OUP user menu

Wernicke's encephalopathy: ‘plus ca change, plus c'est la meme chose’

Allan D. Thomson, Christopher C. H. Cook, Irene Guerrini, Donna Sheedy, Clive Harper, E. Jane Marshall
DOI: http://dx.doi.org/10.1093/alcalc/agm149 180-186 First published online: 24 October 2007


Aims: To develop clinical guidelines to identify individuals who misuse alcohol and are at risk of developing Wernicke's Encephalopathy (WE). Method: Non-systematic literature review of studies which includes a careful clinical record of the development of signs and symptoms of thiamine deficiency and in which the diagnosis of WE has been confirmed at autopsy. Results: The review of the clinical findings in cases of WE, diagnosed at autopsy, shows a consistent pattern of signs and symptoms. The pattern appears to be similar regardless of whether the thiamine deficiency is related to nutritional problems alone or associated with alcohol misuse. Conclusions: The assessment of the degree of thiamine deficiency and the diagnosis of WE remain a clinical evaluation, and guidelines are suggested to help the clinician. Since neurotoxicity due to the metabolism of excessive alcohol in patients with chronic and severe alcohol dependence may be an important factor in determining long-term outcome of treatment, this must form part of the overall evaluation.


Wernicke's Encephalopathy (WE) is a neuropsychiatric condition that can lead to dementia and occurs as a result of thiamine (vitamin B1) deficiency. It has classically been described as presenting with an acute onset, characterized by nystagmus, ophthalmoplegia, ataxia and global confusion, occurring together or in various combinations (Lishman, 1998). WE can be prevented and treated successfully before the onset of irreversible brain damage. It is thought that patients can suffer recurrent episodes of WE, some of which might be subclinical, leading to a more chronic form of the disease (Lishman, 1998). Many of these patients develop a severe amnesic syndrome that is known as Korsakoff's psychosis (KP) (Lishman, 1998). Autopsy studies have shown that the diagnosis of WE and KP is not made during life in 80% of the cases (Harper et al., 1986). Although a great deal has been learnt about the natural history and pathophysiology of WE (Thomson and Marshall, 2006a), early diagnosis still depends upon clinical judgement and experience.

While only approximately 16% of patients with WE have the classic triad of signs (confusion, ataxia and eye signs) described above (Harper et al., 1986; Naidoo et al., 1991), there are often early clinical indicators that the patient is thiamine deficient and at risk of developing WE. Unfortunately, these indicators are not widely known and are often attributed to the effects of excessive alcohol intake (Thomson and Marshall, 2006a). Torvik (1991) commented that ‘The wide spectrum of the clinic symptoms has not been fully appreciated and this may in part explain the low level of diagnostic accuracy of the disease’. In reviewing the literature, we have been impressed by the consistency of the reported signs and symptoms attributed to WE, yet clinicians continue to focus on the triad. In exploring why this is so, we have taken an historical/investigative perspective on the situation and returned to the primary literature sources. A new English translation of Wernicke's original German paper was organized by one of us, and published with a commentary (see Thomson et al., this volume). We further compared Wernicke's original findings with those from a number of subsequent studies of patients with WE, either due to malnutrition alone or in association with alcohol misuse.

The combination of thiamine deficiency and alcohol misuse is a common scenario—in some studies almost 90% of cases of WE have a history of alcohol misuse (Harper et al., 1995). It is important to identify this association with alcohol because it appears to increase significantly the amount of thiamine required to treat the patient successfully compared to individuals with a predominantly nutritional thiamine deficiency. It is also necessary, when alcohol misuse is involved, to give the thiamine parenterally because of alcohol-related malabsorption of thiamine across the intestinal mucosa (Thomson et al., 1970; Thomson, 2000). It is possible that the other thiamine transport systems, particularly those present in the brain, may be similarly affected (Thomson and Marshall, 2006a). Since alcohol neurotoxicity can also interfere with the effective utilization of thiamine, it may be critical to intervene early in thiamine-deficient individuals with alcohol misuse. In other words, success in preventing the amnesic aspects of Wernicke-Korsakoff Syndrome (WKS) requires prophylactic thiamine replacement together with reduction or cessation of alcohol intake at the earliest possible time. With this in mind, we will suggest guidelines for early intervention.

Wernicke's Clinical Descriptions

The clinical findings from Wernicke's original descriptions of patients are summarized in Table 1.

View this table:
Table 1

Recorded clinical findings in patients whose diagnosis of WKS has been confirmed at post-mortem The percentages refer to the frequency with which the signs/symptoms appear in each group have been included where available


The Correlation between Clinical Signs/Symptoms of Wernicke's Encephalopathy and Autopsy Findings

The concept of a ‘classic triad’ of signs and symptoms in WE was based on the original description by Wernicke (1881). The triad consists of an abrupt onset of a confusional state, impairment of consciousness, ataxia and eye signs (nystagmus and ophthalmoplegia). Victor et al. (1989) in their famous monograph concluded: ‘The diagnosis of Wernicke's disease is made most readily on the basis of the acute appearance of ocular palsies, nystagmus, ataxia of gait, and disturbances of consciousness and mentation, which may present singly or in various combinations. More than 80% of patients show signs of polyneuropathy as well, and associated liver disease is found in two thirds of the patients’ (Victor et al., 1989, page 195).

However, as Wernicke himself described and, in subsequent studies, other important clinical signs and symptoms are often present before the later ‘classical'signs appear.

In Vivo’ Clinical Findings in Patients with Wernicke's Encephalopathy, Later Confirmed by Post-Mortem Diagnosis

Table 1 summarizes the clinical findings in the patients reported by Wernicke and Korsakoff, together with those from other studies in which autopsy confirmation of the diagnosis of WE was made. We begin with the premise that: ‘The pathological features of WE are characteristic with little likelihood that the wrong diagnosis could be made’ (Harper et al., 1995). The papers are presented below, in chronological order. In Table 1 symptoms and signs were recorded as being present if they were recorded in any of the patients. Unfortunately, the information available did not allow us to express this as a frequency, except in the papers of Harper et al. (1986) and De Wardener and Lennox (1947), where the percentages have been added to the table. The symptoms and signs selected for recording are, to some extent, arbitrary and dependent upon the knowledge and specialization of the clinicians who saw the patients and what they considered to be the diagnosis at the time. Despite this, there seems to be consistency in those signs and symptoms that were recorded.

Campbell and Biggart (1939) from Edinburgh described 12 cases of WE, aged between 3½ and 68 years: only one was alcoholic (3 male; 9 female). Riggs and Boles (1944) studied 42 cases from Philadelphia (13 men) between 1934 and 1943: only 18 of the patients who ‘volunteered a history of heavy alcohol intake were judged to be alcoholic’. In 1947, De Wardener and Lennox wrote a famous paper on their experience of 52 male soldiers, held as prisoners-of-war by the Japanese, who developed WE due to dietary thiamine deficiency. Barrie (1947) also described WE occurring in three patients in Sheffield. Cravioto et al. (1961) described 28 cases of WE from Bellevue Hospital, New York and provided detailed autopsy findings on the 18 men and 10 women aged between 32 and 82 years (mean 54 years). In 1969, Grunnet compared the earlier findings from Riggs and Boles (1944) series with a later (1965–68) series from the same hospital in Philadelphia.

Victor and colleagues collected a substantial series of patients with WE over several years in Boston, USA, and reported their preliminary findings in 1971 (Victor et al., 1971). Of their final series of 245 patients with WE, 82 had a complete autopsy examination (Victor et al., 1989). These patients, 154 men and 91 women, aged 30–70 years, were predominantly in their fifties to sixties; most were heavy drinkers and malnourished. This work was published as a monograph, sadly now out of print, containing the clinical and pathological findings of these patients. The unique value of Victor et al's (1971) observations was that a substantial proportion of the series survived the acute stage of WE and remained under observation for many months or years. One of us, Clive Harper, a neuropathologist, continued work on WE in Australia, publishing a series of pioneering studies which have done much to establish the incidence of WE in alcoholics and the general population around the world. This work demonstrated that the diagnosis is only made clinically in about 20% of cases prior to autopsy (Harper et al., 1986). Clinical diagnosis prior to autopsy was shown to be as low as 1% in Torvik's (1991) review of autopsies in the United States, Oslo, Norway. Wallis et al. (1978) described four alcoholic patients (two male) who presented in a comatose state.

All of these retrospective studies were based upon the clinical records of patients who often presented late in the course of their illness and in whom the diagnosis was clinically uncertain. It should be emphasized that the clinical signs and symptoms in these patients were usually recorded before the definitive diagnosis was made. In other words, the doctors recorded what they observed and considered relevant often without making the correct diagnosis.

Torvik et al. (1982) described their findings from 8735 autopsies performed over a 5-year period in Oslo. They documented 70 cases of WE (0.8%) and 152 cases of alcoholic cerebellar atrophy (1.7%). The latter was found to be present in 26.8% of all alcoholic patients examined. Twenty-two of their WE cases were active (acute or subacute) and 48 were chronic. Stupor and coma were clinically documented as the dominant symptoms. This study was followed, in 1989, by a paper from Lindboe and Loberg who reported a 5-year study of 6964 autopsies with 52 cases of WE, of which 12 (23%) had occurred in non-alcoholics. There were 18 acute cases in their study and 7 of these (39%) were not alcoholics. Only four of the active (acute) cases of WE had been diagnosed clinically and all of these were alcoholics. The predominant clinical symptoms in this series were disorientation and depressed levels of consciousness, with eye signs being recorded in only three cases. Thiamine was often given too late, prompting the authors to comment that even the slightest suspicion of WE should initiate the immediate parenteral administration of large doses of thiamine. In 2005, Fattal-Valevski et al. reported findings on nine infants (six male; three female, aged 2–12 months) who had received a soy-based formula that was deficient in thiamine. The babies responded promptly to treatment with thiamine but unfortunately two deaths occurred (Vikhanski, 2004). We have also included previously unpublished data on 44 patients with WKS diagnosed on post-mortem which one of us, Clive Harper, has collected more recently.

What These Studies Tell Us about the Clinical Diagnosis of Wernicke's Encephalopathy

The reports from all these studies show a remarkable similarity in patterns of signs and symptoms, even though some patients were not exposed to alcohol. However, care must be taken with the interpretation of these studies as the findings are dependent, in part, on how accurately and carefully the clinical findings were recorded, especially when the diagnosis was not made. Torvik et al. (1982) have emphasized that the classic symptoms of WE may not always be as prominent and striking as generally believed (Victor et al., 1971). Stupor and coma, which have been said to be rare in WE (Victor et al., 1971; Plum and Posner, 1980) develop rapidly and were the main findings in their case series. In their view, this did not imply that there were two different populations of WE.

In Figure 1 we have summarized the development of signs and symptoms of WE in the prisoners-of-war described by De Wardener and Lennox (1947). They also suggested a method for grading the severity of the clinical picture of WE which is highly relevant to the early diagnosis of WE. It is clear from the figure that the mild, moderate and severe/terminal phases of WE have distinctive signs and symptoms. Thus, mild WE was characterized by anorexia, nausea, vomiting and eye signs; the moderate phase by memory loss and emotional changes; and the severe phase by confusion, confabulation and coma. What is even more interesting is that de Warderner and Lennox observed that these symptoms were reversed in approximately the same chronological order following treatment. Unfortunately, we do not have such a carefully documented development of signs/symptoms for alcoholic patients who often present later in the illness and in whom the findings may be modified by alcohol withdrawal or by hepatic encephalopathy (Caine et al., 1997)

Fig. 1

Development of signs/symptoms in non-alcohol related Wernicke's encephalopathy.

In Table 2 we summarize factors predictive of thiamine deficiency and related neurotoxicity (Thomson and Marshall, 2006a). It includes the symptoms and signs associated with early thiamine deficiency and evidence of malnutrition together with factors predisposing to neurotoxicity. This is relevant since there is evidence that alcohol permanently damages the thiamine-depleted brain to a greater degree than thiamine depletion alone (Price et al., 1993; Ciccia and Langlais, 2000; Crowe and El-Hadj, 2002). For further discussion the reader is referred to Thomson and Marshall (2006a).

View this table:
Table 2

Patients at risk of thiamine deficiency and/or neurotoxicity

Factors predisposing to thiamine deficiency
• Weight loss in past year
• Reduced Body Mass Index
• General clinical impression of nutritional status
• High carbohydrate intake
• Recurrent episodes of vomiting in the past month
• Co-occurrence of other nutritionally related conditions (polyneuropathy, amblyopia, pellagra, anaemia)
Factors predisposing to neurotoxicity
• Genetic predisposition to alcohol dependence and neurotoxic effects of alcohol
• Quantity/frequency of alcohol use
• Severity of dependence
• Frequent episodes of acute intoxication
• Withdrawal syndromes
• Concurrent drug use, particularly cocaine
• Alcohol-related liver damage
Early signs-symptoms of thiamine deficiency
• Loss of appetite
• Nausea/vomiting
• Fatigue, weakness, apathy
• Giddiness, diplopia
• Insomnia, anxiety, difficulty in concentration
• Loss of memory
Operational criteria for Wernicke Encephalopathy Two of the following four signs according to Caine et al. (1997):
• Dietary deficiencies
• Oculomotor abnormalities
• Cerebellar dysfunction
• Either altered mental state or mild memory impairment
De Wardener and Lennox (1947)
Caine et al. (1997)
Thomson and Marshall (2006a)

The Diagnosis of Patients at Risk and Those Who Have Developed Wernicke's Encephalopathy

An important advance in the diagnosis of WE was made by Caine et al. (1997) when they developed operational criteria to differentiate between WE alone or in combination with Korsakoff's psychosis or hepatic encephalopathy. The criteria for WE require two of the following signs: (i) dietary deficiency (ii) oculomotor abnormalities (iii) cerebellar dysfunction and (iv) either altered mental state or mild memory impairment. This retrospective study, in which the diagnoses were confirmed at autopsy, showed that using the operational criteria, antemortem identification of WE can be achieved with a high degree of specificity, although the accuracy of diagnosis was reduced when hepatic encephalopathy was present. In our present article, we are concerned with the early identification of patients at risk of developing WE and who have developed the prodromal symptoms of thiamine deficiency. Approximately 10% of patients die in the acute phase of WE and between 56–86% may develop KP when given inadequate parenteral thiamine (Victor et al., 1989; Thomson and Marshall, 2006a). In Table 1, we have seen some of the early prodromal symptoms of thiamine deficiency, their diagnosis having been confirmed at autopsy. If these symptoms are taken in conjunction with other predisposing factors to WE, it may be possible to identify patients who need parenteral thiamine prophylaxis. This would protect them from developing WE and from dying in the acute phase of the disease or being at risk of developing KP or some other less well-defined degree of brain damage. Caine et al. (1997) carefully selected their subjects to include those drinking more than 80 g of ethanol per day for most of their adult life, with neurological and clinical assessments at more than one point, full necropsy, interview with general practitioner or relatives and systematic neuropathological examination to exclude conditions other than those directly attributable to chronic alcohol consumption. In Table 3, we compare the clinical findings from a number of other studies. There is a large variation in the number of cases studied in each group as the collection of clinical data was not always meticulous, patients presented at different stages of their disease, and some received thiamine therapy. However, there is general agreement about the importance of these signs.

View this table:
Table 3

Incidence of clinical signs in WKS grouped according to Caine et al. (1997). Information on dietary deficiency was not complete

AuthorCases (no.)Eye signs (%)Ataxia (%)Mental changes (%)
Harper (unpublished data)4432579
Fattal-Valevski et al. (2005)9332256
Lindboe and Loberg (1989)5221NA6
Harper et al. (1986)97293776
Victor et al., (1989)163/232100100–8199
Wallis et al. (1978)40NA100
Groen and Hoff (1977)50804688
Grunnet (1969)24251375
De Wardener and Lennox (1947)52100NA100
Cravioto et al. (1961)28321893
Barrie (1947)3670100
Riggs and Boles (1944)42241086
Campbell and Biggart (1939)1242092

As new techniques of neuro-imaging are developed and resolution increases, anatomical structures like the mammillary bodies can now be visualized, thus aiding the clinical diagnosis of WE (Aupe et al., 2001; Celik and Kaya, 2004; White et al., 2005). However, these are unlikely to help us identify patients at risk. Recent investigations have identified an abnormal gene present in some patients with WE, which is involved in thiamine transport and may, if future work confirms these findings, be the basis of a diagnostic test (Guerrini et al., 2005). It has become apparent that recovery following thiamine replacement may be limited by too small a dose of thiamine and/or by thiamine being given too late, or via the wrong route (Thomson et al., 2002). However, the neurotoxicity which occurs as a result of heavy alcohol use, and possibly the syndrome of alcohol dependence itself, probably explain why patients misusing alcohol require much larger doses of thiamine than non-alcoholics and often do not recover completely (Thomson and Marshall, 2006a).

We have prepared some guidelines for identifying early thiamine deficiency and to aid the doctor in predicting whether the patient has WE (Table 2).

Although we have concentrated on thiamine replacement, it must not be forgotten that many nutrients are required for normal brain function and that alcoholic patients may need a wide spectrum of nutritional therapy (Bonner et al., 2004).


Review of the clinical findings in cases of WE that have been diagnosed at autopsy show a consistent pattern of signs and symptoms. The patterns appear to be similar regardless of whether the thiamine deficiency is related to nutritional problems alone or associated with alcohol misuse. Ideally, studies would need to be set up to evaluate both alcoholic and non-alcoholic patients carefully to list the order in which symptoms appeared, as well as extremely detailed clinical (including neurological and psychiatric) ongoing assessments. The groups would need to be large enough to allow for the varying stages in the disease process at which the patients present, and a meticulous neuropathological examination would need to be carried out on all patients. Long-term follow-up is required to evaluate the effectiveness of any therapy given to the patient. However, it is also evident that in approximately 80% of cases, the diagnosis of WE was not made prior to death. This means that it is virtually impossible to establish accurate prevalence rates of WE in various populations without significant numbers of autopsies. In recent years, there has been a marked decline in the number of hospitals where Coroner's autopsies have been carried out (Jeganathan et al., 2006). Even in situations where autopsies are performed and the brain is examined, often no sections of the appropriate brain areas are taken. Many families object to an autopsy being carried out, and death certificates are given which would have previously have required an autopsy. There seems to be a general lack of interest by many clinicians for hospital autopsies to be performed (The Royal College of Pathologists of Australasia Autopsy Working Party, 2004). Indeed, in the United Kingdom there is concern about the low numbers of consultant pathologists and the crisis regarding recruiting them into the specialty (United Kingdom Parliament, 2006). It is therefore clear that we will never be able to establish the true incidence of WE by this method, nor how many patients with WE have remained undiagnosed. However, it is possible that future neuro-imaging techniques will be able to diagnose the lesions accurately and thus facilitate treatment. During the past 15 years, attempts have been made to develop post-mortem MR imaging (Pfefferbaum et al., 2004).

The responsibility of the clinician is to identify patients at risk of WE as early as possible and to institute effective prophylactic therapy (Thomson and Marshall, 2006b). The diagnostic criteria in Table 3 will aid in making this assessment. The extent to which the early signs of thiamine deficiency reflect biochemical changes in the body, and when they begin to affect the human brain, remain speculative. Studies in rats, however, have demonstrated that a single episode of thiamine deficiency can selectively damage cortical white matter tracts, while at the same time sparing the thalamus or mammillary bodies, and may be responsible for the behavioural changes observed in alcoholic patients without WE (Langlais and Zhang, 1997).

Doctors have come to rely too much on the ‘classic triad’ of signs for diagnosing WE which occurs only in 10% of patients, although it has been known for many years that other indications suggest that the patient is at risk. It remains to be demonstrated that our early intervention will be effective in preventing permanent brain damage. We will no longer have the gold standard of autopsy changes to help us establish the optimum dose of thiamine required to prevent or treat WE, the duration of treatment necessary, nor the route of administration required. Careful long-term follow-up studies will be required to demonstrate that the patients are not left with KP or are otherwise cognitively impaired. Wernicke could be forgiven for thinking that 120 years after he correctly diagnosed his patients, not much has changed!


View Abstract