Subgroups of migraine with aura patients
N1, P1 and N2 latencies were not significantly different between HV, MA or MA+ (P > 0.05) (Table 2).
In the rm-ANOVA model with N1–P1 peak-peak amplitude as dependent variable, multivariate test was significant for the “group” × “blocks” interaction effect (Wilks’ Lambda = 0.711, F10,140 = 2.608, p = 0.006). After checking that the sphericity assumption was not violated (Mauchley Test: p = 0.126), univariate rm-ANOVAs for N1-P1 peak-peak amplitude confirmed the significant interaction factor effect (F10,370 = 3.025, p = 0.001, partial η2 = 0.076, op = 0.982) observed (see above) at the multivariate test. On post-hoc analysis there was a significant increase of N1-P1 VEP amplitude from the 2nd to the 6th block in MA+ compared with MA, and from the 4th to the 6th block in MA+ compared with HV (row data are shown in Fig. 2). In both MA and MA+, the comparison between the 6th and the 1st N1-P1 amplitude block did not reach the significance level (p > 0.05).
In the rm-ANOVA model with P1–N2 peak-peak amplitude as dependent variable, multivariate test was not significant for the “group” × “blocks” interaction effect (Wilks’ Lambda = 0.834, F10,140 = 1.335, p = 0.218).
The linear regression N1–P1 slope of VEP amplitudes over all blocks differed significantly between the three groups (F2,74 = 12.219, p < 0.0001, partial η2 = 0.248, op = 0.995; raw data are shown in Fig. 2). Post-hoc analysis showed that the slope of N1–P1 VEP amplitude changes over all 6 blocks was less steep in MA and in MA+ patients than in HV (p = 0.0001, p = 0.001 respectively, raw data are shown in Fig. 3), but it was equally steep between MA subgroups (p = 0.894).
The P1–N2 slope of the linear regression analysis was not different between groups (F2,74 = 1.720, p = 0.186, partial η2 = 0.044, op = 0.351; Fig. 3).
In MA+, Pearson’s test disclosed that the N1–P1 habituation slope correlated negatively with attack frequency (r = −0.489, p = 0.034) and positively with days elapsed since last attack (r = 0.578, p = 0.019), correlations that were not found in MA patients (Fig. 4).
Fig. 4 Correlation between the days elapsed between the recordings and the last migraine attack and the slope of N1–P1 VEP amplitude changes over 6 sequential blocks of averaged responses (linear regression: dashed line). This correlation was significant in the group of patients with complex aura (MA+, right panel), but not in patients with pure visual aura (MA, left panel)