Large Virchow-Robin Spaces:

Transcription

1 929 Large Virhow-Robin Spaes: MR-Ciinial Correlation Linda A. Heier 1 Cristel J. Bauer 1 Larry Shwartz 1 Robert D. Zimmerman 1 Susan Morgello 2 Mihael D. F. Dek 1 High-field MR sans frequently show Virhow-Robin spaes, whih onform to the path of the penetrating arteries as they enter either the basal ganglia or the ortial gray matter over the high onvexities. A retrospetive review of 816 MR sans was undertaken to determine the linial signifiane and assoiations (if any) of this finding. The Virhow-Robin spaes were graded, as were the nonspeifi white-matter lesions. The presene of atrophy, infartion, hydroephalus, and misellaneous disease was noted. Large Virhow-Robin spaes were identified in 314 ases. A study sample was reated onsisting of a positive group ontaining all the larger grade 2 and 3 Virhow Robin spaes (67 patients) and a negative or ontrol group of 109 randomly seleted patients from the original 502 who did not have large Virhow-Robin spaes. The harts of this study sample were reviewed and the following patient variables were noted: age, gender, inidental white-matter lesions, infartion, dementia, hypertension, and atrophy. For eah variable, the proportion of patients who were positive for that variable was alulated for eah of the two groups and ompared aross groups by using a Fisher exat test. Multiple logisti regression analysis was used to determine whether any of these variables were jointly assoiated with being "positive" or " negative" for large Virhow-Robin spaes. Some variables were strongly assoiated with being positive for large Virhow-Robin spaes: age, hypertension, dementia, and inidental white-matter lesions. Logisti regression analysis revealed that when all of these variables are onsidered jointly, only age remains signifiant. In onlusion, large Virhow-Robin spaes are another phenomenon of the aging brain. Fators suh as hypertension, dementia, and inidental white-matter lesions were signifiantly assoiated with large Virhow-Robin spaes, but were believed to aompany the aging proess rather than represent independent variables. AJNR 10: , September/ Otober 1989 Reeived November 4, 1988; revision requested January ; final revision reeived February 8, 1989; aepted February 22, Presented at the annual meeting of the Amerian Soiety of Neuroradiology, Chiago, May ' Department of Radiology, New York Hospital/ Cornell Medial Center, 525 E. 68th St., New York, NY Address reprint requests to L. A. Heier. 2 Department of Pathology (Neuropathology), New York Hospital/Cornell Medial Center, New York, NY /89/ Amerian Soiety of Neuroradiology High-field MR sans routinely demonstrate small foi of CSF signal on all pulse sequenes about the base of the brain. These lesions onform to the path of the lentiulostriate arteries as they enter the basal ganglia through the anterior perforated substane and represent large Virhow-Robin spaes, whih aompany the vessel as it penetrates the brain from the subarahnoid spae. Oasionally, these perforations have been seen in the high-onvexity gray matter extending into the entrum semiovale, where they follow the ourse of the penetrating ortial arteries. A retrospetive review of 816 MR sans was undertaken to determine the linial signifiane and assoiations, if any, of this anatomy. Materials and Methods All outpatients who had an MR san of the brain between Otober 1986 and August 1987 on a 1.5-T General Eletri Signa unit were retrospetively reviewed for the presene or absene of large Virhow-Robin spaes. The routine examination onsisted of three spineho sequenes, all done with a 256 x 256 matrix and one exitation: (1) axial images, /20 (TRfTE), with a 5-mm setion thikness and 5-mm gap; (2) axial multieho images,

2 930 HEIER ET AL. AJNR :1 0, September/Otober /40, 80, with a 5-mm setion thikness and 2.5-mm gap; and (3) sagittal images, 800/20, with a 5-mm setion thikness and 5-mm gap. By definition, a Virhow-Robin spae had to be isointense relative to CSF on all pulse sequenes, onform to the path of penetrating arteries. and have no mass effet. The loation of these spaes was reorded as lentiulostriate andjor high onvexity. The number and size of the Virhow-Robin spaes were used to grade their severity. As a rule, mild grade 1 spaes were under 2 mm in diameter, moderate grade 2 spaes were 2-3 mm, and marked grade 3 spaes were over 3 mm. The size of the spaes was measured with alipers on the films, and then the measurement in millimeters was obtained from the 5-m referene sale found on eah image. However, a large number of smaller spaes (four or more) would result in an upgrade to the next level. The presene of both ortial and launar infarts, atrophy, and misellaneous diseases was noted. Atrophy was evaluated qualitatively by two independent observers and not reorded if both observers believed it was appropriate for the patient's age. Only when the atrophy was greater than expeted for the patient's age was it reorded. If the two observers did not agree, the final determination was made by a third neuroradiologist. Nonspeifi white-matter lesions seen only on the long TR images were graded on a sale of 0 to 3 as follows: 0.0 = no lesions, 0.5 = less than three puntate lesions, 1.0 = more than three puntate lesions, 1.5 = more th an seven puntate lesions, 2.0 = pathy foal lesions, and 3.0 = oalesent lesions. Patients with a known ause of white-matter disease-that is, those with multiple slerosis or previous radiation therapy-were not onsidered to have nonspeifi white-matter lesions. A study sample onsisting of all 67 patients with "larger" (grades 2 and 3) Virhow-Robin spaes was seleted from the 314 patients with Virhow-Robin spaes; thi s group was defined as "positive.' The "negative" group (i.e., ontrols) onsisted of a random sample of 109 patients from among the 502 patients who were negative for Virhow Robin spaes. The random seletion was obtained by alphabetizing the 502 patients by name and taking the first 1 09 patients beginning with the letter A. Our statistiian alulated that at least 1 00 ases were required for the sample size to be statistially signifiant. The harts from the patients in the study sample were reviewed for age, gender, linial history of infartion, dementia, and hypertension. The presene of dementia was determined by the examining neurologist who referred the patient for MR. At our institution, neurologists define dementia as a loss of higher integrative faulties suh that it interferes with the ativities of daily living. A patient was onsidered to be hypertensive if he or she was being treated with mediation andjor if diastoli pressures of over 90 mm Hg had been measured on two separate oasions. Finally, several patient variables were reorded for statistial analysis: age, gender, inidental white-matter lesions. infartion (on MR), dementia, hypertension, and atrophy (greater than expeted for age only). For eah variable, the proportion of patients who were positive for that variable (e.g., positive for hypertension) was alulated for eah of the two groups and ompared aross groups by using a Fisher exat test. Multiple logisti regression analysis was used to determine whether any of these variables were jointly assoiated with being positive or negative for large Virhow-Robin spaes. The Mann-Whitney test was used to ompare the age distributions aross groups. A seond phase of the analysis was to define the positive patients as either high onvexity only, lentiulostriate only, or both high onvexity and lentiulostriate and to determine whether membership in these groups was assoiated with any of the linial or demographi fators. These analyses were also arried out by using the Fisher exat test. All results were reported as statistially signifiant if p <.05. Results Large Virhow-Robin spaes were identified in 314 patients. The age distribution of these patients is given in Figure 1, where it is ompared with the age distribution of the 502 ases negative for Virhow-Robin spaes. Note that it is in the two older groups that the perentage of positives exeeds that of the negatives. More speifially, it is in the age group that the perentage of patients positive for Virhow Robin spaes is signifiantly greater than the perentage of those who are negative. Table 1 illustrates the prevalene of the various grades of Virhow-Robin spaes found in eah age group. In general, Virhow-Robin spaes inreased with inreasing age, but larger Virhow-Robin spaes (grades 2 and 3) had a muh stronger orrelation with age while the milder grade 1 spaes were more ubiquitous. Lentiulostriate Virhow-Robin spaes were ommon and mildly age-related, while high-onvexity spaes were rarer and had a muh stronger orrelation with age, as illustrated in Figures 2-4. Two hundred eighty-five patients (35%) had lentiulostriate spaes, 1 02 (13%) had high-onvexity spaes, and 7 4 (9%) had both. In the study sample in whih only the age distribution of the larger Virhow-Robin spaes (grades 2 and 3) was analyzed and ompared with the ontrol group, the orrelation with age was further strengthened (Fig. 5). Figure 6 shows that patients who were positive for Virhow Robin spaes were more likely to have nonspeifi white :::::J Nega ti ve n Positive no31 4 OL_---LU. LD_LL_ Age in yea rs Fig. 1.- Age distribution of the 502 patients without Virhow-Robin spaes and the 314 patients with large Virhow-Robin spaes. TABLE 1: Prevalene of Grades 1-3 Virhow-Robin Spaes by Age Age No. of % by Grade Cases Grade 1 Grades 2 & 3 Total < >

3 AJ NR:1 0, September/Otober 1989 MR OF LARGE VIRCHOW-ROBIN SPACES : (l_ CJ Negative -Posi tive Age in years Age in years Fig. 2.-Age-speifi prevalene of lentiulostriate Virhow-Robin spaes seen on MR. Fig. 5.-Age distribution of patients with large (grades 2 and 3} Virhow Robin spaes vs ontrols : CJ Nega tive n 109 -Positive n Age in years White matter grade Fig. 3.-Age-speifi prevalene of high-onvexity Virhow-Robin spaes seen on MR. Fig. 6.-Prevalene and severity of white-matter lesions in patients with large (grades 2 and 3} Virhow-Robin spaes vs ontrols. HC only - LS only CJ HC + L S Age in years ' J -' Negative Positive Fig. 4.-Age-speifi prevalene of type of Virhow-Robin spae among 67 large Virhow-Robin spaes (grades 2 and 3}. HC = high onvexity; LS = lentiulostriate. Fig.?.-Prevalene of dementia in patients with large (grades 2 and 3} Virhow-Robin spaes vs ontrols.

4 932 HEIER ET AL. AJNR :l 0, September/Otober 1989 u Q:; Q) O L_ L Nego t1ve Posi tl ve Fig. B.-Prevalene of hypertension in patients with large (grades 2 and 3) Virhow-Robin spaes vs ontrols. Q L- L L Negative Positive Fig. 11.-Prevatene of atrophy in patients with large (grades 2 and 3) Virhow-Robin spaes vs ontrols. O L_ Nega tive Pos itive Virhow -Robin Fig. 9.-Prevatene of mate patients with large (grades 2 and 3) Virhow-Robin spaes vs ontrols. Fig. 12.-Simptified illustration of perivasular spae of Virhow-Robin. OL_ L Negative Positive Fig. 10.-Prevalene of stroke in patients with large (grades 2 and 3) Virhow-Robin spaes vs ontrols. matter disease (grades ) than were patients who were negative for Virhow-Robin spaes. Some variables were signifiantly assoiated with being positive for large Virhow-Robin spaes: age (p =.0001 ), hypertension (p =.009), dementia (p =.011 ), and inidental white-matter lesions (p =.0001 ). Gender (more males than females) also was mildly signifiant (p =.04). Although stroke and atrophy ourred with higher frequenies in the positive group, these results were not statistially signifiant (Figs ). All the ortial infartions had a linial history of itus while launar infarts had a history of itus in only half the ases. Owing to small numbers, stroke and atrophy were exluded from the logisti regression analysis. Logisti regression was arried out on age, gender, hypertension, dementia, and white matter. The analysis revealed that, when all these linial variables were onsidered jointly, only age remained signifiant (p =.0002).

5 AJNR :10, September/Otober 1989 MR OF LARGE VIRCHOW-ROBIN SPACES 933 The seond analysis showed no signifiant assoiation of the high onvexity/lentiulostriate grouping with any of the variables, exept for a weak assoiation with hypertension (p =.05); patients with both high onvexity and lentiulostriate Virhow-Robin spaes were more likely to have a history of hypertension. Disussion In the past, the perivasular spae has been the exlusive domain of the anatomists and pathologists, as no imaging tehnique had the resolution required to visualize suh anatomi detail. While Pestalozzi in 1849 (ited in [1 )) was probably the first to desribe a perivasular spae, what has beome known as a Virhow-Robin spae was desribed in 1851 by Virhow [2], a German pathologist, as a subadventitial spae that was ontiguous around the apillaries. In 1859, Robin [3], a Frenh anatomist, desribed a spae he A B regarded as intraadventitial and losed. Today, the eponym refers to the extension of the subarahnoid spae, whih aompanies the vessel penetrating the erebral ortex to the level of the apillaries (Fig. 12) [1, 4). While late-generation CT sanners have enabled the visualization of pathologi proesses dilating the Virhow-Robin spaes [5), it is only with the advent of high-field MR imagers and their greater resolution that presumably "empty" CSF-filled Virhow-Robin spaes have been routinely visualized [6, 7). MR sans obtained at 1.5 T routinely demonstrate small foi of CSF signal on all pulse sequenes on either side of the anterior ommissure at the level of the inferior one third of the basal ganglia; that is, in the anterior perforated substane (Figs. 13 and 14). These Virhow-Robin spaes follow the path of the lentiulostriate arteries as they enter the basal ganglia through the anterior perforated substane (Figs. 15 and 16). Infrequently, Virhow-Robin spaes are seen in the high-onvexity gray matter extending into the entrum semi Fig year-old woman with vertigo. A-C, 800/20 (A), 2000/40 (8), and 2000/80 (C) images. Large grade 3 perivasular spaes (arrows) are seen on either side of anterior ommissure (arrowheads) and are isointense relative to CSF on all pulse sequenes. Fig year-old woman with headahe. A-C, 800/20 (A), 2000/40 (8), and 2000/80 (C) images. Mild grade 1 spaes (arrows) are seen on either side of anterior ommissure (arrowheads).

6 934 HEIER ET AL. AJNR :1 0, September/Otober 1989 ovale (Figs. 17 and 18), where they follow the ourse of the penetrating ortial arterioles (Fig. 19). The artery ontained within the large Virhow-Robin spae is not identified on MR beause the assoiated perivasular spae is so muh larger and the size of the vessel itself (usually 0.4 mm and under) is beyond the limits of the sanner's resolution (0.93 mm with a 5-mm slie and 24-in. [61-m] field of view) (Figs. 168 and 19). Oasionally, it is diffiult to differentiate a launar infart from a large perivasular spae, but in our experiene and others [6-8] infartions our in the upper two thirds of the basal ganglia and are usually not of CSF intensity on all pulse sequenes. Infarts tend to be hyperintense or have hyperintense edges on long TR/short TE images and they tend to be larger, usually measuring 5 mm in diameter or more. Perivasular spaes are far more symmetri and bilateral (85% of our ases) than launae as well. A new lassifiation Fig. 15.-Gross speimen setioned in oronal plane. Lentiulostriate arteries (straight blak arrows) are seen passing through anterior perforated substane (white arrows) at level of anterior ommissure (urved arrows). Perivasular spaes are not enlarged. of erebral launae was proposed in the pathologi literature by Poirier and Derouesne [9]. Type I launae refer to old, small infarts; type II launae refer to old, small hemorrhages; and type Ill launae refer to dilated perivasular spaes. The authors speifially desribed a type IIIC launa orresponding to perivasular dilatation at the entrane of a perforating artery into the lentiular nuleus. This learly orresponds to our dilated Virhow-Robin spaes about the lentiulostriate arteries. Our results indiate that small (grade 1) Virhow-Robin spaes are found in all age groups and probably represent normal anatomy. Grade 1 Virhow-Robin spaes were seen in 23% of ases under age 20, in 33% of ases aged 21-40, in 28% of ases aged 41-60, in 33% of ases aged 61-80, and in 18% of ases aged 80 and over (Table 1 ). Highonvexity spaes are less ommon and muh more agerelated than lentiulostriate spaes, although both are found with inreasing size and frequeny with advaning age (Figs. 1-5). Age, hypertension, dementia, and inidental white-matter lesions were found to be signifiantly assoiated with large Virhow-Robin spaes (grades 2 and 3). Infartion and atrophy ourred more frequently with large spaes as well, but the numbers were not statistially signifiant (Figs ). Atrophy was noted only if it was not appropriate for the patient's age, so it must be remembered that age in this ohort is also an index of atrophy. The multiple logisti regression analysis revealed that when all the above variables were onsidered jointly, only age remained signifiant. The interpretation is that age alone aounts for the differenes between groups and that the other variables do not independently distinguish between groups. This result is not surprising, sine these linial variables are so highly orrelated with age. Beause the Virhow-Robin spae is an extension of the subarahnoid spae, it should be expeted that it would enlarge with age, as the subarahnoid spae does elsewhere in the brain. In onlusion, large Virhow-Robin spaes are another phenomenon of the aging brain. Fig. 16.-A, Severe dilated Virhow-Robin spaes (straight arrows) about anterior ommissure (urved arrow) in gross speimen setioned in oronal plane. B, Aompanying histologi setion with piallined (arrowhead) perivasular spae. Vessel (arrow) measures 0.25 mm in diameter and spae measures 1.05 mm (grade 1). MR was performed on this speimen. (Hand E, x4)

7 AJNR:1 0, September/Otober 1989 MR OF LARGE VIRCHOW-ROBIN SPACES 935 B Fig year-old man with basal ganglia hemorrhage and hypertension. A-C, 800/20 (A), 2000/40 (8), and 2000/80 (C) images show high-onvexity Virhow-Robin spaes perforating ortex (arrows). D-F, 800/20 (D), 2000/40 (E), and 2000/80 (F) images show high-onvexity Virhow-Robin spaes extending into periventriular white matter (arrows). F A B Fig year-old man with vertebral basilar ishemia. A-C, Sagiaal images (800/20) show high-onvexity Virhow-Robin spaes traversing orona radiata (arrows) from lateral to medial.

8 936 HEIER ET AL. AJNR:10, September/Otober 1989 ACKNOWLEDGMENTS We thank Martin L. Lesser for statistial analysis and Susan Cunningham for seretarial assistane. REFERENCES Fig. 19.-Histologi setion of high-onvexity Virhow-Robin spae (straight arrow) ontaining longitudinally setioned arteriole (arrowheads) in entrum semiovale. Note ortiomedullary juntion (urved arrow). MR was not performed in this ase. (Hand E, x4) 1. Woollman OHM, Millen JW. Perivasular spaes of the mammalian CNS. Bioi Rev 1954;29: Virhow R. Uber die Erweiterung kleinerer Gefasse. Virhows Arh [A] 1851 ;3: Robin C. Reherhes sur quelques partiularites de Ia struture des apillaires de l'enephale. J Physiol (Paris) 1859;2: Jones EG. On the mode of entry of blood vessels into the erebral ortex. J Anat 1970;106: Mirfakhraee M, Crofford MJ, Guinto SC Jr, Nauta HAW, Weedn VW. Virhow-Robin spae: a path of spread in neurosaroidosis. Radiology 1986;158 : Braffman BH, Zimmerman RA, Trojanowski JQ, Gonatas NK, Hikey WF, Shlaepfer WW. Brain MR: pathologi orrelation with gross and histopathology. 1. Launar infartion and Virhow-Robin spaes. AJNR 1988; 9: , AJR 1988;151 : Jungreis CA, Kanal E, Hirsh WL, Martinez AJ, Moosey J. Normal perivasular spaes mimiking launar infartion: MR imaging. Radiology 1988; 169: Brown JJ, Hesselink JR, Rothrok JF. MR and CT of launar infarts. AJNR 1988;9: , AJR 1988;151 : Poirier J, Derouesne C. Le onept de laune erebrale de 1838 a nos jours. Rev Neural (Paris) 1985;141 :3-17