Safety and Tolerability of GV150526 (a Glycine Site Antagonist at the N-Methyl-d-Aspartate Receptor) in Patients With Acute Stroke
Background and Purpose—GV150526 is a novel glycine site antagonist at the N-methyl-d-aspartate receptor complex. It is a potent neuroprotective agent in animal models of stroke, including permanent middle cerebral artery occlusion in the rat. Unlike antagonists at the glutamate ligand binding site, GV150526 appears to be free of hemodynamic and central nervous system adverse effects. The purpose of this study was to assess the safety, tolerability, and pharmacokinetics of loading and maintenance infusions of GV150526 in patients with acute stroke.
Methods—This was a randomized, placebo-controlled, parallel-group, ascending-dose study conducted in 2 phases. In part A of the study, loading doses of 50, 100, 200, 400, or 800 mg were administered. In part B, the maximum loading dose from part A was followed by maintenance infusions (5 infusions at 12-hour intervals), aiming to maintain neuroprotective levels. Safety data were collected throughout. The study was not designed to test efficacy, but outcome data (Barthel Index and National Institutes of Health Stroke Scale) were collected.
Results—Sixty-six patients were recruited to the study over 11 months; 18 patients received placebo. GV150526 was well tolerated by the 48 patients who received it. There was no excess of central nervous system or hemodynamic adverse events compared with placebo. Minor abnormalities in liver function tests were observed in association with the higher maintenance doses tested. Four of 7 patients receiving the 800-mg loading dose followed by 400 mg BID and 1 of 6 patients who received the 200-mg BID maintenance dose showed a small rise in bilirubin, and 3 patients had increases in transaminases; the mean values at 72 hours remained under twice the upper limit of normal. These changes were asymptomatic and resolved within 10 days.
Conclusions—GV150526 is an emerging neuroprotective agent, with no apparent significant central nervous system or hemodynamic effects. Dose-limiting effects appear to be restricted to mild transient and asymptomatic rises in bilirubin and/or transaminases, primarily observed at high maintenance doses, and there were no findings that should preclude further clinical development.
The drug GV150526 is a novel antagonist at the glycine site of the N-methyl-d-aspartate (NMDA) receptor complex.1 Antagonism at this site specifically and selectively blocks activation of the NMDA receptor complex. In preclinical models of stroke, GV150526 reduces infarct volume and improves neurological outcome.2
In the rat model of permanent middle cerebral artery occlusion, GV150526 is a potent neuroprotective agent, with a putative neuroprotective plasma concentration of 10 to 30 μg/mL.1 It is free from cardiovascular and behavioral effects at doses of up to 30 mg/kg in rats or mice and 12 mg/kg in dogs.3 4 Animal and human volunteer studies suggest that GV150526 is less likely to cause the adverse effects seen in studies of NMDA antagonists (sedation, agitation, catatonia, nausea, gastrointestinal [GI] upset, and dyspepsia).5 In rats, toxicity of high doses is largely restricted to local injection site irritation and mild sedation. Doses up to 30 mg/kg did not cause neuronal vacuolation.6 Toxicology studies in dogs demonstrated reversible increases in hepatic enzymes, increased liver weight, and minimal bile duct proliferation possibly related to the high concentrations found in the biliary system, in turn due to the more rapid clearance in the dog.
Preclinical pharmacokinetics indicates that the clearance and volume of distribution of GV150526 are low, suggesting limited tissue distribution. The plasma elimination half-life was 6.5 hours in the rat and 2.5 hours in the dog. Most drug was eliminated by biliary excretion. In vitro plasma protein binding studies confirm that >99% is bound to albumin.
Before these studies, GV150526 had been given to 52 young healthy volunteers and 18 elderly volunteers, in doses ranging from 1 to 400 mg (Glaxo S.p.A.,Verona). Only minor adverse events were reported, and there was no change in vital signs or laboratory safety data attributable to treatment. Symptoms potentially attributed to treatment included tiredness, headache, neck ache, flatulence, loose bowels, and sore throat. Clearance and volume of distribution were low in healthy humans, with a terminal half-life of ≈19 hours. Maximum plasma concentration was linearly related to dose over the 1- to 400-mg range.
Subjects and Methods
For administrative reasons, the study was conducted under 2 protocols, but for reporting purposes, these can be considered as 2 parts of a single study. The purpose of the study was to assess safety, tolerability, and clinical pharmacokinetics of GV150526 in patients with acute stroke. Both parts of the study were approved by the West Ethics Committee of the Western Infirmary, Glasgow, Scotland. When written consent could not be obtained, consent was accepted verbally, but only in the presence of an independent witness unconnected to the trial. On rare occasions when patients were unable to give written or verbal consent, assent was accepted from the next of kin in the presence of an independent witness.
Inclusion criteria for entry into the study were as follows: men aged >18 years; postmenopausal or surgically sterilized women; and acute stroke within 12 hours of study entry (for patients with nighttime stroke, time of waking was taken as time of onset).
Patients with the following were excluded from the study: coma (unable to localize painful stimulus); known serious hepatic or renal abnormalities; administration of investigational drug within previous 3 months; acute unstable systemic illness other than acute stroke; known underlying seizure disorder; anticoagulation with warfarin (loading dose study only); and intake of >1 g aspirin in the previous 24 hours. Patients on heparin, warfarin, and high doses of aspirin were initially excluded because of the theoretical risk of pharmacokinetic interactions due to the high protein binding of GV150526; further information became available during the study that allowed the anticoagulant restrictions to be relaxed.
After admission to the acute stroke unit, patients underwent initial clinical and neurological assessment by a member of the medical staff. Blood samples were taken for biochemical screening, urinalysis was performed, and a 12-lead ECG was recorded before dosing and at intervals after dosing. Continuous ECG monitoring was performed throughout all infusions and up to 4 hours after dosing in the loading study. CT or MRI scanning was performed within 3 days to confirm the diagnosis of stroke. Scanning was not a prerequisite to study entry; patients with CT findings inconsistent with a diagnosis of stroke were to have the study drug discontinued and be replaced but would remain within the study for collection of safety data.
In the first part of the study (part A: GlaxoWellcome protocol GLYB2001), intravenous loading doses (50, 100, 200, 400, or 800 mg) were given. In the second part (part B: protocol GLYB2003), the loading dose was fixed at 800 mg GV150526, selected from tolerability data from part A, and this was followed by 5 repeated infusions at doses of 100, 200, or 400 mg, respectively, at 12-hour intervals. Patients were randomized in groups of 8 (6 active, 2 placebo) according to a double-blind, placebo-controlled, parallel-group design.
GV150526 is incompatible with saline, but 5% dextrose solution was found to be a suitable vehicle, with a final concentration of 0.8 mg/mL. The placebo group received 5% dextrose vehicle alone. Infusions were administered at a constant rate of 500 mL/h, and therefore the infusion duration varied according to dose (Table 1⇓). The only exception to the constant rate of infusion was for the 800-mg loading dose, administered to the last group of subjects in part A and to all subjects in part B. This dose was infused at 500 mL/h for 1 hour followed immediately by 167 mL/h for 3 hours (ie, 400 mg in the first hour and 400 mg over the next 3 hours). The first maintenance infusion was started 12 hours after the start of the loading dose.
Adverse event reports were collected at study entry and at intervals throughout the 1 month of study. Brief general physical examinations were performed at baseline, 24 hours, 3 days, 1 week, and 1 month. Blood pressure and pulse recordings were made with the use of semiautomatic oscillometric monitoring equipment (Marquette Electronics Inc) and repeated at frequent intervals during drug dosing. National Institutes of Health Stroke Scale7 assessments were made at baseline, 24 hours, 72 hours, 1 week, and 1 month. Barthel Index8 functional assessment was undertaken at the completion of the study, ie, at 1 month.
Coagulation was assessed by prothrombin time measurement at baseline and after 24 and 72 hours. Biochemistry and hematology testing was performed by local laboratories. Urinalysis was performed with the use of Multistix SG (Bayer Diagnostics), and the presence of blood, protein, and glucose was noted.
Blood samples (2 mL) for pharmacokinetic analysis were taken from a cannula sited in the antecubital fossa of the noninfusion arm at frequent intervals for assay of GV150526 and protein binding. Samples were stored on ice immediately after collection and centrifuged at 3000 rpm for 15 minutes at 4°C. Plasma was then transferred to a plastic vial and frozen (≤−18°C) within 3 hours of collection. Urine was continuously collected when possible for 48 hours, as well as from 60 to 84 hours in part B, and was frozen pending assay for GV150526 and creatinine.
Plasma concentrations of GV150526 were determined at the Drug Metabolism Department, Glaxo S.p.A., Verona, with the use of an online solid-phase extraction, with fully automated cartridge exchange, high-performance liquid chromatography (HPLC) separation, and UV detection. The validated limit of quantification for this analysis was 0.2 μg/mL for total and 0.5 ng/mL for free GV150526.
GV150526 in urine was assayed with the use of HPLC separation with direct injection of diluted urine and UV detection. The validated limit of quantification was 0.025 μg/mL. Standard noncompartmental methods were applied for the pharmacokinetic analysis.
A total of 66 patients were recruited to parts A and B of the study. The demographic details of the patients were comparable across doses and treatment groups (part A: mean age of 74 years and weight of 68.3 kg for placebo versus 69 years and 71 kg for GV150526; part B: 78 years and 69.5 kg for placebo versus 69 years and 62.9 kg for GV150526). All CT scan results were consistent with a diagnosis of acute stroke. Eighteen patients received placebo (12 within part A and 6 within part B). In part A, 6 each received loading doses of 50, 100, 400, and 800 mg GV150526. Because of a dosing error, only 5 patients received the 200-mg dose (the sixth received 400 mg over 30 minutes). In part B, 6 patients were administered 100 and 200 mg BID, and 7 patients received the highest dose of 400 mg BID. Sixty patients were treated within 12 hours of stroke symptom recognition; in 6 patients, practical problems led to delays in drug administration to beyond this.
Safety and Tolerability
Seven of the 66 patients died during the 30-day follow-up period: 6 in part A and 1 in part B. No deaths were thought to be due to administration of study drug. The fatal events are summarized in Table 2⇓. The study death rate of 10.6% over a 1-month follow up period is comparable to the overall 12.2% mortality in our stroke unit for 1996 and published data.9
Within the active arm of part A, a single patient was subsequently demonstrated to have primary intracerebral hemorrhage; in part B, 1 patient in the placebo group and 4 actively treated patients had primary intracerebral hemorrhage diagnosed after treatment. There was no excess of adverse events within the hemorrhagic subgroup. There was minor imbalance in stroke territories: large middle cerebral artery territory strokes (placebo 4, active 10), smaller cortical strokes (placebo 5, active 17), lacunar strokes (placebo 8, active 17), and posterior circulation strokes (placebo 1, active 4).
Seven patients were withdrawn from the study during the infusion phase. In part A, 1 patient developed unstable atrial fibrillation immediately before the 100-mg loading dose, and 1 was withdrawn after a deterioration in level of consciousness during the 800-mg infusion. In part B, 1 subject receiving placebo was withdrawn because of extension of stroke. A patient receiving the 100-mg BID maintenance infusion developed vomiting and occult intestinal bleeding (subsequent investigations showed that GI bleeding had preceded the stroke). One patient receiving the 200-mg BID infusion developed reduced level of consciousness. Two patients in the 400-mg BID dose group were withdrawn because of reduced level of consciousness and phlebitis. One other patient in the 200-mg BID dose group developed mild phlebitis but completed the treatment course.
Patients administered concomitant heparin or warfarin were excluded from part A, but in part B, 1 placebo patient and 3 active arm patients received heparin. An additional 2 patients received warfarin alone, and 1 patient who received heparin also received warfarin. There were no bleeding complications in any patients receiving anticoagulant therapy.
There were no consistently reported or dose-related side effects and no increase in reported adverse events in the actively treated groups (Tables 3⇓ and 4⇓). No serious adverse events reported in either part A or B were thought likely to be the result of drug administration. No changes in blood pressure or heart rate consistent with drug effect were reported. Analysis of ventricular rate and PR, QRS, and QT intervals at the time of maximal drug concentration versus baseline in part A revealed no treatment effect (Table 5⇓).
In part A, transient mild increases in liver enzymes above the normal range (γ-glutamyltransferase [GGT] and alanine aminotransferase [ALT]) were observed in 6 of 29 actively treated patients (21%) compared with 3 of 12 controls (25%). Mild anemia was reported in a total of 4 of 29 actively treated patients (14%) compared with 1 of 12 placebo patients (8.3%), but there was no increased frequency at the higher dose levels, and in each case hematologic indices were consistent with preexisting iron deficiency. One patient in the 800-mg group developed a transient reduction in platelet count 72 hours after commencing drug infusion (142×1012/L at baseline, falling to 81×1012/L, and recovering to 164×1012/L at 1 week). This was not associated with any abnormalities in total red or white cell count that may have suggested marrow failure.
In part B, asymptomatic dose-dependent elevations in bilirubin, GGT, and liver transaminases consistent with a cholestatic pattern were observed after repeated infusions of active treatment but not placebo (Figure 1⇓). One of 6 patients receiving 100 mg BID had a 3-fold rise in GGT and ALT at 24 hours. In the 200-mg group, 2 of 6 patients had liver function abnormalities: 1 patient had an isolated 3-fold increase in ALT at 84 hours, and a second patient had a 2-fold increase in bilirubin, aspartate aminotransferase, and ALT at 84 hours. In the group receiving 400 mg BID, 4 of 7 patients had significant rises in bilirubin, and 3 of 7 had increases in GGT and alkaline phosphatase, consistent with a hepatic cholestatic pattern. All changes resolved without intervention by 10 days.
GV150526 had no effect on plasma glucose levels compared with the control group (data not shown). However, an unplanned analysis showed that the patients treated in part B had higher plasma glucose levels at 24 hours than patients in part A (P<0.001, ANOVA; Figure 2⇓). The effect on plasma glucose at 24 hours correlated with the volume of 5% dextrose infused within the first 24 hours (R2=0.16, P=0.002; Figure 3⇓).
The pharmacokinetics of GV150526 was determined for the single- and multiple-dose regimens. After single doses (part A), the pharmacokinetics of GV150526 was similar to that in healthy volunteers. The median maximum concentration ranged from 16.5 μg/mL after 50 mg to 106 μg/mL after an 800-mg single dose. The median clearance, steady-state volume of distribution, and terminal half-life ranged from 0.27 to 0.38 L/h, 6.7 to 9.9 L, and 23 to 35 hours, respectively, across the dose groups. Renal clearance accounted for <1% of the administered dose. The concentration profile of GV150526 after the 800-mg loading infusion for the multiple-dose regimen (part B) was consistent with that for the 800-mg single dose (part A). The pharmacokinetics of GV150526 after the multiple-dose regimen was determined with the use of compartmental analysis and will be reported separately.
The study was designed to assess safety and not designed to test efficacy. Therefore, as expected, the results of the National Institutes of Health Stroke Scale and Barthel Index at 1-month follow-up failed to demonstrate differences between drug- and placebo-treated patients.
GV150526 had no effect on blood pressure, heart rate, or the ECG. No serious adverse events related to drug therapy were reported, and there was no trend to altered functional outcome.
Plasma concentrations predicted to be neuroprotective in humans from the rat post–MCA occlusion model were successfully exceeded and were well tolerated. Loading doses >800 mg were not examined because 1 L of 5% dextrose was considered the maximum reasonable fluid volume that could routinely be given in a few hours. Maintenance infusions of 200 and 400 mg BID sustained plasma levels of GV150526 above the concentration demonstrated to be neuroprotective in animal models of >10 μg/mL, even after allowing for lower free concentrations in humans.
Interestingly, there were no adverse psychotomimetic or hemodynamic effects, as frequently reported with NMDA antagonists. This is consistent with the preclinical data. However, there is no published clinical evidence that GV150526 penetrates into brain tissue. It may be that antagonism of the glycine binding site of the NMDA receptor complex rather than the glutamate binding site leads to neuroprotection without hemodynamic or psychotomimetic effects. If so, GV150526 would have a significant advantage over other potential neuroprotective agents that are less well tolerated.
Hemodynamic effects are of particular concern to physicians attending stroke patients. Feared complications include hemorrhagic transformation or brain edema in patients with hypertensive responses and exacerbation of the infarction in those with blood pressure falls, either as a direct consequence of neuroprotective treatment or as a result of treatment given to reverse a hypertensive drug effect. GV150526 was free from hemodynamic effects.
Pharmacokinetic analyses in volunteers suggest that GV150526 has a relatively long elimination half-life, permitting 12-hour maintenance dosing. The alternative of constant infusion could restrict early rehabilitation, may exaggerate venous irritation, and requires closer supervision.
The relatively large volume of infusate administered (1 L over 4 hours initially and 1.5 L/d at the highest doses) raises concerns about fluid overload, particularly in elderly patients with coexisting cardiac and renal disease. We did not observe heart failure in these studies. For future studies and clinical use, a more concentrated formulation is now available, which allows the drug to be given in half the volume of fluid.
The infusion of dextrose (the drug is incompatible with saline) also raises theoretical risks of hyperglycemia. Hyperglycemia, even from ≥8 mmol/L, is associated with poorer outcome after stroke.9 GV150526 has no pharmacological effect on plasma glucose. Since the control group in our study received the dextrose vehicle, no firm conclusion can be drawn regarding the effect of the infusions; however, the combined placebo and active data from part A of the study can be compared with part B. This comparison suggests that a small but potentially relevant rise in blood glucose is detectable during the repeated infusions, which was not present in patients who received only a loading infusion. Furthermore, this effect correlates loosely with the total volume of dextrose infused over 24 hours. The effect of this blood glucose change on outcome cannot be determined either by these studies or by the subsequent clinical trials of which we are aware. We have reexamined blood glucose data from a similar trial that used saline as a vehicle: in that study, a less marked rise in plasma glucose also occurred at 24 hours (from 6.9±2.6 to 8.3±2.9 mmol/L; n= 57; P=0.13). Infusion of small volumes of dextrose and/or saline is standard practice to maintain hydration in many stroke patients. The vehicle used for the GV150526 studies reported here appears to cause a slightly greater rise in plasma glucose than would be expected with saline, but the rise is transient and is seen primarily with the highest volumes infused. The new 1.6 mg/mL formulation will reduce total volume and glucose load by 50%.
The abnormalities in liver function reported in the prolonged administration phase of the study merit further examination. It is possible that the changes observed in this study are analogous to the rise in hepatic enzymes associated with increases in liver weight and bile duct proliferation observed in studies in dogs. Reassuringly, in these studies hepatic pathological changes were reversible and had largely resolved by the end of the 10-day recovery period. An alternative explanation that has been advanced concerns a binding interaction between bilirubin and GV150526.10 Patients with elevated bilirubin tended to show time-dependent increases in free fraction of GV150526 and increases in its conjugated metabolites. Potential sites for the GV150526 and bilirubin interaction include altered hepatocyte uptake, glucuronide conjugation, and active transport into bile.10 While rises in liver function tests of 2 to 3 times the normal reference range in themselves may be of little clinical significance, particularly since they were transient and all patients were asymptomatic, monitoring of liver function should be undertaken in the next study with GV150526.
In summary, GV150526 appears to be a well-tolerated glycine antagonist with a favorable pharmacokinetic profile to permit twice daily infusion. Minor changes in hepatic biochemistry require further assessment. From these results, further evaluation of the 800-mg loading infusion followed by either 200 or 400 mg BID maintenance infusions would be appropriate, although data from an additional phase 2 study recently presented suggest that the optimal dose for further study appears to be 800 mg as a loading infusion, followed by 200-mg maintenance infusions twice daily.11 This is the dose now selected for 2 phase III clinical efficacy studies that are currently recruiting patients in North America (GAIN-Americas) and in Europe, Africa, and Australasia (GAIN-International).11
This study was sponsored by Glaxo Research and Development (now GlaxoWellcome). The study would not have been possible without the cooperation of the patients and staff of the Acute Stroke Unit at the Western Infirmary, particularly Sister Elizabeth Colquhoun and her team of dedicated research nurses, and consultant colleagues Dr G.T. McInnes, Professor J.L. Reid, and Dr P.F. Semple. The ECG analysis was undertaken by Dr M.R. Walters. The authors gratefully acknowledge company support in designing the study from Drs Tim Corn and Cliff Preston; in pharmacokinetic analysis from Dr Frank Hoke; in collating data from Jane Lavelle; and in manuscript review from Elizabeth Ashford and colleagues.
- Received January 5, 1999.
- Revision received February 22, 1999.
- Accepted February 22, 1999.
- Copyright © 1999 by American Heart Association
Di Fabio R, Cugola A, Donati D, Feriani A, Gaviraghi G, Ratti E, Trist D, Reggiani A. Identification and pharmacological characterization of GV150526, a novel glycine antagonist as a potent neuroprotective agent. Drugs Future. 1998;23:61–69.
Gaviraghi G, Pietra C, Ratti E, Trist D, Reggiani A. GV150526A: a novel glycine antagonist with neuroprotective activity devoid of side-effects associated to NMDA receptor blockade. Cerebrovasc Dis. 1996;6:41–42. Abstract.
Reggiani A, Pietra C, Tarter G, Tessari M, Ratti E, Trist D, Gaviraghi G. Pre and post-ischaemia effect of GV150526A in rat transient MCA occlusion. Cerebrovasc Dis. 1996;6(suppl 2):5. Abstract.
Muir KW, Lees KR. Clinical experience with excitatory amino acid antagonist drugs. Stroke. 1995;26:503–513.
Goldstein LB, Samsa GP. Reliability of the National Institutes of Health Stroke Scale: Extension to non-neurologists in the context of a clinical trial. Stroke. 1997;28:307–310.
Weir CJ, Murray GD, Dyker AG, Lees KR. Is hyperglycaemia an independent predictor of poor outcome after acute stroke? Results of a long term follow up study. BMJ. 1997;314:1303–1306.
Hoke JF, Dyker AG, McAllister AM, Lees KR. Pharmacokinetics of GV150526A following multiple intravenous doses in acute stroke patients. Cerebrovasc Dis. 1997;7(suppl 4):29. Abstract.
GAIN European Study Group. Safety and tolerability of GV150526 in acute stroke (GLYB2002). Cerebrovasc Dis. 1998;8(suppl 4):20. Abstract.