| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Neuroradiology |
i
ka, MD,
Ji
í Ceral, MD,
Pavel Eliá
, MD,
Jaroslav Tint
ra, PhD,
Ludovít Klzo, MD,
Miroslav Sola, MD and
Libor Straka, MD
1 From Departments of Radiology (J.Z., P.E., L.K.) and Internal Medicine (J.C., M.S.), Charles University Hospital, Sokolská 581, CZ-500 05 Hradec Králové, Czech Republic; MRI Unit, Institute of Clinical and Experimental Medicine, Praha, Czech Republic (J.T.); and Department of Biophysics, Medical Faculty, Charles University, Hradec Králové, Czech Republic (L.S.). From the 2001 RSNA scientific assembly. Received October 4, 2002; revision requested December 12; final revision received May 28, 2003; accepted June 18. Supported in part by grant NA/6169–3 from the Internal Grant Agency, Ministry of Health, Czech Republic. Address correspondence to J.Z. (e-mail: zizka@fnhk.cz).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Forty-three patients with severe essential hypertension and 45 normotensive subjects were matched for age, sex, and body mass index. Magnetic resonance (MR) imaging included transverse and coronal T2-weighted turbo spin-echo (section thickness, 3.0 mm), transverse three-dimensional (3D) time-of-flight MR angiographic (section thickness, 0.8 mm), and 3D constructive interference in steady state (CISS) (section thickness, 1.0 mm) sequences. All MR images were reviewed by two radiologists who were blinded to the hypertensive status of subjects. Presence and degree of NVC at rostral medulla and left/right rostral ventrolateral medulla (RVLM) were evaluated together with conspicuity of anatomic structures on MR images. Differences in prevalence of NVC among normotensive and hypertensive subjects were tested for statistical significance (P < .05) by using nonparametric tests.
RESULTS: Among hypertensive patients, 34 (79%) of 43 showed NVC of rostral medulla at any location, and 14 (33%) of 43 had NVC at the left RVLM. In controls (normotensive subjects), 35 (78%) of 45 showed NVC of rostral medulla, and 17 (38%) of 45 had NVC at left RVLM. Prevalence of NVC was not significantly different between both groups at any location of rostral medulla. Compared with T2-weighted turbo spin-echo and 3D time-of-flight MR imaging sequences, 3D CISS offered better contrast resolution of neural and vascular structures and superior delineation of outer vascular contours.
CONCLUSION: Vascular compression of the rostral medulla oblongata is a frequent finding in both hypertensive and normotensive subjects. Results of this study do not support NVC at left RVLM as an etiologic factor in essential hypertension.
© RSNA, 2003
Index terms: Brainstem, anatomy, 152.136, 152.92 • Brainstem, MR, 152.121411, 152.12142 • Hypertension
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
After Jannetta et al (9,10) initially reported that surgical decompression of a neurovascular contact (NVC) at the left RVLM might relieve hypertension, results of several studies with animal and anatomic models suggested an association between pulsatile NVC at the RVLM and essential hypertension (11–16). The definite proof of a causal relationship between these two conditions might have a considerable effect on possible surgical therapy in a substantial number of patients who have medically intractable hypertension or intolerable side effects of the medication.
Magnetic resonance (MR) imaging is the only noninvasive imaging modality capable of detailed evaluation of the relationships between the brainstem and vascular structures. However, the results of published MR imaging studies in which NVC was evaluated in hypertensive and normotensive subjects widely differ, partially because of methodological differences, root entry zone definition, insufficient spatial resolution, or involvement of a small number of subjects (17–26). Thus, the purpose of our study was to prospectively evaluate the prevalence of NVC at the rostral medulla oblongata in normotensive and hypertensive subjects.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Imaging
All MR imaging examinations were performed with a 1.0-T system (Magnetom Expert; Siemens, Erlangen, Germany) and a quadrature transmit-receive head coil. After acquisition of initial localization images, a set of four dedicated MR imaging sequences was performed to cover the anatomic structures of the posterior fossa and craniocervical junction. The transverse plane was defined as perpendicular to the floor of the fourth ventricle and the posterior border of the brainstem; the coronal plane was parallel to these structures. First, two-dimensional T2-weighted turbo spin-echo images were obtained in the transverse and coronal orientations (repetition time msec/echo time msec, 4,500/120; echo train length, 15); the voxel size was 3.0 x 0.9 x 0.9 mm. Then, for MR angiography, transverse three-dimensional (3D) fast imaging with a steady-state precession time-of-flight imaging sequence (37/9.6; flip angle, 20°) was performed with magnetization transfer and tilted optimized nonsaturating excitation; the voxel size was 0.8 x 0.7 x 0.4 mm. Last, to obtain the highest achievable contrast resolution between the neurovascular structures and the cerebrospinal fluid, a transverse 3D constructive interference in steady state (CISS) sequence was performed (15.7/7.5; flip angle, 70°); the voxel size was 1.0 x 0.7 x 0.4 mm.
Image Review
The MR images were independently reviewed by two radiologists (P.E., J.Z.) whose experience in neuroimaging was 15 and 8 years, respectively. The radiologists were blinded to the relevant clinical data. In instances of disagreement between the reviewers, a consensus interpretation was used for the case. NVC at the lower brainstem was assessed on all MR images, including original transverse 3D time-of-flight MR angiographic images. These images were also used for validation of blood flow within the vessels that were evaluated (ie, vertebral, anterior, and posterior inferior cerebellar arteries) and for generation of maximum intensity projections of the arterial anatomy of the posterior fossa. NVC, if present, was classified as either a simple contact of the artery with the brainstem or an impression of the artery that produced an apparent deformity of the brainstem contour. The definition of the RVLM reflected the anatomic extent of the cranial nerve IX and X root entry zone, which measures 1 cm in rostrocaudal direction and extends over the upper and middle thirds of the olive (14). In the transverse plane, the RVLM was confined to the retro-olivary sulcus, where the C1 adrenergic neurons are located within 1 mm from its surface (4). The anterior border was defined as the transition of the posterior olivary convexity to the concavity of the retro-olivary sulcus; the posterior border was situated at the root entry zone of cranial nerve IX and X fibers. A record was made if NVC was encountered at the left or right RVLM. In addition, NVCs located anteriorly to the left or right RVLM (ie, at the region of the pyramids and olives) were recorded separately. Thus, the number of NVCs per subject could range from zero to four. NVCs at the dorsal medulla, located posterior to the cranial nerve IX and X root entry zone, were not analyzed. The MR sequence or sequences that showed superior conspicuity of neural and vascular structures (ie, in terms of their contrast, spatial, and tissue boundary resolution) were recorded for each case.
Statistical Analysis
When we determined the sample size of our project, we performed a power analysis by using the Fisher exact test for binomial distribution. The estimated proportions for groups 1 and 2 were set as 0.20 and 0.47, with P = .05 and a power of 0.80. The sample size estimation was set as 44 and more. We also took into consideration the sample sizes in other articles published at that time. The demographic data of both groups of subjects were compared by using nonparametric statistics: the 
2 test for sex-related differences and the Mann-Whitney U test for differences in age and body mass index. The prevalence of NVCs was evaluated by using the 2 test; if the number of observations was lower than 10 in any group, the Yates correction was implemented. Interreader agreement was characterized with the 
statistic: = 0 indicated no agreement, and = 1 indicated complete agreement between the readers. The values were calculated from the original data obtained separately from both readers, that is, before consensus. A difference with a P value less than .05 was considered to indicate statistical significance.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
We did not find increased prevalence of NVCs at the left or right RVLM in hypertensive patients. Of 43 hypertensive subjects, 14 (33%) showed an NVC at the left RVLM; and 15 (35%), at the right RVLM (Fig 1). Of 45 normotensive subjects, 17 (38%) showed an NVC at the left RVLM; and 18 (40%), at the right RVLM (Fig 2). The corresponding rates for the NVCs located outside the RVLM were 17 (40%) on the left and 14 (33%) on the right in the group of 43 hypertensive patients, and 19 (42%) on the left and 10 (22%) on the right in the group of 45 normotensive subjects. The prevalence of NVCs did not show statistically significant differences between the groups of hypertensive and control subjects in any location of the rostral medulla. The overall interreader agreement for evaluation of NVCs at the rostral medulla was 0.84.
|
|
|
|
|
|
Among 43 hypertensive and 45 control subjects, the most common offending vessel at the left RVLM was the posterior inferior cerebellar artery in nine (21%) and 11 (24%), followed by the anterior inferior cerebellar artery in three (7%) and five (11%), and the vertebral artery in two (5%) and one (2%) subject, respectively. The corresponding numbers at the right RVLM were 11 (26%) and 12 (27%) for the posterior inferior cerebellar artery, three (7%) and five (11%) for the anterior inferior cerebellar artery, respectively, and one (2%) in both groups for the vertebral artery.
The visualization of neurovascular structures at the lower brainstem was clearly superior with both 3D high-spatial-resolution MR sequences when compared with the standard two-dimensional turbo spin-echo sequence with a 3.0-mm section thickness and a matrix of 240 x 256 (Figs 1, 2). In 36 (41%) of 88 MR imaging studies, the best conspicuity of neurovascular boundaries was rated with the 3D CISS imaging sequence. In 40 (45%) studies, 3D time-of-flight MR imaging offered information equivalent to that obtained with the 3D CISS sequence. In seven (8%) studies, 3D time-of-flight MR images provided more detailed information than did 3D CISS images. The value of two-dimensional turbo spin-echo MR images was rated equal to that of the 3D images in only five (6%) studies. Coronal T2-weighted turbo spin-echo MR images were not found to be of additional help when the NVCs at the lower brainstem were evaluated.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
The impossibility of direct brainstem visualization, together with the invasiveness of digital subtraction angiography, as well as insufficient contrast resolution of computed tomography, make MR imaging the most suitable imaging method for screening patients for the presence of NVC. The MR imaging studies, which were performed after the initial observations, provided somewhat unequivocal results. Naraghi et al (17) examined the RVLM by using a double-echo T2-weighted sequence with a section thickness of 3 mm, together with MR angiography, in a prospective blinded fashion. Twenty (83%) of 24 patients with essential hypertension, two (14%) of 14 patients with renal hypertension, and one (7%) of 14 normotensive subjects had MR imaging evidence of NVC at the left ventrolateral medulla. On the right side, they found an NVC in only four (17%) of 24 patients with essential hypertension. Akimura and colleagues (19) used a 3D fast low-angle shot sequence with effective section thickness of 2.5 mm to demonstrate high prevalence of NVC at the left RVLM. NVCs were present in 25 (78%) of 32 patients with essential hypertension, compared with one (17%) case of left-sided NVC among six patients with secondary hypertension and three (17%) of 18 normotensive subjects. Blinded reading was not used in that study. Morimoto et al (20) reported a high frequency of left- and/or right-sided NVC at the RVLM in the group of patients with essential hypertension (75%) who were compared with the group of patients with secondary hypertension (10%) or with the normotensive (11%) subjects. However, their study with intermediate-weighted images with a section thickness of 3 mm did not show left-sided NVC predominance among the group of 20 patients with essential hypertension: Seven left-sided, seven right-sided, and one bilateral NVC at the RVLM were observed. Findings of these three reports (17,19,20) suggested a causal relationship or at least a close correlation between NVC at the RVLM and essential hypertension. However, a causal relationship might have been confirmed after elimination of the assumed cause, with subsequent disappearance of the effect. Instead, the researchers observed only the coexistence of two events. It should be noted that none of these studies involved more than 18 control subjects. Moreover, the definition of the critical zone (ie, RVLM) was not exactly reproducible, and it differed among the studies.
Even though the concept of neurogenic hypertension, as well as initial results reported by Jannetta et al (9,10), has been in the literature for more than 2 decades, only sporadic reports exist about the surgical microvascular decompression of the left RVLM in patients with essential hypertension. This is the case despite the large size of the affected population. Geiger et al (18) and Levy et al (21) reported the results in patients who had undergone microvascular decompression of the left RVLM primarily for severe refractory hypertension. Researchers in both reports claimed substantial blood pressure reduction in most of the patients postoperatively. However, the results of these studies are less convincing if the following facts are taken into account: First, the studies were retrospective. Second, the number of patients involved was small (eight and 12, respectively). Third, extensive changes in medication before and after the operation made the contribution of the surgical procedure rather uninterpretable. Fourth, only half of the patients achieved long-standing normotension while they were still receiving medication.
In contrast, the report by Watters et al (22) published in 1996 was the first one, to our knowledge, that did not show statistically significant differences in the presence of vascular compression of the left cranial nerve IX and X root entry zone between the groups of hypertensive (57%) and normotensive (55%) subjects. The reviewers were not blinded to the hypertensive status of the subjects in that retrospective MR study, which relied on transverse spin-echo T2-weighted MR images with 5-mm section thickness and 2-mm intersection gap. Colón et al (23) performed a blinded prospective MR imaging study in 30 patients with essential hypertension and 45 normotensive subjects by using 3-mm-thick T1-, T2-, and intermediate-weighted images with a 0.5-mm intersection gap. They detected no significant differences in NVC between the essential hypertensive group (31% at the left, 35% at the right RVLM) and the control group (44% at the left, 30% at the right RVLM).
These two studies might receive some criticism because of limited visualization of structures that were smaller than 1 mm on MR images obtained with section thickness of 3 mm or more. Nevertheless, this problem of partial-volume effects can also be applied to other aforementioned studies. From our experience, the conspicuity of neurovascular boundaries at two-dimensional turbo spin-echo MR imaging with a 3-mm section thickness was rated inferior to any of the high-spatial-resolution 3D MR imaging sequences in 94% of cases.
Another problem should also be addressed when we deal with the details of NVCs. Namely, it is not the lumen of the vessel but the outer contour of the vascular wall that is in contact with the neural structures. Hence, the imaging should be able to exactly delineate the outer vascular margin. However, none of the MR sequences used in all the reports cited was capable of precise delineation of the outer vascular contour, which especially pertains to bright-blood MR angiographic techniques. None of those studies performed high-spatial-resolution MR imaging sequences that might have had sufficient resolution at the neurovascular boundaries. This also partially applies to a prospective blinded study by Thuerl et al (26) published in 2001. The researchers stated that the periphery of the signal void had been taken into account, since the wall of the vessel could not be visualized directly by using two-dimensional turbo spin-echo T2-weighted MR images with 3-mm section thickness, as well as 3D time-of-flight MR angiographic images (effective section thickness of 1.5 mm). In that study, Thuerl et al found no statistically significant differences concerning the prevalence of NVC at the RVLM in patients with essential hypertension (49% at the left and 24% at the right RVLM, respectively), patients with secondary renal hypertension (27% and 13%, respectively), and normotensive volunteers (48% and 40%, respectively).
To overcome these limitations in spatial resolution, we used a combination of four MR imaging sequences in our prospective blinded study; included among them were two high-spatial-resolution 3D MR sequences with in-plane resolution markedly below 1 mm and effective section thickness that did not exceed 1 mm. The 3D CISS sequence offered superior contrast resolution between the neurovascular structures and the cerebrospinal fluid within the medullary cisterns. Therefore, it was possible to delineate the outer vascular contour precisely—what we believe is essential in assessment of NVC, generally. The depiction of cranial nerve fibers was also clearly superior to that with all other MR sequences used.
Although the 3D CISS images offered improved contrast, spatial, and tissue boundary resolution over the resolution of the conventional 3-mm-thick T2-weighted images or transverse 3D time-of-flight MR angiographic images (mostly used in previous studies), we did not prove an increased prevalence of NVC at the left or right RVLM among the patients with essential hypertension. Similar results were reported in the study by Hohenbleicher et al (25), the only other research team who also used this high-spatial-resolution MR imaging technique, to our knowledge; the reported prevalence of NVC at the left RVLM did not significantly differ between hypertensive (23%) and normotensive (16%) subjects (P = .12). Nonetheless, our results are also in general agreement with those of other aforementioned studies (22–24,26), although the contrast and/or spatial resolution in those studies might have been lower. That is, the NVC at the upper medulla and even at the left RVLM is a frequent and rather nonspecific finding in both hypertensive and normotensive subjects.
Thus, it can be assumed that the presence of NVC at the RVLM is not the sole factor responsible for development of essential hypertension throughout the large affected population. Therefore, even a high-spatial-resolution MR image cannot aid patient selection for surgery. We still believe that neurogenic hypertension may exist at least in some patients—for example, those with sympathetically maintained hypertension who show elevated baseline levels of plasma norepinephrine, as proposed by Gajjar et al (24). However, further detailed studies will be required to elucidate these etiopathogenetic mechanisms.
|
FOOTNOTES |
|---|
Author contributions: Guarantors of integrity of entire study, J.Z., J.C.; study concepts, J.C., J.Z., M.S.; study design, J.C., J.Z., L.S.; literature research, J.C., J.Z., M.S.; clinical studies, J.T., L.K., J.Z.; data acquisition, J.Z., L.K.; data analysis/interpretation, J.Z., P.E.; statistical analysis, L.S., J.Z; manuscript preparation and definition of intellectual content, J.Z., J.C.; manuscript editing, J.Z.; manuscript revision/review, P.E., J.C.; manuscript final version approval, J.Z., P.E., J.C.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  M. M. Sendeski, F. M. Consolim-Colombo, C. C. Leite, M. C. Rubira, P. Lessa, and E. M. Krieger Increased Sympathetic Nerve Activity Correlates With Neurovascular Compression at the Rostral Ventrolateral Medulla Hypertension, May 1, 2006; 47(5): 988 - 995. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
