Risk Prediction Model for Isoniazid Dosing in Tuberculosis Meningitis Patients in Southwest China.

Abstract

Purpose: Tuberculosis meningitis (TBM) has emerged as the most lethal type of disease. The prognosis of meningitis is often related to disease severity and early therapeutic intervention.

Methods: Patients were screened for primary TBM and received a quadruple regimen comprising isoniazid (standard dose of 300 mg/day and high dose of 600 mg/day), rifampin, ethambutol, and pyrazinamide. Further, the indices and prognosis factors of diseased patients were analyzed, using 12-month treatment mortality as the primary observation endpoint. Several predictors included demographic data, clinical presentation, ancillary tests, treatment changes, and isoniazid dose. The data were analyzed using a least absolute shrinkage, the selection operator regression, and multi-factor logistic regression.

Results: Among the selected TBM patients (n=119), 18 patients were dead at the end of December. A total of 68 influencing factors were screened, in which 5 clinical factors were included as potential prognostic factors, including older age, presence of nausea, high MRC grade, imaging suggestive of cerebral infarction, and dose of isoniazid (300 mg/day). The AUC value was recorded as 0.8316832. The validation set confirmed the model's robustness, with an AUC of 0.887 and good calibration performance. These findings highlight the model's potential for clinical application in optimizing isoniazid dosing. The model demonstrated the advantage of predicting the therapeutic outcome of patients.

Conclusion: In summary, the model could be suitable for evaluating the risk of death within 12 months in TBM patients towards assessing the severity and treatment needs of patients. The isoniazid dose is an important factor affecting the prognosis of these patients.