Modeling prognostic factors for poor neurological outcome in conservatively treated patients with intracerebral hemorrhage: A focus on TNF-α

Abstract

Objective: Prognostic models for Intracerebral hemorrhage (ICH), mainly based on clinical evaluation, have remained inherently confounded by subjective scoring assessments and limited accuracy. In this study, we aimed at assessing the risk for poor outcome after ICH based on peripheral biochemical markers (TNF-α, glutamate and glucose) and radiological variables (both at admission and five days after patient's care), for modeling purposes of prognostication.

Patients and methods: The defined initial variables of fifty non-comatose conservatively treated ICH patients without severe complications during the hospitalization process (as intraventricular bleeding, or hematoma expansion) were aligned with the evaluated parameters during re-evaluation (3 months later). A comprehensive statistical approach has been applied by using different modeling strategies for prediction of their functional status and outcome.

Results: Higher blood plasma glutamate, TNF-α and initial ICH volume at admission, as well as higher volumes of ICH and perihematomal edema after five days of care were significantly more likely associated with the poor outcome. Nevertheless, in all of the constructed models, TNF-α was estimated as the only significant predictive risk factor, thus outperforming the capacity of the initial ICH volume and the radiological variables after 5 days, both in terms of prognostication of the functional status and the 3-month neurological outcome. The constructed canonical variable that has fairly marked off the different outcomes was also mainly weighed by the admission TNF-α levels. For the first time, we have carefully developed probability functions for the neurological outcome as a response to the admission TNF-α levels; TNF-α levels >110.35 pg/mL were assessed as an optimal cutoff point fairly identifying patients who will fall into the group with poor outcome.

Conclusions: TNF-α based models and admission TNF-α screening might be appropriate as a key component that assists more objective prognostication and management of patient's care in clinical decision making, as rapid initial diagnosis and concentrated management are crucial for secondary prevention of further devastating neurological impairments after ICH.

Keywords: Intracerebral hemorrhage; Poor outcome; Probability function; Prognostication; TNF-α.