- Email: [email protected]
Lack of Association between Periodic Limb Movements during Sleep and Neuroimaging Signatures of Cerebral Small Vessel Disease in Stroke-Free Community-Dwelling Older Adults. The Atahualpa Project
ARTICLE IN PRESS
Lack of Association between Periodic Limb Movements during Sleep and Neuroimaging Signatures of Cerebral Small Vessel Disease in Stroke-Free Community-Dwelling Older Adults. The Atahualpa Project Oscar H. Del Brutto, MD,* Robertino M. Mera, MD, PhD,† Victor J. Del Brutto, MD,‡ and Pablo R. Castillo, MD§
Background: Evidence of the relationship between periodic limb movements during sleep (PLMS) and cerebral small vessel disease (cSVD) is limited and inconsistent. Here, we aimed to assess the independent association between PLMS and the different neuroimaging signatures of cSVD. Methods: Atahualpa residents aged more than or equal to 60 years enrolled in the Atahualpa Project undergoing polysomnography and MRI with time intervals less than or equal to 6 months were included. MRI readings focused on white matter hyperintensities (WMH) of presumed vascular origin, deep cerebral microbleeds (CMB), silent lacunar infarcts (LI), and more than 10 enlarged basal ganglia-perivascular spaces (BG-PVS). Data from single-night polysomnograms were interpreted according to recommendations of the American Academy of Sleep Medicine. Associations between the PLMS index and neuroimaging signatures of cSVD (as dependent variables) were assessed by means of logistic regression models, adjusted for relevant confounders. Results: A total of 146 individuals (mean age: 71.4 § 7.5 years; 64% women) were included. A PLMS index more than or equal to 15 per hour were noted in 48 (33%) participants. Moderate-to-severe WMH were present in 33 individuals (23%), deep CMB in 9 (6%), silent LI in 16 (11%), and more than 10 BG-PVS in 44 (30%). In univariate analyses, silent LI (P = .035) and the presence of more than 10 enlarged BG-PVS (P = .034) were significantly higher among participants with a PLMS index more than or equal to 15 per hour. However, fully-adjusted multivariate models showed no significant association between PLMS index more than or equal to 15 per hour and any of the neuroimaging signatures of cSVD. Conclusions: This study shows no independent association between the PLMS index and neuroimaging signatures of cSVD in stroke-free community-dwelling older adults. Key Words: Periodic limb movements during sleep—cerebral small vessel disease—white matter hyperintensities—cerebral microbleeds—silent lacunar infarcts—enlarged basal ganglia-perivascular spaces © 2019 Elsevier Inc. All rights reserved.
From the *School of Medicine, Universidad Espíritu Santo Ecuador, Samborond on, Ecuador; †Department of Epidemiology, Gilead Sciences, Inc, Foster City, CaliforniaA; ‡Department of Neurology, Miller School of Medicine, University of Miami, Miami, Florida; and §Sleep Disorders Center, Mayo Clinic College of Medicine, Jacksonville, Florida. Received June 10, 2019; revision received October 20, 2019; accepted October 22, 2019. Financial Disclosure: This study was supported by Universidad Espíritu Santo Ecuador. Address correspondence to Oscar H. Del Brutto, MD, School of Medicine, Universidad Espíritu Santo Ecuador, Air Center 3542, PO Box 522970, Miami, FL 33152-2970. E-mail: [email protected]. 1052-3057/$ - see front matter © 2019 Elsevier Inc. All rights reserved. https://doi.org/10.1016/j.jstrokecerebrovasdis.2019.104497
Journal of Stroke and Cerebrovascular Diseases, Vol. &&, No. && (&&), 2019: 104497
1
ARTICLE IN PRESS O.H. DEL BRUTTO ET AL.
2
Introduction Information on the relationship between periodic limb movements during sleep (PLMS) and neuroimaging signatures of cerebral small vessel disease (cSVD) is scarce and inconsistent. Indeed, a PubMED search using relevant keywords disclosed only 3 studies, 2 of them conducted in patients with stroke or transient ischemic attacks1,2 and the other in subjects with sleep-related symptoms referred to a specialized center.3 In addition, such studies included small numbers of patients, and might not be representative of the population at large. Current knowledge does not allow to conclude whether PLMS are associated or not with cSVD, and there is certainly no robust evidence on the direction of this relationship. The finding of a relationship between PLMS and neuroimaging signatures of cSVD would open new venues of research that may help to elucidate pathogenetic mechanisms involved in the development and severity of cSVD. By means of data available through the Atahualpa Project, we aimed to assess whether independent associations between the PLMS index and the different neuroimaging signatures of cSVD exist.
Methods Study Population Atahualpa residents share important characteristics for the study of potential relationships between sleep-related disorders and cSVD, including ethnicity (95% of the populations belong to the Ecuadorian native ethnic group), exposure to 12 daily hours of sunlight all over the year, hot and dry weather, virtually no shift working, limited nighttime light pollution, and socioeconomic status (almost all men belong to the blue-collar class and most women are homemakers).4 The Institutional Review Board of Hospital-Clínica Kennedy, Guayaquil, Ecuador (FWA 00006867) approved the study protocol and a comprehensive informed consent form that individuals must sign before enrollment.
Neuroimaging Protocol MRIs were performed with a Philips Intera 1.5T (Philips Medical Systems, Eindhoven, the Netherlands) at HospitalClínica Kennedy, Guayaquil. Exams included 2-dimensional multislice turbo spin echo T1-weighted, fluid attenuated inversi on recovery (FLAIR), T2-weighted, and gradient-echo sequences in the axial plane, as well as a FLAIR sequence oriented in the sagittal plane; slice thickness was 5 mm with 1 mm gap between slices. This protocol meets the standards proposed by Wardlaw et al5 for research on cSVD. Interest focused on: (1) white matter hyperintensities (WMH) of presumed vascular origin, defined as lesions appearing hyperintense on T2-weighted images that remained bright on FLAIR (without cavitation) and graded according to the modified Fazekas scale,6 (2) deep cerebral microbleeds (CMB),
identified and rated according to the microbleed anatomical rating scale,7 (3) silent lacunar infarcts (LI), defined as fluidfilled cavities measuring 3-15 mm located in the territory of a perforating arteriole,5 and (4) enlarged basal ganglia-perivascular spaces (BG-PVS), defined as less than 3 mm structures of CSF intensity—assessed on the T2-weighted sequence— that followed the orientation of perforating arteries, and rated as abnormal if more than 10 of these lesions were present in a single slice in 1 side of the brain (we used the BG slice with the highest number in 1 side).8 All MRIs were independently read by 2 raters blinded to clinical information (O.H.D. and V. J.D.). Kappa coefficients for interrater agreement were .90 for WMH, .76 for deep CMB, .90 for LI, and .83 for the presence of more than 10 enlarged BG-PVS; discrepancies were resolved by consensus.
Polysomnography (PSG) Diagnostic single-night PSGs were performed at the sleep unit of the Atahualpa Project Community Center. Exams were performed with the use of an Embletta X100 Comprehensive Portable PSG System (Embla Systems, Inc; Thornton, CO). A board-certified sleep medicine neurologist (P.R. C.), blinded to other information, reviewed raw data and interpreted all exams. PSGs were scored based upon recommended by the American Academy of Sleep Medicine scoring guidelines.9 All of the raw data were reviewed in 30 second epochs. Sleep efficiency, total time in bed, total sleep time, sleep architecture (percentages of time spent in N1, N2, N3, and rapid eye movements (REM) sleep), mean O2 saturation, and the apnea/hypopnea index (AHI) were used for objective assessment of PSG-derived parameters. In addition, the PLMS index was defined by the number of periodic leg movement events during sleep that meet PLMS criteria divided by the number of hours of sleep with leg movement recording. We used a nasal pressure transducer to exclude upper airway resistance. Limb movements were not scored within .5 seconds of a respiratory event.
Clinical Covariables Demographics and cardiovascular risk factors of interest (the body mass index, blood pressure, fasting glucose, and total cholesterol blood levels) were selected as confounding variables. These were assessed through interviews and procedures previously described in the Atahualpa Project.10 To exclude patients with an overt stroke and with the restless legs syndrome, rural doctors screened all participants with the use of validated field instruments, and then, certified neurologists confirmed the diagnosis as previously reported.11,12
Statistical Analyses Data analyses are carried out by using STATA version 15 (College Station, TX). In univariate analyses, continuous variables were compared by linear models and
ARTICLE IN PRESS ASSOCIATION BETWEEN PLMS AND NEUROIMAGING SIGNATURES OF CSVD
categorical variables by x2 or Fisher exact test as appropriate. Multivariate logistic regression models were fitted to assess the independent association between the PLMS index and the aforementioned neuroimaging signatures of cSVD (as dependent variables), after adjusting for demographics, cardiovascular risk factors, and other PSGderived information.
Results A total of 188 community-dwellers aged more than or equal to 60 years enrolled in the Atahualpa Project underwent PSG and MRI with time intervals of less than or equal to 6 months, and were eligible for this study. Of these, 42 were excluded (27 had an overt stroke, seven had a diagnosis of restless legs syndrome, and 8 presented technical difficulties precluding proper PSG interpretation). The mean age of the 146 included subjects was 71.4 § 7.5 years (median age: 71 years, age range: 60-95 years) and 94 (64%) were women. A body mass index more than or equal to 30 kg/m2 was noticed in 35 (24%) persons, blood pressure more than or equal to 140/90 mm Hg in 73 (50%), fasting glucose more than or equal to 126 mg/dL in 45 (31%), and total cholesterol levels more than or equal to 240 mg/dL in 27 (18%). Mean values of PSG-derived parameters included: total time in bed 480.9 § 30.6 minutes, total sleep time 365.6 § 64.2 minutes, mean O2 saturation 94.7 § 5.1%, sleep efficiency 75.6 §
3
11.7%, and AHI 11.7 § 12.6 (68% individuals had an AHI > 5 per hour). Mean percentages of time in the different sleep stages were: N1 sleep 4.3 § 4.3%, N2 sleep 67.7 § 14.4%, N3 sleep 12 § 8.8%, and REM sleep 15.9 § 10.8%. A PLMS index more than or equal to 15 per hour was recorded in 48 (33%) individuals. Moderate-to-severe WMH were noticed in 33 (23%) individuals, deep CMB in nine (6%), silent LI in 16 (11%), and more than 10 enlarged BG-PVS in 44 (30%). Table 1 shows characteristics of participants across categories of the PLMS index. In univariate analyses, increasing age (P < .001) and having blood pressure levels more than or equal to 140/90 mm Hg (P = .014) were significantly associated with a high PLMS index. Likewise, percentage of time spent in N2 sleep was lower, but that of time spent in REM sleep was higher among participants with a PLMS index more than or equal to 15 per hour. There were no significant differences in the other PSGderived parameters. Also in univariate analyses, silent LI (P = .035) and the presence of more than 10 enlarged BGPVS (P = .034) were significantly higher among participants with a PLMS index more than or equal to 15 per hour, but no differences were found in subjects with moderate-to-severe WMH or deep CMB across groups. Multivariate logistic regression models showed no significant association between a high PLMS index and any of the neuroimaging signatures of cSVD, after adjusting (in separate models) for clinical covariables, and for
Table 1. Characteristics of study participants across categories of the periodic limb movement index per hour (univariate analysis) PLM index <15 per hour (n = 98)
PLM index 15 per hour (n = 48)
P value
71.4 § 7.5 94 (64) 35 (24) 73 (50) 45 (31) 27 (18)
69.8 § 6.5 63 (64) 25 (26) 42 (43) 28 (29) 18 (18)
74.8 § 8.4 31 (65) 10 (21) 31 (65) 17 (35) 9 (19)
<.001* .972 .534 .014* .400 .955
480.9 § 30.6 365.6 § 64.2 94.7 § 5.1 4.3 § 4.3 67.7 § 14.4 12 § 8.8 15.9 § 10.8 75.6 § 11.7 100 (68)
479.8 § 31.3 363.9 § 65.5 95.1 § 1.7 4.1 § 4.2 69.6 § 13.8 12.2 § 8.8 14.1 § 9.8 75.2 § 12.5 68 (69)
483.3 § 28.9 369.2 § 61.3 94.1 § 8.5 4.9 § 4.5 63.8 § 14.7 11.5 § 8.6 19.5 § 11.9 76.4 § 11.8 32 (67)
.514 .640 .263 .293 .021* .650 .004* .580 .888
33 (23) 16 (11) 9 (6) 44 (30)
19 (19) 7 (7) 5 (5) 24 (24)
14 (29) 9 (19) 4 (8) 20 (42)
.184 .035* .477 .034*
Total series (n = 146) Clinical covariables Age in years, mean § SD Women, n (%) Body mass index 30 Kg/m2, n (%) Blood pressure 140/90 mmHg, n (%) Fasting glucose 126 mg/dL, n (%) Total cholesterol 240 mg/dL, n (%) Polysomnography-derived information Total time in bed in minutes, mean § SD Total sleep time in minutes, mean § SD O2 saturation, mean § SD N1 sleep, %, mean § SD N2 sleep, %, mean § SD N3 sleep, %, mean § SD REM sleep, %, mean § SD Sleep efficiency, %, mean § SD Apnea hypopnea index >5 per hour, n (%) Dependent variables Moderate-to-severe WMH, n (%) Silent lacunar infarctions, n (%) Deep cerebral microbleeds, n (%) >10 enlarged BG-PVS, n (%)
Abbreviations: BG-PVS, basal ganglia-perivascular spaces; PLM, periodic limb movement; WMH: white matter hyperintensities. *Statistically significant result.
ARTICLE IN PRESS O.H. DEL BRUTTO ET AL.
4
Table 2. Adjusted logistic regression models showing lack of association between the periodic limb movement index and the different neuroimaging signatures of cerebral small vessel disease (as dependent variables), after taking into account the effect of clinical confounders (upper panel) and clinical confounders plus PSG-derived information (lower panel) Model adjusted for clinical confounders, using the PLM index as the independent variable
Odds ratio
95% CI
P value
Moderate-to-severe white matter hyperintensities Deep cerebral microbleeds Silent lacunar infarcts >10 enlarged basal ganglia-perivascular spaces
1.31 1.74 2.08 1.36
.50-3.42 .38-7.88 .62-7.00 .56-3.33
.581 .474 .239 .498
Model adjusted for clinical confounders plus PSG-derived information, using the PLM index as the independent variable
Odds ratio
95% CI
P value
Moderate-to-severe white matter hyperintensities Deep cerebral microbleeds Silent lacunar infarcts >10 enlarged basal ganglia-perivascular spaces
1.61 4.61 2.67 1.29
.57-4.58 .58-36.8 .58-12.4 .44-3.79
.372 .148 .210 .648
clinical covariables plus PSG-derived parameters (Table 2). Some covariables remained significant in the fullyadjusted models, including increased age (P = .002), high total cholesterol blood levels (P = .034) and an AHI more than 5 per hour (P = .036) in the model using WMH as the dependent variable, and increased age (P < .001) and high total cholesterol blood levels (P = .047) in the model using more than 10 enlarged BG-PVS as the dependent variable. There were no other covariables remaining significant in the models using silent LI or deep CMB as dependent variables.
Discussion The increasingly recognized evidence of an association between PLMS and cerebrovascular events13,14 motivated the assessment of a potential association between the PLMS index and neuroimaging signatures of cSVD. However, studies reported inconsistent results, which are likely related to differences in study designs, inclusion of selected (and biased) populations, and small samples sizes. The present study is the largest series reported to date (n = 146) and, more importantly, is the first in systematically assessing the possible association between the PLMS index and all the neuroimaging signatures of cSVD in an unbiased sample of community-dwelling older adults with a median age of 71 years, when both PLMS and neuroimaging signatures of cSVD are common. The first study—published by Boulos et al1—included 30 consecutive patients (mean age: 63.7 § 13.5 years) presenting within the first 2 weeks of a minor stroke or a transient ischemic attack. In such study, patients with a PLMS index more than or equal to 5 per hour had a greater burden of WMH (in cm3) than those with a PLMS index less than 5 per hour (7.8 § 3.6 versus 4.2 § 3.8; P = .012), after adjusting for cardiovascular risk factors and PSG-derived
parameters. The study did not assess the relationship between PLMS and the other neuroimaging signatures of cSVD. Thereafter, Manconi et al2 could not replicate those previous findings after studying 68 patients (mean age: 59.8 years) selected from a larger cohort study; participants were stratified in 2 arms (PLMS index <5 and 5 per hour), and the authors found no differences in WMH volumes across groups. Likewise, other neuroimaging signatures of cSVD were not assessed in this second study. The most recent study, conducted by Kang et al,3 enrolled 60 stroke-free subjects (mean age: 61.7 § 14.4 years) visiting a clinic for sleep-related symptoms. Those individuals were stratified in 2 arms, including 31 with a PLMS index more than or equal to 15 per hours and 29 with a PLMS index of less than 5 per hour, and compared the prevalence of the 4 neuroimaging signatures of cSVD across groups, after adjusting for several confounders. The authors found significant associations between the PLMS index and WMH and enlarged perivascular spaces, but not with silent LI or CMB. There was no explanation for the exclusion of subjects with a PLMS index between 5 and 14.9 per hour. This gap between arms might have influenced the positive associations found in the study. Despite the clear cutoff established for defining an abnormal PLMS index in adults (15 per hour),15 there is still controversy regarding which cutoff should better be used to find an association between a high PLMS index and neuroimaging signatures of cSVD, we carried out a poststudy analysis to assess the above-mentioned associations using a PLMS index cutoff of more than or equal to 5 per hour to define high PLMS. In these models, a high PLMS index was not associated with moderate-to-severe WMH, the presence of deep CMB or more than 10 enlarged BG-PVS, after adjusting for clinical covariates or in a fully-adjusted model including both clinical covariates and PSG-derived information. The only exception
ARTICLE IN PRESS ASSOCIATION BETWEEN PLMS AND NEUROIMAGING SIGNATURES OF CSVD
was the finding of a significant association between a high PLMS index and silent LI in the fully-adjusted model (OR: 9.07; 95% C.I.: 1.08-76.10; P = .042). However, this finding does not necessarily means an association between a PLMS index more than or equal to 5 per hour and cSVD for 2 main reasons: (1) the PLMS index was not associated with the other neuroimaging signatures of cSVD, and (2) many silent LI are not related to cSVD but to other pathogenetic mechanisms such as cardiogenic or artery-toartery cerebral embolisms.16-18 As previously noted, there are no convincing arguments supporting the association between PLMS and cSVD or the direction of this relationship. In the nonconsented scenario that PLMS are associated with cSVD, it is possible that PLMS may predispose to the severity of WMH because of fluctuations in nocturnal blood pressure and heart rate, oxidative stress, inflammation, and hypoxia. In the event of a reverse causation phenomenon, that is, WMH causing PLMS, the disruption of neural networks associated with WMH would be in the path for PLMS occurrence. Those proposed mechanisms would only explain a possible association between PLMS and WMH, but there are no clues on the pathogenesis of the association between PLMS and the other signatures of cSVD. In the present study, the significant association between the PLMS index and silent LI and more than 10 enlarged BG-PVS that we found in univariate analysis disappeared in multivariate regression models, probably by the effect of confounders. Of interest, the odds ratio for some of these associations was high in multivariate models, but significances were probably tempered by the high range of 95% CI due to the sample size. Despite the strengths of the present study, such as the unbiased inclusion of participants and the systematic evaluation of PSG-derived parameters and neuroimaging signatures of cSVD by means of internationally accepted standards,5,9 there are potential limitations that include the cross-sectional design and, more importantly, the well-recognized night-to-night variability of PLMS, which was not possible to estimate by the use of a single-night diagnostic PSG. Further studies, preferably longitudinal and conducted in larger number of subjects, are needed to better understand the relationship between PLMS and cSVD.
Conflict of Interest The authors declared they do not have anything to disclose.
References 1. Boulos MI, Murray BJ, Muir RT, et al. Periodic limb movements and white matter hyperintensities in first-ever
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16. 17.
18.
5
minor stroke or high-risk transient ischemic attacks. Sleep 2017;40. https://doi.org/10.1093/sleep/zsw080. Manconi M, Bassetti CL, Ferri R. Periodic limb movements during sleep and white matter hyperintensity in minor stroke or TIA. Sleep 2017;40. https://doi.org/ 10.1093/sleep/zsx031. Kang MK, Koo DL, Shin JH, et al. Association between periodic limb movements during sleep and cerebral small vessel disease. Sleep Med 2018;51:47-52. Del Brutto OH, Mera RM, Castillo PR, et al. Key findings from the Atahualpa Project: what should we learn? Expert Rev Neurother 2018;18:5-8. Wardlaw JM, Smith EE, Biessels GJ, et al. Neuroimaging standards for research into small vessel disease and its contribution to ageing and neurodegeneration. Lancet Neurol 2013;12:822-838. Pantoni L, Basile AM, Pracucci G, et al. Impact of agerelated cerebral white matter changes on the transition to disability: the LADIS study: rationale, design and methodology. Neuroepidemiology 2005;24:51-62. Gregoire SM, Chaudhary UJ, Brown MM, et al. The microbleed anatomical rating scale (MARS). Reliability of a tool to map brain microbleeds. Neurology 2009;73:1759-1766. Doubal FN, MacLullich AMJ, Ferguson KJ, et al. Enlarged perivascular spaces on MRI are a feature of cerebral small vessel disease. Stroke 2010;41:450-454. Berry RB, Budhiraja R, Gottlieb DJ, et al. Rules for scoring respiratory events in sleep: Update of the 2007 AASM manual for the scoring of sleep and associated events. J Clin Sleep Med 2012;8:597-619. Del Brutto OH, Santamaría M, Ochoa E, et al. Population-based study of cardiovascular Health in Atahualpa, a rural village of coastal Ecuador. Int J Cardiol 2013;168:1618-1620. Del Brutto OH, Santamaría M, Zambrano M, et al. Stroke in rural coastal Ecuador: a community-based survey. Int J Stroke 2014;9:365-366. Del Brutto OH, Del Brutto VJ, Zambrano M, et al. Prevalence of Willis-Ekbom disease in rural coastal Ecuador. A two-phase, door-to-door, population-based survey. J Neurol Sci 2014;344:139-142. Lin TC, Zeng BY, Chen YW, et al. Cerebrovascular accident risk in a population with periodic limb movements of sleep: a preliminary meta-analysis. Cerebrovasc Dis 2018;46:1-9. Gadodia G, Rangaraju S, Raza SA, et al. Frequency of and factors associated with obstructive sleep apnea and periodic limb movements in stroke and TIA patients. Neurologist 2018;23:67-70. Haba-Rubio J, Martí-Soler H, Marques-Vidal P, et al. Prevalence and determinants of periodic limb movements in the general population. Ann Neurol 2016; 79:464-474. Millikan C, Futrell N. The fallacy of the lacunar hypothesis. Stroke 1990;21:1251-1257. van Dijk AC, Fonville S, Zadi T, et al. Association between arterial calcifications and nonlacunar and lacunar ischemic strokes. Stroke 2014;45:728-733. Chung JW, Kim BJ, Sohn CH, Yoon BW, Lee SH. Branch atheromatous plaque: a major cause of lacunar infarction (high-resolution MRI study). Cerebrovasc Dis Extra 2012;2:36-44.
Lack of Association between Periodic Limb Movements during Sleep and Neuroimaging Signatures of Cerebral Small Vessel Disease in Stroke-Free Community-Dwelling Older Adults. The Atahualpa Project Oscar H. Del Brutto, MD,* Robertino M. Mera, MD, PhD,† Victor J. Del Brutto, MD,‡ and Pablo R. Castillo, MD§
Background: Evidence of the relationship between periodic limb movements during sleep (PLMS) and cerebral small vessel disease (cSVD) is limited and inconsistent. Here, we aimed to assess the independent association between PLMS and the different neuroimaging signatures of cSVD. Methods: Atahualpa residents aged more than or equal to 60 years enrolled in the Atahualpa Project undergoing polysomnography and MRI with time intervals less than or equal to 6 months were included. MRI readings focused on white matter hyperintensities (WMH) of presumed vascular origin, deep cerebral microbleeds (CMB), silent lacunar infarcts (LI), and more than 10 enlarged basal ganglia-perivascular spaces (BG-PVS). Data from single-night polysomnograms were interpreted according to recommendations of the American Academy of Sleep Medicine. Associations between the PLMS index and neuroimaging signatures of cSVD (as dependent variables) were assessed by means of logistic regression models, adjusted for relevant confounders. Results: A total of 146 individuals (mean age: 71.4 § 7.5 years; 64% women) were included. A PLMS index more than or equal to 15 per hour were noted in 48 (33%) participants. Moderate-to-severe WMH were present in 33 individuals (23%), deep CMB in 9 (6%), silent LI in 16 (11%), and more than 10 BG-PVS in 44 (30%). In univariate analyses, silent LI (P = .035) and the presence of more than 10 enlarged BG-PVS (P = .034) were significantly higher among participants with a PLMS index more than or equal to 15 per hour. However, fully-adjusted multivariate models showed no significant association between PLMS index more than or equal to 15 per hour and any of the neuroimaging signatures of cSVD. Conclusions: This study shows no independent association between the PLMS index and neuroimaging signatures of cSVD in stroke-free community-dwelling older adults. Key Words: Periodic limb movements during sleep—cerebral small vessel disease—white matter hyperintensities—cerebral microbleeds—silent lacunar infarcts—enlarged basal ganglia-perivascular spaces © 2019 Elsevier Inc. All rights reserved.
From the *School of Medicine, Universidad Espíritu Santo Ecuador, Samborond on, Ecuador; †Department of Epidemiology, Gilead Sciences, Inc, Foster City, CaliforniaA; ‡Department of Neurology, Miller School of Medicine, University of Miami, Miami, Florida; and §Sleep Disorders Center, Mayo Clinic College of Medicine, Jacksonville, Florida. Received June 10, 2019; revision received October 20, 2019; accepted October 22, 2019. Financial Disclosure: This study was supported by Universidad Espíritu Santo Ecuador. Address correspondence to Oscar H. Del Brutto, MD, School of Medicine, Universidad Espíritu Santo Ecuador, Air Center 3542, PO Box 522970, Miami, FL 33152-2970. E-mail: [email protected]. 1052-3057/$ - see front matter © 2019 Elsevier Inc. All rights reserved. https://doi.org/10.1016/j.jstrokecerebrovasdis.2019.104497
Journal of Stroke and Cerebrovascular Diseases, Vol. &&, No. && (&&), 2019: 104497
1
ARTICLE IN PRESS O.H. DEL BRUTTO ET AL.
2
Introduction Information on the relationship between periodic limb movements during sleep (PLMS) and neuroimaging signatures of cerebral small vessel disease (cSVD) is scarce and inconsistent. Indeed, a PubMED search using relevant keywords disclosed only 3 studies, 2 of them conducted in patients with stroke or transient ischemic attacks1,2 and the other in subjects with sleep-related symptoms referred to a specialized center.3 In addition, such studies included small numbers of patients, and might not be representative of the population at large. Current knowledge does not allow to conclude whether PLMS are associated or not with cSVD, and there is certainly no robust evidence on the direction of this relationship. The finding of a relationship between PLMS and neuroimaging signatures of cSVD would open new venues of research that may help to elucidate pathogenetic mechanisms involved in the development and severity of cSVD. By means of data available through the Atahualpa Project, we aimed to assess whether independent associations between the PLMS index and the different neuroimaging signatures of cSVD exist.
Methods Study Population Atahualpa residents share important characteristics for the study of potential relationships between sleep-related disorders and cSVD, including ethnicity (95% of the populations belong to the Ecuadorian native ethnic group), exposure to 12 daily hours of sunlight all over the year, hot and dry weather, virtually no shift working, limited nighttime light pollution, and socioeconomic status (almost all men belong to the blue-collar class and most women are homemakers).4 The Institutional Review Board of Hospital-Clínica Kennedy, Guayaquil, Ecuador (FWA 00006867) approved the study protocol and a comprehensive informed consent form that individuals must sign before enrollment.
Neuroimaging Protocol MRIs were performed with a Philips Intera 1.5T (Philips Medical Systems, Eindhoven, the Netherlands) at HospitalClínica Kennedy, Guayaquil. Exams included 2-dimensional multislice turbo spin echo T1-weighted, fluid attenuated inversi on recovery (FLAIR), T2-weighted, and gradient-echo sequences in the axial plane, as well as a FLAIR sequence oriented in the sagittal plane; slice thickness was 5 mm with 1 mm gap between slices. This protocol meets the standards proposed by Wardlaw et al5 for research on cSVD. Interest focused on: (1) white matter hyperintensities (WMH) of presumed vascular origin, defined as lesions appearing hyperintense on T2-weighted images that remained bright on FLAIR (without cavitation) and graded according to the modified Fazekas scale,6 (2) deep cerebral microbleeds (CMB),
identified and rated according to the microbleed anatomical rating scale,7 (3) silent lacunar infarcts (LI), defined as fluidfilled cavities measuring 3-15 mm located in the territory of a perforating arteriole,5 and (4) enlarged basal ganglia-perivascular spaces (BG-PVS), defined as less than 3 mm structures of CSF intensity—assessed on the T2-weighted sequence— that followed the orientation of perforating arteries, and rated as abnormal if more than 10 of these lesions were present in a single slice in 1 side of the brain (we used the BG slice with the highest number in 1 side).8 All MRIs were independently read by 2 raters blinded to clinical information (O.H.D. and V. J.D.). Kappa coefficients for interrater agreement were .90 for WMH, .76 for deep CMB, .90 for LI, and .83 for the presence of more than 10 enlarged BG-PVS; discrepancies were resolved by consensus.
Polysomnography (PSG) Diagnostic single-night PSGs were performed at the sleep unit of the Atahualpa Project Community Center. Exams were performed with the use of an Embletta X100 Comprehensive Portable PSG System (Embla Systems, Inc; Thornton, CO). A board-certified sleep medicine neurologist (P.R. C.), blinded to other information, reviewed raw data and interpreted all exams. PSGs were scored based upon recommended by the American Academy of Sleep Medicine scoring guidelines.9 All of the raw data were reviewed in 30 second epochs. Sleep efficiency, total time in bed, total sleep time, sleep architecture (percentages of time spent in N1, N2, N3, and rapid eye movements (REM) sleep), mean O2 saturation, and the apnea/hypopnea index (AHI) were used for objective assessment of PSG-derived parameters. In addition, the PLMS index was defined by the number of periodic leg movement events during sleep that meet PLMS criteria divided by the number of hours of sleep with leg movement recording. We used a nasal pressure transducer to exclude upper airway resistance. Limb movements were not scored within .5 seconds of a respiratory event.
Clinical Covariables Demographics and cardiovascular risk factors of interest (the body mass index, blood pressure, fasting glucose, and total cholesterol blood levels) were selected as confounding variables. These were assessed through interviews and procedures previously described in the Atahualpa Project.10 To exclude patients with an overt stroke and with the restless legs syndrome, rural doctors screened all participants with the use of validated field instruments, and then, certified neurologists confirmed the diagnosis as previously reported.11,12
Statistical Analyses Data analyses are carried out by using STATA version 15 (College Station, TX). In univariate analyses, continuous variables were compared by linear models and
ARTICLE IN PRESS ASSOCIATION BETWEEN PLMS AND NEUROIMAGING SIGNATURES OF CSVD
categorical variables by x2 or Fisher exact test as appropriate. Multivariate logistic regression models were fitted to assess the independent association between the PLMS index and the aforementioned neuroimaging signatures of cSVD (as dependent variables), after adjusting for demographics, cardiovascular risk factors, and other PSGderived information.
Results A total of 188 community-dwellers aged more than or equal to 60 years enrolled in the Atahualpa Project underwent PSG and MRI with time intervals of less than or equal to 6 months, and were eligible for this study. Of these, 42 were excluded (27 had an overt stroke, seven had a diagnosis of restless legs syndrome, and 8 presented technical difficulties precluding proper PSG interpretation). The mean age of the 146 included subjects was 71.4 § 7.5 years (median age: 71 years, age range: 60-95 years) and 94 (64%) were women. A body mass index more than or equal to 30 kg/m2 was noticed in 35 (24%) persons, blood pressure more than or equal to 140/90 mm Hg in 73 (50%), fasting glucose more than or equal to 126 mg/dL in 45 (31%), and total cholesterol levels more than or equal to 240 mg/dL in 27 (18%). Mean values of PSG-derived parameters included: total time in bed 480.9 § 30.6 minutes, total sleep time 365.6 § 64.2 minutes, mean O2 saturation 94.7 § 5.1%, sleep efficiency 75.6 §
3
11.7%, and AHI 11.7 § 12.6 (68% individuals had an AHI > 5 per hour). Mean percentages of time in the different sleep stages were: N1 sleep 4.3 § 4.3%, N2 sleep 67.7 § 14.4%, N3 sleep 12 § 8.8%, and REM sleep 15.9 § 10.8%. A PLMS index more than or equal to 15 per hour was recorded in 48 (33%) individuals. Moderate-to-severe WMH were noticed in 33 (23%) individuals, deep CMB in nine (6%), silent LI in 16 (11%), and more than 10 enlarged BG-PVS in 44 (30%). Table 1 shows characteristics of participants across categories of the PLMS index. In univariate analyses, increasing age (P < .001) and having blood pressure levels more than or equal to 140/90 mm Hg (P = .014) were significantly associated with a high PLMS index. Likewise, percentage of time spent in N2 sleep was lower, but that of time spent in REM sleep was higher among participants with a PLMS index more than or equal to 15 per hour. There were no significant differences in the other PSGderived parameters. Also in univariate analyses, silent LI (P = .035) and the presence of more than 10 enlarged BGPVS (P = .034) were significantly higher among participants with a PLMS index more than or equal to 15 per hour, but no differences were found in subjects with moderate-to-severe WMH or deep CMB across groups. Multivariate logistic regression models showed no significant association between a high PLMS index and any of the neuroimaging signatures of cSVD, after adjusting (in separate models) for clinical covariables, and for
Table 1. Characteristics of study participants across categories of the periodic limb movement index per hour (univariate analysis) PLM index <15 per hour (n = 98)
PLM index 15 per hour (n = 48)
P value
71.4 § 7.5 94 (64) 35 (24) 73 (50) 45 (31) 27 (18)
69.8 § 6.5 63 (64) 25 (26) 42 (43) 28 (29) 18 (18)
74.8 § 8.4 31 (65) 10 (21) 31 (65) 17 (35) 9 (19)
<.001* .972 .534 .014* .400 .955
480.9 § 30.6 365.6 § 64.2 94.7 § 5.1 4.3 § 4.3 67.7 § 14.4 12 § 8.8 15.9 § 10.8 75.6 § 11.7 100 (68)
479.8 § 31.3 363.9 § 65.5 95.1 § 1.7 4.1 § 4.2 69.6 § 13.8 12.2 § 8.8 14.1 § 9.8 75.2 § 12.5 68 (69)
483.3 § 28.9 369.2 § 61.3 94.1 § 8.5 4.9 § 4.5 63.8 § 14.7 11.5 § 8.6 19.5 § 11.9 76.4 § 11.8 32 (67)
.514 .640 .263 .293 .021* .650 .004* .580 .888
33 (23) 16 (11) 9 (6) 44 (30)
19 (19) 7 (7) 5 (5) 24 (24)
14 (29) 9 (19) 4 (8) 20 (42)
.184 .035* .477 .034*
Total series (n = 146) Clinical covariables Age in years, mean § SD Women, n (%) Body mass index 30 Kg/m2, n (%) Blood pressure 140/90 mmHg, n (%) Fasting glucose 126 mg/dL, n (%) Total cholesterol 240 mg/dL, n (%) Polysomnography-derived information Total time in bed in minutes, mean § SD Total sleep time in minutes, mean § SD O2 saturation, mean § SD N1 sleep, %, mean § SD N2 sleep, %, mean § SD N3 sleep, %, mean § SD REM sleep, %, mean § SD Sleep efficiency, %, mean § SD Apnea hypopnea index >5 per hour, n (%) Dependent variables Moderate-to-severe WMH, n (%) Silent lacunar infarctions, n (%) Deep cerebral microbleeds, n (%) >10 enlarged BG-PVS, n (%)
Abbreviations: BG-PVS, basal ganglia-perivascular spaces; PLM, periodic limb movement; WMH: white matter hyperintensities. *Statistically significant result.
ARTICLE IN PRESS O.H. DEL BRUTTO ET AL.
4
Table 2. Adjusted logistic regression models showing lack of association between the periodic limb movement index and the different neuroimaging signatures of cerebral small vessel disease (as dependent variables), after taking into account the effect of clinical confounders (upper panel) and clinical confounders plus PSG-derived information (lower panel) Model adjusted for clinical confounders, using the PLM index as the independent variable
Odds ratio
95% CI
P value
Moderate-to-severe white matter hyperintensities Deep cerebral microbleeds Silent lacunar infarcts >10 enlarged basal ganglia-perivascular spaces
1.31 1.74 2.08 1.36
.50-3.42 .38-7.88 .62-7.00 .56-3.33
.581 .474 .239 .498
Model adjusted for clinical confounders plus PSG-derived information, using the PLM index as the independent variable
Odds ratio
95% CI
P value
Moderate-to-severe white matter hyperintensities Deep cerebral microbleeds Silent lacunar infarcts >10 enlarged basal ganglia-perivascular spaces
1.61 4.61 2.67 1.29
.57-4.58 .58-36.8 .58-12.4 .44-3.79
.372 .148 .210 .648
clinical covariables plus PSG-derived parameters (Table 2). Some covariables remained significant in the fullyadjusted models, including increased age (P = .002), high total cholesterol blood levels (P = .034) and an AHI more than 5 per hour (P = .036) in the model using WMH as the dependent variable, and increased age (P < .001) and high total cholesterol blood levels (P = .047) in the model using more than 10 enlarged BG-PVS as the dependent variable. There were no other covariables remaining significant in the models using silent LI or deep CMB as dependent variables.
Discussion The increasingly recognized evidence of an association between PLMS and cerebrovascular events13,14 motivated the assessment of a potential association between the PLMS index and neuroimaging signatures of cSVD. However, studies reported inconsistent results, which are likely related to differences in study designs, inclusion of selected (and biased) populations, and small samples sizes. The present study is the largest series reported to date (n = 146) and, more importantly, is the first in systematically assessing the possible association between the PLMS index and all the neuroimaging signatures of cSVD in an unbiased sample of community-dwelling older adults with a median age of 71 years, when both PLMS and neuroimaging signatures of cSVD are common. The first study—published by Boulos et al1—included 30 consecutive patients (mean age: 63.7 § 13.5 years) presenting within the first 2 weeks of a minor stroke or a transient ischemic attack. In such study, patients with a PLMS index more than or equal to 5 per hour had a greater burden of WMH (in cm3) than those with a PLMS index less than 5 per hour (7.8 § 3.6 versus 4.2 § 3.8; P = .012), after adjusting for cardiovascular risk factors and PSG-derived
parameters. The study did not assess the relationship between PLMS and the other neuroimaging signatures of cSVD. Thereafter, Manconi et al2 could not replicate those previous findings after studying 68 patients (mean age: 59.8 years) selected from a larger cohort study; participants were stratified in 2 arms (PLMS index <5 and 5 per hour), and the authors found no differences in WMH volumes across groups. Likewise, other neuroimaging signatures of cSVD were not assessed in this second study. The most recent study, conducted by Kang et al,3 enrolled 60 stroke-free subjects (mean age: 61.7 § 14.4 years) visiting a clinic for sleep-related symptoms. Those individuals were stratified in 2 arms, including 31 with a PLMS index more than or equal to 15 per hours and 29 with a PLMS index of less than 5 per hour, and compared the prevalence of the 4 neuroimaging signatures of cSVD across groups, after adjusting for several confounders. The authors found significant associations between the PLMS index and WMH and enlarged perivascular spaces, but not with silent LI or CMB. There was no explanation for the exclusion of subjects with a PLMS index between 5 and 14.9 per hour. This gap between arms might have influenced the positive associations found in the study. Despite the clear cutoff established for defining an abnormal PLMS index in adults (15 per hour),15 there is still controversy regarding which cutoff should better be used to find an association between a high PLMS index and neuroimaging signatures of cSVD, we carried out a poststudy analysis to assess the above-mentioned associations using a PLMS index cutoff of more than or equal to 5 per hour to define high PLMS. In these models, a high PLMS index was not associated with moderate-to-severe WMH, the presence of deep CMB or more than 10 enlarged BG-PVS, after adjusting for clinical covariates or in a fully-adjusted model including both clinical covariates and PSG-derived information. The only exception
ARTICLE IN PRESS ASSOCIATION BETWEEN PLMS AND NEUROIMAGING SIGNATURES OF CSVD
was the finding of a significant association between a high PLMS index and silent LI in the fully-adjusted model (OR: 9.07; 95% C.I.: 1.08-76.10; P = .042). However, this finding does not necessarily means an association between a PLMS index more than or equal to 5 per hour and cSVD for 2 main reasons: (1) the PLMS index was not associated with the other neuroimaging signatures of cSVD, and (2) many silent LI are not related to cSVD but to other pathogenetic mechanisms such as cardiogenic or artery-toartery cerebral embolisms.16-18 As previously noted, there are no convincing arguments supporting the association between PLMS and cSVD or the direction of this relationship. In the nonconsented scenario that PLMS are associated with cSVD, it is possible that PLMS may predispose to the severity of WMH because of fluctuations in nocturnal blood pressure and heart rate, oxidative stress, inflammation, and hypoxia. In the event of a reverse causation phenomenon, that is, WMH causing PLMS, the disruption of neural networks associated with WMH would be in the path for PLMS occurrence. Those proposed mechanisms would only explain a possible association between PLMS and WMH, but there are no clues on the pathogenesis of the association between PLMS and the other signatures of cSVD. In the present study, the significant association between the PLMS index and silent LI and more than 10 enlarged BG-PVS that we found in univariate analysis disappeared in multivariate regression models, probably by the effect of confounders. Of interest, the odds ratio for some of these associations was high in multivariate models, but significances were probably tempered by the high range of 95% CI due to the sample size. Despite the strengths of the present study, such as the unbiased inclusion of participants and the systematic evaluation of PSG-derived parameters and neuroimaging signatures of cSVD by means of internationally accepted standards,5,9 there are potential limitations that include the cross-sectional design and, more importantly, the well-recognized night-to-night variability of PLMS, which was not possible to estimate by the use of a single-night diagnostic PSG. Further studies, preferably longitudinal and conducted in larger number of subjects, are needed to better understand the relationship between PLMS and cSVD.
Conflict of Interest The authors declared they do not have anything to disclose.
References 1. Boulos MI, Murray BJ, Muir RT, et al. Periodic limb movements and white matter hyperintensities in first-ever
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16. 17.
18.
5
minor stroke or high-risk transient ischemic attacks. Sleep 2017;40. https://doi.org/10.1093/sleep/zsw080. Manconi M, Bassetti CL, Ferri R. Periodic limb movements during sleep and white matter hyperintensity in minor stroke or TIA. Sleep 2017;40. https://doi.org/ 10.1093/sleep/zsx031. Kang MK, Koo DL, Shin JH, et al. Association between periodic limb movements during sleep and cerebral small vessel disease. Sleep Med 2018;51:47-52. Del Brutto OH, Mera RM, Castillo PR, et al. Key findings from the Atahualpa Project: what should we learn? Expert Rev Neurother 2018;18:5-8. Wardlaw JM, Smith EE, Biessels GJ, et al. Neuroimaging standards for research into small vessel disease and its contribution to ageing and neurodegeneration. Lancet Neurol 2013;12:822-838. Pantoni L, Basile AM, Pracucci G, et al. Impact of agerelated cerebral white matter changes on the transition to disability: the LADIS study: rationale, design and methodology. Neuroepidemiology 2005;24:51-62. Gregoire SM, Chaudhary UJ, Brown MM, et al. The microbleed anatomical rating scale (MARS). Reliability of a tool to map brain microbleeds. Neurology 2009;73:1759-1766. Doubal FN, MacLullich AMJ, Ferguson KJ, et al. Enlarged perivascular spaces on MRI are a feature of cerebral small vessel disease. Stroke 2010;41:450-454. Berry RB, Budhiraja R, Gottlieb DJ, et al. Rules for scoring respiratory events in sleep: Update of the 2007 AASM manual for the scoring of sleep and associated events. J Clin Sleep Med 2012;8:597-619. Del Brutto OH, Santamaría M, Ochoa E, et al. Population-based study of cardiovascular Health in Atahualpa, a rural village of coastal Ecuador. Int J Cardiol 2013;168:1618-1620. Del Brutto OH, Santamaría M, Zambrano M, et al. Stroke in rural coastal Ecuador: a community-based survey. Int J Stroke 2014;9:365-366. Del Brutto OH, Del Brutto VJ, Zambrano M, et al. Prevalence of Willis-Ekbom disease in rural coastal Ecuador. A two-phase, door-to-door, population-based survey. J Neurol Sci 2014;344:139-142. Lin TC, Zeng BY, Chen YW, et al. Cerebrovascular accident risk in a population with periodic limb movements of sleep: a preliminary meta-analysis. Cerebrovasc Dis 2018;46:1-9. Gadodia G, Rangaraju S, Raza SA, et al. Frequency of and factors associated with obstructive sleep apnea and periodic limb movements in stroke and TIA patients. Neurologist 2018;23:67-70. Haba-Rubio J, Martí-Soler H, Marques-Vidal P, et al. Prevalence and determinants of periodic limb movements in the general population. Ann Neurol 2016; 79:464-474. Millikan C, Futrell N. The fallacy of the lacunar hypothesis. Stroke 1990;21:1251-1257. van Dijk AC, Fonville S, Zadi T, et al. Association between arterial calcifications and nonlacunar and lacunar ischemic strokes. Stroke 2014;45:728-733. Chung JW, Kim BJ, Sohn CH, Yoon BW, Lee SH. Branch atheromatous plaque: a major cause of lacunar infarction (high-resolution MRI study). Cerebrovasc Dis Extra 2012;2:36-44.