Daily Ards Research Analysis
Three impactful ARDS studies span mechanistic innovation, ECMO-era prognostication, and interpretable AI. A mouse study demonstrates that orally delivered onion-derived mitochondria reprogram lung macrophage bioenergetics to mitigate LPS-induced ALI, while a multicenter ECMO cohort reveals day-14 tidal volume as an independent mortality predictor. An interpretable ML model, aligned with the new global ARDS definition, externally validates 28-day ICU mortality prediction in sepsis-associated ARDS
Summary
Three impactful ARDS studies span mechanistic innovation, ECMO-era prognostication, and interpretable AI. A mouse study demonstrates that orally delivered onion-derived mitochondria reprogram lung macrophage bioenergetics to mitigate LPS-induced ALI, while a multicenter ECMO cohort reveals day-14 tidal volume as an independent mortality predictor. An interpretable ML model, aligned with the new global ARDS definition, externally validates 28-day ICU mortality prediction in sepsis-associated ARDS.
Research Themes
- Mitochondrial therapeutics and macrophage immunometabolism in ALI/ARDS
- Dynamic prognostication and ventilator strategy during ECMO for ARDS
- Interpretable machine learning for ARDS outcomes under the new global definition
Selected Articles
1. Onion-Mitochondria Inhibit Lipopolysaccharide-Induced Acute Lung Injury by Shaping Lung Macrophage Mitochondrial Function.
In LPS-induced ALI mice, orally delivered onion-derived mitochondria trafficked to lungs, were preferentially taken up by macrophages via PA–CR1L interaction, fused with host mitochondria, and reprogrammed bioenergetics. MDHB-enriched O-Mit epigenetically suppressed ND1, dampened complex I-driven oxidative stress, limited DRP1-mediated fission and cardiolipin peroxidation, and ameliorated lung injury.
Impact: This work introduces a cross-kingdom organelle therapy concept with mechanistic depth linking ND1, DRP1, and cardiolipin biology to ALI attenuation. It opens a novel therapeutic avenue targeting macrophage mitochondrial dysfunction in ALI/ARDS.
Clinical Implications: Although preclinical, ingestible plant-derived mitochondria could form the basis of safe, noninvasive immunometabolic therapies for ALI/ARDS by restoring macrophage mitochondrial function. Translation will require rigorous safety, biodistribution, and efficacy studies in large animals and humans.
Key Findings
- Oral onion-derived mitochondria trafficked from gut to lung and were preferentially taken up by lung macrophages via PA–CR1L interaction.
- O-Mit fused with host macrophage mitochondria and reprogrammed energy metabolism to counter LPS-induced dysfunction.
- MDHB-enriched O-Mit epigenetically suppressed ND1, reducing complex I-driven oxidative stress, DRP1-mediated fission, and cardiolipin peroxidation, thereby rescuing LPS-induced ALI.
Methodological Strengths
- Mechanistic depth across molecular (ND1/DRP1/cardiolipin), cellular, and in vivo levels
- Oral delivery with demonstrated organ trafficking and cell-specific uptake
Limitations
- Mouse LPS-ALI model may not fully recapitulate human ALI/ARDS pathobiology
- Safety, immunogenicity, and dose scaling of plant-derived mitochondria in humans remain untested
Future Directions: Quantify biodistribution and persistence, assess immunogenicity and safety in large animals, validate efficacy across diverse ALI/ARDS models, and explore receptor/ligand specificity (PA–CR1L) in human macrophages.
Mitochondrial dysfunction contributes to various inflammatory-related diseases by triggering the release of inflammatory molecules. Targeting mitochondrial dysfunction is emerging as a promising avenue for treating inflammatory diseases. Here, it is demonstrated that dietary plant-derived mitochondria (P-Mit) are capable of rescuing the lung macrophage mitochondrial (M-Mit) dysfunction in lipopolysaccharide (LPS)-induced mouse acute lung injury (ALI). Specifically, oral administration of dietary onion-derived mitochondria (O-Mit) can travel from the gut to the lungs in ALI mice, where preferentially uptake by lung macrophage mediated by the interaction between O-Mit phosphatic acid (PA) and macrophage complement C3b/C4b receptor 1 Like (CR1L), followed by fusing with murine M-Mit and by reprograming the M-Mit energy metabolism in the lungs of ALI mice. Further evidence suggests that O-Mit enriches methyl 3,4-dihydroxybenzoate (MDHB) inhibits M-Mit NADH dehydrogenase subunit 1 (ND1) gene expression in the epigenetic process, which represses LPS-induced complex I-related oxidative stress activation and excessive mitochondrial fission via modulating dynamin-related protein 1 (DRP1) phosphorylation and cardiolipin peroxidation in M-Mit, eventually rescues the LPS-induced ALI. Given LPS-induced mouse model of ALI is widely used to study human ALI and acute respiratory distress syndrome, this finding provides a clinical potential for the treatment of human ALI via edible P-Mit.
2. Tidal volume and mortality during extracorporeal membrane oxygenation for acute respiratory distress syndrome: a multicenter observational cohort study.
In 1137 COVID-19 ARDS patients on ECMO across 29 German centers, ICU mortality was 75%. Predictors evolved over the first 14 days: age and lactate dominated on day 1, while on day 14 tidal volume per predicted body weight independently associated with mortality (aOR 0.693 per 1 mL/kg; p<0.001), with Vt <2 mL/kg marking very high adjusted mortality.
Impact: This large multicenter cohort provides time-sensitive prognostic insights during prolonged ECMO, highlighting tidal volume at day 14 as a robust predictor and informing ventilator strategy research in ECMO-dependent ARDS.
Clinical Implications: Day-14 tidal volume may serve as a dynamic prognostic marker and could guide ventilator targets in ECMO-dependent ARDS. Extremely low Vt (<2 mL/kg PBW) may flag high-risk patients for intensified evaluation and potential strategy adjustment.
Key Findings
- ICU mortality was 75% among 1137 ECMO-treated COVID-19 ARDS patients across 29 centers.
- Mortality predictors shifted over time: age and lactate on day 1; on day 14, tidal volume per PBW independently associated with mortality (aOR 0.693 per 1 mL/kg increase; p<0.001).
- Vt <2 mL/kg at day 14 corresponded to adjusted mortality exceeding 80% (lower 95% CI bound); higher Vt reflected higher compliance but benefit was inconsistent at very low driving pressures.
Methodological Strengths
- Large multicenter cohort (N=1137) with day-by-day granular modeling over 14 days
- Multivariable stepwise logistic regression with clinically relevant endpoints (ICU mortality)
Limitations
- Observational design limits causal inference; ventilation strategies were not randomized
- COVID-19-specific cohort from German centers may limit generalizability to non-COVID ARDS and other settings
Future Directions: Prospective trials to test ventilator targets for ECMO-dependent ARDS beyond the first fortnight; development of dynamic risk tools integrating compliance, driving pressure, and tidal volume.
BACKGROUND: Approximately half of the patients with acute respiratory distress syndrome (ARDS) receiving extracorporeal membrane oxygenation (ECMO) remain ECMO-dependent beyond 14 days after ECMO initiation. The identification of factors associated with mortality during an ECMO run may update prognostic assessment and focus clinical interventions. METHODS: In this observational study, data from 1137 patients with COVID-19 ARDS receiving ECMO support in 29 German centers between January 1st 2020 and July 31st 2021 were analyzed. Multivariable stepwise logistic regression analyses were performed to build survival prediction models with day-by-day data during the first 14 days of an ECMO run. The primary endpoint was all-cause mortality in the intensive care unit. RESULTS: Mortality in this cohort was high (75%). Patients who remained ECMO-dependent on day 14 of their ECMO run showed comparable mortality to all patients receiving ECMO support on day 1. Yet, factors associated with mortality changed during the first 14 days of ECMO support. On day 1 of ECMO support, only patient age and lactate remained in the final mortality prediction model. On day 14 of an ECMO run, tidal volume was independently associated with mortality (adjusted Odds Ratio 0.693 (95%CI 0.564-0.851), p < 0.001 for 1 mL/kg increase in tidal volume per predicted body weight). The adjusted mortality for patients with a tidal volume below 2 mL/kg on day 14 of their ECMO run was above 80% (lower limit of the 95%CI interval). Higher tidal volume was mainly based on higher respiratory system compliance. Yet, the benefit of higher compliance was not observed in some patients who were still ventilated with very low driving pressures despite remaining ECMO-dependent on day 14 of ECMO support. CONCLUSIONS: Mortality predictors change during the course of an ECMO run. In a cohort with high mortality, on day 14 of ECMO support for ARDS, tidal volume may be an independent predictor of mortality. Further analyses on ventilation strategies in patients who remain ECMO-dependent are needed. TRIAL REGISTRATION NUMBER: DRKS00022964, retrospectively registered.
3. Under the background of the new global definition of ARDS: an interpretable machine learning approach for predicting 28-day ICU mortality in patients with sepsis complicated by ARDS.
Using MIMIC-IV 2.2 and a Chinese external cohort aligned to the new global ARDS definition, an interpretable SVC model predicted 28-day ICU mortality in sepsis-associated ARDS (internal AUC 0.792; external AUC 0.816). Lasso-selected 15 variables and SHAP explanations support clinical transparency.
Impact: Provides an externally validated, interpretable prognostic tool built under the updated global ARDS definition, facilitating risk stratification and personalized management of sepsis-associated ARDS.
Clinical Implications: The model can triage high-risk patients early, inform goals-of-care, and prioritize targeted interventions in sepsis-associated ARDS. Integration into ICU workflows with real-time data feeds could enhance situational awareness.
Key Findings
- Developed an interpretable SVC model for 28-day ICU mortality in sepsis-associated ARDS using MIMIC-IV 2.2 with external validation.
- Model achieved internal AUC 0.792 and external AUC 0.816, with good calibration and DCA performance.
- Lasso-selected 15 predictors and SHAP values provided variable importance and directionality for clinical interpretability.
Methodological Strengths
- External validation under the new global ARDS definition
- Model interpretability via SHAP with rigorous algorithm benchmarking
Limitations
- Retrospective datasets with potential selection and information bias
- Single-center external validation may limit generalizability; prospective real-time validation is lacking
Future Directions: Prospective, multi-center deployment with real-time EHR integration, continuous recalibration, and impact evaluation against standard-of-care triage.
BACKGROUND: Acute respiratory distress syndrome (ARDS) is a prevalent clinical complication among patients with sepsis, characterized by high incidence and mortality rates. The definition of ARDS has evolved over time, with the new global definition introducing significant updates to its diagnosis and treatment. Our objective is to develop and validate an interpretable prediction model for the prognosis of sepsis patients complicated by ARDS, utilizing machine learning techniques in accordance with the new global definition. METHODS: This study extracted data from the MIMIC database (version MIMIC-IV 2.2) to create the training set for our model. For external validation, this study used data from sepsis patients complicated by ARDS who met the new global definition of ARDS, sourced from the Affiliated Hospital of Xuzhou Medical University. Lasso regression with cross-validation was used to identify key predictors of patient prognosis. Subsequently, this study established models to predict the 28-day prognosis following ICU admission using various machine learning algorithms, including logistic regression, random forest, decision tree, support vector machine classifier, LightGBM, XGBoost, AdaBoost, and multi-layer perceptron (MLP). Model performance was assessed using ROC curves, clinical decision curves (DCA), and calibration curves, while SHAP values were utilized to interpret the machine learning models. RESULTS: A total of 905 patients with sepsis complicated by ARDS were included in our analysis, leading to the selection of 15 key variables for model development. Based on the AUC of the ROC curve, as well as DCA and calibration curve results from the training set, the support vector classifier (SVC) model demonstrated strong performance, achieving an average AUC of 0.792 in the internal validation set and 0.816 in the external validation set. CONCLUSION: The application of machine learning methodologies to construct prognostic prediction models for sepsis patients complicated by ARDS, informed by the new global definition, proves to be reliable. This approach can assist clinicians in developing personalized treatment strategies for affected patients.