDS, three convergent mechanisms may impair HPV: (1) iNOS-derived excess NO persistently activates guanylate cyclase-cGMP, causing SMC over-relaxation; (2) PGI₂ excess and leukotriene-mediated dilation oppose HPV; (3) TNF-α and IL-6 inhibit K_v channel function and impair endothelin-1 (ET-1) secretion (Gierhardt et al., 2021). The net effect may be persistent perfusion of non-ventilated alveoli with expansion of V/Q=0 shunt. Pulmonary hypertension in ARDS can additionally recruit PFO right-to-left shunting; transpulmonary bubble transit (TPBT) is detected in approximately 20-26% of cases (Boissier et al., 2015), underscoring that anatomical and functional shunt coexist.

Figure 1.

Normal HPV mechanism and impaired HPV in ARDS. Left: Under normoxia, mitochondrial O₂ sensing maintains K_v channel activity in pulmonary arterial SMCs. Alveolar hypoxia inhibits K_v, triggering depolarization, Ca²⁺ influx, SMC contraction, and blood flow redirection toward ventilated zones (HPV). Endothelial eNOS-derived NO and ET-1 modulate vasomotor tone. Right: In ARDS, three mechanisms may impair HPV: (1) iNOS-derived excess NO causes SMC over-relaxation via guanylate cyclase-cGMP; (2) elevated PGI₂ and leukotrienes promote vasodilation; (3) TNF-α/IL-6 suppress K_v and reduce ET-1. The result may be persistent perfusion of non-ventilated alveoli, expanding V/Q=0 shunt. ARDS, acute respiratory distress syndrome; ET-1, endothelin-1; HPV, hypoxic pulmonary vasoconstriction; iNOS, inducible nitric oxide synthase; K_v, voltage-gated potassium channel; NO, nitric oxide; PGI₂, prostacyclin; ROS, reactive oxygen species; SMC, smooth muscle cell; V/Q, ventilation-perfusion ratio.

2.2. Monitoring of pulmonary shunt

Traditional blood gas indices (PaO₂/FiO₂, P(A-a)O₂, Q_s/Q_t) reflect global gas exchange averages and are confounded by cardiac output, haemoglobin, and FiO₂ (Feiner and Weiskopf, 2017). Electrical impedance tomography (EIT) generates bedside regional ventilation maps and, with specialised protocols, estimates perfusion; pre-prone dependent-zone shunt fraction < 8.4% independently predicts prone responsiveness (sensitivity 94%), and V/Q matching improvement during proning is independently associated with reduced ICU mortality (Leali et al., 2024; Wu et al., 2025; Wang et al., 2025). Bubble contrast echocardiography discriminates intracardiac (≤3 cardiac cycles) from intrapulmonary (>3 cycles) shunting (Boissier et al., 2015), and dual-energy CT (DECT) quantifies gas-blood volume mismatch, which correlates inversely with PaO₂/FiO₂ (Busana et al., 2024). EIT is non-invasive and repeatable but requires dedicated equipment, trained acquisition, and attention to motion and belt-position artefacts; DECT provides greater spatial detail but requires transport, radiation, and contrast exposure (Frerichs et al., 2017). The hierarchical structure of these tools within the ICV framework is shown in Figure 2.

Figure 2.

Hierarchical monitoring framework for pulmonary and systemic shunt within the ICV paradigm. Upper half—Pulmonary shunt: Global indices (PaO₂/FiO₂, P(A-a)O₂, Q_s/Q_t) lack regional resolution. EIT generates real-time V/Q maps. Bubble contrast echocardiography discriminates intracardiac from intrapulmonary shunting. DECT quantifies regional gas-blood mismatch. Lower half—Systemic shunt: Low or inappropriately normal O₂ER with elevated P_v-aCO₂ may support impaired extraction within the proposed ICV phenotype; high O₂ER instead suggests global DO₂ insufficiency. These indices are non-specific. IDF microscopy quantifies PVD, PPV, and HI, while CRT and NIRS-VOT provide adjunctive bedside assessment. Arrow denotes increasing spatial resolution. C(a-v)O₂, arteriovenous oxygen content difference; CRT, capillary refill time; DECT, dual-energy computed tomography; EIT, electrical impedance tomography; HI, heterogeneity index; ICV, ineffective circulating volume; IDF, incident dark-field microscopy; NIRS-VOT, near-infrared spectroscopy with vascular occlusion test; O₂ER, oxygen extraction ratio; P(A-a)O₂, alveolar-arterial oxygen gradient; PPV, proportion of perfused vessels; PVD, perfused vessel density; Q_s/Q_t, shunt fraction; ScvO₂, central venous oxygen saturation; P_v-aCO₂, venoarterial CO₂ difference.

2.3. Treatment of pulmonary shunt

Prone positioning promotes dorsal recruitment and optimises V/Q distribution; early application (≤7 days) reduces mismatch, whereas late application may paradoxically increase shunt (Wang et al., 2022; Yuan et al., 2023). PEEP prevents end-expiratory collapse; its net effect on shunt depends on recruitability— overdistension/recruitment (O/R) ratio <15% typically improves V/Q matching (Karbing et al., 2020). For refractory hypoxaemia, VV-ECMO enables ultra-protective ventilation; prone-ECMO combinations further reduce intrapulmonary shunt (Quintel et al., 2020). Inhaled vasodilators (iNO, prostaglandins) selectively dilate vessels in ventilated zones; almitrine reinforces HPV and improved oxygenation in ∼66% of COVID-19 ARDS patients, though at the cost of right ventricular strain at higher doses (Kaisers et al., 2003; Caplan et al., 2021). Sildenafil worsens shunt fraction and is contraindicated in ARDS (Cornet et al., 2010). Anatomical shunts are managed interventionally (PFO closure, antibiotic prophylaxis for PAVM) or via liver transplantation for hepatopulmonary syndrome (Homma et al., 2016). These interventions may reduce the shunt fraction or its physiological consequences without necessarily reversing the underlying inflammatory lung injury.

Pulmonary shunt depresses systemic DO₂ through reduced arterial oxygen content; compensatory cardiac output increases paradoxically amplify shunt-zone blood volume, further diluting CaO₂. The two interfaces are therefore not hemodynamically independent.

3. Systemic shunt: pathophysiology, monitoring, and treatment

The ICV framework adds a third mechanism to the conventional two macrocirculatory explanations of distributive shock (vasodilation-driven relative hypovolaemia; reduced SVR): microcirculatory flow maldistribution, which may cause tissue hypoxia even when macrocirculatory parameters are normalised (Figure 3) (Elbers and Ince, 2006; Ince, 2005; Vincent and De Backer, 2013). Ince’s concept of “hemodynamic coherence” captures this loss of correspondence between macro- and microcirculatory responses (Ince, 2015). Loss of hemodynamic coherence may help explain why normalising global targets did not improve outcomes in ProCESS, ARISE, and ProMISe, although these trials did not directly test the ICV hypothesis (Yealy et al., 2014; Peake et al., 2014; Mouncey et al., 2015).

Figure 3.

Three pathophysiological mechanisms of distributive shock within the ICV framework. Mechanism 1—Relative hypovolemia: Systemic vasodilation increases venous capacitance, reducing MSFP, stroke volume, and MAP. Mechanism 2—Reduced SVR: Arteriolar vasodilation decreases SVR and MAP without necessarily reducing cardiac output. Mechanism 3—Microcirculatory maldistribution (proposed ICV interpretation): Even when MAP and cardiac output are normalised, perfusion heterogeneity may contribute to loss of hemodynamic coherence and reduce effective tissue perfusion despite adequate global flow. ICV, ineffective circulating volume; MAP, mean arterial pressure; MSFP, mean systemic filling pressure; SVR, systemic vascular resistance.

3.1. Mechanisms of microcirculatory functional shunt

In normal microcirculation, metabolic signals (local hypoxia, adenosine, CO₂) are retrogradely transmitted via conducted responses to upstream arterioles, driving targeted dilation that recruits blood to high-demand zones (Pries et al., 2010; Segal, 2005). Red blood cells additionally act as PO₂ sensors: haemoglobin conformation changes release nitrosothiol and ATP, inducing microvascular dilation (Bateman et al., 2003). In septic shock, this regulatory capacity is disrupted through four superimposed mechanisms (Figure 4): (1) spatially heterogeneous iNOS overexpression drives over-vasodilation in high-expression beds (fast non-nutritive flow) while low-expression areas are underperfused (no-flow zones); (2) glycocalyx degradation by TNF-α-activated heparanase causes syndecan-1 shedding, leukocyte adhesion, and capillary permeability increase; (3) oxidative stress reduces RBC deformability, causing stopped flow at 3-5µm capillary entrances; (4) glycocalyx loss exposes tissue factor, initiating platelet-mediated microthrombus formation (Bateman et al., 2003; Joffre et al., 2020; Uchimido et al., 2019). These mechanisms can reduce perfused vessel density (PVD), increase perfusion heterogeneity, and contribute to loss of hemodynamic coherence.

Figure 4.

Four molecular mechanisms of microcirculatory maldistribution in septic shock. Normal (top left): eNOS-derived NO maintains uniform capillary perfusion matched to O₂ demand. iNOS heterogeneity (top right): Spatially heterogeneous iNOS overexpression drives over-vasodilation in highexpression zones (fast non-nutritive flow) and underperfusion in low-expression zones. Glycocalyx degradation (bottom left): TNF-α activates heparanase; leukocyte adhesion, plugging, and oedema impair nutritive perfusion. RBC rigidity and microthrombi (bottom right): Oxidative stress reduces erythrocyte deformability; VWF-platelet microthrombus formation occludes capillary segments. Together these mechanisms can reduce PVD, increase HI, and contribute to loss of hemodynamic coherence; their interpretation as systemic ICV remains hypothetical. eNOS, endothelial nitric oxide synthase; HI, heterogeneity index; iNOS, inducible nitric oxide synthase; NO, nitric oxide; PVD, perfused vessel density; RBC, red blood cell; TNF-α, tumour necrosis factor-alpha; VWF, von Willebrand factor.

3.2. Monitoring and assessment of systemic shunt

Systemic shunt monitoring combines indirect metabolic indices with direct microcirculation imaging (Figure 2). O₂ER and C(a-v)O₂ are the primary Fick-principle indices; SvO₂/ScvO₂ is a composite that does not directly reflect extraction efficiency (Vallet et al., 2013). In distributive shock, low or inappropriately normal O₂ER despite evidence of hypoperfusion may reflect impaired extraction from flow maldistribution or cytopathic hypoxia, whereas high O₂ER more typically indicates global DO₂ insufficiency. Neither pattern is specific for ICV, and no validated O₂ER threshold diagnoses systemic shunt. P_v-aCO₂ (normal < 6mmHg) reflects whether blood flow is adequate relative to CO₂ production; P_v-aCO₂/C(a-v)O₂ >1.4mmHg·dl/ml has been associated with anaerobic metabolism (Mesquida et al., 2015). Thus, low or inappropriately normal O₂ER with elevated P_v-aCO₂ may support, but cannot establish, the proposed systemic ICV phenotype.

Near-infrared spectroscopy with a vascular occlusion test (NIRS-VOT) measures peripheral tissue oxygen saturation (StO₂); the post-occlusion reoxygenation slope provides an index of microvascular recruitment and reactivity. Abnormal responses have been reported in sepsis and critical illness (Lima et al., 2009; Donati et al., 2016). NIRS-VOT is non-invasive but samples superficial regional tissue and is affected by probe site, tissue thickness, oedema, vasoactive therapy, device, and protocol. It should therefore be considered an adjunct rather than a direct measure of systemic ICV.

Sublingual IDF/SDF microscopy directly quantifies PVD, proportion of perfused vessels (PPV), and heterogeneity index (HI); decreased PVD and elevated HI independently associate with organ dysfunction and mortality, persisting even after macrocirculatory normalisation (Sanfilippo et al., 2022; Aykut et al., 2015). Its routine use is limited by operator-dependent acquisition, sublingual sampling, artefacts, and the absence of validated treatment thresholds (Ince et al., 2018). Capillary refill time (CRT) > 3 seconds associates with sublingual microcirculatory abnormalities; ANDROMEDA-SHOCK evaluated CRT-guided resuscitation (Hernandez et al., 2019a), and the 2026 SSC guidelines suggest CRT as an adjunct to other measures of perfusion (Prescott et al., 2026). However, CRT may remain normal in hyperdynamic distributive shock despite substantial microcirculatory maldistribution (Kanoore Edul et al., 2016)—within the ICV framework, a normal CRT does not exclude elevated systemic ICV.

3.3. Treatment of systemic shunt

Tissue hypoxia from systemic shunt is spatially heterogeneous: bypassed zones accumulate lactate anaerobically while hyperperfused zones may fail to extract oxygen due to cytopathic hypoxia (Elbers and Ince, 2006). Persistently elevated lactate despite adequate DO₂ may be consistent with this pattern, but adrenergic glycolysis and impaired clearance are alternative explanations. Three large-scale RCTs (ProCESS, ARISE, ProMISe; n = 4,183 combined) showed ScvO₂-targeted EGDT conferred no benefit over standard care (Yealy et al., 2014; Peake et al., 2014; Mouncey et al., 2015). Microcirculatory maldistribution is one possible interpretation of the limited specificity of ScvO₂, but these trials did not test the ICV framework. Current SSC guidance instead supports serial lactate, dynamic flow assessment, and CRT as an adjunct (Evans et al., 2021; Prescott et al., 2026).

When microcirculatory functional shunt is established, additional fluid may enter non-nutritive pathways and worsen tissue oedema (Ince, 2015; Hjortrup et al., 2016); vasopressors can restore perfusion pressure but may not correct spatial maldistribution (Hernandez et al., 2019b). Dobutamine may improve sublingual capillary perfusion independently of cardiac index changes, though responses are heterogeneous (De Backer et al., 2006). Trials of microcirculatory vasodilators (nitroglycerin, inhaled NO) have not established outcome benefit (Spronk et al., 2002; Boerma et al., 2010). Supportive treatment may improve pressure, flow, or the consequences of maldistribution, whereas source control and treatment of the underlying inflammatory process address its upstream drivers. Whether P_v-aCO₂/C(a-v)O₂, CRT, and PVD can be combined into a composite resuscitation endpoint remains untested.

4. Shared mechanisms and functional analogy at both interfaces

ARDS and distributive shock share a functional analogy: ARDS generates V/Q=0 mismatch (ventilation absent, perfusion preserved), whereas distributive shock can generate perfusion-without-extraction mismatch (flow present, metabolic coupling impaired). The systemic “pO₂ gap” between intestinal microvascular and venous oxygen partial pressures observed in endotoxaemia is analogous to, but not quantitatively equivalent to, the pulmonary P(A-a)O₂ gap (Ince and Sinaasappel, 1999). When both coexist in sepsis-associated ARDS, abnormalities at the two interfaces may compound impaired oxygen transfer and utilization.

4.1. HPV and conducted response as parallel regulatory mechanisms

HPV and the systemic microvascular conducted response provide a functional regulatory analogy. HPV uses alveolar hypoxia to withdraw blood from ineffectively ventilated zones, whereas the conducted response uses tissue hypoxia, adenosine, and CO₂ to recruit blood to high-metabolic-demand zones (Pries et al., 2010; Segal, 2005). In sepsis, iNOS overexpression and glycocalyx degradation may impair both processes. Supportive interventions can modify flow distribution or its consequences, while durable recovery of the regulatory lesion depends on the course and treatment of the primary inflammatory injury.

4.2. Shared molecular mechanisms

Three molecular mechanisms may affect both interfaces. Glycocalyx degradation: TNF-α-activated heparanase drives syndecan-1 shedding; pulmonary consequences include neutrophil adhesion, alveolarcapillary permeability amplification, and ARDS; systemic consequences include capillary leak and functional short-circuit opening (Uchimido et al., 2019). Elevated plasma syndecan-1 is associated with ARDS, circulatory failure, and mortality (Colbert and Schmidt, 2016; Zinter et al., 2018). Heterogeneous iNOS overexpression: spatially heterogeneous iNOS activation may contribute to V/Q maldistribution pulmonarily and microcirculatory perfusion maldistribution systemically (Gierhardt et al., 2021; Rovas et al., 2019). Neutrophil extracellular traps (NETs): pulmonary NETs promote microthrombus formation and alveolar-capillary barrier disruption; systemic NETs cause microvascular occlusion via the plateletthrombin axis (Lefranc¸ais et al., 2018). PAD4 deficiency or DNase treatment reduces NET-mediated vascular injury in experimental models, supporting rather than proving a mechanism spanning both interfaces (McDonald et al., 2017).

These observations suggest that ARDS and distributive shock may share upstream pathways. No randomised trial targeting glycocalyx integrity, selective iNOS inhibition, or NETs degradation has adopted dual-interface ICV as a combined primary endpoint.

The proposed equivalence has important limitations. “Systemic functional shunt” is not widely accepted terminology, and low oxygen extraction, elevated venous saturation, or hyperlactataemia can also reflect low metabolic demand, mitochondrial dysfunction, adrenergic glycolysis, or impaired clearance. Moreover, IDF/SDF and NIRS sample regional tissue, and no validated bedside measure quantifies a systemic ICV fraction. The framework should therefore be regarded as a testable hypothesis rather than a diagnostic or therapeutic target.

5. Conclusions and perspectives

The ICV framework rests on two distinctions. First, cardiac output and effective gas exchange capacity are not interchangeable: pulmonary shunt or microcirculatory maldistribution may persist while total flow is preserved. Second, anatomical and functional abnormalities differ in their susceptibility to intervention: supportive measures may improve the latter’s physiological consequences without correcting their upstream cause.

Validation should proceed in stages. A prospective methods study should first establish reproducibility and confounding of candidate systemic measures such as PVD, HI, NIRS-VOT, O₂ER, CRT, and CO₂-derived indices. A subsequent multicentre cohort should obtain time-synchronised pulmonary shunt and systemic microcirculatory measurements to test convergent validity, temporal responsiveness, and association with organ dysfunction. Only after a reproducible construct is defined should an interventional trial evaluate dual-interface measures, initially as mechanistic secondary endpoints.

The ICV framework’s principal utility is terminological—a single label for blood failing to participate in gas exchange at either interface. Clinical translation requires quantification: pulmonary ICV fraction has an established surrogate in Q_s/Q_t; systemic ICV fraction lacks an equivalent bedside measure. Developing such an index is the unresolved prerequisite for transitioning ICV from a conceptual construct to a prospectively testable endpoint.

Glossary

ARISE

Australasian Resuscitation in Sepsis Evaluation

ARDS

acute respiratory distress syndrome

ATP

adenosine triphosphate

C(a-v)O₂

arteriovenous oxygen content difference

CRT

capillary refill time

DECT

dual-energy computed tomography

EIT

electrical impedance tomography

EGDT

early goal-directed therapy

ET-1

endothelin-1

FiO₂

fraction of inspired oxygen

HI

heterogeneity index

HPV

hypoxic pulmonary vasoconstriction

ICV

ineffective circulating volume

IDF

incident dark-field

iNO

inhaled nitric oxide

iNOS

inducible nitric oxide synthase

K_v

voltage-gated potassium channel

MAP

mean arterial pressure

NETs

neutrophil extracellular traps

NIRS

near-infrared spectroscopy

NIRS-VOT

near-infrared spectroscopy with vascular occlusion test

NO

nitric oxide

O₂ER

oxygen extraction ratio

P(A-a)O₂

alveolar-arterial oxygen gradient

PAD4

peptidylarginine deiminase 4

PaO₂

arterial partial pressure of oxygen

PAVM

pulmonary arteriovenous malformation

PEEP

positive endexpiratory pressure

PFO

patent foramen ovale

PGI₂

prostacyclin

PPV

proportion of perfused vessels

ProCESS

Protocolized Care for Early Septic Shock

ProMISe

Protocolised Management in Sepsis

P_v-aCO₂

venoarterial CO₂ difference

PVD

perfused vessel density

RBC

red blood cell

ScvO₂

central venous oxygen saturation

SDF

sidestream dark-field

SSC

Surviving Sepsis Campaign

StO₂

tissue oxygen saturation

SVR

systemic vascular resistance

SvO₂

mixed venous oxygen saturation

TPBT

transpulmonary bubble transit

V/Q

ventilation-perfusion ratio

VOT

vascular occlusion test

VV-ECMO

venovenous extracorporeal membrane oxygenation.

Funding Statement

The author(s) declared that financial support was received for this work and/or its publication. This work was supported by the Zhejiang Provincial Traditional Chinese Medicine Science and Technology Program [grant number 2025ZR124] (ZF).

Author contributions

SM: Conceptualization, Investigation, Writing – original draft. JY: Visualization, Writing – review & editing. ZF: Investigation, Funding acquisition, Writing – review & editing. ZG: Writing – review & editing. ZZ: Supervision, Writing – review & editing.

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declared that generative AI was used in the creation of this manuscript. During the preparation of this work, the authors used Claude (Anthropic) in order to generate the initial LaTeX source file structure and resolve compilation errors during manuscript typesetting, and used ChatGPT (OpenAI) in order to improve the language fluency and grammatical accuracy of the English manuscript. After using these tools, the authors reviewed and edited the content as needed and take full responsibility for the content of the published article.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.