Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Different effects of cardiometabolic syndrome on brain age in relation to gender and ethnicity

Different effects of cardiometabolic syndrome on brain age in relation to gender and ethnicity Background A growing body of evidence shows differences in the prevalence of cardiometabolic syndrome (CMS) and dementia based on gender and ethnicity. However, there is a paucity of information about ethnic‑ and gender ‑ specific CMS effects on brain age. We investigated the different effects of CMS on brain age by gender in Korean and British cognitively unimpaired (CU) populations. We also determined whether the gender‑specific difference in the effects of CMS on brain age changes depending on ethnicity. Methods These analyses used de‑identified, cross‑sectional data on CU populations from Korea and United Kingdom (UK) that underwent brain MRI. After propensity score matching to balance the age and gender between the Korean and UK populations, 5759 Korean individuals (3042 males and 2717 females) and 9903 individuals from the UK (4736 males and 5167 females) were included in this study. Brain age index (BAI), calculated by the difference between the predicted brain age by the algorithm and the chronological age, was considered as main outcome and presence of CMS, including type 2 diabetes mellitus ( T2DM), hypertension, obesity, and underweight was considered as a predic‑ tor. Gender (males and females) and ethnicity (Korean and UK) were considered as effect modifiers. Results The presence of T2DM and hypertension was associated with a higher BAI regardless of gender and ethnicity (p < 0.001), except for hypertension in Korean males (p = 0.309). Among Koreans, there were interaction effects of gen‑ der and the presence of T2DM (p for T2DM*gender = 0.035) and hypertension (p for hypertension*gender = 0.046) on BAI in Koreans, suggesting that T2DM and hypertension are each associated with a higher BAI in females than in males. In contrast, among individuals from the UK, there were no differences in the effects of T2DM (p for T2DM*gender = 0.098) and hypertension (p for hypertension*gender = 0.203) on BAI between males and females. Conclusions Our results highlight gender and ethnic differences as important factors in mediating the effects of CMS on brain age. Furthermore, these results suggest that ethnic‑ and gender ‑specific prevention strategies may be needed to protect against accelerated brain aging. Keywords Brain age, Cardiometabolic syndrome, Gender, Ethnicity Sung Hoon Kang and Mengting Liu contributed equally to this work. *Correspondence: Sang Won Seo sw72.seo@samsung.com Full list of author information is available at the end of the article © The Author(s) 2023. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. The Creative Commons Public Domain Dedication waiver (http:// creat iveco mmons. org/ publi cdoma in/ zero/1. 0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Kang et al. Alzheimer’s Research & Therapy (2023) 15:68 Page 2 of 10 by gender and ethnicity, we hypothesized that there Background might be differences in the effects of CMS on the BAI in Aging is an important risk factor for cognitive impair- relation to gender and ethnicity. ment and dementia [1]. As aging progresses, brain atro- phy also occurs at a mean volume reduction rate of 0.5% per year after the age of 40 [2, 3]. Age-related brain atro- phy is referred to as the brain age. Cardiometabolic syn- Methods drome (CMS), syndrome X, metabolic syndrome, and Study populations cardiometabolic dysfunction, composed of type 2 dia- We enrolled CU participants aged ≥ 45  years from the betes mellitus (T2DM), hypertension, and obesity, are Health Promotion Center of Samsung Medical Center critical modifiable risk factors for cognitive impairment. (Seoul, Korea) who underwent a comprehensive health There is also a growing body of evidence that CMS has screening exam from September 1, 2008, to October deleterious effects on brain atrophy [4] even in non- 31, 2019. A total of 8227 eligible candidates underwent demented population [5, 6]. Previous studies have sug- a full medical examination, which included cognitive gested that CMS may accelerate brain aging. assessment and 3.0-Tesla MRI, including high-resolution Several cross-sectional studies have shown differences T1-weighted MRI, as part of a standard screening for in brain volume between males and females in cognitively dementia. The medical examination procedure used for unimpaired (CU) populations [7–9]. Previously, changes the participants has been previously described [23]. We in brain age or atrophy were shown to occur differently excluded participants who had any of the following con- depending on gender [3, 10, 11]. Additionally, previous ditions: 728 participants with missing data on years of studies based on Hispanic or Korean populations sug- education or Mini-Mental State Examination (MMSE) gested that CMS-associated brain atrophy was more score [24]; 509 participants with significant cognitive extensive or prominent in females than in males [6, 10]. impairment defined by MMSE scores below the 16th per - However, considering the differences in the prevalence of centile in age-, gender-, and education-matched norms or CMS and dementia between Korean and European popu- through an interview conducted by a qualified neurolo - lations, it would be reasonable to hypothesize that there gist; 312 participants with severe cerebral white matter might be a difference in the gender-specific relation - hyperintensities (deep white matter ≥ 2.5  cm and caps ship between CMS and brain age between Koreans and or band ≥ 1.0 cm) or structural lesions such as territorial Europeans. infarction, lobar hemorrhage, brain tumor, and hydro- Previous studies have analyzed various morphologi- cephalus; 542 participants with missing information on cal features on brain magnetic resonance imaging (MRI), DM, hypertension, or body mass index (BMI); and 377 including cortical thickness [12], regional gray matter participants with unreliable analyses of cortical thickness volume [11], and white matter hyperintensity [13] and due to head motion, blurred MRI, inadequate registra- integrity [14], and investigated the impact of CMS on tion to a standardized stereotaxic space, misclassifica - brain structure in aging populations. Recently, various tion of tissue type, or inexact surface extraction. Finally, machine learning approaches have been developed to 5759 participants (3042 males and 2717 females) were begin accurate prediction of brain age using the afore- included in this study. mentioned brain imaging features [15–18] and provide Similar data for people of British ancestry was obtained a new metric called brain age index (BAI) to measure from the UK Biobank (UKB, http:// www. ukbio bank. how old the brain age is compared to the chorological ac. uk), a population-based prospective cohort study of age at MRI scan. The difference between the predicted approximately 500,000 people in the UK [https:// journ brain age using a deep learning-based algorithm and the al s . plo s . or g/ plo sm e dic i ne/ ar tic le? id= 10. 1371/ j our n chronological age is called the BAI, which explains how al. pmed. 10017 79, https:// www. nature. com/ artic les/ much older or younger an individual brain appears com-s41586- 018- 0579-z]. Of these participants, approximately pared to the current age. A positive BAI is a novel marker 40,000 attended an additional visit during which MRI of an older brain and has been shown to predict compro- brain imaging data was collected in addition to other mised brain health [19], earlier mortality [20], and cogni- health-related data [https:// www. nature. com/ artic les/ tive impairment [21, 22].s41467- 020- 15948-9]. We included non-Hispanic White In the present study, we investigated the different adults only in the present study. We excluded partici- effects of CMS on BAI with respect to the sex of CU pop - pants with a self-reported or hospital record-based his- ulations from Korea and United Kingdom (UK). Next, tory of dementia, Parkinson’s disease, or other central we determined whether CMS affects gender-specific BAI nervous system-related diseases. Finally, 9903 (4736 differently according to ethnicity. Considering that there males and 5167 females) UKB participants were included are differences in incidence of CMS and cortical atrophy after applying the inclusion/exclusion criteria and after K ang et al. Alzheimer’s Research & Therapy (2023) 15:68 Page 3 of 10 random selection of a smaller subset of participants for Field echo (TFE) MRI data using the following imaging brain imaging data processing. parameters: sagittal slice thickness, 1.0  mm with 50% The institutional review board of the Samsung Medical overlap; no gap; repetition time of 9.9  ms; echo time of Center approved this study and adhered to the principles 4.6 ms; flip angle of 8; and matrix size of 240 × 240 pixels of the Declaration of Helsinki. Written informed consent reconstructed to 480 × 480 over a field view of 240 mm. was obtained from all participants in the Health Promo- In the UKB populations, brain MRI scans were tion Center of Samsung Medical Center. Anonymous obtained at one of the three assessment sites using a 3.0 and deidentified data from the UKB was used for analy - Tesla Siemens Skyra MRI Scanner. Among the six brain sis and, therefore, the present study was exempted from imaging modalities acquired was a T1-weighted, sag- obtaining informed consent. ittal 3D magnetization prepared rapid gradient echo (MPRAGE) scan. The following imaging parameters were Measurement of cardiometabolic syndrome used in this T1-weighted acquisition: inversion time of For populations from the Health Promotion Center at 880 ms; repetition time of 2000 ms; 1 × 1 × 1 mm voxel the Samsung Medical Center, a health screening exam size; 208 × 256 × 256 matrix size; and SENSE factor (R) was conducted by a well-trained medical professional of 2.0 [Miller, 2016: https:// www. nature. com/ artic les/ nn. using standardized protocols. Baseline workup included 4393]. blood tests (complete blood cell count, liver/kidney/thy- roid function test, and tumor markers), urine analyses, Image processing and cortical surface extraction abdominal sonography, chest radiography, electrocar- T1-weighted MRI scans from the Health Promotion diogram, pulmonary function test, and gastroduoden- Center in Korea and the UKB were used to reconstruct oscopy. We classified each CMS component using the the inner and outer cortical boundaries using the CIVET following criteria: T2DM was defined as a diagnostic his - pipeline developed at the Montreal Neurological Insti- tory of T2DM or current use of any anti-diabetic medica- tute (http:// www. bic. mni. mcgill. ca/ Servi cesSo ftware/ tion; hypertension was defined as a diagnostic history of CIVET). Cortical morphology was quantitatively char- hypertension or current use of any antihypertensive med- acterized by measuring cortical thickness, sulcal depth, ication; obesity and underweight were defined using the and gray/white intensity ratio [26] on the cortical sur- cut-off for BMI calculated by weight (kilograms)/height face at 81,924 vertices (163,840 polygons). These features (meters) squared at the first visit. According to a previ - were further resampled to the surface template using the ous study [10], populations with BMI < 18.5  kg/m were transformation obtained in the surface registration to labeled as underweight, and those with BMI ≥ 27.5 kg/m allow for inter-subject comparisons. were labeled as obese. For populations from the UKB, the classification of Development of prediction model for relative brain age T2DM, hypertension, and obesity was determined based As illustrated in Fig.  1, we did not use topology-varying on a combination of a touchscreen-based questionnaire, surfaces because of the nature of the graph convolutional a verbal interview, and linked hospital records. Specifi - networks (GCN) model used in this study. Rather, we cally, T2DM was defined as either self-reported T2DM, considered the cortical morphological changes that occur a doctor’s diagnosis of T2DM (Data Field 2443), patients in relation to brain size and gyrification using cortical who were taking insulin (Data Fields 6177, 6153), or a thickness, volume, and sulcal depth. The GCN employed hospital data-linked record of an individual with a diag- in our study requires identical graph/mesh structures for nosis of T2DM. Hypertension was defined as either all individual inputs, whereas the features of the nodes/ self-reported hypertension (Data-Field 20,002) or a hos- vertices can vary. Another advantage of the topology- pital data-linked record of having a primary or secondary kept surface model is that surface nodes are registered diagnosis of hypertension (Data-Fields 41,202, 41,204, across all individuals such that anatomical information is 41,203, and 41,205). BMI was calculated using weight and shared. height measurements in the same way as the Samsung Medical Center data. Populations with a BMI < 18.5  kg/ Brain age index m were categorized as underweight, whereas those with After calculating the predicted brain age for each subject, BMI ≥ 35 kg/m were categorized as obese [25]. we further calculated a metric that reflected the subject’s relative brain health status, called the BAI. BAI was ini- Acquisition of brain MRI tially measured by subtracting the true brain age from the All Korean populations underwent a 3D volumetric brain predicted brain age [27]. Due to regression dilution [28], MRI scan. An Achieva 3.0-Tesla MRI scanner (Philips, however, it is also possible that regression models bias the Best, the Netherlands) was used to acquire 3D T1 Turbo predicted brain age toward the mean, underestimating Kang et al. Alzheimer’s Research & Therapy (2023) 15:68 Page 4 of 10 Fig. 1 The graph‑based convolutional network for brain age prediction the age of older subjects and overestimating the age of populations, we performed linear regression analyses younger subjects [29]. When deriving the BAI, this bias by adding each two-way interaction term (the pres- must be corrected using a strategy introduced in other ence of each CMS component*gender) to covariates in studies [15, 28]. We defined the new BAI as the differ - Korean and UK populations after controlling for the ence between the individual BAI and the expected BAI other CMS components. To assess whether the asso- (measurement fitted over the entire sample set using the ciation between the presence of each CMS component regression model and leave-one-out cross-validation). and brain age might differ by gender and ethnicity, we The BAI was corrected such that the BAIs of the whole performed linear regression analyses with the addition dataset analyzed became unbiased across all age ranges. of each three-way interaction term (the presence of each CMS component*gender*ethnicity) to covariates after controlling for the other CMS components. Propensity score matching False discovery rate (FDR) correction was conducted Propensity score matching was performed to minimize for all statistical analyses to control for p-values, and the differences in the demographics and cardiometa - q-values were obtained after FDR correction. All bolic factors between the UK and KOR participants. The reported p-values and q-values were two-sided and the propensity score was obtained using multivariable logis- significance level was set at 0.05. All analyses were per - tic regression based on age, gender, T2DM, hyperten- formed using R version 4.3.0 (Institute for Statistics and sion, and obesity. A total of 5541 KOR participants were Mathematics, Vienna, Austria; www.R- proje ct. org). matched with 9903 UK participants based on propensity scores using the 1:2 nearest-neighbor matching algo- rithm with caliper of 0.1. A good balance was achieved Results between the KOR and UK participants, with all standard- Demographics of cognitively unimpaired populations ized mean differences (age, gender, T2DM, hypertension, in the UK and Korea and obesity) below 0.1 after matching. After propensity score matching, the demographic characteristics of the two ethnic datasets were similar Statistical analysis (Table  1). Among the 5541 Korean populations, there Independent t-tests and chi-squared tests were used were 2599 (46.9%) females and 2942 (53.1%) males. to compare continuous and categorical variables, Among the 9903 UK populations, there were 5167 respectively. (52.2%) females and 4736 (47.8%) males. There were To explore the association between the presence of some differences in mean age (64.0 and 63.6  years, each CMS component and brain age in females and p < 0.001), female ratio (46.9 and 52.2%, p < 0.001), males among the Korean and UK populations, we per- and the presence of T2DM (17.3 and 9.8%, p < 0.001), formed a linear regression analysis with the presence hypertension (42.7% and 40.6%, p = 0.011), obesity of T2DM, hypertension, obesity, and underweight as (10.7 and 7.9%, p < 0.001), and underweight (1.8 and covariates. To assess whether the association between 0.6%, p < 0.001) between Koreans and participants from the presence of each CMS component and brain the UK. age might differ by gender in the Korean and the UK K ang et al. Alzheimer’s Research & Therapy (2023) 15:68 Page 5 of 10 Table 1 Demographics of populations from UK Biobank and Health Promotion Center in Korea Variables Korea UK p-value Females Males (n = 2942) Total (n = 5541) Females Males (n = 4736) Total (n = 9903) (n = 2599) (n = 5167) Age (years) 63.2 ± 6.9 64.7 ± 6.5 64.0 ± 6.7 63.4 ± 7.1 63.8 ± 7.4 63.6 ± 7.2 0.007 Hypertension (n, 965 (37.1%) 1402 (47.7%) 2367 (42.7%) 1813 (35.1%) 2208 (46.6%) 4021 (40.6%) 0.1 %) T2DM (n, %) 273 (10.5%) 683 (23.2%) 956 (17.3%) 365 (7.1%) 602 (12.7%) 967 (9.8%) < 0.001 2 a BMI (kg/m ) 23.5 ± 2.9 24.5 ± 2.6 24.0 ± 2.8 26.8 ± 5.2 27.7 ± 4.4 27.2 ± 4.9 < 0.001 Obesity (n, %) 227 (8.7%) 336 (12.4%) 593 (10.7%) 450 (8.7%) 336 (7.1%) 786 (7.9%) < 0.001 Underweight (n, %) 70 (2.7%) 32 (1.1%) 102 (1.8%) 49 (0.9%) 7 (0.1%) 56 (0.6%) < 0.001 Propensity score matching was performed to balance the age and gender between the Korean and UK populations, and 9903 out of 17,791 populations in the UK and 5541 out of 5759 populations in Korea were selected for the present study Abbreviations: BMI, body mass index; T2DM, type2 diabetes mellitus; UK, United Kingdom Distribution of age and BMI was compared between the populations of UK and Korea using independent t tests Distribution of education level and presence of hypertension, T2DM, obesity, and underweight between the populations of UK and Korea were tested using the chi-squared test Eec ff ts of cardiometabolic syndrome components on brain Interactive effects of cardiometabolic syndrome age index components on brain age index in relation to gender As shown in Fig.  2, DM was associated with increased and ethnicity BAI for all participants, regardless of gender and eth- We also investigated the interaction of the presence of nicity (q < 0.001 in the four groups) (Table  2). Hyper- each CMS component and gender with BAI in Koreans tension was associated with a significantly higher BAI and participants from the UK. Among Koreans, there for all participants (q < 0.001), except for Korean males were interactions between T2DM and gender with BAI (q = 0.309). Obesity significantly increased the BAI for (q = 0.035) and between hypertension and gender with UK males (q = 0.004). Being underweight increased the BAI (q = 0.046), suggesting that the effects of T2DM and BAI significantly only for UK females (q = 0.002). hypertension on BAI were more prominent in females Fig. 2 BAI distribution between groups regarding gender and ethnicity for healthy participants and participants with different CMS. BAI = 0 indicates that the chronological age is the same as the predicted brain age, with higher values indicating an older‑appearing brain than chronological age. Asterisk symbol ( ) indicates the following: q‑ values, FDR‑ corrected p‑ values, are lower than 0.05. BAI, brain age index; Kor, Korea; UK, United Kingdom; CMS, cardiometabolic syndrome Kang et al. Alzheimer’s Research & Therapy (2023) 15:68 Page 6 of 10 Table 2 Brain age index in controls and four CMS component groups Ethnicity/gender CMS Age BAI β (SE) q-value Korean females Control 61.6 ± 8.1 − 0.67 ± 3.57 T2DM 66.9 ± 9.5 1.51 ± 4.44 1.73 (0.25) < 0.001 Hypertension 66.0 ± 8.0 0.39 ± 3.97 0.80 (0.20) < 0.001 Obesity 64.8 ± 8.5 0.31 ± 3.69 0.26 (0.94) 0.347 Underweight 62.8 ± 7.6 0.30 ± 4.34 0.77 (0.48) 0.145 Korean males Control 64.1 ± 8.2 0.70 ± 3.55 T2DM 65.5 ± 8.6 1.82 ± 4.05 0.88 (0.19) < 0.001 Hypertension 65.3 ± 8.4 1.29 ± 4.13 0.19 (0.18) 0.309 Obesity 63.4 ± 7.7 1.52 ± 4.03 0.37 (0.23) 0.144 Underweight 67.5 ± 10.7 2.17 ± 3.63 1.24 (0.72) 0.171 UK females Control 62.0 ± 7.2 − 0.17 ± 3.50 T2DM 64.4 ± 7.4 1.22 ± 3.87 1.08 (0.20) < 0.001 Hypertension 65.2 ± 6.6 0.45 ± 3.79 0.49 (0.13) < 0.001 Obesity 61.5 ± 6.7 0.53 ± 3.54 0.32 (0.19) 0.088 Underweight 63.8 ± 7.0 1.60 ± 3.87 1.67 (0.53) 0.002 UK males Control 61.5 ± 7.5 0.25 ± 3.66 T2DM 65.8 ± 6.8 2.42 ± 3.90 1.55 (0.17) < 0.001 Hypertension 65.6 ± 6.8 1.35 ± 3.86 0.73 (0.13) < 0.001 Obesity 62.9 ± 7.2 1.72 ± 3.63 0.65 (0.22) 0.004 Underweight 68.0 ± 6.3 − 1.21 ± 2.28 − 1.69 (1.42) 0.235 Values of age and BAI are presented as mean ± standard deviation Abbreviations: BAI, brain age index; CMS, cardiometabolic syndrome; T2DM, type2 diabetes mellitus; UK, United Kingdom q-values, FDR-corrected p-values, were obtained using a linear regression analysis with the presence of hypertension, T2DM, obesity, and underweight as covariates in each group (Korean females, Korean males, UK females and UK males) than in males (Table  3, Fig.  3). Among British partici- Table 3 Interaction effect on the difference in BAI between participants with each CMS and those with control pants, however, there were no interactions of any CMSs and gender with BAI (q range 0.098 to 0.203, Table  3, Ethnicity Two-way interaction (each CMS β (SE) q-value Fig.  3). In fact, there were interactions between gender component*gender) and ethnicity for T2DM (q = 0.004) and hypertension Korea T2DM*gender 0.84 (0.32) 0.035 (q = 0.011, Table 3, Fig. 3). Hypertension*gender 0.61 (0.27) 0.046 Obesity*gender − 0.11 (0.36) 0.770 Underweight*gender − 0.82 (0.85) 0.450 Discussion UK T2DM*gender − 0.52 (0.26) 0.098 In the present study, we systematically investigated the Hypertension*gender − 0.26 (0.18) 0.203 different effects of CMS on BAI in relation to gender Obesity*gender − 0.37 (0.28) 0.191 and ethnic differences in a large sample of Korean and Underweight*gender 3.51 (1.50) 0.076 ¥ UK CU populations. Our major findings are as follows: Three-way β (SE) q-value interaction(each CMS first, among Koreans, the effects of DM and hypertension component*gender*ethnicity) on BAI were higher in females than in males. This indi - Both T2DM*gender*ethnicity 1.36 (0.41) 0.004 cated interaction effects of gender and the presence of Hypertension*gender*ethnicity 0.89 (0.32) 0.011 T2DM and hypertension on BAI in Korean population. Abbreviations: BAI, brain age index; CMS, cardiometabolic syndrome; T2DM, Second, among the UK population, there were no differ - type2 diabetes mellitus; UK, United Kingdom ences in the effects of T2DM and hypertension on BAI q-values, FDR-corrected p-values, were obtained using linear regression between males and females. Overall, there was evidence analyses with adding each two-way interaction term (the presence of each CMS component*gender) to covariates in Korean and UK populations after that ethnicity modified the gender-specific relationship controlling for the other CMS components of T2DM and hypertension with BAI. Taken together, q-values, FDR-corrected p-values, were obtained using linear regression our findings suggest that CMS exerts different effects on analyses with additionally adding each three-way interaction term (the presence brain age depending on gender and ethnicity. Therefore, of each CMS component*gender*ethnicity) to covariates after controlling for the other CMS components K ang et al. Alzheimer’s Research & Therapy (2023) 15:68 Page 7 of 10 Fig. 3 Ethnic‑ and gender ‑specific difference in BAI between participants with and without T2DM and HTN. Values depicted in the bar plot represent the mean of BAI, and values depicted in the error bar represent the standard error of mean. BAI = 0 indicates that the chronological age is the same as the predicted brain age, with higher values indicating an older‑appearing brain than chronological age. Asterisk ( ) symbol indicates the following: q‑ values, FDR‑ corrected p‑ values, were obtained using linear regression analyses with adding each two‑ way interaction term (the presence of each CMS component*gender) to covariates in Korean and UK populations after controlling for the other CMS components. Yen ( ) symbol indicates the following: q‑ values, FDR‑ corrected p‑ values, were obtained using linear regression analyses with additionally adding each three‑ way interaction term (the presence of each CMS component*gender*ethnicity) to covariates after controlling for the other CMS components. BAI, brain age index; Kor, Korea; UK, United Kingdom; T2DM, type 2 diabetes mellitus; HTN, hypertension ethnic- and gender-specific prevention strategies may be with autonomic dysfunction in females than in males necessary to protect against accelerated brain aging. [35]. Similar associations are witnessed in the cases of We found that the presence of T2DM and hypertension microalbuminuria [36] and reduction of heart function was associated with a higher BAI regardless of gender [37]. In particular, females uniquely experience meno- and ethnicity, except for hypertension in Korean males. pause transition, which might accelerate cardiometabolic T2DM and hypertension are well-known risk factors for syndrome, brain aging, or cognitive impairment via sev- brain atrophy, which is an important indicator of brain eral mechanisms including changes in the availability of age [30]. T2DM may have deleterious effects on the brain estrogen [38], estrogen receptor activity, and/or estro- via various mechanisms, including cerebrovascular com- gen-regulated neural networks [39]. Specifically, estrogen plications, glucose toxicity due to insulin resistance, and deficiency in postmenopausal females leads to inflamma - chronic inflammation [31]. Similarly, the positive asso - tory process and vasoconstriction via the dysfunction of ciation between hypertension and BAI may be due to the renin-angiotensin system [40–42]. In fact, a growing several possible mechanisms including cerebral hypoper- body of evidence shows that menopause has a deleteri- fusion, micro- and macrovascular damage in white mat- ous impact on cognitive function, which may contribute ter, and cerebral microinfarcts [32, 33]. to the higher prevalence of dementia in females than in Our first major finding was that Korean females suf - males [43–47]. Additionally, several studies have shown fered more deleterious effects of T2DM and hyperten - that females tend to maintain lifestyles that are more sion on brain age than Korean males. Although the favorable for brain health, with overall lower drinking underlying mechanisms for the gender-specific effects and smoking rates [48–51]. Therefore, our findings might of T2DM and hypertension on brain age are not fully be also related to differences in stress, alcohol consump - understood, our findings might be related to the com - tion, smoking, and dietary habits according to gender. plex effects of biological and socioeconomic differences Our second major finding was that there were interac - [34]. Previous studies have suggested that hypertension tive effects of the presence of T2DM and hypertension, exerts worse effects on multiple organs in females than gender, and ethnicity on BAI. That is, unlike Koreans, in males. This was attributed to differences in sex hor - there were no differences in the effects of T2DM and mones. There are stronger associations of hypertension hypertension on BAI between males and females in the Kang et al. Alzheimer’s Research & Therapy (2023) 15:68 Page 8 of 10 UK population. A few studies have found that brain age hyperintensities, which can also be associated with brain differs depending on ethnicity [52, 53]. However, gen- age. Further studies are needed to identify the effects of der- and ethnicity-specific differences in the effects of CMS on brain aging in relation to the pathophysiologi- T2DM and hypertension on brain age have not been cal processes. Despite the aforementioned limitations, extensively investigated. These differences might be our study is the first report to compare the gender- and related to the biological and socioeconomic differences ethnicity-specific effects of CMS on brain age. between the Korean and UK populations. Previously, a higher frequency of CMS in Korean populations com- Conclusions pared to Europeans has been explained by the fact that In the present study, we highlight gender and ethnic Asians have higher visceral fat and lower subcutaneous differences in the effects of CMS on brain age. Further - fat than Europeans with the same BMI [54]. This might more, our findings suggest that different measures may increase the complication rate of CMS because visceral be needed to prevent accelerated brain aging by CMS fat has more deleterious effects on arteriosclerosis and in terms of gender and ethnic differences. In conclu - brain health than subcutaneous fat. In fact, Asians are sion, CMS exerted different effects on brain age accord - more likely to develop CMS-related complications such ing to the gender and ethnicity of the individuals. Our as coronary artery disease [55], stroke [56], dementia study shows that it is important to control for T2DM and [57–59], or mortality [60]. Another potential explanation hypertension to prevent brain aging. Since the effects of is that there were fewer differences in socioeconomic sta - T2DM and hypertension on brain age were the largest tus and years of education between males and females among Korean females, more careful treatment of these in the UK than in Korea. Further studies are needed to CMS components would be more effective to prevent or investigate the pathomechanism to explain gender differ - mitigate fast brain aging in Korean females. Therefore, ences according to ethnicity. ethnic- and gender-specific prevention strategies may be needed to protect against accelerated brain aging. Limitations Acknowledgements The strengths of our study include a large sample size The UK Biobank Resource was accessed through Application Number 25641. from two different cohorts, well-balanced clinical demo - We thank UK Biobank study participants and UK Biobank for enabling this resource. graphics between the two cohorts after propensity score matching, and a novel measurement of brain age that is Authors’ contributions sensitive to neurodegenerative changes in gray and white Concept and design: S.H. Kang, S.W. Seo, H. Kim. Acquisition of data: S.H. Kang, S.Y. Kim, Lee, Matloff, Zhao, H. Yoo, J.P. Kim, H.J. Kim, Jahanshad, Oh, Koh, Na, matter. However, our study had some limitations. First, Gallacher, S.W. Seo, H. Kim. Analysis or interpretation of data: S.H. Kang, Liu, owing to the cross-sectional study design, the causal S.W. Seo, H. Kim. Drafting of the manuscript: S.H. Kang. Intellectual content: or temporal relationship of the effects of CMS on brain S.H. Kang, M. Liu, Gottesman, S.W. Seo, H. Kim. Statistical analysis: S.H. Kang, M. Liu, H. Yoo. Obtained funding: S.H. Kang, S.W. Seo, H. Kim. Administrative, aging was not determined. In addition, the study did not technical, or material support: S.W. Seo, H. Kim. Supervision: S.W. Seo, H. Kim. have information on exposure time or changes in the The authors read and approved the final manuscript. status of risk factors. Longitudinal studies are needed to Funding identify whether there are dynamic differences from mid- This research was supported by a grant of the Korea Health Technology R&D adulthood to old age in the effects of risk factors on brain Project through the Korea Health Industry Development Institute (KHIDI), aging in elderly CU populations. Second, differences funded by the Ministry of Health & Welfare and Ministry of science and ICT, Republic of Korea (grant number: HU20C0111, HU22C0170); the National in subject selection methods between the two cohorts Research Foundation of Korea (NRF) grant funded by the Korea govern‑ may have confounded the ethnic differences. Third, the ment (MSIT ) (NRF‑2019R1A5A2027340, NRF‑2022R1F1A1063966); Institute presence of T2DM and hypertension was determined of Information & communications Technology Planning & Evaluation (IITP) grant funded by the Korea government (MSIT ) (No.2021–0‑02068, Artificial through the patient history of diagnosis or medications Intelligence Innovation Hub); Future Medicine 20*30 Project of the Samsung and not through clinical examinations including meas- Medical Center [#SMX1230081]; the “Korea National Institute of Health” urement of systolic blood pressure and fasting glucose. research project(2021‑ER1006‑02); Korea University Guro Hospital (KOREA RESEARCH‑DRIVEN HOSPITAL) and grant funded by Korea University Medicine Fourth, obesity was defined using BMI only rather than (K2210201); Basic Science Research Program through the National Research waist circumstance, which has relevance to central obe- Foundation of Korea (NRF) funded by the Ministry of Education (grant sity according to the International Diabetes Federation. number: 2022R1I1A1A01056956); the National Institutes of Health grants (P50NS035902, P01NS082330, R01NS046432, R01HD072074; P41EB015922; We also used different criteria for diagnosing obesity U54EB020406; U19AG024904; U01NS086090; 003585–00001); and HK funded according to ethnicity-specific BMI. This was, however, by Bright Focus Foundation Award (A2019052S). done to abide by a previous consensus on the definition Availability of data and materials of obesity according to ethnicity [25]. Finally, we did not Anonymized data for our analyses presented in this report are available upon consider the brain pathology markers of Alzheimer’s dis- request from the corresponding authors. ease, lacunes, micro-cortical infarcts, and white matter K ang et al. Alzheimer’s Research & Therapy (2023) 15:68 Page 9 of 10 modifiable risk factors for dementia, and cognitive performance. JAMA Declarations Netw Open. 2019;2: e1917257. 12. Choi YY, Lee JJ, Choi KY, Seo EH, Choo IH, Kim H, et al. The Aging slopes Ethics approval and consent to participate of brain structures vary by ethnicity and sex: evidence from a large Approval was obtained from the ethics committee of Samsung Medical magnetic resonance imaging dataset from a single scanner of cognitively Centre. The procedures used in this study adhere to the tenets of the Declara‑ healthy elderly people in Korea. Front Aging Neurosci. 2020;12:233. tion of Helsinki. Written informed consent was obtained from all patients and 13. Scharf EL, Graff‑Radford J, Przybelski SA, Lesnick TG, Mielke MM, Knop ‑ caregivers. man DS, et al. Cardiometabolic health and longitudinal progression of white matter hyperintensity: the Mayo Clinic Study of Aging. Stroke. Consent for publication 2019;50:3037–44. Not applicable. 14. Tamura Y, Shimoji K, Ishikawa J, Matsuo Y, Watanabe S, Takahashi H, et al. Subclinical atherosclerosis, vascular risk factors, and white matter altera‑ Competing interests tions in diffusion tensor imaging findings of older adults with cardio ‑ The authors declare no competing interests. metabolic diseases. Front Aging Neurosci. 2021;13: 712385. 15. Smith SM, Vidaurre D, Alfaro‑Almagro F, Nichols TE, Miller KL. Estimation of Author details brain age delta from brain imaging. Neuroimage. 2019;200:528–39. Departments of Neurology, Samsung Medical Center, Sungkyunkwan Univer‑ 16. Jonsson BA, Bjornsdottir G, Thorgeirsson TE, Ellingsen LM, Walters GB, sity School of Medicine, Seoul, South Korea. Department of Neurology, Korea Gudbjartsson DF, et al. Brain age prediction using deep learning uncovers University Guro Hospital, Korea University College of Medicine, Seoul, South associated sequence variants. Nat Commun. 2019;10:5409. Korea. School of Biomedical Engineering, Sun Yat‑Sen University, Shenzhen, 17. Cole JH, Leech R, Sharp DJ. Alzheimer’s Disease Neuroimaging Initiative China. Keck School of Medicine of University of Southern California, USC Prediction of brain age suggests accelerated atrophy after traumatic Steven Neuroimaging and Informatics Institute, Los Angeles, CA 90033, USA. 5 6 brain injury. Ann Neurol. 2015;77:571–81. Department of Psychiatry, University of Oxford, Oxford, UK. National I nstitute 18. Shokri‑Kojori E, Bennett IJ, Tomeldan ZA, Krawczyk DC, Rypma B. of Neurological Disorders and Stroke Intramural Research Program, National Estimates of brain age for gray matter and white matter in younger and Institutes of Health, Bethesda, MD, USA. Department of Digital Health, SAIHST, older adults: insights into human intelligence. Brain Res. 2021;1763: Sungkyunkwan University, Seoul, South Korea. Department of Health Sci‑ ences and Technology, SAIHST, Sungkyunkwan University, Seoul, South Korea. 19. Franke K, Gaser C. Ten years of brainAGE as a neuroimaging biomarker of Alzheimer’s Disease Convergence Research Center, Samsung Medical Center, brain aging: what insights have we gained? Front Neurol. 2019;10:789. Seoul, South Korea. Department of Intelligent Precision Healthcare Conver‑ 20. Cole JH, Ritchie SJ, Bastin ME, Valdés Hernández MC, Muñoz gence, Sungkyunkwan University, Suwon, South Korea. Maniega S, Royle N, et al. Brain age predicts mortality. Mol Psychiatry. 2018;23:1385–92. Received: 10 December 2022 Accepted: 20 March 2023 21. Liem F, Varoquaux G, Kynast J, Beyer F, KharabianMasouleh S, Huntenburg JM, et al. Predicting brain‑age from multimodal imaging data captures cognitive impairment. Neuroimage. 2017;148:179–88. 22 Franke K, Gaser C. Longitudinal changes in individual BrainAGE in healthy aging, mild cognitive impairment, and Alzheimer’s disease. GeroPsych. References 2012;25:235–45. 1. Wahl D, Solon‑Biet SM, Cogger VC, Fontana L, Simpson SJ, Le Couteur DG, 23. Lee JS, Shin HY, Kim HJ, Jang YK, Jung NY, Lee J, et al. Combined effects et al. Aging, lifestyle and dementia. Neurobiol Dis. 2019;130: 104481. of physical exercise and education on age‑related cortical thinning in 2. Fotenos AF, Snyder AZ, Girton LE, Morris JC, Buckner RL. Normative esti‑ cognitively normal individuals. Sci Rep. 2016;6:24284. mates of cross‑sectional and longitudinal brain volume decline in aging 24. Kang SH, Park YH, Lee D, Kim JP, Chin J, Ahn Y, et al. The cortical neuro‑ and AD. Neurology. 2005;64:1032–9. anatomy related to specific neuropsychological deficits in Alzheimer’s 3. Lee JS, Kim S, Yoo H, Park S, Jang YK, Kim HJ, et al. Trajectories of physi‑ continuum. Dement Neurocogn Disord. 2019;18:77–95. ological brain aging and related factors in people aged from 20 to over‑ 25. WHO Expert Consultation. Appropriate body‑mass index for Asian popu‑ 80. J Alzheimers Dis. 2018;65:1237–46. lations and its implications for policy and intervention strategies. Lancet. 4. Söderlund H, Nyberg L, Nilsson LG. Cerebral atrophy as predictor of 2004;363:157–63. cognitive function in old, community‑ dwelling individuals. Acta Neurol 26. Lewis JD, Evans AC, Tohka J. T1 white/gray contrast as a predictor of Scand. 2004;109:398–406. chronological age, and an index of cognitive performance. Neuroimage. 5. Leritz EC, Salat DH, Williams VJ, Schnyer DM, Rudolph JL, Lipsitz L, et al. 2018;173:341–50. Thickness of the human cerebral cortex is associated with metrics of 27. Cole JH, Poudel RPK, Tsagkrasoulis D, Caan MWA, Steves C, Spector cerebrovascular health in a normative sample of community dwelling TD, et al. Predicting brain age with deep learning from raw imag‑ older adults. Neuroimage. 2011;54:2659–71. ing data results in a reliable and heritable biomarker. Neuroimage. 6. Kotkowski E, Price LR, DeFronzo RA, Franklin CG, Salazar M, Garrett AS, 2017;163:115–24. et al. Metabolic syndrome predictors of brain gray matter volume in 28. Ning K, Zhao L, Matloff W, Sun F, Toga AW. Association of relative brain an age‑stratified community sample of 776 Mexican‑American adults: age with tobacco smoking, alcohol consumption, and genetic variants. results from the genetics of brain structure image archive. Front Aging Sci Rep. 2020;10:10. Neurosci. 2022;14: 999288. 29. Brown CJ, Moriarty KP, Miller SP, Booth BG, Zwicker JG, Grunau RE, et al. 7. Cosgrove KP, Mazure CM, Staley JK. Evolving knowledge of sex dif‑ Prediction of brain network age and factors of delayed maturation in very ferences in brain structure, function, and chemistry. Biol Psychiatry. preterm infants. In: Descoteaux M, Maier‑Hein L, Franz A, Jannin P, Collins 2007;62:847–55. DL, Duchesne S, editors., et al., Medical Image Computing and Computer‑ 8. Lüders E, Steinmetz H, Jäncke L. Brain size and grey matter volume in the Assisted Intervention ‑ MICCAI 2017: 20th International Conference, healthy human brain. NeuroReport. 2002;13:2371–4. Quebec City, QC, Canada, September 11–13, 2017, Proceedings, Part I. 9. Ritchie SJ, Cox SR, Shen X, Lombardo MV, Reus LM, Alloza C, et al. Sex Cham: Springer; 2017. p. 84–91. differences in the adult human brain: evidence from 5216 UK Biobank 30. Whitmer RA. Type 2 diabetes and risk of cognitive impairment and participants. Cereb Cortex. 2018;28:2959–75. dementia. Curr Neurol Neurosci Rep. 2007;7:373–80. 10. Kim SE, Lee JS, Woo S, Kim S, Kim HJ, Park S, et al. Sex‑specific relationship 31. Ninomiya T. Diabetes mellitus and dementia. Curr Diab Rep. 2014;14:487. of cardiometabolic syndrome with lower cortical thickness. Neurology. 32. Walker KA, Power MC, Gottesman RF. Defining the relationship between 2019;93:e1045–57. hypertension, cognitive decline, and dementia: a review. Curr Hypertens 11. Suzuki H, Venkataraman AV, Bai W, Guitton F, Guo Y, Dehghan A, Rep. 2017;19:24. et al. Associations of regional brain structural differences with aging, Kang et al. Alzheimer’s Research & Therapy (2023) 15:68 Page 10 of 10 33. Gąsecki D, Kwarciany M, Nyka W, Narkiewicz K. Hypertension, brain dam‑ 57. Park JE, Kim BS, Kim KW, Hahm BJ, Sohn JH, Suk HW, et al. Decline in the age and cognitive decline. Curr Hypertens Rep. 2013;15:547–58. incidence of all‑ cause and Alzheimer’s disease dementia: a 12‑ year‑later 34. Rocca WA, Mielke MM, Vemuri P, Miller VM. Sex and gender differences in rural cohort study in Korea. J Korean Med Sci. 2019;34: e293. the causes of dementia: a narrative review. Maturitas. 2014;79:196–201. 58. Niu H, Álvarez‑Álvarez I, Guillén‑ Grima F, Aguinaga‑ Ontoso I. Prevalence 35. Sevre K, Lefrandt JD, Nordby G, Os I, Mulder M, Gans RO, et al. Autonomic and incidence of Alzheimer’s disease in Europe: a meta‑analysis. Neurolo ‑ function in hypertensive and normotensive subjects: the importance of gia. 2017;32:523–32. gender. Hypertension. 2001;37:1351–6. 59. Jang J‑ W, Park JH, Kim S, Lee S‑H, Lee S‑H, Kim Y ‑ J. Prevalence and inci‑ 36. Palatini P, Mos L, Santonastaso M, Saladini F, Benetti E, Mormino P, et al. dence of dementia in South Korea: a nationwide analysis of the National Premenopausal women have increased risk of hypertensive target organ Health Insurance Service senior cohort. J Clin Neurol. 2021;17:249–56. damage compared with men of similar age. J Womens Health (Larchmt). 60. Wild SH, Fischbacher C, Brock A, Griffiths C, Bhopal R. Mortality from all 2011;20:1175–81. causes and circulatory disease by country of birth in England and Wales 37. Lim JG, Shapiro EP, Vaidya D, Najjar SS, Turner KL, Bacher AC, et al. Sex dif‑ 2001–2003. J Public Health (Oxf ). 2007;29:191–8. ferences in left ventricular function in older persons with mild hyperten‑ sion. Am Heart J. 2005;150:934–40. Publisher’s Note 38. Brinton RD, Yao J, Yin F, Mack WJ, Cadenas E. Perimenopause as a neuro‑ Springer Nature remains neutral with regard to jurisdictional claims in pub‑ logical transition state. Nat Rev Endocrinol. 2015;11:393–405. lished maps and institutional affiliations. 39. Scheyer O, Rahman A, Hristov H, Berkowitz C, Isaacson RS, Diaz Brinton R, et al. Female sex and Alzheime’’s risk: the menopause connection. J Prev Alzheimers Dis. 2018;5:225–30. 40. Chedraui P, Jaramillo W, Pérez‑López FR, Escobar GS, Morocho N, Hidalgo L. Pro‑inflammatory cytokine levels in postmenopausal women with the metabolic syndrome. Gynecol Endocrinol. 2011;27:685–91. 41. Pfeilschifter J, Köditz R, Pfohl M, Schatz H. Changes in proinflammatory cytokine activity after menopause. Endocr Rev. 2002;23:90–119. 42. Macova M, Armando I, Zhou J, Baiardi G, Tyurmin D, Larrayoz‑Roldan IM, et al. Estrogen reduces aldosterone, upregulates adrenal angiotensin II AT2 receptors and normalizes adrenomedullary Fra‑2 in ovariectomized rats. Neuroendocrinology. 2008;88:276–86. 43. Weber MT, Maki PM, McDermott MP. Cognition and mood in perimeno‑ pause: a systematic review and meta‑analysis. J Steroid Biochem Mol Biol. 2014;142:90–8. 44. Luine VN. Estradiol and cognitive function: past, present and future. Horm Behav. 2014;66:602–18. 45. Kilpi F, Soares ALG, Fraser A, Nelson SM, Sattar N, Fallon SJ, et al. Changes in six domains of cognitive function with reproductive and chronologi‑ cal ageing and sex hormones: a longitudinal study in 2411 UK mid‑life women. BMC Womens Health. 2020;20:177. 46. Yaffe K, Barnes D, Lindquist K, Cauley J, Simonsick EM, Penninx B, et al. Endogenous sex hormone levels and risk of cognitive decline in an older biracial cohort. Neurobiol Aging. 2007;28:171–8. 47. Yaffe K, Sawaya G, Lieberburg I, Grady D. Estrogen therapy in postmeno ‑ pausal women: effects on cognitive function and dementia. JAMA. 1998;279:688–95. 48. Emslie C, Hunt K, Macintyre S. How similar are the smoking and drinking habits of men and women in non‑manual jobs? Eur J Public Health. 2002;12:22–8. 49. Bray RM, Fairbank JA, Marsden ME. Stress and substance use among military women and men. Am J Drug Alcohol Abuse. 1999;25:239–56. 50. Thun M, Peto R, Boreham J, Lopez AD. Stages of the cigarette epidemic on entering its second century. Tob Control. 2012;21:96–101. 51. Cifkova R, Pitha J, Krajcoviechova A, Kralikova E. Is the impact of conven‑ tional risk factors the same in men and women? Plea for a more gender‑ specific approach. Int J Cardiol. 2019;286:214–9. 52. Brickman AM, Schupf N, Manly JJ, Luchsinger JA, Andrews H, Tang MX, et al. Brain morphology in older African Americans, Caribbean Hispanics, and whites from northern Manhattan. Arch Neurol. 2008;65:1053–61. 53. Zahodne LB, Manly JJ, Narkhede A, Griffith EY, DeCarli C, Schupf NS, et al. Re Read ady y to to submit y submit your our re researc search h ? Choose BMC and benefit fr ? Choose BMC and benefit from om: : Structural MRI predictors of late‑life cognition differ across African Ameri‑ cans, Hispanics, and Whites. Curr Alzheimer Res. 2015;12:632–9. fast, convenient online submission 54. Nazare JA, Smith JD, Borel AL, Haffner SM, Balkau B, Ross R, et al. Ethnic thorough peer review by experienced researchers in your field influences on the relations between abdominal subcutaneous and vis‑ ceral adiposity, liver fat, and cardiometabolic risk profile: the International rapid publication on acceptance Study of Prediction of Intra‑Abdominal Adiposity and Its Relationship support for research data, including large and complex data types With Cardiometabolic Risk/Intra‑Abdominal Adiposity. Am J Clin Nutr. • gold Open Access which fosters wider collaboration and increased citations 2012;96:714–26. 55. McKeigue PM, Miller GJ, Marmot MG. Coronary heart disease in south maximum visibility for your research: over 100M website views per year Asians overseas: a review. J Clin Epidemiol. 1989;42:597–609. 56. Eastwood SV, Tillin T, Chaturvedi N, Hughes AD. Ethnic differences in asso ‑ At BMC, research is always in progress. ciations between blood pressure and stroke in South Asian and European Learn more biomedcentral.com/submissions men. Hypertension. 2015;66:481–8. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Alzheimer s Research & Therapy Springer Journals

Loading next page...
 
/lp/springer-journals/different-effects-of-cardiometabolic-syndrome-on-brain-age-in-relation-Z910usM0Pa

References (64)

Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2023
eISSN
1758-9193
DOI
10.1186/s13195-023-01215-8
Publisher site
See Article on Publisher Site

Abstract

Background A growing body of evidence shows differences in the prevalence of cardiometabolic syndrome (CMS) and dementia based on gender and ethnicity. However, there is a paucity of information about ethnic‑ and gender ‑ specific CMS effects on brain age. We investigated the different effects of CMS on brain age by gender in Korean and British cognitively unimpaired (CU) populations. We also determined whether the gender‑specific difference in the effects of CMS on brain age changes depending on ethnicity. Methods These analyses used de‑identified, cross‑sectional data on CU populations from Korea and United Kingdom (UK) that underwent brain MRI. After propensity score matching to balance the age and gender between the Korean and UK populations, 5759 Korean individuals (3042 males and 2717 females) and 9903 individuals from the UK (4736 males and 5167 females) were included in this study. Brain age index (BAI), calculated by the difference between the predicted brain age by the algorithm and the chronological age, was considered as main outcome and presence of CMS, including type 2 diabetes mellitus ( T2DM), hypertension, obesity, and underweight was considered as a predic‑ tor. Gender (males and females) and ethnicity (Korean and UK) were considered as effect modifiers. Results The presence of T2DM and hypertension was associated with a higher BAI regardless of gender and ethnicity (p < 0.001), except for hypertension in Korean males (p = 0.309). Among Koreans, there were interaction effects of gen‑ der and the presence of T2DM (p for T2DM*gender = 0.035) and hypertension (p for hypertension*gender = 0.046) on BAI in Koreans, suggesting that T2DM and hypertension are each associated with a higher BAI in females than in males. In contrast, among individuals from the UK, there were no differences in the effects of T2DM (p for T2DM*gender = 0.098) and hypertension (p for hypertension*gender = 0.203) on BAI between males and females. Conclusions Our results highlight gender and ethnic differences as important factors in mediating the effects of CMS on brain age. Furthermore, these results suggest that ethnic‑ and gender ‑specific prevention strategies may be needed to protect against accelerated brain aging. Keywords Brain age, Cardiometabolic syndrome, Gender, Ethnicity Sung Hoon Kang and Mengting Liu contributed equally to this work. *Correspondence: Sang Won Seo sw72.seo@samsung.com Full list of author information is available at the end of the article © The Author(s) 2023. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. The Creative Commons Public Domain Dedication waiver (http:// creat iveco mmons. org/ publi cdoma in/ zero/1. 0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Kang et al. Alzheimer’s Research & Therapy (2023) 15:68 Page 2 of 10 by gender and ethnicity, we hypothesized that there Background might be differences in the effects of CMS on the BAI in Aging is an important risk factor for cognitive impair- relation to gender and ethnicity. ment and dementia [1]. As aging progresses, brain atro- phy also occurs at a mean volume reduction rate of 0.5% per year after the age of 40 [2, 3]. Age-related brain atro- phy is referred to as the brain age. Cardiometabolic syn- Methods drome (CMS), syndrome X, metabolic syndrome, and Study populations cardiometabolic dysfunction, composed of type 2 dia- We enrolled CU participants aged ≥ 45  years from the betes mellitus (T2DM), hypertension, and obesity, are Health Promotion Center of Samsung Medical Center critical modifiable risk factors for cognitive impairment. (Seoul, Korea) who underwent a comprehensive health There is also a growing body of evidence that CMS has screening exam from September 1, 2008, to October deleterious effects on brain atrophy [4] even in non- 31, 2019. A total of 8227 eligible candidates underwent demented population [5, 6]. Previous studies have sug- a full medical examination, which included cognitive gested that CMS may accelerate brain aging. assessment and 3.0-Tesla MRI, including high-resolution Several cross-sectional studies have shown differences T1-weighted MRI, as part of a standard screening for in brain volume between males and females in cognitively dementia. The medical examination procedure used for unimpaired (CU) populations [7–9]. Previously, changes the participants has been previously described [23]. We in brain age or atrophy were shown to occur differently excluded participants who had any of the following con- depending on gender [3, 10, 11]. Additionally, previous ditions: 728 participants with missing data on years of studies based on Hispanic or Korean populations sug- education or Mini-Mental State Examination (MMSE) gested that CMS-associated brain atrophy was more score [24]; 509 participants with significant cognitive extensive or prominent in females than in males [6, 10]. impairment defined by MMSE scores below the 16th per - However, considering the differences in the prevalence of centile in age-, gender-, and education-matched norms or CMS and dementia between Korean and European popu- through an interview conducted by a qualified neurolo - lations, it would be reasonable to hypothesize that there gist; 312 participants with severe cerebral white matter might be a difference in the gender-specific relation - hyperintensities (deep white matter ≥ 2.5  cm and caps ship between CMS and brain age between Koreans and or band ≥ 1.0 cm) or structural lesions such as territorial Europeans. infarction, lobar hemorrhage, brain tumor, and hydro- Previous studies have analyzed various morphologi- cephalus; 542 participants with missing information on cal features on brain magnetic resonance imaging (MRI), DM, hypertension, or body mass index (BMI); and 377 including cortical thickness [12], regional gray matter participants with unreliable analyses of cortical thickness volume [11], and white matter hyperintensity [13] and due to head motion, blurred MRI, inadequate registra- integrity [14], and investigated the impact of CMS on tion to a standardized stereotaxic space, misclassifica - brain structure in aging populations. Recently, various tion of tissue type, or inexact surface extraction. Finally, machine learning approaches have been developed to 5759 participants (3042 males and 2717 females) were begin accurate prediction of brain age using the afore- included in this study. mentioned brain imaging features [15–18] and provide Similar data for people of British ancestry was obtained a new metric called brain age index (BAI) to measure from the UK Biobank (UKB, http:// www. ukbio bank. how old the brain age is compared to the chorological ac. uk), a population-based prospective cohort study of age at MRI scan. The difference between the predicted approximately 500,000 people in the UK [https:// journ brain age using a deep learning-based algorithm and the al s . plo s . or g/ plo sm e dic i ne/ ar tic le? id= 10. 1371/ j our n chronological age is called the BAI, which explains how al. pmed. 10017 79, https:// www. nature. com/ artic les/ much older or younger an individual brain appears com-s41586- 018- 0579-z]. Of these participants, approximately pared to the current age. A positive BAI is a novel marker 40,000 attended an additional visit during which MRI of an older brain and has been shown to predict compro- brain imaging data was collected in addition to other mised brain health [19], earlier mortality [20], and cogni- health-related data [https:// www. nature. com/ artic les/ tive impairment [21, 22].s41467- 020- 15948-9]. We included non-Hispanic White In the present study, we investigated the different adults only in the present study. We excluded partici- effects of CMS on BAI with respect to the sex of CU pop - pants with a self-reported or hospital record-based his- ulations from Korea and United Kingdom (UK). Next, tory of dementia, Parkinson’s disease, or other central we determined whether CMS affects gender-specific BAI nervous system-related diseases. Finally, 9903 (4736 differently according to ethnicity. Considering that there males and 5167 females) UKB participants were included are differences in incidence of CMS and cortical atrophy after applying the inclusion/exclusion criteria and after K ang et al. Alzheimer’s Research & Therapy (2023) 15:68 Page 3 of 10 random selection of a smaller subset of participants for Field echo (TFE) MRI data using the following imaging brain imaging data processing. parameters: sagittal slice thickness, 1.0  mm with 50% The institutional review board of the Samsung Medical overlap; no gap; repetition time of 9.9  ms; echo time of Center approved this study and adhered to the principles 4.6 ms; flip angle of 8; and matrix size of 240 × 240 pixels of the Declaration of Helsinki. Written informed consent reconstructed to 480 × 480 over a field view of 240 mm. was obtained from all participants in the Health Promo- In the UKB populations, brain MRI scans were tion Center of Samsung Medical Center. Anonymous obtained at one of the three assessment sites using a 3.0 and deidentified data from the UKB was used for analy - Tesla Siemens Skyra MRI Scanner. Among the six brain sis and, therefore, the present study was exempted from imaging modalities acquired was a T1-weighted, sag- obtaining informed consent. ittal 3D magnetization prepared rapid gradient echo (MPRAGE) scan. The following imaging parameters were Measurement of cardiometabolic syndrome used in this T1-weighted acquisition: inversion time of For populations from the Health Promotion Center at 880 ms; repetition time of 2000 ms; 1 × 1 × 1 mm voxel the Samsung Medical Center, a health screening exam size; 208 × 256 × 256 matrix size; and SENSE factor (R) was conducted by a well-trained medical professional of 2.0 [Miller, 2016: https:// www. nature. com/ artic les/ nn. using standardized protocols. Baseline workup included 4393]. blood tests (complete blood cell count, liver/kidney/thy- roid function test, and tumor markers), urine analyses, Image processing and cortical surface extraction abdominal sonography, chest radiography, electrocar- T1-weighted MRI scans from the Health Promotion diogram, pulmonary function test, and gastroduoden- Center in Korea and the UKB were used to reconstruct oscopy. We classified each CMS component using the the inner and outer cortical boundaries using the CIVET following criteria: T2DM was defined as a diagnostic his - pipeline developed at the Montreal Neurological Insti- tory of T2DM or current use of any anti-diabetic medica- tute (http:// www. bic. mni. mcgill. ca/ Servi cesSo ftware/ tion; hypertension was defined as a diagnostic history of CIVET). Cortical morphology was quantitatively char- hypertension or current use of any antihypertensive med- acterized by measuring cortical thickness, sulcal depth, ication; obesity and underweight were defined using the and gray/white intensity ratio [26] on the cortical sur- cut-off for BMI calculated by weight (kilograms)/height face at 81,924 vertices (163,840 polygons). These features (meters) squared at the first visit. According to a previ - were further resampled to the surface template using the ous study [10], populations with BMI < 18.5  kg/m were transformation obtained in the surface registration to labeled as underweight, and those with BMI ≥ 27.5 kg/m allow for inter-subject comparisons. were labeled as obese. For populations from the UKB, the classification of Development of prediction model for relative brain age T2DM, hypertension, and obesity was determined based As illustrated in Fig.  1, we did not use topology-varying on a combination of a touchscreen-based questionnaire, surfaces because of the nature of the graph convolutional a verbal interview, and linked hospital records. Specifi - networks (GCN) model used in this study. Rather, we cally, T2DM was defined as either self-reported T2DM, considered the cortical morphological changes that occur a doctor’s diagnosis of T2DM (Data Field 2443), patients in relation to brain size and gyrification using cortical who were taking insulin (Data Fields 6177, 6153), or a thickness, volume, and sulcal depth. The GCN employed hospital data-linked record of an individual with a diag- in our study requires identical graph/mesh structures for nosis of T2DM. Hypertension was defined as either all individual inputs, whereas the features of the nodes/ self-reported hypertension (Data-Field 20,002) or a hos- vertices can vary. Another advantage of the topology- pital data-linked record of having a primary or secondary kept surface model is that surface nodes are registered diagnosis of hypertension (Data-Fields 41,202, 41,204, across all individuals such that anatomical information is 41,203, and 41,205). BMI was calculated using weight and shared. height measurements in the same way as the Samsung Medical Center data. Populations with a BMI < 18.5  kg/ Brain age index m were categorized as underweight, whereas those with After calculating the predicted brain age for each subject, BMI ≥ 35 kg/m were categorized as obese [25]. we further calculated a metric that reflected the subject’s relative brain health status, called the BAI. BAI was ini- Acquisition of brain MRI tially measured by subtracting the true brain age from the All Korean populations underwent a 3D volumetric brain predicted brain age [27]. Due to regression dilution [28], MRI scan. An Achieva 3.0-Tesla MRI scanner (Philips, however, it is also possible that regression models bias the Best, the Netherlands) was used to acquire 3D T1 Turbo predicted brain age toward the mean, underestimating Kang et al. Alzheimer’s Research & Therapy (2023) 15:68 Page 4 of 10 Fig. 1 The graph‑based convolutional network for brain age prediction the age of older subjects and overestimating the age of populations, we performed linear regression analyses younger subjects [29]. When deriving the BAI, this bias by adding each two-way interaction term (the pres- must be corrected using a strategy introduced in other ence of each CMS component*gender) to covariates in studies [15, 28]. We defined the new BAI as the differ - Korean and UK populations after controlling for the ence between the individual BAI and the expected BAI other CMS components. To assess whether the asso- (measurement fitted over the entire sample set using the ciation between the presence of each CMS component regression model and leave-one-out cross-validation). and brain age might differ by gender and ethnicity, we The BAI was corrected such that the BAIs of the whole performed linear regression analyses with the addition dataset analyzed became unbiased across all age ranges. of each three-way interaction term (the presence of each CMS component*gender*ethnicity) to covariates after controlling for the other CMS components. Propensity score matching False discovery rate (FDR) correction was conducted Propensity score matching was performed to minimize for all statistical analyses to control for p-values, and the differences in the demographics and cardiometa - q-values were obtained after FDR correction. All bolic factors between the UK and KOR participants. The reported p-values and q-values were two-sided and the propensity score was obtained using multivariable logis- significance level was set at 0.05. All analyses were per - tic regression based on age, gender, T2DM, hyperten- formed using R version 4.3.0 (Institute for Statistics and sion, and obesity. A total of 5541 KOR participants were Mathematics, Vienna, Austria; www.R- proje ct. org). matched with 9903 UK participants based on propensity scores using the 1:2 nearest-neighbor matching algo- rithm with caliper of 0.1. A good balance was achieved Results between the KOR and UK participants, with all standard- Demographics of cognitively unimpaired populations ized mean differences (age, gender, T2DM, hypertension, in the UK and Korea and obesity) below 0.1 after matching. After propensity score matching, the demographic characteristics of the two ethnic datasets were similar Statistical analysis (Table  1). Among the 5541 Korean populations, there Independent t-tests and chi-squared tests were used were 2599 (46.9%) females and 2942 (53.1%) males. to compare continuous and categorical variables, Among the 9903 UK populations, there were 5167 respectively. (52.2%) females and 4736 (47.8%) males. There were To explore the association between the presence of some differences in mean age (64.0 and 63.6  years, each CMS component and brain age in females and p < 0.001), female ratio (46.9 and 52.2%, p < 0.001), males among the Korean and UK populations, we per- and the presence of T2DM (17.3 and 9.8%, p < 0.001), formed a linear regression analysis with the presence hypertension (42.7% and 40.6%, p = 0.011), obesity of T2DM, hypertension, obesity, and underweight as (10.7 and 7.9%, p < 0.001), and underweight (1.8 and covariates. To assess whether the association between 0.6%, p < 0.001) between Koreans and participants from the presence of each CMS component and brain the UK. age might differ by gender in the Korean and the UK K ang et al. Alzheimer’s Research & Therapy (2023) 15:68 Page 5 of 10 Table 1 Demographics of populations from UK Biobank and Health Promotion Center in Korea Variables Korea UK p-value Females Males (n = 2942) Total (n = 5541) Females Males (n = 4736) Total (n = 9903) (n = 2599) (n = 5167) Age (years) 63.2 ± 6.9 64.7 ± 6.5 64.0 ± 6.7 63.4 ± 7.1 63.8 ± 7.4 63.6 ± 7.2 0.007 Hypertension (n, 965 (37.1%) 1402 (47.7%) 2367 (42.7%) 1813 (35.1%) 2208 (46.6%) 4021 (40.6%) 0.1 %) T2DM (n, %) 273 (10.5%) 683 (23.2%) 956 (17.3%) 365 (7.1%) 602 (12.7%) 967 (9.8%) < 0.001 2 a BMI (kg/m ) 23.5 ± 2.9 24.5 ± 2.6 24.0 ± 2.8 26.8 ± 5.2 27.7 ± 4.4 27.2 ± 4.9 < 0.001 Obesity (n, %) 227 (8.7%) 336 (12.4%) 593 (10.7%) 450 (8.7%) 336 (7.1%) 786 (7.9%) < 0.001 Underweight (n, %) 70 (2.7%) 32 (1.1%) 102 (1.8%) 49 (0.9%) 7 (0.1%) 56 (0.6%) < 0.001 Propensity score matching was performed to balance the age and gender between the Korean and UK populations, and 9903 out of 17,791 populations in the UK and 5541 out of 5759 populations in Korea were selected for the present study Abbreviations: BMI, body mass index; T2DM, type2 diabetes mellitus; UK, United Kingdom Distribution of age and BMI was compared between the populations of UK and Korea using independent t tests Distribution of education level and presence of hypertension, T2DM, obesity, and underweight between the populations of UK and Korea were tested using the chi-squared test Eec ff ts of cardiometabolic syndrome components on brain Interactive effects of cardiometabolic syndrome age index components on brain age index in relation to gender As shown in Fig.  2, DM was associated with increased and ethnicity BAI for all participants, regardless of gender and eth- We also investigated the interaction of the presence of nicity (q < 0.001 in the four groups) (Table  2). Hyper- each CMS component and gender with BAI in Koreans tension was associated with a significantly higher BAI and participants from the UK. Among Koreans, there for all participants (q < 0.001), except for Korean males were interactions between T2DM and gender with BAI (q = 0.309). Obesity significantly increased the BAI for (q = 0.035) and between hypertension and gender with UK males (q = 0.004). Being underweight increased the BAI (q = 0.046), suggesting that the effects of T2DM and BAI significantly only for UK females (q = 0.002). hypertension on BAI were more prominent in females Fig. 2 BAI distribution between groups regarding gender and ethnicity for healthy participants and participants with different CMS. BAI = 0 indicates that the chronological age is the same as the predicted brain age, with higher values indicating an older‑appearing brain than chronological age. Asterisk symbol ( ) indicates the following: q‑ values, FDR‑ corrected p‑ values, are lower than 0.05. BAI, brain age index; Kor, Korea; UK, United Kingdom; CMS, cardiometabolic syndrome Kang et al. Alzheimer’s Research & Therapy (2023) 15:68 Page 6 of 10 Table 2 Brain age index in controls and four CMS component groups Ethnicity/gender CMS Age BAI β (SE) q-value Korean females Control 61.6 ± 8.1 − 0.67 ± 3.57 T2DM 66.9 ± 9.5 1.51 ± 4.44 1.73 (0.25) < 0.001 Hypertension 66.0 ± 8.0 0.39 ± 3.97 0.80 (0.20) < 0.001 Obesity 64.8 ± 8.5 0.31 ± 3.69 0.26 (0.94) 0.347 Underweight 62.8 ± 7.6 0.30 ± 4.34 0.77 (0.48) 0.145 Korean males Control 64.1 ± 8.2 0.70 ± 3.55 T2DM 65.5 ± 8.6 1.82 ± 4.05 0.88 (0.19) < 0.001 Hypertension 65.3 ± 8.4 1.29 ± 4.13 0.19 (0.18) 0.309 Obesity 63.4 ± 7.7 1.52 ± 4.03 0.37 (0.23) 0.144 Underweight 67.5 ± 10.7 2.17 ± 3.63 1.24 (0.72) 0.171 UK females Control 62.0 ± 7.2 − 0.17 ± 3.50 T2DM 64.4 ± 7.4 1.22 ± 3.87 1.08 (0.20) < 0.001 Hypertension 65.2 ± 6.6 0.45 ± 3.79 0.49 (0.13) < 0.001 Obesity 61.5 ± 6.7 0.53 ± 3.54 0.32 (0.19) 0.088 Underweight 63.8 ± 7.0 1.60 ± 3.87 1.67 (0.53) 0.002 UK males Control 61.5 ± 7.5 0.25 ± 3.66 T2DM 65.8 ± 6.8 2.42 ± 3.90 1.55 (0.17) < 0.001 Hypertension 65.6 ± 6.8 1.35 ± 3.86 0.73 (0.13) < 0.001 Obesity 62.9 ± 7.2 1.72 ± 3.63 0.65 (0.22) 0.004 Underweight 68.0 ± 6.3 − 1.21 ± 2.28 − 1.69 (1.42) 0.235 Values of age and BAI are presented as mean ± standard deviation Abbreviations: BAI, brain age index; CMS, cardiometabolic syndrome; T2DM, type2 diabetes mellitus; UK, United Kingdom q-values, FDR-corrected p-values, were obtained using a linear regression analysis with the presence of hypertension, T2DM, obesity, and underweight as covariates in each group (Korean females, Korean males, UK females and UK males) than in males (Table  3, Fig.  3). Among British partici- Table 3 Interaction effect on the difference in BAI between participants with each CMS and those with control pants, however, there were no interactions of any CMSs and gender with BAI (q range 0.098 to 0.203, Table  3, Ethnicity Two-way interaction (each CMS β (SE) q-value Fig.  3). In fact, there were interactions between gender component*gender) and ethnicity for T2DM (q = 0.004) and hypertension Korea T2DM*gender 0.84 (0.32) 0.035 (q = 0.011, Table 3, Fig. 3). Hypertension*gender 0.61 (0.27) 0.046 Obesity*gender − 0.11 (0.36) 0.770 Underweight*gender − 0.82 (0.85) 0.450 Discussion UK T2DM*gender − 0.52 (0.26) 0.098 In the present study, we systematically investigated the Hypertension*gender − 0.26 (0.18) 0.203 different effects of CMS on BAI in relation to gender Obesity*gender − 0.37 (0.28) 0.191 and ethnic differences in a large sample of Korean and Underweight*gender 3.51 (1.50) 0.076 ¥ UK CU populations. Our major findings are as follows: Three-way β (SE) q-value interaction(each CMS first, among Koreans, the effects of DM and hypertension component*gender*ethnicity) on BAI were higher in females than in males. This indi - Both T2DM*gender*ethnicity 1.36 (0.41) 0.004 cated interaction effects of gender and the presence of Hypertension*gender*ethnicity 0.89 (0.32) 0.011 T2DM and hypertension on BAI in Korean population. Abbreviations: BAI, brain age index; CMS, cardiometabolic syndrome; T2DM, Second, among the UK population, there were no differ - type2 diabetes mellitus; UK, United Kingdom ences in the effects of T2DM and hypertension on BAI q-values, FDR-corrected p-values, were obtained using linear regression between males and females. Overall, there was evidence analyses with adding each two-way interaction term (the presence of each CMS component*gender) to covariates in Korean and UK populations after that ethnicity modified the gender-specific relationship controlling for the other CMS components of T2DM and hypertension with BAI. Taken together, q-values, FDR-corrected p-values, were obtained using linear regression our findings suggest that CMS exerts different effects on analyses with additionally adding each three-way interaction term (the presence brain age depending on gender and ethnicity. Therefore, of each CMS component*gender*ethnicity) to covariates after controlling for the other CMS components K ang et al. Alzheimer’s Research & Therapy (2023) 15:68 Page 7 of 10 Fig. 3 Ethnic‑ and gender ‑specific difference in BAI between participants with and without T2DM and HTN. Values depicted in the bar plot represent the mean of BAI, and values depicted in the error bar represent the standard error of mean. BAI = 0 indicates that the chronological age is the same as the predicted brain age, with higher values indicating an older‑appearing brain than chronological age. Asterisk ( ) symbol indicates the following: q‑ values, FDR‑ corrected p‑ values, were obtained using linear regression analyses with adding each two‑ way interaction term (the presence of each CMS component*gender) to covariates in Korean and UK populations after controlling for the other CMS components. Yen ( ) symbol indicates the following: q‑ values, FDR‑ corrected p‑ values, were obtained using linear regression analyses with additionally adding each three‑ way interaction term (the presence of each CMS component*gender*ethnicity) to covariates after controlling for the other CMS components. BAI, brain age index; Kor, Korea; UK, United Kingdom; T2DM, type 2 diabetes mellitus; HTN, hypertension ethnic- and gender-specific prevention strategies may be with autonomic dysfunction in females than in males necessary to protect against accelerated brain aging. [35]. Similar associations are witnessed in the cases of We found that the presence of T2DM and hypertension microalbuminuria [36] and reduction of heart function was associated with a higher BAI regardless of gender [37]. In particular, females uniquely experience meno- and ethnicity, except for hypertension in Korean males. pause transition, which might accelerate cardiometabolic T2DM and hypertension are well-known risk factors for syndrome, brain aging, or cognitive impairment via sev- brain atrophy, which is an important indicator of brain eral mechanisms including changes in the availability of age [30]. T2DM may have deleterious effects on the brain estrogen [38], estrogen receptor activity, and/or estro- via various mechanisms, including cerebrovascular com- gen-regulated neural networks [39]. Specifically, estrogen plications, glucose toxicity due to insulin resistance, and deficiency in postmenopausal females leads to inflamma - chronic inflammation [31]. Similarly, the positive asso - tory process and vasoconstriction via the dysfunction of ciation between hypertension and BAI may be due to the renin-angiotensin system [40–42]. In fact, a growing several possible mechanisms including cerebral hypoper- body of evidence shows that menopause has a deleteri- fusion, micro- and macrovascular damage in white mat- ous impact on cognitive function, which may contribute ter, and cerebral microinfarcts [32, 33]. to the higher prevalence of dementia in females than in Our first major finding was that Korean females suf - males [43–47]. Additionally, several studies have shown fered more deleterious effects of T2DM and hyperten - that females tend to maintain lifestyles that are more sion on brain age than Korean males. Although the favorable for brain health, with overall lower drinking underlying mechanisms for the gender-specific effects and smoking rates [48–51]. Therefore, our findings might of T2DM and hypertension on brain age are not fully be also related to differences in stress, alcohol consump - understood, our findings might be related to the com - tion, smoking, and dietary habits according to gender. plex effects of biological and socioeconomic differences Our second major finding was that there were interac - [34]. Previous studies have suggested that hypertension tive effects of the presence of T2DM and hypertension, exerts worse effects on multiple organs in females than gender, and ethnicity on BAI. That is, unlike Koreans, in males. This was attributed to differences in sex hor - there were no differences in the effects of T2DM and mones. There are stronger associations of hypertension hypertension on BAI between males and females in the Kang et al. Alzheimer’s Research & Therapy (2023) 15:68 Page 8 of 10 UK population. A few studies have found that brain age hyperintensities, which can also be associated with brain differs depending on ethnicity [52, 53]. However, gen- age. Further studies are needed to identify the effects of der- and ethnicity-specific differences in the effects of CMS on brain aging in relation to the pathophysiologi- T2DM and hypertension on brain age have not been cal processes. Despite the aforementioned limitations, extensively investigated. These differences might be our study is the first report to compare the gender- and related to the biological and socioeconomic differences ethnicity-specific effects of CMS on brain age. between the Korean and UK populations. Previously, a higher frequency of CMS in Korean populations com- Conclusions pared to Europeans has been explained by the fact that In the present study, we highlight gender and ethnic Asians have higher visceral fat and lower subcutaneous differences in the effects of CMS on brain age. Further - fat than Europeans with the same BMI [54]. This might more, our findings suggest that different measures may increase the complication rate of CMS because visceral be needed to prevent accelerated brain aging by CMS fat has more deleterious effects on arteriosclerosis and in terms of gender and ethnic differences. In conclu - brain health than subcutaneous fat. In fact, Asians are sion, CMS exerted different effects on brain age accord - more likely to develop CMS-related complications such ing to the gender and ethnicity of the individuals. Our as coronary artery disease [55], stroke [56], dementia study shows that it is important to control for T2DM and [57–59], or mortality [60]. Another potential explanation hypertension to prevent brain aging. Since the effects of is that there were fewer differences in socioeconomic sta - T2DM and hypertension on brain age were the largest tus and years of education between males and females among Korean females, more careful treatment of these in the UK than in Korea. Further studies are needed to CMS components would be more effective to prevent or investigate the pathomechanism to explain gender differ - mitigate fast brain aging in Korean females. Therefore, ences according to ethnicity. ethnic- and gender-specific prevention strategies may be needed to protect against accelerated brain aging. Limitations Acknowledgements The strengths of our study include a large sample size The UK Biobank Resource was accessed through Application Number 25641. from two different cohorts, well-balanced clinical demo - We thank UK Biobank study participants and UK Biobank for enabling this resource. graphics between the two cohorts after propensity score matching, and a novel measurement of brain age that is Authors’ contributions sensitive to neurodegenerative changes in gray and white Concept and design: S.H. Kang, S.W. Seo, H. Kim. Acquisition of data: S.H. Kang, S.Y. Kim, Lee, Matloff, Zhao, H. Yoo, J.P. Kim, H.J. Kim, Jahanshad, Oh, Koh, Na, matter. However, our study had some limitations. First, Gallacher, S.W. Seo, H. Kim. Analysis or interpretation of data: S.H. Kang, Liu, owing to the cross-sectional study design, the causal S.W. Seo, H. Kim. Drafting of the manuscript: S.H. Kang. Intellectual content: or temporal relationship of the effects of CMS on brain S.H. Kang, M. Liu, Gottesman, S.W. Seo, H. Kim. Statistical analysis: S.H. Kang, M. Liu, H. Yoo. Obtained funding: S.H. Kang, S.W. Seo, H. Kim. Administrative, aging was not determined. In addition, the study did not technical, or material support: S.W. Seo, H. Kim. Supervision: S.W. Seo, H. Kim. have information on exposure time or changes in the The authors read and approved the final manuscript. status of risk factors. Longitudinal studies are needed to Funding identify whether there are dynamic differences from mid- This research was supported by a grant of the Korea Health Technology R&D adulthood to old age in the effects of risk factors on brain Project through the Korea Health Industry Development Institute (KHIDI), aging in elderly CU populations. Second, differences funded by the Ministry of Health & Welfare and Ministry of science and ICT, Republic of Korea (grant number: HU20C0111, HU22C0170); the National in subject selection methods between the two cohorts Research Foundation of Korea (NRF) grant funded by the Korea govern‑ may have confounded the ethnic differences. Third, the ment (MSIT ) (NRF‑2019R1A5A2027340, NRF‑2022R1F1A1063966); Institute presence of T2DM and hypertension was determined of Information & communications Technology Planning & Evaluation (IITP) grant funded by the Korea government (MSIT ) (No.2021–0‑02068, Artificial through the patient history of diagnosis or medications Intelligence Innovation Hub); Future Medicine 20*30 Project of the Samsung and not through clinical examinations including meas- Medical Center [#SMX1230081]; the “Korea National Institute of Health” urement of systolic blood pressure and fasting glucose. research project(2021‑ER1006‑02); Korea University Guro Hospital (KOREA RESEARCH‑DRIVEN HOSPITAL) and grant funded by Korea University Medicine Fourth, obesity was defined using BMI only rather than (K2210201); Basic Science Research Program through the National Research waist circumstance, which has relevance to central obe- Foundation of Korea (NRF) funded by the Ministry of Education (grant sity according to the International Diabetes Federation. number: 2022R1I1A1A01056956); the National Institutes of Health grants (P50NS035902, P01NS082330, R01NS046432, R01HD072074; P41EB015922; We also used different criteria for diagnosing obesity U54EB020406; U19AG024904; U01NS086090; 003585–00001); and HK funded according to ethnicity-specific BMI. This was, however, by Bright Focus Foundation Award (A2019052S). done to abide by a previous consensus on the definition Availability of data and materials of obesity according to ethnicity [25]. Finally, we did not Anonymized data for our analyses presented in this report are available upon consider the brain pathology markers of Alzheimer’s dis- request from the corresponding authors. ease, lacunes, micro-cortical infarcts, and white matter K ang et al. Alzheimer’s Research & Therapy (2023) 15:68 Page 9 of 10 modifiable risk factors for dementia, and cognitive performance. JAMA Declarations Netw Open. 2019;2: e1917257. 12. Choi YY, Lee JJ, Choi KY, Seo EH, Choo IH, Kim H, et al. The Aging slopes Ethics approval and consent to participate of brain structures vary by ethnicity and sex: evidence from a large Approval was obtained from the ethics committee of Samsung Medical magnetic resonance imaging dataset from a single scanner of cognitively Centre. The procedures used in this study adhere to the tenets of the Declara‑ healthy elderly people in Korea. Front Aging Neurosci. 2020;12:233. tion of Helsinki. Written informed consent was obtained from all patients and 13. Scharf EL, Graff‑Radford J, Przybelski SA, Lesnick TG, Mielke MM, Knop ‑ caregivers. man DS, et al. Cardiometabolic health and longitudinal progression of white matter hyperintensity: the Mayo Clinic Study of Aging. Stroke. Consent for publication 2019;50:3037–44. Not applicable. 14. Tamura Y, Shimoji K, Ishikawa J, Matsuo Y, Watanabe S, Takahashi H, et al. Subclinical atherosclerosis, vascular risk factors, and white matter altera‑ Competing interests tions in diffusion tensor imaging findings of older adults with cardio ‑ The authors declare no competing interests. metabolic diseases. Front Aging Neurosci. 2021;13: 712385. 15. Smith SM, Vidaurre D, Alfaro‑Almagro F, Nichols TE, Miller KL. Estimation of Author details brain age delta from brain imaging. Neuroimage. 2019;200:528–39. Departments of Neurology, Samsung Medical Center, Sungkyunkwan Univer‑ 16. Jonsson BA, Bjornsdottir G, Thorgeirsson TE, Ellingsen LM, Walters GB, sity School of Medicine, Seoul, South Korea. Department of Neurology, Korea Gudbjartsson DF, et al. Brain age prediction using deep learning uncovers University Guro Hospital, Korea University College of Medicine, Seoul, South associated sequence variants. Nat Commun. 2019;10:5409. Korea. School of Biomedical Engineering, Sun Yat‑Sen University, Shenzhen, 17. Cole JH, Leech R, Sharp DJ. Alzheimer’s Disease Neuroimaging Initiative China. Keck School of Medicine of University of Southern California, USC Prediction of brain age suggests accelerated atrophy after traumatic Steven Neuroimaging and Informatics Institute, Los Angeles, CA 90033, USA. 5 6 brain injury. Ann Neurol. 2015;77:571–81. Department of Psychiatry, University of Oxford, Oxford, UK. National I nstitute 18. Shokri‑Kojori E, Bennett IJ, Tomeldan ZA, Krawczyk DC, Rypma B. of Neurological Disorders and Stroke Intramural Research Program, National Estimates of brain age for gray matter and white matter in younger and Institutes of Health, Bethesda, MD, USA. Department of Digital Health, SAIHST, older adults: insights into human intelligence. Brain Res. 2021;1763: Sungkyunkwan University, Seoul, South Korea. Department of Health Sci‑ ences and Technology, SAIHST, Sungkyunkwan University, Seoul, South Korea. 19. Franke K, Gaser C. Ten years of brainAGE as a neuroimaging biomarker of Alzheimer’s Disease Convergence Research Center, Samsung Medical Center, brain aging: what insights have we gained? Front Neurol. 2019;10:789. Seoul, South Korea. Department of Intelligent Precision Healthcare Conver‑ 20. Cole JH, Ritchie SJ, Bastin ME, Valdés Hernández MC, Muñoz gence, Sungkyunkwan University, Suwon, South Korea. Maniega S, Royle N, et al. Brain age predicts mortality. Mol Psychiatry. 2018;23:1385–92. Received: 10 December 2022 Accepted: 20 March 2023 21. Liem F, Varoquaux G, Kynast J, Beyer F, KharabianMasouleh S, Huntenburg JM, et al. Predicting brain‑age from multimodal imaging data captures cognitive impairment. Neuroimage. 2017;148:179–88. 22 Franke K, Gaser C. Longitudinal changes in individual BrainAGE in healthy aging, mild cognitive impairment, and Alzheimer’s disease. GeroPsych. References 2012;25:235–45. 1. Wahl D, Solon‑Biet SM, Cogger VC, Fontana L, Simpson SJ, Le Couteur DG, 23. Lee JS, Shin HY, Kim HJ, Jang YK, Jung NY, Lee J, et al. Combined effects et al. Aging, lifestyle and dementia. Neurobiol Dis. 2019;130: 104481. of physical exercise and education on age‑related cortical thinning in 2. Fotenos AF, Snyder AZ, Girton LE, Morris JC, Buckner RL. Normative esti‑ cognitively normal individuals. Sci Rep. 2016;6:24284. mates of cross‑sectional and longitudinal brain volume decline in aging 24. Kang SH, Park YH, Lee D, Kim JP, Chin J, Ahn Y, et al. The cortical neuro‑ and AD. Neurology. 2005;64:1032–9. anatomy related to specific neuropsychological deficits in Alzheimer’s 3. Lee JS, Kim S, Yoo H, Park S, Jang YK, Kim HJ, et al. Trajectories of physi‑ continuum. Dement Neurocogn Disord. 2019;18:77–95. ological brain aging and related factors in people aged from 20 to over‑ 25. WHO Expert Consultation. Appropriate body‑mass index for Asian popu‑ 80. J Alzheimers Dis. 2018;65:1237–46. lations and its implications for policy and intervention strategies. Lancet. 4. Söderlund H, Nyberg L, Nilsson LG. Cerebral atrophy as predictor of 2004;363:157–63. cognitive function in old, community‑ dwelling individuals. Acta Neurol 26. Lewis JD, Evans AC, Tohka J. T1 white/gray contrast as a predictor of Scand. 2004;109:398–406. chronological age, and an index of cognitive performance. Neuroimage. 5. Leritz EC, Salat DH, Williams VJ, Schnyer DM, Rudolph JL, Lipsitz L, et al. 2018;173:341–50. Thickness of the human cerebral cortex is associated with metrics of 27. Cole JH, Poudel RPK, Tsagkrasoulis D, Caan MWA, Steves C, Spector cerebrovascular health in a normative sample of community dwelling TD, et al. Predicting brain age with deep learning from raw imag‑ older adults. Neuroimage. 2011;54:2659–71. ing data results in a reliable and heritable biomarker. Neuroimage. 6. Kotkowski E, Price LR, DeFronzo RA, Franklin CG, Salazar M, Garrett AS, 2017;163:115–24. et al. Metabolic syndrome predictors of brain gray matter volume in 28. Ning K, Zhao L, Matloff W, Sun F, Toga AW. Association of relative brain an age‑stratified community sample of 776 Mexican‑American adults: age with tobacco smoking, alcohol consumption, and genetic variants. results from the genetics of brain structure image archive. Front Aging Sci Rep. 2020;10:10. Neurosci. 2022;14: 999288. 29. Brown CJ, Moriarty KP, Miller SP, Booth BG, Zwicker JG, Grunau RE, et al. 7. Cosgrove KP, Mazure CM, Staley JK. Evolving knowledge of sex dif‑ Prediction of brain network age and factors of delayed maturation in very ferences in brain structure, function, and chemistry. Biol Psychiatry. preterm infants. In: Descoteaux M, Maier‑Hein L, Franz A, Jannin P, Collins 2007;62:847–55. DL, Duchesne S, editors., et al., Medical Image Computing and Computer‑ 8. Lüders E, Steinmetz H, Jäncke L. Brain size and grey matter volume in the Assisted Intervention ‑ MICCAI 2017: 20th International Conference, healthy human brain. NeuroReport. 2002;13:2371–4. Quebec City, QC, Canada, September 11–13, 2017, Proceedings, Part I. 9. Ritchie SJ, Cox SR, Shen X, Lombardo MV, Reus LM, Alloza C, et al. Sex Cham: Springer; 2017. p. 84–91. differences in the adult human brain: evidence from 5216 UK Biobank 30. Whitmer RA. Type 2 diabetes and risk of cognitive impairment and participants. Cereb Cortex. 2018;28:2959–75. dementia. Curr Neurol Neurosci Rep. 2007;7:373–80. 10. Kim SE, Lee JS, Woo S, Kim S, Kim HJ, Park S, et al. Sex‑specific relationship 31. Ninomiya T. Diabetes mellitus and dementia. Curr Diab Rep. 2014;14:487. of cardiometabolic syndrome with lower cortical thickness. Neurology. 32. Walker KA, Power MC, Gottesman RF. Defining the relationship between 2019;93:e1045–57. hypertension, cognitive decline, and dementia: a review. Curr Hypertens 11. Suzuki H, Venkataraman AV, Bai W, Guitton F, Guo Y, Dehghan A, Rep. 2017;19:24. et al. Associations of regional brain structural differences with aging, Kang et al. Alzheimer’s Research & Therapy (2023) 15:68 Page 10 of 10 33. Gąsecki D, Kwarciany M, Nyka W, Narkiewicz K. Hypertension, brain dam‑ 57. Park JE, Kim BS, Kim KW, Hahm BJ, Sohn JH, Suk HW, et al. Decline in the age and cognitive decline. Curr Hypertens Rep. 2013;15:547–58. incidence of all‑ cause and Alzheimer’s disease dementia: a 12‑ year‑later 34. Rocca WA, Mielke MM, Vemuri P, Miller VM. Sex and gender differences in rural cohort study in Korea. J Korean Med Sci. 2019;34: e293. the causes of dementia: a narrative review. Maturitas. 2014;79:196–201. 58. Niu H, Álvarez‑Álvarez I, Guillén‑ Grima F, Aguinaga‑ Ontoso I. Prevalence 35. Sevre K, Lefrandt JD, Nordby G, Os I, Mulder M, Gans RO, et al. Autonomic and incidence of Alzheimer’s disease in Europe: a meta‑analysis. Neurolo ‑ function in hypertensive and normotensive subjects: the importance of gia. 2017;32:523–32. gender. Hypertension. 2001;37:1351–6. 59. Jang J‑ W, Park JH, Kim S, Lee S‑H, Lee S‑H, Kim Y ‑ J. Prevalence and inci‑ 36. Palatini P, Mos L, Santonastaso M, Saladini F, Benetti E, Mormino P, et al. dence of dementia in South Korea: a nationwide analysis of the National Premenopausal women have increased risk of hypertensive target organ Health Insurance Service senior cohort. J Clin Neurol. 2021;17:249–56. damage compared with men of similar age. J Womens Health (Larchmt). 60. Wild SH, Fischbacher C, Brock A, Griffiths C, Bhopal R. Mortality from all 2011;20:1175–81. causes and circulatory disease by country of birth in England and Wales 37. Lim JG, Shapiro EP, Vaidya D, Najjar SS, Turner KL, Bacher AC, et al. Sex dif‑ 2001–2003. J Public Health (Oxf ). 2007;29:191–8. ferences in left ventricular function in older persons with mild hyperten‑ sion. Am Heart J. 2005;150:934–40. Publisher’s Note 38. Brinton RD, Yao J, Yin F, Mack WJ, Cadenas E. Perimenopause as a neuro‑ Springer Nature remains neutral with regard to jurisdictional claims in pub‑ logical transition state. Nat Rev Endocrinol. 2015;11:393–405. lished maps and institutional affiliations. 39. Scheyer O, Rahman A, Hristov H, Berkowitz C, Isaacson RS, Diaz Brinton R, et al. Female sex and Alzheime’’s risk: the menopause connection. J Prev Alzheimers Dis. 2018;5:225–30. 40. Chedraui P, Jaramillo W, Pérez‑López FR, Escobar GS, Morocho N, Hidalgo L. Pro‑inflammatory cytokine levels in postmenopausal women with the metabolic syndrome. Gynecol Endocrinol. 2011;27:685–91. 41. Pfeilschifter J, Köditz R, Pfohl M, Schatz H. Changes in proinflammatory cytokine activity after menopause. Endocr Rev. 2002;23:90–119. 42. Macova M, Armando I, Zhou J, Baiardi G, Tyurmin D, Larrayoz‑Roldan IM, et al. Estrogen reduces aldosterone, upregulates adrenal angiotensin II AT2 receptors and normalizes adrenomedullary Fra‑2 in ovariectomized rats. Neuroendocrinology. 2008;88:276–86. 43. Weber MT, Maki PM, McDermott MP. Cognition and mood in perimeno‑ pause: a systematic review and meta‑analysis. J Steroid Biochem Mol Biol. 2014;142:90–8. 44. Luine VN. Estradiol and cognitive function: past, present and future. Horm Behav. 2014;66:602–18. 45. Kilpi F, Soares ALG, Fraser A, Nelson SM, Sattar N, Fallon SJ, et al. Changes in six domains of cognitive function with reproductive and chronologi‑ cal ageing and sex hormones: a longitudinal study in 2411 UK mid‑life women. BMC Womens Health. 2020;20:177. 46. Yaffe K, Barnes D, Lindquist K, Cauley J, Simonsick EM, Penninx B, et al. Endogenous sex hormone levels and risk of cognitive decline in an older biracial cohort. Neurobiol Aging. 2007;28:171–8. 47. Yaffe K, Sawaya G, Lieberburg I, Grady D. Estrogen therapy in postmeno ‑ pausal women: effects on cognitive function and dementia. JAMA. 1998;279:688–95. 48. Emslie C, Hunt K, Macintyre S. How similar are the smoking and drinking habits of men and women in non‑manual jobs? Eur J Public Health. 2002;12:22–8. 49. Bray RM, Fairbank JA, Marsden ME. Stress and substance use among military women and men. Am J Drug Alcohol Abuse. 1999;25:239–56. 50. Thun M, Peto R, Boreham J, Lopez AD. Stages of the cigarette epidemic on entering its second century. Tob Control. 2012;21:96–101. 51. Cifkova R, Pitha J, Krajcoviechova A, Kralikova E. Is the impact of conven‑ tional risk factors the same in men and women? Plea for a more gender‑ specific approach. Int J Cardiol. 2019;286:214–9. 52. Brickman AM, Schupf N, Manly JJ, Luchsinger JA, Andrews H, Tang MX, et al. Brain morphology in older African Americans, Caribbean Hispanics, and whites from northern Manhattan. Arch Neurol. 2008;65:1053–61. 53. Zahodne LB, Manly JJ, Narkhede A, Griffith EY, DeCarli C, Schupf NS, et al. Re Read ady y to to submit y submit your our re researc search h ? Choose BMC and benefit fr ? Choose BMC and benefit from om: : Structural MRI predictors of late‑life cognition differ across African Ameri‑ cans, Hispanics, and Whites. Curr Alzheimer Res. 2015;12:632–9. fast, convenient online submission 54. Nazare JA, Smith JD, Borel AL, Haffner SM, Balkau B, Ross R, et al. Ethnic thorough peer review by experienced researchers in your field influences on the relations between abdominal subcutaneous and vis‑ ceral adiposity, liver fat, and cardiometabolic risk profile: the International rapid publication on acceptance Study of Prediction of Intra‑Abdominal Adiposity and Its Relationship support for research data, including large and complex data types With Cardiometabolic Risk/Intra‑Abdominal Adiposity. Am J Clin Nutr. • gold Open Access which fosters wider collaboration and increased citations 2012;96:714–26. 55. McKeigue PM, Miller GJ, Marmot MG. Coronary heart disease in south maximum visibility for your research: over 100M website views per year Asians overseas: a review. J Clin Epidemiol. 1989;42:597–609. 56. Eastwood SV, Tillin T, Chaturvedi N, Hughes AD. Ethnic differences in asso ‑ At BMC, research is always in progress. ciations between blood pressure and stroke in South Asian and European Learn more biomedcentral.com/submissions men. Hypertension. 2015;66:481–8.

Journal

Alzheimer s Research & TherapySpringer Journals

Published: Mar 30, 2023

Keywords: Brain age; Cardiometabolic syndrome; Gender; Ethnicity

There are no references for this article.