Swiss Cheese Striatum: Clinical Implications

Importance  Markedly enlarged Virchow-Robin spaces throughout the striatum appear occasionally on magnetic resonance imaging (MRI) scans of the elderly, and this type of striatum is known as the Swiss cheese striatum (SCS); however, its clinical impact is unknown.

Objective  To determine the clinical features associated with SCS detected on MRI scans.

Design, Setting, and Participants  A blinded, retrospective case-control study using medical records from 2000 to 2007 obtained from an MRI database at the Mayo Clinic in Rochester, Minnesota, of residents 40 years of age or older of Olmsted County, Minnesota, who had extensive Mayo Clinic medical records and MRI reports suggestive of SCS. Cases with a severe form of SCS (n = 27) were randomly selected for comparison with age-, sex-, and examination year–matched controls (n = 52) with a minimal form of SCS or no SCS.

Exposure  Magnetic resonance imaging.

Main Outcomes and Measures  Associations of clinical and imaging features with the presence of a severe form of SCS. Medical records were reviewed for clinical features such as parkinsonism, dementia, and vascular risk factors. The MRI scans were visually scored for degree of leukoaraiosis, central atrophy, and cortical atrophy.

Results  No significant differences were found between those with a severe form of SCS and controls in rates of parkinsonism (19% vs 17%; odds ratio, 1.09 [95% CI, 0.28-4.16]) or dementia of any type (30% vs 21%; odds ratio, 1.57 [95% CI, 0.48-5.13]). Vascular risk factors were not significantly different between groups. Swiss cheese striatum correlated with degree of leukoaraiosis (P < .001). Potential associations with visualized cortical atrophy (P = .01), nonobstructive urinary incontinence (18.5% vs 3.9%; P = .04), and syncope (37% vs 9.6%; P = .01) did not hold up after correction for the false discovery rate.

Conclusions and Relevance  Our study suggests that marked cribriform change in the striatum was not associated with the development of extrapyramidal clinical disorders, including parkinsonism. The association of SCS with leukoaraiosis suggests that it is part of a more generalized cerebrovascular process. Skepticism is called for when attributing clinical symptoms to this MRI finding.

Dilated Virchow-Robin spaces may be seen on magnetic resonance imaging (MRI) scans of the brains of elderly individuals. Virchow-Robin spaces represent pial-lined extensions of the subarachnoid space surrounding the perforating arteries. More than just potential spaces, they contain interstitial fluid and macrophages. They are present around all arteries to the level of the capillary, where the pia mater fuses. On MRI scans, Virchow-Robin spaces are hypointense on T1-weighted images, hyperintense on T2-weighted images, and hypointense on fluid-attenuated inversion recovery (FLAIR) images. They can be distinguished from pathological white matter lesions by the persistence of isointensity to cerebrospinal fluid on all sequences, by the lack of enhancement, and by small, sharply defined, circular perimeters.^1,2 The distribution patterns vary, and dilated Virchow-Robin spaces can appear on MRI scans as a single enlarged space (up to 2 cm in diameter) or as hundreds of bilateral, 1- to 2-mm foci in the basal ganglia, subcortical white matter, and subinsular region lateral to the lentiform nucleus. This latter pattern is sometimes called état criblé, or cribriform change. Occasionally, patients have very prominent cribriform changes primarily localized in the striatum on MRI scans; the appearance has quite naturally led to the term Swiss cheese striatum (SCS).

One would intuitively think that marked striatal cribriform change should disrupt the functioning of the basal ganglia. Indeed, a pathological study of 191 adult brains found that 16 (8%) had SCS, and all of these individuals had either parkinsonism or pseudobulbar palsy.³ Recently, SCS has again been associated with parkinsonism in case reports.^4,5 We explored this proposed relationship between SCS and extrapyramidal symptoms to determine the extent to which these striatal imaging findings are associated with neurological dysfunction.

A retrospective case-control study, with reviews of patients’ medical records and MRI scans.

To qualify as a case patient, 2 criteria had to be met: (1) severe striatal cribriform changes in the brain had to be detected by MRI, similar to the changes illustrated in Duker and Espay’s parkinsonism case report,⁴ and (2) adequate medical records sufficient to document the neurologic status of the patient when MRI was performed had to be available. The latter criterion was facilitated by limiting the search to Mayo Clinic patients who were residents of Olmsted County, Minnesota. These participants were identified by querying the Mayo Rochester Radiology Information Management System, a database of all brain MRI reports created at the Mayo Clinic in Rochester, Minnesota, since 1997. With this type of search, we selected brain MRI reports from 2000 to 2007 of individuals 40 years of age or older that included the term perivascular spaces or Virchow-Robin/Virchow Robin spaces and the term basal ganglia. The list of scans was randomized using a list randomizer (http://www.random.org), and then each MRI scan was consecutively viewed by M.S.B., who was blinded to the clinical data, to select those scans of participants’ brains that had the striking état criblé pattern within the striatum, similar to Duker and Espay’s case report.⁴

A control group was assembled from a list of MRI reports that did not contain the above-mentioned search terms. That list was randomized, and then a subset of controls was matched to cases by age, sex, and year of MRI scan, with approximately 2 controls for each case. The control group MRI scans were visually reviewed to confirm the absence of multiple enlarged striatal perivascular spaces.

The research authorization status for inclusion in retrospective studies was confirmed for every participant. The research protocol was approved by the independent review board of the Mayo Clinic. Participants provided written informed consent and did not receive compensation.

Blinded to case-control status and MRI results, M.S.B. searched the medical records of each participant for the presence or absence of a number of clinical measures, including movement disorders, vascular risk factors and events, dementia, sleep disorders, autonomic dysfunction and disorders, and general medical conditions. Such measures were considered present if they were listed in the participant’s medical history or rendered as a formal diagnosis by a Mayo Clinic clinician before, at the time of, or after the MRI scan of interest. Parkinsonism was defined as pertinent symptoms, with confirmation of 2 or more cardinal parkinsonian features of tremor, bradykinesia, rigidity, or extrapyramidal gait disorder/postural instability. Gait disorders were defined as any complaint of imbalance or falls with confirmatory examination findings or a history of use of a gait aid. Sleep disorders were documented if diagnosed by a Mayo Clinic physician, if the patient required a device or medication to assist with a sleep disorder, or if there was a history of an abnormal polysomnogram.

The MRI scans included both 1.5- and 3-T MRI scans. Various protocols were used, but all included axial 2-dimensional FLAIR sequences. The slice thickness varied between 3 and 5 mm. Two investigators (R.J.W. and M.S.B., who were blinded to the clinical data) independently rated approximately one-third of the scans for atrophy, cribriform change, and leukoariosis and then compared their ratings. Once M.S.B.’s ratings were consistently within a half-grade of R.J.W.’s ratings, M.S.B. then reviewed the MRI scans twice, using her last measured value as the final value, with approximately 90% intrarater reliability for leukoariosis and 80% intrarater reliability for atrophy. If a participant had more than 1 MRI scan, the most recent scan was reviewed.

We visually reviewed the MRI scans to select those with the most prominent cribriform changes, similar in degree to the case in Duker and Espay’s case report.⁴ The axial FLAIR series was predominantly used for review, with coronal and axial T1-weighted images (if available) used only for confirmation of the general impression, because many scans lacked an axial T1-weighted series. Perivascular spaces were defined as round or oval, well-demarcated lesions with a diameter typically less than 3 mm with a center intensity matching that of cerebrospinal fluid, which is dark on FLAIR images. The difficult task of distinguishing between enlarged perivascular spaces and lacunar infarcts was addressed by using criteria similar to that of the Cardiovascular Health Study⁶: lacunar infarcts should be brighter on spin density (if obtained) and T2-weighted images than normal gray matter. We focused on evaluating perivascular spaces in the superior basal ganglia and striatum (rostral to the anterior commissure). Typically, the most severe cases had 20 or more perivascular spaces visible in the striatum on the most affected side.

We used the semiquantitative, 0 to 9 grading scale that was used in the Cardiovascular Health Study⁷ to assess leukoaraiosis in the periventricular and subcortical white matter.

Central atrophy was defined here as the degree of ventricular enlargement regardless of the degree of sulcal widening. This way, we assessed degenerative processes affecting the white matter or deep gray matter but not necessarily affecting most of the cortex. Central atrophy was analyzed in 2 ways. First, it was estimated by measuring the third ventricular size, defined as the smallest distance between the thalami (in millimeters), usually at the midpoint of a line extending the length of the third ventricle along the anterior-posterior axis. Measurements were done on the axial FLAIR image where the third ventricle was of maximum length. The second measure of central atrophy was the intercaudate ratio, which was measured as previously described.⁸ In summary, the distance between the caudate nuclei were measured on the caudal-most axial slice where the caudates were prominent. The intercaudate ratio was defined as this distance normalized by the transverse width of the inner table of the skull at the same level. Cortical atrophy was estimated by a visual rating of global sulcal width and gyral volume loss on a semiquantitative scale ranging from 0 to 3, with 1 equal to mild atrophy, 1.5 equal to mild-moderate atrophy, 2 equal to moderate atrophy, 2.5 equal to moderate-severe atrophy, and 3 equal to severe, maximal atrophy.

We compared frequencies of risk factors and imaging findings between the case and control groups using χ² tests or the Fisher exact test, when indicated. Semiquantitative, ordinal scale measures were compared with the Mann-Whitney U test, and continuous measures were compared with the 2-tailed t test. Multiple comparisons were controlled for by adjusting the P value cutoff of less than .05 using the false discovery rate method of Benjamini and Hochberg.⁹

A total of 13 058 MRI scans were performed at the Mayo Clinic on consenting Olmsted County residents aged 40 years or older during the period from 2000 to 2007. Of the 13 058 MRI scan reports, 956 (7.3%) contained the term perivascular spaces or Virchow-Robin/Virchow Robin spaces and the term basal ganglia. The median age of these 956 patients was 70 years (mode, 79 years). We randomly selected a subset of 151 MRI scans, and these were reviewed by M.S.B. in a blinded fashion to identify those participants with severe cribriform changes. Ultimately, a total of 27 of 151 case patients (18%) were found to have severe cribriform change; all had extensive clinical records. The demographic characteristics of this group are shown in Table 1. Typical MRI scans of cases are shown in our Figure. The mean age of these 27 case patients was 83 years; only 6 (22%) were female. All but 1 participant was white, reflecting the demographics of the region.

Case patients with the severe cribriform change were compared with the matched controls. There were no significant differences between the 2 groups in rates of the following features: parkinsonism, any gait disorder, extrapyramidal gait disorders, hyperkinetic movement disorders, mild cognitive impairment, and dementia of any type (Table 2). In addition, there were no significant differences between the 2 groups in the frequency of medical conditions previously reported to be associated with SCS: cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy, traumatic brain injury, cranial radiation therapy, migraine, depression, multiple sclerosis, sarcoidosis, cryptococcosis, adrenoleukodystrophy, mucopolysaccharidosis, and diffuse carcinomatous leptomeningeal metastases (Table 2). In contrast to previous reports,^19,20 vascular risk factors did not vary between groups (case patients vs controls), including history of stroke (33% vs 23%; P = .42), myocardial infarction (48% vs 46%), diabetes mellitus (18.5% vs 23%; P = .78), hypertension (81% vs 67%; P = .29), hyperlipidemia (48% vs 60%; P = .35), or smoking 20 packs or more of cigarettes per year (30% vs 29%). Other clinical symptoms were also assessed, and unexpected associations were found between SCS and nonobstructive urinary incontinence (18.5% vs 3.9%; P = .04) and between SCS and syncope (37% vs 9.6%; P = .006). However, these associations were not significant after correction for multiple comparisons.

Swiss cheese striatum was associated with higher leukoaraiosis scores, and this remained significant when corrected for multiple comparisons. Although SCS was associated with higher visualized cortical atrophy estimates, this relationship did not hold up after correction for multiple comparisons. No differences were found between the 2 groups in quantitative measures of central atrophy (third ventricular width or intercaudate ratio), even before correcting for multiple comparisons (Table 3).

Intuitively, one would expect that prominent striatal cribriform changes should translate into parkinsonism or another extrapyramidal syndrome; this was the assumption in recent case reports.^4,5,21 Because striatal cribriform imaging changes occasionally surface in our movement disorders practice, we were interested in determining whether these are clinically important. We focused on more extreme cases, with marked striatal imaging features, colloquially referred to as Swiss cheese striatum. Although selecting for the most advanced cases, we were unable to find an association with parkinsonism, gait problems in general, hyperkinesias, or cognitive impairment/dementia. These aggregate findings do not exclude the possibility that such cribriform pathology makes some contribution to striatal dysfunction, but they argue against this being a primary or sole cause of extrapyramidal syndromes.

Previous studies linking striatal cribriform changes to parkinsonism have not included comparative control groups and have typically selected the cases. Note that with advancing age, parkinsonism becomes increasingly prevalent; in one community-based study,²² clinical parkinsonism exceeded 50% by age 85 years. Hence, cribriform pathology and parkinsonism may be coincidental in an elderly cohort. Moreover, this coincidental relationship may be increased in pathologic series,³ which are notorious for selection bias.

Our selection of cases with the most severe cribriform change limited the number of case patients (n = 27) and, hence, the power of our study. However, the 95% CIs for both parkinsonism and dementia include 1 patient and are fairly narrow (odds ratios, 1.09 [95% CI, 0.28-4.16] and 1.57 [95% CI, 0.48-5.13], respectively), supporting the idea that the null hypothesis is more likely than not for these factors.

Consistent with our interpretation were findings from an MRI study²³ tabulating Virchow-Robin spaces/cysts in the basal ganglia of patients with Parkinson disease who were scheduled for a pallidotomy. Ironically, parkinsonism tended to be more severe on the side ipsilateral to the larger lesion burden; this is opposite to what would be predicted if such cribriform findings translated into extrapyramidal symptoms.

We investigated risk factors for SCS. Aging was one obvious association, given the mean age of 83 years in the SCS group. In our comparison of case patients with the age- and sex-matched controls, we found that the degree of leukoaraiosis was associated with SCS. This association supports the results of other studies.^24,25 Hypertension has previously been associated with enlarged perivascular spaces in prior investigations,^20,26,27 and such an association has also been found in experimental hypertensive rats.²⁸ Our case patients were more likely than controls to have a hypertension diagnosis (81% vs 67%), although this was not statistically significant. The retrospective nature of our study limited the confident assessment of clinical details, such as hypertension duration, severity, and medication response, which may have shed additional light on our findings. In addition, somewhat unexpectedly in our series, SCS was not associated with other vascular risk factors, such as hyperlipidemia, diabetes, smoking, history of stroke, or coronary artery disease; however, this is consistent with findings in other studies.^29,30 Note that such vascular risk factors are common in elderly populations, and this may overshadow contributory relationships in small series such as ours. To summarize, we found associations with age and leukoaraiosis, plus a nonsignificant but plausible association with hypertension. Other unknown factors must also play a role in the development of SCS because leukoaraiosis is much more common in aging than SCS.

Dilated Virchow-Robin spaces have been associated with vascular dementia,³¹ and enlarged perivascular spaces (especially in the basal ganglia) have correlated with reduced cognitive function in men.³² Virchow-Robin spaces were also found to be more prominent in those with Alzheimer disease and mild cognitive impairment than in cognitively normal adults.³³ Recently, it has been hypothesized that the flow of fluid through the cerebral interstitial and perivascular space serves an important role in the clearance of toxic solutes from the brain, with changes to this flow a possible contributor to dementia.³⁴ Dilated Virchow-Robin spaces may reflect such processes. Hence, we also assessed the frequency of cognitive impairment/dementia in our cases. Although we did find slightly more cases of dementia than cognitive impairment associated with SCS (30% vs 20%), this was not statistically significant. Given the significant association of SCS with leukoaraiosis, plus a trend toward an association with visualized cortical atrophy, this possible link with dementia may point toward SCS as a marker of cerebral and vascular aging.

In a secondary analysis, we also identified significant associations of SCS with syncope and urinary incontinence (presumably neurogenic bladder). However, these findings did not hold up after correction for multiple comparisons. Nevertheless, the association with syncope complements a recent study⁶ detecting a link between syncope and leukoaraiosis.

In conclusion, severe cribriform change in the basal ganglia detected on MRI scans was not associated with extrapyramidal or cognitive disorders. Conceptually, MRI-detected SCS should contribute to clinical conditions (similar to leukoaraiosis); however, it does not appear to be the primary explanation for basal ganglia disorders/parkinsonism, nor for cognitive impairment.

Accepted for Publication: February 4, 2014.

Corresponding Author: Melinda S. Burnett, MD, Division of Movement Disorders, Department of Neurology, Mayo Clinic, 200 First St SW, Rochester, MN 55905 (burnett.melinda@mayo.edu).

Published Online: April 14, 2014. doi:10.1001/jamaneurol.2014.286.

Author Contributions: Dr Burnett had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Burnett, Ahlskog.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Burnett.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Burnett.

Conflict of Interest Disclosures: None reported.

Additional Contributions: We thank Mandie Maroney-Smith for assistance with MRI analysis, Allen Omdahl and Vicki Schmidt for assistance with the MRI database, and Jay Mandrekar for statistical advice. Written permission for inclusion in the acknowledgments has been obtained from all persons named, and none received any monetary compensation for their work.