Highlights
-
•
In a large, international, multicenter registry of 621 patients, admission organ perfusion pressure (OPP), defined as mean arterial pressure-central venous pressure, was a potent and independent predictor of in-hospital mortality in cardiogenic shock.
-
•
An admission OPP < 57 mmHg was associated with a 3.2-fold increase in the risk of death, providing a specific hemodynamic target for early risk stratification.
-
•
The prognostic value of the 57 mmHg threshold remained consistent across diverse shock phenotypes, including both acute myocardial infarction and chronic heart failure etiologies.
-
•
OPP remained a significant independent predictor (OR 1.03 per mmHg decrease) even after rigorous multivariable adjustment.
-
•
Given its ease of calculation and physiological robustness, OPP serves as a practical, scalable metric to identify subclinical hypoperfusion and guide clinical escalation protocols in the intensive care unit.
ABSTRACT
Background
The organ perfusion pressure (OPP) is a surrogate for end-organ hypoperfusion and was shown to predict outcomes in a multicenter cohort of cardiogenic shock (CS) patients. This investigation was conducted to externally validate the independent prognostic efficacy of admission OPP in predicting in-hospital mortality in CS.
Methods
This was a retrospective analysis of a Multicenter International Registry that enrolled consecutive patients admitted for CS from July 2023 to October 2024. Only patients with CS related to heart failure (HF) or acute myocardial infarction (AMI) were included. Admission OPP was calculated as the difference between mean arterial pressure and central venous pressure. The primary outcome was in-hospital all-cause death.
Results
About 621 patients were considered (mean age 62 ± 13 years; 138 [22.2%] female): 518 (83.4%) patients presented with SCAI stage ≥ C severity. In-hospital all-cause death occurred in 247 (39.8%) individuals. As compared to survivors, nonsurvivors had significantly lower OPP (59 mmHg [IQR 49-69 mmHg] vs 70 mmHg [60-80 mmHg], P -value <.001). In univariable analysis, low OPP (<57 mmHg) was associated with significantly higher in-hospital all-cause mortality (OR 3.20 [95% CI 2.25-4.56], P -value <.001); this result was consistent across both AMI-CS and HF-CS cohorts. In a multivariable logistic regression analysis including age, diabetes, SCAI stage, Sequential Organ Failure Assessment Score, creatinine, lactates, Vasoactive Inotropic Score, OPP, central venous pressure and cardiac arrest, lower OPP significantly predicted the primary outcome (OR per mmHg decrease: 1.03 [95% CI 1.01-1.06], P -value =.020). The C-index for OPP as a predictor of in-hospital mortality was 0.691 (slope = 1.01; intercept = 0.01).
Conclusions
In this multicenter CS cohort, admission OPP was an independent prognostic marker of in-hospital mortality, irrespective of underlying CS etiology. Its inherent simplicity and demonstrated clinical robustness may support its integration into risk stratification and CS protocols.
Graphical abstract
External validation of organ perfusion pressure as a predictor of outcomes in cardiogenic shock.
Background
Cardiogenic shock (CS) remains the leading cause of death among patients hospitalized with acute myocardial infarction (AMI) and a critical challenge across diverse etiologies of heart failure (HF). Despite significant advancements in early diagnosis, pharmacotherapy, and mechanical circulatory support (MCS), mortality rates associated with CS remain prohibitively high, frequently exceeding 40%.
The cornerstone of CS management is prompt restoration of adequate tissue perfusion. Current hemodynamic guidelines prioritize increasing mean arterial pressure (MAP) primarily via the use of vasopressors and inotropes However, MAP serves only as a surrogate for macrocirculatory flow and does not fully capture the complex interplay of factors influencing capillary perfusion. Crucially, this traditional target often ignores the opposing force of venous drainage pressure, as quantified by the central venous pressure (CVP). High CVP can significantly impede capillary flow, negating the benefit of an adequate MAP and maintaining a state of organ hypoperfusion even when the MAP target is met. organ perfusion pressure (OPP), calculated as the difference between MAP and CVP, is a physiologically derived metric that represents the net driving pressure available to perfuse vital organs. Theoretically, OPP provides a superior and more complete assessment of the effective pressure gradient across capillary beds than MAP alone. Prior investigations have suggested that lower sustained OPP values are independently associated with poor outcomes, with a proposed optimal threshold set at 57 mmHg in the CS population. ,
To confirm the clinical utility and generalizability of this proposed hemodynamic target, external validation in a diverse, independent patient population is essential. The present study was designed to externally validate the association between OPP and critical outcomes in CS.
Methods
This investigation was a retrospective analysis of a multicenter, observational study designed to enrol consecutive adult patients (≥18 years) presenting with CS. Enrolment took place at 3 specialized tertiary cardiac centers located in Italy and Mexico, spanning the period from July 2023 to October 2024 (detailed in the Supplementary Appendix). Diagnosis of CS was established at each enrolling site utilizing the most recent consensus definitions, and all patients were subsequently stratified according to Society for Cardiovascular Angiography and Interventions (SCAI) stages. ,,
The present analysis focused exclusively on patients with CS attributable to either AMI (AMI-CS) or HF (HF-CS). Specifically, AMI-CS was defined as CS complicating an acute coronary syndrome HF-CS was defined as CS resulting from the acute decompensation of chronic or de novo HF.
OPP was calculated as the difference between MAP and CVP. MAP was measured invasively via an arterial line, and CVP was obtained via an invasive central venous catheter or a pulmonary artery catheter accurately positioned at the junction of the superior vena cava and the right atrium, which was confirmed radiographically. In patients supported by an intra-aortic balloon pump, MAP was obtained directly from the pressure transducer integrated into the intra-aortic balloon pump console. To ensure accuracy, MAP and CVP measurements were taken simultaneously. Only patients possessing complete data for both assessments were included in the final analysis.
Comprehensive data—including clinical, laboratory, hemodynamic, medication, and device information, along with follow-up data—were collected for all included patients. This information was captured electronically using a secure, anonymous digital platform.
The primary endpoint of the study was the external validation of the prognostic performance of OPP in predicting in-hospital all-cause mortality. According to existing studies, patients were dichotomized into 2 cohorts: a low OPP group (<57 mmHg) and a high OPP group (≥57 mmHg). Although the optimal statistical cutoff for in-hospital mortality in our cohort was 67 mmHg (via the Youden Index), we utilized a prespecified threshold of 57 mmHg for our primary analysis. This decision was made to prioritize clinical specificity in identifying the highest-risk patients and to facilitate comparison with established literature. By utilizing a consistent threshold across independent registries, we aimed to validate the generalizability of this “perfusion floor” as a standardized hemodynamic marker in CS.
Also, the OPP trajectory was assessed and the predictive capability of OPP measured at different time points (6, 12, 24, and 48 hours after admission) was assessed. To account for repeated measurements within individual patients, OPP values were analyzed using generalized linear mixed-effects models with a logistic link and random intercepts for each patient, assessing the association between OPP and in-hospital death over time.
The study protocol received approval from the Institutional Review Board of each participating center, and written informed consent was secured from all participants. The research was executed in strict adherence to the ethical principles outlined in the Declaration of Helsinki, the International Conference on Harmonization for Good Clinical Practice, and prevailing ethical regulations.
Statistical analysis
Continuous variables are reported as mean (standard deviation) or median (interquartile range), as appropriate, while categorical variables are reported as percentages. The presence of normal distribution was verified by the Shapiro–Wilk test.
Univariable analysis was performed to assess predictors of in-hospital all-cause mortality. Comparisons between baseline characteristics of in-hospital survivors and deceased and between patients with low and high OPP were performed using Student t-test for parametric continuous variables, Mann–Whitney U test for nonparametric continuous variables for and Chi-square test or Fisher’s exact test for categorical variables, accordingly. Survival curves were generated using the Kaplan and Meier method ; patients were followed until the occurrence of the primary endpoint or hospital discharge. Univariable distributions were compared with logistic regression analysis, and odds ratios (OR) and 95% confidence intervals (CI) were calculated. Also, multivariable logistic regression analysis was performed to assess OPP in relation to other potential predictors of the primary outcome. Multicollinearity among between MAP, CVP, and OPP was assessed by calculating the variance inflation factors. When significant collinearity with OPP was detected (ie, variance inflating factor greater than 5), only OPP was retained in the model (Supplementary Table 1). A multivariable model for the primary outcome was developed using a hybrid selection strategy; to ensure clinical relevance and minimize the bias inherent in purely data-driven variable selection, we “forced-in” prognostic markers identified as having at least moderate grading of recommendations, assessment, development, and evaluation certainty in recent large-scale meta-analyses; these were analyzed alongside statistically significant univariable predictors to provide a comprehensive and externally validated assessment of in-hospital mortality risk. Also, other multivariable models for the primary outcome were evaluated, in line with the original work of our group : Model 1 was built adjusting for significant ( P -value <.05) clinical data available at first bedside assessment, namely age and Sequential Organ Failure Assessment score; Model 2 investigated OPP alongside age, Sequential Organ Failure Assessment score and CVP; Model 3 evaluated hemodynamic parameters available at initial bedside assessment, namely OPP, heart rate and CVP. A separate logistic regression multivariable model including OPP, CVP and vasoactive inotropic score was assessed, using both raw units and standardized coefficients (per 1 standard deviation change); MAP was not included in these models due to collinearity with OPP.
The discriminative ability of the OPP for outcome risk was assessed with Harrell’s C-statistic. To determine the calibration of the predictive model for in-hospital mortality, calibration plots comparing predicted probabilities with observed event rates were generated. Predicted probabilities were obtained from a logistic regression model with in-hospital mortality as the dependent variable of interest and OPP as the independent variable. Patients were grouped into deciles based on predicted risk, and for each decile, the mean predicted probability was plotted against the observed mortality rate. The 45-degree identity line represented perfect calibration. Calibration-in-the-large and calibration slope were evaluated by fitting a logistic regression model using the logit of the predicted probabilities (ie, linear predictor) as the sole independent variable. In this model, a calibration intercept of 0 and a slope of 1 indicate perfect agreement between predicted and observed outcomes. To address potential overfitting and provide a robust estimate of model performance within the derivation cohort, we performed an internal validation procedure using 200 bootstrap resamples. This approach allowed for the estimation and correction of optimism in the calibration slope and intercept, ensuring the reliability of the model’s predictive accuracy.
Univariable analyses to assess predictors of in-hospital all-cause mortality were run separately for the individual AMI-CS and HF-CS subgroups and for patients with and without MCS, separately; Kaplan–Meier curves were generated, and survival distributions were compared using univariable Cox regression analyses stratified by OPP in each subset.
Also, the OPP was measured at 6, 12, 24, and 48 hours after hospital admission and compared between survivors and nonsurvivors using the Mann–Whitney U test. Kaplan–Meier curves were generated and Cox regression analysis was performed. The difference between OPP at different time points and baseline OPP (ie, delta-OPP) was calculated and its predictive yield was assessed. An exploratory analysis of the receiver operating characteristics curve to identify the optimal OPP cutoff at different timepoints was performed using Youden’s index. Secondary outcomes were assessed, including acute kidney injury, renal replacement therapy, and sepsis.
Also, the prognostic yield of OPP on in-hospital mortality was tested against those of the SCAI classification and the SOFA score alone and the corresponding net reclassification indexes were calculated.
Separate univariable and multivariable analyses assessing the predictive yield of MAP were performed; a receiver operating characteristics curve on baseline MAP was also drawn.
A 2-sided P -value <.05 was considered statistically significant. Statistical analyses were performed using SPSS 28.0 (SPSS, Chicago, IL) and R software version 3.5.3 (R Foundation).
Results
From the 752 patients initially assessed for eligibility, 145 individuals were excluded according to the predefined exclusion criteria. The study cohort comprized 621 patients, with a mean age of 62 ± 13 years ( Table 1 ); 138 (22.2%) were female. The mean body mass index was 27.0 ± 4.5 kg/m 2. A total of 468 (75.4%) patients presented with AMI-CS, while 153 (24.6%) presented with HF-CS. Upon admission, 518 (83.4%) patients were classified as SCAI stage C or worse CS with median lactate levels of 2.3 mmol/L (IQR 1.5-3.8 mmol/L). Median MAP was 79 mmHg (IQR 70-89 mmHg), median CVP was 13 mmHg (IQR 9-18 mmHg) and median OPP was 66 mmHg (IQR 55-76 mmHg). On echocardiography, median left ventricular ejection fraction was 30% (IQR 20%-40%) and mean tricuspid annular plane systolic excursion was 16.9 ± 4.4 mm. A total of 279 (45.0%) patients were on intra-aortic balloon pump, 25 (4.0%) on microaxial flow pump (Impella) and 5 (0.8%) on extracorporeal membrane oxygenation (Supplementary Table 2).
Table 1
Baseline characteristics of the study population
| Variables | All patients ( n = 621) | In-hospital all-cause death | P -value | |
|---|---|---|---|---|
| Yes ( n = 247) | No ( n = 374) | |||
| Age—y | 62 ± 13 | 63 ± 12 | 62 ± 13 | .329 |
| Female sex—no. (%) | 138 (22.2%) | 57 (23.2%) | 81 (21.7%) | .658 |
| BMI—kg/m 2 | 27.0 ± 4.5 | 26.9 ± 4.7 | 27.1 ± 4.4 | .719 |
| Smoking habit—no. (%) | 171 (27.5%) | 70 (28.3%) | 101 (27.0%) | .716 |
| Arterial hypertension—no. (%) | 348 (56.0%) | 135 (54.7%) | 213 (57.0%) | .573 |
| Diabetes mellitus—no. (%) | 286 (46.1%) | 126 (51.0%) | 160 (42.8%) | .044 |
| Dyslipidemia—no. (%) | 223 (35.9%) | 82 (33.2%) | 141 (37.7%) | .252 |
| Prior stroke/TIA—no. (%) | 234 (37.7%) | 108 (43.7%) | 126 (33.7%) | .012 |
| Chronic kidney disease—no. (%) | 90 (14.5%) | 38 (15.4%) | 52 (13.9%) | .608 |
| Cancer history—no. (%) | 36 (19.3%) | 12 (25.3%) | 24 (17.1%) | .436 |
| Prior PCI—no. % | 178 (28.7%) | 70 (28.3%) | 108 (28.9%) | .885 |
| Prior CABG—no. (%) | 75 (12.1%) | 28 (11.3%) | 47 (12.6%) | .645 |
| Atrial fibrillation—no. (%) | 131 (21.1%) | 51 (20.6%) | 80 (21.4%) | .824 |
| ICD —no. (%) | 57 (9.2%) | 19 (7.7%) | 38 (10.2%) | .297 |
| CS etiology—no. (%) | ||||
| AMI-CS | 468 (75.4%) | 198 (80.2%) | 270 (72.2%) | .086 |
| HF-CS | 153 (24.6%) | 49 (19.8%) | 94 (25.1%) | |
| ACS-CS type—no. (%) | ||||
| STEMI | 378 (80.4%) | 172 (85.6%) | 206 (76.3%) | .016 |
| NSTEMI | 92 (19.6%) | 28 (14.4%) | 64 (23.7%) | |
| HF-CS type—no. (%): | ||||
| Advanced HF-CS | 143 (23.0%) | 49 (19.8%) | 94 (25.1%) | .086 |
| De novo HF-CS | 10 (1.6%) | 0 (0.0%) | 10 (2.7%) | |
| Cardiocirculatory arrest—no. (%) | 30 (4.8%) | 13 (5.3%) | 17 (4.5%) | .683 |
| Positive pressure ventilation—no. (%) | 318 (51.3%) | 128 (51.8%) | 190 (50.8%) | .803 |
| SOFA score | 5.4 ± 2.9 | 6.1 ± 3.0 | 4.8 ± 2.7 | <.001 |
| Vasoactive inotropic score | 5.4 ± 7.3 | 7.1 ± 10.0 | 4.9 ± 6.1 | .149 |
| SCAI shock stage—no. (%) | ||||
| Stage A | 14 (2.3%) | 1 (0.4%) | 13 (3.5%) | <.001 |
| Stage B | 89 (14.3%) | 16 (6.5%) | 73 (19.5%) | |
| Stage C | 95 (15.3%) | 25 (10.1%) | 70 (18.7%) | |
| Stage D | 170 (27.4%) | 47 (19.0%) | 123 (32.9%) | |
| Stage E | 253 (40.7%) | 158 (64.0%) | 95 (25.4%) | |
| Hemodynamics | ||||
| Systolic blood pressure—mmHg | 110 ± 23 | 104 ± 24 | 115 ± 26 | <.001 |
| MAP—mmHg | 79 (IQR 70-89) | 75 (IQR 66-83) | 83 (IQR 73-93) | <.001 |
| Diastolic blood pressure—mmHg | 65 ± 15 | 64 ± 16 | 69 ± 16 | <.001 |
| Heart rate—bpm | 93 ± 25 | 94 ± 25 | 93 ± 26 | .523 |
| CVP—mmHg | 13 (IQR 9-18) | 15 (IQR 11-19) | 12 (IQR 8-16) | <.001 |
| Central venous oxygen saturation—% | 55 ± 13 | 54 ± 14 | 55 ± 13 | .785 |
| OPP—mmHg | 66 (IQR 55-76) | 59 (IQR 49-69) | 70 (IQR 60-80) | <.001 |
| Laboratory tests | ||||
| Creatinine—mg/dL | 1.75 ± 1.09 | 1.98 ± 1.25 | 1.60 ± 0.94 | <.001 |
| Total bilirubin—mg/dL | 3.0 ± 6.1 | 5.4 ± 10.8 | 2.0 ± 0.2 | .177 |
| Alanine transaminase—IU/L | 295 ± 835 | 470 ± 1183 | 190 ± 504 | <.001 |
| High-sensitivity troponin I—ng/mL | 4,329 ± 14,698 | 4,057 ± 16,712 | 4,564 ± 12,730 | .734 |
| NT-pro–BNP—pg/mL | 8,725 (IQR 3,359-20,180) | 12,316 (IQR 4,221-26,942) | 6,579 (IQR 2,374-16,464) | <.001 |
| Lactate—mmol/L | 2.3 (IQR 1.5-3.8) | 2.5 (IQR 1.8-4.1) | 2.1 (IQR 1.4-3.4) | <.001 |
| Heart failure home medications | ||||
| Beta-blocker—no. (%) | 133 (21.7%) | 42 (17.4%) | 91 (24.5%) | .034 |
| RAASi—no. (%) | 88 (14.4%) | 18 (7.5%) | 70 (18.8%) | <.001 |
| MRA—no. (%) | 51 (8.3%) | 14 (5.8%) | 37 (9.9%) | .070 |
| Echocardiographic data | ||||
| LVEDD—mm | 58 ± 12 | 60 ± 12 | 58 ± 12 | .598 |
| LV ejection fraction—% | 30 (IQR 20-40) | 30 (IQR 20-38) | 30 (IQR 20-40) | .514 |
| Average E/e′ | 14.4 ± 5.3 | 15.0 ± 5.1 | 14.2 ± 5.4 | .506 |
| At least moderate-to-severe mitral regurgitation—no. (%) | 48 (26.0%) | 11 (24.4%) | 37 (26.6%) | .181 |
| TAPSE—mm | 16.9 ± 4.4 | 16.4 ± 3.9 | 17.1 ± 4.5 | .359 |
| At least moderate-to-severe tricuspid regurgitation—no. (%) | 47 (26.1%) | 13 (27.5%) | 34 (24.9%) | .793 |
| Estimated PASP—mmHg | 52 ± 16 | 57 ± 18 | 51 ± 15 | .077 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
