OUP user menu

High ethanol intake and malnutrition in alcoholic cerebellar shrinkage

J.M. Nicolás, J. Fernández‐Solà, J. Robert, E. Antúnez, M. Cofán, C. Cardenal, E. Sacanella, R. Estruch, A. Urbano‐Márquez
DOI: http://dx.doi.org/10.1093/qjmed/93.7.449 449-456 First published online: 1 July 2000


To determine the influence of chronic ethanol intake and nutritional status on cerebellar shrinkage in alcoholism, we studied 12 undernourished patients with acute Wernicke's encephalopathy (WE), 12 undernourished and 24 well‐nourished asymptomatic chronic alcoholics, and 24 age‐matched well‐nourished controls, using morphometric analysis of MRI scans with volumetry of the cerebellum. Alcoholics reported a mean daily intake of ethanol of 177±8 g over a period of 27±1 years. Most undernourished alcoholics and half of the well‐nourished alcoholics, compared to one‐tenth of the controls, showed a significant reduction in cerebellar volume (p≤0.01, both). Alcoholics with cerebellar shrinkage (n=33) were older (p=0.05) and tended to report greater daily ethanol intake than alcoholics without cerebellar shrinkage (n=15), although not significantly so (p=0.09). Cerebellar volume correlated negatively with age in controls and asymptomatic alcoholics (r≥0.52, p≤0.01, both), with a significantly greater shrinkage for age in the latter (p=0.003). Logistic regression analysis showed that malnutrition (OR 6.6 [95%CI 1.7–25.6], p=0.005) and a daily ethanol intake of more than 140 g over ten years (OR 6.1 [95%CI 1.8–20.5], p=0.003) were independently associated with the development of cerebellar shrinkage.


Ethanol misuse has been related to a wide range of deleterious effects on the central nervous system.1–,5 Besides the frequently reported effects of ethanol on supratentorial brain, cerebellar involvement is also commonly observed in chronic alcoholics.6–,8 Some 25–50% of severe alcoholic patients show cerebellar shrinkage in necropsy studies, predominantly involving the superior vermis, due to dendritic atrophy and loss of Purkinje cell density.7–,9 Previous qualitative neuroradiological studies have documented cerebellar and vermian shrinkage in patients with Wernicke‐Korsakoff syndrome, as well as in a variable portion of asymptomatic alcoholics.9–,11 However, these studies included patients with different nutritional statuses, and did not take into account the severity of alcoholism and the presence of liver cirrhosis. Thus, the relationship between cerebellar degeneration and nutrition, vitamin deficiencies and the frequency and amount of ethanol consumed remains uncertain. We, therefore, aimed to evaluate how these factors, in controlled conditions, influence the development of cerebellar shrinkage in chronic alcoholics, assessing in particular the effect of acute Wernicke's encephalopathy.


Patient and control selection

Wernicke's encephalopathy

We included 12 male patients under 65 years with acute Wernicke's encephalopathy (WE) only attributable to alcoholism, consecutively admitted to our hospital from 1994 to 1996. Diagnosis was based on clinical features of acute WE and confirmed by definite improvement under thiamine therapy. All these patients were undernourished.

Asymptomatic chronic alcoholics

Patients under the age of 65 years, who reported a daily ethanol consumption of at least 100 g for the previous eight years were randomly evaluated from a total of 710 patients seen in the Alcoholism Unit of our Institution who fulfilled the diagnostic criteria for chronic alcoholism.12 We selected patients based on the presence/absence of malnutrition (see below). Out of 43 patients initially selected, seven were excluded: three patients because of liver cirrhosis, two subjects presenting withdrawal syndrome, one case for being positive for the human immunodeficiency virus, and another for using illicit drugs. Thus, 12 undernourished and 24 well‐nourished asymptomatic chronic alcoholic patients were included in the study.


This group was comprised of 24 well‐nourished asymptomatic males who did not report ethanol consumption. They were recruited from friends and relatives of the alcoholics. This group was matched for age (±3 years) with the group of well‐nourished alcoholics, and the members were studied in the same manner as the alcoholics.

None of the patients and controls showed causes of brain damage other than alcoholism, and none reported any events of head trauma. No patient or control objected to inclusion in the study, and all gave informed consent for the various procedures. Patients and controls were all Caucasian males of Spanish descent who lived with their families in or around Barcelona and had histories of stable employment. Only three patients with WE were indigent. The study protocol was approved by the Institutional Review Board of the Hospital Clínic.

Clinical studies

A complete neurological examination was performed in all subjects. A detailed history of ethanol intake and dietary habits was obtained for each patient using a structured questionnaire. Data were confirmed in consultation with family members. Patients with acute WE were interviewed regarding improvement of symptoms. Both the frequency and amount of ethanol intake were recorded. Life events such as marriage, military service and work posts were used as ‘anchor points’ to help data recollection (‘time‐line follow‐back method’13). Withdrawal symptoms were evaluated according to the Clinical Institute for Withdrawal Assessment scale.14 None of the patients or controls received any medication before or during the investigation, except for thiamine supplementation in WE patients.

Laboratory and nutritional studies

Blood samples were obtained the day following admission for measurement of markers of alcohol intake and nutritional status. Patients were considered to have caloric malnutrition if their body weight was less than 90% of their ideal weight or if the calculated lean body mass was more than 10% below the normal value. Protein malnutrition was diagnosed when the patients had abnormal values for three of the following parameters: haemoglobin, lymphocyte count, total protein, albumin, prealbumin, retinol‐binding protein, and transferrin.15,,16 Patients were considered undernourished/well‐nourished in the presence/absence of both caloric and protein malnutrition. Thiamine deficiency was determined indirectly by measuring erythrocyte transketolase activity (TPP effect). In addition, serological assays for the Venereal Disease Research Laboratory test, Fluorescent Treponemal Antibody Absorption test, and human immunodeficiency virus antibodies (ELISA) were performed.


MR imaging of the brain was obtained in all alcoholics within 48 h of hospital admission, with all WE patients being studied within the first 24 h of thiamine treatment. A brain MRI scan was performed on a 1.5‐T system (Signa MRI system, General Electric Medical Systems), operating on a matrix 256×256 with slices of 5 mm thickness and 1.5 mm intersection gap. We used standard neuroanatomic landmarks to bring each brain into a unified system of coordinates and to correct the deviations in all three orthogonal planes. Acquisition of T1‐weighted spin echo images were obtained from sagittal and axial planes (TR/TE, 600/17 ms) and for the coronal plane we used an IRC (TR/TE/TI 1500/20/650 ms). None of the subjects had any substantial loss of data owing to technical MRI problems. MRI pictures were digitized with a densitometer (Vilber Loumart), transfered as TIFF format images, and processed using the NIH Image software by manually tracing all the regions of interest (ROIs).

Morphometric measurements in the mid‐sagittal plane included the vermis, posterior fossa, corpus callosum body and supratentorial intracranial areas (Figure 1). Because the uvula was not always clearly demarcated, it was not included in the vermian area, and was left out of the lower border of the vermis. Volumetry of the cerebellum (vermis plus cerebellar hemispheres) and cerebellar posterior fossa (posterior fossa excluding brain stem) were calculated from the sum of the volume of each sagittal slice until the cerebellar hemispheres were no longer observed, plus the calculated volume corresponding to the intersection gap. The volume of the cerebellum in comparison with the volume of the posterior fossa, mid‐sagittal supratentorial area and the height of the subject were considered as indices for cerebellar shrinkage. The ratio of the corpus callosum area with the supratentorial sagittal area was considered as index of cortical atrophy.5 Two physicians (JMN and JR) who had no knowledge of the clinical data obtained the measurements, with intraobserver and interobserver coefficients of variation within 1% and 2%, respectively.

Figure 1.

Brain MRI scans were performed on a Signa 1.5 T System with slices of 5 mm in thickness and a 1.5 mm intersection gap. MRI pictures were digitized and ROIs were traced manually using the NIH Image software. In the mid‐sagittal slice we measured the cross‐sectional areas of the corpus callosum (A), mid‐sagittal cerebrum (B), cerebellar vermis (C), and cerebellar posterior fossa (D).

Statistical analysis

Standard statistical methods from the SPSS Statistical Analysis System V7.5 were used. Morphometric indices were considered abnormal when the changes were >2 SD of the control values, or when greater than the 95% CI for the regression line for the variables that correlated with age in the non‐alcoholics. Group comparisons were performed by Mann‐Whitney analysis, ANOVA, Student's t‐test, χ2 and Fisher's exact tests when indicated. Correlation studies were obtained by Pearson's correlation coefficient and regression analysis. Partial correlation was used to determine the relationship between ethanol intake and cerebellar measurements adjusted for age. When two or more variables were significant in the univariate study, multiple stepwise regression analysis was performed. Stepwise logistic regression was used for evaluating the particular influence of each of the factors determining cerebellar shrinkage in alcoholics. Since most of the variables followed a normal distribution, they are expressed as means±SEM, and p<0.05 was taken as significant.


Clinical and epidemiological data

At diagnosis, acute WE patients presented with sudden onset of ataxia (83%), nystagmus (92%), ocular motor abnormalities (75%), and global confusion (83%). None had had previous episodes of WE. All WE patients were rapidly treated with 100 mg/day thiamine for one week, with marked clinical improvement of nine patients being observed within the initial 72 h of admission. The remaining three WE patients improved progressively over the first week of treatment, although one case continued to have gait ataxia. Asymptomatic chronic alcoholics and controls had normal neurological examination.

Epidemiological data are shown in Table 1. The age of the alcoholics ranged from 26 to 64 years (mean 47±1 years). The mean reported a daily intake of ethanol was 177±8 g (range 100–310 g) over a period of 27±1 years (range 8–44). Alcoholic beverages were generally consumed continuously as part of everyday life and only occasional binges were reported. Ethanol intake was mainly in the form of wine, beer or brandy, and less frequently as anisette, whiskey or gin. No relationship was observed between the type of beverage consumed and any of the variables studied.

View this table:
Table 1

Epidemiological and nutritional data of alcoholics and controls (data are expressed as means±SEM)

WE patients (n=12)Asymptomatic chronic alcoholics Controls (n=24)
Undernourished (n=12)Well‐nourished (n=24)
Age (y)49.7±2.652.7±2.745.2±2.0*47.5±2.5
Daily ethanol intake (g)197±14172±21170±14NA
TLDE (kg ethanol)1626±1491687±1871383±98NA
TLDE (kg ethanol/kg body weight)33.6±3.327.9±3.219.2±1.4**NA
Ideal body weight (%)74.1±5.988.5±3.3*102.5±2.0**105.7±2.2
Lean body mass (kg)41.0±1.445.8±1.451.4±0.8*52.7±0.7
Tricipital skin fold (cm)0.54±0.060.64±0.060.84±0.03*1.14±0.05***
Haemoglobin (g/l)124±6127±4143±4145±3
Lymphocytes (106/l)1590±1751495±921780±1081772±110
Total protein (g/l)61.5±1.664.9±1.768.4±1.671.5±1.3
Transferrin (mg/dl)208±13204±15268±12**257±16
Albumin (g/l)34.8±1.937.7±2.245.2±0.9*44.4±0.7
Prealbumin (mg/dl)19.4±2.819.6±2.431.4±1.9**31.2±1.4
Retinol‐binding protein (mg/l)31±332±547±3**51±5
Erythrocyte transketolase activity (%)a26.2±3.1b20.3±4.117.9±2.215.9±1.9
  • WE, patients with acute Wernicke's encephalopathy, all of whom were undernourished; TLDE, total lifetime dose of ethanol consumed; NA, not applicable.

    a Percentage difference between the enzymic activity with and without the addition of thiamine pyrophosphate (TPP effect).

    b Data refer to only five Wernicke's patients, prior to thiamine treatment.

    * p<0.05 compared with left‐column group; ** p<0.01 compared with the left‐column group; *** p<0.001 compared with the left‐column group.

Laboratory and nutritional data

As classified, undernourished alcoholics had worse nutritional values than well‐nourished alcoholics (Table 1). Well‐nourished asymptomatic alcoholics exhibited similar nutritional data to controls, except for a thinner tricipital skin fold (p<0.001). Acute WE patients tended to show worse nutritional status than undernourished chronic alcoholics, specially for a lower ratio of actual:ideal body weight (p=0.03). None of the well‐nourished alcoholics reported having had any episode of low body weight since they had begun ethanol misuse, and no dietary changes were reported. Erythrocyte transketolase activity (TPP effect) was markedly decreased in all five acute WE patients in whom the assay could be done prior to thiamine treatment. Mild elevations of hepatic enzymes after two months of ethanol abstinence were found in four alcoholics, as was echographic liver enlargement in another three cases. Liver biopsy was performed in these seven patients and the following diagnoses were made: normal liver in three cases, fatty liver in three cases, and alcoholic hepatitis in the remaining subject.

MRI scan data

None of the subjects showed focal MRI brain lesions, except for increased T2 signals in the paraventricular and periaqueductal regions in half of the WE patients. The volume of both cerebellar hemispheres were equivalent (r≥0.86, p<0.001), and the volume of the cerebellar hemispheres correlated with the vermian area (r≥0.72, p<0.001) in all the groups of patients. Morphometric analysis showed that the cerebellum, vermis and corpus callosum were significantly reduced in all the groups of alcoholics, compared to controls (Table 2). The vermis and cerebellum were smaller in undernourished than well‐nourished alcoholics, and in WE patients with respect to undernourished asymptomatic alcoholics. The degree of cerebellar shrinkage correlated with supratentorial atrophy, considered as the mid‐sagittal corpus callosum area, both in controls and asymptomatic alcoholics (r≥0.41, p≤0.03, all indices).

Cerebellar volume correlated inversely with age in the controls (r=−0.52, p=0.008) and asymptomatic alcoholics (r=−0.57, p<0.001), but not in WE patients (Figure 2). Interestingly, reduction in cerebellar volume with age was greater in asymptomatic alcoholics than controls, as evidenced by a significantly steeper slope of the regression line (−0.78±0.19 vs. −0.49±0.16, respectively, p=0.003). A greater reduction with age was also observed between asymptomatic alcoholics and controls regarding the slopes of the vermian area (p=0.001) and other cerebellar indices (p≤0.03, all).

Figure 2.

Correlation between cerebellar volume and age in controls (•), asymptomatic chronic alcoholics (▪), and acute WE patients (▴). Cerebellar volume correlated negatively with age in controls (r=−0.52, p=0.008) (solid line) and asymptomatic alcoholics (r=−0.57, p<0.001) (dash line). No correlation of age and cerebellar volume was obtained in WE patients (r=−0.02, p=0.94) (thin line). Notice that the slope between cerebellar volume and age in asymptomatic chronic alcoholics is steeper than that in controls (−0.78±0.19 and −0.49±0.16, p=0.003).

View this table:
Table 2

Quantitative morphometric brain data in chronic alcoholics and controls (data expressed as mean±SEM)

WE patients (n=12)Asymptomatic chronic alcoholics Controls (n=24)
Undernourished (n=12)Well‐nourished (n=24)
Corpus callosum surface (cm2)5.0±0.25.1±0.36.1±0.2*7.2±0.1**
Corpus callosum (% of supratentorial area)4.1±0.24.1±0.14.8±0.1*5.8±0.1**
Vermis surface (cm2)8.1±0.29.1±0.2*9.9±0.2*11.1±0.3**
Fossa‐adjusted vermis surface (%)40.0±1.644.9±1.3*46.1±1.149.8±1.1*
Cerebellum volume (ml)105±2114±3*127±3**136±2*
Cerebellum volume height‐adjusted (ml/m)63±170±1*76±1*79±1
Fossa‐adjusted cerebellum volume (%)53±158±1*62±1*65±1*
  • WE, patients with acute Wernicke's encephalopathy, all of whom were undernourished.

    * p<0.05 compared with the left‐column group; ** p<0.001 compared with the left‐column group.

Influence of nutrition and ethanol intake on cerebellar shrinkage

We evaluated the influence of different factors on the development of cerebellar atrophy. Regarding to nutrition, cerebellar volume was reduced by more than 95%CI of the regression line for the corresponding age of controls (cerebellar atrophy) in most of the undernourished alcoholics (12/12 WE and 10/12 asymptomatic alcoholics), compared with only 11/24 well‐nourished alcoholics (p=0.001). Nutritional status also affected the vermis area (atrophy in 11/12 WE patients, and 9/12 and 9/24 undernourished and well‐nourished asymptomatic alcoholics, respectively) and the other cerebellar indices (data not shown).

Well‐nourished alcoholics showed a significantly smaller cerebellum and vermis than controls (Table 2), occurring in 11/24 vs. 3/24 (p=0.01) and 9/24 vs. 3/24 subjects (p=0.04), respectively, suggesting that factors other than nutrition were probably involved in cerebellar degeneration in alcoholism. Alcoholics with cerebellar shrinkage (n=33) were older (49.2±1.4 vs. 43.0±1.6 years, p=0.05) and had a (non‐significant) trend to report greater daily ethanol intake (182±11 vs. 154±8 g, p=0.09) than alcoholics without cerebellar shrinkage (n=15). They also reported a non‐significantly greater duration of alcoholism and TLDE.

Stepwise regression analysis found age (p≤0.01), daily ethanol intake (p≤0.02), and the presence of malnutrition (p≤0.01) to be the only significant independent factors related to the degree of cerebellar and vermian shrinkage in alcoholics. The statistical significance of these variables remained when each group of alcoholics was evaluated separately, except for a lack of significance of age in WE patients. Daily ethanol intake and malnutrition globally accounted for about half of the variability of cerebellum values in alcoholics (r2 0.38–0.57, all cerebellar indices). In alcoholics, reduction in cerebellar and vermian measurements adjusted for age correlated significantly with daily ethanol intake (r≥0.31, p≤0.05, all). This relationship was even higher when only considering well‐nourished alcoholics (r≥0.52, p<0.001, all).

Finally, we used stepwise logistic regression for analysing the influence of each factor in the presentation of cerebellar atrophy. We observed that the presence of malnutrition was associated with a six‐fold increase in the risk for a significant reduction in cerebellar volume (OR 6.6 [95%CI 1.7–25.6], p=0.005). Likewise, a daily ethanol intake of more than 140 g over 10 years was independently associated with a higher probability of cerebellar atrophy (OR 6.1 [95%CI 1.8–20.5], p=0.003).


The pathogenesis of ethanol‐related neurological disorders has been considered to be multifactorial and attributed to genetic predisposition, nutritional factors, and the neurotoxic effects of ethanol or its metabolites.3 Most clinical neuroimaging studies have been focused on the deleterious effect of alcoholism on the cerebrum,1–,8 and only a few have specifically assessed its repercussion on the cerebellum. In the current study we provide morphometric evidence of cerebellar shrinkage in all patients with acute WE and in more than half of the asymptomatic chronic alcoholics without liver cirrhosis. Our results suggest that, besides the effect of aging, alcohol exerts a toxic effect on the cerebellum in a dose‐dependent manner. Moreover, malnutrition may act on the cerebellum as an additive deleterious factor in addition to alcoholism, similar to what has previously been suggested for the brain4 and peripheral nerves.16,,17

Moderate shrinkage of the vermis and cerebellar hemispheres was observed on post‐mortem examination in 27–42% of patients with alcohol misuse,7,,18 and in those with Korsakoff syndrome.19 The histological counterpart was decreased Purkinje cell density,7,,8 ranging from 15% in moderate alcoholics to 33% in heavy drinkers,8,,20 a reduction of the molecular and granular layers10,,20 and decreases in dendritic arborization and the number of synapses in the Purkinje cells.21 However, these studies included up to one third of alcoholics with liver cirrhosis, which is itself associated with cerebellar degeneration.22

Preliminary qualitative CT neuroradiological studies reported vermian and cerebellar shrinkage in alcoholics,23–,25 but it was not until the introduction of MRI that the cerebellum was properly assessed. Although initial MRI studies presumed that cerebellar shrinkage was restricted to thiamine‐deficient patients,26 Blansjaar et al.10 reported cerebellar shrinkage not only in Korsakoff syndrome, but also in some non‐amnesic alcoholics. Shear et al. noted widened cerebellum sulci in about 80% of alcoholics compared to 50% of controls, affecting Korsakoff patients and non‐amnesic alcoholics to a similar degree.9 By contrast, in another qualitative study, vermian and cerebellar shrinkage was observed in all patients with acute WE, compared to only one third of the asymptomatic alcoholics, and none of the controls.11 To clarify these aspects, cerebellar shrinkage has been assessed in the current study by quantitative methods. We observed significant cerebellar shrinkage in all subjects with acute WE, in more than half of the asymptomatic chronic alcoholics, and in about one‐tenth of the well‐nourished controls. The vermis and the cerebellar hemispheres were equally affected. Moreover, the degree of cerebellar and cerebral shrinkage ran parallel with supratentorial atrophy, except for WE patients, who showed a greater cerebellar shrinkage compared with corpus callosum atrophy (Table 2).

An interesting point is the relationship between age and brain volume in alcoholism. Despite some controversy,27 and in consonance with the current study, a number of authors have found an age‐related shrinkage of the cerebral and cerebellum in the general population.28–,30 Interestingly, we also observed that asymptomatic alcoholics had a greater reduction in cerebellar volume than that expected for their age, a situation similar to that observed for cortical brain atrophy.4 Therefore, older alcoholic patients may show more severe cerebellar involvement simply because they have been exposed to prolonged overdrinking. A reduction has been reported in dendritic arborization and the number of synapses in Purkinje cells in ethanol‐fed rats.31,,32

Additionally, cerebellar shrinkage was observed in most of the undernourished alcoholics compared with only 45% of the well‐nourished alcoholics, despite a similar daily ethanol consumption. Malnutrition was formerly believed by many investigators to be paramount in the development of alcohol‐related diseases. Malnutrition in the setting of marasmus33 and anorexia nervosa34 results in brain atrophy. Likewise, thiamine deficiency in humans, as well as pyrithiamine‐induced thiamine deficiency in experimental animals, causes selective neuronal loss of diencephalic structures.35 Interestingly, almost half of the well‐nourished alcoholics showed cerebellar atrophy which can only be related to ethanol intake, although past episodes of malnutrition resulting in brain damage36 cannot be totally excluded.

In summary, this study evaluated cerebellar involvement in alcoholics and the risk factors that determine cerebellar shrinkage in these patients. We found that almost half of the well‐nourished chronic alcoholics and all the malnourished patients showed significant cerebellar shrinkage. Age, daily ethanol intake, and the presence/absence of malnutrition were independent factors influencing cerebellar shrinkage in chronic alcoholism.


We would like to thank Mrs Anna Colomé for her assistance in all secretarial aspects of this study. Supported with grants from Fondo de Investigaciones Sanitarias FIS 98/0330 and CUR 1999/SGR‐0279 from Generalitat de Catalunya.


  • Address correspondence to Dr J.M. Nicolás, Department of Internal Medicine, Hospital Clínic, Villarroel 170, 08036 Barcelona, Spain. e‐mail: nicolasmedicina.ub.es


View Abstract