Relationship of Serum Uric Acid to Hematoma Volume and Prognosis in Patients with Acute Supratentorial Intracerebral Hemorrhage

Enhancing Robustness of Machine Learning Integration With Routine Laboratory Blood Tests to Predict Inpatient Mortality After Intracerebral Hemorrhage

Objective: The accurate evaluation of outcomes at a personalized level in patients with intracerebral hemorrhage (ICH) is critical clinical implications. This study aims to evaluate how machine learning integrates with routine laboratory tests and electronic health records (EHRs) data to predict inpatient mortality after ICH.

Methods: In this machine learning-based prognostic study, we included 1,835 consecutive patients with acute ICH between October 2010 and December 2018. The model building process incorporated five pre-implant ICH score variables (clinical features) and 13 out of 59 available routine laboratory parameters. We assessed model performance according to a range of learning metrics, such as the mean area under the receiver operating characteristic curve [AUROC]. We also used the Shapley additive explanation algorithm to explain the prediction model.

Results: Machine learning models using laboratory data achieved AUROCs of 0.71–0.82 in a split-by-year development/testing scheme. The non-linear eXtreme Gradient Boosting model yielded the highest prediction accuracy. In the held-out validation set of development cohort, the predictive model using comprehensive clinical and laboratory parameters outperformed those using clinical alone in predicting in-hospital mortality (AUROC [95% bootstrap confidence interval], 0.899 [0.897–0.901] vs. 0.875 [0.872–0.877]; P <0.001), with over 81% accuracy, sensitivity, and specificity. We observed similar performance in the testing set.

Conclusions: Machine learning integrated with routine laboratory tests and EHRs could significantly promote the accuracy of inpatient ICH mortality prediction. This multidimensional composite prediction strategy might become an intelligent assistive prediction for ICH risk reclassification and offer an example for precision medicine.

Prestroke Metformin Use on the 1-Year Prognosis of Intracerebral Hemorrhage Patients with Type 2 Diabetes

Background

Although recent studies have focused on the use of metformin in treating ischemic stroke, there is little literature to support whether it can treat intracerebral hemorrhage (ICH). Therefore, this study is aimed at evaluating the possible effects of prestroke metformin (MET) on ICH patients with type 2 diabetes.

Methods

From January 2010 to December 2019, all first-ever ICH patients with type 2 diabetes from our hospitals were included. All discharged patients would receive a one-time follow-up at 1 year after admission. Death, disability, and recurrence events were recorded.

Results

We included 730 patients for analysis (the median age: 65 [IQR, 56-72] years and 57.7% was men). Of those patients, 281 (38.5%) had received MET before ICH (MET+), whereas 449 (61.5%) had not (MET−). MET (+) patients had a lower median baseline hematoma volume than did MET (-) patients (9.6 ml [IQR, 5.3-22.4 ml] vs. 14.7 ml [IQR, 7.9-28.6 ml]; P < 0.001). The inhospital mortality events were not significantly reduced in the MET (+) group compared with the MET (-) group (6.4% vs 8.9%, respectively; absolute difference, −2.5% [95% CI, −3.9% to −0.7%]; OR, 0.70 [95% CI, 0.39 to 1.27]; P = 0.22). The 1-year mortality events were not significantly reduced in the MET (+) group compared with the MET (-) group (14.1% vs 17.4%, respectively; absolute difference, −3.3% [95% CI, −5.1% to −1.8%]; OR, 0.73 [95% CI, 0.47 to 1.14]; P = 0.16). The 1-year disability events were not significantly reduced in the MET (+) group compared with the MET (-) group (28.4% vs 34.1%, respectively; absolute difference, −5.7% [95% CI, −8.2% to −3.3%]; OR, 0.77 [95% CI, 0.52 to 1.13]; P = 0.18). Finally, the recurrence rates in those two groups were not significantly different (MET [+] vs. MET [-]: 6.4% vs. 5.9%; absolute difference, 0.5% [95% CI, 0.2% to 1.3%]; OR, 1.08 [95% CI, 0.51 to 2.28]; P = 0.84).

Conclusions

Pre-ICH metformin use was not associated with inhospital mortality and 1-year prognosis in diabetic ICH patients.

Comparing the inter-observer reliability of the Tada formula among neurosurgeons while estimating the intracerebral haematoma volume

OBJECTIVE

To compare the inter-observer reliability among neurosurgeons while estimating the intracerebral haematoma (ICH) volume by the Tada formula and assess its influence on predicting the severity and prognosis of various ICHs.

METHODS

We obtained clinical data from 262 consecutive patients with spontaneous ICH. The haematoma volume was independently calculated and compared by 3D Slicer and eight neurosurgeons. The inter-observer reliability was obtained by calculating the intraclass correlation coefficients (ICC) and Cohen's kappa score (kappa), within different shape and volume ICH subgroups. We conducted the receiver operating characteristic analysis to assess the predictive value of the ICH volume evaluated for clinical features, including the Glasgow Coma Scale at the onset of the disease, ICH-related surgical treatments, the length of stay in the intensive care unit, the length of hospitalisation, the modified Rankin Scale score at discharge, and in-hospital deaths.

RESULTS

The median haematoma volume was 17.4 ml (range, 7.3-34.7 ml). The estimated volumes were significantly different among neurosurgeons (p < 0.001). Six out of eight neurosurgeons demonstrated obvious deviations from the 3D Slicer software (p < 0.001). Round (ICC: 0.947) and tapered (ICC: 0.954) haematomas were more consistently evaluated between the neurosurgeons. We observed a substantial strength of agreement between neurosurgeons with kappa> 0.693 and ICC: 0.938 in the entire volume range, and slight to fair strength of agreement with kappa> 0.175 and ICC: 0.689 between 20 ml and 40 ml volume interval. All estimated volumes had a positive predictive value for clinical features, with the area under the curve > 0.5 (p < 0.05). However, the 3D Slicer software performed relatively better than most neurosurgeons.

CONCLUSIONS

There exists a significant inter-observer variability among neurosurgeons when utilizing the Tada formula, thus demonstrating significant implications for ICH-related clinical practices and researches.

Temporal metabolomic alteration in rat brains of experimental intracerebral hemorrhage

BACKGROUND

Intracerebral hemorrhage (ICH) is the top lethal and disabling form of stroke. The pathophysiology of ICH is not fully understood yet. Metabolites are indicators and regulators of cellular processes. However, the overall brain metabolic pattern and the temporal alterations after ICH remain unknown.

METHODS

A total of 40 male rats were randomly assigned to sham group and ICH group. ICH was induced by collagenase Ⅶ. Body weight was assessed. Neurological deficits were evaluated by modified neurological severity score. Then, the perihematomal brain tissues were collected for metabolites detection using high-performance liquid chromatography-tandem mass spectrometry (HPLC-MS). The metabolic profiles were displayed by principal component analysis (PCA), partial least-squares-discriminant analysis (PLS-DA) and cluster analysis. The significant differential metabolites were screened by fold change > 2.0, the false discovery rate (FDR) < 0.05 and Variable Importance of Projection (VIP) > 1. Next, the relevant metabolic pathways were discerned by MetaboAnalyst website. A metabolite-protein interaction network was subsequentially constructed to further annotate the function of differential metabolites.

RESULTS

Rats suffered from compromised body weight increasement and impaired neurological function. The metabolomics profiles of brain tissues in the post-ICH rats were markedly different from those in the sham group on days 3 and 14. Thirty-four metabolites (bilirubin, uric acid, 6-Methylnicotinamide et al.) were abnormally upregulated in the acute stage, while 27 metabolites were disturbed in the recovery stage, including bilirubin, uric acid, and histamine et al. Seven and three metabolic pathways altered in the acute and recovery stage, respectively. Metabolite-protein interaction analysis revealed that the disturbed metabolites may participate in ICH pathophysiology by altering amino acid metabolism, peroxisome proliferators-activated receptor signaling pathway, fatty acid metabolism and urea cycle in the acute stage, while influencing amino acid metabolism, urea cycle and peroxisome in the recovery stage.

CONCLUSIONS

Our study mapped the pathological metabolomics profiles of the post-ICH rat brains in the acute and recovery phases. This work will assist in discovering novel therapeutic targets and treatments for ICH.