Prehospital Transdermal Glyceryl Trinitrate for Ultra-Acute Intracerebral Hemorrhage

Supplemental Digital Content is available in the text.


SUPPLEMENTAL METHODS (additional information)
Study design and study population • No ambulances had a CT scanner to provide immediate diagnosis before hospital.
• The study was approved by the UK regulator (Medicines and Healthcare products Regulatory Agency, reference: 03057/0064/001-0001; Eudract 2015-000115-40) and national research ethics committee (IRAS: 167115) and was adopted by the National Institute for Health Research Clinical Research Network.

Randomisation and consent
• Randomisation was stratified by ambulance station with blocks of 4 packs (2 active, 2 control) in random permuted order. • Paramedics managed the primary consent process, and patients with capacity gave written informed consent that covered the whole trial. If capacity was absent, proxy consent was obtained from an accompanying relative, carer or friend if present, or from the paramedic if no accompanying person was present. Participants who regained capacity were then given the opportunity to consent in hospital. • The final diagnosis was made after arrival at hospital by the principal investigator based on clinical and neuroimaging findings and was categorised as ICH, ischaemic stroke, TIA, or non-stroke/TIA mimic.

Outcome measures
• The primary analysis involved a comparison of the distribution of all 7 levels of the mRS ('shift analysis') between the treatment groups. • Neurological deterioration was defined as an increase in the National Institutes of Health stroke scale, NIHSS ≥4 points from hospital admission to day 4 or worsening conscious level in the NIHSS consciousness domain item Ia.

Neuroimaging outcomes (additional information)
• Imaging outcomes on admission to hospital included the presence of pre-stroke features (atrophy, periventricular lucencies, old strokes, frailty score and small vessel disease score), haematoma location, size (categorised maximum diameter), volume, 1 extension (to subarachnoid spaces or ventricles), mass effect including midline shift (assessed on a validated 7-point scale), 2 and qualitative measures suggestive of active acute bleeding (e.g. hypodensity sign, island sign and swirl sign, 3 and spot sign if CT angiography was performed 4 ). • A further CT or MR scan was performed on day 2 to assess safety; the same factors were assessed blind to treatment assignment and baseline scan findings.

Sample size
• We estimated that a total sample size of 850 participants (425 in each arm) was required to detect a shift in mRS with common odds ratio 0.70, 5 significance level 5%, power 90%, assumed distribution of mRS scores as shown in the appendix of the main publication, 6 loss to follow-up 3%, joint mimic and TIA rate of 20%, and reduction for baseline co-variate adjustment 20%. 7 We assumed that approximately 15% of participants would have a final diagnosis of ICH. 8 • The sample size was determined for the whole trial and not for this or any other subgroup of patients. • During the trial, the non-stroke diagnosis rate was found to exceed 30%, and the sample size was increased from 850 to 1050 to maintain statistical power.

Statistical analysis (additional information)
• A post hoc global analysis based on dichotomous outcomes (mRS >2, BI <30, TICS-M <5, ZDS >80, HSUV <0.2) was performed as in the NINDS alteplase trial. 9 • No adjustments are made for multiplicity of testing since all secondary analyses were hypothesis-generating and designed to support the primary analysis. • Data are shown as number (%), median [interquartile range, IQR], mean (standard deviation, SD), and odds ratio, hazard ratio, difference in means or Mann-Whitney difference (test of global outcome) with 95% confidence intervals (CI). • Analyses were done as randomised in participants with ICH and with observed outcome data, and performed with SAS software (version 9.3).

Data sharing
• Individual participant data will be shared with the Blood pressure in Acute Stroke Collaboration (BASC) and Virtual International Stroke Trials Archive (VISTA). • From Jan 1, 2021, the Chief Investigator (with approval from the Trial Steering Committee as necessary) will consider other requests to share individual participant data via email at: right-2@nottingham.ac.uk. We will require a protocol detailing hypothesis, aims, analyses, and intended tables and figures. Where possible, we will perform the analyses; alternatively, de-identified data and a data dictionary will be supplied for the necessary variables for remote analysis. Any sharing will be subject to a signed data access agreement. • Ultimately, the entire trial dataset will be published.

Demographics
• 7 (5%), 52 (36%) and 114 (79%) of participants received study drug within 30, 60 and 120 minutes of symptom onset, respectively (Table 1). • Although adherence was excellent with 100% of participants receiving the first randomised treatment, adherence on days 2, 3 and 4 was considerably lower at 61%, 59% and 55% respectively (Supplemental Table I). • At baseline in the ambulance, blood pressure was mean 176 (SD 27)/100 (SD 22) mmHg (Table 1) and fell in both treatment groups over the four days after randomisation (Supplemental Figure I). In hospital management and treatment • A majority of participants (69%) received open-label antihypertensive therapy between admission and day 4, typically between 4 and 5 hours after randomisation (Supplemental Table II). This included a variety of antihypertensive drug classes with the most common being ß-receptor antagonists, calcium channel blockers and nitrates. The use of these did not differ between GTN and sham treatment groups.

Neuroimaging findings
• The presence of pre-stroke imaging features did not differ between the treatment groups (Supplemental Table III). • A majority of haematoma were deep, and location did not differ between the treatment groups. • In an analysis suggested by a reviewer, the effect of GTN vs sham on perihaematomal oedema remained significantly increased at day 2 with the addition of haematoma size to the primary analysis model (mean difference 6.87, 95% CI 0.26, 13.48; p=0.041); • There was no interaction between treatment and prior antithrombotic therapy on haematoma size (MD 0.71, 95% CI -0.22, 1.65) • Haematoma were one-to-three fold larger than in previous ICH trials of pharmacological agents, and approached the size studied in surgical trials (Supplemental Table IV). • Few CT angiograms were performed, and a positive spot sign was infrequent (12%); this rate did not differ between GTN and sham.

Relationship between imaging variables and outcome
• Although none of the pre-stroke imaging features were associated with haematoma expansion by day 2-4, or death or mRS at day 90 (Supplemental Table VI), the presence of periventricular lucencies, and higher frailty and SVD scores, appeared to either modify or mediate the effect of GTN on outcome. • Multiple haematoma characteristics at hospital admission were associated with haematoma expansion by day 2-4, and death or mRS at day 90, including the maximum length (Supplemental Figure VII) and volume, and presence of irregular shape and density, island sign, hypodensity sign, swirl sign, intraventricular haemorrhage, mass effect, perihaematomal oedema, midline shift, and hydrocephalus (Supplemental Table V). Of these, haematoma shape, intraventricular volume, presence of hydrocephalus appeared to either modify or mediate the effect of GTN on haematoma expansion or outcome.

Meta-analysis of completed trials
• Addition of the results for participants with ICH in RIGHT-2 to those in RIGHT and ENOS-early resulted in neutral effects for end-of-trial death or dependency (mRS>2), OR 0.98 (95% CI 0.25-3.86) and death, OR 0.55 (95% CI 0.12-2.47) (Supplemental Figure VIII). • Overall, there was no difference between GTN and sham when data were analysed by time from onset to randomisation for 0-2 hours and 2-6 hours for either outcome although both death or dependency (mRS>2) and death at day 90 were increased in RIGHT-2 participants randomised within two hours of ICH onset (Supplemental Figure VIII). • Heterogeneity was present for both outcomes (death or dependency I 2 53%; death I 2 70%) and the point estimates were in opposite directions such that very early treatment (0-2 hours) with GTN might be hazardous whilst slightly later treatment (2-6 hours) might be beneficial (Supplemental Figure VIII).

SUPPLEMENTAL DISCUSSION
• Different mechanisms of action may explain the different effects on outcome throughout the ultra-acute, hyper-acute and acute phases of stroke. For example, vasodilators may have negative, neutral or positive effects depending on when they are started over the first 48 hours after ICH, as seen in recent BP-lowering trials (Supplemental Table VI). • The negative effect apparent in the first 2 hours after ICH (as seen here in RIGHT-2) may reflect attenuation of the first phase of haemostasis based on rapid but transient vasoconstriction; 10 mechanisms for driving vasoconstriction involve: o reduced nitric oxide and prostacyclin from damaged endothelium; o thromboxane A2 liberated from activated platelets; o nervous reflexes from "pain"; o direct injury to vascular smooth muscles. Hence, administration of GTN would oppose reduced endothelial NO synthesis and so attenuate the vasoconstrictory response. • Additionally, most vasodilators have mild antiplatelet effects, as may be true for GTN, 11 and so prevent the second phase of haemostasis based on plateletplugging. • The combination of less vasoconstriction and platelet plugging will lead to increased bleeding. • The interaction of haematoma irregular shape on the effect of GTN on haematoma expansion and death supports this interpretation since irregular finger-like projections are also a marker of cerebral amyloid angiopathy-associated ICH which is a condition in which micro-vessels have reduced vasomotor activity. 12 • The subsequent positive effect of vasodilators between 2 and 6 hours as seen in INTERACT-2 and GTN-early-ICH 13,14 presumably reflects the benefit of lowering BP and so limiting haematoma expansion. • If treatment is started later than about 6 hours then there is no benefit since haematoma expansion has largely finished. 15 • If this pattern of effect is correct, then GTN, and potentially other vasodilators, should not be given to lower BP very early, say within 2 hours, after ICH. This may have major implications for the very early management of high BP in ICH. • It is interesting to note a similar time-dependent effect for another intervention, namely physical therapy for stroke with hazard if started acutely, 16 benefit if started in the sub-acute period, and then no benefit if started weeks or months later.
This subgroup report highlights several findings that may inform future trial design in ICH. 1. A simple measure of haematoma size, namely maximum length, is both prognostic for functional outcome and sensitive to a treatment effect. This measure could also be easily used in clinical practice. 2. It could be argued that patients with large ICH (>5 cm in maximum length) should be excluded from drug-based trials in view of their very poor outcome although how this would be identified in the majority of ambulances without a CT scanner remains unclear at present. 3. Other haematoma characteristics such as shape may be sensitive measures of treatment effect. 4. Pre-hospital treatment can alter hospital treatment activity such as critical care and therapy utilisation; these measures should become secondary outcomes in future ambulance-based trials. 5. A global outcome (based on ordinal or continuous data and analysed using the Wei-Lachin test) may be more sensitive to treatment effects and avoid the need to identify a single outcome measure; whether such global analysis should comprise the primary outcome of trials needs further testing and discussion. 6. It is questionable whether GCS and FAST offer sufficient granularity in quantifying severity and acting as covariates in adjusted statistical analyses; we note that several alternative pre-hospital scoring systems for severity are in development.

CONTRIBUTORS
• The trial was conceived and designed by the grant applicants, and they wrote the protocol. • The trial was overseen by a Trial Steering Committee (which included three independent members and a patient-public representative), and advice was given by an International Advisory Committee. • The day-to-day conduct of the trial was run by a Trial Management Committee, which was based at the Stroke Trials Unit in Nottingham, UK. Study data were collected and quality-assured by the RIGHT-2 Coordinating Centre in Nottingham. • Analysis, interpretation, and report writing were performed independently of the funder and sponsor. • PMB was chief investigator, a grant applicant, participated in the Steering Committee, collected, verified, and analysed data and drafted this report, and is project guarantor. • LJW was trial statistician, involved in the design of the trial, participated in the Steering Committee, and analysed data. • KK performed brain scan measurements in participants with intracerebral haemorrhage. • JPA was the trial physician supporting the chief investigator and trial delivery.
• CSA was an international adviser who provided guidance on trial delivery and interpretation. • EB was an international adviser who provided guidance on trial delivery and interpretation. • LC adjudicated brain scans.
• MD was the national paramedic lead coordinating ambulance service trial delivery.
• TJE was a grant applicant, participated in the Steering Committee, and advised on trial delivery. • PJG was statistician to the Data Monitoring Committee.
• TH wrote the approval documents and information sheets and provided statistical advice. • GM adjudicated brain scans.
• AAM was a statistician, grant applicant, and participated in the Steering Committee. • SJP was an international adviser who provided guidance on trial delivery and interpretation. • JP was a grant applicant, participated in the Steering Committee, and advised on trial delivery. • CIP was a grant applicant, participated in the Steering Committee, and advised on trial delivery by ambulance services. • MR adjudicated serious adverse events.
• TGR was a grant applicant, participated in the Steering Committee, and advised on trial delivery. • CR was a grant applicant, participated in the Steering Committee, and advised on trial delivery. • PMR was a national adviser who provided guidance on trial delivery and interpretation. • ECS was an international adviser who provided guidance on trial delivery and interpretation. • NSa was an international adviser who provided guidance on ambulance trial delivery and interpretation. • JLS was an international adviser who provided guidance on ambulance trial delivery and interpretation.
• ANS was a grant applicant, participated in the Steering Committee, and advised on trial delivery by ambulance services. • GV was independent chair of the Steering Committee.
• JMW was a grant applicant, participated in the Steering Committee, and led adjudication of brain scans. • NSp was deputy chief investigator, a grant applicant, and participated in the Steering Committee. • The corresponding author wrote the first draft of the manuscript, and this was edited and commented on by the Writing Committee. All members of the Writing Committee commented on the analyses and drafts of this report and have seen and approved the final version of the report. Title  Page  Tables  I  Adherence and reasons