Septic shock remains a leading cause of mortality in intensive care, and while fluid resuscitation (FR) is a cornerstone of early management, its benefit-risk balance is highly variable. Excessive fluid administration can cause venous congestion and organ dysfunction, while insufficient resuscitation risks hypoperfusion. Current strategies often rely on fluid responsiveness (i.e., increased cardiac output after fluids), but this does not guarantee improved outcomes, particularly if congestion ensues.

This prospective, multicenter, observational study aims to assess the clinical impact of FR in septic shock patients who are fluid responsive. The primary objective is to evaluate changes in tissue perfusion and venous congestion markers following FR. Patients will be categorized into four response profiles based on the presence or absence of perfusion improvement and congestion worsening.

Secondary objectives include exploring the prognostic implications of each profile (organ dysfunction, mortality), identifying pre-FR predictors of adverse responses, evaluating changes in congestion markers after passive leg raising (PLR), and performing phenotypic clustering and mediation analyses.

Eligible patients are adults with septic shock requiring vasopressors and mechanical ventilation, with confirmed fluid responsiveness via echocardiography. Each patient will undergo standardized pre- and post-FR assessments, including cardiac ultrasound, Doppler of hepatic/portal veins (VeXUS), CVP, perfusion markers, and blood gases.

Data on SOFA scores, organ support duration, and 28-day mortality will be collected. Approximately 170 patients will be enrolled across five ICUs experienced in advanced hemodynamic monitoring. Statistical analyses will include multivariate modeling, clustering, ROC curves, and mediation analyses.

By identifying phenotypes of fluid-responsive but fluid-intolerant patients, the study aims to refine fluid management strategies and improve outcomes through more personalized care in septic shock.

Background and Rationale Sepsis and septic shock are among the leading causes of morbidity and mortality in intensive care units worldwide. Septic shock is defined as a subset of sepsis characterized by profound circulatory, cellular, and metabolic abnormalities, associated with higher risk of death. One of the cornerstone interventions in the early management of septic shock is fluid resuscitation (FR), which aims to restore tissue perfusion and oxygen delivery through an increase in cardiac output.

However, the benefit-risk balance of fluid administration remains a major challenge. While inadequate fluid resuscitation can perpetuate hypoperfusion, organ ischemia, and multi-organ failure, excessive fluid administration has been consistently associated with increased morbidity and mortality. This paradox stems largely from venous congestion and tissue edema caused by fluid overload, which may impair oxygen diffusion and organ function.

Several large randomized trials have failed to show a survival benefit of one fluid strategy over another, potentially due to heterogeneous patient profiles and a lack of individualized approaches. These studies typically did not consider patients' baseline volume status, nor did they incorporate markers of venous congestion when guiding fluid administration.

An emerging concept in critical care is that of fluid tolerance, which reflects a patient's ability to receive intravenous fluids without developing the harmful effects of fluid overload, such as venous congestion or capillary leakage. However, assessing fluid tolerance remains difficult due to the absence of standardized tools, thresholds, or clinical guidelines.

Traditionally, fluid responsiveness-defined as a significant increase in cardiac output following fluid administration-has been the main criterion used to determine the appropriateness of FR. Yet, growing evidence suggests that being fluid responsive does not guarantee clinical benefit, especially if venous congestion worsens post-resuscitation.

Ultrasound-based venous congestion assessment tools, particularly the VeXUS (Venous Excess Ultrasound Score) protocol, have gained attention for their potential to identify fluid-intolerant patients. These include analysis of hepatic and portal vein Doppler waveforms, along with central venous pressure (CVP). Despite promising associations with renal injury and mortality, these tools have not yet been evaluated prospectively in the context of fluid challenge in septic shock patients.

The primary objective of the study will be to evaluate the evolution of tissue perfusion markers and venous congestion indices after fluid resuscitation in patients with septic shock who are deemed fluid responsive (i.e., with increased cardiac output after FR).

Patients will be classified into a four-category classification: Improvement in tissue perfusion markers without worsening of congestion markers / Improvement in tissue perfusion markers with worsening of congestion markers / No improvement in tissue perfusion markers without worsening of congestion markers / No improvement in tissue perfusion markers with worsening of congestion markers.

Tissue perfusion improvement will be defined as an improvement in at least one of the following criteria: decrease in mottling score by ≥1 point / Reduction in lactate level by >10% if initial value >2 mmol/L / Increase in central venous oxygen saturation (≥ 4%) / Improvement in the arteriovenous pCO₂ gradient (≥ 2 mmHg) / improvement in capillary refill time (≥ 1 second if > 3 seconds).

Venous congestion worsening will be defined by the presence of at least two of the following criteria: worsening of portal or hepatic vein pulsatility index (VeXUS grading) / CVP > 8 mmHg Increase in CVP (ΔCVP) > 2 mmHg

This classification will enable the identification of patients who, despite being fluid responsive, do not improve perfusion and show increased congestion-a group that may derive no net benefit or even harm from fluid administration.

Secondary Objectives:

To evaluate the prognostic implications (organ dysfunction and mortality) of these four response profiles.

To assess predictive factors, available before fluid administration, associated with adverse profiles.

To explore the evolution of congestion markers after a passive leg raising (PLR) test.

To investigate the relationship between venous congestion and organ support duration, SOFA scores, and 28-day mortality.

To perform phenotypic clustering analyses independent of the predefined classification.

To assess the mediating role of venous congestion in the relationship between fluid volume and organ failure.

Study Design and Methodology This is a prospective, multicenter, observational cohort study conducted in five intensive care units (ICUs) with prior experience in advanced hemodynamic monitoring and echocardiographic research.

Adults patients admitted to the ICU for septic shock (SEPSIS-3 definition), under mechanical ventilation and equipped with a central venous catheter (superior vena cava territory) and an arterial catheter and demonstrating fluid responsiveness (≥10% increase in stroke volume or cardiac output measured via echocardiography after fluid challenge) will be included.

Study Protocol:

Each patient will undergo two standardized evaluations: before and 30-60 minutes after fluid resuscitation. These will include:

Echocardiography (to assess cardiac output)

Doppler ultrasound of portal and hepatic veins (to evaluate venous congestion)

Central venous pressure (CVP)

Clinical perfusion markers: capillary refill time, mottling score

Gasometric markers: lactate, ScvO₂, pCO₂ gap

Additional data:

SOFA score and fluid balance at 24h

Organ support duration (mechanical ventilation, vasopressors, renal replacement therapy)

Mortality at Day 28

Patient Classification:

Based on post-FR changes, patients will be grouped into:

Improved perfusion / No worsening congestion

Improved perfusion / Worsening congestion

No improvement / No worsening congestion

No improvement / Worsening congestion

Outcome Measures:

Changes in SOFA score (renal, respiratory, cardiovascular) at 24h and Day 28

Duration of organ support

28-day mortality

Identification of predictive factors for high-risk profiles using multivariate logistic regression

Clustering and PCA to detect fluid response phenotypes

Mediation analysis to determine if venous congestion mediates the effect of fluid resuscitation on organ dysfunction

Sample Size and Statistical Analysis Based on preliminary data suggesting that ~40% of fluid responders exhibit venous congestion post-resuscitation, a total of 170 patients will be enrolled. This will ensure a minimum of 25 patients in the smallest of the four classification groups, sufficient for comparative analysis.

Analysis Plan:

Descriptive statistics of clinical and hemodynamic variables

Intergroup comparisons using:

Student's t-test or Wilcoxon test (continuous variables)

Chi² or Fisher's exact test (categorical variables)

Multivariable models:

Logistic regression for binary outcomes (e.g., mortality)

Generalized linear models for continuous outcomes (e.g., SOFA scores)

Cox proportional hazards models for survival analysis

Use of LASSO or stepwise selection for predictive modeling

ROC analysis to evaluate model discrimination (AUC with 95% CI)

Clustering or PCA for exploratory identification of phenotypes

Mediation analysis to explore causality between fluid volume, congestion, and organ dysfunction

All analyses will be performed using R statistical software, with significance set at p < 0.05.

Operational Feasibility and Ethics All participating ICUs are experienced in advanced hemodynamic and echocardiographic research (e.g., HEMOPRED, PRODIASYS).

Study classified as low-risk interventional research (RIPH 3).

Echocardiography and congestion assessment via ultrasound are already routine practices in these centers.

Non-opposition consent will be obtained from patients' next of kin.

Follow-Up:

Patients will be followed until ICU discharge or Day 28, whichever occurs later.

Clinical Relevance and Innovation This study addresses an unmet need in precision fluid therapy for septic shock. By combining hemodynamic, ultrasound, and clinical perfusion data, it moves beyond the traditional binary model of fluid responsiveness and seeks to establish a risk-benefit stratification for fluid administration.

The identification of high-risk phenotypes-fluid responsive but intolerant patients-will inform future individualized treatment strategies and potentially guide therapeutic trials using phenotypically driven protocols. The ultimate goal is to improve clinical outcomes by avoiding unnecessary or harmful fluid administration in vulnerable patients.

Inclusion Criteria:

• Septic shock (sepsis-3 criteria)
• Under mechanical ventilation
• Central venous catheter in the superior vena cava territory and arterial catheter in place
• Response to fluid resuscitation as per standard definition (i.e., an increase in cardiac output >10% assessed by subaortic velocity time integral [VTI] using echocardiography)

Exclusion Criteria:

• Patient under 18 years of age
• Lack of social security coverage / adult under legal protection
• Pregnant woman
• Inability to obtain non-opposition consent

Patients in septic shock, defined according to the Sepsis-3 criteria (7), who are under mechanical ventilation, have a central venous catheter in the superior vena cava territory and an arterial catheter, present with hypotension requiring vasopressor support and hyperlactatemia, and demonstrate a response to fluid resuscitation as per standard definition (i.e., an increase in cardiac output >10% assessed by subaortic velocity time integral [VTI] using echocardiography).