Seventy patients were identified: 53 women and 17 men with a mean age of 35.63 years (range, 16-48 years; mean age: men, 35.3 years; women, 35.7 years) (Table 1). Nine patients had a relevant family history (5 with MS and 4 with migraine). Twenty-five patients had a medical background: migraine (n = 12), depression (n = 7), endocrinologic disorders (n = 4), atopic eczema (n = 2), and breast cancer (n = 1). The first brain MRI was usually ordered by the general practitioner for various medical events, including headache (n = 24), migraine with (n = 4) or without (n = 10) aura, craniocerebral trauma (n = 2), depression (n = 7), tinnitus (n = 7), radiculalgia (n = 5), endocrinopathies (n = 3), childhood epilepsy (n = 2), cognitive complaint (n = 1), anosmia (n = 1), and dystonia (n = 1); 3 patients had volunteered to undergo MRI. None of these complaints were related to suspicion of a demyelinating event. Even if depression and radiculalgia rarely precede MS, it cannot be definitively disregarded as a CIS. The mean follow-up time for the whole cohort was 5.2 years (range, 3-6.4 years). The mean time between the first consultation and the first brain MRI was 5 months (range, 1-48 months). Seventeen patients had initial gadolinium enhancement, 58 had at least 9 T2-hyperintense lesions, and 45 had infratentorial lesions (Table 1). Cerebrospinal fluid analysis revealed a mean cell count of 3 cells/mm3 (range, 0-16 cells/mm3; with 28 patients having >4 cells/mm3). Thirty patients had oligoclonal bands and 28 had an increased IgG index. Forty-five patients had an abnormal asymptomatic VEP. For the 12 patients with fewer than 9 T2-hyperintense signals on brain MRI, CSF abnormalities were seen in 2 cases associated with an abnormal VEP. Among the 58 patients with at least 9 T2-hyperintense signals, 38 of 58 (65.5%) had an abnormal VEP and 48 of 58 (82.7%) had a positive CSF result. The mean time until the second brain MRI was 6 months (range, 3-30 months) (3-6 months in 34 patients, 6-12 months in 14 patients, 12-24 months in 21 patients, and 30 months in 1 patient). Sixty-four patients (91%) had dissemination in time, including 20 of 64 patients (37%) with new gadolinium enhancement and 47 with at least a new T2 lesion. Patients with dissemination in time were considered to have MS on MRI. During follow-up, 23 of 70 patients were found to have clinical conversion to the following conditions: optic neuritis (n = 6), myelitis (n = 6), brainstem (diplopia or internuclear ophthalmoplegia) symptoms (n = 5), sensitive symptoms (paresthesias in lower or upper limbs) (n = 4), cerebellar symptoms (n = 1), and cognitive deterioration (n = 1). No progressive form of MS was detected. No particular event (infection or vaccination) was found in the 3 months before the CIS. The mean time between the first brain MRI and CIS was 2.3 years (range, 0.8-5.0 years). Twelve patients had been treated with immunomodulators after the CIS: 8 with interferon and 4 with glatiramer acetate (4 patients after the second relapse). Of patients who had MS diagnosed according to McDonald criteria,10,11 4 had their first clinical event during the first year after the initial MRI, 6 patients had it during the second year, 11 had it during the third and fourth years, and 2 had it at 5 years (Figure 1). The mean age at development of a CIS was 34.2 years. Eight of 23 patients had 2 relapses, with a mean delay of 9 months. To date, the mean Expanded Disability Status Scale (EDSS) score for the 23 patients who had a CIS is 1.5 (range, 0-4), with a mean follow-up after the CIS of 2.1 years. Considering MRI criteria, we found that only gadolinium enhancement on MRI and infratentorial lesions were statistically significant for clinical conversion in a univariate analysis (P = .01) (Table 1). Associated MRI criteria (≥9 T2-hyperintense lesions and gadolinium enhancement) predicted CIS (P < .05). When comparing patients with initial symptoms that are unlikely to represent demyelinating events with those who had symptoms that possibly represented a CIS, only gadolinium enhancement was statistically significant (P = .005) (Table 2). The CSF criteria were only statistically significant when associated with 9 or more T2-hyperintense lesions on the first MRI on multivariate analysis (P = .02) (Figure 2). An abnormal VEP at presentation was a significant pejorative marker for developing CIS in a univariate analysis (P < .05) (Table 3).