Glyceryl Trinitrate versus Sham in the Treatment of Hyperacute Stroke: The Efficacy of Nitric Oxide in Stroke-2 (ENOS-2) Feasibility Randomised Controlled Trial

Abstract

Introduction

High blood pressure (BP) is associated with a poor outcome after stroke. Trials of transdermal glyceryl trinitrate (GTN), a nitric oxide donor, have suggested that treatment between 3 and 5 h might improve functional outcome.

Methods

We randomly assigned hospitalised patients with an acute ischaemic or haemorrhagic stroke to 2 days of transdermal GTN (5 mg/day) or sham, started between 3 and 5 h after onset. The primary feasibility outcome was recruitment rate; proof-of-concept (PoC) was assessed at 90 days by central observers blinded to treatment assignment using the modified Rankin Scale (mRS). Data are number (%), median [interquartile range], or mean (standard deviation). Comparisons were assessed by multiple linear regression.

Results

Thirty-nine of an intended 120 participants were recruited. A total of 3,314 people were excluded, commonly related to presentation >5 h of onset or outside of researcher working hours, no eligible symptoms/signs or an unclear onset time. Mean age was 72 (SD 13) years, females were 41%, BP was 161.8 (18.4)/80.8 (14.9) mm Hg, and time from onset at baseline was 216 [186, 251] minutes. The fall in BP over 24 h did not differ between GTN versus sham. mRS at 3 months did not differ between the groups (difference in means, DIM −0.20, 95% confidence intervals −1.30, 0.90; p = 0.72). GTN was associated with improved cognition on the telephone interview of cognition scale (DIM 8.0, 95% CI 1.3, 14.8; p = 0.020). Although headache was more common, GTN was associated with fewer serious adverse events (p = 0.020).

Conclusion

Recruitment limitations in this small single-centre trial prevented demonstration of feasibility for patients in the time window of 3–5 h post ictus. GTN demonstrated some evidence of PoC and was safe. A multicentre trial needs to further test this hypothesis.

Introduction

High blood pressure (BP) is present in 60% of patients with acute ischaemic stroke (AIS) or acute intracerebral haemorrhage (ICH) [1] and is associated with a worse outcome manifest as increased early recurrence or death and later death and dependency [2–4]. Lowering BP acutely after AIS might improve outcome providing that cerebral perfusion is not reduced in the presence of dysfunctional cerebral autoregulation. BP should not be lowered if already in the normal or low range since a U-shaped relationship exists between BP and outcome, i.e., low BP is also associated with poor outcome [1, 5]. Lowering BP in ICH is associated with reduced haematoma expansion.

Nitric oxide (NO) donors are candidate treatments for acute stroke through multiple potentially beneficial effects: NO is a vasodilator and so lowers BP and might improve cerebral perfusion; modulates endothelial, smooth muscle cell, platelet and leucocyte neuronal function; modulates neuronal function and is neuroprotective; and inhibits apoptosis. In preclinical studies of AIS, early administration of NO donors reduced infarct lesion size and improved regional cerebral blood flow and functional outcome [6–8]. Four small phase-2 trials of transdermal glyceryl trinitrate (GTN, also known as nitroglycerin, a NO donor) reported that GTN lowered systemic and 24-h BP; did not alter platelet activity, middle cerebral artery blood flow velocity, or regional cerebral blood flow; improved aortic compliance; could be given to patients with dysphagia; and was safe in patients with acute stroke [9–12]. When started within 4 h of stroke onset, GTN was associated with improved functional outcome [12].

Three phase-3 randomised controlled trials of GTN have been reported. The large hospital-based ENOS trial recruited 4,011 patients and had a recruitment time window of 48 h. Although the trial was neutral overall [13], the subgroup of participants treated within 6 h of stroke onset appeared to benefit [14]. An individual patient data meta-analysis found that GTN appeared to be beneficial for both AIS and ICH when treatment was started within 6 h [15]. Hence, a large prehospital ambulance-based trial, RIGHT-2, recruited 1,149 patients with probable stroke within 4 h (median 71 min). This trial was neutral overall [16], although patients with ICH (when treated within 2 h) and AIS (when treated within 1 h) appeared to fare worse, while those with a stroke mimic had a better outcome [17, 18]. Another prehospital trial, MR ASAP, randomised 325 patients up to 3 h after stroke onset (median ∼62 min) and was also neutral overall [19]; nevertheless, it too suggested outcome was worse in ICH with GTN. Taking the results together suggests that early but not too early treatment might be better, i.e., after 2 but before 6 h.

We report here the findings of the Efficacy of Nitric Oxide in Stroke-2 (ENOS-2) randomised controlled trial, the objective of which was to assess the feasibility, safety, and proof-of-concept (PoC) when GTN was administered between 3 and 5 h after stroke onset.

Methods

Trial Design

ENOS-2 was a prospective randomised sham-controlled blinded-endpoint parallel-group single-centre hospital-based feasibility and proof-of-concept trial. In brief, patients with an acute stroke were randomised equally to 2 days of transdermal GTN or a sham patch. The trial broadly followed the design of ENOS [13] but without assessment of stopping versus continuing pre-stroke antihypertensive drugs. A preprint edition of the trial report is published [20].

Participants

Patients admitted with a clinical stroke syndrome due to ischaemia or haemorrhage to Nottingham University Hospitals NHS Trust, a comprehensive stroke centre, were eligible for the trial if they were: aged 18 or over; had a stroke syndrome comprising limb weakness, dysphasia, neglect or hemianopia; had systolic BP >120 mm Hg; and could be treated between 3 and 5 h after onset. The diagnosis of ischaemic or haemorrhagic stroke was confirmed with computed tomography or magnetic resonance imaging done before, or soon after, enrolment using standard techniques; the scan had to show acute ICH, AIS, or be normal.

Key exclusion criteria included the following: preceding severe dependency (modified Rankin scale [mRS] 4 or 5), plasma glucose <3 mmol/L, coma (Glasgow Coma Scale <8), witnessed seizure at presentation, known life expectancy <6 months, pure sensory symptoms, a condition mimicking stroke, other cause of intracranial haemorrhage, known allergy to GTN patch or sham dressing, planned for palliative care, recent use of a phosphodiesterase 5-inhibitor, or previous enrolment in ENOS-2. Co-enrolment was allowed with mutually agreed trials: MAPS-2 (ISTRCTN40512746), OPTIMIST (clinicaltrials.com NCT03734640), and TICH-3 (ISRCTN97695350).

Randomisation and Masking

Investigators entered baseline and follow-up data into a database via a secure web-based randomisation system (https://stroke.nottingham.ac.uk/enos-2/live/enos-2_login.php). Data were checked to confirm the patient’s eligibility in real time, and the computer system then assigned a participant to treatment with GTN or sham with allocation 1:1. Treatment assignment comprised stratification on stroke type (IS, ICH) and minimisation on key prognostic baseline variables: age (≥72 years), severity (NIHSS ≥10/42), time from onset to randomisation (≥4 h), systolic BP (≥160 mm Hg) and candidate for or received reperfusion therapy (“no”); minimisation included a random element in 5% of patients. Minimisation was used to ensure that the groups were balanced for prognostic factors, and the random element reduced predictability. The randomisation algorithm then presented a treatment allocation as either “GTN” or “no GTN”; GTN/sham was prescribed via the hospital system.

Intervention

Treatment was started immediately after randomisation, given daily for 2 days and consisted of a GTN transdermal patch (Transiderm-Nitro, 5 mg) or a non-matching sham dressing (DuoDERM hydrocolloid, ConvaTec) with both covered by a gauze dressing to conceal allocation. The patch/dressing and gauze were changed daily (07:00–09:00 h) with rotation of the site of placement on the shoulders or back. The dose of GTN was not adjusted during treatment and patches were left on for 24 h, i.e., there was no overnight GTN-free period [13].

Study agents could be stopped if the patient withdrew consent, for safety reasons, or if unacceptable adverse events developed. Non-trial nitrates and other antihypertensive agents could not be given during the 2-day treatment period unless deemed necessary by the local investigator. Randomised treatment was given in addition to standard care; thrombolysis and/or mechanical thrombectomy were permitted in patients with AIS as per the admitting physician’s plan. Systematic use of antihypertensive agents (all patients, after 2 days), and oral antithrombotic and lipid lowering agents (patients with AIS) were recommended for secondary prevention.

Assessments

Demographic and clinical characteristics were recorded online during enrolment. The severity of the stroke was assessed with the National Institutes of Health Stroke Scale (NIHSS, range from 0 to 42, with higher scores indicating a more severe neurologic deficit) [21] at baseline, 2 days, and discharge. BP was measured using a validated automated monitor as used routinely at the site. The final diagnosis was confirmed at the time of discharge/in-hospital death. The site was asked to provide a trained member of their research staff who was unaware of treatment assignment to perform post-randomisation in-hospital assessments. Computed tomography and magnetic resonance imaging scans were evaluated using the routine hospital report as read by hospital neuroradiologists who were blinded to treatment; a natural language programme was used to confirm interpretation of the report. Investigator-reported serious adverse events (SAEs), including cause-specific deaths, were validated by the chief investigator who were blinded to treatment assignment. Patients who did not receive their assigned treatment or who did not adhere to the protocol were followed up in full. Final follow-up was performed centrally at 90 days by telephone from the coordinating centre, blinded to treatment allocation. If the patient could not be contacted, a questionnaire covering the same outcome measures was sent by post.

Outcome Measures

The primary outcomes were as follows: feasibility assessed as recruitment of 120 participants (100 with AIS, 20 with ICH); mechanistic assessed as BP and heart rate; and clinical assessed as functional outcome using the modified Rankin Scale (mRS) at 90 days after enrolment [22]. Scores on the mRS range from 0 to 6, with a score of 0 indicating no symptoms, 5 severe dependency, and 6 death. Prespecified secondary clinical outcomes included disability (Barthel Index [BI] [23]); cognition (telephone interview for cognition scale-modified [TICS-M] [24]); mood (short Zung Depression Score [ZDS] [25]); health-related quality of life (EuroQol-5 dimensions-5 levels [EQ-5D-5L], from which health utility status [HUS] using UK weightings was calculated, and EuroQoL-visual analogue scale [EQ-VAS]); at 90 days. These outcomes were measured by central observers blinded to treatment assignment. At discharge from initial hospitalisation, duration of stay and discharge destination (to institution or home) were recorded. The safety outcomes of interest were SAEs up to day 2 (i.e., 48 h after 1st patch applied), fatal SAEs to day 90, and all-cause case fatality.

Study Oversight

The trial was conceived and designed by the grant applicants, and they wrote the protocol (available at https://stroke.nottingham.ac.uk/enos-2/, last version 4 dated 1 March 2024). The trial was overseen by a Trial Steering Committee (which included 3 independent members) and conducted by the Trial Management Committee based at the coordinating centre in Nottingham. An independent Data Monitoring Committee reviewed unblinded data approximately annually. There was no commercial support for the trial. Study data were monitored and analysed by the ENOS-2 Coordinating Centre. Analysis, interpretation, and report writing were performed independent of the funders and sponsor. The corresponding author wrote the first draft of the manuscript, with input from the writing committee, all of whom approved the decision to submit the manuscript for publication. The corresponding author and statisticians (L.J.W., I.M.) had full access to all the data in the study; additionally, the corresponding author had final responsibility for the decision to submit for publication and is the guarantor for the study.

Statistical Analyses

Since ENOS-2 was a feasibility study, there was no formal sample size calculation. However, recruitment of 100 patients with AIS and 20 with ICH would provide sufficient information on feasibility, safety, and PoC. With a sample size of 100 and assuming alpha = 0.05, power 0.80 and mRS distribution as in ENOS-early [14], an adjusted common odds ratio of 0.51 for mRS would be detectable in PoC analyses.

Statistical analyses of the effect of the treatments on clinical outcomes were planned to be analysed using adjusted binary logistic regression, Cox proportional hazards regression (for time to events, e.g., death), ordinal logistic regression and multiple linear regression, with adjustment for the randomisation factors used in stratification and minimisation [26]. The mRS would be reported as a common odds ratio with 95% confidence intervals and calculated using ordinal logistic regression. For sensitivity purposes, the primary analyses were also to be analysed unadjusted, and as a binary outcome (death or dependency, mRS >2). We would assess the heterogeneity of the treatment effects on the primary outcome in pre-specified subgroups by adding an interaction term to an unadjusted ordinal logistic regression model. Death was to be analysed using Kaplan-Meier and Cox regression models. A global outcome assessing original data from the mRS, BI, TCS-M, EQ-5D-5L, death, and SAEs was to be analysed using the Wei-Lachin test [27]. Since outcomes such as mRS, EQ-5D-HUS, and BI include death (scores of 6, 0, and −5, respectively) and in case treatment was associated with asymmetric effects on death and other outcome measures (e.g., more death and less impairment), an extreme value for death would be added to the other outcome scales: EQ-VAS −1, t-MMSE −1, SSS −1, TICS-M −1, ZDS 102.5, DSRS 13. The nominal level of significance for all analyses was p < 0.05 since no interim analysis were to be performed. No adjustments were to be made for multiplicity of testing. Data would be shown as number (%), median [interquartile range], or mean (standard deviation). As a feasibility trial, no interim analyses or efficacy/futility stopping rules were set.

Due to the paucity of recruited participants, the analyses specified in the SAP were restricted. First, tabulations were abbreviated to avoid reporting an excess number of variables. Second, unadjusted rather than adjusted analyses (using the stratification and minimisation variables) were performed to avoid failure of regression convergence and statistical over-fitting. Third, we used Fisher’s exact test to analyse binary variables where one group had zero/very low numbers of events. Last, analyses in subgroups were not performed. Analyses were performed with SAS software (SAS Institute, Cary, North Carolina USA, version 9.4) according to the intention-to-treat principle.

Results

Study Participants

Between September 2021 and June 2024, 39 patients were included (GTN 21, no GTN 18) at 1 UK site as shown in Figure 1. Two interruptions to recruitment occurred due to lack of investigational drugs (online suppl. Fig. 1; for all online suppl. material, see https://doi.org/10.1159/000550499) and eight protocol amendments were made, as shown in online supplementary Table 1. The trial ceased when the last batch of GTN patches went out of date. A total of 3,314 people were excluded, as shown in online supplementary Table 2.

Fig. 1.

CONSORT diagram, enrolment, and follow-up. Data are number (%). GTN, glyceryl trinitrate.

Seven consultant stroke physicians, two registrars, and two allied health professionals on the trial log recruited the participants, with two of these recruiting more than 50% (see online suppl. Fig. 2). Baseline characteristics were well balanced between the treatment groups at baseline, as shown in Table 1. The mean (standard deviation) age was 72.3 (13.2) years, 16 (41.0%) were females, median [IQR] symptom onset to randomisation was 216 [186–251] minutes (27 (69.2%) recruited within 4 h of onset), and NIHSS score was 9.1 (6.3). The final diagnosis was AIS 33 (84.6%), ICH 1 (2.6%), transient ischaemic stroke 4 (10.3%), and functional mimic 1 (2.6%). Consent was provided by 24 (61.5%) participants and proxy assent by 13 (33.3%) relatives or close friends 8 (20.5%) (shown in online suppl. Table 3); emergency waiver of consent was used in 2 (5.1%) participants. A majority of neuroimaging was normal (21 [53.8%]) at baseline with the remaining scans showing ischaemic (16 [41.0%]) or haemorrhagic (1 [2.6%]) changes (see online suppl. Table 4). Evidence of previous stroke, atrophy, and/or white matter disease was present in a minority of participants. All participants were scanned prior to randomisation.

Table 1.

Baseline characteristics of the patients: glyceryl trinitrate versus no glyceryl trinitrate

Characteristic	N	All	GTN	Sham
Patients, n	​	39	21	18
Age, years1	39	72.3 (13.2)	71.0 (11.5)	73.7 (15.1)
Sex, female, n (%)	39	16 (41.0)	9 (42.9)	7 (39.9)
Ethnicity, n (%)
Asian	39	2 (5.1)	2 (9.5)	0 (0.0)
White	39	37 (94.9)	19 (90.5)	18 (100.0)
Time onset to randomisation, min1	39	216 [186, 251]	218 [192, 264]	196 [186, 245]
<4 h/240 min, n (%)	39	27 (69.2)	14 (66.7)	13 (72.2)
Time onset to scanning, min	39	136 [104, 178]	150 [116, 178]	113 [101, 142]
Candidate for reperfusion1, n (%)	39	22 (56.4)	11 (52.4)	11 (61.1)
Premorbid mRS >0, %	39	0.0 [0.0, 1.0]	0.0 [0.0, 1.0]	1.0 [0.0, 1.0]
Medical history, n (%)
Treated hypertension	39	24 (61.5)	11 (52.4)	13 (72.2)
ACE inhibitor	39	13 (33.3)	4 (19.0)	9 (50.0)
Angiotensin-II receptor antagonist	39	5 (12.8)	3 (14.3)	2 (11.1)
Alpha-blocker	39	2 (5.1)	1 (4.8)	1 (5.6)
Beta-blocker	39	8 (20.5)	4 (19.0)	4 (22.2)
Calcium channel blocker	39	9 (23.1)	4 (19.0)	5 (27.8)
Diuretic	39	4 (10.3)	2 (9.5)	2 (11.1)
Renin	39	1 (2.6)	0 (0.0)	1 (5.6)
Other antihypertensives	39	2 (5.1)	1 (4.8)	1 (5.6)
Stroke	39	10 (25.6)	5 (23.8)	5 (27.8)
Transient ischaemic attack	38	7 (18.4)	3 (15.0)	4 (22.2)
Ischaemic heart disease	39	8 (20.5)	4 (19.0)	4 (22.2)
Diabetes mellitus	39	9 (23.1)	6 (28.6)	3 (16.7)
Hyperlipidaemia	39	15 (38.5)	8 (38.1)	7 (38.9)
Smoking, current, %	37	9 (24.3)	7 (33.3)	2 (11.1)
Alcohol >21 units per week	36	3 (7.7)	2 (9.5)	1 (5.6)
AF, current/previous	39	7 (17.9)	4 (19.0)	3 (16.7)
COVID-19, definite/possible	39	3 (7.7)	2 (9.5)	1 (5.6)
Nitrate use before stroke, n (%)	39	2 (5.1)	2 (9.5)	0 (0.0)
Side of lesion, right, n (%)	32	18 (56.3)	8 (44.4)	10 (71.4)
NIHSS (/42)1	39	9.1 (6.3)	7.6 (4.2)	10.9 (7.9)
GCS <15, n (%)	39	11 (28.2)	5 (23.8)	6 (33.3)
OCSP classification, n (%)
Total anterior	39	11 (28.2)	5 (23.8)	6 (33.3)
Partial anterior	39	18 (46.2)	9 (42.9)	9 (50.0)
Lacunar	39	10 (25.6)	7 (33.3)	3 (16.7)
Posterior	39	0 (0.0)	0 (0.0)	0 (0.0)
Qualifying event2, n (%)
Ischaemic stroke3	39	33 (84.6)	19 (90.5)	14 (77.8)
Cardioembolic	33	12 (36.4)	7 (36.8)	5 (35.7)
Large vessel	33	8 (24.2)	3 (15.8)	5 (35.7)
Small vessel	33	6 (18.2)	4 (21.1)	2 (14.3)
Other	33	9 (27.3)	6 (31.6)	3 (21.4)
Not determined	33	1 (3.0)	0 (0.0)	1 (7.1)
Primary ICH	39	1 (2.6)	1 (4.8)	0 (0.0)
Transient ischaemic attack	39	4 (10.3)	1 (4.8)	3 (16.7)
Stroke mimic	39	1 (2.6)	0 (0.0)	1 (5.6)4

Data are number (%), median [interquartile range], or mean (standard deviation). Percentages exclude missing values from denominators. Unused antihypertensive drug classes are not shown. In a post hoc analysis, the NIHSS did not differ between GTN and sham: −3.3 (95% CI −7.32, 0.72), p = 0.11.

AF, atrial fibrillation; BP, blood pressure; bpm, beats per minute; GCS, Glasgow Coma Scale; ICH, intracerebral haemorrhage; mRS, modified Rankin Scale; NIHSS, National Institutes of Health Stroke Scale; OCSP, Oxfordshire Community Stroke Project classification.

¹Minimisation variable.

²Qualifying event is determined by investigator by discharge.

³Stratification variable.

⁴Functional neurological disorder.

⁵Sum may exceed 100% because of mixed aetiology.

Feasibility

Altogether, 39 (32.5%) patients were recruited out of the intended 120 patients (Table 2). The trial struggled to recruit for multiple reasons, these including factors internal to the trial design and protocol such as a significant number of participants who presented >5 h after onset and those with no eligible symptoms or unknown/unclear time of onset (online suppl. Table 2). External non-protocol-related factors included lack of IMP on two occasions, presentation out-of-hours when no research staff were present and an unwillingness by some medical staff to have the trial offered to their patients due to perceived safety concerns, especially prior to potential mechanical thrombectomy (although no such concerns were identified by the DMC).

Table 2.

Feasibility measures

Measure	Metric	Achieved, n/N (%)
Recruitment
Overall	​	39/120 (32.5)
Target patients	100 with ischaemia	37/100 (37.0)
	20 with ICH	1/20 (5.0)
Primary
Retention of enrollees at day 90 (follow-up)	>70/120 (58.3)	38/39 (97.4)
Treatment compliance/adherence
Received GTN/sham	All 2 days treatment	33/39 (84.6)
	First day treatment	37/39 (94.9)
Co-enrolment	MAPS-2 3	5/39 (12.8)
	OPTIMIST 1	​
	TICH-3 1	​

Adherence

Adherence with GTN versus no GTN was good with 37 (94.9%) participants receiving their first treatment and 33 (84.6%) both patches/dressings as randomised (online suppl. Table 5). There were no differences in the use of secondary antihypertensive drugs over the first 2 days between the two groups (online suppl. Table 6).

Blood Pressure

The mean BP at baseline was 161.8/80.8 mm Hg and fell in both GTN and no GTN groups over the 2 days of treatment (online suppl. Table 7). However, following the first dose of treatment, BP fell more with the first treatment in the sham group than in the GTN group; difference in means 17.2 (95% CI 1.2–33.3)/5.8 (95% CI −1.37 to 12.9) mm Hg with sham versus 13.2 (1.4–24.9)/0.31 (−10.0 to 10.6) mm Hg with GTN. Nevertheless, BP did not differ at day 1 between the groups. The frequency of clinical hypotension (4 cases each, all reported as SAEs) and clinical hypertension (GTN 1, sham 2) did not differ by day 2 (Table 3). Baseline heart rate started at 81.9 bpm. In both groups, heart rate fell then increased over the first day, although the initial fall was less in the control group (p = 0.052) (online suppl. Table 6).

Table 3.

Outcomes at days 2 and 90: glyceryl trinitrate versus no glyceryl trinitrate

Outcome	N	All	GTN	Sham	DIM/HR/MD/PD/OR (95% CI)	2p
Participants	39	39	21	18	​	​
Day 2 (or discharge)
Death, all cause, n (%)	39	1 (2.6)	0 (0.0)	1 (5.6)	−5.56 (−16.14, 5.03)	0.462
Clinical neurological deterioration, n (%)1	39	2 (5.1)	0 (0.0)	2 (11.1)	−11.11 (−25.63, 3.41)	0.212
Recurrent stroke, n (%)1	39	1 (2.6)	1 (4.8)	0 (0.0)	4.76 (−4.35, 13.87)	1.002
Symptomatic ICH, n (%)1	39	1 (2.6)	0 (0.0)	1 (5.6)	−5.56 (−16.14, 5.03)	0.462
Clinical hypotension, n (%)1	39	8 (20.5)	4 (19.0)	4 (22.2)	−3.17 (−28.69, 22.34)	1.002
Clinical hypertension, n (%)1	39	3 (7.7)	1 (4.8)	2 (11.1)	−6.35 (−23.49, 10.79)	0.592
Headache, n (%)1	39	7 (17.9)	6 (28.6)	1 (5.6)	23.02 (0.99, 45.05)	0.0982
NIHSS (/43)	33	7.0 (10.1)	4.1 (4.2)	10.5 (13.7)	−6.36 (−12.79, 0.08)	0.053
Dysphagia severity rating scale (/13)	39	2.4 (4.2)	2.0 (3.7)	3.0 (4.9)	−1.00 (−3.66, 1.66)	0.46
Discharge data
Hospital stay after randomisation, days	39	5.0 [1.0, 13.0]	7.0 [1.0, 20.0]	4.0 [1.0, 10.0]	3.00 (−6.31, 12.31)	0.39
Death or not discharged home, n (%)	39	7 (17.9)	2 (9.5)	5 (27.8)	0.27 (0.05, 1.63)	0.15
Day 90
modified Rankin Scale (/6)	38	2.9 (1.8)	2.8 (1.7)	3.0 (1.8)	−0.20 (−1.30, 0.90)	0.72
Death, n (%)	39	5 (12.8)	1 (4.8)	4 (22.2)	0.21 (0.02, 1.92)	0.17
Barthel Index (/100)	38	72.0 (38.8)	71.8 (35.8)	72.2 (43.0)	−0.47 (−24.88, 23.93)	0.97
t-MMSE score (/22)	31	16.1 (7.8)	18.2 (5.7)	14.2 (9.2)	4.01 (−1.23, 9.25)	0.13
TICS-M score (/39)	31	20.2 (10.6)	24.3 (8.5)	16.3 (11.0)	8.02 (1.27, 14.78)	0.020
Verbal fluency (animal naming)	32	14.9 (9.6)	18.1 (8.5)	12.1 (9.8)	6.01 (−0.20, 12.21)	0.058
EQ-5D-5L	38	0.5 (0.3)	0.5 (0.4)	0.5 (0.3)	0.02 (−0.20, 0.24)	0.85
EQ-visual analogue scale	37	56.6 (29.5)	62.2 (26.7)	50.8 (31.9)	11.43 (−6.96, 29.82)	0.22
Zung depression scale	32	57.3 (23.3)	50.8 (20.6)	62.9 (24.7)	−12.11 (−27.47, 3.26)	0.12
Participants with ≥1 SAE, n (%)	39	16 (41.0)	8 (38.1)	8 (44.4)	0.77 (0.21, 2.77)	0.69
Global outcome	31	​	​	​	−0.12 (−0.38, 0.14)	0.37

Data are number of patients (%), median [interquartile range], or mean (standard deviation) score. Comparisons were assessed by Fisher’s exact test, Mann-Whitney U test, or unadjusted Cox proportional hazards, logistic regression, linear regression, shown as odds ratio (OR, with 95% confidence intervals), hazard ratio (HR), difference in medians (DIM), percentage difference (PD), or mean difference (MD). National Institutes Health stroke scale (NIHSS): 0 (normal neurological status) to 42 (coma with quadriplegia), 43 (death) [1]. Dysphagia severity rating scale: 0 (normal oral intake), 12 (nil by mouth), 13 (death) [2]. Barthel Index: −5 (death) to 0 (severe disability) to 100 (no disability) [3]. Modified telephone Mini-Mental State Examination (t-MMSE): −1 (death), 0 (severe dementia) to 18 (normal). Modified Telephone Interview for Cognitive Status (TICS-M): −1 (death, 0 (severe dementia) to 37 (normal) [4]. Verbal fluency (number of animals named in 1 min): −1 (death), 0 (none named) to infinity.

Health utility status (HUS, derived from European Quality of Life-5 dimensions, EQ-5D): −0.5 (very poor quality of life, 0 (death) to 1.0 (perfect quality of life). European Quality of Life-Visual Analogue Scale (EQ-VAS): −1 (death), 0 (very poor) to 100 (excellent). Zung Depression Scale (ZDS): 0 (normal), 100 (severe depression) to 102.5 (death) [5]. Global outcome: continuous – mRS, BI, TICS-M, EQ-5D; binary – death, SAEs.

BI, Barthel Index; DIM, difference in medians; EQ, EuroQol; HR, hazard ratio; ICH, intracranial haemorrhage; MD, mean difference; mRS, modified Rankin Scale; NIHSS, National Institutes of Health stroke scale; Neuro. det., neurological deterioration; OR, odds ratio; t-MMSE, modified telephone Mini-Mental State Examination; SAE, serious adverse event; TICS-M, Modified Telephone Interview for Cognitive Status; VAS, Visual Analogue Scale; ZDS, Zung Depression Scale.

¹Clinical events determined by the investigator: recurrent stroke, sICH, hypotension, hypertension.

²Percentage difference and Fisher’s exact test.

Clinical Outcomes

There were no significant differences in death, clinical neurological deterioration, recurrent stroke, symptomatic haemorrhage, or swallowing impairment at day 2 (Table 3). NIHSS at day 2 was nonsignificantly lower GTN 4.1 (4.2) versus sham 10.5 (13.7), p = 0.053. Hospital length of stay and death in hospital or not discharged home did not differ between GTN and sham.

By day 90, there were no statistical differences in dependence (mRS), death, disability (BI), quality of life (EQ-5D-5L, EQ-VAS), or mood (ZDS), as shown in Figure 2 and Table 3. Nevertheless, less cognitive impairment, measured using the TICS-M (p = 0.020) and verbal fluency (animal naming) (p = 0.058), was seen with GTN.

Fig. 2.

Distribution of modified Rankin Scale scores at 90 days: glyceryl trinitrate versus sham. Comparison of GTN versus sham by unadjusted multiple linear regression: difference in means −0.20 (−1.30, 0.90) p = 0.72.

Serious Adverse Events

Although the number of participants with one or more SAEs did not differ with eight in each group (Table 3), there were less SAEs in the GTN group (9) than the sham group (15) (Table 4). The majority occurred during treatment and four in each group were considered to be possibly, probably, or definitely related to treatment. No fatal SAEs were considered to be associated with treatment. One death occurred in the GTN group and was related to sepsis; the sham group had three deaths related to the nervous system and one to the respiratory system (Table 4). The CONSORT checklist is given in the online supplementary material.

Table 4.

Serious adverse events by treatment group and relationship

​	All	All			p value	All	Fatal			p value
		GTN	sham	OR (95% CI)			GTN	sham	OR (95% CI)
Participants	39	21	18	​	​	39	21	18	​	​
Number of SAEs	24	9	15	–	0.020	5	1	4	​	​
Treatment relationship
Before	0 (0.0)	0 (0.0)	0 (0.0)	–	–	0 (0.0)	0 (0.0)	0 (0.0)	–	–
During	22 (91.7)	8 (88.9)	14 (93.3)	–	0.71	3 (60.0)	0 (0.0)	3 (75.0)	–	–
After	2 (8.3)	1 (11.1)	1 (6.7)	–	0.71	2 (40.0)	1 (100.0)	1 (25.0)	–	1.00
Relationship to treatment	​	​	​	​	0.59	​	​	​	​	–
Not related	2 (8.3)	1 (11.1)	1 (6.7)	–	–	0 (0.0)	0 (0.0)	0 (0.0)	–	–
Improbable	14 (58.3)	4 (44.4)	10 (66.7)	–	–	5 (100.0)	1 (100.0)	4 (100.0)	–	–
Possible	6 (25.0)	3 (33.3)	3 (20.0)	–	–	0 (0.0)	0 (0.0)	0 (0.0)	–	–
Probable	1 (4.2)	0 (0.0)	1 (6.7)	–	–	0 (0.0)	0 (0.0)	0 (0.0)	–	–
Definite	1 (4.2)	1 (11.1)	0 (0.0)	–	–	0 (0.0)	0 (0.0)	0 (0.0)	–	–
Organ
Cardiovascular	8 (20.5)	4 (19.0)	4 (22.2)	0.82 (0.17, 3.90)	0.81	0 (0.0)	0 (0.0)	0 (0.0)	–	–
Nervous system	4 (10.3)	1 (4.8)	3 (16.7)	0.25 (0.02, 2.65)	0.25	3 (7.7)	0 (0.0)	3 (16.7)	–	–
Respiratory	5 (12.8)	1 (4.8)	4 (22.2)	0.18 (0.02, 1.74)	0.14	1 (2.6)	0 (0.0)	1 (5.6)	–	–
Gastrointestinal	2 (5.1)	1 (4.8)	1 (5.6)	0.85 (0.05, 14.64)	0.91	0 (0.0)	0 (0.0)	0 (0.0)	–	–
Genitourinary	1 (2.6)	0 (0.0)	1 (5.6)	–	–	0 (0.0)	0 (0.0)	0 (0.0)	–	–
Haematological/immunological	0 (0.0)	0 (0.0)	0 (0.0)	–	–	0 (0.0)	0 (0.0)	0 (0.0)	–	–
Metabolic/endocrine	0 (0.0)	0 (0.0)	0 (0.0)	–	–	0 (0.0)	0 (0.0)	0 (0.0)	–	–
Musculoskeletal/cutaneous	1 (2.6)	0 (0.0)	1 (5.6)	–	–	0 (0.0)	0 (0.0)	0 (0.0)	–	–
Miscellaneous	3 (7.7)	2 (9.5)	1 (5.6)	1.79 (0.15, 21.54)	0.65	1 (2.6)	1 (4.8)	0 (0.0)	–	–
Any SAE	16 (41.0)	8 (38.1)	8 (44.4)	0.77 (0.21, 2.77)	0.69	5 (12.8)	1 (4.8)	4 (22.2)	0.18 (0.02, 1.74)	0.14

Data are number (%). Comparisons were assessed by binary logistic regression. Miscellaneous SAEs: tongue haematoma, coffee ground vomiting, fall.

Discussion

ENOS-2 assessed the effect of GTN vs sham in patients with acute stroke within a narrow time window defined by apparent safety concerns with GTN when given before 2 h of stroke onset [16, 17, 28] and futility beyond 6 h [13, 15]. The trial found that feasibility was challenged by multiple recruitment causes, some internal to the trial design and some external, such that only 39 of 120 planned participants were recruited. However, GTN appeared safe with tendencies to fewer deaths and SAEs and less cognitive impairment.

External factors potentially relate to issues at a single centre with lack of equipoise by some clinicians. Hence, it cannot necessarily be concluded that a further study is not warranted, particularly if future testing was done across multiple sites. A multicentre design would have been more robust but was not feasible due to the limited funding available. Such testing remains desirable since the present data are still compatible with the positive findings seen for treatment in the range 3–5 h after onset, as shown in online supplementary Figure 2 [12–15].

The findings that GTN did not lower BP are surprising for a vasodilator and not easily explained since all previous GTN trials have found that it lowered BP; chance due to the limited sample size may explain this, i.e., type II error. Nevertheless, GTN did raise heart rate as would be expected for a mixed venous and arterial dilator. And similarly, GTN increased the rate of headache as anticipated; interestingly, the development of headache may be a positive prognostic factor as seen acutely in stroke with GTN and with chronically with dipyridamole, a modulator of nitric oxide used for secondary stroke prevention [29, 30].

The present trial has several strengths. First, it was randomised in design and used a sham treatment in contrast to earlier GTN trials. Second, the trial was a pragmatic test of real-world approaches to BP management since recruitment could be started in the Emergency Department prior to neuroimaging. Third, participants could have stroke of varying characteristics: ischaemic or haemorrhagic; severe or mild; cortical or lacunar; and anterior or posterior circulation. Fourth, outcomes were measured centrally masked to treatment assignment by observers who were not involved with clinical care. Last, transdermal administration of a drug in acute stroke is advantageous since it can be started immediately, is not confounded by dysphagia which limits the immediate use of oral treatment, does not need intravenous access and so does not need intensive monitoring, and can be stopped and restarted according to clinical need.

Nevertheless, there are several limitations present. First, the final trial size was much smaller at 39 than the intended 120 participants. Some of the explanations are unrelated to the protocol and might be resolved in a multicentre rather than single-centre design. A site concern was that GTN might be unsafe in patients proceeding to mechanical thrombectomy; this could not be assessed due to the limited number (5) of these procedures (online suppl. Table 6). Second, the trial recruited from one site only and so is not generalisable to other sites or countries. Trial sites vary in their ability to recruit and so having multiple participating sites can average out recruitment limitations at any one site. Third, a key limitation was the narrow time window of 3–5 h after stroke onset, although this was deliberate on the basis of previous data; potentially, any future trial could extend this to, say, 2.5–5.5 h. Fourth, some patients assigned to GTN did not receive it for the protocolised 2 days of treatment. And fifth, some outcomes (cognition, mood) could not be measured in all participants at day 90, further limiting the power of some statistical analyses.

In conclusion, feasibility was not demonstrated in this small trial and, surprisingly, GTN did not appear to lower BP. Nevertheless, GTN was safe, SAEs were reduced, and point estimates for treatment effects were all in a direction supporting benefit. A meta-analysis of the three trials that have examined the effect of GTN on mRS (ENOS-1/2, RIGHT-2) in the time window of 3–5 h suggests efficacy: mean difference −0.45 (95% CI −0.91, 0.02), p = 0.06 with no heterogeneity, see online supplementary Figure 3. Since the results support, potentially, the use of GTN in hyperacute stroke, a multicentre trial testing feasibility of GTN in the time window of 3–5 h is now warranted.

Acknowledgments

We thank the research staff in Nottingham for their support and the Nottingham University Hospitals NHS Trust Charity which was the primary funder. We also thank Professor Pip Logan for access to her NIHR Senior Investigator award to facilitate NIHR eligibility and access to research coordinators. ENOS-2 acknowledges the support of the UK National Institute for Health and Care Research (NIHR) for support through the clinical research network. Declaration of generative AI and AI-assisted technologies in the writing process: No AI was used in the analysis of the study or preparation of this report.

Statement of Ethics

This study protocol was reviewed and approved by the Northwest – Greater Manchester South Research Ethics Committee/UK Health Research Authority (20/NW/0246, 30 June 2020), the Medicines & Healthcare products Regulatory Agency (CTA 03057/0072/001-0001, 24 August 2020) and was adopted by the UK NIHR Clinical Research Network. Written informed consent was obtained from each patient, or from a relative or close friend if the patient lacked capacity, prior to enrolment and in accordance with UK regulations.

Conflict of Interest Statement

P.M.B. is Stroke Association Professor of Stroke Medicine and is an Emeritus NIHR Senior Investigator; he was chief investigator of ENOS, RIGHT, and RIGHT-2. He has received honoraria from CoMind, DiaMedica, and Phagenesis and holds stock options in DiaMedica. The remaining authors have no declarations.

Funding Sources

The trial was funded by a grant from the Nottingham University Hospitals Charity (CRF-BATH-NOV 2019). The trial was sponsored by the University of Nottingham and registered: ISRCTN17654248 (31 March 2021), EudraCT 2020-001304, IRAS 281728, WHO U1111-1259-7057.

Author Contributions

Conceptualisation and funding acquisition: Philip M. Bath, Nikola Sprigg, Timothy J. England, Lisa J. Woodhouse, and Kailash Krishnan. Methodology – study coordinators: Amanda Buck, Lucy Gray, Amanda Hedstrom, and Camille Hutchinson. Analysis: Lisa J. Woodhouse and Iris Mhlanga. Writing – original draft preparation: Philip M. Bath. Writing – review and editing: Philip M. Bath, Nikola Sprigg, Timothy J. England, Lisa J. Woodhouse, Kailash Krishnan, Iris Mhlanga, Tiffany Hamilton, and Diane Havard. Resources: Tiffany Hamilton and Diane Havard. Supervision: Philip M. Bath and Kailash Krishnan.

Funding Statement

The trial was funded by a grant from the Nottingham University Hospitals Charity (CRF-BATH-NOV 2019). The trial was sponsored by the University of Nottingham and registered: ISRCTN17654248 (31 March 2021), EudraCT 2020-001304, IRAS 281728, WHO U1111-1259-7057.

Data Availability Statement

The dataset generated during the current study are not currently publicly available due their integration into an update of our individual patient data meta-analysis of GTN. They will be shared with the VISTA Collaboration [31]: https://www.virtualtrialsarchives.org. The CI, with approval from the TSC as necessary, will consider all reasonable requests to share individual participant data on provision of a protocol detailing aims, hypotheses, analyses, tables, figures and publication plan. Where possible, we will perform the analyses; alternatively, de-identified data and a data dictionary will be provided for remote analyses, subject to signed data access agreement. Further enquiries may be directed to the corresponding author.