Abstract
Introduction
White matter is made up of axons, different types of glial cells, and blood vessels. The destruction of white matter integrity will lead to a less efficient and effective electrical impulse conduction along the white fiber tracts, and hence, interfere with the normal functioning of the brain. White matter lesions (WML), also known as white matter hyperintensities or leukoaraiosis (Leuko, meaning white, and araiosis meaning rarefaction, in Greek), represent one of the extreme forms of white matter injury and are prevalent in the aging brain. It was estimated that only 5% –13% of brains of older adults are free from WML (de Leeuw et al., 2001). Moreover, WML severity can account up to 15% of the variance in general cognitive functioning in the healthy elderly population (Brickman et al., 2011). Hence, the impact of WML should not be neglected or underestimated.
WML are also common in individuals with mild cognitive impairment (MCI). MCI describes a transitional stage between normal ageing and dementia. Research on MCI has received much attention in the past decade as researchers and clinicians have pushed for early identification and treatment of dementia. Current research suggests that older adults with MCI are more likely to progress to dementia than their counterparts without MCI (Albert, Blacker, Moss, Tanzi & McArdle, 2007; Busse, Hensel, Gühne, Angermeyer & Riedel-Heller, 2006; Landau et al., 2010; Petersen et al., 2001). Petersen and colleagues (2001) reported that most MCI subjects with memory impairment progressed to Alzheimer’s disease at a rate of 10% to 15% per annum, in comparison to those healthy controls whose progressed a rate was 1% to 2 %. It was estimated that the prevalence of WML in MCI ranges from 70% to 100 % and the distribution of lesions was more extensive than that in the healthy aging population (Son et al., 2012; Targosz-Gajniak, Siuda, Ochudło, & Opala, 2009; Wallin et al., 2016).
Recently the International Association of Gerontology and Geriatrics Congress issued a guideline for early diagnosis of cognitive impairment (Morley et al., 2015), suggesting an inclusion of WML investigation as part of the screening for early diagnosis of cognitive impairment. However, reviews focusing on WML in subtypes of MCI are scarce. The findings in the clinical significance of WML in Mild Cognitive Impairment is inconsistent in the literature, with some studies reporting an associated cognitive decline and dementia (B.T. et al., 2012; Bolandzadeh, Davis, Tam, Handy & Liu-Ambrose, 2012; Coutu, Goldblatt, Rosas & Salat, 2015), while others fail to demonstrate such association (Mortamais et al., 2013).
The objectives of the present review are two-folded. First, we aimed to extend the previous reviews by considering the impact of WML in the subtypes of MCI, namely 1) the Amnestic-type mild cognitive impairment single domain (aMCI- single domain), 2) Amnestic-type mild cognitive impairment-multiple domains (aMCI- multiple domains); 3) Non-amnesic mild cognitive impairment-single non-memory domain (non-aMCI- single domain), and 4) Non-amnesic mild cognitive impairment multiple domains (non-aMCI-multiple domains) (Petersen, 2004). Subtyping MCI in the study of WML has a few advantages. It allows us to compare the findings across studies using a more homogenous group of population. It can also allow us to explore whether there is a differential impact of WML on the pathology of different subtypes of MCI. Second, given that the integrity of white matter is crucial for intact cognitive functioning and healthy aging, we also aimed to review the evidence on white matter plasticity and its potential application in intervening or delaying the progress of WML in MCI.
Methods
The literature search consisted of two parts. First, we conducted a literature search in PubMed from 2011 to Sept 2016. Terms for searching included Mild Cognitive Impairment AND white matter lesion*, Mild cognitive impairment AND white matter hyperintensit*, Mild cognitive impairment AND leukoaraiosis. Only studies between years 2011 and 2016 were included in this review as the neuroimaging technology, especially in diffusion imaging, in recent 5 years has become stable and matured. Studies were included if 1) they were published as a peer-reviewed journal article, 2) in English, 3) conducted in humans, 4) they investigated WML in any subtypes of MCI, and 5) the articles included both neuroimaging data and neuropsychological test results. Neuropsychological assessment was required in the same study as we aimed to determine the clinical significance of neuroimaging findings through correlations with neuropsychological evaluation. Studies were excluded if 1) there is only one MCI group, 2) studies with a focus on other areas such as Depression, Parkinson’s disease, Alzheimer’s disease; 3) they were single case studies. Figure 1 describes the procedure in identification and selection of studies. We identified a total of 12 articles for our review in this paper.
Flow chart of identification and selection of studies.
Study characteristics
Table 1 shows the characteristics of studies included in this review. Eight of them included participants with amnestic MCI. The other four included compared MCI who progressed to dementia with those who did not. Of the eight studies that investigate aMCI, only three studies further characterized them into single-domain aMCI and multiple-domain aMCI. Other studies included both single domain and multiple domains in defining their group of aMCI. Only two studies included participants with non-amnesic MCI. All of them adopted the Petersen’s criteria or modified Petersen’s criteria in classifying MCI and its subtypes, which corresponds to 1) memory complaint preferably corroborated by an informant, 2) objective memory impairment which is inconsistent with age and education 3) relatively normal general cognitive functioning, 4) relatively normal activities of daily living and 5) not diagnosed with dementia according to the definition of The Diagnostic and Statistical Manual of Mental Disorders-Fourth edition (DSM-IV) or the International Statistical Classification of Diseases -10th revision (ICD-10) (Petersen, 2004; Petersen et al., 2001).
Characteristics of studies
Characteristics of studies
Note. aMCI = amnesic MCI; dMCI = dysexecutive MCI; aMCI-SD = amnesic MCI – single domain; aMCI-MD = amnesic MCI-multiple-domain; MCI-MCDE = multiple MCI with executive dysfunction; a/eMCI = attention/executive MCI; sMCI = MCI patients who remained cognitively stable in follow-up; pMCI = MCI patients who progressed to dementia at follow-up.
Table 2 summarizes the findings in neuroimaging data of the reviewed studies. All three studies using diffusion imaging indicate microscopic abnormalities of white matter integrity in patients with MCI in comparison to healthy controls. For instance, the non-amnesic attention/executive MCI group had a higher radial diffusivity values in rostral middle frontal, medial orbitofrontal, caudal anterior cingulate, and entorhinal regions than those of the healthy controls (Grambaite et al., 2011). Similarly, Bosch and colleagues (2012) reported a significant increase in radial diffusivity in parts of the inferior longitudinal, occipitofrontal fasciculi, posterior cingulum, right longitudinal superior and uncinate fasciculus comparing between a group of aMCI patients and a group of healthy controls. Uncinate fasciculus, which connects anterior temporal lobe to lateral orbitofrontal cortex, and the cingulate bundles are white matter tracts that have been shown to be involved in memory encoding and retrieval (Wendelken et al., 2015).
Findings on neuroimaging data in the studies
Findings on neuroimaging data in the studies
Note. ARWMC = Age-Related White Matter Change rating scale; CHIPs = The Cholinergic Pathways Hyper Intensities Scale; DTI – TBSS = Diffusion Tensor Imaging – Tract-based Spatial Statistics; FA = Fractional anistropy; MD = Mean diffusivity; DR = Radial diffusivity; DA = Axial diffusivity; FLAIR = Fluid Attenuation Inversion Recovery; aMCI = amnesic MCI; dMCI = dysexecutive MCI; aMCI-SD = amnesic MCI – single domain; aMCI-MD = amnesic MCI-multiple-domain; MCI-MCDE = multiple MCI with executive dysfunction; a/eMCI = attention/executive MCI; sMCI = MCI patients who remained cognitively stable in follow-up; pMCI = MCI patients who progressed to dementia at follow-up; AD = Alzhimer’s disease; HC = healthy controls.
Relative to aMCI single domains, individuals suffering from aMCI-multiple domain showed a more extensive white matter abnormality. Li and colleagues (2013) reported a reduction in fractional anisotropy in distributed brain regions including the body of corpus callosum, fornix, bilateral anterior internal capsule, posterior internal capsule, and tapetum.
Nine studies examined the macrostructure of WML in subtypes of MCI using T2 or FLAIR images. Seven studies of them used visual rating scales of WML and only two computed the volumetric measure in quantifying WML. The findings regarding the macrostructure of WML were less consistent than those from the DTI studies. There were conflicting results regarding the volumetric load or visual ratings of WML in various subtypes of MCI.
Comparing to healthy controls, only one study reviewed here reported a significant increase in WML in aMCI (Naranjoa et al., 2015). The rest of the studies did not find a significant difference in volumetric load or visual ratings of WML between aMCI and healthy controls. Of the three studies which compare MCI who progressed to dementia and those who did not, only one study reported a higher periventricular WML in MCI who progressed to dementia (Defrancesco et al., 2013). However, other research groups failed to replicate these findings(e.g. Eckerström et al., 2015; Nolze-Charron, Mouiha, Duchesne & Bocti, 2015). In sum, studies measuring the macrostructure of WML suggest that WML is a weak radiological marker in differentiating MCI from healthy controls and AD, or other subtypes of MCI.
Cognitive correlates of neuroimaging of WML
Table 3 descirbes the cognitive assessment tools that were employed by the studies. Of those who identified a significant association between WML and aMCI diagnosis, it was consistently reported that periventricular WML was associated with poor performance on cognitive measures such as memory, language, psychomotor speed, attention and executive functions (Defrancesco et al., 2013; Makino et al., 2014; Naranjoa et al., 2015). Total WML burden significantly predicted general cognitive status measured by the Mini-Mental State Examination after controlling for age, education, and gender of the amnesic MCI participants (Kim et al., 2015).
Outcome measures: Cognitive tests used in the studies
Outcome measures: Cognitive tests used in the studies
Note. MMSE = Mini-mental State Examination; WAIS = Wechsler Abbreviated Scale of Intelligence; RAVLT = Rey Auditory Verbal Learning test; COWAT = Controlled Oral Word Association Test; VOSP = Visuo-Object and Space Perception Battery; CAMCOG = Cambridge Cognitive Examination; RCFT = Rey Complex Figure Test; SNSB = Seoul Neuropsychological Screening Battery; TMT = Trail Making Test; AVLT = Auditory Verbal Learning Test; BNT = Boston Naming Test; CERAD = Consortium to Establish a Registry for Alzheimer’s Disease; CWIT = Color-Word Interference Test from Delis-Kaplan Executive Function System; WASI = Wechsler Abbreviated Scale of Intelligence.
In diffusion imaging studies, the association between imaging indices and cognitive measures were reported in all three studies. For instance, mean diffusivity value of the corpus callosum was associated with MMSE and tasks on processing speed in a group of aMCI and with response/switching task in a group of a/e MCI (Grambaite et al., 2011). The FA index was significantly associated with the memory performance in aMCI and AD patients (Bosch et al., 2012).
In this review, we identified 12 recent articles to explore the neuroimaging and neuropsycholgoical correlates of white matter lesions in different subtypes of MCI. Our findings show that 1) WM abnormality was identified between different subtypes of MCI and healthy controls on diffusion imaging; 2) visual ratings of WML or its volumetry did not reliably differentiate different subtypes of MCI or its prognosis to dementia; and 3) WML is associated with a general reduction in cognitive functioning in MCI, and cognitive correlates of WML was evident in the domains of memory, language, psychomotor speed, attention and executive functions in aMCI.
The findings of the studies reviewed in this article concur with the consensus in the literature on MCI, in which the microstructural change in the white matter of MCI are extensive and affects various brain regions (Radanovic et al., 2013; Sexton et al., 2016). For instance, significant changes in the white matter microstructure were reported in the genu and splenium of corpus callosum, middle and posterior cingulum, parahippocampal cingulum, uncinate fasciculus, superior longitudinal fasciculus, and parahippocampal gyrus in AD and MCI individuals (Clerx, Visser, Verhey & Aalten, 2012).
Previous studies using diffusion imaging to investigate the manifestation of WML in different subtypes of MCI suggests that aMCI and non-amnesic MCI follows a different pattern of WM abnormality. For instance, Zhuang and colleagues (2012) reported that WM abnormality was relatively spared in the temporal region of non-amnesic MCI and their WML was anatomically more widespread than aMCI (Zhuang et al., 2012). Findings of Li (2013)’s study gave support to their observation, suggesting that the pathology in amnesic MCI is different from that of non-amnesic MCI and they may possibly follow a different clinical progression.
The discrepancy in the findings between the microstructure and macrostructure of WML in subtypes of MCI calls for further clarification in research. The positive findings in the diffusion imaging studies may suggest that diffusion tensor imaging can detect subtle changes in the white matter. The inconsistent findings in the macrostructure of WML in MCI may also stem from the different methodologies in collecting data and different experimental designs in the studies. For instance, some studies recruited participants from university memory clinics but some recruited the participants from the community. The age range for specifying MCI varies across studies. For instance, Eckerstrom (2015) studied participants of ages between 41 and 78 while Nolze-Charron (2015) studied participants of ages between 69 and 85. It was also difficult to make direct comparison across the studies because of the different inclusion and exclusion criteria. For example, Markino and colleagues (2014) excluded aMCI participants with Fazekas rating greater than two while other studies include all MCI participants in their analysis regardless of their ratings on the Fazekas scale. Hence, the findings may not be generalized to other MCI populations.
Apart from the difference in experimental designs, the null findings in different MCI subgroups may also reflect the heterogeneous pathologies of WML even within the same subtype of MCI. There is still much debate about the pathophysiology of WML and several mechanisms have been proposed, including ischemia, blood- brain barrier alternations, apoptosis, chronic edema, genetic factors or a combination of the above (Pantoni, 2002). One theory proposed that WML occurs as a result of Wallerian degeneration, in which loss of axons runs parallel to the gray matter pathology and begins from the hippocampus and entorhinal region to the temporal and parietal association cortex (Sexton, Kalu, Filippini, Mackay & Ebmeier, 2011). Conversely, another theory follows the retrogenesis model and proposes that white matter change is independent of the gray matter pathology. The white matter will follow the reverse order of myelinogenesis, in which white matter will degenerate first in the late-myelinating WM tracts, such as the inferior and super longitudinal fasciculus and the uncinate fasciculus (Meier et al., 2012). In short, our current results suggest that WML detected in T2 or FLARE images of MCI cannot differentiate different subtypes of MCI and is yet to be a reliable prognostic marker in determining the risk of dementia. The various mechanisms and theories suggest a high heterogeneity in the pathology of WML and its manifestation in MCI is complex.
Our review on the cognitive correlates of WML supports the two prevalent notions in the existing literature. First, WML is associated with reduced cognitive functioning in MCI. It is widely accepted that a severe grade of WML has a detrimental impact on cognitive functioning including executive function, attention, processing speed, and global cognition (Jokinen, Lipsanen & Schmidt, 2012; Meier et al., 2012; Pantoni, Poggesi, & Inzitari, 2007; Prins & Scheltens, 2015; Xiong & Mok, 2011). Second, the location or the spatial distribution of the WML has a differential impact on cognitive functioning. Specifically, periventricular WML, not subcortical WML, has a negative impact on cognitive functions in the MCI population. One speculation is that the periventricular region houses numerous long associating fibers that connect the cortex to various subcortical nuclei and other distant brain regions. Hence damage in the periventricular WML results in impairment in executive functions and processing speed (Bolandzadeh et al., 2012).
WML and white matter plasticity
Despite the gradual deterioration of white matter integrity with age, a growing number of studies have shown the white matter plasticity in the aging brain. Research in the past decade has identified some factors that modulate the decline and enhance its integrity in healthy older adults, including physical exercise (Fleischman et al., 2015; Gow et al., 2012; Sen et al., 2012), cognitive activities (Wirth, Haase, Villeneuve, Vogel & Jagust, 2014), and complex leisure activities (Saczynski et al., 2008). In one study, older adults who participated in a life-long high volume and high-intensity exercise training showed a significant 83% reduction in deep white matter lesions and a 44% reduction in total white matter hyperintensities volume relative to their sedentary counterparts. Moreover, this group of physically fit older adults also showed a more preserved front-to-back white matter network that is important for visuospatial function, motor control and coordination (Tsang et al., 2013).
In addition, experience-induced white matter plasticity was also evident in healthy older adults. Lovden and colleagues (2012) found a decrease in the mean diffusivity signals in older participants who were trained to play a spatial navigation video game while walking on a treadmill over a period of 4 months (52 sessions and 50 mins/session). Loveden and colleagues (2010) demonstrated an experience-dependent plasticity of white matter integrity by training older adults on working memory, episodic memory and perceptual speed tasks for half a year. These older adults showed a decrease in RD especially from the anterior part of the corpus callosum. In a randomized control study, Engvig and colleagues (2012) showed that an 8-week of intensive memory training can increase the FA signal in the anterior regions of the brains of the older participants as well as an enhanced performance on the verbal memory tasks. A more recent study on a one-week 3-session of visual perceptual training can induce white matter change beneath the visual cortex (Yotsumoto et al., 2014). These findings suggest an intact ability of the white matter plasticity in the aging brains.
Despite the promising evidence of white matter plasticity, there is currently no effective pharmacological or behavioral intervention for curing white matter lesions or delaying its progress. However, emerging evidence has shown that the brain is capable of regenerating white matter connectivity after a traumatic loss of white matter. For instance, adult rodents with parts of their corpus callosum surgically removed showed a remarkable recovery after 28 days and restored most of its functional connectivity measured by the resting state fMRI connectivity (Zhou et al., 2014).
To date, only a few studies have investigated the plasticity of white matter in older adults with MCI, and inconsistent findings were found. For instance, Podewils et al. (2006) found no association between physical activity and the progression of white matter lesions in MCI. Doi and colleagues (2015) demonstrated that physical exercise was a neuroprotective factor that protects against brain atrophy in a group of older adults with MCI independent of their level of white matter lesions. More research effort is called to investigate the potential plasticity of white matter in older adults with MCI, given the benefits of it may potentially increase the cognitive reserve and lessen the impact of the pathology of the neuro-degenerative disease.
To our best knowledge, there are two on-going rigorous studies that investigate the plasticity of white matter in older adults with MCI. In Flak’s study (Flak et al., 2014), the authors employ a randomized control study design to investigate whether a 5-week 45 minutes of computerized working memory training will relate to changes in their white matter of the brain in patients with MCI. In another study, Cyarto and colleagues (2012) evaluate whether 24 months of moderate, 150 minutes/week home-based physical exercise could delay the progression of white matter changes in older adults with MCI. These two pre-registered studies will offer some insights into the white matter plasticity in older adults with MCI.
Directions for future research
Through reviewing recent evidence of WML in different subtypes of MCI, our review suggests that the clinical significance of WML in subtypes of MCI remains incompletely defined. There is a dearth of research studying the plasticity of white matter in the MCI population and there was insufficient evidence to support the prognostic role of WML as a radiological marker for dementia. However, our review supports that the distribution and the load of WML are associated with cognitive functioningin MCI.
More research is needed to differentiate the impact of WML on different subtypes of MCI, especially the non-amnesic subtype of MCI. Moreover, there is no gold standard in quantifying WML in the literature. The different quantifying strategy in different studies may give rise to the inconsistent findings and make comparison among studies difficult. The Fazekas Scale (Fazekas et al., 1987), the Sacheltens scale (Scheltens et al., 1993), and the Age-related White matter Change Scale (ARWMC; Wahlund et al., 2001) are commonly used semi-quantitative visual rating scale used to identify hyperintense on MRI images. A previous study compared the three visual rating scales on quantifying WML in AD and MCI patients. Their results suggested that all three rating scales have good face validity and correlate well with the WML volumetry, but the more complex rating scales correlate better with cognitive measures (Gao et al., 2011). However, one caveat of using visual rating scale is that these scales may subject to a ceiling effect and lead to underestimation of the severity of WML. Hence, quantitative volumetry of WML has an advantage over the conventional visual rating methods, and future studies are encouraged to consider both volumetry and visual rating when quantifying WML.
Effective interventions for WML are lacking. On top of the plasticity of white matter, the potential role of cognitive or brain reserve in mitigating the impact or progress of WML may shed light into the intervention of WML. In the context of normal aging, positive evidence has been established in the association between reserve, WML and cognitive functioning (Brickman et al., 2011). In a large population study of healthy elderly subjects aged 64 to 76 years, education, as a proxy for cognitive reserve, was found to modulate the consequences of WML on their cognitive performance (Dufouil et al., 2003). In other words, those with higher cognitive or brain reserves can withstand more severe WML than those with a lower reserve. The empirical evidence for the resilient effect of reserve in the MCI group awaits further substantiation. In a recent study by Serra et al. (2015), the authors showed that cognitive reserve could modulate the impact of the AD pathology in aMCI patients with WML.
After almost three decades of extensive research in the area of WML, the clinical significance of WML has been increasingly recognized. WML is part of the aging process, but the presence of WML should not be taken as merely incidental, especially in the cognitive at-risk population i.e. MCI. More research is needed to clarify the nature of WML in various subtypes of MCI in order to improve the clinical management of WML. Given the evidence of the plasticity in the white matter of aging brain and the potential role of cognitive or brain reserve in modulating the impact of WML, more research effort is called for to harness this remarkable resilience of our brains and protect it from the deleterious impact of WML.
Conflict of interest
There are no conflicts of interest to report.
Footnotes
Acknowledgements
This project was supported by the KKHo International Charitable Foundation.