Cerebral hypoxia, along with hypotension, is one of the most critical factors worsening secondary brain damage after TBI, and particularly following diffuse TBI [6, 13]. Despite this clinical relevance, the underlying mechanisms by which hypoxia aggravates neurological outcome following TBI have not been studied adequately.
Using focal or mixed focal-diffuse models, systemic hypoxia following TBI in rats exacerbates neurological deficit [32, 37] and increases the lesion size, neuronal death [[33, 34, 37, 64]] and brain edema, while reducing cerebral blood flow [35, 51]. However, the role of post-traumatic hypoxia elicited after diffuse brain injury has rarely been addressed. Therefore, we explored the impact of hypoxia using a model of diffuse TAI [[40, 65, 66]] followed by a 30-min hypoxic ventilation. Using this combinatorial insult model, we previously reported enhanced axonal damage and macrophage infiltration within the corpus callosum and the brain stem . Thus, in this follow-up study we further investigated changes in neurological outcome, brain edema, ventricle enlargement, cerebral cytokines, and energy metabolism.
We found that in comparison to TAI alone, an additional hypoxic insult enhanced sensorimotor deficits on the Rotarod, beam walk and tape removal tests, reduced spontaneous exploratory behavior, and delayed recovery. These data closely relate to clinical studies on TBI patients showing that post-traumatic hypoxia worsens neurological outcome and prolongs the recovery period [[7, 8, 67]]. The behavioural data in this model of TAI are consistent with similar deficits shown at day 1 in previous studies using diffuse or focal TBI models in combination with hypoxia [[32, 34, 36–38, 68]]. However, in extension of this early work, our results show that an additional hypoxic insult has a detrimental effect on behaviour, inflammatory and metabolic outcomes for an extended period of time.
Brain swelling is a major contributor for the development of secondary ischemia causing raised ICP and decreased cerebral perfusion pressure . Enlargement of the brain due to edema  and/or obstruction of CSF flow  is a common event in severe TBI patients and a frequent cause of death. Cytotoxic edema results from excessive accumulation of ion and water within the cell, while vasogenic edema is caused by increased vascular permeability and subsequent fluid extravasation into the parenchyma. Here, we demonstrated that at 2 h after TAI, brain water content was similar to sham animals, but it increased to a peak between 24 and 48 h, and remained elevated until 72 h. Although hypoxia following TAI exacerbated sensorimotor deficit, it did not further increase cerebral edema when compared with TAI only animals, corroborating previous observations using diffuse-weighted imaging . Interestingly, using MRI, others demonstrated that acute brain swelling after TAI (both with and without hypoxia), as early as 60 min post-injury, was associated with increased extracellular fluid and BBB dysfunction, indicative of vasogenic edema [[72–75]]. This early brain swelling was transient, with values quickly returning to sham levels [[53, 58, 75]]. Since the earliest timepoint examined in our study was 2 h, it is likely that we missed this initial peak in edema, as no differences were detected between TAI, TAI+Hx and sham rats later on. However, other studies have also demonstrated that a modest, widespread second edematous response occurs at 24 h after TAI despite the intact BBB, which suggests ongoing cytotoxic edema [58, 75]. Our results are consistent with this modest yet significant increase of edema at 24 h, which was maintained until 48 h. It is possible that the peak in brain water content observed at 24 h in both the TAI and TAI+Hx rats (approximately 79.3%) reflects a sort of saturation level, with the brain unable to tolerate any further water accumulation. Other studies also demonstrated peak edema of similar degree after TBI [[59, 76, 77]].
An interesting observation was the enlargement of the lateral ventricles after TAI, and even greater following TAI+Hx. Recent clinical neuroimaging studies have shown correlations between ventricular enlargement and long-term neurological impairment [[78–80]]. The prognostic value of ventricular dilatation had high sensitivity and specificity for the prediction of cognitive outcome [[80–83]]. In this study, we showed that the lateral ventricles are markedly enlarged at 1 day post-injury after TAI and even larger in TAI+Hx animals, when compared to sham or rats with isolated hypoxia. Although we did not examine the mechanism leading to ventricular enlargement after TAI, imaging studies on TBI patients suggested that white matter degeneration around the lateral ventricle may be a contributing factor . However, since ventricular enlargement in TAI rats was an early and transient effect, it could be most likely attributed to the onset of post-traumatic hydrocephalus, caused by impaired CSF circulation due to edema compressing the aqueduct of sylvius.
Neuroinflammation has been extensively investigated in hypoxia-ischemia and TBI in both humans and animal models  and all these studies have reported a robust elevation of cytokines in the central nervous system [[19, 28, 86–89]]. More relevant for this study, our preliminary data on severe TBI patients with additional hypoxic insult have shown enhanced and prolonged production of cytokines in the CSF (Yan et al: Neuroinflammation and brain injury markers in TBI patients: Differences in focal and diffuse brain damage, and normoxic or hypoxic status on neurological outcome; manuscript in preparation). Consistently, here we demonstrated exacerbated production of IL-6, IL-1β, and TNF in the brains after TAI with additional hypoxia.
IL-1β is a key mediator of the inflammatory response, which exacerbates neuronal injury and induces BBB dysfunction by stimulating matrix metalloproteinases . IL-1β mRNA is upregulated within minutes after TBI, and increased protein levels are detectable within an hour after TBI [[21, 91–93]]. In this study, IL-1β increased early after TAI alone, peaking at 2 h. Post-TAI hypoxia significantly enhanced IL-1β concentration at 2 h compared to TAI-only rats. In addition, whilst the elevation of IL-1β in TAI-only rats appeared to be transient, in TAI+Hx rats IL-1β was still significantly elevated at 24 h, suggesting that the addition of hypoxia prolongs neuroinflammation.
The neurotoxic effects of IL-1β are synergistically enhanced in the presence of TNF , as both share many of the same physiologic effects. However, the role of TNF following TBI is controversial, neuronal toxicity of TNF has been demonstrated with local TNF administration inducing breakdown down of the BBB and increased leukocyte recruitment [[95–98]]. Clinically, high levels of TNF in the CSF of brain-injured patients correlated with BBB dysfunction . TNF inhibition also reduced cerebral ischemia/reperfusion injury , decreased TBI induced neuronal damage , and ameliorated BBB dysfunction after closed head injury . However, studies on TNF deficient mice demonstrated an early functional improvement between 24-48 h after TBI, but failed to produce further amelioration at 4 weeks . Taken together, these studies suggest that TNF may be deleterious in the acute phase post-injury, but beneficial for long-term recovery. In accordance with Kamm et al. , no changes in TNF levels were detected in rats subjected to TAI alone, whereas the combination of TAI and hypoxia elicited a significant early increase in TNF at 2 h post-injury, lasting up to 72 h post-injury. These early enhancement in the TAI+Hx rats possibly reflects a more severe brain damage in this combined insult model.
Similar to IL-1β and TNF, at 24 h IL-6 was significantly higher in TAI+Hx rats compared to TAI alone. IL-1β is an early mediator inducing the production of IL-6 at both mRNA and protein levels . IL-6 displays pleiotropic functions with both deleterious and beneficial effects in the injured brain [[104–106]]. Using the mild severity (250 g/2 m) of the Marmarou model, we showed that IL-6 increased in rat CSF within 24 h and IL-6 protein and mRNA was found expressed on neurons . Studies of IL-6 gene-deficient mice have provided more information in regards to the protective function of IL-6, by having a compromised immune response, increased oxidative stress and neurodegeneration . In this study, we demonstrate significantly heightened IL-6 levels in the TAI+Hx rats at 24 h, which remained elevated above TAI levels until 96 h. Altogether, the increased acute production of IL-1β and TNF may be associated with disruption of BBB integrity and consequently formation of cerebral edema, while late elevation of IL-6 may trigger repair mechanisms [24, 99].
We also investigated changes in energy metabolism in this combinatorial insult model. Due to the nature of the impact acceleration injury, it is impractical to implant a microdialysis probe prior to injury without compromising the integrity of the trauma. It is also difficult to implant the probe directly after trauma as it resulted in higher mortality rate. Carré and colleagues implanted the probe 2 weeks prior to injury, but without success . We therefore allowed the rats to recover for 4 hours after TAI before implanting the microdialysis probe. In accordance with others , our study has shown that in sham rats energy metabolism is altered during the first 24 h following microdialysis probe implantation, therefore we chose to examine only the data from 20 h onwards to reduce the "probe effect".
At 21 h, the glucose values for TAI+Hx rats were substantially lower compared to TAI or sham rats, and dropped to extremely low levels from 57 h onwards. These low levels of cerebral glucose could be the result of low glucose availability and/or hyperglycolysis in the acute post-injury phase. Hyperglycolysis has previously been shown as common early event following neurotrauma both experimentally and in the clinic [109, 110]. It is often followed by a prolonged period of metabolic depression beginning as early as 6 h post-injury, remaining for as long as 5 days [111, 112], a phenomenon which has also been demonstrated in the present study. Interestingly, rats subjected to TAI experienced only a brief period of glucose depletion between 39 h and 57 h, at which time glucose levels returned to sham values for the remaining duration of monitoring. It is possible that the additional hypoxic insult depleted available glucose stores in the TAI+Hx animals, and thus a prolonged compensatory period of anaerobic respiration occurred to provide essential ATP and generate lactate as by-product. Our experiments have demonstrated that this is a protracted process, lasting for 51 h after TAI. Lactate may be utilized by the brain during periods of increased brain energy requirements in which ATP and glucose stores are exhausted, such as following TBI [113, 114]. In a situation of prolonged glucose depletion, high concentrations of lactate and high-level energy usage for neuronal repair or alternative metabolic pathways may further reduce the ATP reserves, with a subsequent mismatch between glucose transport, uptake and ATP production [115, 116]. This may explain the further drop in glucose concentrations at 57 h post TAI, in that the restoration of aerobic metabolism decreases lactate concentration but further reduces glucose. Post-traumatic impairment in energy metabolism is a major contributor to cytotoxic edema, and interestingly, the period of elevated lactate in the TAI+Hx rats between 21 h and 57 h overlaps with the peak of increased brain water content. As edema begins to reside, lactate levels in these rats return to sham values. This prolonged period of metabolic crisis also extends to glutamate production, which was depressed below sham levels for TAI, and particularly TAI+Hx rats, for the duration of the monitoring by microdialysis.