Accepted Manuscript Title: MRI Evidence for Preserved Regulation of Intracranial Pressure in Patients with Cerebral Arteriovenous Malformations Author: Felix G. Meinel Judith Fischer Andreas Pomschar Natalie W¨ohrle Inga K. Koerte Denise Steffinger R¨udiger P. Laubender Alexander Muacevic Maximilian F. Reiser Noam Alperin Birgit Ertl-Wagner PII: DOI: Reference:
S0720-048X(14)00247-2 http://dx.doi.org/doi:10.1016/j.ejrad.2014.05.011 EURR 6781
To appear in:
European Journal of Radiology
Received date: Accepted date:
21-3-2014 6-5-2014
Please cite this article as: Meinel FG, Fischer J, Pomschar A, W¨ohrle N, Koerte IK, Steffinger D, Laubender RP, Muacevic A, Reiser MF, Alperin N, Ertl-Wagner B, MRI Evidence for Preserved Regulation of Intracranial Pressure in Patients with Cerebral Arteriovenous Malformations, European Journal of Radiology (2014), http://dx.doi.org/10.1016/j.ejrad.2014.05.011 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
*Conflict of Interest
Disclosure of potential conflicts of interest:
Ac ce p
te
d
M
an
us
cr
ip t
Noam Alperin is a shareholder of Alperin Noninvasive Diagnostics, Inc.
Page 1 of 29
*Title Page (including paper title & Complete corresponding author details)
MRI Evidence for Preserved Regulation of Intracranial Pressure in Patients with Cerebral Arteriovenous Malformations
ip t
Felix G. Meinel1, Judith Fischer1, Andreas Pomschar1, Natalie Wöhrle1, Inga K. Koerte1, Denise Steffinger1, Rüdiger P. Laubender2, Alexander Muacevic3, Maximilian F. Reiser1, Noam Alperin4
cr
& Birgit Ertl-Wagner1
1
us
Authors’ Affiliations
Institute for Clinical Radiology, Ludwig-Maximilians-University Hospital, Marchioninistr. 15,
2
an
81377 Munich, Germany.
Institute of Medical Informatics, Biometry and Epidemiology, Ludwig-Maximilians-University,
M
Marchioninistr. 15, 81377 Munich, Germany.
European Cyberknife Center Munich, 81377 Munich, Germany.
4
Department of Radiology, Miller School of Medicine, University of Miami, Miami, FL 33136,
d
3
Ac ce p
te
USA.
Corresponding Author:
Birgit Ertl-Wagner, M.D.
Institute of Clinical Radiology
Ludwig-Maximilians-University, Großhadern Campus, Marchioninistr. 15
81377 Munich, Germany.
[email protected]
Page 2 of 29
Abstract Purpose The purpose of this study was to investigate intracranial pressure and associated hemo- and
ip t
hydrodynamic parameters in patients with cerebral arteriovenous malformations AVMs.
cr
Methods
us
Thirty consecutive patients with arteriovenous malformations (median age 38.7 years, 27/30 previously treated with radiosurgery) and 30 age- and gender-matched healthy controls
an
were investigated on a 3.0 Tesla MR scanner. Nidus volume was quantified on dynamic MR
M
angiography. Total arterial cerebral blood flow (tCBF), venous outflow as well as aqueductal and craniospinal stroke volumes were obtained using velocity-encoded cine-phase contrast
ed
MRI. Intracranial volume change during the cardiac cycle was calculated and intracranial pressure (ICP) was derived from systolic intracranial volume change (ICVC) and pulse
Results
ce pt
pressure gradient.
TCBF was significantly higher in AVM patients as compared to healthy controls (median 799
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
vs. 692 mL/min, p=0.007). There was a trend for venous flow to be increased in both the ipsilateral internal jugular vein (IJV, 282 vs. 225 mL/min, p=0.16), and in the contralateral IJV (322 vs. 285 mL/min, p=0.09), but not in secondary veins. There was no significant difference in median ICP between AVM patients and control subjects (6.9 vs. 8.6 mmHg, p=0.30) and ICP did not correlate with nidus volume in AVM patients (ρ=-0.06, p=0.74). There was a
2 Page 3 of 29
significant positive correlation between tCBF and craniospinal CSF stroke volume (ρ=0.69, p=0.02).
Conclusions
ip t
The elevated cerebral blood flow in patients with AVMs is drained through an increased flow
cr
in IJVs but not secondary veins. ICP is maintained within ranges of normal and does not
an
us
correlate with nidus volume.
Key words
ce pt
ed
M
Arteriovenous Malformations; MRI; phase-contrast MRI; blood flow; cerebrospinal fluid.
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
3 Page 4 of 29
List of Abbreviations Arteriovenous malformations
CSF
Craniospinal fluid
ICP
Intracranial pressure
ICVC
Intracranial volume change
tCBF
Total (arterial) cerebral blood flow
ICAs
Internal carotid arteries
VAs
Vertebral arteries and
IJVs
Internal jugular veins
VVs
Vertebral veins
EVs
Epidural veins
DCVs
Deep cervical veins
ce pt
ed
M
an
us
cr
ip t
AVMs
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
4 Page 5 of 29
Introduction It has been shown that cerebral AVMs are associated with a local dysfunction in autoregulation of cerebral blood flow.1,
2
Vascular injury, abnormal endothelial signaling,
microshunt formation, and venous hypertension have been identified as potential 3
The dysbalance of
ip t
mechanisms of impaired autoregulation in cerebral AVMs.2,
cr
autoregulation may drive growth and remodeling of AVMs.2 Other authors have hypothesized that cerebral AVMs may originate as a compensatory response to a local
us
imbalance in autoregulation of blood flow.4 A recent study suggests that impaired
an
cerebrovascular reserve in peri-nidal brain areas associated with venous congestion may account for seizures.5 Other authors, however, have described preserved autoregulatory
M
responsiveness in brain areas adjacent to AVMs.6, 7
ed
While there is extensive research about local autoregulation in perinidal areas, little is known about how cerebral AVMs affect global cerebral autoregulation and ICP. Papilledema
ce pt
caused by intracranial hypertension is a rare but recognized presentation of unruptured cerebral AVMs.8-12 High-volume shunts may potentially exhaust the venous drainage capacity and thus lead to congestion of the venous system, associated with an increase in cerebral blood volume, impairment of CSF absorption, and increase in CSF production.10, 13
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
Direct mass effect may be another mechanism of AVM-associated intracranial hypertension.9,
10
In one recent case series, 3 out of 4 cases of intracranial hypertension
(identified as papilledema) associated with cerebral AVMs were considered to be due to overload of draining veins, while 1 case was caused by thrombotic occlusion of the straight sinus.8 All four cases occurred in patients with large cerebral AVMs. The diagnosis of intracranial hypertension had immediate therapeutic consequences in these patients, as partial endovascular embolization was performed and immediately reduced the massive 5 Page 6 of 29
overload of the venous system and thus relieved neurological symptoms.8 Identifying patients with cerebral AVMs at risk for intracranial hypertension and monitoring their ICP may therefore be of significant clinical relevance in order to prevent serious complications. We aimed to assess how the presence of a cerebral AVM affects ICP depending on nidus
ip t
volume. Moreover, we aimed to analyze relevant hydro- and hemodynamic parameters that
ce pt
ed
M
an
us
cr
may compensate the increased flow in the AVM.
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
6 Page 7 of 29
Methods Informed Consent and Ethical Approval Institutional Review Board approval was obtained prior to the commencement of the study.
ip t
Written consent was obtained from all patients and healthy individuals (parents or legal
cr
guardians for minors) prior to enrollment.
us
Patient Population and Control Subjects
an
30 consecutive patients with cerebral AVMs referred to MR imaging were enrolled in the study. All examinations were performed for clinical indications. Both treatment-naïve
M
patients and patients previously treated with radiosurgery were included. For each patient, a healthy individual of the same gender and with no more than 12 months of difference in age
ed
served as a paired reference. To minimize confounding of perfusion parameters, exclusion criteria for both patients and control participants included a history of diabetes, vascular
ce pt
disease, stroke or intracranial hemorrhage. General exclusion criteria for MR imaging included claustrophobia, ferromagnetic implants, cochlear implants and cardiac pacemakers.
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
MRI Examination Protocol All subjects underwent MR imaging on a 3-Tesla MR scanner (Magnetom Verio, Siemens Healthcare, Erlangen, Germany), using a 12-element phased-array head coil. 3D magnetization-prepared rapid-acquisition gradient echo imaging (MP-RAGE) was obtained for structural information using the following parameters: repetition time (TR) = 11 ms; echo time (TE) = 4.76 ms; field of view (FOV) = 230 mm; voxel size = 0.9 x 0.9 x 0.9 mm 3. 160 7 Page 8 of 29
sagittal slices parallel to the falx cerebri were acquired, covering the entire brain. The data were acquired with a parallel acquisition technique (iPAT, acceleration factor 2). A dynamic contrast-enhanced angiography sequence (TWIST) was obtained using the following
ms, flip angle 21°, base resolution 256 x 256, phase resolution 80%.
ip t
acquisition parameters: field of view 280 mm, slice thickness 2.0 mm, TR 2.33 ms, TE 0.94
cr
In order to obtain blood flow to and from the brain, two retrospectively gated, velocity-
us
encoded cine-phase contrast scans were performed (FOV = 140 mm, matrix = 256 x 179, voxel size = 0.8 x 0.5 x 6 mm, TR = 40 ms, TE 4.05 ms, FA = 20 degrees, acquisition time = 32
an
cardiac cycles equaling approximately 3 minutes as determined by the individual heart rate). First, a high-velocity encoding (70 cm/s) was used to quantify the high-velocity blood flow in
M
the internal carotid arteries (ICAs), vertebral arteries (VAs), and internal jugular veins (IJVs). The sequence was positioned at the upper level of the 2nd cervical vertebra with an
ed
orientation perpendicular to the main four arteries (ICAs and VAs) and the IJVs. To assess the
ce pt
low-velocity secondary venous flow, a sequence with low-velocity encoding (7–9 cm/s) was applied at the same level. To visualize primary and secondary venous channels, a 2D time-offlight MR venography of the infratentorial and upper cervical regions was performed (slice thickness 2.0 mm, FOV 160 mm, TR 23 ms, TE 5.43 ms and FA 40°).
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
Post-processing and Data Analysis Before performing quantitative analysis, all MRI datasets were reviewed to confirm diagnostic image quality. 32 images of the pulsatile flow per cardiac cycle were derived from the applied cine-phase contrast sequence. Time-dependent volumetric flow rates were calculated by integrating the flow velocities inside the luminal cross-sectional areas over all 8 Page 9 of 29
32 images, using the pulsatility-based segmentation of lumens conducting non-steady flow algorithm. Mean flow rates were obtained for each of the four main cervical arteries (left and right ICAs and left and right VAs), for the primary venous pathways (left and right IJVs), and the secondary venous pathways including vertebral veins (VVs), epidural veins (EVs), and
ip t
deep cervical veins (DCVs).
cr
Flow waveforms were obtained for each of the four main cervical arteries (left and right
us
internal carotid artery (LICA, RICA) and left and right vertebral artery (LVA, RVA), for the primary venous pathways, the left and right internal jugular vein (LIJV, RIJV), and for the
an
secondary venous pathways, the vertebral veins (VV), epidural veins (EV), and deep cervical veins (DCV). In addition, cervical CSF stroke volume, i.e., the volume of CSF that flows back
M
and forth between the cranium and the spinal canal, and aqueductal CSF stroke volume were obtained by time integration of the CSF flow waveforms. Total arterial cerebral blood
ed
flow (tCBF) was obtained by summation of the flows through the four arteries supplying the
ce pt
brain (left and right ICA, left and right VA). Flow in the basilar artery was calculated as the sum of the flow in both VAs. Secondary venous flow was defined as the sum of the flow through the three main secondary venous channels (VVs, EVs, and DCVs). Total venous outflow was determined as the total flow in all detected veins (IJVs and secondary veins).
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
Details of the derivation of the intracranial compliance and pressure have been described previously14, 15. Briefly, based on the physical definition of compliance as a ratio of volume and pressure changes, intracranial compliance is estimated from the ratio of the maximal (systolic) intracranial volume and pressure fluctuations during the cardiac cycle. The change in intracranial volume (ICVC) is obtained from the momentary differences between volumes of blood and CSF entering and leaving the cranium as shown in equations 1 and 2,
9 Page 10 of 29
ICVC (i) [ f A (i) f V (i) f CSF (i)] t
ICVC (i) [ f
cardiac cycle
A
(1)
(i) f V (i ) f CSF (i )] t 0
(2)
cardiac cycle
ip t
f f where f A is arterial inflow, V is venous outflow, and CSF is the craniospinal CSF flow. Equation 2 states that in steady state, the intracranial volume is on average constant over an
cr
entire cardiac cycle.
us
The pressure change is derived from the amplitude of the pressure gradient (PG) waveform
an
obtained using the Navier-Stokes relationships between derivatives of CSF velocities and the CSF pressure gradient. An MRI equivalent of ICP (MRICP) is then obtained based on the
M
reported inverse relationship between compliance and ICP16. Volumetric blood and CSF flow rate waveforms and derived parameters were obtained using a dedicated software tool
ed
(MRICP version 1.4.35 Alperin Noninvasive Diagnostics, Miami, FL).
ce pt
All MR images were reviewed by a board-certified neuroradiologist with >12 years of experience in neuroradiology and MR angiography (initials blinded). All images were carefully reviewed to detect any potential confounders such as evidence of microangiopathy, stroke, intracranial hemorrhage, gliosis or parenchymal defect zones. AVM
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
nidus size was measured in three dimensions on dynamic MR angiographic datasets (TWIST sequence); the time point, in which the nidal volume was largest, was chosen for the measurements. AVM nidus volume was calculated using the spheric volume formula 4/3 x π x (l/2) x (h/2) x (w/2) with l, h, and w being the length, height and width of the AVM nidus. As an approximation, the three-dimensional shape of the gliar scar surrounding the AVM nidus can be regarded as a spherical shell, i. e. the region between two concentric spheres of differing radii. Therefore, the maximum thickness of the glial scar around the nidus t and the 10 Page 11 of 29
maximum diameter oft the lesion including nidus and glial scar d was measured. Glial scare volume was then calculated using the volume formula for a spherical shell as 4/3 x π x (d/2)3 – 4/3 x π x (d/2 – t)3. For parenchymal defect zones adjacent to the AVM nidus, the
cr
calculated using the spheric volume formula 4/3 x π x (d/2)3.
ip t
maximum diameter of the defect d was measured and the approximate volume was
us
Statistical Analysis
Data were summarized by properly chosen measures of location and spread for continuous
an
variables and by proportions for discrete variables. Correlations between variables were
M
estimated by Spearman’s correlation coefficient ρ. Linear mixed effects regression models were used to test for differences between AVM patients and healthy controls. For
ed
aqueductal stroke volume we further separately evaluated AVMs with supratentorial location. In these models we allowed the intercept to vary by the matched subjects (random
ce pt
effects term). For AVM patients, the impact of AVM nidus volume and total cerebral blood flow on flow parameters was estimated and adjusted for the volume of parenchymal defects and for the volume of glial scar by multivariate linear regression analysis. For all regression models the corresponding dependent variables were transformed by appropriately chosen
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
powers or by the natural logarithm to satisfy the normality assumption. A two-sided level of significance of α=0.05 was applied. All statistical analyses were performed using R (version 2.13.0).
11 Page 12 of 29
Results Thirty consecutive patients were included into the data analysis (table 1). Two patients underwent MRI as part of their evaluation prior to stereotactic radiosurgery. 28 patients were referred for MR imaging for follow-up after stereotactic radiosurgery. In median, the
ip t
radiosurgical procedure had been performed 20 (interquartile range 26) months prior to the
cr
MRI examination. All patients received a single procedure using CyberKnife radiosurgery
us
(Accuray, Inc., Sunnyvale, CA) with a median dose of 18.5 Gy (range 15-24 Gy). 12 (40%) patients were female. Median age was 38.7 years for AVM patients and 38.1 years for
an
control subjects. Ages ranged from 17 to 62 years.
AVM nidus volumes ranged from 0.0 to 51.5 cm3 with a median of 0.8 cm3 and an
M
interquartile range of 3.4 cm3 (See Table 1). 7/30 patients did not have a measurable perfused nidus volume due to successful radiosurgical treatment. 9/30 patients (30%) had an
ed
AVM nidus volume >2 cm3. Most frequently, AVMs were located in the frontal (n=10),
ce pt
parietal (n=6), and temporal lobe (n=6). AVMs were also found in the occipital lobe, thalamus, brainstem (n=2 each) and cerebellum (n=1). In 19 cases, the remaining AVM nidus was surrounded by a glial scar. The median glial scar volume was 5.2 cm3 with a range of 0.1 to 49.9 cm3. 13 patients had developed a parenchymal defect with a median volume 2.6 cm3
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
with a range of 0.1 to 54.4 cm3. TCBF to the brain was found to be significantly higher in AVM patients as compared to healthy controls (median 799 vs. 692 mL/min, p=0.007; see Table 2). In AVM patients, tCBF ranged from 566 to 2222mL/min as compared to a range of 502 to 903 mL/min in healthy individuals. Specifically, arterial flow was significantly higher in the internal carotid artery (ICA) on the side of the AVM localization (293 vs. 248 mL/min, p=0.02). In contrast, there 12 Page 13 of 29
were no significant differences in arterial blood flow in the contralateral ICA between patients and healthy volunteers (255 vs. 256 mL/min, p=0.20) and the basilar artery (221 vs. 200, p=0.05). In AVM patients, there was a trend for flow to be higher in the ICA ipsilateral to the AVM nidus than in the contralateral ICA (293 vs. 255 mL/min, p=0.10). AVM nidus
ip t
volume correlated positively with total cerebral arterial blood flow (ρ=0.43, p=0.02).
cr
Corresponding to the increased arterial blood flow, median total venous outflow (TVO) was significantly higher in AVM patients than in healthy participants (675 vs. 567 mL/min,
us
p=0.007, see Table 2). The recorded median TVO values were 13-19% lower than tCBF. There
an
was a trend for venous flow to be increased in both the ipsilateral internal jugular vein (IJV, 282 vs. 225 mL/min, p=0.16), and in the contralateral IJV (322 vs. 285 mL/min, p=0.09),
M
although these differences did not reach the level of significance. There was no increase in
38 mL/min, p=0.28).
ed
the venous drainage by secondary veins (epidural, vertebral and deep cervical veins; 23 vs.
ce pt
There was a trend towards a positive correlation between AVM nidus volume and total venous outflow (ρ=0.34, p=0.06) as well as contralateral IJV flow (ρ=0.41, p=0.02; see Table 3). No relevant correlation was found between nidus volume and blood flow in ipsilateral IJV (ρ=0.11, p=0.56) and secondary veins (ρ=0.04, p=0.83).
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
In AVM patients, ICP ranged from 3.7 to 16.1 mmHg. A very similar range of 3.3 to 15.7 mmHg was found in the control group. There was no significant difference in median ICP between AVM patients and healthy subjects (6.9 vs. 8.6 mmHg, p=0.30, see Table 2) and ICP did not correlate with nidus volume (ρ=-0.06, p=0.74; see Table 3), tCBF (ρ=0.13, p=0.50; Table 4).
13 Page 14 of 29
There were no significant differences between AVM patients and the control group regarding CSF stroke volume, intracranial volume change or aqueductal stroke volume during the cardiac cycle (See Table 2). Moreover, these parameters showed no significant correlation with nidus volume (See Table 3). No correlation of intracranial volume change or
ip t
aqueductal stroke volume with TCBF was observed (Table 4). There was, however, a significant correlation between TCBF and CSF stroke volume in AVM patients (ρ=0.69,
cr
p=0.02; see Table 4). For aqueductal stroke volume we further separately evaluated AVMs
us
with supratentorial location (n=27). Again, no significant difference between AVM patients
ce pt
ed
M
an
and healthy individuals (p=0.27) was found.
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
14 Page 15 of 29
Discussion Unruptured cerebral AVMs can cause intracranial hypertension by high shunt volumes exhausting the venous drainage capacity or by direct mass effect.8-12 Since AVM-induced intracranial hypertension alters management8, it is desirable to identify AVM patients at risk
ip t
for intracranial hypertension. To our knowledge this is the first systematic study evaluating
cr
intracranial pressure in patients with AVM in relation to AVM nidus volume and cerebral
us
flow parameters.
In agreement with other studies2, 10, 13, we found a significantly elevated tCBF in patients
an
with AVMs. AVM patients had a significantly higher flow in their ipsilateral ICA and basilar artery but not in their contralateral ICA. Contralateral ICA flow did, however, show a
M
moderate but significant positive correlation with AVM nidus size indicating that AV shunting
ed
recruits blood volume from the circle of Willis thus moderately increasing flow in the contralateral ICA. These results are in line with a recent report that ipsilateral ICA flow is
ce pt
significantly higher than contralateral ICA flow in AVM patients and that this difference is leveled out 2-4 years after radiosurgery17. We have further shown that the increased cerebral blood flow is drained by an increased flow in the ipsilateral and contralateral IJVs but not by an enhanced secondary venous flow.
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
Overall, total venous outflow from the brain must equal total arterial cerebral blood flow. We determined TVO by adding blood flow in both internal jugular veins and all identified secondary veins. Thus, recorded median TVO values were 13-19% lower than tCBF. This is most likely due to small secondary veins that are missed when manually placing regions of interest into all discernible epidural, vertebral and deep cervical veins. Similar rates of
15 Page 16 of 29
unmeasured secondary venous outflow have been reported in other studies 18 and are likely due to drainage through smaller secondary veins not measured (eg, ocular veins). In our study, intracranial pressure showed no correlation with AVM nidus size and the range of ICP values was virtually equal in AVMs patients and healthy individuals. This may be due
ip t
to nidus sizes and / or flow rates too small to induce intracranial hypertension. In a recent
cr
case series 3 out of 4 patients with AVM-induced intracranial hypertension had “large” AVMs (volume not further specified)8. However, our study did include large AVMs with nidus
us
volumes up to 51.5 cm3 and total cerebral blood flow as high as 2222 mL/min. Even if the
an
number of large AVMs in our study was relatively low with only 9/30 patients (30%) showing an AVM nidus volume >2 cm3, we would have expected to see at least a trend if such a
M
correlation existed. Although large nidus size may be a predisposing factor for intracranial hypertension, other factors such as the AVM nidus angioarchitecture, reduced venous
ed
drainage capacity or altered CSF production and absorption may play a more important role. Given the rare incidence of intracranial hypertension in AVM patients, larger studies or case-
ce pt
control-studies are needed to identify such predisposing factors. Because compliance of the spinal canal exceeds intracranial compliance, a small but measurable amount of CSF is shifted through the foramen magnum into the spinal canal
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
during each cardiac cycle. Similarly, there is a small pulsatile CSF shift through the cerebral aqueduct into the fourth ventricle and subarachnoid space.19 We had hypothesized AVM patients to show a shunt volume-dependent increase in craniospinal CSF shift in order to maintain constant intracranial pressure despite their elevated cerebral blood flow. Surprisingly, we found no significant differences between AVM patients and healthy individuals in aqueductal and craniospinal CSF stroke volume of intracranial volume change and these parameters showed no correlation with nidus volume. There was, however, a 16 Page 17 of 29
significant correlation of total cerebral blood flow with CSF stroke volume in AVM patients. Therefore, increased craniospinal CSF pulsation may be a compensatory mechanism in order to avoid pulsatile elevations in ICP in AVM patients with high cerebral blood flow. One limitation of our study is that all but two patients had been previously treated with
ip t
stereotactic radiosurgery. As reported above, patients in our study were examined at a
cr
median of 20 months after undergoing radiosurgical treatment. Considering that obliteration of the AVM nidus is usually completed within 1-3 years after radiosurgery20-24, most patients
us
showed at least partially obliterated nidi and a varying degree of reactive changes such as
an
gliosis or parenchymal defects. Parenchymal defect zones and glial scars represent areas of hypoperfusion and therefore represent a potential confounder of blood flow parameters.
M
However, we did find a significantly higher tCBF in AVM patients than in healthy individuals and tCBF correlated with nidus volume. Therefore, a significant part of our patient
ed
population had significant residual nidi with intact AV shunting. Also, correlation of nidus volume with cerebral flow parameters was adjusted for potential confounding by glial scars
ce pt
and defect zones using multivariate linear regression analysis. Neither glial scar nor defect zone volume significantly influenced flow parameters and intracranial pressure. Therefore, we expect similar results regarding tCBF and intracranial pressure in an untreated patient population. On the other hand, it has recently been shown that the circle of Willis shows
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
significant remodeling with altered flow patterns after stereotactic surgery in patients with cerebral AVMs.25 This remodeling may influence the relative flow distribution among the three major arteries supplying the brain as assessed in our study. In this respect, it is possible that our results are specific for patients after radiosurgery and may not represent flow patterns in treatment-naïve patients.
17 Page 18 of 29
Conclusions In conclusion, we have shown that the elevated cerebral blood flow in patients with AVMs is drained through an increased flow in IJVs but not secondary venous flow. ICP did not correlate with nidus volume in our patient cohort. Large nidus volume may be one risk factor
ip t
but other determinants need to be identified in further studies to identify those AVM
ce pt
ed
M
an
us
cr
patients at risk for intracranial hypertension.
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
18 Page 19 of 29
Tables Table 1. Characteristics of Study Participants Healthy Controls
30
30
Number of Subjects Age [years]
ip t
AVM Patients
38.7 (17.1) 12 (40%)
12 (40%)
cr
Females (% of participants)
38.1 (17.1)
n =28, stereotactic radiosurgery n=2, none
n/a
20 (26)
n/a
us
Prior therapy
an
Time after Radiosurgery [months]
n=10, frontal lobe n=6, parietal lobe n=7, temporal lobe n=2, occipital lobe n=2, thalamus n=1, cerebellum n=2, brain stem
n/a
0.8 (3.4)
n/a
5.2 (12.3)
n/a
2.6 (24.0)
n/a
Nidus Volume [cm3] (n=30)
ce pt
Glial Scar Volume [cm3] (n=19)
ed
M
Main Localization of AVM Nidus
Parenchymal Defect Volume [cm3] (n=13)
Table 1. Key characteristics of AVM patients and healthy controls are listed. Unless indicated
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
otherwise, median values are provided followed by interquartile ranges in parentheses.
19 Page 20 of 29
Table 2. Cerebral Flow Parameters in AVM patients and Healthy Controls Healthy Controls
p-value
799 (153)
692 (135)
0.007
Ipsilateral ICA
293 (98)
248 (60)
0.02
Contralateral ICA
255 (69)
256 (63)
Basilar Artery
221 (69)
200 (47)
675 (273)
567 (134)
0.007
Ipsilateral IJV
282 (240)
225 (267)
(0.16)
Contralateral IJV
322 (288)
285 (247)
(0.09)
Secondary Veins
23 (56)
38 (77)
(0.28)
6.9 (4.1)
8.6 (4.7)
(0.30)
0.54 (0.30)
0.50 (0.20)
(0.88)
0.56 (0.30)
0.54 (0.23)
(0.46)
33.7 (16.6)
(0.16)
Intracranial Pressure [mmHg] CSF Stroke Volume [mL]
ed
Intracranial Volume Change [mL]
M
cr
an
Total Venous Outflow
us
Total Cerebral Arterial Blood Flow
43.4 (40.1)
(0.20)
(0.05)
ce pt
Aqueductal Stroke Volume [μL]
ip t
AVM Patients
Table 2. Cerebral flow parameters of AVM patients (n=30) are compared to age- and gendermatched healthy controls (n=29). A linear mixed-effects model was used to test for differences between both groups. Median flow values in mL/min are provided followed by interquartile ranges in parentheses. P-values which did not meet the level of statistical
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
significance (p
S0720-048X(14)00247-2 http://dx.doi.org/doi:10.1016/j.ejrad.2014.05.011 EURR 6781
To appear in:
European Journal of Radiology
Received date: Accepted date:
21-3-2014 6-5-2014
Please cite this article as: Meinel FG, Fischer J, Pomschar A, W¨ohrle N, Koerte IK, Steffinger D, Laubender RP, Muacevic A, Reiser MF, Alperin N, Ertl-Wagner B, MRI Evidence for Preserved Regulation of Intracranial Pressure in Patients with Cerebral Arteriovenous Malformations, European Journal of Radiology (2014), http://dx.doi.org/10.1016/j.ejrad.2014.05.011 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
*Conflict of Interest
Disclosure of potential conflicts of interest:
Ac ce p
te
d
M
an
us
cr
ip t
Noam Alperin is a shareholder of Alperin Noninvasive Diagnostics, Inc.
Page 1 of 29
*Title Page (including paper title & Complete corresponding author details)
MRI Evidence for Preserved Regulation of Intracranial Pressure in Patients with Cerebral Arteriovenous Malformations
ip t
Felix G. Meinel1, Judith Fischer1, Andreas Pomschar1, Natalie Wöhrle1, Inga K. Koerte1, Denise Steffinger1, Rüdiger P. Laubender2, Alexander Muacevic3, Maximilian F. Reiser1, Noam Alperin4
cr
& Birgit Ertl-Wagner1
1
us
Authors’ Affiliations
Institute for Clinical Radiology, Ludwig-Maximilians-University Hospital, Marchioninistr. 15,
2
an
81377 Munich, Germany.
Institute of Medical Informatics, Biometry and Epidemiology, Ludwig-Maximilians-University,
M
Marchioninistr. 15, 81377 Munich, Germany.
European Cyberknife Center Munich, 81377 Munich, Germany.
4
Department of Radiology, Miller School of Medicine, University of Miami, Miami, FL 33136,
d
3
Ac ce p
te
USA.
Corresponding Author:
Birgit Ertl-Wagner, M.D.
Institute of Clinical Radiology
Ludwig-Maximilians-University, Großhadern Campus, Marchioninistr. 15
81377 Munich, Germany.
[email protected]
Page 2 of 29
Abstract Purpose The purpose of this study was to investigate intracranial pressure and associated hemo- and
ip t
hydrodynamic parameters in patients with cerebral arteriovenous malformations AVMs.
cr
Methods
us
Thirty consecutive patients with arteriovenous malformations (median age 38.7 years, 27/30 previously treated with radiosurgery) and 30 age- and gender-matched healthy controls
an
were investigated on a 3.0 Tesla MR scanner. Nidus volume was quantified on dynamic MR
M
angiography. Total arterial cerebral blood flow (tCBF), venous outflow as well as aqueductal and craniospinal stroke volumes were obtained using velocity-encoded cine-phase contrast
ed
MRI. Intracranial volume change during the cardiac cycle was calculated and intracranial pressure (ICP) was derived from systolic intracranial volume change (ICVC) and pulse
Results
ce pt
pressure gradient.
TCBF was significantly higher in AVM patients as compared to healthy controls (median 799
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
vs. 692 mL/min, p=0.007). There was a trend for venous flow to be increased in both the ipsilateral internal jugular vein (IJV, 282 vs. 225 mL/min, p=0.16), and in the contralateral IJV (322 vs. 285 mL/min, p=0.09), but not in secondary veins. There was no significant difference in median ICP between AVM patients and control subjects (6.9 vs. 8.6 mmHg, p=0.30) and ICP did not correlate with nidus volume in AVM patients (ρ=-0.06, p=0.74). There was a
2 Page 3 of 29
significant positive correlation between tCBF and craniospinal CSF stroke volume (ρ=0.69, p=0.02).
Conclusions
ip t
The elevated cerebral blood flow in patients with AVMs is drained through an increased flow
cr
in IJVs but not secondary veins. ICP is maintained within ranges of normal and does not
an
us
correlate with nidus volume.
Key words
ce pt
ed
M
Arteriovenous Malformations; MRI; phase-contrast MRI; blood flow; cerebrospinal fluid.
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
3 Page 4 of 29
List of Abbreviations Arteriovenous malformations
CSF
Craniospinal fluid
ICP
Intracranial pressure
ICVC
Intracranial volume change
tCBF
Total (arterial) cerebral blood flow
ICAs
Internal carotid arteries
VAs
Vertebral arteries and
IJVs
Internal jugular veins
VVs
Vertebral veins
EVs
Epidural veins
DCVs
Deep cervical veins
ce pt
ed
M
an
us
cr
ip t
AVMs
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
4 Page 5 of 29
Introduction It has been shown that cerebral AVMs are associated with a local dysfunction in autoregulation of cerebral blood flow.1,
2
Vascular injury, abnormal endothelial signaling,
microshunt formation, and venous hypertension have been identified as potential 3
The dysbalance of
ip t
mechanisms of impaired autoregulation in cerebral AVMs.2,
cr
autoregulation may drive growth and remodeling of AVMs.2 Other authors have hypothesized that cerebral AVMs may originate as a compensatory response to a local
us
imbalance in autoregulation of blood flow.4 A recent study suggests that impaired
an
cerebrovascular reserve in peri-nidal brain areas associated with venous congestion may account for seizures.5 Other authors, however, have described preserved autoregulatory
M
responsiveness in brain areas adjacent to AVMs.6, 7
ed
While there is extensive research about local autoregulation in perinidal areas, little is known about how cerebral AVMs affect global cerebral autoregulation and ICP. Papilledema
ce pt
caused by intracranial hypertension is a rare but recognized presentation of unruptured cerebral AVMs.8-12 High-volume shunts may potentially exhaust the venous drainage capacity and thus lead to congestion of the venous system, associated with an increase in cerebral blood volume, impairment of CSF absorption, and increase in CSF production.10, 13
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
Direct mass effect may be another mechanism of AVM-associated intracranial hypertension.9,
10
In one recent case series, 3 out of 4 cases of intracranial hypertension
(identified as papilledema) associated with cerebral AVMs were considered to be due to overload of draining veins, while 1 case was caused by thrombotic occlusion of the straight sinus.8 All four cases occurred in patients with large cerebral AVMs. The diagnosis of intracranial hypertension had immediate therapeutic consequences in these patients, as partial endovascular embolization was performed and immediately reduced the massive 5 Page 6 of 29
overload of the venous system and thus relieved neurological symptoms.8 Identifying patients with cerebral AVMs at risk for intracranial hypertension and monitoring their ICP may therefore be of significant clinical relevance in order to prevent serious complications. We aimed to assess how the presence of a cerebral AVM affects ICP depending on nidus
ip t
volume. Moreover, we aimed to analyze relevant hydro- and hemodynamic parameters that
ce pt
ed
M
an
us
cr
may compensate the increased flow in the AVM.
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
6 Page 7 of 29
Methods Informed Consent and Ethical Approval Institutional Review Board approval was obtained prior to the commencement of the study.
ip t
Written consent was obtained from all patients and healthy individuals (parents or legal
cr
guardians for minors) prior to enrollment.
us
Patient Population and Control Subjects
an
30 consecutive patients with cerebral AVMs referred to MR imaging were enrolled in the study. All examinations were performed for clinical indications. Both treatment-naïve
M
patients and patients previously treated with radiosurgery were included. For each patient, a healthy individual of the same gender and with no more than 12 months of difference in age
ed
served as a paired reference. To minimize confounding of perfusion parameters, exclusion criteria for both patients and control participants included a history of diabetes, vascular
ce pt
disease, stroke or intracranial hemorrhage. General exclusion criteria for MR imaging included claustrophobia, ferromagnetic implants, cochlear implants and cardiac pacemakers.
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
MRI Examination Protocol All subjects underwent MR imaging on a 3-Tesla MR scanner (Magnetom Verio, Siemens Healthcare, Erlangen, Germany), using a 12-element phased-array head coil. 3D magnetization-prepared rapid-acquisition gradient echo imaging (MP-RAGE) was obtained for structural information using the following parameters: repetition time (TR) = 11 ms; echo time (TE) = 4.76 ms; field of view (FOV) = 230 mm; voxel size = 0.9 x 0.9 x 0.9 mm 3. 160 7 Page 8 of 29
sagittal slices parallel to the falx cerebri were acquired, covering the entire brain. The data were acquired with a parallel acquisition technique (iPAT, acceleration factor 2). A dynamic contrast-enhanced angiography sequence (TWIST) was obtained using the following
ms, flip angle 21°, base resolution 256 x 256, phase resolution 80%.
ip t
acquisition parameters: field of view 280 mm, slice thickness 2.0 mm, TR 2.33 ms, TE 0.94
cr
In order to obtain blood flow to and from the brain, two retrospectively gated, velocity-
us
encoded cine-phase contrast scans were performed (FOV = 140 mm, matrix = 256 x 179, voxel size = 0.8 x 0.5 x 6 mm, TR = 40 ms, TE 4.05 ms, FA = 20 degrees, acquisition time = 32
an
cardiac cycles equaling approximately 3 minutes as determined by the individual heart rate). First, a high-velocity encoding (70 cm/s) was used to quantify the high-velocity blood flow in
M
the internal carotid arteries (ICAs), vertebral arteries (VAs), and internal jugular veins (IJVs). The sequence was positioned at the upper level of the 2nd cervical vertebra with an
ed
orientation perpendicular to the main four arteries (ICAs and VAs) and the IJVs. To assess the
ce pt
low-velocity secondary venous flow, a sequence with low-velocity encoding (7–9 cm/s) was applied at the same level. To visualize primary and secondary venous channels, a 2D time-offlight MR venography of the infratentorial and upper cervical regions was performed (slice thickness 2.0 mm, FOV 160 mm, TR 23 ms, TE 5.43 ms and FA 40°).
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
Post-processing and Data Analysis Before performing quantitative analysis, all MRI datasets were reviewed to confirm diagnostic image quality. 32 images of the pulsatile flow per cardiac cycle were derived from the applied cine-phase contrast sequence. Time-dependent volumetric flow rates were calculated by integrating the flow velocities inside the luminal cross-sectional areas over all 8 Page 9 of 29
32 images, using the pulsatility-based segmentation of lumens conducting non-steady flow algorithm. Mean flow rates were obtained for each of the four main cervical arteries (left and right ICAs and left and right VAs), for the primary venous pathways (left and right IJVs), and the secondary venous pathways including vertebral veins (VVs), epidural veins (EVs), and
ip t
deep cervical veins (DCVs).
cr
Flow waveforms were obtained for each of the four main cervical arteries (left and right
us
internal carotid artery (LICA, RICA) and left and right vertebral artery (LVA, RVA), for the primary venous pathways, the left and right internal jugular vein (LIJV, RIJV), and for the
an
secondary venous pathways, the vertebral veins (VV), epidural veins (EV), and deep cervical veins (DCV). In addition, cervical CSF stroke volume, i.e., the volume of CSF that flows back
M
and forth between the cranium and the spinal canal, and aqueductal CSF stroke volume were obtained by time integration of the CSF flow waveforms. Total arterial cerebral blood
ed
flow (tCBF) was obtained by summation of the flows through the four arteries supplying the
ce pt
brain (left and right ICA, left and right VA). Flow in the basilar artery was calculated as the sum of the flow in both VAs. Secondary venous flow was defined as the sum of the flow through the three main secondary venous channels (VVs, EVs, and DCVs). Total venous outflow was determined as the total flow in all detected veins (IJVs and secondary veins).
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
Details of the derivation of the intracranial compliance and pressure have been described previously14, 15. Briefly, based on the physical definition of compliance as a ratio of volume and pressure changes, intracranial compliance is estimated from the ratio of the maximal (systolic) intracranial volume and pressure fluctuations during the cardiac cycle. The change in intracranial volume (ICVC) is obtained from the momentary differences between volumes of blood and CSF entering and leaving the cranium as shown in equations 1 and 2,
9 Page 10 of 29
ICVC (i) [ f A (i) f V (i) f CSF (i)] t
ICVC (i) [ f
cardiac cycle
A
(1)
(i) f V (i ) f CSF (i )] t 0
(2)
cardiac cycle
ip t
f f where f A is arterial inflow, V is venous outflow, and CSF is the craniospinal CSF flow. Equation 2 states that in steady state, the intracranial volume is on average constant over an
cr
entire cardiac cycle.
us
The pressure change is derived from the amplitude of the pressure gradient (PG) waveform
an
obtained using the Navier-Stokes relationships between derivatives of CSF velocities and the CSF pressure gradient. An MRI equivalent of ICP (MRICP) is then obtained based on the
M
reported inverse relationship between compliance and ICP16. Volumetric blood and CSF flow rate waveforms and derived parameters were obtained using a dedicated software tool
ed
(MRICP version 1.4.35 Alperin Noninvasive Diagnostics, Miami, FL).
ce pt
All MR images were reviewed by a board-certified neuroradiologist with >12 years of experience in neuroradiology and MR angiography (initials blinded). All images were carefully reviewed to detect any potential confounders such as evidence of microangiopathy, stroke, intracranial hemorrhage, gliosis or parenchymal defect zones. AVM
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
nidus size was measured in three dimensions on dynamic MR angiographic datasets (TWIST sequence); the time point, in which the nidal volume was largest, was chosen for the measurements. AVM nidus volume was calculated using the spheric volume formula 4/3 x π x (l/2) x (h/2) x (w/2) with l, h, and w being the length, height and width of the AVM nidus. As an approximation, the three-dimensional shape of the gliar scar surrounding the AVM nidus can be regarded as a spherical shell, i. e. the region between two concentric spheres of differing radii. Therefore, the maximum thickness of the glial scar around the nidus t and the 10 Page 11 of 29
maximum diameter oft the lesion including nidus and glial scar d was measured. Glial scare volume was then calculated using the volume formula for a spherical shell as 4/3 x π x (d/2)3 – 4/3 x π x (d/2 – t)3. For parenchymal defect zones adjacent to the AVM nidus, the
cr
calculated using the spheric volume formula 4/3 x π x (d/2)3.
ip t
maximum diameter of the defect d was measured and the approximate volume was
us
Statistical Analysis
Data were summarized by properly chosen measures of location and spread for continuous
an
variables and by proportions for discrete variables. Correlations between variables were
M
estimated by Spearman’s correlation coefficient ρ. Linear mixed effects regression models were used to test for differences between AVM patients and healthy controls. For
ed
aqueductal stroke volume we further separately evaluated AVMs with supratentorial location. In these models we allowed the intercept to vary by the matched subjects (random
ce pt
effects term). For AVM patients, the impact of AVM nidus volume and total cerebral blood flow on flow parameters was estimated and adjusted for the volume of parenchymal defects and for the volume of glial scar by multivariate linear regression analysis. For all regression models the corresponding dependent variables were transformed by appropriately chosen
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
powers or by the natural logarithm to satisfy the normality assumption. A two-sided level of significance of α=0.05 was applied. All statistical analyses were performed using R (version 2.13.0).
11 Page 12 of 29
Results Thirty consecutive patients were included into the data analysis (table 1). Two patients underwent MRI as part of their evaluation prior to stereotactic radiosurgery. 28 patients were referred for MR imaging for follow-up after stereotactic radiosurgery. In median, the
ip t
radiosurgical procedure had been performed 20 (interquartile range 26) months prior to the
cr
MRI examination. All patients received a single procedure using CyberKnife radiosurgery
us
(Accuray, Inc., Sunnyvale, CA) with a median dose of 18.5 Gy (range 15-24 Gy). 12 (40%) patients were female. Median age was 38.7 years for AVM patients and 38.1 years for
an
control subjects. Ages ranged from 17 to 62 years.
AVM nidus volumes ranged from 0.0 to 51.5 cm3 with a median of 0.8 cm3 and an
M
interquartile range of 3.4 cm3 (See Table 1). 7/30 patients did not have a measurable perfused nidus volume due to successful radiosurgical treatment. 9/30 patients (30%) had an
ed
AVM nidus volume >2 cm3. Most frequently, AVMs were located in the frontal (n=10),
ce pt
parietal (n=6), and temporal lobe (n=6). AVMs were also found in the occipital lobe, thalamus, brainstem (n=2 each) and cerebellum (n=1). In 19 cases, the remaining AVM nidus was surrounded by a glial scar. The median glial scar volume was 5.2 cm3 with a range of 0.1 to 49.9 cm3. 13 patients had developed a parenchymal defect with a median volume 2.6 cm3
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
with a range of 0.1 to 54.4 cm3. TCBF to the brain was found to be significantly higher in AVM patients as compared to healthy controls (median 799 vs. 692 mL/min, p=0.007; see Table 2). In AVM patients, tCBF ranged from 566 to 2222mL/min as compared to a range of 502 to 903 mL/min in healthy individuals. Specifically, arterial flow was significantly higher in the internal carotid artery (ICA) on the side of the AVM localization (293 vs. 248 mL/min, p=0.02). In contrast, there 12 Page 13 of 29
were no significant differences in arterial blood flow in the contralateral ICA between patients and healthy volunteers (255 vs. 256 mL/min, p=0.20) and the basilar artery (221 vs. 200, p=0.05). In AVM patients, there was a trend for flow to be higher in the ICA ipsilateral to the AVM nidus than in the contralateral ICA (293 vs. 255 mL/min, p=0.10). AVM nidus
ip t
volume correlated positively with total cerebral arterial blood flow (ρ=0.43, p=0.02).
cr
Corresponding to the increased arterial blood flow, median total venous outflow (TVO) was significantly higher in AVM patients than in healthy participants (675 vs. 567 mL/min,
us
p=0.007, see Table 2). The recorded median TVO values were 13-19% lower than tCBF. There
an
was a trend for venous flow to be increased in both the ipsilateral internal jugular vein (IJV, 282 vs. 225 mL/min, p=0.16), and in the contralateral IJV (322 vs. 285 mL/min, p=0.09),
M
although these differences did not reach the level of significance. There was no increase in
38 mL/min, p=0.28).
ed
the venous drainage by secondary veins (epidural, vertebral and deep cervical veins; 23 vs.
ce pt
There was a trend towards a positive correlation between AVM nidus volume and total venous outflow (ρ=0.34, p=0.06) as well as contralateral IJV flow (ρ=0.41, p=0.02; see Table 3). No relevant correlation was found between nidus volume and blood flow in ipsilateral IJV (ρ=0.11, p=0.56) and secondary veins (ρ=0.04, p=0.83).
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
In AVM patients, ICP ranged from 3.7 to 16.1 mmHg. A very similar range of 3.3 to 15.7 mmHg was found in the control group. There was no significant difference in median ICP between AVM patients and healthy subjects (6.9 vs. 8.6 mmHg, p=0.30, see Table 2) and ICP did not correlate with nidus volume (ρ=-0.06, p=0.74; see Table 3), tCBF (ρ=0.13, p=0.50; Table 4).
13 Page 14 of 29
There were no significant differences between AVM patients and the control group regarding CSF stroke volume, intracranial volume change or aqueductal stroke volume during the cardiac cycle (See Table 2). Moreover, these parameters showed no significant correlation with nidus volume (See Table 3). No correlation of intracranial volume change or
ip t
aqueductal stroke volume with TCBF was observed (Table 4). There was, however, a significant correlation between TCBF and CSF stroke volume in AVM patients (ρ=0.69,
cr
p=0.02; see Table 4). For aqueductal stroke volume we further separately evaluated AVMs
us
with supratentorial location (n=27). Again, no significant difference between AVM patients
ce pt
ed
M
an
and healthy individuals (p=0.27) was found.
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
14 Page 15 of 29
Discussion Unruptured cerebral AVMs can cause intracranial hypertension by high shunt volumes exhausting the venous drainage capacity or by direct mass effect.8-12 Since AVM-induced intracranial hypertension alters management8, it is desirable to identify AVM patients at risk
ip t
for intracranial hypertension. To our knowledge this is the first systematic study evaluating
cr
intracranial pressure in patients with AVM in relation to AVM nidus volume and cerebral
us
flow parameters.
In agreement with other studies2, 10, 13, we found a significantly elevated tCBF in patients
an
with AVMs. AVM patients had a significantly higher flow in their ipsilateral ICA and basilar artery but not in their contralateral ICA. Contralateral ICA flow did, however, show a
M
moderate but significant positive correlation with AVM nidus size indicating that AV shunting
ed
recruits blood volume from the circle of Willis thus moderately increasing flow in the contralateral ICA. These results are in line with a recent report that ipsilateral ICA flow is
ce pt
significantly higher than contralateral ICA flow in AVM patients and that this difference is leveled out 2-4 years after radiosurgery17. We have further shown that the increased cerebral blood flow is drained by an increased flow in the ipsilateral and contralateral IJVs but not by an enhanced secondary venous flow.
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
Overall, total venous outflow from the brain must equal total arterial cerebral blood flow. We determined TVO by adding blood flow in both internal jugular veins and all identified secondary veins. Thus, recorded median TVO values were 13-19% lower than tCBF. This is most likely due to small secondary veins that are missed when manually placing regions of interest into all discernible epidural, vertebral and deep cervical veins. Similar rates of
15 Page 16 of 29
unmeasured secondary venous outflow have been reported in other studies 18 and are likely due to drainage through smaller secondary veins not measured (eg, ocular veins). In our study, intracranial pressure showed no correlation with AVM nidus size and the range of ICP values was virtually equal in AVMs patients and healthy individuals. This may be due
ip t
to nidus sizes and / or flow rates too small to induce intracranial hypertension. In a recent
cr
case series 3 out of 4 patients with AVM-induced intracranial hypertension had “large” AVMs (volume not further specified)8. However, our study did include large AVMs with nidus
us
volumes up to 51.5 cm3 and total cerebral blood flow as high as 2222 mL/min. Even if the
an
number of large AVMs in our study was relatively low with only 9/30 patients (30%) showing an AVM nidus volume >2 cm3, we would have expected to see at least a trend if such a
M
correlation existed. Although large nidus size may be a predisposing factor for intracranial hypertension, other factors such as the AVM nidus angioarchitecture, reduced venous
ed
drainage capacity or altered CSF production and absorption may play a more important role. Given the rare incidence of intracranial hypertension in AVM patients, larger studies or case-
ce pt
control-studies are needed to identify such predisposing factors. Because compliance of the spinal canal exceeds intracranial compliance, a small but measurable amount of CSF is shifted through the foramen magnum into the spinal canal
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
during each cardiac cycle. Similarly, there is a small pulsatile CSF shift through the cerebral aqueduct into the fourth ventricle and subarachnoid space.19 We had hypothesized AVM patients to show a shunt volume-dependent increase in craniospinal CSF shift in order to maintain constant intracranial pressure despite their elevated cerebral blood flow. Surprisingly, we found no significant differences between AVM patients and healthy individuals in aqueductal and craniospinal CSF stroke volume of intracranial volume change and these parameters showed no correlation with nidus volume. There was, however, a 16 Page 17 of 29
significant correlation of total cerebral blood flow with CSF stroke volume in AVM patients. Therefore, increased craniospinal CSF pulsation may be a compensatory mechanism in order to avoid pulsatile elevations in ICP in AVM patients with high cerebral blood flow. One limitation of our study is that all but two patients had been previously treated with
ip t
stereotactic radiosurgery. As reported above, patients in our study were examined at a
cr
median of 20 months after undergoing radiosurgical treatment. Considering that obliteration of the AVM nidus is usually completed within 1-3 years after radiosurgery20-24, most patients
us
showed at least partially obliterated nidi and a varying degree of reactive changes such as
an
gliosis or parenchymal defects. Parenchymal defect zones and glial scars represent areas of hypoperfusion and therefore represent a potential confounder of blood flow parameters.
M
However, we did find a significantly higher tCBF in AVM patients than in healthy individuals and tCBF correlated with nidus volume. Therefore, a significant part of our patient
ed
population had significant residual nidi with intact AV shunting. Also, correlation of nidus volume with cerebral flow parameters was adjusted for potential confounding by glial scars
ce pt
and defect zones using multivariate linear regression analysis. Neither glial scar nor defect zone volume significantly influenced flow parameters and intracranial pressure. Therefore, we expect similar results regarding tCBF and intracranial pressure in an untreated patient population. On the other hand, it has recently been shown that the circle of Willis shows
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
significant remodeling with altered flow patterns after stereotactic surgery in patients with cerebral AVMs.25 This remodeling may influence the relative flow distribution among the three major arteries supplying the brain as assessed in our study. In this respect, it is possible that our results are specific for patients after radiosurgery and may not represent flow patterns in treatment-naïve patients.
17 Page 18 of 29
Conclusions In conclusion, we have shown that the elevated cerebral blood flow in patients with AVMs is drained through an increased flow in IJVs but not secondary venous flow. ICP did not correlate with nidus volume in our patient cohort. Large nidus volume may be one risk factor
ip t
but other determinants need to be identified in further studies to identify those AVM
ce pt
ed
M
an
us
cr
patients at risk for intracranial hypertension.
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
18 Page 19 of 29
Tables Table 1. Characteristics of Study Participants Healthy Controls
30
30
Number of Subjects Age [years]
ip t
AVM Patients
38.7 (17.1) 12 (40%)
12 (40%)
cr
Females (% of participants)
38.1 (17.1)
n =28, stereotactic radiosurgery n=2, none
n/a
20 (26)
n/a
us
Prior therapy
an
Time after Radiosurgery [months]
n=10, frontal lobe n=6, parietal lobe n=7, temporal lobe n=2, occipital lobe n=2, thalamus n=1, cerebellum n=2, brain stem
n/a
0.8 (3.4)
n/a
5.2 (12.3)
n/a
2.6 (24.0)
n/a
Nidus Volume [cm3] (n=30)
ce pt
Glial Scar Volume [cm3] (n=19)
ed
M
Main Localization of AVM Nidus
Parenchymal Defect Volume [cm3] (n=13)
Table 1. Key characteristics of AVM patients and healthy controls are listed. Unless indicated
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
otherwise, median values are provided followed by interquartile ranges in parentheses.
19 Page 20 of 29
Table 2. Cerebral Flow Parameters in AVM patients and Healthy Controls Healthy Controls
p-value
799 (153)
692 (135)
0.007
Ipsilateral ICA
293 (98)
248 (60)
0.02
Contralateral ICA
255 (69)
256 (63)
Basilar Artery
221 (69)
200 (47)
675 (273)
567 (134)
0.007
Ipsilateral IJV
282 (240)
225 (267)
(0.16)
Contralateral IJV
322 (288)
285 (247)
(0.09)
Secondary Veins
23 (56)
38 (77)
(0.28)
6.9 (4.1)
8.6 (4.7)
(0.30)
0.54 (0.30)
0.50 (0.20)
(0.88)
0.56 (0.30)
0.54 (0.23)
(0.46)
33.7 (16.6)
(0.16)
Intracranial Pressure [mmHg] CSF Stroke Volume [mL]
ed
Intracranial Volume Change [mL]
M
cr
an
Total Venous Outflow
us
Total Cerebral Arterial Blood Flow
43.4 (40.1)
(0.20)
(0.05)
ce pt
Aqueductal Stroke Volume [μL]
ip t
AVM Patients
Table 2. Cerebral flow parameters of AVM patients (n=30) are compared to age- and gendermatched healthy controls (n=29). A linear mixed-effects model was used to test for differences between both groups. Median flow values in mL/min are provided followed by interquartile ranges in parentheses. P-values which did not meet the level of statistical
Ac
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
significance (p
