Hypertension and experimental stroke therapies

Abstract

Hypertension is an established target for long-term stroke prevention but procedures for management of hypertension in acute stroke are less certain. Here, we analyze basic science data to examine the impact of hypertension on candidate stroke therapies and of anti-hypertensive treatments on stroke outcome. Methods: Data were pooled from 3,288 acute ischemic stroke experiments (47,899 animals) testing the effect of therapies on infarct size (published 1978–2010). Data were combined using meta-analysis and meta-regression, partitioned on the basis of hypertension, stroke model, and therapy. Results: Hypertensive animals were used in 10% of experiments testing 502 therapies. Hypertension was associated with lower treatment efficacy, especially in larger infarcts. Overall, anti-hypertensives did not provide greater benefit than other drugs, although benefits were evident in hypertensive animals even when given after stroke onset. Fifty-eight therapies were tested in both normotensive and hypertensive animals: some demonstrated superior efficacy in hypertensive animals (hypothermia) while others worked better in normotensive animals (tissue plasminogen activator, anesthetic agents). Discussion: Hypertension has a significant effect on the efficacy of candidate stroke drugs: standard basic science testing may overestimate the efficacy which could be reasonably expected from certain therapies and for hypertensive patients with large or temporary occlusions.

Introduction

Hypertension—the elevation of blood pressure—dramatically increases the risk of stroke, with an estimated 52% of all strokes attributable to hypertension.¹ Hypertension can be controlled and numerous interventions are used in the long-term management of hypertension to prevent both first-time strokes and their recurrence.²

Less well understood is how to tackle high blood pressure within the first 24 to 48 hours after stroke and how the presence of hypertension affects the efficacy of neuroprotective treatments administered within this period. Lowering blood pressure is not without risks, and it is generally recommended that blood pressure only be lowered if extreme; nevertheless, there is no consensus on the optimal blood pressure thresholds to trigger initiation of treatment, rates of reduction and blood pressure targets, or how hypertension should be managed with concurrent therapies such as tissue plasminogen activator.^{2, 3, 4} Clinical trials are ongoing.⁵

By examining hypertension and anti-hypertensive treatments in experimental (animal) stroke models, we may develop a better understanding of how to manage hypertension in the acute stroke setting. Knowledge of hypertension in basic science is also important in its own right. The external validity of animal testing depends upon the ability to generalize to the clinical setting: it has been a frequent criticism of preclinical drug testing that the paradigm has failed to capture the impact of important comorbidities—such as hypertension—that are highly prevalent in human stroke.^{6, 7}

Thus, our aim was to examine the role of hypertension in acute stroke drug development with a view to determining: the prevalence of hypertension in tests of acute stroke therapies; the overall effect of hypertension on acute stroke therapies tested in animals; the relative efficacy of individual stroke treatments and classes of treatments tested in normotensive and hypertensive cohorts; the interaction of stroke model and hypertension on therapeutic efficacy and the relative efficacy of clinical anti-hypertensive drugs compared with other treatments in both normotensive and hypertensive animals.

Materials and methods

Data

Data was pooled from two systematic reviews,^{8, 9} which identified studies reporting the efficacy of interventions in animal experiments modelling stroke. Full details of the search strategies and inclusion criteria can be found within these publications. In brief, the included data were published between 1978 and September 2010 and were extracted from papers identified from Web of Science, Current Contents, Biosis Previews, Pubmed, CAB abstracts and Society for Neuroscience abstracts. Hand searching of abstracts and identification through reference lists was also undertaken for one set of data.⁸

The scope encompassed controlled experiments testing potential therapies in non-human in vivo studies of focal cerebral ischemia. Analysis was restricted to a global measure of brain damage post stroke (infarct size) because it is a relatively standardized measure, it is almost universally reported and it a measure that translates across species. We extracted infarct size summary statistics (mean, s.d. error and sample size); therapy (name and treatment, mechanism of action) and study characteristics (species and stroke model). Where volumetric data were not available, estimates were based on areas. Where data were presented only graphically, values were estimated using an electronic ruler.

Within each experiment, means and s.d. of infarct sizes in treatment groups were expressed as a percentage of control group outcomes.¹⁰ Sample sizes in the control groups were adjusted for multiple comparisons.¹⁰

Data Analysis

Meta-analysis was used to combine results from individual experiments testing the effect of acute stroke therapies on infarct size. To account for anticipated data heterogeneity, estimates of therapeutic efficacy of the acute stroke treatments given to animals were calculated using the conservative DerSimonian and Laird random-effects meta-analysis.¹¹ Data were partitioned (i.e., grouped into subcategories) on the basis of animal type, treatment type, and stroke model as described below. Meta-regression (STATA v.10, Statacorp, College Station, TX, USA)^{12, 13} was then used to test whether differences in therapeutic efficacy depended upon the blood pressure of the animal (hypertensive or not) and the type of treatment (anti-hypertensive or not).

Efficacy in Hypertensive Animals

Data were partitioned according to whether experiments used normotensive or hypertensive animals. For all therapies combined, for each individual therapy and for classes of therapeutic targets, we tabulated the effect size, standard error, 95% confidence intervals, and number of experiments for both normotensive and hypertensive cohorts of animals. Results are reported for individual treatments tested in both normotensive and hypertensive animals. Results are also presented for normotensive and hypertensive animals according to target: excitotoxicity; inflammation; metabolism; oxidation; apoptosis/regeneration; blood flow; blood clot; nootropics; fluid management; temperature, and other targets.

Efficacy in Different Stroke Models

As some stroke models induce cerebral hypotension with the occlusion of one or both common carotid arteries (CCA)— in addition to a focal middle cerebral artery (MCA) occlusion—data were further partitioned according to stroke model. Two factors were used to differentiate stroke models, location of occlusion and degree of cerebral hypotension, with four types of stroke models in total: (a) Distal MCA branch occlusion; (b) MCA origin/internal carotid artery (ICA) occlusion; (c) Tandem occlusion (occlusion of the MCA and one CCA), and (d) three-vessel occlusion (occlusion of the MCA and both CCAs). Occlusion of the MCA origin/ICA is the most common site of occlusion and typically results in the largest area of brain damage compared with the other models. In contrast, the standard tandem and three-vessel occlusion models involve focal occlusion of the distal MCA with differing degrees of hemispheric or whole-brain ischemia. Occlusion of the carotid arteries in these models is typically temporary; although such models mimic elements of global ischemia models, these models typically induce focal ischemic lesions around a distal MCA branch and the outcomes are typically reported in the style of focal models.

Efficacy of Anti-hypertensive Treatments

Data were also partitioned according to the category of anti-hypertension treatment: (1) angiotensin-converting-enzyme inhibitors; (2) beta (adrenergic) blockers; (3) calcium channel blockers; (4) vasodilators; (5) angiotensin II (ANG II) antagonists; (6) diuretics; (7) anti-α adrenergic agents; (8) aldosterone antagonists or (9) all other treatment types. Treatments were only ascribed to a particular anti-hypertensive treatment class if the therapy had been used clinically—not just experimentally—for this purpose. Thus, the ‘all other treatment' category included experimental therapies, which may also have effects related to hypertension (such as calcium channels and vasodilation) but which have not been used clinically for this purpose. Therapeutic effect sizes were then calculated for each class, both overall, and for the normotensive and hypertensive animals.

Quality of Data

Heterogeneity was tested across all experiments using I², a statistic derived from Q but adjusting for degrees of freedom.¹⁴ Publication bias was tested using Egger's test, a statistical analog of the funnel plot and Begg's test (STATA v.10, Metabias).^{15, 16} Significant results (P<0.05) suggest bias. Bias-corrected estimates were calculated using the trim and fill technique (STATA, v.10), which accounts for unpublished data by imputing likely values for missing data.¹⁷ Study quality wasassessed in a subset of experiments with a maximum score of 10, with randomization and outcome blinding statistics reported separately.¹⁰

Results

Efficacy in Hypertensive Animals

We analyzed data from 3,288 experiments of which 90% used normotensive animals (N=2,946 experiments, 43,339 animals) and 10% used hypertensive animals (N=342 experimental comparisons, 4,560 animals) (Figure 1). Identified studies only used the genetically hypertensive strain, the spontaneous hypertensive rat (SHR) and its variants (e.g., the stroke-prone SHR). Overall, efficacy was lower in hypertensive animals (23.7±0.4, 95% CI 22.8 to 24.6) than all normotensive animals (25.2±0.2, 95% CI 24.9–25.6) (Adj R²=1.3%, P<0.0001, N=3,288 experiments). Although not a planned comparison, it is worth noting that compared with hypertensive animals (all rats), therapeutic efficacy in normotensive rats was high (26.5±.2, 95% CI 26.2–26.9, N=2,286 experiments), as was efficacy in the SHRs' putative control, the Wistar (27.7±0.4, 95% CI 27.0–28.4, N=767 experiments).

Figure 1.

Testing of acute stroke treatments in animal models. Piechart of data from 3,288 controlled experimental tests of therapies in acute stroke in animal models (47,899 animals). Hypertensive animals were used in 332 experiments (10.4%). Normotensive animals were used in 2,946 experiments (89.6%).

Effects of Individual Therapies and Classes of Therapies Tested in Normotensive and Hypertensive Animals.

Of the 502 therapies tested in acute experimental stroke and included in the dataset, 474 had been tested in normotensive animals, 86 had been tested in hypertensive animals, and 58 had been tested in both. Effect sizes for the 58 therapies tested in both normotensive and hypertensive animals are given in Table 1, with therapies listed alphabetically. The same data are presented graphically, in the order of ascending efficacy within hypertensive animals (Figure 2). Bars for effect sizes in normotensive animals are presented in light gray. Bars for effect sizes in hypertensive animals are presented in red, dark gray or blue, depending upon their efficacy relative to the performance in normotensive animals as determined by the span of the 95% confidence intervals.

Table 1. Individual treatments in normotensive and hypertensive animals.

	Normotensive animals			Hypertensive animals
Treatment name	Effect size±s.e.	95% CI	Number of experiments	Effect size±s.e.	95% CI	Number of experiments
3-Morpholino-sydnonimine	54±4	[47, 61]	9	32±7	[18, 45]	1
Ac-YVAD.cmk	36±2	[32, 39]	5	16±7	[1, 30]	3
Adrenomedullin	46±9	[28, 64]	1	19±7	[6, 33]	1
Albumin	31±2	[26, 35]	20	−16±13	[−41, 9]	2
Aminoguanidine	34±3	[29, 40]	16	13±3	[7, 18]	10
ARL-15896	35±5	[25, 46]	10	40±6	[29, 51]	4
Artificial CSF	35±47	[−57, 126]	1	−2±11	[−23, 19]	1
Candesartana	61±4	[52, 69]	3	35±7	[22, 48]	5
Citicoline	14±2	[10, 17]	34	55±3	[49, 61]	1
D-Arginine	-5±4	[−13, 4]	1	−14±15	[−43, 17]	1
Dexanabinol	56±26	[4, 107]	1	41±2	[37, 45]	10
Dizocilpine	35±1	[33, 37]	95	5±8	[−10, 20]	2
DY-9760e	52±11	[31, 73]	4	15±5	[5, 24]	3
Electrical stimulation	40±4	[31, 48]	24	37±5	[27, 46]	7
Enlimomab	11±5	[3, 20]	7	−2±3	[−9, 5]	2
Enrichment/exercise	31±1	[29, 33]	4	−15±21	[−55, 26]	1
Halothane	40±6	[29, 52]	4	−6±3	[−13, 0]	10
Hyperbaric oxygen	35±3	[28, 41]	15	18±7	[7, 32]	1
Hypercapnia	2±9	[−16, 20]	1	−2±23	[−48, 43]	1
Hypothermia	33±1	[32, 34]	161	56±4	[48, 64]	8
Ifenprodil	18±7	[5, 31]	3	21±4	[14, 29]	1
Isoflurane	31±4	[23, 39]	20	−67±10	[−86, −48]	4
Isradipinea	3±2	[−1, 8]	7	32±1	[30, 34]	17
L-arginine	−18±4	[−26, −10]	4	21±6	[9, 33]	3
L-NAME	20±2	[16, 25]	28	39±4	[31, 47]	5
MDL 101, 002	51±12	[27, 74]	8	22±10	[3, 41]	3
Methohexital	33±12	[9, 56]	2	23±4	[15, 31]	8
Minocycline	44±5	[33, 55]	3	26±6	[14, 37]	1
Nicardipinea	53±8	[38, 68]	3	8±7	[−5, 22]	3
Nicotinamide	24±2	[20, 29]	56	14±5	[3, 24]	5
Nilvadipine	15±3	[9, 22]	10	9±7	[−4, 22]	4
Nimodipinea	23±2	[19, 26]	26	30±4	[23, 37]	7
Normobaric oxygen	17±4	[10, 25]	8	18±10	[−1, 38]	1
NXY-059	41±2	[37, 46]	23	18±6	[6, 29]	2
Ozagrel	19±4	[10, 27]	6	16±6	[4, 28]	6
Pentastarch	57±8	[42, 72]	2	17±13	[−9, 43]	1
Pentobarbital	13±5	[3, 22]	7	−7±4	[−14, 0]	6
Phenylephrine	20±14	[−7, 47]	3	4±7	[−9, 17]	3
Progesterone	21±11	[−1, 43]	8	23±9	[5, 40]	4
Propranolola	−1±4	[−9, 8]	4	−2±24	[−49, 44]	1
Rapamycin	43±26	[−7, 93]	1	5±23	[−39, 49]	1
SB203580	19±6	[6, 32]	4	25±7	[10, 39]	7
SB-221420-A	17±1	[15, 19]	9	25±3	[19, 31]	3
SB-239063	21±6	[8, 34]	3	28±1	[26, 31]	21
Sipatrigine	37±1	[35, 39]	35	32±14	[5, 59]	1
SNX-111	55±6	[43, 67]	4	52±7	[38, 66]	3
Sodium nitroprusside	40±17	[6, 73]	3	19±3	[13, 25]	2
Stem cells	22±3	[16, 27]	11	8±18	[−26, 43]	1
Tacrolimus	25±1	[22, 27]	48	13±2	[8, 18]	15
Tempol	34±8	[18, 49]	5	50±3	[43, 56]	4
Thiopentone	4±16	[−27, 35]	1	31±4	[23, 38]	8
Tirilazad	32±3	[25, 38]	17	14±7	[0, 29]	2
TNF receptor inhibitor	20±3	[14, 26]	7	26±4	[17, 35]	4
TNF-alpha	−3±7	[−16, 10]	2	−73±9	[−91, −55]	4
tPA	17±1	[16, 19]	178	7±3	[1, 13]	17
VML 588	−14±42	[−97, 69]	2	18±4	[10, 26]	3
YM90K	30±2	[26, 35]	18	34±13	[8, 61]	1
Z-DEVD-FMK	40±2	[36, 44]	22	24±12	[0, 48]	3

^aclassified as clinical anti-hypertensive treatments.

Figure 2.

Individual treatment efficacy in normotensive and hypertensive animals. Effect sizes of acute ischemic stroke for the 58 therapies, which had been tested in both normotensive and hypertensive animals (see also Table 1). Data are presented in order of ascending efficacy within hypertensive animals. Scale bars for effect sizes in normotensive animals are presented in light gray. Scale bars for effect sizes in hypertensive animals are presented in red, dark gray or blue, depending upon their efficacy relative to the performance in normotensive animals as determined by the span of the 95% confidence intervals.

Twelve therapies were less effective in hypertensive animals compared to normotensive animals (TNF-α, isoflurane, albumin, exercise, pentobarbital, halothane, dizocilpine, tissue plasminogen activator, nicardipine, aminoguanidine, Ac-YVAD.cmk and NXY-059; Figure 2 red bars) and six therapies were more effective (hypothermia, citicholine, L-Name, isradipine, SB-221420-A and L-arginine; Figure 2 blue bars). When outcomes were grouped according to therapeutic target, hypertensive animals showed relatively beneficial outcomes when the treatment targeted temperature (hypothermia) but worse outcomes when therapies targeted excitotoxicity, inflammation, and thrombolysis (see Table 2).

Table 2. Therapeutic targets in normotensive and hypertensive animals.

	Normotensive animals			Hypertensive animals
Therapeutic target	Effect size±s.e.	95% CI	Number of experiments	Effect size±s.e.	95% CI	Number of experiments
Excitotoxicity	29±0.3	[29, 30]	710	13±1	[11, 16]	65
Inflammation	37±0.7	[26, 29]	228	13±1	[10, 15]	39
Metabolism	16±1	[13, 18]	41			0
Oxidation/free radicals	24±0.6	[23, 25]	384	25±1	[22, 27]	34
Apoptosis/regeneration	28±0.4	[27, 29]	483	25±1	[23, 28]	57
Blood flow	24±0.6	[23, 25]	156	26±1	[24, 27]	71
Blood clot (thrombolysis)	20±0.6	[19, 22]	298	12±2	[8, 16]	41
Nootropics	−7±1	[−10, −4]	53	1±5	[−8, 9]	2
Fluid management	24±1	[22, 27	51	−1±7	[−15, 12]	4
Temperature (hypothermia)	28±0.4	[27, 28]	380	41±2	[38, 45]	20
Other	20±2	[17, 24]	45	37±5	[27, 46]	7

Effect of Stroke Model on Efficacy in Normotensive and Hypertensive Animals

The effect of stroke model and hypertension on therapeutic efficacy is shown in Figure 3. In distal MCA occlusion models, there was little difference in therapeutic efficacy for normotensive (22.6±0.3, 95% CI 22.1–23.1) and hypertensive animals (23.4±0.7, 95% CI 22.124.7). However, in MCA origin or ICA occlusion (i.e., embolic and filament occlusion) models, efficacy was lower in the hypertensive animals (normotensive 30.1±0.2, 95% CI 29.6–30.1; hypertensive 23.4±0.7, 95% CI 25.0, 23.8–26.3). Similarly, in the potentially hypotensive tandem occlusion models, therapies were more effective in the normotensive animals (normotensive 28.3±0.7, 95% CI 26.9–29.6; hypertensive 17.7±0.7, 95% CI 14.6–20.7). For the three-vessel occlusion model, we only found reports of experiments using normotensive animals.

Figure 3.

Efficacy in different animals and stroke models. Therapeutic efficacy of 502 treatments in animal models of stroke portioned on the basis of classes of animal (hypertensive and normotensive) and method of stroke induction (middle cerebral artery (MCA) branch occlusion, occlusion at the origin of the MCA or within the internal carotid artery, tandem occlusion of the MCA and common carotid artery (CCA) and occlusion of three vessels, the MCA and both CCAs).

When testing for significance using a combined statistical model of animal hypertension status and occlusion location (MCA origin/ICA occlusion versus other types), the blood pressure status of the animal was no longer significant, however, the location of occlusion was associated with differences in outcome (F(1,3287)=7.3, P=0.007, partial η²=0.002). Further, there was a significant interaction between location of occlusion and hypertension (F(1,3287)=7.8, P=0.005, partial η²=0.002) with treatments less effective in hypertensive animals with MCA origin/ICA occlusions versus other models. Similarly, efficacy differed for hypertensive and normotensive animals depending on whether a hypotensive occlusion model was used (F(1,3287)=530.9, P<0.001, partial η²=0.14). Use of a hypotensive model also had a direct effect on outcome (F(1,3287)=531.0, P<0.001, partial η²=0.14).

Efficacy of Anti-hypertensive Treatments

Five hundred and two therapies were tested in in vivo stroke models, of which 15 (102 experiments) involved clinically recognized anti-hypertensive drugs. Across all animals, there was no difference in the performance of clinical anti-hypertensive treatments in acute stroke (24.7±0.6, 95% CI 23.5–25.9) versus all other therapies (25.1±0.2, 95% CI 24.8–25.4, Adj R²=0%, P>0.05, N=3,288 experiments). However, in hypertensive animals, anti-hypertensive therapies were more effective than other interventions tested (29.8±0.9, 95CI% 28.0–31.6 v 22.6±0.5, 95% CI 21.6–23.6), whereas in normotensive animals the converse was true (19.8±0.9, 95% CI 18.0–21.5 v 25.4±0.2, 95% CI 25.1–25.8) (Figure 4A) although the interaction was not significant (F(1,3287)=0.04, P>0.05, partial η²=0.00).

Figure 4.

Relationship between hypertension and acute experimental stroke outcome. (A) Effect size of 15 clinical anti-hypertensive therapies and 487 other types of therapies in hypertensive and normotensive animals undergoing acute ischemic stroke. (B) Effect size of different subclasses of clinical anti-hypertensive therapies in acute experimental stroke.

To explore whether the timing of anti-hypertensive treatment might influence these observations, we conducted a post hoc investigation. For anti-hypertensive treatments, the median time of delivery was 45 minutes before stroke onset, compared with 10 minutes after stroke onset for all other treatments. The interaction between the use of clinical anti-hypertensive treatments and animal type was significant when the anti-hypertensive treatments were delivered before stroke onset (F(1,3287)=6965, P=0.014, partial η²=0.005) but not after (F(1,3287)=0.07, P>0.05, partial η²=0.00). When we excluded treatments given before stroke onset, there was still a net benefit for anti-hypertensive treatments in hypertensive animals, although the advantage was smaller (anti-hypertensive treatments 25.5±1.2, 95% CI 23.2–27.9 versus other treatments 23.6±0.7, 95% CI 22.3–24.9). Further, for the normotensive cohort, anti-hypertensive treatments delivered after stroke onset also demonstrated marginal superiority compared with other treatments (anti-hypertensive treatments 26.7±3.2, 95% CI 20.4–32.9 versus other treatments 24.6±0.2, 95% CI 24.2–25.0). Anti-hypertensive treatments were at their least advantageous for normotensive animals (compared with other treatments) when delivered within 24 hours before stroke onset.

Statistical testing did not suggest that any one particular subclass of anti-hypertensive treatments was more beneficial than others (F(4,3283)=1.47, P>0.05, N=3,288); however, the number of experiments in each class were low (Figure 4B). Although not statistically significant, post-stroke outcome was worse after the administration of angiotensin-converting-enzyme inhibitors (8.5±2.3, 95% CI 3.8–13.2) and vasodilators (16.0±2.9, 95% CI 10.4–21.7) compared with non-anti-hypertensive drugs (25.1±0.2, 95% CI 24.8–25.4). Conversely, acute administration of angiotensin II antagonists resulted in a superior outcome (52.5±3.7, 95% CI 45.3–59.7) (Figure 4B). No included studies involved testing of diuretics, anti-adrenergic (α) antagonists, or aldosterone antagonists, which have been used to treat clinical hypertension.

Post hoc analyses were undertaken to investigate whether outcomes would vary if the anti-hypertensive treatment class definition was expanded to include treatments not used clinically as anti-hypertensive drugs but belonging to the same class of drugs. Outcome only changed for beta blockers, with the reduction in infarct volume dropping from 25.3 (25.3±1.6, 95% CI 22.2–28.3, N=13) to 15.5 (15.5±0.8, 95% CI 13.9–17.0, N=48); the performance of all other classes did not alter by more than half a percentage point. Experimental diuretics and anti-adrenergic antagonists yielded an efficacy of 18.6 (18.6±2.4, 95% CI 13.7–23.3, N=22) and 15.4 (15.4±2.4, 95% CI 10.7–20.1, N=15), respectively.

Quality of Data

Data were heterogeneous (I²=87.7%). Eggers's test showed evidence of publication bias (P<0.001), while Begg's test did not (P>0.05). Using the trim and fill technique (STATA v.10), estimates of effect size were adjusted downward from the original estimates by 5.2% (original data N=3,288, improvement in infarct volume 27.1, 95% CI: 26.2–28.1; adjusted data N=3,693: improvement in infarct volume 21.9, 95% CI 20.9–22.8). It was reported that animals were randomized to treatment group in 33% of experiments and that outcomes were assessed by operators blind to treatment allocation in 38% of experiments. For the subset of data for which a complete quality assessment was undertaken (N=708 experiments), the median reported quality score was 4 out of 10 and sample size calculations were reported by 3%.

Discussion

Hypertension has numerous effects upon the brain, notably, compromise of the blood–brain barrier,¹⁸ distributed damage to the white matter,¹⁹ and post-stroke edema.²⁰ Such physiologic changes create a different milieu for the ischemia-challenged brain. With the high prevalence of hypertension as both a risk factor and sequelae of stroke, it is an important consideration in acute stroke management.

Testing in Hypertensive Models

Within the basic science testing of acute stroke therapies, 10% of studies had been undertaken in hypertensive animals—substantially less than the estimated prevalence of hypertension in the stroke population. Failure to capture the clinical prevalence of hypertension within the acute stroke models may limit the generalisablity of results from animal studies of acute stroke therapies.

To the extent that acute stroke therapies have been tested in hypertensive animals, the testing has been undertaken predominantly in the genetically hypertensive rat. This strain typically shows a smaller variance and larger infarct than comparable strains, with the large stroke sizes in stroke-prone SHRs attributed to poor collateral circulation post-stroke.²¹ In the particular set of studies examined here, estimated stroke sizes in the hypertensive rats occupy around 10% more of the hemisphere compared normotensive rats irrespective of which model was used (distal, proximal, or three-vessel occlusion). This would suggest that it may be possible to salvage similar absolute amounts of tissue in hypertensive rats, but the relative proportion of salvageable tissue may be less. However, in careful experiments undertaken by McGill et al ²² where size of initial infarct was comparable between stroke-prone SHRs and Wistar Kyoto rats, the hypertensive rats still fared worse than their normotensive counterparts. Why hypertensive rats fare worse may be related to the presence of pre-existing end-organ damage (brain, heart, and kidneys) and to increased pulsatile pressure on the vasculature in hypertensive rats rendering them less able to maintain autoregulation in the face of ischemia.

When therapies were given to hypertensive animals after distal MCA occlusions, overall, there was little difference in efficacy compared with normotensive animals. However, this may have been confounded by species; efficacy tends to be lower in mice models and mice models frequently use the MCA distal branch model. A post hoc review of the data for normotensive rats alone revealed that occlusion of the distal MCA results in greater efficacy (25±0.3) compared with hypertensive rats (23±0.7), with an even more marked difference when only Wistar rats were considered (32±0.9).

In MCA origin/ICA occlusion models, efficacy was around 7% less effective in hypertensive animals; thus, standard basic science testing regimes in normotensive animals may overestimate efficacy, which could be expected for acute stroke treatments in patient groups with large infarcts. These large infarcts models were around 30 times more likely to be temporary than the distal MCA models (not counting embolic). Although a statistical comparison of occlusion duration was not planned, it may be worth noting that efficacy in normotensive rats was typically much higher in temporary models compared with permanent models (32±0.3 versus 24±0.3), yet this difference was not seen in hypertensive rats (22±1 versus 23±0.6). Consequently, basic science testing regimes in normotensive animals may also overestimate predicted neuroprotective therapeutic efficacy in patient groups where recanalisation has occurred (equivalent to temporary models).

In models with potential cerebral hypotension (the tandem MCA/CCA occlusion model), efficacy was around 10% lower in the hypertensive animals when compared with normotensive animals. The situation may be contrasted with the specific findings for anti-hypertensive therapies; such therapies might be relatively beneficial in hypertensive animals-at least when delivered before stroke onset. Thus, it may not be a reduction in pressure per se, that is beneficial in hypertensive animals; any relative advantage from anti-hypertensive treatments for hypertensive animals might be due to a pressure reduction that is systemic, small and/or delivered before stroke onset, or might arise through mechanisms other than pressure. While anti-hypertensive treatments might still have a place in treatment regimens for normotensive animals in the acute phase of stroke, they were relatively less useful when delivered before stroke onset.

Effects of Different Therapies

Fifty-eight therapies were tested in both normotensive and hypertensive animals in acute experimental stroke. Hypothermia plus several of the clinically used anti-hypertensive agents and nitric oxide system modulators resulted in a superior outcome for hypertensive animals compared with normotensive animals. Thus, targeting the endothelial system could be a useful therapeutic option in hypertensive patients.

Conversely, the thrombolytic tissue plasminogen activator resulted in a worse outcome for hypertensive animals, consistent with clinical findings from registry data.²³ Anesthetic agents (halothane, isoflurane, and pentobarbital) also performed relatively worse in hypertensive animals compared with normotensive animals. As anesthetic agents are deployed in most stroke experiments, confounding effects of anesthesia and lack of modelling of hypertension may reduce the generalizability and external validity of many basic science findings.

Effects of Anti-hypertensives

Acute administration of angiotensin II antagonists resulted in a superior outcome, in contrast to the findings of the clinical trial Scandinavian Candesartan Acute Stroke Trial 1;³ however, in animal studies, drug delivery was commenced before stroke onset in all but one of these tests—often days before (in contrast to the Scandinavian Candesartan Acute Stroke Trial³). By contrast, angiotensin-converting-enzyme inhibitors tended to result in lower therapeutic efficacy although differences were not significant. It is worth noting however, that anti-hypertensive therapies may also have actions relevant to stroke outcome, other than a reduction in hypertension. For instance, angiotensin-converting-enzyme inhibitors reported to reduce aspiration pneumonia post stroke.²⁴

Limitations

This review was broad rather than deep and is limited in terms of conclusion by the heterogeneity of the data. Because of the heterogeneity, we used random-effects meta-analyses, prespecified the comparison of interest, and restricted the analysis to the most common and standardized outcome for basic science stroke experiments (infarct size). Nevertheless, other outcomes and analyses are also relevant to acute stroke.

Classification of treatment mechanisms is notoriously difficult with treatments with multiple mechanisms of action and no objective measure of the strength of action in a domain. For this reason, we defined anti-hypertensive treatments by reference to clinical use; however, this will not give a clean delineation based on scientific method of action.

Finally, all studies identified using hypertensive animals were limited to the SHR model. This raises a flag regarding the generalizability of results and leaves open animal species as a possible confounder. The basic science evidence demonstrating the effect of hypertension on acute stroke therapies would be strengthened if validated in models other than the SHR.^{25, 26}

Conclusions

The basic science literature suggests that hypertension may be relevant to acute ischemic stroke management in at least four ways: First, that the effect of hypertension on treatment outcomes may be more pronounced in larger strokes or temporary occlusions and that standard testing regimes in animal models may overestimate the efficacy, which could be expected in patient groups with large strokes or recanalisation. Second, that substantial reduction in cerebral blood perfusion (as accompanies the CCA occlusion used in tandem and three-vessel occlusion models) might compromise the effect of acute stroke therapies in hypertensive animals. Third, that anti-hypertensive therapy given before stroke may be relatively beneficial for hypertensive cohorts but relatively detrimental where there was no pre-existing hypertension. Fourth, that while some treatments work well in hypertensive animals (e.g., hypothermia, nitric oxide synthase agents), other treatments perform better in normotensive animals (e.g., anesthetic agents, tissue plasminogen activator).

From the perspective of translational medicine, the results of animal and human data are not inconsistent when taking into account hypertension and looking at the relative—rather than the absolute—levels of therapeutic efficacy. The generalisabilty and validity of basic science drug development paradigms could be increased by embracing hypertension as an important covariate in the experimental design.