Magnetic Resonance Imaging Measurement of Transmission of Arterial Pulsation to the Brain on Propranolol Versus Amlodipine
Background and Purpose—Cerebral arterial pulsatility is associated with leukoaraiosis and depends on central arterial pulsatility and arterial stiffness. The effect of antihypertensive drugs on transmission of central arterial pulsatility to the cerebral circulation is unknown, partly because of limited methods of assessment.
Methods—In a technique-development pilot study, 10 healthy volunteers were randomized to crossover treatment with amlodipine and propranolol. At baseline and on each drug, we assessed aortic (Sphygmocor) and middle cerebral artery pulsatility (TCDtranscranial ultrasound). We also performed whole-brain, 3-tesla multiband blood-oxygen level dependent magnetic resonance imaging (multiband factor 6, repetition time=0.43s), concurrent with a novel method of continuous noninvasive blood pressure monitoring. Drug effects on relationships between cardiac cycle variation in blood pressure and blood-oxygen level dependent imaging were determined (fMRI Expert Analysis Tool, fMRIB Software Library [FEAT-FSL]).
Results—Aortic pulsatility was similar on amlodipine (27.3 mm Hg) and propranolol (27.9 mm Hg, P diff=0.33), while MCA pulsatility increased nonsignificantly more from baseline on propranolol (+6%; P=0.09) than amlodipine (+1.5%; P=0.58). On magnetic resonance imaging, cardiac frequency blood pressure variations were found to be significantly more strongly associated with blood-oxygen level dependent imaging on propranolol than amlodipine.
Conclusions—We piloted a novel method of assessment of arterial pulsatility with concurrent high-frequency blood-oxygen level dependent magnetic resonance imaging and noninvasive blood pressure monitoring. This method was able to identify greater transmission of aortic pulsation on propranolol than amlodipine, which warrants further investigation.
Cerebral arterial pulsatility is associated with white matter hyperintensities1 and cognitive decline2 and is largely determined by central arterial pulsatility and arterial stiffness.1 No therapies have been specifically developed to affect cerebral arterial pulsatility, but such effects may explain the greater reduction in stroke risk with calcium channel blockers than β-blockers, despite similar blood pressure (BP) reductions.3,4
Cerebral arterial pulsatility is usually measured by transcranial Doppler,5 with good temporal but limited spatial resolution, whereas current magnetic resonance imaging (MRI) sequences have good spatial but limited temporal resolution. Furthermore, there are practical difficulties in continuous BP measurement during MRI. We piloted a novel method of continuous BP measurement during high temporal resolution MRI (multiband–blood-oxygen level dependent imaging [MB-BOLD])6,7 and determined its potential utility by assessing transmission of arterial pulsations on amlodipine versus propranolol.
Ten healthy adult subjects were randomized (according to CAMARADES recommendations for nonclinical studies8) to 1 week of daily amlodipine 10 mg or propranolol-LA 160 mg in a crossover design, with a 2-week washout. This physiological protocol was assessed by the Medicines and Healthcare products Regulatory Authority. At baseline and on each drug, carotid–femoral pulse wave velocity and aortic BP (Sphygmocor) were measured.1 Transcranial Doppler ultrasound (DWL-Doppler Box) was performed with a handheld 2 MHz probe on the same side as carotid applanation, at 50 mm or at the depth of the optimal waveform. Gosling’s pulsatility index (middle cerebral artery–pulsatility index, MCA-PI=(systolic cerebral blood flow velocity-diastolic cerebral blood flow velocity)/mean cerebral blood flow velocity) and MCA transit time were calculated.1 All waveforms were visually inspected.
On a 3T Siemens Verio scanner, a volume-acquisition T1 multiplanar reconstruction (1.5×1.5×1.5 mm voxels) and a 12-minute multiband BOLD-MRI (multiband factor=6, 30 slices, 3×3×3 mm voxels, echo time=40 ms, repetition time=0.43 s; Figure I in the online-only Data Supplement)6,7 were acquired. Continuous, noninvasive brachial BP was simultaneously acquired by a novel method (Figure 1). Cardiac cycles were marked at the maximum of the second differential of BP.
Multiband BOLD sequences were motion-corrected (motion correction FMRIB linear image registration tool [MCFLIRT]–FSL) to a presaturated BOLD volume, spatially smoothed (fMRI Expert Analysis Tool, fMRIB Software Library [FEAT-FSL]9), registered to T1 (FMRIB non-linear image registration tool [FNIRT]-FSL9), and then the MNI-152 brain for group analysis (FNIRT-FSL9). Nonphysiological artefactual components were identified and removed manually by independent component analysis.9 Voxel-to-voxel differences in pulse arrival time were measured by event-related summation of each time series to the peripheral BP marker (Figure 2), phase shifting voxels by differences in peak arrival time with interpolation by piecewise cubic hermitte interpolation.
For each voxel, power spectra (Welch, 350 volume segments, 50% overlap, nfft 512 volumes) and mean-squared coherence with BP were derived. Correlation between cardiac frequency BP and BOLD signal was determined on each treatment and compared between treatments (FSL-FEAT).
Of 5 men and 5 women (median age=29, range=18–41) recruited, all completed the protocol, half receiving amlodipine first. Both drugs reduced aortic BP, pulsatility and pulse wave velocity, and cerebral blood flow velocity similarly (Table), although MCA-PI increased nonsignificantly more with propranolol.
The coherence (frequency-specific relationship) between the BOLD signal and the peripheral BP at the cardiac cycle frequency was greatest in the ventricles and venous sinuses, but was also present throughout gray matter. Averaging BOLD responses for each voxel across all cardiac cycles produced identifiable arterial waveforms (Figure 1). The peripheral cardiac cycle frequency BP waveform was more strongly associated with BOLD signal in gray matter on propranolol than amlodipine (Figure 2). This was unchanged when excluding individual subjects from the analysis.
We simultaneously acquired noninvasive, continuous BP and high-frequency BOLD MRI, demonstrating a direct relationship at the frequency of the cardiac cycle. This relationship was stronger on propranolol than amlodipine, despite similar effects on aortic BP and pulsatility.
Cerebral artery pulsatility is associated with chronic white matter disease,1,3 potentially because of increased transmission of aortic pulsatility to the brain through stiff vessels.1 However, investigation of dynamic cerebral blood flow changes is limited by poor temporal resolution of standard MRI sequences, low spatial resolution of transcranial Doppler, and practical difficulties in continuous BP measurement during MRI scanning. We used a recently developed high-frequency MRI sequence6,7 and developed a novel method of concurrent, continuous BP monitoring. With refinement, this technique could allow detailed assessment of transmission of rapid fluctuations in systemic BP on region-specific cerebral blood flow. Indeed, we found a stronger association with the cardiac cycle waveform on propranolol than on amlodipine. This may reflect less dampening of systemic BP which could expose the brain to greater arterial pulsatility. This is a potential explanation for differences in cerebrovascular physiology and stroke risk between the 2 drugs4,5,10 that warrants further investigation.
Our study has several limitations. First, subjects were healthy volunteers with less arteriopathy than more elderly patients at a greater risk of stroke. As such, the effects of drugs in this study cannot be extrapolated to clinical populations. Second, BOLD is affected by blood flow, blood oxygenation, and blood volume. However, the associations we demonstrated with cardiac frequency BP variation reflect BOLD variation at a higher frequency than neurovascular coupling and is therefore likely to be dependent primarily on blood flow. Third, the BP measurement method is susceptible to artefactual slow drifts in BP, which were filtered out offline. This has minimal impact on the high-frequency BP fluctuations we addressed, but limits the technique for investigating slower fluctuations in BP. Fourth, the stronger association between cardiac cycle frequency BP fluctuations and the BOLD signal on propranolol superficially follows a different pattern to the distribution of greatest cerebral pulsation, likely reflecting highest absolute brain perfusion in the gray matter and limiting the sensitivity of the analysis for effects on white matter perfusion. This may reflect correlations between BP and BOLD not directly dependent on the magnitude of BP pulsatility, but this pattern of cerebral pulsation has also been demonstrated using a surrogate of peripheral BP.11 Finally, given a repetition time=0.43, heart rates above 70 bpm result in aliasing of the cardiac pulsation. Only one subject had an excess mean resting heart rate, and excluding this individual did not alter the results (data not shown). However, in a broader population, multiband imaging with a shorter TR11 would limit aliasing.
We piloted a novel method of concurrent, continuous noninvasive BP measurement during high-frequency BOLD MRI to assess transmission of arterial pulsatility to the brain, demonstrating a stronger association on propranolol than amlodipine. This needs further development but with refinement could enable systematic MRI-based assessment of rapid BP fluctuations effects on the cerebral circulation.
We gratefully acknowledge the Cardiovascular Clinical Research Facility, the Acute Vascular Imaging Centre, and Michael Kelly from the Oxford Centre for functional Magnetic Resonance Imaging of the Brain (fMRIB).
Sources of Funding
P.M. Rothwell has National Institute for Health Research and Wellcome Trust Senior Investigator Awards and receives funding from the Oxford Biomedical Research Centre. A.J.S. Webb had a Medical Research Council Clinical Training Research Fellowship.
The multiband-BOLD pulse sequence was used under an agreement with the sequence developers at the University of Minnesota. The authors report no conflicts.
Guest Editor for this article was Christopher L.H. Chen, FRCP.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.115.012411/-/DC1.
- Received December 11, 2015.
- Revision received March 2, 2016.
- Accepted March 21, 2016.
- © 2016 American Heart Association, Inc.
- Webb AJ,
- Simoni M,
- Mazzucco S,
- Kuker W,
- Schulz U,
- Rothwell PM
- Lee KY,
- Sohn YH,
- Baik JS,
- Kim GW,
- Kim JS
- Rothwell PM,
- Howard SC,
- Dolan E,
- O’Brien E,
- Dobson JE,
- Dahlöf B,
- et al
- Webb AJ,
- Rothwell PM
- Moeller S,
- Yacoub E,
- Olman CA,
- Auerbach E,
- Strupp J,
- Harel N,
- et al