Diabetes Care 1

Clinical and MRI Features of Lena M. Thorn,1–3 Sara Shams,4 Daniel Gordin,1–3,5 Ron Liebkind,6 Cerebral Small-Vessel Disease in Carol Forsblom,1–3 Paula Summanen,1–3,7 Stefanie Hagg-Holmberg,¨ 1–3 Type 1 Diabetes Turgut Tatlisumak,6,8,9 Oili Salonen,10 Jukka Putaala,6 Juha Martola,4,10 and 1–3,11 https://doi.org/10.2337/dc18-1302 Per-Henrik Groop, on behalf of the FinnDiane Study Group

OBJECTIVE To assess the prevalence of cerebral small-vessel disease (SVD) in subjects with type 1 diabetes compared with healthy control subjects and to characterize the diabetes-related factors associated with SVD.

RESEARCH DESIGN AND METHODS This substudy was cross-sectional in design and included 191 participants with type 1 diabetes and median age 40.0 years (interquartile range 33.0–45.1) and 30 1Folkhalsan¨ Institute of Genetics, Folkhalsan¨ Re- healthy age- and sex-matched control subjects. All participants underwent search Center, Helsinki, Finland clinical investigation and brain MRIs, assessed for cerebral SVD. 2Abdominal Center Nephrology, University of Helsinki and Helsinki University Hospital, RESULTS Helsinki, Finland 3 Cerebral SVD was more common in participants with type 1 diabetes than in Research Programs Unit, Diabetes and Obesity, University of Helsinki, Helsinki, Finland healthy control subjects: any marker 35 vs. 10% (P = 0.005), cerebral microbleeds 4Department of Radiology, Karolinska University (CMBs) 24 vs. 3.3% (P = 0.008), white matter hyperintensities 17 vs. 6.7% (P = Hospital, and Department of Clinical Neurosci- 0.182), and lacunes 2.1 vs. 0% (P = 1.000). Presence of CMBs was independently ence, Karolinska Institute, Stockholm, Sweden 5 associated with systolic blood pressure (odds ratio 1.03 [95% confidence interval Joslin Diabetes Center, Harvard Medical School, Boston, MA 1.00–1.05], P = 0.035). 6Department of Neurology, Helsinki University DIABETES IN COMMUNICATIONS NOVEL Hospital, Helsinki, Finland CONCLUSIONS 7Department of Ophthalmology, Helsinki Univer- Cerebral SVD, CMBs in particular, is more common in young people with type 1 sity Hospital, Helsinki, Finland 8 diabetes compared with healthy control subjects. Department of Clinical Neuroscience/Neurology, Institute of Neuroscience and Physiology, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden Type 1 diabetes is associated with a fivefold increased risk of stroke (1), with cerebral 9Department of Neurology, Sahlgrenska Univer- small-vessel disease (SVD) as the most common etiology (2). Cerebral SVD in type 1 sity Hospital, Gothenburg, Sweden 10 diabetes, however, remains scarcely investigated and is challenging to study in vivo Department of Radiology, Helsinki University per se owing to the size of affected vasculature (3); instead, MRI signs of SVD are Hospital, Helsinki, Finland 11Department of Diabetes, Central Clinical studied. In this study, we aimed to assess the prevalence of cerebral SVD in subjects School, Monash University, Melbourne, Australia with type 1 diabetes compared with healthy control subjects and to characterize Corresponding author: Per-Henrik Groop, per-henrik diabetes-related variables associated with SVD in stroke-free people with type 1 .groop@helsinki.fi diabetes. Received 16 June 2018 and accepted 29 October 2018 RESEARCH DESIGN AND METHODS L.M.T. and S.S. made equal contributions. J.M. All study participants are part of the Finnish Diabetic Nephropathy (FinnDiane) Study and P.-H.G. made equal contributions. (4). Participants attending the Helsinki University Hospital (HUH) study center were © 2018 by the American Diabetes Association. consecutively recruited and underwent brain MRI as part of their study visit. In 2011– Readers may use this article as long as the work 2017, we studied 191 participants with type 1 diabetes. Inclusion criteria were age 18– is properly cited, the use is educational and not for profit, and the work is not altered. More infor- 50 years and type 1 diabetes onset at age ,40 years. Exclusion criteria were presence mation is available at http://www.diabetesjournals of end-stage renal disease, clinical signs of cerebrovascular disease, or contraindications .org/content/license. Diabetes Care Publish Ahead of Print, published online December 14, 2018 2 Cerebral Small-Vessel Disease in Diabetes Diabetes Care

for MRI. We studied 30 healthy age- and Gradient Echo), and magnetic resonance presented as medians (interquartile sex-matched control subjects with mean angiography time-of-flight. An experi- range). We analyzed categorical variables fasting glucose 4.4 6 0.4 mmol/L. enced neuroradiologist (J.M.) assessed with x2 test or Fisher exact test when The study protocol was approved by the the images, and repeated the assessment appropriate. We performed logistic re- ethics committee of the HUH, and the study three times per participant, for markers gression to determine independent asso- was carried out in accordance with the of SVD (5,6): presence of cerebral micro- ciations with cerebral SVD and present Declaration of Helsinki. Each participant bleeds (CMBs), cortical superficial side- results as odds ratio (OR) with 95% CIs. signed a written informed consent form. rosis, white matter hyperintensities Statistical significance was P , 0.05. All All participants underwent a clinical (WMHs) (Fazekas scale used, with cate- analyses were performed with IBM SPSS study visit (4). Brain MRI was performed gory $1 considered to be significant Statistics, version 22.0, software (IBM, with a 3.0 Tesla scanner (Achieva; Philips, burden), or lacunes. The neuroradiolo- Armonk, NY). Best, the Netherlands) at the Helsinki gist was blinded to clinical data but not Medical Imaging Center, HUH, and in- to whether the participant had diabetes. RESULTS cluded T1, T2, fluid-attenuated inversion We analyzed parametric variables with Clinical characteristics and MRI findings recovery, susceptibility-weighted imaging, t tests with results presented as means 6 of the 191 participants with type 1 di- T2*, diffusion-weighted imaging, T1 SD and nonparametric variables with abetes and the 30 age- and sex-matched MPRAGE (Magnetization-Prepared RApid Mann-Whitney U tests with results control subjects appear in Table 1. Of

Table 1—Clinical characteristics and MRI findings in participants with type 1 diabetes and control subjects matched for age and sex Subjects with type 1 diabetes Control subjects P N 191 30 Clinical characteristics Female sex 101 (53) 17 (57) 0.699 Age, years 40.0 (33.0–45.1) 38.4 (31.4–43.2) 0.443 Diabetes duration, years 21.7 (18.3–30.9) dd BMI, kg/m2 26.7 6 4.2 24.5 6 3.2 0.002 Systolic blood pressure, mmHg 130 6 14 121 6 11 0.001 Diastolic blood pressure, mmHg 77 (71–82) 76 (74–85) 0.387

HbA1c, % (mmol/mol) 8.2 6 1.1 (66 6 12) 5.1 6 0.2 (33 6 2) ,0.001 Creatinine, mmol/L 68 (61–79) 74 (68–81) 0.067 Total cholesterol, mmol/L 4.4 (4.0–4.9) 4.6 (4.2–5.4) 0.178 LDL cholesterol, mmol/L 2.4 (2.0–2.9) 2.6 (2.3–3.3) 0.017 HDL cholesterol, mmol/L 1.50 (1.25–1.80) 1.46 (1.26–1.68) 0.362 Triglycerides, mmol/L 0.90 (0.68–1.38) 0.84 (0.69–1.26) 0.398 Antihypertensive medication 68 (36) 0 ,0.001 Statin therapy 42 (22) 0 0.002 Aspirin therapy 15 (7.9) 0 0.232 Albuminuria 30 (16) 0 0.018 Retinal photocoagulation 42 (22) 0 0.002 Coronary heart disease 1 (0.5) 0 1.000 Current smoking 15 (7.9) 5 (17) 0.118 Other autoimmune disease 63 (33) 6 (20) 0.149 MRI findings Cerebral SVD 67 (35) 3 (10) 0.005 CMBs 45 (24) 1 (3.3) 0.008 Number of CMBs 1.000 1, N (% of those with CMBs) 27 (60) 1 (100) 2, N (% of those with CMBs) 6 (13) 0 $3, N (% of those with CMBs) 12 (27) 0 Topography of CMBs Strictly lobar, N (% of those with CMBs) 38 (84) 1 (50) 0.315 Strictly deep or infratentorial, N (% of those with CMBs) 3 (6.7) 1 (50) 0.165 Mixed, N (% of those with CMBs) 4 (8.9) 0 1.000 Cortical superficial siderosis 0 0 d Any WMH 44 (23) 2 (6.7) 0.051 Fazekas category 1 33 (17) 2 (6.7) 0.182 Lacunes 4 (2.1) 0 1.000 Stenosis of carotid arteries 2 (1.0) 0 1.000 Incidental findings 20 (10) 4 (13) 0.751 Unremarkable MRI scan 107 (56) 24 (80) 0.013 Data are n (%), median (interquartile range), or mean 6 SD unless otherwise indicated. care.diabetesjournals.org Thorn and Associates 3

the 67 participants with type 1 diabetes including 67 participants with type 1 di- results, with some reporting more WMHs and SVD, 32 (48%) had only CMBs, abetes and 33 control subjects, there was in type 1 diabetes (14) and others not 20 (30%) only WMHs, 11 (16%) both no difference in WMH prevalence but a (9,15). Most WMHs in our study were CMBs and WMHs, 2 (3%) both CMBs higher prevalence of CMBs in partici- classified as Fazekas category 1, and and lacunes, and 2 (3%) both WMHs and pants with type 1 diabetes and retinop- only four participants had lacunesdall lacunes (Supplementary Fig. 1). athy compared with control subjects (9). of them with type 1 diabetes. Participants with type 1 diabetes and In the current study, CMBs were not The strengths of our study include CMBs more often had presence of albu- associated with retinopathy but were, standardized enrollment of participants, minuria (27 vs. 12%, P = 0.021), were on on the other hand, associated with al- age- and sex-matched healthy con- antihypertensive medication (49 vs. 32%, buminuria, a strong marker of general- trol subjects, and standardized imaging P = 0.033), and had higher systolic blood ized microvascular disease. In addition, and assessment. Although our study is pressure (135 6 17 vs. 129 6 13 mmHg, CMBs were independently associated the largest on type 1 diabetes and SVD P = 0.009) (Supplementary Table 1). Only with higher systolic blood pressure. Hy- prevalence to date, we did not have systolic blood pressure was indepen- pertension has also been associated with sufficient power to detect minor differ- dently associated with CMBs: OR for CMBs in the general population (10), but ences between the groups. The cross- 1-mmHg increment 1.03 (95% CI 1.00– other studies show conflicting results sectional study setting also limits the 1.05); P = 0.035. Eight participants (11). In type 1 diabetes, albuminuria interpretation of causal relationships. had .10 CMBs (range 10–105). These and systolic blood pressure indepen- We conclude that cerebral SVD is more eight participants were older than 38 dently increase the risk for both ischemic common in type 1 diabetes than healthy years; 50% had albuminuria, 75% history and hemorrhagic stroke (12). control subjects. Future studies will focus of retinal photocoagulation, and 88% hy- Cerebral amyloid angiopathy and hy- on longitudinal development of SVD in pertension; 25% were on aspirin therapy; pertensive vasculopathy are the two type 1 diabetes and the associations with and none used anticoagulant therapy. most common pathogenetic processes brain health and cognition. Participants with WMHs (Fazekas $1) underlying cerebral SVD. Both diseases were significantly older (median 44.9 cause microaneurysms, vessel disrup- years [interquartile range 40.8–47.6] tion, microthrombosis, and arteriolo- Acknowledgments. We acknowledge the vs. 38.6 years (32.5–44.2); P , 0.001) sclerosis, leading to permeable and skilled technical assistance of Anna Sandelin, and had a higher systolic blood pressure disrupted microvasculature (3). In the Jaana Tuomikangas, and Mira Korolainen. We (mean 6 SD 137 6 15 vs. 129 6 current study, CMBs were mainly ob- also gratefully thank Pentti Pol¨ onen,¨ Depart- ment of Radiology, HUH, for performing the 14 mmHg, P = 0.005) (Supplementary served in the lobar brain regions, which MRI scans. We are indebted to Markku Kaste, Table 2). Only age was independently has been associated with cerebral amy- Department of Neurology, HUH, for help in the associated with WMHs (OR for age per loid angiopathy, a condition generally initiation of the study. 1-year increment 1.11 [95% CI 1.04– affecting the elderly, whereas CMBs in Funding. The FinnDiane Study was supported ¨ 1.19]; P = 0.003). the deeper parts associate with hyper- by grants from the Folkhalsan Research Founda- tion, Academy of Finland, Wilhelm and Else tensive vasculopathy (3). It is, however, Stockmann Foundation, Liv och Halsa¨ Society, unlikely that the lobar predominance of Novo Nordisk Foundation, and Paivikki¨ and Sakari CONCLUSIONS CMBs in our younger participants would Sohlberg Foundation and by HUH state research Cerebral SVD is more common in partic- indicate cerebral amyloid angiopathy. funding (erityisvaltionosuus). L.M.T. was sup- ported by personal grants from the Medical ipants with type 1 diabetes than in The majority of those with several Society of Finland and the Dorothea Olivia, healthy control subjects. CMBs especially CMBs had hypertension and presence Karl Walter and Jarl Walter Perklen´ Foundation. are more prevalent and are indepen- of microvascular diabetes complications, D.G. was supported by the Biomedicum Helsinki dently associated with hypertension. indicating a more generalized vasculop- Foundation, the Finnish Medical Foundation, and Our results indicate that cerebral SVD athy, although CMBs were not associated the Swedish Cultural Foundation in Finland. P.S was supported by Silmas¨ a¨ati¨ o¨ Foundation. J.P. starts early in type 1 diabetes but is not with other vascular risk factors, such as was supported by Diabetes Wellness Finland explained solely by diabetes-related vas- diabetes duration, BMI, glycemic control, and the Diabetes Research Foundation. cular risk factors or the generalized mi- or smoking. Lobar CMBs mostly lie on None of funding bodies had any role in the crovascular disease that takes place in the proximal course of medullary end study design; collection, analysis, or interpre- diabetes (7). arteries supplying the white matter of tation of data; writing of the manuscript; or the decision to submit the manuscript for There are only small-scale studies on the brain. We hypothesize that CMBs publication. cerebral SVD, especially CMBs, in type 1 represent pathology similar to that ob- Duality of Interest. P.S. has received lecture diabetes. Compared with the current served in diabetic retinopathy, e.g., rup- honoraria from Bayer and Santen. T.T. is an study, one study with similar diabetes ture of a microaneurysm (13). Chronic advisory board member of Boehringer Ingel- fi characteristics (i.e., diabetes duration, inflammation or some factor associated heim, Bayer, P zer, and Lumosa Therapeutics and has received speaker honoraria from the glycemic control, and blood pressure solely with type 1 diabetes itself, e.g., University of Donau (Austria) and the Finnish levels) as in the current study, but lacking autoimmunity, could also contribute to Neurological Association. P-H.G. has received a control population, showed a higher the increase in CMBs observed in our lecture honoraria from AstraZeneca, Boehringer prevalence of WMHs, with more than half study. Ingelheim, Eli Lilly, Elo Water, Genzyme, Med- scape, Merck Sharp & Dohme (MSD), Novartis, of the participants affected, but similar We showed a trend toward more Novo Nordisk, and Sanofi and is an advisory board prevalence of lacunes and lower preva- WMHs in type 1 diabetes. Previous stud- member of AbbVie, Boehringer Ingelheim, Eli lence of CMBs (8). In another study, ies in type 1 diabetes show conflicting Lilly, Janssen, Medscape, MSD, Novartis, Novo 4 Cerebral Small-Vessel Disease in Diabetes Diabetes Care

Nordisk, and Sanofi. No other potential conflicts 4. Thorn LM, Forsblom C, Fagerudd J, et al.; diabetes is associated with cerebral microbleeds, of interest relevant to this article were reported. FinnDiane Study Group. Metabolic syndrome in which is part of generalized microangiopathy. Author Contributions. L.M.T., S.S., D.G., R.L., type 1 diabetes: association with diabetic ne- Diabetes Care 2014;37:1165–1168 C.F., P.S., S.H.-H., T.T., O.S., J.P., J.M., and P.-H.G. phropathy and glycemic control (the FinnDiane 10. Vernooij MW, van der Lugt A, Ikram MA, contributed to the study design and acquisition study). Diabetes Care 2005;28:2019–2024 et al. Prevalence and risk factors of cerebral of data, as well as the interpretation of data. 5. Fazekas F, Kleinert R, Roob G, et al. Histopath- microbleeds: the Rotterdam Scan Study. Neu- L.M.T. and S.S. had the main responsibility ologic analysis of foci of signal loss on gradient- rology 2008;70:1208–1214 for analyzing data and writing the first draft echo T2*-weighted MR images in patients with 11. Jeerakathil T, Wolf PA, Beiser A, et al. Ce- of the paper. D.G., R.L., C.F., P.S., S.H.-H., T.T., spontaneous intracerebral hemorrhage: evidence rebral microbleeds: prevalence and associations O.S., J.P., J.M., and P.-H.G. critically revised of microangiopathy-related microbleeds. AJNR with cardiovascular risk factors in the Framing- the manuscript. P.-H.G. is the guarantor of Am J Neuroradiol 1999;20:637–642 ham Study. Stroke 2004;35:1831–1835 this work and, as such, had full access to all 6. Wardlaw JM, Smith EE, Biessels GJ, et al.; 12. Hagg¨ S, Thorn LM, Forsblom CM, et al.; the data in the study and takes responsibility STandards for ReportIng Vascular changes on FinnDiane Study Group. Different risk factor for the integrity of the data and the accuracy of nEuroimaging (STRIVE v1). Neuroimaging stan- profiles for ischemic and hemorrhagic stroke the data analysis. dards for research into small vessel disease and in type 1 diabetes mellitus. Stroke 2014;45: its contribution to ageing and neurodegenera- 2558–2562 References tion. Lancet Neurol 2013;12:822–838 13. Gardiner TA, Archer DB, Curtis TM, Stitt 1. Janghorbani M, Hu FB, Willett WC, et al. 7. Deckert T, Feldt-Rasmussen B, Borch-Johnsen AW. Arteriolar involvement in the microvascular Prospective study of type 1 and type 2 diabetes K, Jensen T, Kofoed-Enevoldsen A. Albuminuria lesions of diabetic retinopathy: implications for and risk of stroke subtypes: the Nurses’ Health reflects widespread vascular damage. The Steno pathogenesis. Microcirculation 2007;14:25–38 Study. Diabetes Care 2007;30:1730–1735 hypothesis. Diabetologia 1989;32:219–226 14. Nunley KA, Ryan CM, Orchard TJ, et al. 2. Putaala J, Liebkind R, Gordin D, et al. Diabetes 8. van Elderen SG, Brandts A, Westenberg JJ, White matter hyperintensities in middle-aged mellitus and ischemic stroke in the young: clin- et al. Aortic stiffness is associated with cardiac adults with childhood-onset type 1 diabetes. ical features and long-term prognosis. Neurol- function and cerebral small vessel disease in Neurology 2015;84:2062–2069 ogy 2011;76:1831–1837 patients with type 1 diabetes mellitus: assess- 15. Brands AM, Kessels RP, Hoogma RP, et al. 3. Pantoni L. Cerebral small vessel disease: from ment by magnetic resonance imaging. Eur Radiol Cognitive performance, psychological well- pathogenesis and clinical characteristics to 2010;20:1132–1138 being, and brain magnetic resonance imaging therapeutic challenges. Lancet Neurol 2010;9: 9. Woerdeman J, van Duinkerken E, Wattjes in older patients with type 1 diabetes. Diabetes 689–701 MP, et al. Proliferative retinopathy in type 1 2006;55:1800–1806