Abstract
Background:
Cognitive frailty (CF) is defined as simultaneous presence of physical frailty (PF) and cognitive impairment among older adults without dementia. Although white matter hyperintensities (WMH) as expressions of cerebral small vessel disease are associated with physical and cognitive decline and could manifest as CF, this association remains yet to be clarified.
Objects:
To clarify the association between CF and WMH among memory clinic patients.
Methods:
The subjects of this cross-sectional study were 121 cognitively normal (CN) and 212 mildly cognitively impaired (MCI) patients who presented to the Memory Clinic at the National Center for Geriatrics and Gerontology of Japan. PF status was defined based on the definition proposed by Fried and colleagues. CF was defined as simultaneous presence of pre-PF or PF and MCI. WMH volumes were measured using an automatic segmentation application. Multiple liner regression analyses with adjustment for cardiovascular risk factors were performed.
Results:
Of all subjects, 77 (63.6%) and 22 (18.2%) CN patients and 132 (62.3%) and 65 (30.7%) MCI patients were categorized into pre-PF and PF, respectively. Multiple liner regression analysis showed that those with CF had higher WMH volumes than those without (β= 0.23). When categorized into six groups according to PF and cognitive status, the PF/CN (β= 0.15), pre-PF/MCI (β= 0.41), and PF/MCI (β= 0.34) groups had higher WMH volumes than the non-PF/CN group.
Conclusions:
This study showed increased WMH volumes in CF and PF, indicating that WMH could be one of the key underlying brain pathologies of CF.
Keywords
INTRODUCTION
Physical frailty (PF) and cognitive impairment are common health problems in older adults, and simultaneous presence of PF and cognitive impairment (clinical dementia rating = 0.5) is proposed as “cognitive frailty (CF)” in 2013 by the international consensus group comprised of experts from the International Academy of Nutrition and Aging (IANA) and the International Association of Gerontology and Geriatrics (IAGG) [1]. Conceptually, CF is considered to be a state of reduced cognitive reserve which, unlike physiological brain aging, is supposed to be caused not by neurodegenerative disorders but by physical factors [1]. Moreover, as with the concept of PF, CF is also characterized by its potential for reversibility. CF could represent a useful target for early intervention against dependency and dementia in older adults [1]. Recently, the new operational definition, which includes two subtypes, “reversible” CF and “potential reversible” CF, has been proposed [2]. This new operational definition includes subjects with pre-PF and those at pre-clinical stage of dementia such as subjective cognitive decline and positive biomarkers [2]. Although there is as yet no consensus on the definition of CF, several studies demonstrated that CF would be a risk factor of disability, poor quality of life, death, and incident dementia [3].
The underlying mechanisms of CF remain unclear. The close association between PF and cognitive impairment suggests the presence of an underlying mechanism common to these conditions, which may include clinical markers (e.g., older age, lower education and income, alcohol intake, cardiovascular risk factors, and nutritional problems) and inflammatory/immunity markers (e.g., C-reactive protein [CRP], interleukin-6, fibrinogen, homocysteine, and cortisol) [4, 5]. In this context, these clinical and inflammatory markers have been reported as risk factor for white matter hyperintensities (WMH) recognized as one of the cerebral small vessel diseases [6, 7]. WMH can be observed as hyperintensities in T2-weighted and fluid-attenuated inversion recovery (FLAIR) images. The underlying pathology of WMH reflects demyelination and axonal loss as a consequence of chronic ischemia in a majority of older adults. To date, clear evidence exists that WMH causes cognitive decline, in particular executive functions, speed and motor control, and attention, and increases the risk of dementia [8–10]. Current evidence also suggests that, besides cognitive decline and dementia, WMH is associated with geriatric syndromes including physical dysfunction, falls, and urinary dysfunction [9, 10]. Therefore, WMH, which is associated with both physical and cognitive dysfunction, is considered as one of the key underlying mechanisms of CF [11]. However, no studies have investigated the association of CF with WMH.
The aim of this study was therefore to clarify the association of CF with WMH among memory clinic outpatients. In this study, we defined CF as simultaneous presence of pre-PF or PF and cognitive impairment [1, 2]. Identifying common underlying mechanisms of CF, PF, and cognitive impairment may help develop effective strategies for preventing progression of disability and dementia among older adults.
METHODS
Design and subjects
The subjects of this cross-sectional study were outpatients first presenting to the Memory Clinic at the National Center for Geriatrics and Gerontology (NCGG) of Japan during the period from October 2010 to January 2017. We included the 333 patients (aged 65–89 years) who were clinically diagnosed as cognitively normal (CN, n = 121) or mildly cognitively impaired (MCI, n = 212). MCI was diagnosed based on the criteria of the National Institute on Aging-Alzheimer’s Association (NIA/AA) workgroups [12]. All patients completed comprehensive geriatric assessment, brain magnetic resonance imaging (MRI), and blood tests at presentation. The Ethics Committee of the NCGG approved the study protocol. The purpose, nature, and potential risks of the study were fully explained to the subjects, and all subjects gave written informed consent before participating in the study.
Definition of physical frailty and cognitive frailty
We defined PF based on the frailty phenotype proposed by Fried et al in the Cardiovascular Health Study [13]. The components of frailty phenotype include shrinking, weakness, slowness, self-reported exhaustion, and low physical activity. Shrinking was defined as the lowest quintile for body mass index (BMI ≤19.5). Weakness was defined as low hand grip strength measured with a digital force gauge (ZP-500N; Imada, Toyohashi, Japan). Low hand grip strength was defined as hand grip strength less than 18 kg in women and 26 kg in men [14]. Subjects answering “yes” to the question “Do you think you walk slower than before?” were considered as having slowness. Exhaustion was defined as a negative response to the following question in the geriatric depression scale [15]: “Do you feel full of energy?”. Low physical activity was defined as those subjects who did not engage in physical exercise at least once a week. Subjects meeting none of the criteria were classified as non-PF, those who met 1-2 criteria as pre-PF, and those who met 3–5 criteria as PF [13]. Then, in this study, pre-PF or PF individuals with MCI were considered to represent CF [1, 2].
Imaging data and evaluation of white matter hyperintensity
All patients underwent 1.5 T brain MR imaging (Siemens Avanto, Munich, Germany; or Philips Ingenia, Eindhoven, The Netherlands) with T1- and T2-weighted and FLAIR sequence. The individual intracranial volume (IC), brain parenchyma (PAR), and WMH volume were quantified using the automatic Software for Neuro-Image Processing in Experimental Research (SNIPER) segmentation application (Department of Radiology, Leiden University Medical Center, Netherlands). The protocols for brain MRI and SNIPER have been described in detail elsewhere [16, 17]. In analyses, the WMH volume was divided by the PAR volume to adjust for brain atrophy (WMH/PAR, %).
Other variables
Information on the subjects’ age, sex, education, and BMI were obtained from their clinical charts. We assessed global cognitive function and basic activities of daily living were assessed by using the Mini-Mental State Examination (MMSE) [18] and Barthel Index [19], respectively. Cardiovascular factors assessed included hypertension, diabetes mellitus, dyslipidemia, coronary artery disease, smoking status (current smoker or not), and drinking status (drinking daily or not), given that these factors are considered to play a key role in the etiology of WMH [6]. Moreover, in blood testing, all subjects were assessed for their estimated glomerular filtration rate (eGFR), homocysteine, brain natriuretic peptide (BNP), and C-reactive protein (CRP) values.
Statistical analysis
In univariate analyses, those with CF and those without were examined for differences in clinical characteristics and WMH/PAR by using the Kruskal-Wallis test and χ2 test. Additionally, in order to clarify the association of PF and cognitive status with WMH/PAR in detail, subjects were categorized into six groups: non-PF/CN; pre-PF/CN; PF/CN; non-PF/MCI; pre-PF/MCI; and PF/MCI. Then, the 6 groups were also examined for differences in clinical characteristics and WMH/PAR by using the Kruskal-Wallis test and χ2 test.
In multivariate analyses, to clarify the association of CF with WMH, multiple liner regression analysis was performed, where presence or absence of CF was entered as an independent variable. WMH/PAR was entered as a dependent variable. Age, sex, education, and other confounding variables (smoking status, drinking status, hypertension, diabetes mellitus, dyslipidemia, coronary artery disease, eGFR, homocysteine, BNP, and CRP) were entered as covariates. Moreover, to minimize the bias for each patient’s head size, IC volume was also entered as a covariate. WMH/PAR, homocysteine, BNP, and CRP were log-transformed before regression analysis, given that these variables showed skewed distributions. Moreover, to clarify the association of PF and cognitive status with WMH/PAR in detail, multiple liner regression analysis was also performed, with PF and cognitive status (reference group = non-PF/CN) as independent variables.
All statistical analyses were carried out by using STATA 14.2 (Stata Corp, College Station, Texas, USA). p values <0.05 were considered statistically significant.
RESULTS
The mean age of the subjects was 74.7±5.5 years, and 206 (61.9%) subjects were females. The mean MMSE scores of CN and MCI were 28.5±1.8 and 25.4±2.5, respectively. Of all subjects, 63.6% and 18.2% in CN and 62.3% and 30.7% in MCI were shown to represent pre-PF and PF, respectively. Of all subjects, therefore, 197 (59.2%) subjects were shown to represent CF.
Characteristics of subjects with and without cognitive frailty
Table 1 shows the characteristics of the subjects. The values were compared between those with CF and those without. There were differences in age, education, MMSE, drinking status, homocysteine, BNP, and PF components, except for shrinking and exhaustion, among the two groups. Those with CF had lower IC and PAR volumes and higher WMH and WMH/PAR volumes than those without (Table 1).
Clinical characteristics of the subjects with and without CF
Data are presented as n (%), mean (standard deviation), or median (interquartile range). BNP, brain natriuretic peptide; CF, cognitive frailty; CRP, C-reactive protein; eGFR, estimated glomerular filtration rate; IC, intracranial; MMSE, Mini-Mental State Examination; PAR, parenchyma; WMH, white matter hyperintensities.
Supplementary Table 1 shows the characteristics of subjects according to PF and cognitive status. There were differences in age, education, BMI, MMSE, dyslipidemia, homocysteine, BNP, and PF components among the 6 groups (Supplementary Table 1). PAR, WMH, and WMH/PAR volumes were significantly different between 6 groups, and the WMH/PAR volume was shown to be highest among those with MCI and PF (Supplementary Table 1 and Supplementary Figure 1).
Association between cognitive frailty and white matter hyperintensities
Table 2 shows the results of multiple regression analysis, where presence or absence of CF was entered as an independent variable. As a result, those with CF showed a significantly higher WMH/PAR volumes compared to those without (coefficient, 0.57; standard error [SE], 0.13; p < 0.001). The other factors associated with WMH/PAR volumes were older age (coefficient, 0.06; SE, 0.01; p < 0.001) and higher CRP (coefficient, 0.16; SE, 0.06; p = 0.013). To allow interpretation of the association of CF with WMH/PAR volumes, these coefficients were back-transformed into the original metric [20]. Multiple regression analysis thus showed that WMH/PAR volumes were 76.6% [(e0.569 - 1)×100] higher for those with CF than for those without. Additionally, a 1-year increase in age was related to a 6.3% [(e0.061 - 1)×100] increase in WMH/PAR volume. A 1% increase in CRP was related to a 0.2% [(1.01)0.158 – 1]×100] increase in WMH/PAR volume. Moreover, multiple regression analysis, which included PF and cognitive status (reference group = non-PF/CN) as independent variables, showed that the PF/CN (coefficient, 0.77; SE, 0.17; p = 0.026), pre-PF/MCI (coefficient, 1.05; SE, 0.27; p < 0.001), and PF/MCI (coefficient, 1.06; SE, 0.29; p < 0.001) groups had significantly higher WMH/PAR volumes compared to the non-PF/CN group (Supplementary Table 2).
Distribution of WMH/PAR among subjects with and without CF. CF, cognitive frailty; PAR, parenchyma; WMH, white matter hyperintensities.
Association of cognitive frailty with WMH (log transformed)
BNP, brain natriuretic peptide; CRP, C-reactive protein; eGFR, estimated glomerular filtration rate; IC, intracranial; SE, standard error; WMH, white matter hyperintensities.
DISCUSSION
The present study investigated the association between CF and WMH among patients presenting to the Memory Clinic. This study is the first report demonstrating greater WMH volumes in patients with CF than in those without CF.
While a significant link has been widely reported between WMH and cognitive impairment in previous studies [8, 21], the association between PF and WMH remains to be fully elucidated. Recently, among 176 community-dwelling older adults without dementia, the WMH volume was shown to vary between the frailty phenotypes and to be higher in frail persons than in non-frail persons [22]. In the Tasmanian Study of Cognition and Gait population-based study, WMH was shown to be associated with higher frailty index (FI) values in 388 older adults, while this index included cognitive function and comorbidities (such as hypertension) [23]. In this study, there was no significant association between WMH and the frailty phenotypes [23]. Among 87 patients presenting to the university hospital, multidimensional FI including cognitive function was shown to be associated with WMH as determined by the modified Fazekas scale [24]. However, some studies showed disparate results and found marginal [25, 26] or no significant association between WMH and PF [27, 28]. These inconsistent results may be due to differences in the target populations and the operational definitions used for frailty. In some studies that found no association, the focus was on younger subjects who were less frail and had less WMH burden [27, 28]. Indeed, Siejka et al. reported that a greater than moderate WMH burden likely had the strongest association with frailty [23]. In the current study focused on the Memory Clinic patients, PF subjects had higher WMH volumes compared to non-PF subjects among patients without MCI even after adjustment for confounding factors that are considered to play a key role in the etiology of WMH. Moreover, our study was the first to demonstrate increased WMH volumes in CF and even in pre-PF patients with cognitive impairment. This finding suggests that WMH may be among the key common underlying mechanisms of PF and cognitive impairment.
To date, PF has been reported to increase a risk of incident dementia, in particular vascular dementia or non-Alzheimer’s disease (AD) [29–31]. Moreover, several recent studies also demonstrate a significant link between CF and incident dementia [32–35], especially vascular dementia [32]. Additionally, a very recent population-based study operationalized a biopsychosocial frailty model (PF plus psychosocial frailty) and showed that this new construct demonstrated increases in the risk of incident dementia, particularly vascular dementia, during 3.5-year and 7-year follow-up. These previous studies indicate that various frailty phenotypes may be closely associated with cognitive decline and dementia due to vascular pathologies. Our finding that PF or CF subjects have higher volumes of WMH is consistent with these previous studies, given that WMH is associated with a higher risk of incident stroke and dementia [37].
In contrast, a recent study that focused on 91 patients with amnestic MCI, a higher FI was associated with a higher risk of conversion to AD [38]. Additionally, in the Gait and Brain study, combined slow gait and cognitive impairment predicted cognitive decline and incident dementia [39]. In this study, almost all subjects progressing to dementia were diagnosed with AD [39]. Moreover, the level of frailty has been reported to be associated with AD pathologies [40–42]. These results indicate that heterogeneous conditions involving AD pathologies as well as non-neurodegenerative disease may manifest as CF. Further studies investigating the relevance of neurodegenerative disease including AD may deepen our understanding of CF.
The present study has several limitations that should be noted and addressed in future studies. First, since our study was a cross-sectional study, the temporal association between WMH and CF remains unclear. Second, given the focus on outpatients presenting to the Memory Clinic, the sample size of subjects with CN was relatively small in this study, and the study results may not be readily generalizable to other populations. Moreover, the study did not assess other cerebral small vessel diseases including cerebral microbleeds and perivascular space. Despite these limitations, however, the present study is the first to demonstrate a significant link between CF and WMH among the Memory Clinic patients.
In conclusion, this study provides evidence for cross-sectional association between increased WMH and CF, indicating that WMH could be among the key underlying brain pathologies of CF. Further longitudinal studies are needed to confirm our findings.
Footnotes
ACKNOWLEDGMENTS
The authors thank the BioBank at National Center for Geriatrics and Gerontology for quality control of the clinical data. This work was supported by the Research Funding of Longevity Sciences (30-1) from the National Center for Geriatrics and Gerontology, and a Grant-in-Aid for JSPS Research Fellow (No. 17J03037) from Japan Society for the Promotion of Science. The funders had no role in study design, methods, data collections, analysis, and preparation of paper.
