differences between Loganoside injured and control dogs may have been confounded by the constitutive activity of MMP-2 in CSF that may have masked any increase in MMP-9. There are likely a number of possible explanations for why GM6001 failed to improve neurological recovery in spinal cord injured dogs. First, while GM6001 has been shown to improve neurological outcomes in various rodent models of brain and spinal cord injury, no studies to date have evaluated efficacy in dogs. Thus, there may be species differences in responsiveness to GM6001 and/or MMP-directed pathogenesis. Additionally, effects of GM6001 demonstrated in rodents may not be sufficiently robust to positively influence outcome under the clinical conditions of this study. Second, the drug was active beyond the first several days post-injury and as such could have interfered with mechanisms underlying recovery in SCI. Pharmacokinetics in healthy dogs demonstrated that plasma concentration of GM6001, present at even the 96-hour timepoint, NSC 693255 chemical information approximated or exceeded that necessary to block MMP-9 in vitro. As some MMPs modulate the formation of a glial scar and axonal plasticity, their subacute/chronic blockade may result in adverse neurological outcomes. Third, the timing between SCI and administration of GM6001 may not have been optimal. The strong association between MMP-9 expression and neutrophils suggests that an optimal therapeutic window for GM6001 is defined by the early trafficking of neutrophils into the injured cord. Such a position is supported by evidence of pronounced neurological recovery when the drug was given beginning 3 hours post-injury in a murine model of SCI. In dogs treated with GM6001, median delay between injury and enrollment was 12 hours, which may have exceeded the window of efficacy for GM6001. Finally, while the use of dogs with thoracic and lumbar spinal cord lesions could have influenced our ability to detect drugrelated effects, the proportion of dogs with lumbar lesions was similar amongst treatment groups. Additionally, the inclusion of lesion location in multivariable generalized linear modeling did not alter the significance or magnitude of observed treatment effects. We f
