Article

A Pilot Study of Normobaric Oxygen Therapy in Acute Ischemic Stroke

Harvard University, Cambridge, Massachusetts, United States
Stroke (Impact Factor: 5.72). 04/2005; 36(4):797-802. DOI: 10.1161/01.STR.0000158914.66827.2e
Source: PubMed
ABSTRACT
Therapies that transiently prevent ischemic neuronal death can potentially extend therapeutic time windows for stroke thrombolysis. We conducted a pilot study to investigate the effects of high-flow oxygen in acute ischemic stroke.
We randomized patients with acute stroke (<12 hours) and perfusion-diffusion "mismatch" on magnetic resonance imaging (MRI) to high-flow oxygen therapy via facemask for 8 hours (n=9) or room air (controls, n=7). Stroke scale scores and MRI scans were obtained at baseline, 4 hours, 24 hours, 1 week, and 3 months. Clinical deficits and MR abnormalities were compared between groups.
Stroke scale scores were similar at baseline, tended to improve at 4 hours (during therapy) and 1 week, and significantly improved at 24 hours in hyperoxia-treated patients. There was no significant difference at 3 months. Mean (+/-SD) relative diffusion MRI lesion volumes were significantly reduced in hyperoxia-treated patients at 4 hours (87.8+/-22% versus 149.1+/-41%; P=0.004) but not subsequent time points. The percentage of MRI voxels improving from baseline "ischemic" to 4-hour "non-ischemic" values tended to be higher in hyperoxia-treated patients. Cerebral blood volume and blood flow within ischemic regions improved with hyperoxia. These "during-therapy" benefits occurred without arterial recanalization. By 24 hours, MRI showed reperfusion and asymptomatic petechial hemorrhages in 50% of hyperoxia-treated patients versus 17% of controls (P=0.6).
High-flow oxygen therapy is associated with a transient improvement of clinical deficits and MRI abnormalities in select patients with acute ischemic stroke. Further studies are warranted to investigate the safety and efficacy of hyperoxia as a stroke therapy.

Full-text

Available from: Ferdinando S Buonanno, Mar 21, 2016
A Pilot Study of Normobaric Oxygen Therapy in Acute
Ischemic Stroke
Aneesh B. Singhal, MD; Thomas Benner, PhD; Luca Roccatagliata, MD; Walter J. Koroshetz, MD;
Pamela W. Schaefer, MD; Eng H. Lo, PhD; Ferdinando S. Buonanno, MD;
R. Gilberto Gonzalez, MD, PhD; A. Gregory Sorensen, MD
Background and Purpose—Therapies that transiently prevent ischemic neuronal death can potentially extend therapeutic
time windows for stroke thrombolysis. We conducted a pilot study to investigate the effects of high-flow oxygen in acute
ischemic stroke.
Methods—We randomized patients with acute stroke (12 hours) and perfusion-diffusion “mismatch” on magnetic
resonance imaging (MRI) to high-flow oxygen therapy via facemask for 8 hours (n9) or room air (controls, n7).
Stroke scale scores and MRI scans were obtained at baseline, 4 hours, 24 hours, 1 week, and 3 months. Clinical deficits
and MR abnormalities were compared between groups.
Results—Stroke scale scores were similar at baseline, tended to improve at 4 hours (during therapy) and 1 week, and
significantly improved at 24 hours in hyperoxia-treated patients. There was no significant difference at 3 months. Mean
(SD) relative diffusion MRI lesion volumes were significantly reduced in hyperoxia-treated patients at 4 hours
(87.822% versus 149.141%; P0.004) but not subsequent time points. The percentage of MRI voxels improving
from baseline “ischemic” to 4-hour “non-ischemic” values tended to be higher in hyperoxia-treated patients. Cerebral
blood volume and blood flow within ischemic regions improved with hyperoxia. These “during-therapy” benefits
occurred without arterial recanalization. By 24 hours, MRI showed reperfusion and asymptomatic petechial hemor-
rhages in 50% of hyperoxia-treated patients versus 17% of controls (P0.6).
Conclusions—High-flow oxygen therapy is associated with a transient improvement of clinical deficits and MRI
abnormalities in select patients with acute ischemic stroke. Further studies are warranted to investigate the safety and
efficacy of hyperoxia as a stroke therapy. (Stroke. 2005;36:797-802.)
Key Words: magnetic resonance imaging
neuroprotection
oxygen
stroke
I
dentifying strategies to extend the thrombolysis time win-
dow is an important area of stroke research.
1
One approach
is to arrest the transition of ischemia to infarction (“buy
time”) until reperfusion can be achieved. Hyperoxia might be
a useful physiological therapy that slows down the process of
infarction and has shown promise in studies of myocardial
infarction.
2
Tissue hypoxia is a key factor contributing to cell
death after stroke and oxygen easily diffuses across the
blood– brain barrier. Moreover, oxygen has multiple benefi-
cial biochemical, molecular, and hemodynamic effects.
3–5
Hyperbaric oxygen therapy (HBO) has been widely studied
because it significantly raises brain tissue pO
2
(ptiO
2
). Tran
-
sient “during-therapy” clinical improvement was documented
40 years ago,
6
and HBO proved effective in animal studies.
7–9
However, the failure of 3 clinical stroke trials
10 –12
has
reduced the enthusiasm for using HBO in stroke.
In light of the difficulties with HBO, we have begun to
investigate normobaric oxygen therapy (NBO), or the deliv-
ery of high-flow oxygen via a facemask. NBO has several
advantages: it is simple to administer, noninvasive, inexpen-
sive, widely available, and can be started promptly after
stroke onset (for example, by paramedics). Whereas brain
ptiO
2
elevation with NBO is minor as compared with HBO,
the critical mitochondrial oxygen tension is extremely low,
13
and even small increases in ptiO
2
might suffice to overcome
thresholds for neuronal death. Recent studies indicate that
brain ptiO
2
increases linearly with rising concentrations of
inspired oxygen,
14
and increases nearly 4-fold over baseline
have been documented in brain trauma patients treated with
NBO.
3
A recent in vivo electron paramagnetic resonance
oximetry study has shown that NBO significantly increases
ptiO
2
in “penumbral” brain tissue.
15
In rodents, NBO therapy
during transient focal stroke attenuates diffusion-weighted
MRI (DWI) abnormalities, stroke lesion volumes, and neu-
robehavioral outcomes
4,16,17
without increasing markers of
oxidative stress.
16
Based on preclinical results, we conducted
Received December 13, 2004; revision received January 11, 2005; accepted January 12, 2005.
From the Departments of Neurology (A.B.S., W.J.K., F.S.B.) and Radiology (T.B., L.R., P.W.S., E.H.L., R.G.G., A.G.S.), Massachusetts General
Hospital and Harvard Medical School, Boston, Mass.
Correspondence to Aneesh B. Singhal, MD, VBK-802, Stroke Service, Department of Neurology, Massachusetts General Hospital, Boston, MA 02114.
E-mail asinghal@partners.org
© 2005 American Heart Association, Inc.
Stroke is available at http://www.strokeaha.org DOI: 10.1161/01.STR.0000158914.66827.2e
797
by guest on September 24, 2015http://stroke.ahajournals.org/Downloaded from
Page 1
a pilot clinical study to examine the risks and benefits of NBO
in stroke. We hypothesized that clinical and MRI parameters
of ischemia would transiently improve during NBO.
Materials and Methods
This randomized, placebo-controlled study with blinded MRI anal-
ysis was approved by our hospital’s Human Research Committee.
The inclusion criteria were: (1) nonlacunar, anterior circulation
ischemic stroke presenting 12 hours after witnessed symptom
onset or 15 hours after last seen neurologically intact; (2) ineligible
for intravenous/intra-arterial thrombolysis; (3) National Institutes of
Health Stroke Scale (NIHSS) score 4; (4) pre-admission modified
Rankin scale (mRS) score 1, and (5) mean transit time (MTT)
lesion larger than DWI lesion (perfusion– diffusion “mismatch”)
with evidence for cortical hypoperfusion on MRI. To minimize time
to treatment, “mismatch” was assessed during the initial MRI, using
a visual estimate for 20% difference between DWI and MTT lesion
size. The exclusion criteria were: (1) active chronic obstructive
pulmonary disease; (2) 3 L/min oxygen required to maintain
peripheral arterial oxygen saturation (SaO
2
) 95% as per current
stroke management guidelines;
18
(3) rapidly improving neurological
deficits; (4) medically unstable; (5) pregnancy; (6) inability to obtain
informed consent; and (7) contraindication for MRI. Eligible patients
gave consent and were randomized by opening sealed envelopes
containing treatment allocation to the NBO group (humidified
oxygen via simple facemask at flow rates of 45 L/min) or the control
group (room air or nasal oxygen 1 to 3 L/min if necessary to maintain
SaO
2
95%). NBO was stopped after 8 hours; however, nasal
oxygen was continued if clinically warranted.
National Institutes of Health Stroke Scale (NIHSS), mRS, and
Scandinavian Stroke Scale (SSS) scores were recorded after the
admission MRI. NIHSS scores and MRI scans were repeated at 4
hours (range, 2.5 to 5.5 hours); 24 hours (range, 20 to 28 hours); 1
week (range, 5.5 to 8.5 days); and 3 months (range, 80 to 115 days).
SSS and mRS scores were repeated at 3 months. The unblinded
clinical investigator monitored patients during therapy. Imaging
technique details are presented in the appendix.
Manual MRI analysis was performed by 2 neuroradiologists
blinded to clinical presentation, treatment group, clinical course, and
medications. Stroke volumes were calculated from DWI images
except for 1-week and 3-month time points, when fluid-attenuated
inversion recovery images were used. Lesions were outlined on each
axial slice using a commercially available image analysis program
(ALICE; Perceptive Informatics, Waltham, Mass) to yield total
volumes. Reperfusion (defined as clear identification of a previously
occluded artery on magnetic resonance angiography [MRA] or
50% decrease in MTT lesion volume in patients without arterial
cutoff on initial MRA) was determined on 4-hour and 24-hour MRIs.
Postischemic hemorrhage was ascertained on 24-hour gradient-echo
MRIs.
Automated MRI analysis was performed to determine the fate of
individual voxels on apparent diffusion coefficient (ADC) maps, as
per their change in signal intensity above or below a threshold of
60010
6
mm
2
/s (45% of normal
19
) from baseline to the 4-hour
and 24-hour time points. Voxels with signal intensity constantly
above threshold were considered “never-abnormal;” remaining vox-
els were grouped as follows: (1) no reversal, signal intensity below
threshold at all time points; (2) temporary early reversal, signal
intensity below threshold at baseline, improving to an above-
threshold value at 4 hours, but reverting at 24 hours; (3) sustained
early reversal, signal intensity below threshold at baseline, improv-
ing to an above-threshold value at 4 hours and 24 hours; (4) late
reversal, signal intensity below threshold at baseline and 4 hours,
improving to an above-threshold value at 24 hours; and (5) progres-
sion to ischemia, signal intensity above threshold at baseline,
worsening to a below-threshold value at 4 hours or 24 hours. We
further analyzed voxels with “sustained early reversal” for “late
secondary decline”
19
on the 1-week MRI.
For each patient, outlines of the baseline MTT lesion were
transferred onto coregistered perfusion maps at each time point, and
relative cerebral blood volume, relative cerebral blood flow, and
relative cerebral MTT values were calculated within these regions
after normalizing to a region of gray matter in the contralateral
hemisphere.
The prespecified primary outcome was a comparison of DWI
lesion growth at 4 hours between groups. Secondary outcomes were
mean NIHSS scores and perfusion parameters at 4 hours, the
percentage of ADC voxels undergoing reversal at 4 hours or 24
hours, brain hemorrhage at 24 hours, and 3-month stroke lesion
volumes and NIHSS and mRS scores. We initially planned to enroll
40 patients in this pilot study to allow formal power calculations. The
interim analysis showed positive results, which are presented herein.
Statistical Analysis
SPSS for Windows v11.0 (SPSS) was used for the “intention to
treat” statistical analysis. All values are reported as median (range)
or meanSD. For intergroup comparisons, we applied the Student t
test, Mann–Whitney U test, or Fisher exact test; for intragroup
comparisons, we applied the paired t test or Wilcoxon rank-sum test
as appropriate. P0.05 was considered significant.
Patient Data
Characteristic
Hyperoxia
(n9)
Controls
(n7)
Age, y (mean, range) 67 (37–88) 70 (49–97)
Female 5 (56%) 4 (57%)
Stroke etiology
Cardioembolic 6 5
ICA atherosclerosis/thrombosis 3 0
ICA dissection 0 1
Cryptogenic embolism 0 1
Intravenous heparin on day 1 5 (56%) 5 (71%)
Stroke Scale Scores (median, range)
Admission NIHSS 14 (4–22) 11 (8–21)
4-h NIHSS 12 (2–15) 13 (10–26)
24-h NIHSS 6 (4–16) 15 (11–26)
1-wk NIHSS 6 (0–22) 14 (7–23)
3-mo NIHSS 3 (0–19) 13 (1–19)
Admission Scandinavian Stroke Scale 27 (6–55) 32 (2–39)
3-mo Scandinavian Stroke Scale 47 (16–60) 32 (30–56)
3-mo mRS (meanSD) 3.22.2 4.11.6
MRI Characteristics (median, range)
Time intervals
Onset to MRI-1, h 7.4 (1.6–13.4) 6.8 (3.5–8.9)
MRI-1 to MRI-2, h 4 (2.6–4.7) 4.5 (3.5–5.7)
MRI-1 to MRI-3, h* 24.4 (21.3–26.5) 25 (22.5–27.7)
MRI-1 to MRI-4, d* 6.6 (3.7–8.2) 6.2 (4.0–9.9)
MRI-1 to MRI-5, d* 99 (54–106) 116 (107–152)
Postischemic hemorrhage on MRI-2 1 (asymptomatic) 1 (fatal)
Postischemic hemorrhage on MRI-3* 4 (50%) 1 (17%)
Reperfusion
MRI-1 to MRI-2 0 (0%) 1 (14%)
MRI-2 to MRI-3* 4 (50%) 0 (0%)
*Excluding 1 patient per group with postischemic hemorrhage.
MRI-1 indicates first MRI; MRI-2, second MRI, MRI-3, third MRI; MRI-4,
fourth MRI; MRI-5, fifth MRI.
798 Stroke April 2005
by guest on September 24, 2015http://stroke.ahajournals.org/Downloaded from
Page 2
Results
We randomized 9 patients to the NBO group and 7 to the
control group. Hypoventilation did not develop in any patient.
None reported discomfort from the facemask. Mean blood
glucose, mean arterial BP at baseline, 4 hours, and 24 hours,
and anticoagulant and antiplatelet use were not significantly
different between groups. Arterial blood gases were drawn
for clinical reasons in 3 patients: the PaO
2
(mm Hg) was 368
and 420 in 2 NBO patients and was 99 in 1 control patient.
The Table shows patient characteristics. Soon after the
admission MRI, 1 control patient had a massive brain
hemorrhage and died;
20
1 NBO patient had an asymptomatic
brain hemorrhage temporally associated with a supra-
therapeutic partial thromboplastin time from intravenous
heparin treatment. Individual patient data are available online
(Appendix; see http://stroke.ahajournals.org).
Median NIHSS, SSS, and mRS scores are presented in the
Table, and intergroup comparisons of mean NIHSS scores are
shown in Figure 1A. In the NBO group, clinical improvement
was noted as early as 15 to 20 minutes after starting the
8-hour hyperoxia therapy. As compared with baseline, mean
NIHSS scores were significantly lower at 4 hours (P0.016),
24 hours (P0.03), and 3 months (P0.03).
All patients had ICA and/or proximal MCA occlusion with
substantial perfusion deficits (MTT lesion volume 90 mL in
13 of 16 patients). Mean MTT (NBO, 125.965 mL versus
control, 130.581 mL; P0.9) and DWI (NBO, 29.322
mL; control, 27.139 mL; P0.89) lesion volumes were
comparable at baseline. At 4 hours, reperfusion was evident
in 1 control patient; however, mean MTT lesion volumes
were not significantly different between groups (P0.4). At
Figure 1. A, NIHSS scores. B, Percent change in relative stroke lesion
volumes. C, Penumbral salvage or the ratio of acutely hypoperfused
tissue salvaged from infarction [(baseline MTT volume) (infarct vol-
ume at later time point)] to the acute tissue at risk for infarction [(base-
line MTT volume) (DWI volume at baseline)].
21
Controls, white bars;
NBO, black bars; meanSD.
Figure 2. Serial MRI findings in a patient with cardio-embolic
right MCA stroke treated with NBO for 8 hours. Top, Baseline
(pre-NBO) MRI, 13.1 hours after symptom onset, shows a large
DWI lesion, a larger MTT lesion, and MCA occlusion (arrow) on
head MRA. Middle, A second MRI after 3.75 hours (during NBO)
shows 36% reduction in the DWI lesion, stable MTT deficit, and
persistent MCA occlusion. Bottom, A third MRI after 24 hours
(post-NBO) shows reappearance of DWI abnormality in some
areas of previous reversal; MTT image shows partial reperfusion
(39% MTT volume reduction, mainly in the ACA territory); MRA
shows partial MCA recanalization.
Singhal et al Oxygen Therapy in Acute Stroke 799
by guest on September 24, 2015http://stroke.ahajournals.org/Downloaded from
Page 3
24 hours, 4 NBO-treated patients but no additional control
patients showed reperfusion on MRI, and mean MTT lesion
volumes were significantly lower than baseline in the NBO
group (87.848 mL versus 125.965 mL; P0.04).
Asymptomatic petechial hemorrhages were evident on
24-hour MRI scans in 4 NBO patients and in 1 control patient
(P0.6), were located in the deep MCA territory, and were
associated with arterial recanalization (3 patients) and previ-
ous microbleeds (1 patient).
At 4 hours (during therapy), relative DWI lesion volumes
decreased in 6 NBO-treated patients, with 20% reduction in
3 patients. DWI reversal was most evident in the lesion
periphery (Figure 2) and was not associated with regions of
tissue reperfusion. Among controls, only 1 patient had a
smaller DWI volume at 4 hours, and the reduction was minor
(5%). Mean relative DWI volumes were significantly smaller
in the NBO group as compared with controls at 4 hours
(87.822% versus 149.141%; P0.004), but not signifi-
cantly different at 24 hours, 1 week, and 3 months (Figure
1B). Penumbral salvage
21
was significantly higher in the
NBO group at 4 hours (Figure 1C).
Voxels showing temporary and sustained ADC reversal
were located mainly in gray matter and white matter regions
in the lesion periphery (Figure 3A). The NBO group tended to
have a higher <