Neuropsychological aspects of portal-systemic encephalopathy.

An extensive psychometric test program was performed in 96 patients with proven liver cirrhosis and clinical signs of portal hypertension as well as in 20 patients with alcoholic pancreatitis, in 19 patients without cirrhosis but with alcoholic cerebral atrophy and in 163 normal controls. The study population comprised six groups of subjects as follows: Group 1. 27 patients with non-alcoholic cirrhosis and normal EEG pattern. Group 2. 48 patients with alcoholic cirrhosis and normal EEG pattern. Group 3. 21 patients with cirrhosis and minimal EEG changes. Group 4. 20 patients with alcoholic pancreatitis. Group 5. 19 patients without cirrhosis but with alcoholic cerebral atrophy. Group 6. 163 normal controls. A one way analysis of variances comparing asymptomatic patients (group 1, 2 and 4) with controls (group 6) revealed no significant differences between patients with alcoholic and non-alcoholic cirrhosis, both cirrhotic groups scoring significantly lower than patients with alcoholic pancreatitis and normal controls, who did not differ significantly. Comparing symptomatic patients (group 3 and 5) with normal controls both patient groups scored significantly lower than controls, the cirrhotic group (group 3) scoring significantly lower than patients with alcoholic cerebral atrophy. A two way analysis of variances revealed that in clinically asymptomatic patients cerebral functional defects revealed by psychometry are only due to cirrhosis and that in patients with clinical evidence of cerebral impairment the factors alcohol and cirrhosis are additive - not synergistic. A multiple group stepwise discriminant analysis revealed that tests evaluating psychomotor functions contributed most to the discrimination. Especially "line tracing " proved to be most sensitive and most specific followed by dexterity, steadiness, aiming, digit symbols in sensitivity and by reaction time, steadiness and dexterity in specificity. A test program for clinical use is proposed.