Clinical Trials in Traumatic Brain Injury: Past Experience and Current Developments

Introduction 

Randomized controlled trials (RCT) are considered the gold standard for proving the efficacy of new treatments. In the field of traumatic brain injury (TBI), however, not a single multicenter phase III RCT on neuroprotective agents has convincingly shown benefit.1, 2, 3 As a consequence, a sense of disappointment prevails; basic science researchers are increasingly frustrated that their hard and promising work performed under clean experimental conditions does not seem to transfer into the “dirty” clinical situation, and there is some reluctance by pharmaceutical companies to embark on a new high-cost venture in TBI. However, trials conducted so far have not definitively established the lack of possible benefit. To the contrary, various trials have shown a trend toward efficacy, and this offers hope. Basic research has greatly advanced our knowledge of what happens in the brain after TBI, offering opportunities to limit processes involved in the development of secondary brain damage. Many experimental studies on a multitude of agents have shown encouraging results, demonstrating efficacy of different agents on histologic endpoints, intracranial pressure, brain edema, cerebral blood flow, metabolism, and also on functional outcome. Translating advances from basic research into clinical benefit has proven complex.

In the experimental situation, studies are performed under tightly controlled conditions. TBI in the clinical situation, however, includes a complex spectrum of pathologies, often superimposed with systemic second insults (e.g., hypoxia and/or hypotension). Approaches to treatment may vary, and uncertainty exists as to whether pathophysiologic processes targeted are indeed active in individual patients and, if so, at what time after injury. It is here that more basic and clinical research is needed before targeting therapy to the individual needs of a patient can become a clinical reality. TBI populations are also heterogeneous in terms of clinical severity and baseline prognostic risk. This heterogeneity poses complex methodological challenges. It is here that methodological approaches may be optimized to increase statistical power.

The challenge to demonstrate benefit of a novel agent or treatment strategy in TBI is great, but the rewards will be correspondingly high.

TBI is a field with one of the greatest unmet needs in medicine and public health.4 Not only is it a major cause of death and disability, incurring great personal suffering to victims and relatives, but it also leads to huge direct and indirect costs to society. In the United States, the annual burden of TBI has been estimated at over $60 billion.5 Globally, the incidence is increasing, and TBI resulting from blast injuries is being increasingly recognized in military personnel returning from the Middle East conflicts. In the Western world, the epidemiology of TBI is slowly changing with a relative increase in middle-aged and elderly patients injured by falls, resulting typically in contusional brain damage.4 Contusional brain injury is different from diffuse axonal injury, with a much larger component of inflammatory response, and this is often characterized by lesion progression. Approximately 25 to 45% of cerebral contusions will enlarge significantly,6, 7 and even higher occurrences are reported if the initial computed tomographic scan is performed within 2 h of injury.8 The more frequent use of anticoagulant medication and platelet aggregation inhibitors in older patients may further increase the risk of lesion progression. The frequent progression of contusive brain injury indicates that this may constitute a subpopulation of TBI more likely to benefit from neuroprotective agents by limiting processes involved in secondary brain damage. Other mechanisms, and consequently different approaches may be more relevant in patients with diffuse axonal injury, and we emphasize that neuroprotection in a more broad sense also includes strategies and therapies aimed at promoting regeneration or replacement of lost neuronal and glial cells, neuronal circuits, and stimulation of neuroplasticity.

The aim of this article is to review past and current experiences in clinical trials of TBIs, to discuss limitations and challenges that have been recognized, and to summarize current directions. The focus is on studies of patients with severe and moderate TBIs.

Clinical Trials in TBI: Past Experience 

The history of multicenter, randomized controlled trials of TBI really started in the mid-1980s. Major pillars of clinical research, such as the Glasgow coma scale, the Glasgow outcome scale and computed tomographic scanning were only introduced in the 1970s. A further boost toward trials resulted from advances in basic research with the identification of various neuroprotective agents. Prior to 1980, the majority of trials that were reported consisted of single-center studies with evaluation of treatment results that often used historical controls. In this period, a particular interest focused on the efficacy of steroids. These trials were primarily initiated by investigators from scientific interest; in contrast, multicenter trials on neuroprotective agents from the mid-1980s on were largely initiated by pharmaceutical companies. The collaboration between investigators and pharmaceutical companies proved beneficial to both sides; investigators contributed disease-specific expertise, and the participation in trials greatly facilitated international contact and collaboration. Indirectly this has probably led to improved quality of care.

We conducted a systematic literature search of TBI clinical trials for the period of 1980 to 2009. We included studies meeting the following criteria:

We excluded the following criteria:

We selected studies focusing on adults (>15 years) and studies that included at least 100 patients (minimum, 50 per arm). In total, 27 studies met these criteria. Collectively, we were aware of another 6 unpublished, large phase III TBI trials from which we had sufficient knowledge to present details. Therefore, these studies were added to the selection, resulting in a total of 33 studies. A complete overview of the studies is presented in Table 1. This overview permits a number of interesting observations as follows.

Table 1. Overview

	Publication (Funding)	Agent/Intervention (Mechanism)	Centers	Study Population	No.	Year of Study	Status	Results
	Braakman et al.,26 1983 (inv. initiated)	High dose dexamethasone (various processes)	2	Comatose patients after nonmissile TBI	161	1978–1981	Completed	No sign. Tx. effect
	Dearden et al.,27 1986	Dexamethasone (various processes)	1	Severe head injury	130	1980–1983	Completed	No sign. Tx. effect
	Grumme et al.,20 1995 (Inv. initiated)	Triamcinolone (various processes)	9	Severe head injury, not further defined	396	1985–1990	Completed	No sign. Tx. effect
	Bailey et al.,18 1991 (Bayer - HIT I)	Nimodipine (Ca- mediated damage)	6	Not obeying commands	351	1987–1989	Completed	No sign. Tx. effect
	Eur study group,19 1994 (Bayer - HIT II)	Nimodipine (Ca- mediated damage)	21	Not obeying commands	852	1989–1991	Completed	No significant effect in overall population
	Rockswold et al.,28 1992 (Inv. initiated)	Hyperbaric oxygen (cerebral ischemia)	1	GCS ≤9	168	1983–1989	Completed	Reduced mortality
	Wolf et al.,29 1993 (NIH:12587)	Tromethamine (THAM) (cerebral acidosis)	2	GCS ≤8	149	1988–1989	Completed	No overall treatment effect
	Gaab et al.,30 1994 (inv. initiated)	Dexamethasone (various processes)	10	GCS ≤13	300	1986–1989	Completed	No sign. Tx. effect
	Unpublished: tiriliazad-domestic (Upjohn)	Tirilazad (lipid peroxidation)	36	GCS ≤8: 72% GCS 9–12: 28%	1155	1991–1994	Terminated	No sign. Tx. effect reported
	Marshall et al.,31 1998 Tirilazad-International (Upjohn)	Tirilazad (lipid peroxidation)	50	GCS ≤8: 85% GCS 9–12: 15%	1120	1992–1994	Completed	No sign. Tx. effect
	Young et al.,21 1996 (Sanofi-Winthrop)	PEGSOD (free radical damage)	29	GCS ≤8	1562	1993–1995	Completed	No sign. Tx. effect
	Unplublished (SyntheLabo)	Eliprodil (glutamate exitotoxicity)	20+	GCS 4–8	452	1993–1995	Completed	No sign. Tx. effect
	Harders et al.,9 1996 (Bayer - HIT III)	Nimodipine (Ca- mediated damage)	21	tSAH	123	1994	Completed	Significant reduction in unfavorable outcome
	Robertson et al.,32 1999 (Inv. initiated, NIH NS27616)	CBF vs. ICP directed management (cerebral ischemia)	1	Motor score ≤5	189	1994–1997	Completed	No difference in neurologic outcome. Decrease in episodes of jugular desaturation
	Morris et al.,33 1999 (Ciba-Geigy, Novartis)	Selfotel (glutamate exitotoxicity)	95	GCS 4–8	693	1994–1996	Terminated	No sign. Tx. effect
	Clifton et al.,16 2001 (Inv. initiated, NIH NS 32786)	Hypothermia - NABIS (various processes)	11	GCS 3–8 Motor score 1–5	392	1994–1998	Halted	No effects on outcome. Reduced incidence of ICP >30
	Unpublished (Sandoz, Novartis)	D-CPP-ene - Saphir (glutamate exitotoxicity)	51	Not obeying commands, ≥ one reactive pupil	924	1995–1997	Completed	No sign. Tx. effect
	Marmarou et al.,22 1999 (SmithKlineBeecham/Cortech Inc.)	Bradycor/CP-0127(bradykinine antagonist)	31	GCS 3–8	139	1996	Terminated	12% improvement in favorable outcome (p = 0.26)
	Unpublished (Cambridge Neuroscience)	Cerestat/aptiganel (glutamate exitotoxicity)	38	GCS 4–8 GCS 3 if pupils reactive	532	1996–1997	Terminated	No sign. Tx. effect
	Unpublished (Parke Davis)	SNX-111(Glutamate exitotoxicity)	? (multi-center)	GCS 4–8	237	1997–1998	Terminated	Higher mortality
	Unpublished (Bayer HIT IV)	Nimodipine (Ca- mediated damage)	36	GCS <15 + tSAH	592	1997–1999	Completed	No significant effect
	Yurkewicz et al.,23 2005 (Pfizer)	Traxoprodil (CP-101606) (Ca- channel blocker)	? (multi-center)	GCS 4–8	404	1998–2001	Completed	Higher mortality, nonsignificant effect
	Cooper et al.,34 2004 (inv. initiated, MRC-Aus: 124330)	Hypertonic saline (hypovolemia)	12	GCS ≤8	229	1998–2002	Completed	No sign. Tx. effect
	Temkin et al.,35 2007 (inv. initiated, NIH: NS 19643)	Magnesium sulfate (multiple mechanisms)	1	GCS ≤ 12	499	1998–2004	Completed	Poorer outcome in treated group
	Maas et al.,17 2006 (Pharmos Corp.)	Dexanabinol (multiple processes)	86	Motor score 2–5 + CT abnormalities	861	2000–2004	Completed	No sign. Tx. effect
	Edwards et al.,11 2005 (MRC UK)	Methylprednisolone (multiple mechanisms)	239	GCS ≤ 14	10008	1999–2004	Terminated	Higher mortality in Tx. group (p = 0.001)
	Cruz et al.,36⁎ 2001 (Investigator initiated)	High-dose mannitol (raised ICP)	1	ASDH	178	1997–2000	Completed	Significant better outcome (p < 0.01)
	Cruz et al.,37⁎ 2002 (Investigator initiated)	High-dose mannitol (raised ICP)	1	Temporal lobe hemorrhage with abnormal pupils	141	1997–2001	Completed	Significant treatment effect
	Zhi et al.,12 2003 (Investigator initiated)	Mild hypothermia (various processes)	1	GCS ≤ 8	396	1997–2001	Completed	Reduction of mortality and improved outcome
	Lu et al.,13 2003 (Investigator initiated)	Decompr. craniectomy (raised ICP)	1	GCS ≤ 8	230	1998–2001	Completed	Significant reduction of mortality
	Jiang et al.,14 2005 (Investigator initiated)	Standard trauma craniectomy vs. limited craniectomy (raised ICP)	5	GCS ≤ 8 + refractory intracranial hypertension	468	1998–2001	Completed	Better outcome with large craniectomy
	Jiang et al.,15 2006 (Investigator initiated)	Long-term mild hypothermia (multiple mechanisms)	3	GCS ≤ 8	215	2000–2003	Completed	5-day mild hypothermia is more efficacious than 2-day short term
	Xiao et al.,10 2008 (Investigator initiated)	Progesterone (multiple mechanisms)	1	GCS ≤ 8	159	2004–2007	Completed	More favorable outcome in Tx. group (p = 0.048)

ASDH = acute subdural hematoma; CBF = cerebral blood flow; CT = computed tomography; GCS = Glasgow coma score; HIT = Head Injury Trial; ICP = intracranial pressure; NABIS = National Acute Brain Injury Study; PEGSOD = polyethylene glycol-conjugated bovine superoxide dismutase; Tx. = treatment.

⁎ The studies reported by Cruz et al.36, 37, 39 have been subjected to severe criticism, and the reliability and validity of the results have been questioned.

First, there is a peak incidence of trial initiations in the mid-1990s, with a sharp decline in the period from 2000 to 2004. This holds true for both studies with neuroprotective agents and studies using therapeutic strategies (e.g., hypothermia or decompressive craniectomy). We recognize that studies initiated in the latter years may not all have been published yet, and we are aware of four phase III studies on therapeutic strategies (two on decompressive craniectomy, one on hypothermia, and one on the antifibrinolytic agent tranexamic acid [Clinical Randomization of an Antifibrinolytic in Significant Hemorrhage (CRASH 2)]), but of only one trial with a neuroprotective agent (recombinant human erythropoietin [rhEPO]) that is currently enrolling or was recently completed (Table 2). FIG. 1 summarizes the number of phase III RCTs initiated per 5-year period in moderate and severe TBIs since 1980. The decline in studies on neuroprotective agents does not seem to be caused by a lack of new agents with great potential, but it is more likely due to the difficulties experienced in previous trials combined with the high costs.

Table 2. Recently Completed, Ongoing, and Expected Studies

	Funding Sponsoring	Study + Agent	Mechanism Targeted	Type Study	Study Population	Target No.	Start Year	Status
	Investigator initiated (Zhejiang and Hangzhou Health Department, China)	Decompressive craniectomy	Raised ICP	Phase II	GCS ≤8	80	2003	Completed Dec 2008
	NIH-NINDS	Cyclosporin A	Mitochondrial dysfunction	Phase II safety	GCS 3–8	50	2003	Completed (Mazzeo et al.38 2009)
	Biotherapeutics	Oxycyte (oxygen carrier)	Increase cerebral oxygenation	Phase II	GCS ≤9	8	2005	Completed Jan 2008
	Key Neurotek A.G.	KN 38-7271	Cannabinoid receptor agonist	Phase II a	Severe TBI	97	2006	Completed Dec 2008
	Investigator initiated (Department of Health, Taipei City Government, Taiwan)	Multiple cerebral monitoring	Multiple	Phase II b	GCS 3–8	320	2006	Ongoing
	Xytis Pharmaceuticals	XY 2405	Bradykinine (beta 2 receptor) antagonist	Phase II	GCS ≤12	400	2007	Stopped early June 2008
	Investigator initiated (Cambridge University Hospitals NHS Foundation Trust, UK)	Interleukin-1 receptor antagonist	Inflammatory response	Phase II safety	GCS ≤8	26	2008	Ongoing
	Biotherapeutics	Oxycyte (oxygen carrier)	Cerebral ischaemia	Phase II safety	Severe TBI	128	July 2009	Ongoing
	Solvay Pharmaceuticals	SLV334 (ECE/NEP inhibitor) preventing formation of the vasoconstrictor ET-1	Cerebral vasoconstriction; atrial natriuretic peptide	Phase II	Moderate and severe TBI	72	2009	Ongoing
	Neuren Pharmaceuticals	NNZ 2566	Multiple	Phase II dose escalation	?	?	2009?	Expected to start October
	Vasopharm	VAS 203	Nitric oxyde synthesis	Phase II	Severe	?	?	Expected
	Investigator initiated (MRC Australia)	Decompressive craniectomy (DECRA)	Raised ICP	Phase III	Severe TBI: ICP >20 mmHg for 15 min in 1 h refractory to med. measures	165	2004	Ongoing
	Investigator intiated (NIH-NINDS)	NABIS: H IIR Hypothermia	Multiple	Phase III	GCS 3–8	240	2005	Stopped
	Investigator initiated (MRC UK)	RescueICP (decompressive craniectomy)	Raised ICP	Phase III	ICP >25 mmHg for 1 h refractory to med. measures	500	2005	Ongoing
	NIH-NINDS	Rrecombinant Human Erythropoetin (rhEPO)	Multiple	Phase III	GCS motor ≤5	200	2006	Ongoing
	Investigator initiated	Antifibrinolytic therapy with tranexamic acid (CRASH-2)	Fibrinolysis	Phase III	All adult TBI patients at risk for developing intracranial hemorrhage	20,000	2007	Ongoing
	European Society Intensive Care Medicine (ESICM)	Eurotherm hypothermia	Multiple	Phase III	Severe	1800	2009	Expected
	Investigator initiated	Surgical trial in traumatic intracerebral hemorrhage (STITCH)	Contusions with mass effect	Phase III	Traumatic supratentorial ICH or contusion >25 mL	840	2010	Expected
	BHR Pharma	BHR100 (progesterone)	Multiple	Phase III	GCS 4–8	1180	2010	Expected
	NIH-NINDS	ProTECT (progesterone)	Multiple	Phase III	GCS 4–12	1140	2010	Expected

ICP = intracranial pressure; NABIS = National Acute Brain Injury Study; TBI = traumatic brain injury.

View Large Image Download to PowerPoint

FIG. 1. Numbers of initiated randomized controlled trials on moderate to severe traumatic brain injury per 5-year time periods. Trials were grouped by studies on neuroprotective agents and studies on therapeutic strategies.

Second, most studies reported a significant treatment effect concerning those using a therapeutic strategy as an investigational modality. Seven of 10 studies on therapeutic strategies demonstrated statistical significance versus only 3 of 23 on neuroprotective agents (Table 3). The three studies on neuroprotective agents that show a significant treatment effect are the Head Injury Trial (HIT) III nimodipine study,9 the study reported by Xiao et al10 on progesterone (both of which show beneficial effects), and the CRASH study,11 which demonstrated higher mortality in the steroid-treated patients. These three trials had features that distinguished them from the traditional design of TBI trials. The HIT III study targeted a subgroup of the overall population, which only included patients with traumatic subarachnoid hemorrhage. The progesterone study reported by Xiao et al.10 concerned a single-center study. The CRASH trials represented the only “mega trial” conducted in the field of TBI.

Table 3. Trial Results: Differentiated for Type of Investigational Therapy

	Interventional Therapy	Significant Effect	No Significant Effect
	Neuroprotective agent	3	20
	Therapeutic strategy	7	3

Third, we noted that six of the nine single-center studies showed a significant treatment effect versus 4 of the 24 multicenter studies (Table 4).

Table 4. Trial Results: Single versus Multicenter Trials

	Type of Study	Significant Effect	No Significant Effect
	Single-center trial	6	3
	Multicenter trial	4	20

Fourth, a certain shift can be noted during the past 5 to 10 years of studies from the Western hemisphere toward the Far East. These primarily concern investigator-initiated studies, which we find extremely encouraging, but we also note an increasing interest of pharmaceutical companies for moving trials to include centers in the Far East. This shift is motivated by both higher patient recruitment and lower costs. Further collaboration is highly recommended, but at the same time we wish to introduce a word of caution that it is yet uncertain as to how the results obtained in a different setting may translate into a more global perspective. This is important, for example, because of various trials conducted in mainland China that have shown beneficial effects of decompressive craniectomy, hypothermia, and progesterone.10, 12, 13, 14, 15 These were all well-designed, high-quality studies. However, it is conceivable that differences may exist in referral policies, potential for selection bias, access to health care, acute and post-acute treatments, and outcome. We consider it a priority to investigate these issues in a comparative, observational study.

Two other issues deserve special consideration: 1) the number of unpublished studies and 2) the reasons why some studies were prematurely ended. In total, six studies included in the overview have not been published (tirilazad-domestic, Eliprodil, Saphir, Cerestat, Parke-Davis and HIT IV). These were all large, pivotal trials that we could include in the overview due to our knowledge of the field. However, there may be more “negative” trials to which we are not aware, particularly single-center studies. It is regrettable that these studies have not yet been published, as we consider it a moral obligation to patients and relatives who consented in participating and to the community in general to publish these results, even if a trial does not yield a positive result.

In total, seven trials were prematurely terminated. The National Acute Brain Injury Study (NABIS) hypothermia trial was originally initiated with the goal of enrolling 500 patients, but it was prematurely halted after enrollment of 392 patients after an unscheduled futility analysis showing that the likelihood for demonstrating benefit on continuation was low. The tirilazad domestic trial was prematurely halted after enrollment of 1191 of the planned 1212 patients (98.3%) on the recommendation of the data safety monitoring committee having demonstrated a significant difference in mortality between the treatment groups. A more detailed subsequent analysis found that this difference in mortality could be explained by differences in baseline characteristics; however, because the target enrollment had nearly been reached, the decision was made to terminate the study. The Selfotel trials (United States and international) were stopped after an enrollment of 693 patients, because of concerns of the safety and monitoring committee regarding an increased number of deaths and severe brain-related adverse events that had occurred in two contemporary trials in stroke. Such adverse events were not noted in the TBI trials, but as analysis indicated a low likelihood of demonstrating benefit on pursuing the trial to completion, the decision was made to definitively stop the enrollment. The Cerestat trial originally aimed to enroll 700 patients, but was halted after enrollment of 532 patients after a planned interim analysis conducted on the first 340 patients. This decision was also made against the background of concern about the effects of the agent in patients with stroke. The Parke Davis study on SNX-111 was halted prematurely when the data safety monitoring board observed a 10% increase of mortality in treated patients; the occurrence of hypotension as a complication of the use of calcium channel blockers was also noted in the study. The Bradycor trial was designed to enroll 160 evaluable patients, but was stopped after 139 patients were enrolled, and after the results of animal toxicology studies conducted during the course of the trial; these results were largely inconclusive after repeat studies, and the decision to terminate the trial was perhaps all the more regrettable as a strong trend toward benefit was noted on the Glasgow outcome score at 6 months (12% improvement) and was supported by positive trends seen in intracranial pressure, therapy intensity level, and neuropsychological tests. The CRASH mega trial was originally aimed to include a total of approximately 20,000 patients, but was stopped in May 2004 after enrollment of 10,008 patients because of an excess in early mortality (day 14). These observations demonstrate that there are many internal and external factors that influence the conduct of TBI trials. In three cases, the reasons were safety concerns within the trial. In three cases, external factors included experience in simultaneously conducted stroke trials, the results in preclinical studies, and in only one instance (ie, the NABIS hypothermia trial) this resulted from the futility analysis. Indirectly, however, aspects of futility also probably influenced the decision making in tirilazad, Selfotel, and Cerestat.

The observation that the majority of studies showing efficacy of the interventional therapy concerns single-center studies, while the overwhelming majority of multicenter studies failed to demonstrate effect is of interest and raises the question as to whether this may simply reflect publication bias or whether the center effects may be a factor confounding chances of demonstrating efficacy. Considerable between center differences in 6-month unfavorable outcome have been reported in the NABIS hypothermia trial.16 Significant differences in mortality were also found between high- and low-enrolling centers in one of the more recent trials in TBI, investigating the efficacy of dexanabinol.17 The International Mission on Prognosis and Clinical Trial Design in TBI (IMPACT) studies have further shown a 3.3-fold difference after correction for random variation in the odds of having an unfavorable outcome between centers (p < 0.001), which was not explained by patient characteristics (Lingsma et al., submitted for publication). These data illustrate some of the complexities involved in multicenter trials. The advantage of multicenter trials in comparison with single-center studies is enhanced generalizability. Two different approaches may be pursued to minimize variability in choice and sequence of basic therapeutic approaches: 1) rigorous standardization of treatment and 2) recruiting so many patients into a mega trial that variability is of less concern. An example of the latter approach is the CRASH study. However, the relative merits of these two contrasting approaches needs to be further determined.

Recently Completed, Ongoing, and Expected Studies 

General information on “new” and recently completed trials can be found on the following websites: see www.clinicaltrials.gov (U.S. National Institutes of Health trial registry) and www.controlled-trials.com (nongovernmental registry initiated by Springer Science and Business Media supervised by an international advisory group). Central registration of clinical trials via these registries or the Eudra clinical trials database in the European Union has served to promote quality in the design and reporting of clinical trials. Data from the European Medicines Agency (EMEA) registry (EudraCT) are not yet publicly available, but this is anticipated for 2010. Regulatory agencies consider formal registration of the trial mandatory when considering drug approval, and journal editors may refuse publication of trial results if the trial has not been registered prior to initiation. An overview of recently completed, ongoing, and expected studies is presented in Table 2. Much of this knowledge is now in the public domain by virtue of the central trial registries.

This overview indicates that many of the studies concern early stage clinical development. Currently, we believe that there are five phase III trials that are ongoing (four on therapeutic strategies and one on the neuroprotective agent recombinant human erythropoietin). A phase III trial on hypothermia is expected to start soon, supported by the European Society for Intensive Care Medicine. Preparations for a further two trials on progesterone are ongoing, and it is expected that these studies will be open for enrollment early in 2010.

Compared with the past decade, the number of studies on neuroprotective agents being taken forward into efficacy-oriented studies is dismal. We do not consider this due to lack of potential agents, but rather due to reluctance to embark on a high-risk venture, which is moreover extremely costly. In times of “trial crisis,” there is believed to be a clear need for improved trial methodology to decrease the risks, and we strongly believe that if we wish to continue developing the field of neuroprotection in the clinical situation, attempts should be made to reduce costs. In particular, the substantial overhead costs often imposed in academic institutes is concerning.