Glyceryl Trinitrate for Acute Intracerebral Hemorrhage

Introduction

Spontaneous intracerebral hemorrhage (ICH) is a severe form of stroke with more than two-third of survivors disabled at 3 months and less than one-half surviving the first year.¹ High blood pressure (BP) is common in acute ICH and is associated independently with a worse outcome,² in part mediated through expansion of the hematoma.^3,4 In the large Intensive Blood Pressure Reduction in Acute Cerebral Hemorrhage Trial 2 (INTERACT-2) intensive BP lowering during the first 6 hours was associated with a trend to improved functional outcome in comparison with guideline BP lowering.⁵ In contrast, in a subgroup analysis of patients with acute ICH enrolled into the Scandinavian Candesartan Acute Stroke Trial (SCAST), treatment with oral candesartan was associated with a worse functional outcome.⁶ Hence, the management of high BP in acute ICH remains uncertain.

Transdermal glyceryl trinitrate (GTN), a nitric oxide donor is a candidate treatment for acute ICH because it can lower BP without changing cerebral blood flow, has no negative effects on platelet function^7–9 and can be given to patients with dysphagia, a common clinical complication of stroke.¹⁰ The Efficacy of Nitric Oxide in Stroke (ENOS) trial assessed the safety and efficacy of BP lowering with transdermal GTN in 4011 patients with acute stroke¹¹; nearly one-fifth of these presented with spontaneous ICH.¹¹ Although the main analysis showed that GTN did not improve death or dependency at 90 days after acute ischemic stroke or ICH, apparent benefit was observed in patients randomized within 6 hours of stroke onset.¹¹ This result mirrors a result seen in the prehospital pilot Rapid Intervention With Glyceryl Trinitrate in Hypertensive Stroke Trial (RIGHT), where GTN was administered by paramedics within 4 hours of onset.¹⁰ In this prespecified analysis,¹² we have further assessed the effect of GTN in the subgroup of patients randomized into the ENOS trial after ICH, both overall (here called ENOS-ICH) and within 6 hours; the time window of 6 hours matches that for recruitment into the INTERACT-2 trial⁵ and encompasses the time window studied in RIGHT.¹⁰

Methods

The ENOS trial protocol, statistical analysis plan, baseline data, and main results have been published.^11–14 In brief, ENOS compared the effect of transdermal GTN (5 mg daily) versus no GTN, given for 1 week, in patients with acute stroke (randomization within 48 hours of ictus) and high systolic BP (140–220 mm Hg). Patients taking antihypertensive drugs before their stroke were also randomized to continue or stop these temporarily for 1 week. During treatment, BP was measured daily using a validated automatic BP monitor (Omron 705CP) supplied to each site.¹⁵ The trial was registered (ISRCTN99414122) and approved by the ethics committees and competent authorities in all participating countries as appropriate. Patients or relatives gave consent or proxy consent, respectively. In the present analysis, we included all patients enrolled into ENOS with ICH (ENOS-ICH).

Brain Imaging

Participants had a baseline computed tomographic or magnetic resonance imaging scan as part of clinical care, usually before randomization. Where possible, a second research computed tomography or magnetic resonance imaging was performed at day 7+1 (end of treatment). Uncompressed DICOM, JPEG, PNG, or GIF image files were sent to the coordinating center, either uploaded via a secure website or on a compact disc. Images sent on film were digitized using a VICOM digitizer. Scans were assessed centrally using validated scales by expert neuroradiologists or trained neurologists (A.A., L.A.C., A.C., J.L.B., R.A.D., P.K., and J.M.W.) for the presence of hemorrhage, its location in the brain, the presence and amount of mass effect, the presence of blood in the subarachnoid space or ventricles, and underlying changes in the brain, including cerebral atrophy,¹⁶ leukoaraiosis,¹⁶ and the presence of old lesions.

Hematoma volume (measured manually by ABC/2 formula¹⁷ using Osirix software¹⁸ on a 26-inch Apple iMac), shape and density (using an ordered 5-point categorical scale),¹⁹ shape index (perimeter of hematoma/4π×area),²⁰ density index (SD/mean attenuation),²¹ and the presence of blood in the subarachnoid space or ventricles using the Graeb score, and its modified version^22,23 were also measured. All imaging assessments were performed by K.K. masked to clinical data and treatment assignment.

Outcomes

The primary outcome was functional outcome assessed using the modified Rankin Scale (mRS) at 90 days after randomization. Secondary outcomes studied at day 90 included activities of daily living (Barthel Index²⁴), cognition (modified telephone Mini-Mental State Examination²⁵), Telephone Interview for Cognition Scale,²⁶ and health-related quality of life (European Quality of Life-5 dimensions-3 level,²⁷ from which health utility status was calculated,²⁸ and mood [short Zung Depression Scale score²⁹]). Safety outcomes comprised all-cause mortality and case-specific fatality, early neurological deterioration (defined as a decrease of at least 5 points or decrease in consciousness of >2 points from baseline to day 7 on the Scandinavian Stroke Scale [SSS]), recurrent stroke by day 7, symptomatic hypotension, hypertension,¹¹ and serious adverse events. Outcomes at day 90 were assessed via telephone by trained investigators at national coordinating centers, who were masked to treatment allocation.

Analyses

Statistical analysis was performed by intention-to-treat and followed the trial’s statistical analysis plan and analysis approaches used in the primary publication.^11,12 Analyses were performed for all patients with ICH in ENOS and separately for those with ICH, who were randomized within 6 hours of onset. Data are shown as number (%), median [interquartile range], or mean (SD). Patients who died were allocated an extreme score: −5: Barthel Index; −1: EQ-Visual Analogue Scale, SSS, telephone Mini-Mental State Examination, Telephone Interview for Cognition Scale, verbal fluency; 0: health utility status (derived from European Quality of Life-5 dimensions-3 level); 6: mRS; and 102.5: Zung Depression Scale.^11,30 Comparisons between the treatment groups were performed with binary logistic regression, ordinal logistic regression, Cox proportional regression, or multiple linear regression. Statistical models were adjusted for prognostic baseline covariates: age, systolic BP, SSS score, time from symptom onset to randomization, hematoma volume, and treatment assignment (GTN versus no GTN). Odds ratio, hazard ratio, or mean difference, with 95% confidence intervals, is given, and statistical significance was set at P≤0.05. Heterogeneity of treatment effect was assessed by including an interaction term in adjusted models. Analyses were performed using SPSS (version 21) on an Apple Mac.

Results

Baseline Characteristics

Between July 2001 and October 2013, a total of 629 participants with ICH were enrolled in the trial; 310 participants were randomly assigned to receive GTN and 319 participants to no GTN (Table 1). Baseline characteristics were well matched between the 2 groups. The average age was 67 years; 66% of patients were men; 54% were enrolled from the United Kingdom; mean time from onset to recruitment was 25 hours; and mean stroke severity was 30.5 (SD, 12.4) on the SSS, equivalent to National Institute of Health Stroke Scale (NIHSS) score 12.6 (5.3).³¹

A majority (71%) of patients had their baseline scans performed within 12 hours of onset (Table 1). Eighty-seven percent of hemorrhages were deep-seated in the lacunar and striatocapsular brain regions, most hematomas (63%) caused mass effect (moderate or extreme swelling), and many (67%) patients had leukoaraiosis. Evidence of a previous stroke was present in 50% of patients, and brain atrophy was seen in 62% of the available scans. The mean hematoma volume was 13.3 cm³, and 153 (26%) patients had an intraventricular hemorrhage. Additional information on baseline neuroimaging is given in Table I in the online-only Data Supplement.

Blood Pressure

Mean BP at baseline was 172.1/93.4 mm Hg and fell in both treatment groups over the first week. Following the first dose of GTN versus no GTN, BP was significantly lower by 7.5/4.2 mm Hg (P=0.02/0.05, respectively); BP did not differ thereafter (Figure I in the online-only Data Supplement).

Clinical Outcomes

At day 90, the median mRS was 3 (interquartile range, 2) in both the GTN and no GTN groups and did not differ in an adjusted analysis (common odds ratio, 1.04; 95% confidence interval, 0.78–1.38; Table 2; Figure 1) or unadjusted analysis (data not shown). A test of goodness-of-fit showed that the assumption of proportional odds was not violated (P=0.09). When assessed in subgroups defined by baseline clinical or neuroimaging factors, there were significant interactions between treatment and mRS for time to randomization and stroke subtype (Figures II and III in the online-only Data Supplement).

• Download figure
• Download PowerPoint

The cumulative risk of all causes of death during follow-up did not differ between GTN and no GTN (adjusted hazard ratio, 1.02; 95% confidence interval, 0.67–1.56, P=0.92; Figure IV in the online-only Data Supplement). There were no significant differences between the 2 groups in any of the secondary outcomes studied at day 7 or day 90, including measures of disability, cognition, mood, and quality of life (Table 2). The number of patients with a serious adverse event during follow-up did not differ between the treatment groups (24.2% versus 21.9%; P=0.50; Table II in the online-only Data Supplement).

Relationship Between Baseline Neuroimaging and mRS at Day 90

Table 3 shows the association between baseline neuroimaging characteristics and the primary outcome of mRS. Imaging measures that were significantly associated with outcome on both univariate and covariate-adjusted analyses comprised the presence of intraventricular hemorrhage or atrophy, irregular hematoma shape, and heterogeneous density.

Patients Randomized Within 6 Hours

Of the 629 patients with ICH, 61 (9.7%) participants were randomized within 6 hours; the average time to treatment was 4.4 (1.2) hours (Table 1). Patients were less likely to be enrolled in Asia or other non-UK countries, had a larger initial hemorrhage volume (mean, 16.9 cm³), and were more likely to have intraventricular hemorrhage.

Patients randomized to GTN (versus no GTN) had less impairment (higher SSS) at day 7 and were less likely to die in hospital (Table 2). At day 90, GTN was associated with an improved functional outcome assessed using the mRS, manifest as a shift to less dependency (Table 2; Figure 2). Similarly, participants randomized to GTN were less disabled and had significantly better quality of life, mood, and cognition scores (Table 2). A trend to a reduction in death was seen in those patients randomized to GTN versus no GTN (adjusted hazard ratio, 0.19; 95% confidence interval, 0.03–1.01; P=0.051).

• Download figure
• Download PowerPoint

Effect of GTN on Hemorrhage Measures at Day 7

One hundred and eighty-one patients had repeat brain imaging at 1 week for an assessment of the effect on hemorrhage characteristics (Table 4). Of these, 93 patients received GTN and 88 to no GTN. When adjusted for baseline value, treatment with GTN was associated with a smaller hematoma volume (mean difference, −4.3 cm³; P=0.06).

Discussion

In this subgroup of patients enrolled into the ENOS trial with ICH, functional outcome (assessed using the mRS) did not improve with GTN when compared with no GTN. This result mirrors that across the main study and is in spite of GTN reducing BP by 7.5/4.2 mm Hg. Furthermore, no benefit was seen in key secondary outcomes, including activities of daily living, cognition, mood, and quality of life. The absence of significant differences in the rates of deaths or serious adverse events between the 2 treatment groups suggests that treatment with GTN is safe. In a prespecified analysis of the effect of treatment in patients randomized within 6 hours of ICH onset, GTN reduced early impairment; late dependency, disability, and death; and improved late cognition, mood, and quality of life.

These findings may have several explanations. First, patients could be randomized ≤48 hours after stroke onset. The large INTERACT-2 trial suggested that intensive BP lowering might be effective in patients enrolled within 6 hours⁵ so the time window for recruitment into ENOS (mean 25.1 hours, maximum 48 hours) may have been too long. Supporting this is the observation that patients randomized within 6 hours of ICH onset into ENOS seemed to benefit with less dependency, disability, impairment and mood disturbance, and better cognition and quality of life. Very early BP lowering might help limit hematoma expansion.^32,33 Second, the degree by which BP was lowered may have been too small; INTERACT-2 achieved a reduction of 14 mm Hg by 6 hours and showed a near-significant effect on functional outcome.⁵ Finally, the duration of BP control in ENOS-ICH was limited to 3 days as tachyphylaxis developed, a known feature of organic nitrate therapy. Nevertheless, these explanations are confounded by results from the subgroup of patients with ICH randomized into the large SCAST,⁶ where oral candesartan was associated with a worse functional outcome.⁶ Here, treatment could be started up to 30 hours after stroke onset, and the difference in BP between active and placebo groups was smaller than in ENOS-ICH at 6.3/3.3 mm Hg.

The reduction in ICH volume in the GTN group (4 cm³) was similar to the effect observed in trials of recombinant activated clotting factor VII, although the agent was tested in earlier time windows.^34,35 Mechanisms by which nitric oxide donors might reduce hematoma volume and improve functional outcome if given early include lowering BP, neuroprotection, and improving collateral blood flow. The latter 2 effects have been seen in experimental models of brain ischemia^36,37 and may be of relevance after ICH.

This subgroup analysis of ENOS has several strengths, including recruitment of patients with ICH from multiple ethnic groups across 5 continents over a wide time window representative of routine clinical practice. Baseline neuroimaging and clinical outcomes were assessed masked to treatment assignment,¹³ follow-up was near complete,¹¹ and the analysis was prespecified.¹² However, exclusion of patients with low or normal BP (systolic BP <140 mm Hg) or very high (>220 mm Hg), those with reduced consciousness (GCS score <8), and those without motor signs may have limited the external validity of the findings and especially the number of patients with large hematoma.

In conclusion, this subgroup analysis of ENOS was neutral and did not identify any beneficial effect or harm in lowering BP with GTN in patients with acute ICH. Transdermal GTN seems to be safe and modestly effective in lowering BP in acute ICH, which can be useful in patients who are unable to swallow. The results in those patients randomized within 6 hours of ICH onset and recent guidelines^38,39 support ongoing or planned trials of lowering BP in the ultra-acute and hyperacute periods after stroke, including the Antihypertensive Treatment of Acute Cerebral Hemorrhage II (ATACH-2) Trial of intravenous nicardipine⁴⁰ and RIGHT-2 of GTN administered in the prehospital phase of stroke (ISTRCTN26986053).

Acknowledgments

We thank the patients included in the study, the investigators (listed in the baseline and primary result publications^11,14), and research staff who enrolled patients into Efficacy of Nitric Oxide in Stroke (ENOS).

Footnotes