Extracellular brain pH with or without hypoxia is a marker of profound metabolic derangement and increased mortality after traumatic brain injury

Abstract

Cerebral hypoxia and acidosis can follow traumatic brain injury (TBI) and are associated with increased mortality. This study aimed to evaluate a relationship between reduced pH(bt) and disturbances of cerebral metabolism. Prospective data from 56 patients with TBI, receiving microdialysis and Neurotrend monitoring, were analyzed. Four tissue states were defined based on pH(bt) and P(bt)O(2): 1--low P(bt)O(2)/pH(bt), 2--low pH(bt)/normal P(bt)O(2), 3--normal pH(bt)/low P(bt)O(2), and 4--normal pH(bt)/P(bt)O(2)). Microdialysis values were compared between the groups. The relationship between P(bt)O(2) and lactate/pyruvate (LP) ratio was evaluated at different pH(bt) levels. Proportional contribution of each state was evaluated against mortality. As compared with the state 4, the state 3 was not different, the state 2 exhibited higher levels of lactate, LP, and glucose and the state 1--higher LP and reduced glucose (P<0.001). A significant negative correlation between LP and P(bt)O(2) (rho=-0.159, P<0.001) was stronger at low pH(bt) (rho=-0.201, P<0.001) and nonsignificant at normal pH(bt) (P=0.993). The state 2 was a significant discriminator of mortality categories (P=0.031). Decreased pH(bt) is associated with impaired metabolism. Measuring pH(bt) with P(bt)O(2) is a more robust way of detecting metabolic derangements.

Figure 1

Mean (±s.e.) values of (A) lactate/pyruvate (LP) ratio, (B) cerebral extracellular lactate, (C) extracellular glucose, and (D) extracellular glycerol by the tissue state categories.

Figure 2

Scatterplot depicting relationship between lactate/pyruvate (LP) ratio and PbtO₂ depending on the level of pHbt. The dotted fit line represents the subgroup of data with low pHbt (<7.1 in pericontusional brain tissue and <7.15 in less injured brain) and the solid gray line corresponds to the subgroup with normal pHbt (SPSS LOWESS (locally weighted scatterplot smoothing) method of fitting regression lines to the scatterplot was used).

Figure 3

(A) Mean (±s.d.) percentages of monitoring time belonging to the tissue state 2 (low pHbt and normal PbtO₂) by mortality categories, (B) ROC curve representing a duration of time spent in tissue state 2 as a predictor of mortality (P=0.031, AUC: 0.765). AUC, area under the curve; ROC, receiver operating characteristic.