Hemodynamic-metabolic phenotyping of septic shock: a single-center observational study

ABSTRACT Introduction. Septic shock is a heterogeneous syndrome in which conventional severity scores fail to capture the underlying diversity of oxygen transport and utilization. Bedside- measurable variables such as central venous oxygen saturation (ScvO2) and the venous-arterial PCO2 gap (DpCO2) reflect complementary aspects of the oxygen delivery-consumption balance. This study aimed to determine whether phenotypes derived from ScvO2 and DpCO2 identify septic shock subgroups with distinct physiological and prognostic profiles, and whether an oxygen conductance-based approach better characterizes the dominant site of resistance to oxygen transport. Materials and Methods. We performed a retrospective, single-center observational study of 199 adult patients admitted with septic shock to our ICU. Using ScvO2 (threshold 70%) and DpCO2 (threshold 6 mmHg), four hemodynamic-metabolic phenotypes were identified at baseline: Anemic-Micro (n=43), Cytopathic (n=73), Cardiogenic (n=52), and Cardio-Cyto (n=31). In parallel, applying the oxygen conductance equation to arterial and venous blood gas data, patients were classified into three resistive phenotypes: Fl (lung-dominant, n=105), Fq (cardiovascular- dominant, n=51), and Ft (tissue-dominant, n=43). Both classifications were applied at baseline and day 2. Continuous variables were reported as median (IQR) and compared using the Kruskal-Wallis test; categorical variables with Pearson's chi-square or Fisher's exact test. Results. Hemodynamic-metabolic phenotypes differed significantly in ICU mortality (p=0.041): Anemic-Micro (60.5%) and Cytopathic (56.2%) showed the highest rates, versus Cardiogenic (36.5%) and Cardio-Cyto (38.7%). SOFA and SAPS II were comparable across all phenotypes. DpCO2 emerged as a stronger prognostic discriminant than ScvO2: phenotypes with low DpCO2 carried the highest mortality despite apparently reassuring venous saturation, while lactate and pH were paradoxically most altered in Cardio-Cyto, the phenotype with the lowest mortality. Resistive phenotypes were strongly associated with metabolic phenotype distribution (p<0.001): Ft was markedly enriched in Cytopathic patients (65.1%), while Fq predominated in Cardiogenic patients (43.1%). The Fl phenotype was uniformly distributed across all metabolic phenotypes. ICU mortality differed significantly across resistive phenotypes (Fl 59.0%, Ft 44.2%, Fq 33.3%; p=0.008). Fl-dominant patients showed the lowest P/F ratio (120.2), the strongest association with pulmonary infection (60%), and the longest ICU stay (median 13 days); 81% remained Fl-dominant at day 2, with worsening outcomes on persistence. Conclusions. Both classifications identify septic shock subgroups with distinct physiological profiles and clinically meaningful mortality differences, independent of conventional severity scores. The uniform distribution of Fl across metabolic phenotypes confirms that ScvO2 and DpCO2 — being post-pulmonary venous parameters — are inherently blind to pulmonary-level resistance. A convergent pattern across both classifications suggests that deficits in oxygen supply carry higher short-term mortality than deficits in oxygen utilization. Importantly, the phenotypic profiles corresponding to conventional ICU resuscitation targets (ScvO2 ≥70%, DpCO2 <6 mmHg) align with the Cytopathic and Ft-dominant groups, which nonetheless retain high mortality, highlighting that achieving standard endpoints does not guarantee adequate tissue oxygenation in the presence of cellular dysoxia or mitochondrial dysfunction.