Random number generation in patients with symptomatic and presymptomatic Huntington's disease.

Random number generation (RNG), which involves producing lists of numbers at random, requires response selection and inhibition, which are aspects of executive function mediated by the frontal lobe. This study examined RNG performance in a group of mild to moderate Huntington's disease (HD) patients, asymptomatic gene-positive HD patients, and controls. The HD symptomatic group was significantly less random in their choices than controls. The gene-positive asymptomatic HD patients also showed a tendency toward abnormal RNG measures compared with controls, although this did not reach statistical significance after correction for multiple comparisons. However, analyzing the HD group as a whole showed that the frequency of errors, preferential selection of favorite numbers, and counting in 1's correlated significantly with the degree of symptom severity. Furthermore, counting in 1's correlated with degree of vulnerability to symptoms. A subset of patients and controls were retested 6 months later, and all RNG measures showed high correlations over time, highlighting the robust reliability of this test. The data therefore suggest that RNG is a useful, brief and reliable test of executive function and may be of value in studies investigating the longitudinal effects of disease-modifying treatment in HD.