A Retrospective Analysis of Central and Peripheral Metabolic Characteristics in Patients with Cryptococcal Meningitis

Machine Learning-Based Diagnosis of Cryptococcal Meningitis and Tuberculous Meningitis: A Single-Center Retrospective Clinical Study

Background

Differentiating cryptococcal meningitis from tuberculous meningitis is critical for initiating appropriate treatment and improving patient outcomes. However, overlapping pathogenic mechanisms, clinical presentations, laboratory findings, and imaging features complicate accurate diagnosis.

Methods

We performed a retrospective single-center study at the First Hospital of Changsha, analyzing medical records from January 2021 to November 2024. A total of 271 patients were included: 125 diagnosed with cryptococcal meningitis and 146 with tuberculous meningitis. Using LASSO regression (least absolute shrinkage and selection operator), we identified 11 potential predictors, which were subsequently refined to 6 key variables: extracranial fungi, cerebrospinal fluid pressure, age, erythrocyte sedimentation rate, albumin levels, and India ink staining. A diagnostic nomogram was constructed through multivariate logistic regression and validated by receiver operating characteristic analysis, calibration plots, decision curve analysis, and SHAP tools (Shapley additive explanations) for model interpretability.

Results

The optimized model incorporated 6 variables: extracranial fungal presence, cerebrospinal fluid pressure, patient age, erythrocyte sedimentation rate, albumin levels, and India ink staining. According to SHAP analysis, the significance of these factors in prediction was underscored. The model achieved an area under the curve of 0.919 for the training cohort and 0.921 for the validation cohort, demonstrating robust sensitivity and specificity across both data sets. Additionally, calibration plots and decision curve analysis validated the model's precision and its applicability in clinical settings.

Conclusions

The developed nomogram accurately distinguishes cryptococcal meningitis from tuberculous meningitis with superior discriminative ability and clinical relevance, providing a valuable diagnostic tool for clinicians.

Lactate-Dehydrogenase-5 May Play a Key Role in the Disturbance of Brain Energy Caused by Tuberculous Meningitis

BACKGROUND

The conversion of pyruvate to lactate is primarily catalyzed by lactate-dehydrogenase-5 (LDH-5), which comprises four lactate-dehydrogenase-A (LDHA) subunits. However, the mechanism of LDH-5 in tuberculous meningitis (TBM) remains elusive.

METHODS

Thirty-two samples of cerebrospinal fluid (CSF) were collected, including 15 from individuals without central nervous system (CNS) infectious diseases (control group) and 17 from individuals with TBM (TBM group). Based on the results of brain imaging, nine patients with TBM with meningeal enhancement were included in the meninges group. Eight patients with TBM with lesions in the brain parenchyma were included in the brain parenchyma group. The levels of adenosine triphosphatase (ATP), lactate, LDH-1, pyruvate and LDH-5 in the CSF were assessed. Subsequently, the levels of ATP, pyruvate and lactate, as well as the amplitude and frequency of action potentials (APs) in neurons overexpressing LDHA, were investigated.

RESULTS

Reduced levels of pyruvate and ATP and elevated levels of lactate and LDH-5 were observed in the CSF of individuals with TBM. The ATP level was decreased in the brain parenchyma group. In neurons with LDHA overexpression, the lactate level increased, while ATP and pyruvate levels, as well as the amplitude and frequency of APs, decreased.

CONCLUSION

Elevated levels of LDH-5 in the CNS of individuals with TBM may lead to a disturbance in brain energy and negatively affect neuronal activity.

NDUFA11 may be the disulfidptosis-related biomarker of ischemic stroke based on integrated bioinformatics, clinical samples, and experimental analyses

Background

Programmed cell death plays an important role in neuronal injury and death after ischemic stroke (IS), leading to cellular glucose deficiency. Glucose deficiency can cause abnormal accumulation of cytotoxic disulfides, resulting in disulfidptosis. Ferroptosis, apoptosis, necroptosis, and autophagy inhibitors cannot inhibit this novel programmed cell death mechanism. Nevertheless, the potential mechanisms of disulfidptosis in IS remain unclear.

Methods

The GSE16561 dataset was used to screen for differentially expressed disulfidptosis-related biomarkers (DE-DRBs). A correlation between the DE-DRBs was detected. The optimal machine-learning (ML) model and predictor molecules were determined. The GSE58294 dataset was used to verify the accuracy of the optimal ML model. The DE-DRB expression was detected in the blood of patients with IS. Based on IS models, experimental analyses were performed to verify DE-DRB expression and the correlation between DE-DRBs.

Results

Leucine-rich pentatricopeptide repeat-containing (LRPPRC) and NADH dehydrogenase [ubiquinone] 1 alpha subcomplex subunit 11 (NDUFA11) were identified as DE-DRBs. The NADH: ubiquinone oxidoreductase core subunit S1 (NDUFS1) interacted with NDUFA11 and LRPPRC. The support vector machine (SVM) model was identified as the optimal ML model. The NDUFA11 expression level in the blood of patients with IS was 20.9% compared to that in normal controls. NDUFA11 expression was downregulated in the in vitro/in vivo models of IS. The number of formed complexes of NDUFS1 and NDUFA11 decreased in the in vitro/in vivo models of IS.

Conclusion

This research suggests that NDUFA11 is a specific DRB for IS and demonstrates alterations in the disulfidptosis-related protein complexes NDUFS1-NDUFA11.