Lei Li

Sepsis has heterogeneous clinical trajectories, but conventional severity scores offer only static risk estimates. Timely, dynamic prediction could enable personalized intervention. In this multicenter retrospective study of 47,936 ICU patients meeting Sepsis-3 criteria from one institutional and two public datasets (MIMIC-III, eICU; sensitivity in MIMIC-IV), group-based trajectory modeling identified latent recovery patterns. An ensemble machine-learning model incorporating dynamic physiological variability was trained, temporally validated, and externally tested; clinical impact was assessed following implementation. Three trajectories emerged: rapid recovery (41.5%), slow recovery (36.4%), and clinical deterioration (22.1%). In the final binary classification task, AUROC was 0.92 (development), 0.89 (internal), 0.84 (MIMIC-III) and 0.77 (eICU); median warning time before deterioration was 17.6 h (Overall pooled across all cohorts). Reduced heart rate variability (SD < 10 bpm) predicted mortality (adjusted HR 2.17). Implementation reduced ICU stay by 1.8 days, machanical ventilation by 2.3 days, and 28-day mortality by 5.7%. This externally validated trajectory-based model offers accurate, early risk stratification for sepsis, supporting proactive, individualized critical care.

Background: Colorectal cancer (CRC) is a prevalent gastrointestinal malignancy, ranking third in incidence and second in cancer-related mortality. Despite therapeutic advances, challenges such as chemotherapy toxicity and drug resistance persist. Thus, there is an urgent need for novel CRC treatments. However, developing new drugs is time-consuming and resource-intensive. As a more efficient approach, drug repurposing offers a promising alternative for discovering new therapies. Methods: In this study, we screened 1068 small molecular compounds from an FDA-approved drug library in CRC cells. Menadione was selected for further study based on its activity profile. Mechanistic analysis included a cell death pathway PCR array, differential gene expression, enrichment, and network analysis. Gene expressions were validated by RT-qPCR. Results: We identified menadione as a potent anti-tumor drug. Menadione induced three programmed cell death (PCD) signaling pathways: necroptosis, apoptosis, and autophagy. Furthermore, we found that the anti-tumor effect induced by menadione in CRC cells was mediated through a key gene: MAPK8. Conclusions: By employing methods of cell biology, molecular biology, and bioinformatics, we conclude that menadione can induce multiple forms of PCD in CRC cells by activating MAPK8, providing a foundation for repurposing the “new use” of the “old drug” menadione in CRC treatment.

Background: Drug-induced immune hemolytic anemia (DIIHA) is a rare but serious condition, with an estimated incidence of one in 100,000 cases, associated with various antibiotics. This study reports on a case of ceftizoxime-induced hemolysis observed in a patient in China.Case description: A Chinese patient diagnosed with malignant rectal cancer underwent antimicrobial therapy after laparoscopic partial recto-sigmoid resection (L-Dixon). After receiving four doses of ceftizoxime, the patient developed symptoms including rash, itchy skin, and chest distress, followed by a rapid decline in hemoglobin levels, the presence of hemoglobin in the urine (hemoglobinuria), renal failure, and disseminated intravascular coagulation. Laboratory analysis revealed high-titer antibodies against ceftizoxime and red blood cells (RBCs) in the patient’s serum, including immunoglobulin M (IgM) (1:128) antibodies and immunoglobulin G (IgG) (1:8) antibodies, with noted crossreactivity to ceftriaxone. Significant improvement in the patient’s hemolytic symptoms was observed following immediate discontinuation of the drug, two plasma exchanges, and extensive RBC transfusion.Conclusion: This case, together with previous reports, underscores the importance of considering DIIHA in patients who exhibit unexplained decreases in hemoglobin levels following antibiotic therapy. A thorough examination of the patient’s medical history can provide crucial insights for diagnosing DIIHA. The effective management of DIIHA includes immediate cessation of the implicated drug, plasma exchange, and transfusion support based on the identification of specific drug-dependent antibodies through serological testing.

Sepsis involves endothelial cell (EC) dysfunction, which contributes to multiple organ failure. To improve therapeutic prospects, elucidating molecular mechanisms of vascular dysfunction is of the essence. ATP-citrate lyase (ACLY) directs glucose metabolic fluxes to de novo lipogenesis by generating acetyl-Co-enzyme A (acetyl-CoA), which facilitates transcriptional priming via protein acetylation. It is well illustrated that ACLY participates in promoting cancer metastasis and fatty liver diseases. Its biological functions in ECs during sepsis remain unclear. We found that plasma levels of ACLY were increased in septic patients and were positively correlated with interleukin (IL)-6, soluble E-selectin (sE-selectin), soluble vascular cell adhesion molecule 1 (sVCAM-1), and lactate levels. ACLY inhibition significantly ameliorated lipopolysaccharide challenge-induced EC proinflammatory response in vitro and organ injury in vivo. The metabolomic analysis revealed that ACLY blockade fostered ECs a quiescent status by reducing the levels of glycolytic and lipogenic metabolites. Mechanistically, ACLY promoted forkhead box O1 (FoxO1) and histone H3 acetylation, thereby increasing the transcription of c-Myc (MYC) to facilitate the expression of proinflammatory and gluco-lipogenic genes. Our findings revealed that ACLY promoted EC gluco-lipogenic metabolism and proinflammatory response through acetylation-mediated MYC transcription, suggesting ACLY as the potential therapeutic target for treating sepsis-associated EC dysfunction and organ injury.

We aimed to develop an early warning system for real-time sepsis prediction in the ICU by machine learning methods, with tools for interpretative analysis of the predictions. In particular, we focus on the deployment of the system in a target medical center with small historical samples. Light Gradient Boosting Machine (LightGBM) and multilayer perceptron (MLP) were trained on Medical Information Mart for Intensive Care (MIMIC-III) dataset and then finetuned on the private Historical Database of local Ruijin Hospital (HDRJH) using transfer learning technique. The Shapley Additive Explanations (SHAP) analysis was employed to characterize the feature importance in the prediction inference. Ultimately, the performance of the sepsis prediction system was further evaluated in the real-world study in the ICU of the target Ruijin Hospital. The datasets comprised 6891 patients from MIMIC-III, 453 from HDRJH, and 67 from Ruijin real-world data. The area under the receiver operating characteristic curves (AUCs) for LightGBM and MLP models derived from MIMIC-III were 0.98 − 0.98 and 0.95 − 0.96 respectively on MIMIC-III dataset, and, in comparison, 0.82 − 0.86 and 0.84 − 0.87 respectively on HDRJH, from 1 to 5 h preceding. After transfer learning and ensemble learning, the AUCs of the final ensemble model were enhanced to 0.94 − 0.94 on HDRJH and to 0.86 − 0.9 in the real-world study in the ICU of the target Ruijin Hospital. In addition, the SHAP analysis illustrated the importance of age, antibiotics, net balance, and ventilation for sepsis prediction, making the model interpretable. Our machine learning model allows accurate real-time prediction of sepsis within 5-h preceding. Transfer learning can effectively improve the feasibility to deploy the prediction model in the target cohort, and ameliorate the model performance for external validation. SHAP analysis indicates that the role of antibiotic usage and fluid management needs further investigation. We argue that our system and methodology have the potential to improve ICU management by helping medical practitioners identify at-sepsis-risk patients and prepare for timely diagnosis and intervention. Trial registration: NCT05088850 (retrospectively registered).

Background The lack of information interoperability between different devices and systems in the ICU hinders further utilization of data, especially for early warning of specific diseases in the ICU. Objectives We aimed to establish a real-time early warning system for sepsis based on a data integration system that can be implemented at the bedside of the intensive care unit (ICU), named SEPRES. Methods Data is collected from bedside devices through the integration hub and uploaded to the integration system through the local area network. The data integration system was designed to integrate vital signs data, laboratory data, ventilator data, demographic data, pharmacy data, nursing data, etc. from multiple medical devices and systems. It integrates, standardizes, and stores information, making the real-time inference of the early warning module possible. The built-in sepsis early warning module can detect the onset of sepsis within 5 hours preceding at most. Results Our data integration system has already been deployed in Ruijin Hospital, confirming the effectiveness of our system. Conclusions We highlight that SEPRES has the potential to improve ICU management by helping medical practitioners identify at-sepsis-risk patients and prepare for timely diagnosis and intervention.

Severe cytokine release syndrome (CRS) and immune effector cell-associated neurotoxicity syndrome (ICANS) strongly hampered the broad clinical applicability of chimeric antigen receptor T cell (CAR-T) therapy. Vascular endothelial activation has been suggested to contribute to the development of CRS and ICANS after CAR-T therapy. However, therapeutic strategies targeting endothelial dysfunction during CAR-T therapy have not been well studied yet. Here, we found that tumor necrosis factor α (TNFα) produced by CAR-T cells upon tumor recognition and interleukin 1β (IL1β) secreted by activated myeloid cells were the main cytokines in inducing endothelial activation. Therefore, we investigated the potential effectiveness of TNFα and IL1β signaling blockade on endothelial activation in CAR-T therapy. The blockade of TNFα and IL1β with adalimumab and anti-IL1β antibody respectively, as well as the application of focal adhesion kinase (FAK) inhibitor, effectively ameliorated endothelial activation induced by CAR-T, tumor cells, and myeloid cells. Moreover, adalimumab and anti-IL1β antibody exerted synergistic effect on the prevention of endothelial activation induced by CAR-T, tumor cells, and myeloid cells. Our results indicate that TNFα and IL1β blockade might have therapeutic potential for the treatment of CAR-T therapy-associated CRS and neurotoxicity.

Objective: We aimed to develop an early warning system for real-time sepsis prediction in the ICU by equipping with interpretation analysis and transfer learning tools to improve the feasibility to deploy the sepsis prediction system, particularly to target cohorts. Design: Retrospective and observational study. Setting: Medical Information Mart for Intensive Care (MIMIC) dataset, the private Historical Database of local Ruijin Hospital (HDRJH), and data collected from Ruijin real-world study. Patients: 6891 patients from MIMIC dataset and 453 patients from HDRJH for model development and 67 cases from Ruijin real-world data for model evaluation. Interventions: None. Measurements and Main Results: Light Gradient Boosting Machine (LightGBM) and multilayer perceptron (MLP) were trained on MIMIC dataset and then finetuned on HDRJH using transfer learning technique. Ultimately, the performance of the sepsis prediction system was further evaluated in the real-world study in the ICU of the target Ruijin Hospital. The area under the receiver operating characteristic curves (AUCs) for LightGBM and MLP models derived from MIMIC were 0.98–0.98 and 0.95–0.96 respectively on MIMIC dataset, and, in comparison, 0.82–0.86 and 0.84–0.87 respectively on HDRJH, from 1–5h preceding. After transfer learning and ensemble learning, the AUCs of the final ensemble model were enhanced to 0.94–0.94 on HDRJH and to 0.86–0.9 in the real-world study in the ICU of the target Ruijin Hospital. In addition, the Shapley additive explanation (SHAP) analysis illustrated the importance of age, antibiotics, net balance, and ventilation for sepsis prediction, making the model interpretable. Conclusions: Our machine learning model allows accurate real-time prediction of sepsis within 5-h preceding. Transfer learning can effectively improve the feasibility to deploy the prediction model in the target cohort, effectively ameliorating the model performance for external validation. SHAP analysis may illuminate the importance of optimizing antibiotic use and restricting fluid management. Trial registration: NCT05088850 (retrospectively registered). Key Points: Question: We aimed to develop an early warning system for real-time sepsis prediction in the ICU and to improve the feasibility to deploy the system to target cohorts. Findings: Transfer learning technique effectively enhanced the AUCs for LightGBM and MLP models on the target cohort, HDRJH, from 0.82–0.86 and 0.84–0.87 to 0.93-0.94 and 0.92-0.93 for 1-5 hour preceding. Additionally, SHAP analysis illuminated the importance of optimizing antibiotic use and restricting fluid management. Meaning: Transfer learning can improve the feasibility to deploy the prediction model to the target cohort, and SHAP analysis made the prediction model interpretable.

To determine whether infection-prevention and control (IPC) interventions can reduce the colonisation and infection of intensive care unit (ICU)-acquired carbapenem-resistant Klebsiella pneumoniae (CRKP) in a general ICU ward in China. We used a quasi-experimental before-and-after study design. The study was conducted in 4 stages: baseline period, January 2013–June 2013; IPC interventions period including de-escalation and targeted bundle interventions, July 2013–June 2014; modified IPC interventions period, July 2014–June 2015; and follow-up period, July 2015–June 2016. We used modified de-escalation interventions according to patient-risk assessments to prevent the transmission of CRKP. A total of 629 patients were enrolled in study. The incidence of ICU-acquired CRKP colonisation/infection was 10.08 (4.43–16.43) per 1000 ICU patient-days during the baseline period, and significantly decreased early during the IPC interventions, but the colonisation/infections reappeared in April 2014. During the modified IPC intervention and follow-up periods, the incidence of ICU-acquired CRKP colonisations/infections reduced to 5.62 (0.69–6.34) and 2.84 (2.80–2.89), respectively, with ongoing admission of cases with previously acquired CRKP. The incidence of ICU-acquired CRKP catheter-related bloodstream infections decreased from 2.54 during the baseline period to 0.41 during the follow-up period. The incidence of ventilator-associated pneumonia and skin and soft tissue infections showed a downward trend from 2.84 to 0.41 and from 3.4 to 0.47, respectively, with slight fluctuations. Comprehensive IPC interventions including de-escalation and targeted bundle interventions showed a significant reduction in ICU-acquired CRKP colonisations/infections, despite ongoing admission of patients colonised/infected with CRKP.