Hyperlactataemia is a marker of reduced exercise capacity in heart failure with preserved ejection fraction.

AIMS: Heart failure with preserved ejection fraction (HFpEF) is associated with an array of central and peripheral haemodynamic and metabolic changes. The exact pathogenesis of exercise limitation in HFpEF remains uncertain. Our aim was to compare lactate accumulation and central haemodynamic responses to exercise in patients with HFpEF, non-cardiac dyspnoea (NCD), and healthy volunteers. METHODS AND RESULTS: Right heart catheterization with mixed venous blood gas and lactate measurements was performed at rest and during symptom-limited supine exercise. Multivariable analyses were conducted to determine the relationship between haemodynamic and biochemical parameters and their association with exercise capacity. Of 362 subjects, 198 (55%) had HFpEF, 103 (28%) had NCD, and 61 (17%) were healthy volunteers. This included 139 (70%) females with HFpEF, 77 (75%) in NCD (P = 0.41 HFpEF vs. NCD), and 31 (51%) in healthy volunteers (P < 0.001 HFpEF vs. volunteers). The median age was 71 (65, 75) years in HFpEF, 66 (57, 72) years in NCD, and 49 (38, 65) years in healthy volunteers (HFpEF vs. NCD or volunteer, both P < 0.001). Peak workload was lower in HFpEF compared with healthy volunteers [52 W (interquartile range 31-73), 150 W (125-175), P < 0.001], but not NCD [53 W (33, 75), P = 0.85]. Exercise lactate indexed to workload was higher in HFpEF at 0.08 mmol/L/W (0.05-0.11), 0.06 mmol/L/W (0.05-0.08; P = 0.016) in NCD, and 0.04 mmol/L/W (0.03-0.05; P < 0.001) in volunteers. Exercise cardiac index was 4.5 L/min/m2 (3.7-5.5) in HFpEF, 5.2 L/min/m2 (4.3-6.2; P < 0.001) in NCD, and 9.1 L/min/m2 (8.0-9.9; P < 0.001) in volunteers. Oxygen delivery in HFpEF was lower at 1553 mL/min (1175-1986) vs. 1758 mL/min (1361-2282; P = 0.024) in NCD and 3117 mL/min (2667-3502; P < 0.001) in the volunteer group during exercise. Predictors of higher exercise lactate levels in HFpEF following adjustment included female sex and chronic kidney disease (both P < 0.001). CONCLUSIONS: HFpEF is associated with reduced exercise capacity secondary to both central and peripheral factors that alter oxygen utilization. This results in hyperlactataemia. In HFpEF, plasma lactate responses to exercise may be a marker of haemodynamic and cardiometabolic derangements and represent an important target for future potential therapies.

Cardiac Substrate Utilization and Relationship to Invasive Exercise Hemodynamic Parameters in HFpEF.

The authors conducted transcardiac blood sampling in healthy subjects and subjects with heart failure with preserved ejection fraction (HFpEF) to compare cardiac metabolite and lipid substrate use. We demonstrate that fatty acids are less used by HFpEF hearts and that lipid extraction is influenced by hemodynamic factors including pulmonary pressures and cardiac index. The release of many products of protein catabolism is apparent in HFpEF compared to healthy myocardium. In subgroup analyses, differences in energy substrate use between female and male hearts were identified.

Influence of angiotensin II on the gut microbiome: Modest effects in comparison to experimental factors.

INTRODUCTION: Animal models are regularly used to test the role of the gut microbiome in hypertension. Small-scale pre-clinical studies have investigated changes to the gut microbiome in the angiotensin II hypertensive model. However, the gut microbiome is influenced by internal and external experimental factors which are not regularly considered in the study design. Once these factors are accounted for, it is unclear if microbiome signatures are reproduceable. We aimed to determine the influence of angiotensin II treatment on the gut microbiome using a large and diverse cohort of mice and to quantify the magnitude by which other factors contribute to microbiome variations. METHODS AND RESULTS: We conducted a retrospective study to establish a diverse mouse cohort resembling large human studies. We sequenced the V4 region of the 16S rRNA gene from 538 samples across the gastrointestinal tract of 303 male and female C57BL/6J mice randomised into sham or angiotensin II treatment from different genotypes, diets, animal facilities, and age groups. Analysing over 17 million sequencing reads, we observed that angiotensin II treatment influenced α-diversity (P = 0.0137) and ß-diversity (i.e., composition of the microbiome, P < 0.001). Bacterial abundance analysis revealed patterns consistent with a reduction in short-chain fatty acid-producers, microbial metabolites that lower blood pressure. Furthermore, animal facility, genotype, diet, age, sex, intestinal sampling site, and sequencing batch had significant effects on both α- and ß-diversity (all P < 0.001). Sampling site (6.8%) and diet (6%) had the largest impact on the microbiome, while angiotensin II and sex had the smallest effect (each 0.4%). CONCLUSIONS: Our large-scale data confirmed findings from small-scale studies that angiotensin II impacted the gut microbiome. However, this effect was modest relative to most of the other factors studied. Accounting for these factors in future pre-clinical hypertensive studies will increase the likelihood that microbiome findings are replicable and translatable.

No-Reflow Prediction in Acute Coronary Syndrome During Percutaneous Coronary Intervention: The NORPACS Risk Score.

BACKGROUND: Suboptimal coronary reperfusion (no reflow) is common in acute coronary syndrome percutaneous coronary intervention (PCI) and is associated with poor outcomes. We aimed to develop and externally validate a clinical risk score for angiographic no reflow for use following angiography and before PCI. METHODS: We developed and externally validated a logistic regression model for prediction of no reflow among adult patients undergoing PCI for acute coronary syndrome using data from the Melbourne Interventional Group PCI registry (2005-2020; development cohort) and the British Cardiovascular Interventional Society PCI registry (2006-2020; external validation cohort). RESULTS: A total of 30 561 patients (mean age, 64.1 years; 24% women) were included in the Melbourne Interventional Group development cohort and 440 256 patients (mean age, 64.9 years; 27% women) in the British Cardiovascular Interventional Society external validation cohort. The primary outcome (no reflow) occurred in 4.1% (1249 patients) and 9.4% (41 222 patients) of the development and validation cohorts, respectively. From 33 candidate predictor variables, 6 final variables were selected by an adaptive least absolute shrinkage and selection operator regression model for inclusion (cardiogenic shock, ST-segment-elevation myocardial infarction with symptom onset >195 minutes pre-PCI, estimated stent length ≥20 mm, vessel diameter <2.5 mm, pre-PCI Thrombolysis in Myocardial Infarction flow <3, and lesion location). Model discrimination was very good (development C statistic, 0.808; validation C statistic, 0.741) with excellent calibration. Patients with a score of ≥8 points had a 22% and 27% risk of no reflow in the development and validation cohorts, respectively. CONCLUSIONS: The no-reflow prediction in acute coronary syndrome risk score is a simple count-based scoring system based on 6 parameters available before PCI to predict the risk of no reflow. This score could be useful in guiding preventative treatment and future trials.

Electrocardiographic patterns and clinical outcomes of acute coronary syndrome cardiogenic shock in patients undergoing percutaneous coronary intervention - A propensity score analysis.

OBJECTIVES: To determine the influence of presenting electrocardiographic (ECG) changes on prognosis in acute coronary syndrome cardiogenic shock (ACS-CS) patients undergoing percutaneous coronary angiography (PCI). BACKGROUND: The effect of initial ECG changes such as ST-elevation myocardial infarction (STEMI) versus non-STEMI among patients ACS-CS on prognosis remains unclear. METHODS: We analysed data from consecutive patients with ACS-CS enrolled in the Victorian Cardiac Outcomes registry between 2014 and 2020. Inverse probability of treatment weighting analysis (IPTW) was used to assess the effect of ECG changes on 30-day mortality. RESULTS: Of 1564 patients with ACS-CS who underwent PCI, 161 had non-STEMI and 1403 had STEMI on ECG. The mean age was 66 ± 13 years, and 74 % (1152) were males. Patients with non-STEMI compared to STEMI were older (70 ± 12 vs 65 ± 13 years), had higher rates of diabetes (34 % vs 21 %), prior coronary artery bypass graft surgery (14 % vs 3.3 %), peripheral arterial disease (10.6 % vs 4.1 %, p < 0.01), and lower baseline eGFR (53.8 [37.1, 75.4] vs 65.3 [46.3, 87.8] ml/min/1.73m2), all p ≤ 0.01. Non-STEMI patients were more likely to have a culprit left circumflex artery (29 % vs 20 %) and more often underwent multivessel percutaneous coronary intervention (30 % vs 20 %) but had lower rates of out-of-hospital cardiac arrest (21 % vs 39 %), all p ≤ 0.01. Propensity score analysis with IPTW confirmed that non-STEMI ECG was associated with lower odds for 30-day all-cause mortality (OR 0.47 [0.32, 0.69], p < 0.001), and 30-day major adverse cardiovascular and cerebrovascular events (OR 0.48 [0.33, 0.70]). CONCLUSIONS: In patients undergoing PCI, Non-STEMI as compared to STEMI on index ECG was associated with approximately half the relative risk of both 30-day mortality and 30-day MACCE and could be a useful variable to integrate in ACS-CS risk scores.

Comparison of Characteristics and Outcomes in Patients With Acute Decompensated Heart Failure Admitted Under General Medicine and Cardiology Units.

BACKGROUND: Acute decompensated heart failure (ADHF) is a leading cause of cardiovascular disease hospitalisations associated with significant morbidity and mortality. In hospitals, HF patients are typically managed by cardiology or physician teams, with differences in patient demographics and clinical outcomes. This study utilises contemporary HF registry data to compare patient characteristics and outcomes in those with ADHF admitted into General Medicine and Cardiology units. METHODS: The Victorian Cardiac Outcomes Registry was utilised to identify patients hospitalised with ADHF 30-day period in each of four consecutive years. We compared patient characteristics, pharmacological management and outpatient follow-up of patients admitted to General Medicine and Cardiology units. Primary outcome measures included in-hospital mortality, 30-day readmission, and 30-day mortality. RESULTS: Between 2014 and 2017, a total of 1,253 patients with ADHF admissions were registered, with 53% admitted in General Medicine units and 47% in Cardiology units. General Medicine patients were more likely to be older (82 vs 71 years; p<0.001), female (51% vs 34%; p<0.001), and have higher prevalence of comorbidities and preserved left ventricular function (p<0.001). There were no differences in primary outcome measures between General Medicine and Cardiology in terms of: in-hospital mortality (5.0% vs 3.9%; p=0.35), 30-day readmission (23.4% vs 23.6%; p=0.93), and 30-day mortality (10.0% vs 8.0%; p=0.21). CONCLUSIONS: Hospitalised patients with HF continue to have high mortality and rehospitalisation rates. The choice of treatment by General Medicine or Cardiology units, based on the particular medical profile and individual needs of the patients, provides equivalent outcomes.

The Impact of Diabetes on Haemodynamic and Cardiometabolic Responses in Heart Failure With Preserved Ejection Fraction.

AIMS: Heart failure with preserved ejection (HFpEF) and diabetes mellitus (DM) commonly co-exist. However, it is unclear if DM modifies the haemodynamic and cardiometabolic phenotype in patients with HFpEF. We aimed to interrogate the haemodynamic and cardiometabolic effects of DM in HFpEF. METHODS: We compared the haemodynamic and metabolic profiles of non-DM patients and patients with DM-HFpEF at rest and during exercise using right heart catheterisation and mixed venous blood gas analysis. RESULTS: Of 181 patients with HFpEF, 37 (20%) had DM. Patients with DM displayed a more adverse exercise haemodynamic response vs HFpEF alone (mean pulmonary arterial pressure: 47 mmHg [interquartile range {IQR} 42-55] vs 42 [38-47], p<0.001; workload indexed pulmonary capillary wedge pressure indexed: 0.80 mmHg/W [0.44-1.23] vs 0.57 [0.43-1.01], p=0.047). HFpEF-DM patients had a lower mixed venous oxygen saturation at rest (70% [IQR 66-73] vs 72 [69-75], p=0.003) and were unable to enhance O2 extraction to the same extent (Δ-28% [-33 to -15] vs -29 [-36 to -21], p=0.029), this occurred at a 22% lower median workload. Resting mixed venous lactate levels were higher in those with DM (1.5 mmol/L [IQR 1.1-1.9] vs 1 [0.9-1.3], p<0.001), and during exercise indexed to workload (0.09 mmol/L/W [0.06-0.13] vs 0.08 [0.05-0.11], p=0.018). CONCLUSION: Concurrent diabetes and HFpEF was associated with greater metabolic responses at rest, with enhanced wedge driven pulmonary hypertension and relative lactataemia during exercise without appropriate augmentation of oxygen consumption.

Indole-3-Propionic Acid Protects Against Heart Failure With Preserved Ejection Fraction.

BACKGROUND: Heart failure with preserved ejection fraction (HFpEF) is a common but poorly understood form of heart failure, characterized by impaired diastolic function. It is highly heterogeneous with multiple comorbidities, including obesity and diabetes, making human studies difficult. METHODS: Metabolomic analyses in a mouse model of HFpEF showed that levels of indole-3-propionic acid (IPA), a metabolite produced by gut bacteria from tryptophan, were reduced in the plasma and heart tissue of HFpEF mice as compared with controls. We then examined the role of IPA in mouse models of HFpEF as well as 2 human HFpEF cohorts. RESULTS: The protective role and therapeutic effects of IPA were confirmed in mouse models of HFpEF using IPA dietary supplementation. IPA attenuated diastolic dysfunction, metabolic remodeling, oxidative stress, inflammation, gut microbiota dysbiosis, and intestinal epithelial barrier damage. In the heart, IPA suppressed the expression of NNMT (nicotinamide N-methyl transferase), restored nicotinamide, NAD+/NADH, and SIRT3 (sirtuin 3) levels. IPA mediates the protective effects on diastolic dysfunction, at least in part, by promoting the expression of SIRT3. SIRT3 regulation was mediated by IPA binding to the aryl hydrocarbon receptor, as Sirt3 knockdown diminished the effects of IPA on diastolic dysfunction in vivo. The role of the nicotinamide adenine dinucleotide circuit in HFpEF was further confirmed by nicotinamide supplementation, Nnmt knockdown, and Nnmt overexpression in vivo. IPA levels were significantly reduced in patients with HFpEF in 2 independent human cohorts, consistent with a protective function in humans, as well as mice. CONCLUSIONS: Our findings reveal that IPA protects against diastolic dysfunction in HFpEF by enhancing the nicotinamide adenine dinucleotide salvage pathway, suggesting the possibility of therapeutic management by either altering the gut microbiome composition or supplementing the diet with IPA.

Association of socioeconomic status in the incidence, quality-of-care metrics, and outcomes for patients with cardiogenic shock in a pre-hospital setting.

AIMS: The relationship between lower socioeconomic status (SES) and poor cardiovascular outcomes is well described; however, there exists a paucity of data exploring this association in cardiogenic shock (CS). This study aimed to investigate whether any disparities exist between SES and the incidence, quality of care or outcomes of CS patients attended by emergency medical services (EMS). METHODS AND RESULTS: This population-based cohort study included consecutive patients transported by EMS with CS between 1 January 2015 and 30 June 2019 in Victoria, Australia. Data were collected from individually linked ambulance, hospital, and mortality datasets. Patients were stratified into SES quintiles using national census data produced by the Australian Bureau of Statistics.A total of 2628 patients were attended by EMS for CS. The age-standardized incidence of CS amongst all patients was 11.8 [95% confidence interval (95% CI), 11.4-12.3] per 100 000 person-years, with a stepwise increase from the highest to lowest SES quintile (lowest quintile 17.0 vs. highest quintile 9.7 per 100 000 person-years, P-trend < 0.001). Patients in lower SES quintiles were less likely to attend metropolitan hospitals and more likely to be received by inner regional and remote centres without revascularization capabilities. A greater proportion of the lower SES groups presented with CS due to non-ST elevation myocardial infarction (NSTEMI) or unstable angina pectoris (UAP), and overall were less likely to undergo coronary angiography. Multivariable analysis demonstrated an increased 30-day all-cause mortality rate in the lowest three SES quintiles when compared with the highest quintile. CONCLUSION: This population-based study demonstrated discrepancies between SES status in the incidence, care metrics, and mortality rates of patients presenting to EMS with CS. These findings outline the challenges in equitable healthcare delivery within this cohort.

Hypothermic oxygenated perfusion (HOPE) safely and effectively extends acceptable donor heart preservation times: Results of the Australian and New Zealand trial.

BACKGROUND: Cold static storage preservation of donor hearts for periods longer than 4 hours increases the risk of primary graft dysfunction (PGD). The aim of the study was to determine if hypothermic oxygenated perfusion (HOPE) could safely prolong the preservation time of donor hearts. METHODS: We conducted a nonrandomized, single arm, multicenter investigation of the effect of HOPE using the XVIVO Heart Preservation System on donor hearts with a projected preservation time of 6 to 8 hours on 30-day recipient survival and allograft function post-transplant. Each center completed 1 or 2 short preservation time followed by long preservation time cases. PGD was classified as occurring in the first 24 hours after transplantation or secondary graft dysfunction (SGD) occurring at any time with a clearly defined cause. Trial survival was compared with a comparator group based on data from the International Society of Heart and Lung Transplantation (ISHLT) Registry. RESULTS: We performed heart transplants using 7 short and 29 long preservation time donor hearts placed on the HOPE system. The mean preservation time for the long preservation time cases was 414 minutes, the longest being 8 hours and 47 minutes. There was 100% survival at 30 days. One long preservation time recipient developed PGD, and 1 developed SGD. One short preservation time patient developed SGD. Thirty day survival was superior to the ISHLT comparator group despite substantially longer preservation times in the trial patients. CONCLUSIONS: HOPE provides effective preservation out to preservation times of nearly 9 hours allowing retrieval from remote geographic locations.

Sex Differences in Pharmacotherapy and Long-Term Outcomes in Patients With Ischaemic Heart Disease and Comorbid Left Ventricular Dysfunction.

BACKGROUND: Left ventricular (LV) dysfunction and ischaemic heart disease (IHD) are common among women. However, women tend to present later and are less likely to receive guideline-directed medical therapy (GDMT) compared with men. METHODS: We analysed prospectively collected data (2005-2018) from a multicentre registry on GDMT 30 days after percutaneous coronary intervention in 13,015 patients with LV ejection fraction <50%. Guideline-directed medical therapy was defined as beta blocker, angiotensin-converting enzyme inhibitor/angiotensin receptor blocker±mineralocorticoid receptor antagonist. Long-term mortality was determined by linkage with the Australian National Death Index. RESULTS: Women represented 20% (2,634) of the total cohort. Mean age was 65±12 years. Women were on average >5 years, with higher body mass index and higher rates of hypertension, diabetes, renal dysfunction, prior stroke, and rheumatoid arthritis. Guideline-directed medical therapy was similar between sexes (73% vs 72%; p=0.58), although women were less likely to be on an angiotensin-converting enzyme inhibitor/angiotensin receptor blocker (80% vs 82%; p=0.02). Women were less likely to be on statin therapy (p<0.001) or a second antiplatelet agent (p=0.007). Women had higher unadjusted long-term mortality (25% vs 19%; p<0.001); however, there were no differences in long-term mortality between sexes on adjusted analysis (hazard ratio 0.99; 95% confidence interval 0.87-1.14; p=0.94). CONCLUSIONS: Rates of GDMT for LV dysfunction were high and similar between sexes; however, women were less likely to be on appropriate IHD secondary prevention. The increased unadjusted long-term mortality in women was attenuated in adjusted analysis, which highlights the need for optimisation of baseline risk to improve long-term outcomes of women with IHD and comorbid LV dysfunction.

Clinical Features and Outcomes Among Patients With Refractory Out-of-Hospital Cardiac Arrest and an Initial Shockable Rhythm.

BACKGROUND: Clinical features among patients with refractory out-of-hospital cardiac arrest (OHCA) and initial shockable rhythms of ventricular fibrillation/pulseless ventricular tachycardia are not well-characterized. METHODS: We compared clinical characteristics and coronary angiographic findings between patients with refractory OHCA (incessant ventricular fibrillation/pulseless ventricular tachycardia after ≥3 direct-current shocks) and those without refractory OHCA. RESULTS: Between 2014 and 2018, a total of 204 patients with ventricular fibrillation/pulseless ventricular tachycardia OHCA (median age 62; males 78%) were divided into groups with (36%, 74/204) and without refractory arrest (64%, 130/204). Refractory OHCA patients had longer cardiopulmonary resuscitation (23 versus 15 minutes), more frequently required ≥450 mg amiodarone (34% versus 3.8%), and had cardiogenic shock (80% versus 55%) necessitating higher adrenaline dose (4.0 versus 1.0 mg) and higher rates of mechanical ventilation (92% versus 74%; all P<0.01). Of 167 patients (82%) selected for coronary angiography, 33% (n=55) had refractory OHCA (P=0.035). Significant coronary artery disease (≥1 major vessel with >70% stenosis) was present in >70% of patients. Refractory OHCA patients frequently had acute coronary occlusion (64% versus 47%), especially left circumflex (20% versus 6.4%) and graft vessel (7.3% versus 0.9%; all P<0.05) compared with those without refractory OHCA. Refractory OHCA group had higher in-hospital mortality (45% versus 30%, P=0.036) and greater new requirement for dialysis (18% versus 6.3%, P=0.011). After adjustment, refractory OHCA was associated with over 2-fold higher odds of in-hospital mortality (odds ratio, 2.28 [95% CI, 1.06-4.89]; P=0.034). CONCLUSIONS: Refractory ventricular fibrillation/pulseless ventricular tachycardia OHCA was associated with more intensive resuscitation, higher rates of acute coronary occlusion, and poorer in-hospital outcomes, underscoring the need for future studies in this extreme-risk subgroup.

Effect of Concomitant Cardiac Arrest on Outcomes in Patients With Acute Coronary Syndrome-Related Cardiogenic Shock.

Patients with acute coronary syndrome (ACS)-related cardiogenic shock (CS) with or without concomitant CA may have disparate prognoses. We compared clinical characteristics and outcomes of patients with CS secondary to ACS with and without cardiac arrest (CA). Between 2014 and 2020, 1,573 patients with ACS-related CS with or without CA who underwent percutaneous coronary intervention enrolled in a multicenter Australian registry were analyzed. Primary outcome was 30-day major adverse cardiovascular and cerebrovascular events (MACCE) (composite of mortality, myocardial infarction, stent thrombosis, target vessel revascularization and stroke). Long-term mortality was obtained through linkage to the National Death Index. Compared with the no-CA group (n = 769, 49%), the CA group (n = 804, 51%) was younger (62 vs 69 years, p <0.001) and had fewer comorbidities. Patients with CA more frequently had ST-elevation myocardial infarction (92% vs 86%), occluded left anterior descending artery (43% vs 33%), and severe preprocedural renal impairment (49% vs 42%) (all p <0.001). CA increased risk of 30-day MACCE by 45% (odds ratio 1.45, 95% confidence interval 1.05 to 2.00, p = 0.024) after adjustment. CA group had higher 30-day MACCE (55% vs 42%, p <0.001) and mortality (52% vs 37%, p <0.001). Three-year survival was lower for CA compared with no-CA patients (43% vs 52%, p <0.001). In Cox regression, CS with CA was associated with a trend toward greater long-term mortality hazard (hazard ratio 1.19, 95% confidence interval 1.00 to 1.41, p = 0.055). In conclusion, concomitant CA among patients with ACS-related CS conferred a particularly heightened short-term risk with a diminishing legacy effect over time for mortality. CS survivors continue to exhibit high sustained long-term mortality hazard regardless of CA status.

ECMO PAL: using deep neural networks for survival prediction in venoarterial extracorporeal membrane oxygenation.

PURPOSE: Venoarterial extracorporeal membrane oxygenation (VA-ECMO) is a complex and high-risk life support modality used in severe cardiorespiratory failure. ECMO survival scores are used clinically for patient prognostication and outcomes risk adjustment. This study aims to create the first artificial intelligence (AI)-driven ECMO survival score to predict in-hospital mortality based on a large international patient cohort. METHODS: A deep neural network, ECMO Predictive Algorithm (ECMO PAL) was trained on a retrospective cohort of 18,167 patients from the international Extracorporeal Life Support Organisation (ELSO) registry (2017-2020), and performance was measured using fivefold cross-validation. External validation was performed on all adult registry patients from 2021 (N = 5015) and compared against existing prognostication scores: SAVE, Modified SAVE, and ECMO ACCEPTS for predicting in-hospital mortality. RESULTS: Mean age was 56.8 ± 15.1 years, with 66.7% of patients being male and 50.2% having a pre-ECMO cardiac arrest. Cross-validation demonstrated an inhospital mortality sensitivity and precision of 82.1 ± 0.2% and 77.6 ± 0.2%, respectively. Validation accuracy was only 2.8% lower than training accuracy, reducing from 75.5% to 72.7% [99% confidence interval (CI) 71.1-74.3%]. ECMO PAL accuracy outperformed the ECMO ACCEPTS (54.7%), SAVE (61.1%), and Modified SAVE (62%) scores. CONCLUSIONS: ECMO PAL is the first AI-powered ECMO survival score trained and validated on large international patient cohorts. ECMO PAL demonstrated high generalisability across ECMO regions and outperformed existing, widely used scores. Beyond ECMO, this study highlights how large international registry data can be leveraged for AI prognostication for complex critical care therapies.