Developmental changes of the contingent negative variation in migraine and healthy children

Abstract

It has been hypothesized that abnormalities of information processing in migraine may be attributed to impairment of cerebral maturation. However, the most evidences for this hypothesis have come from cross-sectional studies during childhood. We performed a longitudinal study and recorded contingent negative variation (CNV), an event-related slow cortical potential, in migraine children (n = 27) and age-matched healthy individuals (n = 23) in 1998 and 8 years later (2006). Amplitudes of all CNV components were reduced and habituation of the initial CNV (iCNV) increased in the observed time. However, the reduction of the iCNV amplitude was more pronounced in migraine patients who were in remission in 2006 and in healthy subjects and less pronounced in migraineurs with persisting headaches. Patients with the worsened migraine demonstrated the most pronounced loss of iCNV habituation in 1998 and significantly increased iCNV amplitudes in 2006. This longitudinal study supports the hypothesis of impaired cerebral maturation in migraine and shows that migraine manifestation is a key factor interfering with the natural maturation process of central information processing.

Fig. 1

Amplitudes (mean and standard deviation) of different CNV components which were determined at T1 (1998) and T2 (2006) in healthy subjects and migraine patients (with remission, improved or worsened headaches). Reaction time (mean and standard deviations in parentheses) is given in the figure with PINV. P values demonstrate difference (paired t tests) between amplitudes of the iCNV obtained at T1 and T2. These differences did not survive Bonferroni alpha adjustment for the late CNV component and the total CNV

Fig. 2

Course of habituation of the iCNV determined at T1 (1998) and T2 (2006) in healthy subjects and migraine patients (with remission, improved or worsened headaches). Amplitudes in eight blocks of recordings are given as mean and standard error. Regression lines are supported by equations y = A(z) x + B(up) where A is a mean of regression coefficient and z its standard deviation; B an intercept and up its standard deviation