The majority of reports examining contributions of the two COX isoforms to neuroinflammation have focused on a causative role for the COX-2 isoform. Much of the emphasis on COX-2 may result from the availability and therapeutic success of COX-2-selective inhibitors. However, studies utilizing COX-1-deficient  or COX-2-deficient  mice have shown that inactivation of COX-1 reduces neuroinflammation whereas COX-2 inactivation worsens the neuroinflammatory response. Furthermore, in both humans and animal models, TBI results in the prolonged accumulation of COX-1 expressing microglia in the region of the developing lesion [2, 3]. Therefore, findings from previous reports would support the strategy of COX-1 inactivation as a potential method for improving outcomes following TBI. However, as described in the current study using the MWM method of determining cognitive functioning, there was no identifiable effect of COX-1 deficiency on behavioral outcomes following TBI in mice.
The strategy of using pharmacological inhibition of COX-2 for amelioration of behavioral outcomes resulting from TBI has produced differing results. In rats, COX-2 inhibition has been shown to improve cognitive functioning as determined by the MWM following a severe 3 mm CCI  and by the Barnes maze following an impact-acceleration model of diffuse TBI . However, Dash et al.  did not observe any differences in MWM performance between administration of the COX-2 inhibitor celecoxib or vehicle, following a milder 2 mm CCI. Another experimental brain injury study failed to see an improvement in the MWM following COX-2 inhibitor administration to juvenile rats . When the weight drop model was used to induce brain injury, the COX-2 inhibitor nimesulide did not improve Neurological Severity Score (NSS) . Furthermore, the MWM was recently used as an outcome measure following CCI performed on COX-2 null mutant mice. As we observed in our current study, Ahmed et al. found no significant cognitive effect resulting from COX-2 gene deletion following a 1.2 mm CCI in mice .
In addition to the behavioral results discussed above, there have also been inconsistent findings with studies using biochemical and histological endpoints to examine the effectiveness of COX-2 inhibition following TBI. Gopez et al.  showed that COX-2 inhibition decreased neuronal expression of the apoptotic marker, activated caspase-3, and reduced PGE2 levels in the brain. Hickey et al.  also found that COX-2 inhibition decreased PGE2 production, but this decrease was not accompanied by decreased brain edema or increased tissue sparing, an effect also seen by Koyfman et al. . Further investigation on neuroprotective effects of COX-2 inhibition up to 72 hours following injury showed no significant change in degenerating neuronal cell bodies, as detected with fluoro-jade B, or in DNA fragmentation, as determined by TUNEL staining . In contrast, COX-2 null mutants subjected to a 1.2 mm CCI did show a significant decrease in TUNEL positive cells at 24 hours after injury . In our current study, we observed a 6% difference in the amount of cortical tissue spared between the COX-2 +/+ mice and COX-2 -/- mice. However, cortical tissue surrounding the epicenter of the cortical impact is heavily damaged, both in cell bodies and axonal processes [22, 33]. Because we did not identify improvements in cognitive outcomes resulting from the deficiency of COX-2, the biological significance of the observed 6% improvement in cortical tissue sparing in COX-2-deficient mice is currently not clear.
Recently, there have been conflicting reports as to the specificity of [3H]-PK11195 for labeling activated microglia. Multiple neurodegeneration studies have shown a strong correlation between ex vivo binding of racemic or the (R) enantiomer of [3H]-PK11195 and immunohistochemical markers of microglia [27, 28, 34–36]. Some of these studies have also shown a correlation between [3H]-PK11195 binding and immunohistochemical markers of reactive astrocytes, though the correlation coefficient is much lower than with activated microglia [25, 27]. Additionally, [3H]-PK11195 binding sites have been found on neutrophils , although these cell types are significantly decreased or undetectable by seven days post-injury [38–41]. Therefore, although the increase in [3H]-PK11195 binding following brain injury is most often associated with activated microglia, the contribution of other cell types cannot be excluded. Recently, the brains of COX-2-deficient mice were shown to exhibit increased activation of both microglia and astrocytes following lipopolysaccharide administration . Thus, the increased [3H]-PK11195 binding that we observed in COX-2-deficient mice may indicate an increased inflammatory response involving multiple cell types.