Therapeutic Time Window is a Key Element of Drugs to Treat TBI
Despite decades of research, there are currently no treatments for TBI other than palliative care (Diaz-Arrastia et al., 2014). The reasons for the lack of therapeutics are many; drug may have failed in clinical trials since most preclinical studies dose drugs immediately or soon after experimental TBI (Diaz-Arrastia et al., 2014). This experimental design fails to take into account the well-documented clinical phenomenon of treatment gap: the time individuals wait before seeking medical care after head trauma (Tanielian and Jaycox, 2008; Demakis and Rimland, 2010). In 1991, one quarter of an estimated 1.5 million patients in America did not seek medical care after receiving a TBI that did not result in death or long-term institutionalization (Sosin et al., 1996). The multiple reasons given to postpone or avoid treatment include perceived symptom resolution, as well as the time and cost of treatment (Demakis and Rimland, 2010). Military personnel are particularly at risk for TBI. Lack of access to safe and accessible transportation for deployed military personnel can delay treatment up to 72 h after TBI (Farmer et al., 2017). Thus, treatment gap likely contributes to negative outcomes after TBI. Despite the importance of treatment gap, we know relatively little about the time course of pathophysiological events that can be successfully targeted with drugs first dosed many hours to days following TBI.
The treatment of thromboembolic stroke using tissue plasminogen activator illustrates the importance of time window in neurodegenerative diseases with a rapid onset. Thromboembolic stroke produces a complex and rapidly evolving injury with an overlapping, yet distinct, pathophysiology to TBI. Tissue plasminogen activator (t-PA) is highly effective if dosed within 4.5 h of a stroke, yet its utility drops sharply after 4.5 h due to the increased probability of hemorrhage (Ahmed et al., 2013). Despite its established ability to prevent injury, only 2–5% of stroke patients receive t-PA (Miller et al., 2011). A major reason for the limited use of t-PA is its short time window (Miller et al., 2011). The experience of clinicians with t-PA to treat stoke suggests similar difficulties will arise if drugs to treat TBI have similarly short therapeutic time windows that fall off sharply.
Since no drug has received FDA-approval, a key unanswered question is: what is a clinically relevant therapeutic window for a TBI drug? Clinical trials at designated trauma centers have enrolled patients 4–7 h after a moderate to severe TBI. Even with the high skill of the clinical staff at these trauma centers; many patients could not be enrolled because treatment could not be initiated within 4–7 h. Less specialized hospitals are likely to have even longer treatment delays. To treat the largest number of patients, a drug or drug combination will likely need to retain high efficacy when first dosed 12 h after moderate to severe TBI. In contrast to those with severe or moderate TBI, patients with mild TBI often delay seeking medical help for days after injury until their symptoms do not abate (Sosin et al., 1996; Tanielian and Jaycox, 2008; Demakis and Rimland, 2010). Thus, drugs will need to retain efficacy when dosed days after injury to treat large numbers of patients with mild TBI.
Traumatic brain injury has a complex pathophysiology whether initiated by a blunt impact, penetration through the skull into the brain, or exposure to explosive blast (Dixon, 2017). TBI produces mechanical injury within seconds to neurons, glia, and vessels. This primary injury rapidly triggers a secondary injury that evolves for weeks to months (Dixon, 2017). Both primary and secondary injury damages both gray and white matter. Within minutes after primary injury, neurons lose the ability to control ion homeostasis, which results in accumulation of intracellular calcium, cell depolarization, excitotoxic release of glutamate and additional disruptions of ionic gradients (Weber, 2012; Guerriero et al., 2015). Impaired mitochondrial function leads to energy failure; calcium accumulation and elevated reactive oxygen species are additional early events in secondary injury (Bains and Hall, 2012; Weber, 2012; Hill et al., 2017). Damage to vessels reduces cerebral blood flow resulting in hypoxia, hypoglycemia, and breakdown of the blood-brain barrier (Price et al., 2016). Inflammation rapidly follows TBI and persists for weeks to months after injury (Hinson et al., 2015). Acute inflammation is initiated by release by necrotic cells of damage associated molecular patterns (DAMPS) that activate astrocytes and microglia. Release of proinflammatory cytokines and chemokines lead to further breakdown of the blood brain barrier and recruitment of peripheral inflammatory cells. Microglia and astrocyte activation occurs rapidly in both gray and white matter; neuroinflammation may become chronic and continue to injure brain for months or years after injury. Later events in secondary injury include induction of cytogenic and vasogenic edema, increased intracerebral pressure, oxidative damage and necrotic and apoptotic cell loss (Bains and Hall, 2012; Hill et al., 2017). Early events in white matter include damage to axons, impaired transport and swelling. Damage to oligodendrocytes leads to demyelination and oligodendrocyte loss (Narayana, 2017). White matter damage evolves for weeks resulting in Wallerian axonal degeneration.
The pathophysiological events of secondary injury are highly interconnected. If dosed before, or soon after TBI, a variety of drugs with diverse modes of action (anti-oxidants, glutamate receptor antagonists, and anti-inflammatories) greatly limit the scope of secondary injury (Diaz-Arrastia et al., 2014). These drugs are effective despite targeting only one component of secondary injury. This suggests that, early after TBI, multiple pathophysiological events trigger the spread of secondary injury. Thus, early blockade of any one of these many injury mechanisms results in a substantial, long-term therapeutic effect. As secondary injury evolves, the efficacy of most drugs rapidly diminish through loss of drug targets; the intensification of secondary injury greatly diminishes the therapeutic effect of inhibiting one injury mechanism. Drugs that retain efficacy when dosed at longer intervals after injury may target pathophysiological events that initiate slowly after injury. Alternatively, drugs with good therapeutic windows have multiple targets that can still reduce secondary injury even after its intensification over time.
The importance of therapeutic time window in treating TBI is illustrated by comparing the preclinical testing of progesterone and CDP-choline with the design of Phase III clinical trials testing the same drugs. Progesterone was tested in two recent Phase III trials for TBI. The PROTECT III trial (NCT00822900) recruited patients with moderate to severe TBI (Glasgow Coma Score 4–12) within 4 h post-injury while the SYNAPSE trial (NCT01143064) recruited patients with severe TBI (Glasgow Coma Score < 7) within 7 h (Skolnick et al., 2014; Wright et al., 2014). Both trials were unable to demonstrate a therapeutic effect for progesterone. Prior to Phase III testing, numerous laboratories demonstrated a diverse set of therapeutic effects of progesterone in multiple rodent TBI models (Stein and Sayeed, 2018). Progesterone reduced glutamate release, prevented vasogenic edema, restored the blood brain barrier, improved aerobic respiration, and increased myelin and neurotrophin synthesis. Most importantly, progesterone reduced brain damage and improved multiple functional outcomes. Few drugs have had such wide preclinical testing on so many therapeutic outcomes. Virtually all of these studies, however, first dosed progesterone within 1 h or less after injury (Stein and Sayeed, 2018). Only three studies dosed progesterone between 4 and 6 h after injury and none of these studies performed a careful analysis of how the efficacy of progesterone changed after injury (Peterson et al., 2012, 2015). A first dose of progesterone 4 h after experimental TBI decreased gray matter damage, improved motor function and limited astrocyte activation (Peterson et al., 2012, 2015). A first dose at 6 h produced small improvements on expression of Nogo-A, GFAP, and GAP-43 (Liu et al., 2014). None of these studies examined multiple therapeutic time windows so it remains unknown how the efficacy of progesterone changed with increasingly longer times to first dose. A study of a first dose of progesterone 1 or 6 h post-stroke showed good retention of drug efficacy in a rat cerebral ischemia model (Yousuf et al., 2014). Little is understood, however, of how the analysis of therapeutic time window in animal models of stroke tells us whether an equivalent therapeutic window exists for TBI. The PROTECT III and SYNAPSE trials provided important information of how rapidly we could recruit and treat patients after moderate to severe TBI, however, due to the lack of appropriate preclinical testing, we do not know if progesterone retained sufficient potency to treat TBI when first dosed at 4–7 h after injury.
The Phase 3 COBRIT study tested the efficacy of CDP-choline on mild, moderate and severe TBI (Zafonte et al., 2012). Most patients (86%) received drugs within the first 24 h after injury. The COBRIT study did not show improvement in any outcome measures. Compared to progesterone, there was relatively little preclinical testing of CHP-choline. Dixon et al. showed that a first dose of CDP-choline beginning 24 h after injury produced mild improvements on beam balance and beam walk, and on acquisition of Morris water maze (Dixon et al., 1997). Two additional studies that dosed CDP-choline immediately after injury reported decreased lesion volume, increased neuroprotection, improvements in neurological tests, edema and protection of the blood brain barrier (Başkaya et al., 2000; Dempsey and Raghavendra Rao, 2003). A potential hypothesized mechanism of action of CDP-choline was to improve lipid metabolism, yet no study examined whether CDP-choline limited white matter injury. As with progesterone, there are no studies examining the efficacy of CDP-choline at different therapeutic time windows. Thus, inadequate drug potency at the time when patients were treated may have contributed to the futility of the PROTECT III, SYNAPSE, and COBRIT trials.
TBI Pathophysiology is a Major Determinant of Therapeutic Time Window
The speed of secondary injury after TBI results in the rapid appearance and disappearance of drug targets (Dixon, 2017). Studies of the therapeutic time windows of methyl-d-aspartate (NMDA) receptor agonists and antagonists illustrate how therapeutic time windows arise from the interaction of drugs with changes in TBI pathophysiological changes over time (Guerriero et al., 2015). Excessive glutamate release activates NMDA receptors within minutes after the onset of TBI (Guerriero et al., 2015). NMDA receptor activation produces calcium overload and activation of calcium-activated catabolic enzymes (Weber, 2012). If dosed soon after injury, NMDA antagonists prevent this calcium overload and prevent neuronal loss (Shohami and Biegon, 2013). The short therapeutic time window of NMDA receptor antagonists is the consequence of the speed of the calcium overload after TBI (Shohami and Biegon, 2013; Campos-Pires et al., 2015). Ongoing secondary injury subsequently produces a long-lasting downregulation of NMDA receptor expression. The loss of NMDA receptor function impairs synaptic plasticity and results in cognitive and memory deficits. The partial NMDA receptor agonist D-cycloserine when first dosed at 24 or 72 h post-injury improves Neurological Severity Score (NSS). A first dose of d-cycloserine at 24 h PI also improved performance of hippocampal-dependent tasks (Temple and Hamm, 1996; Adeleye et al., 2010; Sta Maria et al., 2017). A first dose of cycloserine at 24 h post-injury was also effective in rat model of pediatric TBI (Sta Maria et al., 2017). D-cycloserine improved performance on Novel Object Recognition and produced a mild improvement in acquisition, but not retention of Morris Water Maze (Sta Maria et al., 2017). Earlier dosing of d-cycloserine was ineffective at 8 or 16 h post-injury when NMDA receptors were downregulated (Adeleye et al., 2010). Thus, the different therapeutic time windows of NMDA receptor antagonists and agonists results from the differential consequences of NMDA receptor activation after TBI (Shohami and Biegon, 2013).
Are Studies of Therapeutic Time Window in Animal Models Relevant to Human TBI?
Animal models of TBI have been invaluable for our understanding of TBI pathophysiology (Xiong et al., 2013). Most of the secondary injury events that occur in clinical TBI also occur in animal models. This has validated the use of animal models to find drug targets to treat TBI. Virtually all studies of therapeutic time window have used rodent TBI models (Table 1 and Supplementary Table 1). Studies of therapeutic time window in rodents not only assume similar TBI pathophysiology in animals and people, but that these pathophysiological events occur with similar kinetics. Both humans and rodents rapidly develop edema, elevated extracellular glutamate, excitotoxicity and elevated intracellular Ca++2 after TBI or experimental TBI (Palmer et al., 1993; Bullock et al., 1995; Vespa et al., 1998; Markgraf et al., 2001; Hutchinson, 2005; Weber, 2012). The increase in reactive oxygen species and its accompanying oxidative damage also occurs rapidly in animals and people (Bains et al., 2013; Cornelius et al., 2013). A variety of plasma biomarkers (GFAP, UCh-1, Tau, and S100β) show similar kinetics in rodent TBI models and clinical TBI (Mondello et al., 2016; Caprelli et al., 2017; Korley et al., 2018; Shahjouei et al., 2018). In both human TBI and TBI animal models, there is an acute and rapid increase in the levels of pro-inflammatory markers (Clausen et al., 2018; Huie et al., 2018). These data suggest that studies using rodent TBI model can provide important insights into the therapeutic window of a drug to treat clinical TBI.