Perispinal Etanercept for Post-Stroke Neurological and Cognitive Dysfunction: Scientific Rationale and Current Evidence

Experimental evidence implicating TNF in stroke pathophysiology was published in 1994, and has continued through the present [80‐89]. A recent study investigated the long-term consequences of subarachnoid hemorrhage (SAH) on behavior, neuroinflammation, and damage to gray and white matter in Wistar rats through day 21 post-insult [90]. Severe SAH induced significant gray- and white-matter damage and changes in multiple cytokines, including increased expression of TNF at 48 h post-insult [90]. Neuroinflammation, including microglial activation, was “very long-lasting and still present at day 21” and accompanied by changes in sensorimotor behavior [90].

TNF was identified as a mediator of post-stroke focal ischemic brain injury 2 decades ago [80‐82, 89]. Specific inhibition of TNF, using antibodies or other recombinant TNF inhibitors, was found to reduce neurological damage from stroke, improving stroke outcomes [80‐82, 88, 89].

In 2013, inhibition of TNF using three different molecular approaches yielded favorable results in three separate animal models [42‐44]. Researchers from Duke summarized the scientific rationale and their results as follows:

Intracerebral hemorrhage is a devastating stroke subtype characterized by a prominent neuroinflammatory response. Antagonism of pro-inflammatory cytokines by specific antibodies represents a compelling therapeutic strategy to improve neurological outcome in patients after intracerebral hemorrhage … Post-injury treatment with the TNF-alpha antibody CNTO5048 resulted in less neuroinflammation and improved functional outcomes in a murine model of intracerebral hemorrhage …. TNF-alpha does not serve as a simple “biomarker” of inflammation, but rather plays a central role in mediating and extending neuronal injury after insult … Monoclonal antibodies against TNF-alpha make sense as a therapeutic strategy in intracerebral hemorrhage due to the marked neuroinflammatory effects seen in this disease [43].

Increased peri-hematomal expression of TNF has been functionally associated with neurovascular injury in multiple species and experimental models of intracerebral hemorrhage (ICH) [91‐96]. These findings are consistent with clinical reports that found elevated cerebrospinal fluid and plasma concentrations of TNF directly correlated with acute hematoma enlargement, edema development, and poor patient outcomes after ICH [97‐102]. In contrast to the early clinical success of biologic inhibitors, which directly bind TNF as a decoy receptor, small molecule inhibitors of TNF signaling pathways remain largely unexplored after ICH. TNF induces biological activity via stimulation of the TNF receptors (TNFR1 and TNFR2) [103, 104]. Post-ICH administration of R-7050, a novel cell-permeable triazoloquinoxaline compound that prevents the association of TNFR with intracellular adaptor molecules [105], reduced vasogenic edema and improved neurological outcomes in a mouse model of ICH [42]. These studies raise the possibility that small molecule inhibitors of TNF-TNFR signaling may possess therapeutic potential after ICH.

A further mechanism to not only mitigate TNF-mediated actions and signaling after ICH but also to aid in defining their roles is to inhibit TNF generation. The controversial sedative, thalidomide, has immunomodulatory actions that are mediated, in large part, by lowering the rate of TNF synthesis [106, 107]. Recent analogs that more effectively achieve this include 3,6′-dithiothalidomide (3,6′-DT) [108], which readily enters the brain [109] and suppresses TNF synthesis post-transcriptionally at the level of translational regulation via the 3′-untranslated region of its messenger RNA (mRNA) [108, 110] as well as through down-regulation of the eukaryotic elongation initiation factor (eIF)-4E [111] to allow its rapid degradation.

In a mouse model of focal ischemic stroke in which brain TNF levels were found to be rapidly elevated within both ipsi- and contralateral brain, 3,6′-DT fully ameliorated this rise and reduced infarct volume, neuronal death, and neurological deficits [112]. This neuroprotection was accompanied by reduced inflammation, with 3,6′-DT lowering the expression of interleukin (IL)-1Beta and inducible nitric oxide synthase, reducing activated microglia/macrophages, astrocyte, and neutrophil numbers, and decreasing the expression of intercellular adhesion molecule (ICAM)-1 within ischemic brain tissue [112]. TNF plays a role in the induction of ICAM-1 expression and also promotes BBB leakage by inducing the expression of matrix metalloproteinase (MMP)-9 [113, 114], which degrades BBB tight junction proteins [115, 116]. Mitigating the rise in TNF by 3,6′-DT treatment suppressed the known TNF-induced activation of MMP-9 [117] and, thereby, decreased stroke-induced BBB disruption by preserving junction proteins [112]. In support of a major role of TNF in processes mediating stroke as well as TNF inhibition as the primary mechanism for the neuroprotective action of 3,6′-DT, the ability of 3,6′-DT to decrease ischemic brain damage was abolished in mice lacking TNF receptors [112].

The mechanisms underlying the detrimental effects of TNF signaling after ICH remain poorly defined and could provide additional therapeutic targets upon elucidation. Emerging data suggest that TNF induces necroptosis, a novel form of cell death with characteristic features of apoptosis, necrosis, and type 2 autophagic death [118‐121]. In an experimental model, hemorrhagic injury increased TNF expression and promoted necroptotic cell death in cultured glial cells [122]. This effect was reversed by inhibition of receptor-interacting serine/threonine-protein kinase (RIPK)-1, a multi-functional protein kinase that interacts with TNFR to activate the pro-inflammatory transcription factor, nuclear factor (NF)-κB [123‐125]. In line with this finding, it was observed that necrostatin-1, a pharmacological inhibitor of RIPK [124, 125], similarly limited neurovascular injury and improved outcomes in a pre-clinical model of ICH [126]. This finding is also consistent with reports showing necrostatin-1 is neuroprotective in experimental models of ischemic stroke and TBI [125, 127, 128]. Taken together, these experimental results support the assertion that TNF induces detrimental effects after neurological injury and suggests that directed targeting of TNF and downstream signaling pathways may improve patient outcomes.

Additional research involving multiple animal models of stroke and TBI provides documentation of a favorable therapeutic response to TNF inhibition [42‐45, 60, 61, 81, 82, 86, 129]. As an example, brain TNF levels were found to have elevated rapidly (within 1 h) following concussive (weight drop-induced) mild TBI in mice, and were maximal at 12 h [109]. Inhibition of this TBI-induced rise by administration of a single dose of the TNF synthesis inhibitor 3,6′-DT fully ameliorated cognitive impairments evaluated both 7 and 30 days later; supporting both a role for TNF in TBI-induced neuroinflammation/cognitive impairment and its targeting for treatment [109]. Most recently, inhibition of phosphoinositide 3-kinase delta, a molecule that controls intracellular TNF trafficking in macrophages, was shown to reduce TNF secretion and neuroinflammation and confer protection in a mouse cerebral stroke model [130].

In 1988, researchers used autoradiography to investigate the effects of cerebral infarction induced by unilateral middle cerebral artery occlusion in rats. The radiolabeled ligand PK11195 that binds primarily to activated microglia was used. Seven days after stroke, [³H]PK11195 bound significantly in the cortical and striatal regions surrounding the focus of cerebral infarction with smaller increases in the ventrolateral and posterior thalamic complexes and in the substantia nigra, all ipsilateral to the occlusion [131].

In 1991, increased [³H]PK11195 binding in the thalamus during the second week after experimentally induced stroke in rats was found using ex vivo autoradiography, at a time when [³H]PK11195 binding around the primary lesion was beginning to subside [132].

In 2000, a multi-national European academic collaboration of neurologists, neuroscientists, and nuclear medicine specialists demonstrated that brain inflammation may persist for months or years after stroke in humans [133]. The physicians and scientists investigated the potential of positron emission tomography (PET) using [¹¹C]PK11195 to assess the microglial reaction in secondary thalamic lesions in patients with infarcts in the territory of the middle cerebral artery. All patients studied were found to have increased [¹¹C]PK11195 binding in the ipsilateral thalamus, indicating microglial activation in projection areas remote from the primary lesion [133]. The only patient studied more than 7 months after stroke was a 50-year-old patient with a primary stroke involving the left temporo-parietal region, and he demonstrated bilateral thalamic microglial activation 24 months after stroke [133].

In 2004, an international collaboration of neuroscientists found pathological evidence of a long-term intracerebral inflammatory response after TBI in a series of patients who had sustained blunt head injury. They described microglial hyperplasia and hypertrophy with major histocompatibility complex (MHC) class II upregulation, and inflammatory changes up to 16 years after the injury [134].

In 2005, Gerhard et al. [135], in another international collaboration of academic scientists and physicians, studied a series of patients between 3 and 150 days after onset of ischemic stroke in order to measure the time course of microglial activation. Utilizing (R)-[¹¹C]-PK11195 PET, they found that brain inflammation was long-lasting after stroke, with (R)-[¹¹C]-PK11195 binding involving both the area of the primary lesion and areas distant from the primary lesion site [135]. They described the spread of the glial response beyond the ischemic core as closely resembling the progression of microglial activation in animal experiments, with “early recruitment of microglia in the ischemic border zone and later involvement of the neocortex and thalamus” [135].

In 2006, Price et al. [136], in a multi-center academic collaboration, used (R)-[¹¹C]-PK11195 imaging to study a series of patients after stroke. Using this imaging methodology, they documented persistent neuroinflammation in the stroke penumbra and elsewhere in the brain in patients following stroke, and recognized that this neuroinflammatory response might represent a therapeutic opportunity that extends beyond time windows traditionally reserved for neuroprotection [136].

In 2011, Folkersma et al. [137], studied microglial activation in patients with moderate and severe TBI using (R)-[¹¹C]-PK11195 brain PET, 6 months after trauma. In both whole-brain and regional analysis, increased (R)-[¹¹C]-PK11195 binding potential was found compared with age- and sex-matched healthy controls. From these series, increased (R)-[¹¹C]-PK11195 binding potential was found not only in the ipsilateral but also in the contralateral hemisphere, indicating prolonged and widespread microglia activation after TBI.

Subsequent studies, using either PET imaging or pathologic examination, have confirmed the existence of chronic intracerebral glial activation that has been documented to last for 17 years after even a single acute brain injury [138, 139].

Microglial imaging using (R)-[¹¹C]-PK11195 brain PET can be of meaningful clinical and diagnostic value in terms of visualization and quantification of active neuroinflammatory and neurodegenerative disease processes and in elucidation of the long-term effects of neuroinflammatory sequelae and its implications for neurological outcome [137]. Taken together, along with additional research showing that pathologic TNF mediates neurotoxicity in the ischemic penumbra, these data suggest that chronic microglial activation and neuroinflammation may be a common pathological response to stroke and other forms of acute brain injury [86, 133‐140].

There is a need to understand the long-term relationship between late microgliosis and TNF. Although the PET data discussed in this section do not describe TNF actions or changes, PET imaging before and after therapeutic intervention with TNF inhibitors that can quantify and describe patterns of microglial activation promises to be a fertile area for future investigation. As suggested by Price et al. [136], the accumulating evidence indicates that chronic glial activation after acute brain injury represents a therapeutic target that persists far longer than the time windows traditionally reserved for neuroprotection. This evidence provides a scientific basis for considering pharmacologic therapeutic intervention that targets chronic glial activation months or years after stroke, and supports the plausibility of achieving a therapeutic response in patients with chronic post-stroke neurological dysfunction by targeting pathologic TNF concentration [86, 133‐139].

In an in vitro model of microglial activation and propagated neuron killing in the stroke penumbra, TNF inhibition using a soluble TNF receptor reduced neurotoxicity [86]. In addition, experimental data suggest that TNF functions as a gliotransmitter that is involved in the mechanisms whereby glia modulate synaptic transmission and neuronal network function [141‐155].

The plausibility of beneficial effects of etanercept for treatment of chronic post-stroke neurological dysfunction is supported by the fact that, in addition to its known role as a potent biologic inhibitor of TNF, etanercept has also been shown to be capable of reducing glial activation in multiple experimental models [45, 64‐67]. The known physiological effects of etanercept on TNF and glial activation make it a well matched candidate to address the chronic glial activation and pathologic TNF that may be a long-lasting consequence of stroke [45, 64‐67, 86, 133, 135, 136].

Statistically significant improvements in pain, including improvements in hyperesthesia, allodynia, pain associated with spasticity, post-stroke shoulder pain, and neuropathic pain were reported in the 617-patient stroke cohort [5]. These results are supported by a long series of experiments documenting the effects of etanercept and other biologic TNF inhibitors in experimental models and in the clinic.

In 1998 and thereafter, Sommer and colleagues [156‐158], in a series of basic science experiments, demonstrated the central involvement of TNF in the pathophysiology of neuropathic pain and the favorable effects of anti-TNF antibody treatment in these models. In 1999, a separate group of investigators [159] showed that neuropathic pain was mediated by brain-derived TNF. Subsequent studies provided further supportive evidence [159‐165]. In 2001, etanercept was shown to reduce hyperalgesia in experimental painful neuropathy [166]. In 2003 and 2004, the first human evidence of the effectiveness of etanercept for treating neurological spinal pain was published [167‐170]. Many of these early studies were performed by the authors and their colleagues. Subsequently, four randomized controlled clinical trials have provided favorable data supporting the efficacy of etanercept for neurological spinal pain, and TNF inhibition is emerging as a treatment strategy for intractable sciatica and other forms of intervertebral disc-related pain [171‐176]. The accumulated evidence is substantial [65, 156‐175, 177‐179]. This evidence, taken together, suggests by analogy the plausibility of pain improvement following etanercept in patients with chronic post-stroke pain.

Statistically significant reduction in cognitive impairment is reported in the 617-patient stroke cohort following perispinal etanercept treatment [5]. The data included improvement in a standardized instrument, the Montreal Cognitive Assessment, with p-values less than 0.0001 immediately post-treatment and 1 and 3 weeks later [5]. The cognitive improvement documented in the etanercept stroke studies is supported, by analogy, by substantial scientific evidence that suggests that TNF is centrally involved in the pathophysiology of chronic brain dysfunction.

Beginning in the 1980s, and continuing into the present, TNF has been implicated in the pathophysiology of multiple diseases and disorders associated with chronic brain dysfunction, including cerebral malaria [32, 51, 56, 180‐185]; TBI [45, 52, 60, 61, 129, 186, 187]; Alzheimer’s disease [32, 34, 53, 55, 59, 149, 188‐203], frontotemporal dementia [54]; primary progressive aphasia [62, 204]; sarcoidosis [205]; rheumatoid arthritis [206]; surgery-induced cognitive decline [57]; and a wide variety of additional diseases and disorders [32, 58, 67, 207]. For example, an increasing body of evidence supports a major role for central neuroinflammatory mechanisms in the pathogenesis of hepatic encephalopathy, a neuropsychiatric complication of both acute and chronic liver failure. Microglial activation in liver failure has been attributed to the accumulation of lactate in the brain, and focal accumulation of brain lactate is a common feature of stroke, TBI, and status epilepticus, conditions that are known to result in significant neuroinflammation [67, 208]. Neuroinflammation characterized by microglial activation and increased expression of pro-inflammatory cytokines in the brain has been reported in both human and experimental liver failure of diverse etiology, including viral hepatitis [208] and biliary cirrhosis [209], as well as in acute liver failure resulting from toxic [210] or ischemic [211] liver injuries. Microglial activation and increased pro-inflammatory cytokine expression are significantly correlated with the grade of encephalopathy in these disorders. Moreover, slowing of hepatic encephalopathy progression has been demonstrated following inhibition of microglial activation by hypothermia [211] or minocycline [212] and following the use of anti-TNF strategies such as etanercept [213]. TNFR gene deletion delays the progression of hepatic encephalopathy in mice with acute liver failure resulting from toxic liver injury [210].

Additionally, it is notable that among the 69 patients who participated in the early phase I studies of prolonged (24-h or 5-day) intravenous infusions of recombinant human TNF, three developed transient focal neurological symptoms. One patient developed diplopia, lethargy, and expressive dysphasia after receiving recombinant TNF at 2.0 × 10⁵ U/m²/d for 2 days, with return to baseline neurologic status within 48 h without sequelae [214]. The second study, involving a 24-h infusion of human recombinant TNF documented two cases of neurological toxicity, as follows:

Two elderly patients had transient episodes of focal neurological deficits. One patient had an isolated loss of recent memory, while the other had transient expressive aphasia. No abnormalities were noted upon computerized tomography brain scan or cerebrospinal fluid analysis. In each case, the symptoms occurred near the completion of treatment and resolved without sequelae within 6 h. These two toxic events occurred at doses of 182 and 327 μg/m² and did not represent dose-limiting toxicity [215].

These early cases of focal neurological toxicity following TNF infusion provide further scientific support for the involvement of excess TNF in the pathophysiology of post-stroke neurological dysfunction and the perispinal etanercept results.

Etanercept has demonstrated favorable effects in neuroinflammatory disorders, both in the clinic and in multiple experimental models [4, 5, 10, 35, 45, 58, 60, 61, 64‐69, 146, 166‐173, 177‐179, 204, 207, 216‐221].

TNF levels in the cerebrospinal fluid 25 times higher than in controls have been found in patients with Alzheimer’s disease [222]. In patients with mild cognitive impairment (MCI) followed prospectively, “only MCI patients who progressed to Alzheimer’s disease at follow up, showed significantly higher CSF levels of TNF-alpha than controls … Indicating that CNS inflammation is a early hallmark in the pathogenesis of AD” [223]. A later study from these investigators supported this conclusion regarding the role of TNF in Alzheimer’s disease pathogenesis [224].

In 2006, the clinical results of a prospective, single-center, open-label, pilot clinical trial of perispinal etanercept for Alzheimer’s disease was reported by the senior author and colleagues [216]. The authors included two neurologists, a rheumatologist, and an internist, and the study included 15 patients treated with perispinal etanercept weekly over a period of 6 months [216]. The main outcome measures included three standard instruments for measuring cognition: the Mini-Mental State Examination (MMSE), the Alzheimer’s Disease Assessment Scale-Cognitive subscale (ADAS-Cog), and the Severe Impairment Battery (SIB). There was significant improvement with treatment, as measured by all of the primary efficacy variables, through 6 months: MMSE increased by 2.13 ± 2.23 (p < 0.003), ADAS-Cog improved (decreased) by 5.48 ± 5.08 (p < 0.006), and SIB increased by 16.6 ± 14.52 (p < 0.04).

In 2008, rapid cognitive improvement in a patient with Alzheimer’s disease following treatment with perispinal etanercept was reported by the senior author and a neurologist [218]. Sue Griffin, co-editor of the Journal of Neuroinflammation, reported her independent observations after witnessing rapid clinical improvement in additional patients with Alzheimer’s disease following treatment with perispinal etanercept [217]. Subsequent publications by the senior author and colleagues documented cognitive improvement in patients with Alzheimer’s disease and other forms of dementia following treatment with perispinal etanercept [10, 68, 69, 146, 204, 219].

In a basic science study conducted by the senior author and Stanford scientists and published in 2009, perispinal administration of radiolabeled etanercept followed by head-down positioning was discovered to deliver radiolabeled etanercept into the choroid plexus and cerebrospinal fluid within the cerebral ventricles within minutes of injection, as visualized by PET scan [31].

In 2010, Chio et al. [45] studied etanercept in an experimental model of TBI. They found that etanercept caused attenuation of TBI-induced cerebral ischemia, reduction of motor and cognitive function deficits, and reduction of microglial activation [45].

Chen et al. [206] studied the effects of anti-TNF treatment on cognition in 15 patients with rheumatoid arthritis over a period of 6 months with subcutaneous anti-TNF treatment: eight received etanercept 25 mg twice weekly and seven received adalimumab 40 mg twice monthly. Cognitive function determined by MMSE scores was significantly improved in the patient cohort [206].

Elfferich et al. [205] studied 343 sarcoidosis patients over a period of 6 months, with all patients completing the Cognitive Failure Questionnaire (CFQ) at baseline and at 6 months [206]. Patients were separated into three groups: (1) no immunomodulating drugs; (2) prednisone with or without methotrexate; and (3) anti-TNF drugs. Only patients receiving anti-TNF drugs demonstrated a significant improvement in CFQ score [205].

Chou et al. [225] presented the results of their review of medical and pharmacy claims data from January 2000 to November 2007 for a commercially insured cohort of 8.5 million adults throughout the USA. They derived a sub-cohort of 42,193 patients with a pre-existing diagnosis of rheumatoid arthritis. In this population of adults with rheumatoid arthritis, they found a 55 % decreased incidence in Alzheimer’s in those patients treated with TNF inhibitors, but not with other disease-modifying agents used for treatment of rheumatoid arthritis [225]. When they further analyzed the risk according to the individual anti-TNF agent used, they found that only etanercept was significantly (p = 0.024) associated with reduced risk [226].

In 2011, Shi et al. [195, 196] reported cognitive improvement in a woman with Alzheimer’s disease following intrathecal administration of infliximab, a chimeric TNF monoclonal antibody, following the favorable results of the use of infliximab in an experimental Alzheimer’s model [195, 196].

In 2012, Gabbita et al. [227] found that early intervention with a small molecule inhibitor of TNF prevented cognitive deficits and improved the ratio of resting to reactive microglia in the hippocampus in a murine triple transgenic model of Alzheimer’s disease. Belarbi et al. [228] found that a TNF protein synthesis inhibitor restored neuronal function and reversed cognitive deficits induced by chronic neuroinflammation. McNaull et al. [197, 229] and Butchart and Holmes [197, 229] discussed the rationale for TNF inhibition as a treatment approach for Alzheimer’s disease in their review articles [197, 229].

Bassi and De Filippi [207] reported verbal, cognitive, and behavioral improvement in a patient with long-standing neurological dysfunction, in whom etanercept was used for treatment of psoriasis. The beneficial effect on cognition and social interaction was a surprising side effect of etanercept used to treat the cutaneous psoriasis [207].

In 2013, Cheong et al. [197, 229] studied etanercept in an experimental model of TBI. They found that neurological and motor deficits, cerebral contusion, and increased brain TNF-alpha contents caused by TBI were attenuated by etanercept [60].

In 2014, Detrait et al. [197, 229] reported favorable effects of etanercept administered systemically in a basic science Alzheimer’s model [230]. However, the only dose that was effective across all measures of efficacy was the highest dose, 30 mg/kg given every 2 days (for a total dose of 90 mg/kg given during the first week). This 90-mg/kg weekly dose is more than 100 times the normal human etanercept dose. Etanercept doses of 3 mg/kg every 2 days, about 15 times the usual human dose, were not effective. The lack of efficacy of systemically administered etanercept in this Alzheimer’s disease model at doses closer to the usual human therapeutic dose is consistent with a previous Alzheimer’s disease clinical trial in which etanercept administered systemically at a dose of 25 mg twice weekly was not found to be effective [231].

The totality of this evidence suggests, by analogy, the plausibility of cognitive improvement following perispinal administration of etanercept in patients with chronic post-stroke cognitive impairment.

Finally, rapid neurological improvement following perispinal etanercept has been witnessed first-hand by independent third parties, including several of the authors of this commentary as well as others [11, 35, 216, 217, 232]. A new report has documented that a single dose of perispinal etanercept produced an immediate, profound, and sustained improvement in expressive aphasia, speech apraxia, cognitive dysfunction, and left hemiparesis in a patient with chronic, intractable, debilitating neurological dysfunction present for more than 3 years after acute brain injury [11]. Replication of experimental results with validation by different observers is a time-honored cardinal scientific principle supporting the reliability of a scientific observation [39].