Transdermal Glyceryl Trinitrate Lowers Blood Pressure and Maintains Cerebral Blood Flow in Recent Stroke

Methods

We performed a prospective, single-center, patient- and measurement-blinded, randomized, controlled trial. Approval was obtained from the local research ethics committee (EC01/69) and Medicines Control Agency (DDX MF8000/13187). Patients gave written informed consent to participate; assent was obtained from the next of kin if the patient was unable to give consent. Enrollment into the study was made before neuroimaging or estimation of CBF. The study conformed to the Declaration of Helsinki and International Conference on Harmonisation guidelines. Enrollment took place between September 2002 and September 2004.

Eighteen previously independent (modified Rankin Scale, 0 to 2) adult patients with a clinical stroke syndrome and limb weakness (Scandinavian Stroke Scale, arm and/or leg <6) were recruited within 5 days of ictus (Figure 1). All of the subjects had an elevated systolic BP (140 to 220 mm Hg) at enrollment. Subjects were excluded if they had a requirement for, or a contraindication to, nitrate therapy; had a definite need for previous antihypertensive therapy or vasoactive drugs; or could not cooperate with scanning.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Trial flow profile. Withdrawal because of intolerance of xenon CT procedure.

Subjects were randomly assigned using computerized minimization (on age, sex, baseline systolic BP, baseline Scandinavian Stroke Scale, hours from onset, and the presence of a visible stroke lesion on computed tomography [CT]) to receive either a 5-mg GTN patch (Transiderm-Nitro 5, Novartis Pharmaceuticals) or control in a ratio of 2:1. Treatment was given once daily for 7 days. Patients and the assessor who performed hemodynamic and xenon CT measurements (M.W.) were blinded to treatment by placement of a large gauze dressing over the patch or empty skin to conceal treatment status. GTN patches were changed at 8:00 am each day and kept on for a full 24 hours. Any previous antihypertensive medication was discontinued at the time of admission, as is routine at our institution.

Quantitative regional CBF was measured using the stable xenon CT method (Diversified Diagnostic Products XeCT system 2).¹⁶ A baseline CT head scan was performed to confirm the diagnosis of stroke (ischemia, hemorrhage) and to obtain a “scout” image; subsequent scans were composed of 4 adjacent 10-mm–thick slices chosen to encompass the maximum axial dimensions of the stroke lesion. If no lesion was visible, slices included the basal ganglia and internal capsule. The patient was connected to the XeCT system via a face mask and breathed room air for 30 seconds while 2 baseline sets of images were recorded. Subsequently, the patient breathed a mix of xenon (28%) and oxygen (25%) with monitoring of end tidal xenon concentration. Additional sets of images were then collected over 5 minutes. A second xenon CT scan series was performed 1 hour after administration of GTN or control; careful positioning of the patient ensured that near-identical brain slices as determined by neuroradiological landmarks were imaged in pre-GTN and post-GTN scans.

CBF was calculated on a PC using XeCT software in an identical manner for pre-GTN and post-GTN scans. Images with excessive movement artifact or where an old stroke lesion was present were discarded. Analyses (by M.W.) were blinded to treatment and concentrated on the slice with the largest visible area of stroke lesion. If no lesion was visible, an appropriate level was selected according to the clinical presentation. Global and hemispheric regions of interest (ROIs) were sited using a rectangular-shaped template, the former touching the inside of the skull anteroposteriorly and laterally and the latter dividing the skull at the midline. Anterior, middle, and posterior cerebral artery territory ROIs were placed over the cortex using a template generated by the XeCT software (Figure 2a).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. (A) Example of template for anterior cerebral artery (areas 1 and 6), middle cerebral artery (2 and 5), and posterior cerebral artery (3 and 4) cortical ROIs in a patient with a large right hemisphere infarct. (B) Measurement of cerebral blood flow in an ROI encompassing an area of oligaemia (20 to 36 mL/min per 100 g) in a patient with a left hemisphere infarct. Similar CBF filters were used to identify lesion core (CBF<10 mL/min per 100 g) and penumbra (CBF 10 to 20 mL/min per 100 g).

An additional pixel-based analysis was used to assess the effect of GTN on CBF in the stroke lesion, if visible. A rectangular ROI was placed to cover the whole of the stroke lesion and surrounding brain tissue (Figure 2b). A CBF filter was then used to determine the number of pixels within the ROI matching prespecified CBF values for “core” (<10 mL/min per 100 g), “penumbra” (10 to 19 mL/min per 100 g), and “oligemia” (20 to 36 mL/min per 100 g).^17–20 Matching ROIs were sited on pretreatment and posttreatment scans to ensure consistency, and the areas (in pixels) of reduced CBF were compared.

Middle cerebral artery blood velocity was determined bilaterally by transcranial Doppler (Nicolet EME Companion, Kleimoftheim) with the transtemporal window accessed at depths from 30 to 60 mm.²¹ Duplicate mean, systolic, and diastolic velocities and pulsatility index were measured separately for affected and unaffected hemispheres.^10,11

BP was measured immediately before the baseline xenon CT scan and immediately after the posttreatment scan. Duplicate measurements of peripheral systolic and diastolic BP were measured in the nonhemiparetic arm with a validated digital readout oscillometric device (Omron HEM-705CP, Omron Corp).²² Central BP was assessed by applanation tonometry of the left radial artery and using the Pulse Wave Analysis system (Sphygmocor).¹¹ Duplicate recordings were taken, composed of a screen of data satisfying the quality control criteria of the Pulse Wave Analysis system (pulse height and diastolic variability ≤10%). The recorded radial artery pressure wave was transformed to the corresponding central wave using a validated transfer function and central BP derived automatically.

CPP and ZFP were estimated noninvasively from measures of middle cerebral artery blood flow velocity (FV), assessed using transcranial Doppler, and peripheral BP using the following method^9,23: equation equation

The study assumed that a between-group difference in CBF of 10 mL/min per 100 g (SD, 6) would be of clinical relevance. We calculated a sample size of 18, assuming significance 0.05, power 0.80, and 2:1 randomization, and accepted that this would lead to results with large CIs. Data were entered and analyzed by intention-to-treat using SPSS (Apple Mac version 11.0.2, SPSS Inc). Mean (SD), median (interquartile range) or frequencies (%) are given. All of the comparisons of hemodynamic and BP data between the treatment groups were analyzed with adjustment for baseline values using ANCOVA. Linear regression was used to assess the relationship between on-treatment CBF and BP with adjustment for baseline values. Significance was set at P<0.05.

Results

Subject characteristics were balanced between the 2 groups, except that more patients randomly assigned to GTN were female or had a cortical-based stroke on CT scan (Table 1). Most patients had a mild–moderate stroke severity reflecting the fact that only one third had a middle cerebral artery cortical lesion visible on CT scan; the remaining patients had a lacunar stroke. The time between stroke onset and treatment was similar between the 2 groups [median (range), control 77 (22–92) hours; GTN, 79 (28–116) hours]. Peripheral BP was elevated at baseline and higher than central BP. No patients had significant carotid stenosis. Although the mean areas of oligemia, penumbra, and core were similar in size between the 2 groups, baseline measures of global, hemispheric, and regional CBF, both ipsilateral and contralateral to the lesion, and ZFP were nonsignificantly higher in patients randomly assigned to GTN (Table 2). Patients randomly assigned to control tended to have higher CPP.

GTN lowered peripheral and central systolic BP by 23 mm Hg (14%) and 22 mm Hg (13%), respectively (P=0.034; P=0.048). Nonsignificant reductions in peripheral and central diastolic BP were present, at 4 mm Hg (3%) for each (P=0.47; P=0.55). GTN did not alter heart rate. In contrast to BP, GTN did not alter any measure of global, hemispheric, or regional CBF whether on the side ipsilateral or contralateral to the lesion (Table 3 and Figures 2a, 3, and 4⇓⇓). Nevertheless, the CIs were wide, and GTN might have reduced ipsilateral hemispheric CBF by 7.6 mL/min per 100 g or increased it by 4.9 mL/min per 100 g. When defined by blood flow levels, GTN had no effect on the size of presumptive core, penumbra, or oligemic areas (Figure 2b). Similarly, GTN did not significantly alter estimates of CPP and ZFP (Table 3). There was no association between on-treatment measures of ipsilateral CBF and systolic BP (P=0.83, with adjustment for baseline CBF and systolic BP) in patients randomized to GTN.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3. (A) Patient with a left hemisphere borderzone infarct (bottom left); cerebral blood flow before (top left) and 1 hour after GTN (top right); cerebral blood flow has increased with GTN in and around the hypodense lesion, and no cerebral steal is present, as shown by the difference between prescans and postscans (bottom right). (B) Patient with a left intracranial hemorrhage (bottom left); cerebral blood flow before (top left) and 1 hour after GTN (top right); cerebral blood flow has increased with GTN around the bleed, and no cerebral steal is present, as shown by the difference between prescans and postscans (bottom right).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4. Cerebral blood flow difference scans (post-GTN, baseline) for all patients. Patients receiving active treatment: rows a and b; control: row c. Axial slices are shown at level of the largest lesion diameter or an appropriate level for the clinical presentation if no lesion was visible on the CT scan. The side of lesion (R indicates right; L, left) and Oxfordshire Community Stroke Project classification (LACS indicates lacunar syndrome; PACS, partial anterior circulation syndrome; TACS, total anterior circulation syndrome) are given for each scan.

All of the patients completed 7 days of treatment. One serious adverse event occurred during the treatment phase; a patient receiving GTN had a nonhypotension-related fall leading to trauma of the affected arm. Headache occurred in 1 GTN patient. There were no deaths, and the modified Rankin Scale (telephone assessment by B.W. blinded to treatment) did not differ between the groups at 90 days [median (interquartile range): GTN, 2 (1); control, 2 (1); Mann–Whitney U test²⁴ 2P=0.89).

Discussion

We have shown that transdermal GTN lowers BP modestly (by 14.3%) without having any detrimental effect on CBF or CPP in patients with recent stroke. The neutral effect on CBF was present whether global CBF, ipsilateral and contralateral hemispheric CBF, or ipsilateral and contralateral arterial ROIs were studied. Similarly, GTN did not alter the size of lesion core or penumbral or oligemic areas, as defined by prespecified CBF levels, and did not seem to cause a “cerebral steal” effect.

Examination of the results for individual patients (Figure 3a) suggest that GTN may, in some cases, increase CBF both within and around the hypoattenuated tissue and both superficially and deep within the brain. Similarly, an increase in perilesional CBF was seen in a patient with a primary intracerebral hemorrhage (Figure 3b).

Acute stroke, whether of ischemic or hemorrhagic type, is associated with dysfunctional cerebral autoregulation so that, in the extreme, cerebral perfusion becomes dependent on systemic BP. The mechanism by which GTN can lower BP while maintaining CBF was not addressed in this study. However, GTN forms NO, which is a potent modulator of cerebrovascular reactivity, especially in collateral vessels, such as pial arteries.^25,26 Vascular NO levels are low in stroke,²⁷ so collateral vessels may not be maximally dilated. Hence, CBF might be held constant with GTN if moderate reductions in systemic BP were counterbalanced by increases in collateral blood supply, which would be potentially beneficial in acute stroke. Clearly a larger study including patients at a broader range of stroke subtypes and of acute times after stroke is required to confirm these observations.

It is commonly held that cerebral vasodilators will reduce CPP through increasing cerebral blood volume and ICP. However, this hypothesis neglects the effect that such drugs will have on ZFP, a measure of downstream pressure. We estimated CPP and ZFP and found that GTN did not alter either significantly. The observation that GTN did not alter CPP is new and contrary to the expectation that cerebral vasodilators inevitably increase ICP and, therefore, reduce CPP; presumably, venodilation increases blood flow out of the cranium, thereby maintaining CPP (and ICP).

Several caveats should be placed on our study. First, it was small, and the CIs for the hemodynamic effects of GTN were wide, such that GTN could have moderately reduced or increased CBF and CPP. Nevertheless, the point estimates for differences in a variety of CBF measurements (and lesion areas) for GTN and control-treated patients all lay on or close to zero. Second, we assessed the effect of GTN on CBF and CPP in patients with recent stroke (ie, acute and subacute phases of stroke, median time to randomization, 78 hours) and cannot comment on the effect that GTN has on CBF during the hyperacute period. Third, we included patients irrespective of stroke type, because the CT diagnosis of stroke type was only made at the time of study to reduce radiation doses; because only 2 patients with primary intracerebral hemorrhage were included (1 in each group), we cannot specifically address the effect of GTN on CBF in primary intracerebral hemorrhage, although blood flow was seen to increase adjacent to the hemorrhage in 1 patient. Fourth, few patients with cortical syndromes were included, reflecting the problems of intensive studying of such patients who are often agitated and confused. Fifth, we did not titrate the dose of GTN to BP response, instead choosing to use a fixed-dose GTN patch, so cerebral and systemic hemodynamic measures will have varied somewhat between patients. Finally, we did not directly measure ICP, but rather used an indirect estimate of CPP^9,23 to avoid the need for measuring invasive pressure transducers.