Cerebral Microbleeds Remain for Nine Years: A Prospective Study with Yearly Magnetic Resonance Imaging

ARTICLE IN PRESS

Cerebral Microbleeds Remain for Nine Years: A Prospective Study with Yearly Magnetic Resonance Imaging Tsukasa Saito, MD, PhD, Yuichiro Kawamura, MD, PhD, Nobuyuki Sato, MD, PhD, Eitaro Sugiyama, MD, Motoi Okada, MD, PhD, Toshiharu Takeuchi, MD, Kazumi Akasaka, MD, PhD, and Naoyuki Hasebe, MD, PhD

Background: Cerebral microbleeds (CMBs) are refined neuroimaging findings detected on T2*-weighted gradient echo (GRE) magnetic resonance imaging (MRI) and are widely accepted as an important marker of the vulnerability of cerebral small vessels. It is necessary to further clarify the natural history of CMBs by a longitudinal study. This study aimed to reveal the natural history of CMBs and find a better way to track CMBs by a prospective long-term observation. Methods: We performed yearly brain MRI assessments for 7 or more years in 8 nonvalvular atrial fibrillation Japanese outpatients with CMBs detected in the baseline MRI. We began to use a 3.0T MRI scanner from 2012 as well. Results: We followed up 3 patients for 9 years, 2 for 8 years, and 3 for 7 years. In all patients, the CMBs at baseline did not disappear during the follow-up period. Importantly, the CMB in 1 patient seemed to disappear during the sixth imaging using 1.5T T2*weighted GRE but was detected again during the seventh imaging with 3.0T susceptibility weighted imaging and ninth imaging with 3.0T T2* GRE. Moreover, in a patient implanted with a pacemaker, which is only applicable for 1.5T MRI at present, the CMB seemed to disappear and appeared once again with a 1.5T T2*-weighted GRE at a slice thickness of 2.5 mm instead of 5 mm. Conclusions: From this prospective study, we obtained 2 absolutely new findings that CMBs remained for as long as 9 years and a high-field or thin-slice MRI can detect concealed CMBs. Key Words: T2*-weighted GRE—SWI—cerebral microbleeds—Atrial fibrillation—cerebral hemorrhage. © 2017 National Stroke Association. Published by Elsevier Inc. All rights reserved.

Introduction Cerebral microbleeds (CMBs, also known as cerebral microhemorrhages) are defined as small, from 2 to 5 mm

From the Department of Internal Medicine, Cardiovascular, Respiratory and Neurology Division, Asahikawa Medical University, Asahikawa, Japan. Received June 29, 2017; revision received August 25, 2017; accepted September 3, 2017. Address correspondence to Tsukasa Saito, MD, PhD, Department of Internal Medicine, Cardiovascular, Respiratory and Neurology Division, Asahikawa Medical University, Midorigaoka Higashi 2-1-1-1, Asahikawa, Hokkaido 078-8510, Japan. E-mail: tsukasa@asahikawa -med.ac.jp. 1052-3057/$ - see front matter © 2017 National Stroke Association. Published by Elsevier Inc. All rights reserved. https://doi.org/10.1016/j.jstrokecerebrovasdis.2017.09.001

in diameter, round foci with a hypointensity detected on T2*-weighted gradient echo (GRE) magnetic resonance imaging (MRI).1,2 CMBs represent a perivascular accumulation of hemosiderin-containing macrophages and pericytes on histopathologic examinations,3-6 and therefore reflect a hemorrhage-prone microvascular fragility. There is consistent evidence of a significant association between the presence of CMBs at baseline and the risk of recurrent strokes.7 We found that CMBs are frequently observed in atrial fibrillation (AF) patients taking warfarin and can predict the subsequent increase in CMBs and the occurrence of asymptomatic cerebral infarctions.8,9 Stroke specialists know well that we can easily detect scars from cerebral hemorrhages by T2*-weighted GRE imaging as black areas over the years. Applying that idea, CMBs that reflect minute bleeding could also remain over the years. It is possible that the CMBs were created by

Journal of Stroke and Cerebrovascular Diseases, Vol. ■■, No. ■■ (■■), 2017: pp ■■–■■

1

ARTICLE IN PRESS T. SAITO ET AL.

2

an environment in the past and continued to exist until we discovered them. In other words, CMBs do not necessarily accurately explain the state at the time of the imaging. We believe that it is more important to capture a new emergence rather than the presence of an already existing CMB, and we have eagerly conducted a prospective study. Based on the hypothesis that CMBs do not easily disappear, we carried out this prospective longterm follow-up with MRI exams performed yearly. Furthermore, we discovered some interesting and important findings about the natural history of CMBs. The purpose of our study was to reveal the natural history of CMBs and to find a better way to track CMBs by a prospective long-term observation.

Methods The entire study protocol was approved by the ethical review board of Asahikawa Medical University, and all patients gave their written informed consent for the study.

Patient Enrollment and the Longitudinal Study A brain MRI assessment was performed on Japanese outpatients aged older than 45 years with AF who visited the Cardiovascular, Respiratory and Neurology Divisions of Asahikawa Medical University Hospital, and they were consecutively enrolled in the study from 2008. Valvular AF patients were excluded. A baseline MRI examination was performed on the patients enrolled, and yearly MRI exams were repeated thereafter. When a patient refused an MRI in a certain year, we recommended that the patient receive an MRI the following year so that the patient would not drop out of this study. As a result, even if the patient eventually skipped an MRI, we did not treat him or her as a censored case if he or she continued to undergo MRI exams.

MRI Assessment We adopted the recommended criteria for the identification of CMBs proposed by Greenberg et al1 to precisely assess the behavior of CMBs: (1) a black lesion on the T2*-weighted MRI; (2) round or ovoid lesions (rather than linear); (3) a blooming effect of the T2*-weighted MRI; (4) signals devoid of hyperintensity on the T1-weighted or T2-weighted sequences; (5) at least half of the lesion was surrounded by brain parenchyma; (6) distinct from other potential mimicking conditions such as iron or calcium deposits, bone, or vessel flow voids; and (7) a clinical history excluding any traumatic diffuse axonal injury. The MRI assessments were performed by a trained neuroradiologist with 10 years of experience in reading neurological magnetic resonance images.

MRI Protocol Imaging of the brain was initially performed on 3 1.5T MRI scanners at our hospital, which included Signa Excite (General Electric Medical Systems, Waukesha, WI), Magnetom Sonata (Siemens Medical Solutions, Munich, Germany), and Achieva MR (Philips Healthcare, Bothwell, WA) scanners. The GRE sequence parameters for the Signa Excite were as follows: 22 axial images; field of view, 220 mm; slice thickness, 5.5 mm; interslice gap, 1.0 mm; 256 × 256 matrix; echo time, 20 ms; repetition time, 640 ms; and flip angle, 20°. For the Magnetom Sonata (Siemens Medical Solutions), the parameters were 22 axial images; repetition time, 600 ms; echo time, 18 ms; slice thickness, 5.5 mm; interslice gap, 1.0 mm; field of view, 175 × 200 mm; and flip angle, 20°. For the Achieva MR scanner (Philips Healthcare), the parameters were repetition time, 700 ms; echo time, 20 ms; slice thickness, 5.5 mm; interslice gap, 1.0 mm; field of view, 210 mm; and flip angle, 20°. From 2012, we started to use a 3.0T MRI scanner, the Discovery MR750w (General Electric Healthcare) instead of the Magnetom Sonata (Siemens Medical Solutions, Waukesha, WI). The GRE sequence of this 3.0T MRI scanner was called a multiple-echo combined gradient echo, and its parameters were as follows: 160 axial images; field of view, 220 mm; slice thickness, 2.0 mm; 512 × 512 matrix; repetition time, 30 ms; echo time, 12.4 ms; and flip angle, 5°.

Results Since 2008, 131 nonvalvular AF patients were enrolled, and 33 of them completed a follow-up period of 7 years or more. Of the 33 patients, 8 had CMBs during the baseline MRI. Importantly, the CMBs in all 8 patients at baseline did not disappear during the followup period. We followed up 3 cases for 9 years (Fig 1), 2 for 8 years (Fig 2), and 3 for 7 years (Fig 3). Table 1 shows the medications and new MRI findings during the follow-up period in those 8 cases. In 3 of the 8 patients with baseline CMBs, 1 CMB newly appeared respectively during the follow-up period. On the other hand, in 3 of 25 patients without baseline CMBs, 1 CMB appeared respectively during the follow-up period. The Fisher exact test revealed that there was a tendency that CMBs appeared more frequently in the group with baseline CMBs, but it was not significant (P = .078). No symptomatic cerebral infarctions or hemorrhages occurred in 33 patients. Asymptomatic cerebral infarctions occurred in 2 of 8 cases with baseline CMBs and in 2 of 25 cases without CMBs, and there was no significant difference between the 2 groups (P = .561). A brief description of the natural histories of the CMBs in each case are given below. The CMB of the right putamen in case 1, a 69-year-old man, was confirmed at baseline and continued to be observed for 9 years (Fig 1). The CMB of the right parietal lobe in case 2 of a 56-year-old woman was confirmed at

ARTICLE IN PRESS CMBS REMAIN FOR 9 YEARS

3

Figure 1. Time series display of MRI T2*weighted gradient echo images in 3 cases observed for 9 years. In each image, the magnetic field intensity of the MRI used and slice thickness are described. An arrow indicates a cerebral microbleed that can be clearly pointed out. The imaging in case 3 in 2013 and 2015 was skipped. Note that the imaging in case 2 in 2015 was performed exceptionally with 3.0T SWI. Abbreviations: MRI, magnetic resonance imaging; SWI, susceptibility weighted imaging.

Figure 2. Time series display of MRI T2*weighted gradient echo images in 2 cases observed for 8 years. In each image, the magnetic field intensity of the MRI used and slice thickness are described. The arrow indicates a cerebral microbleed that can be clearly pointed out. The imaging in case 4 in 2010 was skipped. Abbreviation: MRI, magnetic resonance imaging.

Table 1. Drugs and newly appearing MRI findings of the enrolled patients during the follow-up period

Case 1 Case 2 Case 3 Case 4 Case 5 Case 6 Case 7 Case 8

Age at baseline

Sex

Follow-up years

Antiplatelets

Anticoagulants

Development of new ACI

Development of new CMB

69 56 74 59 65 78 68 76

M W M M M M W M

9 9 9 8 8 7 7 7

− Aspirin − − Aspirin Clopidogrel − −

War War 4 y → Dab 5 y War 4 y → Riv 5 y War 4 y → Riv 4 y − − War 4y → Riv 3y War

− + − − − + − −

+ − − − − + − +

Abbreviations: ACI, asymptomatic cerebral infarct; CMB, cerebral microbleed; Dab, dabigatran; M, man; Riv, rivaroxaban; W, woman; War, warfarin.

ARTICLE IN PRESS 4

T. SAITO ET AL.

Figure 3. Time series display of MRI T2*weighted gradient echo images in 3 cases observed for 7 years. In each image, the magnetic field intensity of the MRI used and slice thickness are described. The arrow indicates a cerebral microbleed that can be clearly pointed out. The imaging in case 8 in 2013 was skipped. Abbreviation: MRI, magnetic resonance imaging.

baseline and became almost invisible in 2014 (Fig 1). However, the CMB became visible under 3.0T susceptibility weighted imaging (SWI) experimentally performed in the ensuing year. Moreover, after disappearing once again in 2016 under 1.5T T2*-weighted GRE, the CMB made an appearance at last in 2017 by means of 3.0T T2*-weighted GRE. The CMB of the right putamen in case 3, a 74-yearold man at baseline, was detected until 2014; however, it seemed to have almost disappeared in 2016 (Fig 1). Because a permanent pacemaker that was not compatible with 3.0T MRI was implanted in the patient in late 2016, we had to shoot a thin-slice T2*-weighted GRE as thin as possible using 1.5T MRI at a slice thickness of 2.5 mm instead of 5 mm, and then the CMB reappeared again in 2017. Depending on the patient’s request, imaging was not performed in 2013 and 2015. Case 4 was a 59-year-old, and case 5 was a 65-year-old, both men. The CMBs in

the right temporal lobe and left thalamus in cases 4 and 5, respectively (Fig 2), continued to be observed from 2009 to 2016. Imaging in case 4 in 2010 was skipped. Case 6 was a 78-year-old man, case 7 was a 68-year-old woman, and case 8 was a 76-year-old man. A total of 3 CMBs in cases 6, 7, and 8 (Fig 3) in the right cerebellum, left hypothalamus, and right parietal lobe, respectively, were confirmed for 7 years. The imaging in case 8 in 2013 was skipped. Furthermore, we presented 2 additional cases, both 63year-old men, with CMBs that seemed to have hidden themselves but have become unmasked by 3.0T MRI, but the men’s CMBs were not present at baseline and have not remained that long yet (Fig 4). The CMB in case 9 that appeared in 2011 in the right frontal lobe seemed to disappear once in 2015; however, it was brought to light by 3.0T MRI in 2016. The CMB in case 10 in the left parietal lobe was detected by 3.0T MRI initially in 2013;

Figure 4. Time series display of MRI T2*weighted gradient echo images in 2 cases with CMBs that once disappeared but were unmasked by 3.0T MRI. However, their CMBs were not present at baseline. In each image, the magnetic field intensity of the MRI used and slice thickness are described. The arrow indicates a CMB that can be clearly pointed out. The imaging in case 10 in 2015 was skipped. Abbreviations: CMB, cerebral microbleed; MRI, magnetic resonance imaging.

ARTICLE IN PRESS CMBS REMAIN FOR 9 YEARS

however, it concealed itself during the following year’s 1.5T MRI. After 2015, when the imaging was skipped, the CMB exposed itself again in 2016 with 3.0T MRI.

Discussion This is the first study to show that CMBs can continue to exist for as long as 9 years using yearly performed MRIs in a prospective manner. Further, we proved that the CMBs in 2 patients seemed to disappear at some point using 1.5T MRI but became detected afterward with 3.0T MRI. Moreover, in the patient with an implanted pacemaker only applicable for 1.5T MRI, the CMB that seemed to disappear once was detected during imaging with a slice thickness of 2.5 mm.

5

wall and were surrounded by a basement membrane that also contained dense deposits of iron.6 The real identity and natural history of CMBs are still a mystery. If CMBs detected by T2*-weighted GRE someday disappear, it means that hemosiderin-laden macrophages or pericytes moved to somewhere from a certain place. In fact, both macrophages and pericytes have a migratory function.15 This is total speculation, but otherwise brain macrophages might release iron into the peripheral blood as they do when they are in the liver or spleen. As we know, most subcutaneous hemorrhages macroscopically disappear without any scar after a while. However, on the other hand, we stroke clinicians know well that the scar of cerebral hemorrhages detected by T2*-weighted GRE survive without disappearing for many years. Macrophages might behave differently in the brain and other places.

CMBs Exist for a Long Time One of the most important findings of this study was the fact that CMBs continue to be detected for as long as 9 years by T2*-weighted GRE. Although many reports regarding CMBs have been published until now, reports discussing the natural history of CMBs are surprisingly few. In particular, with respect to the disappearance of CMBs, there are only a few fragmentary reports derived from prospective studies during a short period of followup. Greenberg et al first reported that 2.3% of CMBs at baseline became undetectable after an average followup period of 17.4 ± 7.6 months using 1.5T T2*-weighted GRE with a 6-mm slice thickness.10 Goos et al reported that 2% of patients with CMBs present at baseline had fewer CMBs visible after 1.9 ± .9 years using 1.0T T2*weighted GRE with a 5-mm slice thickness.11 The Rotterdam Scan Study reported that CMBs that were present at baseline seemed to disappear in 12 persons (5.9% of the patients with CMBs at baseline) after a mean follow-up period of 3.4 years using 1.5T T2*-weighted GRE with a 5-mm slice thickness.12 Gregoire et al reported that 1 of 62 baseline CMBs in 8 patients disappeared after 5.6 years; however, they concluded that it was due to artifact.13 In the present study, in contrast to those previous findings, we revealed invulnerable CMBs through an unprecedented observation period of 9 years using yearly performed MRI exams with a high-field or thin-slice thickness that minimized the possibility of overlooking CMBs.

Elusive CMBs A radiological-pathological study revealed that CMBs detected by T2*-weighted GRE in autopsied brains correspond pathologically to clusters of hemosiderin-laden macrophages 14 ; however, both CMBs seen on T2*weighted GRE, but not on histopathologic examination, and CMBs that are not detected by T2*-weighted GRE identified on histopathologic examination have also been described.5 Moreover, a report by Fisher et al using electron microscopy revealed pericytes attached to the capillary

Knacks of Evaluating CMBs Interestingly and of next importance, the CMBs that seemed to disappear with 1.5T T2*-weighted GRE again became obvious with 3.0T T2*-weighted GRE or a thin-slice 1.5T T2*-weighted GRE in the present study. On the basis of these findings, we think that CMBs, which were concluded to have disappeared in the previous reports, were merely not detected, simply because of methodological issues. Some studies have already reported that imaging with 3.0T T2*-weighted GRE is superior to 1.5T for the detection of CMBs.16,17 Further, if the other conditions are the same, sequences with a thinner slice thickness can detect more CMBs regardless of whether the MRI used was 1.5T or 3.0T.18 From our experience, we recommend using 3.0T MRI as much as possible to evaluate CMBs. However if it is impossible, we also recommend 1.5T MRI with a slice thickness as thin as possible, such as 2.5 mm. Further, the most important thing in order not to lose sight of CMBs is to repeat the MRI several times, as we have done. Recently many studies have recommended SWI to evaluate CMBs.19 In the present study, 3.0T SWI could detect CMBs that were no longer detected by 1.5T T2*-weighted GRE, similar to the 3.0T T2*-weighted GRE. Because of the 3-dimensional (3D), higher field strength, thin-slice thickness, and phase information, conventional SWI can yield a greater CMB detection than GRE T2*.18 However, we must be aware of the confidence in SWI for diagnosing CMBs regarding the point that enhanced detection by SWI does not improve the clinical relevance.20 It is also known that the 3D multiecho GRE T2*-weighted angiography sequence we used in this study is known to be equivalent to the SWI-spoiled gradient-recalled echo sequence regarding the diagnostic ability of detecting CMBs.21

Anticoagulants and Natural History of CMBs The appearance of new CMBs seemed to be more common in patients who had continuously taken warfarin throughout the follow-up period. No new CMBs occurred

ARTICLE IN PRESS T. SAITO ET AL.

6

in the patients whose anticoagulants were changed to direct oral anticoagulants (DOACs) in the course of the followup period of this study. Although this was determined from a small number of cases, this was consistent with our previous report that concluded that warfarin increased CMBs and DOACs did not.9 The influence of anticoagulants on the new appearance of CMBs is becoming apparent as we have shown in previous reports. On the other hand, it is completely unknown how anticoagulants affect the long-term survival in patients with CMBs. Unlike our previous report,8 the presence of baseline CMBs was not significantly associated with the subsequent appearance of CMBs and asymptomatic cerebral infarctions. It is possible that this result may be affected by the fact that in half of the 8 patients with baseline CMBs the anticoagulants had been replaced with DOACs, which were less likely to cause CMBs. This study had several limitations. First, in the present study we could not conclude in the end whether the CMBs disappear. Although we observed the fact that the CMBs remained for as long as 9 years and that a high-field or thin-slice MRI could reveal hidden CMBs, we still could not deny the possibility that the CMBs could disappear 20 or 30 years later. In order to draw a conclusion to this problem, further long-term follow-up is required. Second, the results of this research have limited effectiveness because of the small number of cases. Not all CMBs will always behave the same. In other words, there is even the possibility that both disappearing CMBs and amaranthine CMBs will exist. Large prospective studies are needed to address the natural history of CMBs.

Conclusions We conclude that CMBs remain for as long as 9 years using yearly performed MRIs in a prospective manner. Further, we revealed the fact that 3.0T MRI or a thinslice thickness, such as 2.5 mm, can detect CMBs that once disappeared in the 1.5T MRI.

References 1. Greenberg SM, Vernooij MW, Cordonnier C, et al. Cerebral microbleeds: a guide to detection and interpretation. Lancet Neurol 2009;8:165-174. 2. Offenbacher H, Fazekas F, Schmidt R, et al. MR of cerebral abnormalities concomitant with primary intracerebral hematomas. AJNR Am J Neuroradiol 1996; 17:573-578. 3. Cole FM, Yates PO. Comparative incidence of cerebrovascular lesions in normotensive and hypertensive patients. Neurology 1968;18:255-259. 4. Shoamanesh A, Kwok CS, Benavente O. Cerebral microbleeds: histopathological correlation of neuroimaging. Cerebrovasc Dis 2011;32:528-534.

5. Fazekas F, Kleinert R, Roob G, et al. Histopathologic analysis of foci of signal loss on gradient-echo T2*weighted MR images in patients with spontaneous intracerebral hemorrhage: evidence of microangiopathyrelated microbleeds. AJNR Am J Neuroradiol 1999;20:637642. 6. Fisher M, French S, Ji P, et al. Cerebral microbleeds in the elderly: a pathological analysis. Stroke 2010;41:27822785. 7. Charidimou A, Kakar P, Fox Z, et al. Cerebral microbleeds and recurrent stroke risk: systematic review and metaanalysis of prospective ischemic stroke and transient ischemic attack cohorts. Stroke 2013;44:995-1001. 8. Saito T, Kawamura Y, Tanabe Y, et al. Cerebral microbleeds and asymptomatic cerebral infarctions in patients with atrial fibrillation. J Stroke Cerebrovasc Dis 2014;23:1616-1622. 9. Saito T, Kawamura Y, Sato N, et al. Non-vitamin K antagonist oral anticoagulants do not increase cerebral microbleeds. J Stroke Cerebrovasc Dis 2015;24:13731377. 10. Greenberg SM, O’Donnell HC, Schaefer PW, et al. MRI detection of new hemorrhages: potential marker of progression in cerebral amyloid angiopathy. Neurology 1999;53:1135-1138. 11. Goos JD, Henneman WJ, Sluimer JD, et al. Incidence of cerebral microbleeds: a longitudinal study in a memory clinic population. Neurology 2010;74:1954-1960. 12. Poels MM, Ikram MA, van der Lugt A, et al. Incidence of cerebral microbleeds in the general population: the Rotterdam Scan Study. Stroke 2011;42:656-661. 13. Gregoire SM, Brown MM, Kallis C, et al. MRI detection of new microbleeds in patients with ischemic stroke: five-year cohort follow-up study. Stroke 2010;74:1954-1960. 14. Tatsumi S, Shinohara M, Yamamoto T. Direct comparison of histology of microbleeds with postmorterm MR images: a case report. Cerebrovasc Dis 2008;56:537-539. 15. Dore-Duffy P, Owen C, Balabanov R, et al. Physiologic angiodynamics in the brain. Antioxid Redox Signal 2000;9: 1363-1371. 16. Scheid R, Ott DV, Roth H, et al. Comparative magnetic resonance imaging at 1.5 and 3 Tesla for the evaluation of traumatic microbleeds. J Neurotrauma 2007;24:18111816. 17. Stehling C, Wersching H, Kloska SP, et al. Detection of asymptomatic cerebral microbleeds: a comparative study at 1.5 and 3.0 T. Acad Radiol 2008;15:895-900. 18. Nandigam RNK, Viswanathan A, Delgado P, et al. MR imaging detection of cerebral microbleeds: effect of susceptibility-weighted imaging, section thickness, and field strength. AJNR Am J Neuroradiol 2013;34:984-989. 19. Shams S, Martola J, Cavallin L, et al. SWI or T2*: which MRI sequence to use in the detection of cerebral microbleeds? The Karolinska Imaging Dementia Study. AJNR Am J Neuroradiol 2015;36:1089-1095. 20. Goos JD, van der Flier WM, Knol DL, et al. Clinical relevance of improved microbleed detection by susceptibility-weighted magnetic resonance imaging. Stroke 2011;42:1894-1900. 21. Guo LF, Wang G, Zhu XY, et al. Comparison of ESWAN, SWI-MPGR, and 2D T2*-weighted GRE sequence for depicting cerebral microbleeds. Clin Neuroradiol 2013;23:121-127.