Artículo original Anales de Radiología México 2016 Apr;15(2):94-102.

T2 gradient echo sequence versus susceptibility-weighted sequence in detecting microhemorrhages in hypertensive patients.

Torres-Gómez E1, Onofre-Castillo JJ2, Santana-Vela IA1, Hernández-Salazar JJ3, García-Concha A4, Martínez-Aparicio JS4, Cuituny-Romero AK1

Abstract INTRODUCTION: cerebral microhemorrhages are deposits of in small blood vessels produced by microangio- pathic processes and represent a risk for the development of ; in magnetic resonance they appear as rounded focal areas with lack of signal.

OBJECTIVE: compare T2 gradient echo and SWAN (suscepti- bility-weighted angiography) magnetic resonance sequences in describing cerebral microhemorrages in hypertensive patients.

MATERIAL AND METHODS: a transverse, descriptive, obser- 1Médico Especialista en Radiología e Imagen. vational study in a group of hypertensive patients admitted to 2Médico Especialista en Radiología e Imagen, Profe- Hospital Christus Muguerza Alta Especialidad de Monterrey and sor Titular y Asesor del Departamento de Radiología. 3Residente de cuarto año de Imagenología Diag- who underwent magnetic resonance. nóstica y Terapéutica. 4Residente de tercer año de Imagenología Diag- nóstica y Terapéutica. RESULTS: fifty-seven patients were included (35 men and 22 women) with mean age of 63 years. Seventeen had microhe- Departamento de Radiología e Imagen del Hospital Christus Muguerza Alta Especialidad, Hidalgo 2525, morrhages detected with susceptibility-weighted angiography Col. Obispado, C.P. 64060, Monterrey, N.L. (SWAN) sequence and 15 with T2 gradient echo sequence, the majority with lobar localization. On comparing the procedures Received: May 10, 2016 one-to-one, the SWAN showed a larger number of microhemor- Accepted: May 14, 2016 rhages. Correspondence Enrique Torres-Gómez CONCLUSION: both sequences are useful to detect the pres- [email protected] ence, localization, and degree of cerebral microhemorrhages, but the susceptibility-weighted angiography sequence proved to be This article must be quoted as Torres-Gómez E, Onofre-Castillo JJ, Santana-Vela IA, more useful in detecting the total number of microhemorrhages. Hernández-Salazar JJ, García-Concha A, Martínez- Aparicio JS et al. Secuencia T2 eco de gradiente KEYWORDS: microhemorrhages; magnetic resonance; high versus secuencia ponderada de susceptibilidad angiológica en la detección de microhemorragias en blood pressure; T2 gradient echo; susceptibility-weighted angi- pacientes hipertensos. Anales de Radiología México ography sequence 2016;15(2):94-102.

94 www.nietoeditores.com.mx Torres-Gómez E, et al. Screening of microhemorrhages in hypertensive patients

INTRODUCTION transformation after an ischemic vascular event and with recurrence of spontaneous intracerebral Cerebral microhemorrhages are hemosiderin bleeds. There are also controversial results that deposits in the brain due to red blood cell leaks suggest that cerebral microhemorrhages involve in small blood vessels without any symptoms. an increased risk for complications Cerebral microbleeds are neuroimaging findings from thrombolytic therapy or the use of platelet defined as small areas with void signal on MRI agents.8 imaging as a consequence of old bleeding foci.1 They are thought to be the result of microangio- Hemosiderin deposits lead to a lack of uniformity pathic processes either from atherosclerosis or in the areas where the lesions are found, leading fibrohyalinosis or from amyloid cerebral angi- to a signal breakdown in MRI (called magnetic opathy. The main contributors of these lesions susceptibility effect) so there is a signal loss or are cardiovascular risk factors, underlying ac- low intensity areas.9 In magnetic resonance im- tive angiopathy disease mechanisms or perhaps aging, cerebral microhemorrhages are defined as changes due to aging. multiple oval or circular foci with a substantial loss of signal in T2 weighted images,10-12 mainly Previous reports have shown that the number of in echo-gradient sequences that have proven a microhemorrhages is associated to the severity of higher sensitivity for detection of old changes in the and with the number compared to conventional spin-echo sequences; of lacunar infarcts, believed to happen as a result such images should always be differentiated from of small vessel disease in the brain.2 Walker et bloodflow voids and calcifications in the brain. al3 found evidence of microhemorrhages with cortical-subcortical distribution, a characteristic Concerning size, the criteria have been incon- feature in 15.5% of patients over 70 years of age. sistent and most of the studies indicate that There is substantial evidence that supports an im- microhemorrhages tend to be smaller than 5 mm portant association between microhemorrhages or, 10 mm as the upper limit.7,10 Traditionally, and hypertensive and atherosclerotic vasculopa- echo-gradient T2 weighted sequence has been thy. These lesions are found in more than 71% used for detection of microhemorrhages; howev- of individuals that present with intracerebral er, Haacke and his group13 presented a resonance hemorrhage and in 20 to 68% of patients that sequence called susceptibility weighted image are admitted with an ischemic event.2,4-6 and called it (SWI) to further improve the effect of T2 weighted sequences using detection and Talking specifically about cerebral hyperten- measurement of and other substances that sive angiopathy, which is caused by intimal alter the magnetic field.14 This sequence provides hyperplasia and hyalinosis in the deep cerebral information of any tissue that has a different sus- arteriolar branches as a result of hypertension, ceptibility in its surrounding structures such as it involves pathophysiologic changes that lead deoxygenated blood, hemosiderin, and to cerebral microhemorrhages, mainly in the calcium. There are multiple neurologic disorders thalamus, basal ganglia, cerebellum and pons.7 in which we can dramatically benefit from this Based on these findings, it is postulated that ce- highly sensitive method to monitor the amount rebral microhemorrhages are an important risk of iron in the brain, either as deoxyhemoglobin, factor for a subsequent intracerebral hematoma. ferritin or hemosiderin. Such sequence has the Also, microhemorrhages have been found to be peculiarity of being able to present the data with associated with a greater risk of hemorrhagic maximal or minimal projection intensity (MIP/

95 Anales de Radiología México 2016 abril;15(2)

MinIP) in a given image, to get a better assess- cooperate and whose images were suboptimal ment of the vascular tracts. for proper evaluation and interpretation.

The General Electric units offer susceptibility The evaluation using the sequences mentioned weighted image (SWI) and call it susceptibility above was done at the Hospital Christus Mu- weighted angiography (SWAN), that helps to guerza Alta Especialidad, with a General Electric clearly outline the small blood vessels, micro- Healthcare MRI scanner, Signa HDx model, 3 hemorrhages and large vascular structures in the tesla, with trained radiologists having over two brain, aside from visualizing iron and calcium years experience, and obtaining the sequences deposits due to the paramagnetic properties according to the standardized protocols (Ta- already mentioned.15 ble 1). Echo-gradient T2 weighted sequences and susceptibility weighted angiography sequences Timely detection of microhemorrhages in hy- (SWAN) were obtained from all the participants, pertensive patients plays an important role in from the skull base to the convexity of both brain the prognosis and probability of ischemic event hemispheres. risk with probable hemorrhagic transformation or a new hemorrhagic vascular event in patients Table 1. MRI sequence parameters with previous events. We must know, detect, characterize and quantify microhemorrhages, so Parameter 2d Echo gradient SWAN 3D 3T the echo-gradient T2 weighted images, as well Repetition time(ms) 650 41.2 as SWAN, play a crucial role in such an assess- Echo time (ms) 15.0 24.7 ment. The objective of this study is to compare Flip angle(grados) 20 15 echo-gradient T2 weighted MRI sequences and Band width (kKz) 20.8 62.5 susceptibility weighted angiography sequence Field of vision (cm) 24 20 (SWAN) in the description of cerebral microhe- Matrix size 256 × 224 320 × 224 morrhages in hypertensive patients. Flow compensation Yes Yes Section thickness 4.0 2.0 MATERIAL AND METHODS (5 mm) Gap No No An observational, descriptive and cross-sectional Number of sections 30 62 hasta 128 study was conducted in hypertensive patients Acquisition time 3:42 3:20 who were admitted to the Hospital Christus Muguerza Alta Especialidad de Monterrey who SWAN: susceptibility-weighted angiography sequence. had an MRI of the brain with a 3 Tesla unit from March, 2014 to September, 1915. Patients di- agnosed with hypertension who would have an Classification of microhemorrhages echo-gradient T2 sequence and a susceptibility weighted angiography sequence (SWAN) were The MRI sequences were reviewed and classified included, regardless of the cause. Patients who based on the presence, location, number and had a history or suspicion of a vascular malfor- grade of microhemorrhages, defining micro- mation, brain tumor on CT or MRI, or history hemorrhages as focal areas of very low signal of previous surgery, or recent trauma were ex- intensity under 10 mm in size. Based on previous cluded; as well as patients with metal objects in studies2,16 the microhemorrhages were classified the brain or skull base and patients who did not into one of the following locations: 1) lobar:

96 Torres-Gómez E, et al. Screening of microhemorrhages in hypertensive patients

involvement of the cortical gray matter and the Microsoft Excel 2012 and the statistical software subcortical white matter; 2) deep: involvement of of the Social Science Statistics web site were used the deep gray matter (basal ganglia and thalamus) for statistical measurements.18 and the deep white matter (corpus callosum); 2) infratentorial: involvement of the cerebellum and RESULTS the bulbo-medullary junction. Fifty-seven patients with a median of 63 years The signal from vessels in the sulci, symmetrical were included (DE ±14), 35 (61%) were males calcifications in the deep gray matter, calci- and 22 (39%) were females. A summary of the fied choroidal plexus, calcified pineal gland as distribution of the main demographic variables well as signal disturbances from the were in the evaluation is shown on Table 2. Out of excluded because they appear like microhemor- the total number of hypertensive patients, 26 had rhages. Finally, the grade of microhemorrhages comorbidities (Table 3), and headaches were the was assessed according to the classification from main reason for consultation in 40% of cases, Lee et al:17 while the rest are basically divided into hemipa- resis, vertigo, amnesia, disorientation, syncope a. Grade I: absence of microhemor- and aphasia in consecutive order of frequency rhages. (Table 4). Cerebral microhemorrages (Figure 1) were found in 15 patients with echo-gradient b. Grade II: mild, 1 or 2 microhemor- T2 sequences and in 17 susceptibility weighted rages. angiography sequence (SWAN) (Table 5).

c. Grade III: moderate, 3-10 microhe- There was not enough evidence to rule out the morrages. null hypothesis concerning the difference in microhemorrhages between the sequences, with d. Grade IV: severe, 10 or more micro- 0.5 McNemar test result. In both sequences, hemorrages. echo-gradient T2 as well as susceptibility weight- ed angiography (SWAN), most of the subjects had Multivariate logistic regression analysis was used microhemorrhages with lobar location; however, to evaluate the statistical significance (c2, K-S) in the SWAN sequence more lobar and deep concerning the main demographic variables, and microhemorrhages were found, infratentorially normal distribution, respectively. Prevalence, the amount was the same. (Table 5). multiplicity and grade of microhemorrhages Table 2. Main population in the study and their association were estimated in both sequences. The statistical to presence of absence of microhemorrhages significance was evaluated using the McNemar non-parametric test for matched proportions. Variables Patients Patients with p Besides, in both sequences the rate of indi- with mi- no microhem- value viduals with microhemorrhages was described crohemor- orrhages rhages (n=40) per location (lobar, deep or infratentorial). For (n=17) those in which both sequences showed micro- Age 68.5 ± 12.59 61.27 ± 14.07 0.918 hemorrhages, the difference (in the number of Male gender 13 (37%) 22 (62%) 0.12 microhemorrhages in one or the other sequence) Female gender 4 (18%) 18 (82%) was analyzed to see if it was statistically sig- Arterial 17 (30%) 40 (70%) 0.08 nificant using the non-parametric Wilcoxon test. hypertension

97 Anales de Radiología México 2016 abril;15(2)

Table 3. Main associated comorbidities in hypertensive patients

Comorbidities Diabetes mellitus Meningioma Hypothyroidism Pituitary adenoma Parkinson Esophageal cancer

n=25 patients 18 (31%) 2 (3.5%) 2 (3.5%) 1 (1.7%) 1 (1.7%) 1 (1.7%)

Table 4. Main cause of consultation in hypertensive patients

Cause of Headache Vertigo Amnesia Hemiparesis Miscellaneous consultation (probable ) (aphasia, diplopia, syncope and disorientation) n=57 patients 23 (40%) 7 (12.2%) 3 (5.2%) 12 (21%) 12 (21%)

Table 5. Prevalence of cerebral microbleeds on MRI

A B Prevalence T2 echo SWAN p value gradient (n=57) (n=57) Microbleeds 15 (26.3) 17 (29.8) <0.05 *M=0.5 A singe 7 (12.2) 6 (10.5) <0.05 *M=0 microbleed Multiple (≥ 2) 8 (14.0) 13 (22.8) <0.05 microbleeds *M=3.2 Location£ Lobar 8 (30.4) 11 (36.9) Deep 9 (34.2) 11 (36.9) Figure 1. MRI (3 Tesla), axial sections with sequences: Infratentorial 4 (15.2) 4 (13.4) A) T2 echo-gradient B) susceptibility weighted angiog- raphy sequence (SWAN), in the same patient. A greater Note: the data are numbers of patients with rates in paren- number of microhemorrhages are seen (arrows) in the thesis. SWAN sequence with mainly a lobar distribution. * McNemar test: if the M value es less than 3.84, there is not enough evidnce to reject H0. £Rates were estimated based on the proportion of participants One to one comparison with microhemorrhage in any sequence (15 for T2 echo gradient, 17 for SWAN) with microbleeding in any location. The rates do not add up to 100% because some patients had In the one to one comparison of sequences microbleedings in more than one location. there were no microhemorrages seen in echo- gradient T2 sequence that could not be shown in the susceptibility weighted angiography microhemorrhages not seen with echo-gradient sequence (SWAN); but in the patients in whom T2 sequence; there was also a difference in the microbleeds were shown (n=15) in both se- classification of the different grades of microhe- quences significantly more microhemorrhages morrhage (Table 6). were documented, visualized with a critical W value of 8, with an average difference of around DISCUSSION 4 and p < 0.05, in the SWAN sequence, with Wilcoxon test. Also, it must be mentioned that The susceptibility weighted angiography in two patients the SWAN sequence revealed sequence (SWAN) revealed more cerebral

98 Torres-Gómez E, et al. Screening of microhemorrhages in hypertensive patients

Table 6. Prevalence of cerebral microbleeds on MRI in both investigated sequences, blood products may be clearly identified as lineal structures, Exact number and T2 echo SWAN p value where, using susceptibility weighted angiogra- grade of gradient (n=17) microbleeds (n=15) phy sequence and with reformatting and MinIp and MIP post processing, they can be better Exact number 62 90 <0.05* depicted. On the other hand, calcifications Gradeα in the brain have a typical location and when n=57 located in the basal ganglia they tend to have a symmetrical distribution. Therefore, we believe I 42 (73.6) 40 (70.0) that microhemorrhages are properly character- II 8 (14.0) 10 (17.5) ized without overestimating their occurrence, III 7 (12.2) 5 (8.77) location, number or grade. IV 0 2 (3.50)

Note. The exact number of microhemorrhages is globally In both sequences we might have overlooked described in patients. some microhemorrhages at the skull base due * Wilcoxon test. to artifact susceptibility, either in echo-gradient αRates were calculated based on the total sample of patients since grade I includes patients with no microbleeds, this rate T2 sequence as well as susceptibility weighted does add up to 100% in both sequences. angiography sequence (SWAN), due to air or SWAN: susceptibility-weighted angiography sequence. bone interphases.

We found microhemorrhages in 17 hypertensive microhemorrhages than the gradient-echo T2 patients, (30%) of the sample, a figure that agrees weighted sequences in two patients, however, with the distribution rate described in previous this difference was not statistically significant in reviews.6 We also found a high frequency com- the assessment concerning prevalence. But in the pared to other morbidities, similar to the study patients in whom the total amount of microhe- conducted by Poels et al24 where they confirm morrhages was assessed, the SWAN sequence an association between hypertension and occur- revealed greater amount of microbleeds. This is rence of microhemorrhages. important because a greater number of micro- hemorrhages suggests that the microangiopathy Because of the increased risk of adverse has progressed and reached a stage where the neurologic events (recurrence of a previous 19-23 blood vessels tend to bleed. spontaneous hemorrhagic event, hemorrhagic transformation of an ischemic vascular event) Before interpreting our data, an explanation is an adequate assessment of microhemorrhages is needed about our method and potential limita- important. A larger sample must be collected in tions in our trial. Both MRI sequences were used future studies, aside from trying to evaluate the in hypertensive patients, resulting in a limited inter and intraobserver reliability. sample of only 57 patients. Also, the classifica- tion or definition of microhemorrhages with MRI CONCLUSION might be thought to be overestimated because other brain structures (deoxygenated blood Both echo-gradient T2 sequence and susceptibil- in small veins, calcification of basal ganglia, ity weighted angiography sequence are useful in choroidal plexus and pineal gland) may appear the localization and determination of the grade like cerebral microhemorrhages. Nevertheless, of cerebral microhemorrhages; however, the

99 Anales de Radiología México 2016 abril;15(2)

susceptibility weighted angiography sequence MR with deoxyhemoglobin as an intrinsic (SWAN) proved to be more efficient to determine contrast agent. 1997;204:272–277. the exact number of microhemorrhages found in 11. Tsushima Y, Aoki J, Endo K. Brain microhemorrhages detected on weighted gradient-echo MR images. Am J each patient, having an impact on the evaluation Neuroradiol 2003;24:88-96 of microangiopathic diseases. 12. Koennecke HC. Cerebral microbleeds on MRI: preva- lence, associations, and potential clinical implications. REFERENCES Neurology 2006;66:165-71. 13. Haacke EM, Xu Y, Cheng YC, et al. Susceptibility weight- ed imaging (SWI). Magn Reson Med 2004;52:612–18. 1. Fazekas F, Kleinert R, Roob G, Kleinert G, Kapeller P, 14. Haacke EM, Cheng NY, House MJ, et al. Imaging iron Schmidt R, Hartung HP. Histopathologic analysis of foci stores in the brain using magnetic resonance imaging. of signal loss on gradient-echo T2*-weighted MR images Magn Reson Imaging 2005;23:1–25. in patients with spontaneous intracerebral hemorrhage: Evidence of microangiopathy-related microbleeds. Am 15. Sowers J. SWAN Designed for excellent visualization of J Neuroradiol. 1999;20:637-642. vasculature and blood products http://www3.gehealth- care.com/en/products/categories/magnetic_resonancei- 2. Jeerakathil T, Wolf PA, Beiser A, Hald JK, Au R, Kase CS, maging/neuro_imaging/swan et al. Cerebral microbleeds: prevalence and associations with cardiovascular risk factors in the Framingham 16. Roob G, Schmidt R, Kapeller P, Lechner A, Hartung Study. Stroke 2004;35:1831-5. HP, Fazekas F MRI evidence of past cerebral micro- bleeds in a healthy elderly population. Neurology 3. Walker DA, Broderick DF, Kotsenas AL, Rubio FA. 1999b;52:991-4 Routine use of gradient-echo MRI to screen for cerebral amyloid angiopathy in elderly patients. AM J Roentgenol 17. Lee SH, Bae HJ, Yoon BW, Kim H, Kim DE, Roh 2004;182:1547-1550. JK. Low concentration of serum total cholesterol is associated with multifocal signal loss lesions on 4. Kinoshita T, Okudera T, Tamura H, Ogawa T, Hatazawa gradient-echo magnetic resonance imaging: analysis J, Assessment of lacunar hemorrhage associated with of risk factors for multifocal signal loss lesions. Stroke. hypertensive stroke by echo-planar gradient-echo T2*- 2002;33:2845–2849. weighted MRI. Stroke. 2002;31:1646-1650. 18. Stangroom J, Social Science Statistics [en línea] 2014. 5. Kato H, Izumiyama M, Izumiyama K, Takahashi A, [fecha de consulta: 23 de octubre del 2014]. Disponible Itooyama Y. Silent cerebral microbleeds on T2*-weight- en: http://www.socscistatistics.com/Default.aspx ed MRI: correlation with stroke subtype, stroke recur- 19. Kidwell CS, Saver JL, Villablanca JP. et al. Magnetic rence, and leukoaraiosis. Stroke. 2002;33:1536-1540. resonance imaging detection of microbleeds before 6. Nighoghossian N, Hermier M, Adeleine P, Blanc-Laserre thrombolysis. Stroke. 2002;33:95-98. K, Derex L, Honnorat J, Philippeau F, Dugor JF, Fro- 20. Leblanc R, Haddad G, Robitaille Y. Cerebral hemor- ment JC, Trouillas P. Old microbleeds are a potential rhage from amyloid angiopathy and coronary throm- risk factor for cerebral bleeding after isquemic stroke: bolysis. 1992;31(3):586-590. a gradient-echo T2*-weighted brain MRI study. Stroke. 21. Chalela JA, Kang DW, Warach S. Multiple cerebral 2002;33:735-742. microbleeds: MRI marker of a diffuse hemorrhage-prone 7. Blitstein MK, Tung GA. MRI of cerebral microhemor- state. J Neuroimagin 2004;14(1):54-57. rhages. Am J Roentgenol 2007;189:720–725. 22. McCarron MO, Nicoli JA. Cerebral amyloid angiopathy 8. Wong KS, Chan YL, Liu JY, Gao S, Lam WW. Asymptom- and thrombolysis-related intracerebral haemorrhage. atic microbleeds as a risk factor for aspirin-associated Lancet Neurol 2004;3(8):484-492. intracerebral hemorrhages. Neurology 2003;60(3):511- 23. Greenberg SM. Cerebral amyloid angiopathy: pros- 513. pects for clinical diagnosis and treatment. Neurology 9. Greenberg SM, Vernooij MW, Cordonnier C, et al. 1998;51(3):690-694. Cerebral microbleeds: a guide to detection and inter- 24. Poels, M.M., Vernooij, M.W., Ikram, M.A., Hofman, pretation. Lancet Neurol 2009;8:165-74. A., Krestin, G.P., van der Lugt, A. et al. Prevalence and 10. Reichenbach JR, Venkatesan R, Schillinger DJ, Kido risk factors of cerebral microbleeds: an update of the DK, Haacke EM. Small vessels in the : Rotterdam scan study. Stroke. 2010;41:S103–S106.

100