Alcohol-related dementia: an update of the evidence

The characteristics of dementia relating to excessive alcohol use have received increased research interest in recent times. In this paper, the neuropathology, nosology, epidemiology, clinical features, and neuropsychology of alcohol-related dementia (ARD) and alcohol-induced persisting amnestic syndrome (Wernicke-Korsakoff syndrome, or WKS) are reviewed. Neuropathological and imaging studies suggest that excessive and prolonged use of alcohol may lead to structural and functional damage that is permanent in nature; however, there is debate about the relative contributions of the direct toxic effect of alcohol (neurotoxicity hypothesis), and the impact of thiamine deficiency, to lasting damage. Investigation of alcohol-related cognitive impairment has been further complicated by differing definitions of patterns of alcohol use and associated lifestyle factors related to the abuse of alcohol. Present diagnostic systems identify two main syndromes of alcohol-related cognitive impairment: ARD and WKS. However, 'alcohol-related brain damage' is increasingly used as an umbrella term to encompass the heterogeneity of these disorders. It is unclear what level of drinking may pose a risk for the development of brain damage or, in fact, whether lower levels of alcohol may protect against other forms of dementia. Epidemiological studies suggest that individuals with ARD typically have a younger age of onset than those with other forms of dementia, are more likely to be male, and often are socially isolated. The cognitive profile of ARD appears to involve both cortical and subcortical pathology, and deficits are most frequently observed on tasks of visuospatial function as well as memory and higher-order (executive) tasks. The WKS appears more heterogeneous in nature than originally documented, and deficits on executive tasks commonly are reported in conjunction with characteristic memory deficits. Individuals with alcohol-related disorders have the potential to at least partially recover - both structurally and functionally - if abstinence is maintained. In this review, considerations in a clinical setting and recommendations for diagnosis and management are discussed.

in combination with nutritional defi ciencies). Attempts to clarify this have been hindered by confounding factors that often accompany the lifestyles of alcohol abusers, such as head injury, psychiatric and other substance abuse co-morbidities, and a higher rate of vascular risk factors [6]. In some countries, terms such as 'alcoholrelated brain damage' or 'alcohol-related brain injury' are preferred over ARD to refl ect the heterogeneity of alcohol-related cognitive disorders in both etiology and clinical presentation [7].

The pathophysiology of alcohol-related dementia
Autopsy evaluations suggest that up to 78% of individuals with diagnosed alcoholism demonstrate some degree of brain pathology [8]. Neuroimaging and neuropathological evidence show prominent white matter loss (most notable in the prefrontal cortex, corpus callosum, and cerebellum) and neuronal loss in the superior frontal asso ciation cortex, hypothalamus, and cerebellum [1,9]. Th e frontal lobes of individuals with diagnosed alcoholism appear particularly susceptible to damage, with evidence of markedly decreased neuron density, volume shrinkage, and altered glucose metabolism and perfusion [10]. Partial recovery of white matter disturbances can occur with abstinence, and magnetic resonance imaging studies indicate early reversibility of white matter shrinkage that is accompanied by clinical improve ment in cognitive and motor abilities [11,12]. Th e mechanism behind recovery from white matter damage is thought to involve the restoration of myeli na tion and axonal integrity, but is vulnerable to repeated disruption if drinking is resumed [11]. Harper [1] concludes that the brain pathology of abusers of alcohol likely has two compo nents: one of permanent change, the other transient.
While these fi ndings refl ect general structural changes in chronic abusers of alcohol, what lesions characterize a clinically identifi able dementia and the neuropathological process that underlies this process remain in dispute. Th e 'neurotoxicity' hypothesis suggests that the direct physiological eff ects of chronic alcohol exposure can cause neuronal loss through glutamate excitotoxicity, oxidative stress, and the disruption of neurogenesis [13]. In particular, drinking patterns of repeated binges and withdrawal periods may enhance neuronal injury through increased vulnerability of upregulated N-methyl-Daspar ate (NDMA) receptors to glutamate-induced excitotoxicity. Support for the neurotoxicity hypothesis emerges from animal studies, which have demonstrated doserelated ethanol-induced damage to brain struc turesincluding the hippocampus, hypothalamus, and cerebellum -that correspond with impairments in memory and learning [14,15]. Cholinergic neuro trans mission in the basal forebrain, which plays a key role in attention, learning, and memory, also appears to be impacted by prolonged intake of alcohol. Imaging studies of 'uncomplicated alcoholics' -individuals with no history of nutritional defi ciency, hepatic failure, or other indirect forms of brain injury -confi rm structural abnormalities, including changes to the corpus callosum, pons, and cerebellum [12]. However, the permanence of such changes, and whether they relate to neurotoxicity in isolation, remains to be established [1].
Another hypothesis is that thiamine (vitamin B 1 ) defi ciency is primarily responsible for the development of ARD. Individuals with alcohol use disorders are at particularly high risk of thiamine defi ciency, not only from poor dietary nutrition but because alcohol directly compromises thiamine metabolism [16]. Th iamine defi ciency can lead to Wernicke's encephalopathy (WE), an acute neurological disorder characterized by the clinical triad of oculomotor abnormalities, cerebellar dysfunction, and altered mental state and by the traditionally defi ned pathology of neuronal loss and hemorrhagic lesions in the paraventricular and periaqueductal grey matter. Not all individuals with WE show the triad of neurological symptoms, and the severity of signs is likely related to the extent of the underlying pathology [17]. To increase diagnostic accuracy of WE, refi ned operational criteria specify a minimum of two symptoms for diagnosis, a guideline recently endorsed by the European Federation of Neurological Societies (EFNS) [18,19]. Long-term outcomes of WE can include development of a syndrome of profound memory impairment -Korsakoff syndrome (KS) -that appears to be related to additional disruption to diencephalic and hippocampal circuitry. As KS shares similar pathological substrates and often follows an episode of WE, it is commonly referred to as the Wernicke-Korsakoff syndrome [1]. Increasing evidence suggests that the WKS encompasses a spectrum of pathological, neurological, and cognitive impairments resulting from thiamine defi ciency [4]. Th e heterogeneity in presentation of the WKS, in combination with a lack of distinct pathological evidence for ARD, has led to the suggestion that cases of ARD are variants of the WKS [20]. Other evidence suggests that ARD and WKS are distinct disorders with overlapping clinical symptoms and associations such as peripheral neuropathology and ataxia [21].

The relationship of Wernicke-Korsakoff syndrome to alcohol-related dementia: pathophysiology
In cases found to have WKS pathology at autopsy, the syndrome was correctly diagnosed in only around 20% of individuals prior to death [22]. Overlap with ARD has been noted, and it has been hypothesized that inactive (chronic) WE is likely to be the main underlying pathology in both KS and ARD [23,24]. Individuals with diagnosed KS and ARD show greater loss of neurons in the nucleus basalis than individuals with uncomplicated alcoholism, although this needs replication [25,26]. Graded defi cits in regional brain volumes, in which 'uncomplicated alcoholics' demonstrate similar but less severe lesions than individuals with WKS (mammillary bodies, thalamus, cerebellar hemispheres, and vermis), also suggest that subclinical bouts of thiamine defi ciency contribute to mild structural changes in alcohol-related disorders [17]. In individuals with diagnosed KS, reports of abnormalities in brain regions beyond those associated with thiamine depletion also provide support for a 'spectrum' of impairment in WKS that likely relates to damage from both thiamine and direct neurotoxicity or a compound eff ect of both [17].
Preliminary reports from animal models suggest that thiamine defi ciency and direct alcohol neurotoxicity produce similar brain eff ects. Th ese include loss of cells in the basal forebrain, hippocampal acetylcholine hypofunction, and shrinkage of frontal grey and white matter, with thiamine defi ciency characterized by additional lesions in the diencephalon [27]. Vetreno and colleagues [27] suggested that the interaction between ethanol and thiamine defi ciency does not produce more behavioral or neural pathology, with the exception of reduced white matter, than long-term thiamine defi ciency alone; however, synergic eff ects have been noted elsewhere [28]. Notably, pure cases of thiamine defi ciency, unaccompanied by chronic and excessive alcohol consumption (such as in cases of malnutrition), show a low rate of progression to KS [29], giving credence to the idea that an interaction of causative factors is responsible for the lasting cognitive defi cits seen in alcohol-related disorders.

How much is too much?
Th e relationship between the amount of alcohol use and cognitive outcomes is complicated by diff ering defi nitions of drinking levels in the literature, and this complication relates in part to the varying defi nitions of a 'standard drink' from country to country. For example, a standard drink in the United Kingdom contains a relatively low 8 grams of alcohol, compared with 10 grams in Australia, 14 grams in the US, and 19.75 grams in Japan [30].
'High' levels of alcohol consumption can range from 10 'standard' drinks a week [31] to more than 9 'standard' drinks a day [32]. Reduced frontal lobe volume has been associated with an amount of 418 grams a week but has not correlated with lower levels of consumption [14]. One review suggested that consumption of fi ve to six drinks per day (which, by US standards, equates to 70 to 84 grams) over extended periods results in 'cognitive ineffi ciencies' , while consumption of 10 or more standard drinks a day manifests as moderate cognitive defi cits equivalent to that found in individuals with diagnosed alcoholism [33].
Th e diff ering elements of drinking patterns (for example, duration and severity of abuse, binge, and withdrawal periods) as well as diffi culties gaining an accurate selfreport of past drinking have further complicated attempts to link drinking levels to later cognitive impairment. Estimates of past drinking habits of individuals diagnosed with ARD have included up to 60 years of drinking (and up to 120 drinks a week at heaviest), although there is signifi cant variability in length and severity of drinking [34]. Oslin and colleagues [35] suggest ed that a fi ve-year history of consuming 35 standard drinks a week for men and 28 for women constitutes a suffi cient level of neurotoxic burden to risk the development of ARD, but this needs verifi cation.
Alcohol has been examined as a risk factor for other dementia syndromes. Th ere are suggestions of a U-or J-shaped relationship between alcohol consumption and dementia, with low-moderate drinking levels reducing the risk of overall dementia but heavy use increasing the risk [14]. Low to moderate alcohol use is thought to reduce the risk of coronary artery disease and ischemic stroke through the inhibitory eff ect of ethanol on platelet aggregation and reduction of infl ammatory markers and by alteration of the serum lipid profi le [36]. Th e antioxidant eff ect of polyphenols or ethanol itself might also provide neuroprotection [14]. Alternatively, heavier drinking may contribute to adverse cerebrovascular changes (hypertension and raised triglycerides) and increased risk of arterial thrombosis, cardiac disorders, and strokes [6]. A meta-analysis examining the association of ethanol and incident dementia concluded that small amounts of alcohol likely protects against Alzheimer's disease but not against vascular dementia [37]. However, others suggest that the benefi t of moderate drinking applies to all forms of dementia [38]. Th is uncertainty was emphasized in a recent 20-year study (n = 1,300 women at least 65 years old), which reported that moderate alcohol consump tion was not protective against dementia. Furthermore, women who increased their alcohol consumption over the course of the study had an increased risk of developing dementia [39]. Interestingly, animal models have shown that low concentrations of alcohol protect cultured cortical and hippocampal neurons against the synapse damage induced by amyloid-β and α-synuclein, providing a pathological explanation for reports that alcohol consumption protects against the development of specifi c dementia syndromes [40].

Nosology?
Th e current DSM-IV (Diagnostic and Statistical Manual of Mental Disorders, 4th edition) criteria for 'alcoholinduced persisting dementia' specify the persistence of cognitive and functional decline following cessation of alcohol consumption, with all other causes of dementia excluded [41]. Oslin and colleagues [35] attempted to improve the validity and reliability of ARD diagnosis by standardizing alcohol consumption criteria for a 'probable' diagnosis of ARD (length and severity of alcohol use) and specifying a minimum abstinence time for a dementia diagnosis to be considered. Th ese guidelines were not meant to be defi nitive and were designed with the intention of stimulating further research. Even so, further use of their criteria has been limited to a handful of studies [42][43][44] that have inconsistently adopted elements of the criteria (for example, some have excluded participants with previous acute symptomatology of WKS) and diff ered in participants' ages, education level, and global cognitive function. Th us, current diagnostic criteria for ARD have been inadequately tested.
WKS is classifi ed in the DSM-IV as 'alcohol-induced persisting amnestic disorder' , and memory disturbance is the key diagnostic feature [41]. International Statistical Classifi cation of Diseases and Related Health Problems (ICD-10) criteria for 'alcohol-induced amnesic syndrome' [45] are more descriptive and detail impairment of recent memory and disturbance of time sense in the absence of impaired immediate recall or generalized cognitive impairment.
Another term that is becoming widely used is 'alcoholrelated brain damage' (ARBD) (also called alcohol-related brain impairment), which in some countries has superseded the use of DSM-IV nosology [7]. Some authors do not encompass specifi c neurological disorders that present with a distinct clinical and neuropathological presentation (for example, Marchiafava-Bignami disease) within this terminology [20], whereas others group these under the ARBD banner [7]. For many clinicians, this term is preferential to ARD because it better accounts for the heterogeneity of presentations, avoids the stigma associated with the term dementia, and distinguishes the non-progressive nature of ARD from other degenerative disorders. DSM-5 looks likely to adopt a similar inclusive approach, with plans to have categories of major and minor 'neurocognitive disorder due to substance disorder' , which are likely to include an 'amnesticconfabulatory (Korsakoff )' subtype, although this is yet to be fi nalized [46].

Prevalence of alcohol-related cognitive disorders
Given the lack of operationally defi ned diagnostic criteria, it is not surprising that incidence and prevalence estimates of ARD vary in the literature. Epidemiological fi ndings generally have been derived from population studies that relate patterns of alcohol consumption and dementia. A review by Ritchie and Villebrun [47] established that studies have indicated a high prevalence of alcohol abuse in patients with dementia (9% to 22%) and high rates of dementia in alcohol abusers (10% to 24%), although most studies did not specify the type of dementia. Th e range in rates may be explained at least partially by diff erences in criteria for 'heavy' alcohol use, varying age limits, and diff erences in sampling. ARD cases generally have a younger age of onset, and consequently studies that exclude those under 60 years of age may miss a signifi cant proportion of cases [47]. Prevalence studies of dementia subtypes in nursing homes have reported ARD to account for 10% to 24% of all dementias [43,47,48], which is likely higher than in the general population. Rates of ARD in dementia cases identifi ed in neurology and memory clinics tend to be lower (around 3% to 5% [49,50]), which may indicate the lack of referrals of these patients to such clinics. Rates of ARD of around 10% were found in an English epidemiological study of younger-onset dementia in specifi c London districts (onset of less than 65 years) [51]. Th e eff ect of age was demonstrated in an Australian analysis of hospital admissions of more than 20,000 dementia patients who were at least 50 years old: ARD was found in 1.4% of all patients with diagnosed dementia but in 22% of dementia patients under 65 [52].
Individuals with ARD are often male, have co-morbid mental and physical conditions (including liver and digestive diseases), and are likely to be identifi ed through hospital admissions [43,52]. Social isolation appears to be a signifi cant factor in the poor identifi cation and treatment of ARD, and a high proportion of patients are unmarried or do not have the support of family or friends [43,53]. Th ese gender and social fi ndings are consistent with reported characteristics of individuals who are heavy users of alcohol [2]. Th ere has been little examination of the prevalence of co-morbid substance abuse, head injuries, or psychological co-morbidities in the ARD popu lation despite evidence that these are linked to the presence and maintenance of substance use disorders in both younger and older adults [13,54].
Most cases of WKS in developed countries relate to the misuse of alcohol, although WKS syndromes following gastrointestinal disorders and systemic diseases can also contribute. While there is no direct correlation between the prevalence of WE and per capita consumption of alcohol, the introduction of thiamine supplementation programs in some countries, as well as general dietary habits, also infl uences overall rates [16]. Prevalence rates of WKS identifi ed post-mortem are thought to be between 1% and 2% of the general population and around 10% of alcohol misusers in Western countries [16,19]. A study of KS in Th e Netherlands reported a prevalence of 48 per 100,000 inhabitants [55], and incidence rates of KS in the East End of Glasgow, Scotland, were estimated at around 8 per 100,000 in 1995, a seven-fold increase from 1990 [56]. A study of hospital admissions of patients at least 50 years old identifi ed 126 cases of KS (0.05% of all admissions) and 77 cases of WE (0.03% of admissions), although there was some overlap in diagnostic groups [52]. Th ere is a need for further epidemiological study of this population by using standardized criteria for diagnosis to increase accuracy in identifi cation of underlying WKS neuropathology and allow overall prevalence rates to be established.

Neuropsychological fi ndings
Alcohol can have acute and chronic eff ects on cognitive function. Direct intoxication impairs most cognitive skills and in excess may lead to stupor and respiratory depression. Acute withdrawal in long-term alcohol abusers can result in tremor, hallucinations, seizures, agitation, and fl uctuating levels of alertness [14].
Cognitive and behavioral changes specifi c to ARD have received limited investigation. Unlike patients with other dementia syndromes, patients with alcohol-induced dementia are unlikely to demonstrate language impairment, an observation incorporated into the ARD criteria proposed by Oslin and colleagues [21,35]; however, preserved confron ta tional naming has not been shown in all studies [44]. In comparison with individuals with Alzheimer's disease, ARD groups generally performed better on semantic tasks (confrontational naming, category fl uency, and general knowledge) and on verbal memory recognition measures despite generally equivalent verbal learning and overall delayed recall to Alzheimer's disease groups [34,44,57]. However, the ARD groups had poorer performance on visuospatial measures, including clock drawing and copy ing tasks. Defi cits on tasks of working memory [44], motor speed [34], and executive function (verbal abstract reasoning [44] and letter fl uency [34]) have also been observed in ARD samples in comparison with healthy controls. While fi ndings are limited by small samples (which in some cases overlapped between studies) as well as diff erences in group characteristics (that is, global dementia severity), Munro and colleagues [57] proposed that the clinical profi le of ARD refl ects both cortical and subcortical pathology. Th is was supported by a recent SPECT (single-photon emission computed tomography) study that reported reduced regional cerebral blood fl ow in the frontal cortices, basal ganglia, and thalami of patients with ARD [42].
Th e neuropsychology of WKS has been the subject of more extensive investigation. Patients typically demonstrate profound anterograde amnesia and impaired recall of past events, with a temporally graded defi cit in which recall is better for more remote time periods [58]. Implicit memory and procedural memory are comparatively spared. Other cognitive functions apart from memory may be disturbed, and impaired executive functions, visuoperceptual diffi culties, and disturbed working memory have been observed [59]. Executive defi cits have been identifi ed in 80% of patients with KS [60]. Difficulties are most frequently detected on tasks assessing higher-order organization, planning, and cognitive fl exibility (for example, verbal fl uency and divided attention) [60,61]. Th ere is also evidence for variable intellectual function in WKS [62]. In a review of evidence for variability in WKS, Bowden [4] remarks that empirical evidence suggests that the chronic phase of WKS is more accurately described as 'dementia-like deterioration' rather than severe and selective amnesia.
Abstinence for as little as a week typically resolves many of the defi cits associated with heavy alcohol consumption, and further recovery of cognitive abilities can continue over several years. Th e pattern and rate of cognitive recovery are not yet fully understood; however, there is some suggestion that verbal defi cits resolve faster than visuospatial diffi culties [13]. Executive function, working memory, and perceptual and motor impairments commonly endure following short-term abstinence, which has been proposed to partly refl ect compromised frontocortico-cerebellar functional networks [10]. Recovery of cognitive skills appears to be linked to amount of recent alcohol use and duration of abstinence rather than lifetime alcohol consumption. Multiple withdrawals and binge drinking may signifi cantly exacerbate cognitive defi cits [32,63]. Older drinkers show greater alcoholrelated cognitive changes and are less likely to recover function once they cease drinking, even after drinking history is controlled for [13]. Other factors such as gender and education also likely contribute to vulnerability to cognitive impairment, and lower levels of education are associated with less recovery over time and females appear to be more vulnerable to cognitive impair ment despite generally lower drinking levels than men [13,63]. It is unclear whether education is a protective factor or whether low pre-morbid intelligence is a risk for both cognitive impairment and poor educational attainment [13]. It has also been suggested that genetic infl uences -such as a family history of alcoholism -may be a risk factor for development of alcohol use cognitive disorders, with cognitive dysfunction potentially predating alcohol misuse [13].
Another feature common to ARD and WKS is stability, and even improvement, in cognitive functioning over time provided that abstinence is maintained. A two-year follow-up of nursing home residents with dementia reported that patients with ARD demonstrated a stabilization of cognitive and functional status, as opposed to individuals with Alzheimer's disease or vascular dementia who experienced a general decline [43]. Th is is consistent with reports from a Scottish study [64] of stability and even improvement in some cognitive domains for an ARD group over a one-year follow-up. Similarly, indi viduals with WKS also have potential for recovery, although this is not commonly recognized [4]. Improvements in general knowledge, visual long-term memory, and verbal fl uency have been found in abstinent individuals with KS over two years, and cognitive improvement is associated with higher pre-morbid education and fewer detoxifi cations in the past [65]. Early observations by Victor and colleagues [66] also suggest that a large proportion (21%) of individuals with KS can make a full recovery. Th ese clinical fi ndings are consistent with current neuroimaging evidence that suggests the possibility of at least partial structural and functional recovery from alcohol-related brain damage if abstinence is maintained. A recent animal study reported that while the eff ects of chronic ethanol exposure (including working memory and episodic memory impairment) can recover with prolonged abstinence, the defi cits of thiamine defi ciency (spatial memory impairment and increased perseverative behavior) are more persistent [28].

Clinical considerations
Th e assessment and management of individuals with ARD and WKS entail a number of clinical issues. Firstly, a thorough nutritional and drinking history should be taken, with confi rmation from an informant if possible. Secondly, all individuals with any evidence of chronic alcohol misuse and suspected of having WE should be treated immediately with parenteral thiamine [16]. Treatment with oral thiamine is ineff ective because it does not achieve an adequate plasma concentration [67]. Furthermore, alcohol directly interferes with thiamine uptake [67]. While there is no consensus as to the optimum dose, frequency, route, and duration of thiamine treatment, the EFNS recom mends that in cases of suspected WE, thiamine be given in doses of 200 mg three times daily, preferably intravenously. Th is treatment should be continued until no further improvement in signs and symptoms is evident [19]. Th irdly, assessment of cognitive status should be conducted on an ongoing basis as this will allow any improvement, stabilization, or deterioration to be detected. Acute intoxication and withdrawal may exacerbate cognitive defi cits, so assessment following this period (which usually lasts no longer than 2 weeks) may allow a more accurate baseline to be established. Fourthly, key characteristics associated with alcohol-related cognitive disorders may assist with diff eren tiation from neurodegenerative conditions. Th ese typically involve stabilization or improvement in cognition with abstinence; a cognitive profi le involving executive, visuo spatial, and memory diffi culties with spared language function; and neurological symptoms such as ataxia. Neuroimaging may suggest atrophy in the mammil lary bodies, thalamus and cerebellum, and ventricular enlargement, although this may vary from case to case.
Patients with ARD and WKS have shown cognitive improve ment following treatment with memantine, although these fi ndings require replication [68,69]. Finally, these socially isolated patients are often hospital ized for another health condition and this presents an ideal opportunity for screening, identifi cation, and intervention.

Conclusions
Th e evidence refl ects signifi cant commonality between ARD and WKS. Neuropsychological studies have largely attempted to diff erentiate these syndromes by limiting individuals with more global cognitive impairment from WKS investigations and by excluding individuals with past symptoms of WKS from ARD studies. However, the validity of this distinction is now being brought into question. While attention has been given to the notion that ARD is likely related to underlying WKS pathology, it appears strange that more credence has not been given to whether WKS may clinically be a form of dementia, given that it typically entails multiple domains of cognitive impairment alongside functional defi cits. While 'dementia' in current neurological settings is typically used to describe a progressive disease of the brain, it perhaps more accurately encompasses a deterioration of intellectual or cognitive function that may or may not be progressive in nature [70]. Debate over the utility of the term ARD in the clinical setting ensues [4]. It remains to be seen whether the term 'neurocognitive disorder' will satisfactorily resolve these nosological issues, dispel some of the stigma associated with this illness, and bring a better understanding of alcohol-related cognitive disorders.