Abstract
White matter dementia (WMD) is a syndrome introduced in 1988 to highlight the potential of cerebral white matter disorders to produce cognitive loss of sufficient severity to qualify as dementia. Neurologists have long understood that such a syndrome can occur, but the dominance of gray matter as the locus of higher function has strongly directed neurobehavioral inquiry to the cerebral cortex while white matter has received less attention. Contemporary neuroimaging has been crucial in enabling the recognition of white matter abnormalities in a host of disorders, and the correlation of these changes with cognitive performance. Comprising about half the brain, white matter is prominently or exclusively involved in well over 100 disorders, in each of which white matter dysfunction can potentially cause or contribute to dementia. Neuropsychological findings from ten categories of white matter disorder lead to a convergence of findings that document remarkable neurobehavioral commonality among the dementias produced. More recently, the syndrome of mild cognitive dysfunction (MCD) has been introduced to expand the concept of WMD by proposing a precursor syndrome related to early white matter neuropathology. WMD and MCD inform the understanding of how white matter contributes to normal and abnormal cognition, and the specific neuroanatomic focus of these syndromes may enhance the diagnosis and treatment of many disabling disorders that do not primarily implicate the cerebral cortex. Forming essential connections within widely distributed neural networks, white matter is critical for rapid and efficient information transfer that complements the information processing of gray matter. As neuroimaging continues to advance, further information on white matter structure can be expected, and behavioral neurology will play a central role in elucidating the functional significance of these emerging data. By emphasizing the contribution of myelinated systems to higher function, the study of white matter and cognition represents investigation of the basic neuroscience of human behavior.
Keywords
Introduction
Behavioral neurology is commonly described as the study of higher cortical function, an association held so firmly by many neuroscientists that a “corticocentric” view of behavior is often adopted without question (Parvizi, 2009). However, a wealth of evidence supports the notion that brain-behavior relationships extend far beyond the syndromes resulting from damage to the cerebral cortex (Geschwind, 1965; Schmahmann et al., 2008; Filley, 2010; Filley, 2011a). Whereas the importance of the cortical mantle in elaborating human behavior is firmly established, the contributions of non-cortical regions – the subcortical gray matter and the white matter – cannot be neglected. Experienced neurologists recognize the potential for damage to these regions to produce significant neurobehavioral dysfunction, but the traditional dominance of the cortical gray matter in the neuroscience of behavior has to some extent hampered investigation of the full range of brain-behavior relationships. This limitation is nowhere more evident than in the study of dementia.
The growing problem of Alzheimer’s disease (AD) overshadows the entire field of dementia, and its well-known neuropathology naturally directs attention to the cerebral cortex in the investigation of etiopathogenesis, clinical phenomenology, and treatment (Querfurth and LaFerla, 2010). But many other dementias result from neuropathology subjacent to the cortex. The subcortical dementias, the most prominent being Huntington’s disease (HD) and Parkinson’s disease (PD), are associated with neuropathology in the basal ganglia and other deep gray matter structures, and despite some criticism, the concept of subcortical dementia has persisted as a useful contrasting clinical syndrome to the cortical dementia of AD. White matter disorders are often included in the subcortical dementias, alternatively known as frontal-subcortical dementias (Bonelli and Cummings, 2008), but uncertainty lingers about the importance of white matter damage in producing neurobehavioral effects, and demented patients with white matter lesions are frequently assumed to have coexistent cortical neuropathology to explain their cognitive loss.
White matter, however, merits systematic consideration as a brain component critical to behavioral neurology (Filley et al., 1988; Schmahmann et al., 2008; Filley, 2010; Filley, 2011a; Filley, 2012). Comprising about half the human brain, white matter provides the essential connectivity of distributed neural networks coursing within and between the hemispheres to subserve a host of neurobehavioral functions. White matter expands the operational capacity of neurons by enabling the rapid and efficient transfer of information that complements the information processing of gray matter.
The increasing interest in white matter in behavioral neurology has been stimulated by impressive advances of neuroimaging, beginning with the introduction of magnetic resonance imaging (MRI) in the early 1980s. For the first time, white matter could be visualized and correlated with clinical phenomena without the need for autopsy. This advantage allowed application of the lesion method, a time-honored approach at the core of behavioral neurology, to lesions of white matter in the same manner as has been the case for gray matter. An impressive data base has since been generated to support the role of white matter in cognition and emotion. In addition to the well-recognized sensorimotor deficits that can accompany white matter involvement – such as visual loss, paresis, spasticity, ataxia, gait disorder, and incontinence – a wide variety of neurobehavioral dysfunction can occur, including focal disconnection syndromes, neuropsychiatric disorders including depression, and cognitive impairment often sufficiently severe to merit the term dementia (Filley, 2012).
White matter has a legitimate position in the study of dementia. The neuropathology of white matter disorders is typically diffuse or widespread, thus disrupting many networks simultaneously and producing a multi-domain syndrome that merits the term dementia. Beyond the well-recognized focal neurobehavioral syndromes of white matter (Geschwind, 1965), dementia resulting from white matter lesions is fact far more common (Filley, 2011a). This article will review and update the syndrome of white matter dementia (WMD), a term introduced more than two decades ago (Filley et al., 1988) to call attention to the cognitive sequelae of brain white matter disorders.
A brief neuropathologic overview of dementia
The acquired loss of cognitive competence meant by the term dementia implies that brain function is compromised to the extent that multiple cognitive domains are affected and neither normal cognition nor customary social and occupational function can be maintained (McKhann et al., 1984; American Psychiatric Association, 1994). One approach to the understanding of dementia can begin with focusing on the brain regions in which dysfunction is thought to occur (Filley, 2011b). A classification based on neuropathology not only informs diagnostic evaluation but highlights the diversity of systems that each contribute to the phenomenology of cognitive function. To begin, the cerebral cortex, including the hippocampus and entorhinal region, is regarded as the primary site of neuropathology in AD, the classic cortical dementia, and neuritic plaques and neurofibrillary tangles remain the targets of most investigation directed at this increasingly threatening medical and societal challenge. The subcortical dementias, meanwhile, including but not limited to HD and PD, are associated with major neuropathology in deep gray matter nuclei such as the caudate and substantia nigra (Bonelli and Cummings, 2008). A third category can also be invoked, that of mixed dementia, in which neuropathology affects both cortical and deep cerebral structures; examples include multi-infarct dementia and Creutzfeldt-Jakob Disease. A fourth and least familiar category is one characterized by significant neuropathology in white matter. Dementia in this context results from diffuse or widespread involvement of association and commissural tracts that provide critical intra- and interhemispheric connectivity enabling the structure and function of distributed neural networks mediating cognition (Schmahmann et al., 2008; Filley, 2012). Variable degrees of overlap undoubtedly exist between these four categories, to some extent blurring neuropathologic distinctions. White matter has been found abnormal in AD, for example, and conversely, cortical involvement can be detected in multiple sclerosis (MS). However, AD is characterized by massive cortical neuronal loss, with secondary white matter changes from Wallerian degeneration, whereas in MS, primary white matter damage occurs as a consequence of inflammatory demyelination. These neuropathologic differences imply that each dementia category features a unique profile of cognitive impairment, and substantial evidence exists for these distinctions (Filley, 2011b; Filley, 2012). Taken as a whole, therefore, this structurally-based classification assists in the understanding the diagnosis, etiopathogenesis, and treatment of dementia in all its forms (Filley, 2011b; Filley 2012).
The syndrome of white matter dementia
Neurologists since at least the time of Jean-Martin Charcot in the 19th century have appreciated that cerebral white matter disease may produce important neurobehavioral sequelae, including dementia. In 1988, based largely on experience with toluene leukoencephalopathy (Filley et al., 1990) and MS (Franklin et al., 1989), the syndrome of WMD was proposed to describe the dementia.that can accompany cerebral white matter involvement. Implicit in the idea was that white matter disorders typically involve widespread areas of the brain, and the disruption of myelinated tracts affects cognition regardless of the specific neuropathology present. WMD is thus defined as a dementia syndrome resulting from diffuse or multifocal cerebral white matter damage (Filley et al., 1988; Filley, 1998; Schmahmann et al., 2008).
WMD can be caused by disorders within ten categories, constituting a wide array of neuropathologic states with the notable exception of neurodegenerative disease (Table 1). Well over 100 disorders can lead to WMD, a list that is constantly expanding as MRI discloses new entities and older ones are better understood (Filley, 2012). These diseases, injuries, and intoxications may occur at any age and in a wide variety of medical settings, highlighting the broad relevance of the WMD concept. Discussion of all these entities, beyond the scope of this review, can be found elsewhere (Filley, 2012). Perusal of current neurologic literature, however, quickly discloses the breadth and depth of the investigations now being conducted. A particularly vigorous research area is cerebrovascular disease, and the phenomenon of leukoaraiosis has stimulated much study as a cause of cognitive dysfunction (Inzitari et al., 2009; Debette and Markus, 2010).
Neuropathological categories of white matter disorder, with selected examples.
The clinical profile of WMD has been investigated by considering a variety of white matter disorders in parallel. Initial efforts were directed at distinguishing WMD from both cortical and subcortical dementia, and studies comparing MS with AD (Filley et al., 1989) and MS with HD (Lafosse et al., 2007) were instructive. These and other reports helped establish that WMD differs from both cortical and subcortical dementia, particularly in early stages of the disorder before severe dementia has developed. WMD is distinguished from cortical dementia by relative normalcy of language and declarative memory encoding while cognitive speed, executive function, and sustained attention are impaired, and from subcortical dementia by sparing of procedural memory and extrapyramidal function (Filley, 1988; Filley, 1998; Filley, 2012). Core distinctions between the three syndromes, shown in Table 2, are most apparent in specific aspects of memory and language.
Core distinctions between cortical, white matter, and subcortical dementia.
As time progressed and new data emerged, the importance of cognitive slowing became apparent. Whereas slowed cognition was long associated with the subcortical dementias (Albert et al. 1974; McHugh and Folstein, 1975), and indeed all dementia patients may struggle with slowed processing speed, the advent of advanced neuroimaging allowed detailed study of white matter tracts in vivo, and cognitive processing speed has been tightly linked with the integrity of myelinated systems (Turken et al., 2008; Penke et al., 2010). Disorders that impair central impulse conduction thus produce slowed cognition – a reasonable assumption now increasingly supported by neuroimaging evidence – and cognitive slowing has become a distinctive feature of WMD (Filley, 2012). With this addition to the list of cognitive deficits and areas of relative strength, the current formulation of the clinical profile of WMD is shown in Table 3.
The profile of white matter dementia.
One of the recurring questions about the WMD concept is the degree to which co-existent gray matter neuropathology may explain the cognitive deficits. Relatively few cognitive disorders affect white matter exclusively, and as discussed elsewhere (Filley, 2012), the relative contributions of white and gray matter changes may both be relevant. However, the accumulated weight of correlational evidence justifies the existence of WMD in many patients with many disorders (Filley, 2012). Toluene leukoencephalopathy, for example, is a convincing cause of WMD, with confirmatory neuroimaging, neuropsychological (Yücel et al., 2008) and neuropathologic (Al-Hajri and Del Bigio, 2010) evidence continuing to support the earlier reports making this claim (Hormes et al. 1986; Rosenberg et al., 1988; Filley et al., 1990; Filley et al., 2004). Even in MS, in which gray matter neuropathology has long been recognized, the WMD model appears justified, at least in the initial stages of the disease. Cognitive decline begins early in MS (Amato et al., 2010), and while all patients have white matter lesions, cortical lesions are found in only 8% of children (Absinta et al., 2011) and 38% of adults with newly apparent disease (Lucchinetti et al., 2011). Moreover, the cognitive profile of MS resembles other that of white matter disorders more than AD (Filley et al., 1989), suggesting that tract demyelination is the major determinant of cognitive decline, at least until later in the course.
Mild cognitive dysfunction
WMD is a syndrome based on the correlation of white matter lesions visible on conventional MRI with a specific profile of cognitive deficits. The early years of the MRI era made possible this research, but recent advances in neuroimaging have revealed an intriguing new finding: the normal-appearing white matter (NAWM) on conventional MRI may not be normal when examined with more sensitive techniques. These newer techniques, among them magnetic resonance spectroscopy (MRS), magnetization transfer imaging (MTI), and diffusion tensor imaging (DTI), permit the investigation of NAWM and regularly uncover microstructural abnormalities. A rapidly expanding list of conditions can now be assembled in which the NAWM is not normal when queried by one or more of these methods (Table 4). Given the neurobehavioral importance of macrostructural white matter lesions, an appropriate next question concerns the potential effects of more subtle abnormalities in NAWM.
Some conditions in which the normal-appearing white matter may be abnormal.
Studies of WMD as well as clinical experience suggest the existence of a less severe cognitive syndrome associated with less obvious white matter neuropathology. A good example of the potentially wide range of white matter involvement can be seen in toxic leukoencephalopathy, in which a spectrum of MRI and neuropathological changes occurs, with the resultant neurobehavioral dysfunction being concomitantly mild, moderate, or severe (Filley and Kleinschmidt-DeMasters, 2001; Filley et al., 2004). The concept of a spectrum of white matter alterations can be extended to suggest that NAWM abnormalities may be associated with subtle cognitive loss even before macrostructural lesions appear on conventional MRI.
In this context, the category of inflammatory white matter disease became relevant as a clinical model for studying the effects of NAWM changes on cognition. Systemic lupus erythematosus (SLE) presented itself an ideal disease for this purpose as an illness that typically affects young adults with little or no other neuropathology, features white matter hyperintensities as its most common conventional MRI finding, and frequently affects cognition at an early stage (Kozora and Filley, 2011). MRS studies of frontal NAWM in fact disclosed that increased choline (Ch), a marker of inflammation and demyelination, correlated with impaired processing speed, attention, and executive dysfunction in SLE patients whose gray and white matter volumes did not differ from controls (Filley et al., 2009). The elevation of Ch suggested myelinopathy, presumably immune-mediated, since the axonal marker N-acetyl aspartate was found to be normal (Filley et al., 2009).
From these and other findings, the term mild cognitive dysfunction (MCD) was introduced (Kozora and Filley, 2011). MCD is an early cognitive syndrome in SLE patients, and the correlation of cognitive slowing, executive dysfunction, and inattention with increased Ch suggests that in some patients MCD results from insidious white matter neuropathology. More study is needed to confirm the relationship of microstructural white matter abnormalities and cognition, but this model may inform the understanding of how subtle white matter changes disrupt cognition, Described specifically in SLE, MCD may also apply to other disorders in which white matter is involved, and many possibilities in improving early diagnosis and treatment are apparent. From a cognitive perspective, the deficits of MCD are remarkably similar to, but less severe than, those of WMD, again supporting the legitimacy of WMD as a dementia syndrome.
The relationship of MCD to WMD can be seen as based in white matter dysfunction that includes both macrostructural and microstructural involvement. Figure 1 depicts in schematic form the proposed general relationship between cognitive decline and white matter involvement of varying severity. The categories of MCD and WMD require further study using a range of disease models, as distinctions between these syndromes need clarification and correlation with neuropathological changes, but the notion of increasing white matter change producing advancing degrees of cognitive loss serves as a paradigm for exploring the entire range of white matter involvement.
Proposed schematic relationship between white matter and cognition (WM – white matter, NAWM – normal-appearing white matter; MRI – magnetic resonance imaging).
An obvious parallel to MCD in behavioral neurology is mild cognitive impairment (MCI; Petersen, 2011), a syndrome being intensively studied as a putative precursor to AD. MCD does not readily correspond to one of the subtypes of MCI (Petersen, 2011), and its specific profile merits a separate category. Many differences in fact exist between MCD and MCI (Table 5), and while the understanding of MCD is still preliminary, a syndrome that captures the cognitive effects of white matter dysfunction of any etiology may prove useful in contrast to MCI as a syndrome of gray matter degeneration.
Mild cognitive dysfunction compared with mild cognitive impairment.
Clinical implications
The clinical impact of WMD and MCD will be apparent in terms of improving diagnosis, prognostication, and treatment options. As in every clinical setting, maintaining suspicion for the problem will improve the likelihood of its detection, and thus improve all further aspects of care.
The diagnosis of cognitive decline related to white matter involvement can be challenging because language is typically normal or nearly so in affected patients, and deficits are more apparent in cognitive speed, executive function, and attention. The relative subtlety of these deficits can mean that many impaired patients remain undetected as other sensorimotor features of the illness often dominate the clinical encounter. Detailed neurobehavioral or neuropsychological evaluation is sensitive to these deficits but may not always be feasible, and so brief clinical measures may be required. The familiar Mini-Mental State Examination (MMSE; Folstein et al., 1975), while useful in the assessment of AD, is heavily weighted toward language and relatively insensitive to the executive dysfunction of patients with white matter disorders (Franklin et al., 1988; Swirsky-Sacchetti et al., 1992). Other more useful tests include the Montreal Cognitive Assessment (MoCA; Nasreddine et al., 2005), the Frontal Assessment Battery (FAB; Dubois et al., 2000), and the Clock Drawing Test (CDT; Cosentino et al., 2004). The MoCA (Griebe et al., 2011), the FAB (Kanno et al., 2011), and the CDT (Kim et al., 2009) have all been found sensitive to white matter dysfunction in various disorders. The MoCA may be most convenient of these measures as it incorporates executive function and clock drawing tasks into a 30-point format; another advantage is that it enables the testing of memory retrieval, a clinical feature central to the assessment of WMD (Table 2).
Determining the prognosis of WMD and MCD will be improved by greater understanding of white matter neurobiology. The wide range of white matter disorders of course means that prognosis depends most critically on the disease, intoxication, or injury present, the age of the patient, and impact of co-existent problems. Generalizations about all white matter disorders, however, are already possible. It is now known, for example, that axonal loss clearly implies a worse prognosis than myelin damage alone (Trapp et al., 1998; Medana and Esiri, 2003). In addition, as the MCD construct is applied to white matter disorders at an early stage, an intriguing prospect is the opportunity to use neuroimaging biomarkers together with clinical data to detect dysfunction at a time when the prognosis for recovery is far more favorable (Filley et al., 2009; Kozora and Filley, 2011).
Treatment of WMD and MCD is also likely to improve. Each disorder causing these syndromes will be addressed by investigation within the relevant research community, and preventive, medical, surgical, psychiatric, and rehabilitative interventions may all be relevant. Among many questions is whether immunosuppressive and immunomodulatory treatment in MS will effect cognitive improvement, as has been suspected (Fischer et al., 2000; Tumani and Uttner, 2007). Other promising therapeutic avenues for white matter disorders include bone marrow transplantation (Krivit et al., 1999), gene therapy (Leone et al., 2000), and stem cell therapeutics (Goldman, 2007; Tran et al., 2010). An interesting approach to many psychiatric disorders has been proposed with the use of deep brain stimulation (DBS) within white matter structures; DBS was shown effective for intractable depression when applied to the subgenual white matter (Mayberg et al., 2005) and for obsessive-compulsive disorder after stimulation of the ventral anterior internal capsule (Greenberg et al., 2006). The rehabilitation of white matter disorders may also be influenced by new understanding of the response to neuropathologic insults. Evidence has been presented that white matter may be reparable by intrinsic plasticity that can remyelinate axonal segments through activity-dependent myelination (Wake, Lee, and Fields, 2011; see below).
Research perspectives
Many new directions can be imagined from a consideration of the role of white matter in cognition. These include pursuing a better understanding of normal white matter neuroanatomy as it participates in the architecture of cognition, novel innovations for diagnosis and treatment of many dementing disorders, and, perhaps most intriguingly, the hypothesis that AD may in fact begin in the white matter (Bartzokis, 2011).
An important outcome of work on myelinated systems is that the normal white matter will be much better understood. In recent years, classic neuroanatomy has been greatly advanced by more detailed understanding of the origin, course, and termination of white matter tracts in the rhesus monkey (Schmahmann and Pandya, 2006), and new neuroimaging methods promise to address these questions in humans with increasing sophistication. A productive collaboration can be developed between neuroimaging and traditional neuroanatomy to refine knowledge of normal white matter (Schmahmann and Pandya, 2006; Mesulam, 2011).
These investigations will in turn clarify the neuroanatomy of distributed neural networks, the central organizational feature of the brain as it subserves the wide array of cognitive and emotional operations comprising the human behavioral repertoire. White matter is the connecting tissue between cortical and subcortical gray matter regions within and between the hemispheres, and recent data also point to the importance of white matter in networks involving the cerebellum (Schmahmann and Pandya, 2008). The concept of the connectome – “a comprehensive structural description of the network of elements and connections forming the human brain” – has recently been proposed to capture the spirit of this endeavor (Sporns, 2011), and mapping of human brain connectivity is underway.
A fascinating new idea is the notion of white matter plasticity. Long known to be a capacity of gray matter related to synaptic function, plasticity has recently been observed to occur in white matter as well (Fields, 2010). This phenomenon has been demonstrated in both normal individuals, such as piano players whose pyramidal tract integrity correlated with number of hours practiced (Bengtsson et al., 2005), and in neurologic patients, such as those with Broca’s aphasia in whom the right arcuate fasciculus volume increased as Melodic Intonation Therapy improved language performance (Schlaug et al, 2009). White matter plasticity may be mediated by newly discovered glutamatergic axo-oligodendroglial synapses within white matter, which appear to be capable of myelinating distal axonal segments upon stimulation of the proximal axon (Alix and Domingues, 2011). More study is needed to understand the potential applications of white matter plasticity to clinical populations, but the implications for recovery after any myelin lesion may be profound.
The possibility exists that AD may originate in cerebral white matter (Bartzokis, 2011). This “myelin hypothesis” posits that the vulnerability of aging myelin to a host of insults such as ischemia, trauma, and iron toxicity may initiate a cascade of events culminating in dementia; plaque and tangle formation are interpreted not as primary events, but instead as markers of failed myelin repair processes (Bartzokis, 2011) This novel idea merits attention given the still uncertain etiopathogenesis of AD, and the failure of amyloid-targeted immunotherapies to effect improvement in dementia that have raised doubts about the amyloid hypothesis (Hardy, 2009). Much work is needed to evaluate the myelin hypothesis, but provocative neuroimaging studies have found evidence of microstructural white matter damage in women genetically at risk for AD before the onset of dementia and the appearance of medial temporal atrophy (Gold et al., 2010).
Summary
The syndrome of WMD has now existed for more than 20 years, and its utility as an organizational concept appears to be expanding. WMD, and its more recent companion MCD, are not disease-specific diagnoses, and neither incidence nor prevalence data are available for either syndrome, but steadily mounting evidence for the role of white matter in cognition helps support their legitimacy. As a measure of how much this field has grown since the coining of the term WMD (Filley et al., 1988), a basic
The clinical and research benefits of investigating WMD may be substantial. The diagnosis of many patients with white matter disorders affecting cognition can be enhanced, particularly if the putative precursor syndrome of MCD proves useful in identifying those with early involvement. Treatment involving existing and many evolving modalities will continue to advance, and the opportunity to treat patients at the early stage of MCD may substantially improve outcome. Research will be invigorated by a host of innovations that consider the white matter as the primary site of neuropathology.
Finally, from a theoretical perspective, the study of white matter and cognition offers the appealing prospect of furthering the pursuit of fundamental brain-behavior relationships. It is commonly held that basic science necessarily involves reductionistic investigation in the laboratory, where a tightly defined problem can be isolated, described, and modeled as a basis for understanding a larger system. But when the larger system is the vast panoply of human behavior, humans must be the objects of investigation (Bear, 1997), and it is critical to understand the cerebral origins of behavioral dysfunction by assembling the sequelae of structural damage from many disorders, and then seeking to combine these data into a clinical syndrome with wide generalizability. Cataloguing and synthesizing the cognitive effects of white matter dysfunction can thus be seen as basic science as much as examining the impact of a gene mutation, viral infection, or inflammatory cascade. Consideration of the contributions of white matter to cognition and its decline significantly expands the fundamental study of human mentation that is the essence of behavioral neurology.
Footnotes
Funding
This work was in part supported by two research grants: 1 R01 AR04915-01A2 8/15/04-6/30/08 National Institute for Arthritis, Musculoskeletal and Skin Disease MRI/MRS Correlates of Cognitive Function in Systemic Lupus Erythematosus 3 RO1 AR049152-02S1 7/1/05-6/30/08 National Institute for Arthritis, Musculoskeletal and Skin Disease Immune Correlates of Cognitive Function in Systemic Lupus Erythematosus.