[Sensitivity of criteria for MRI interpretation in patients with multiple sclerosis]

Abstract

Magnetic resonance imaging (MRI) has become the preferred imaging technique in multiple sclerosis (MS). Areas of increased signal (AIS) are detected on T2-weighted (T2W) pulse sequencies in 70-100% of patients with clinically definite MS. However, AIS similar to those seen in MS have also been described in healthy elderly subjects and patients with various neurological disorders including several vasculitises, migraine, and trauma. The aim of this study was to test and compare the sensitivity of different currently used MRI criteria for the diagnosis of MS.

Method: The study comprised 49 patients wiht clinically definite MS diagnosed according to McAlpine's criteria. Cranial MRI was performed on a 1.5 T Magnetom with spin-echo pulse technique defined by relaxation times: T1W, T2W and proton density. Gadolinium-DTPA was not administred. We interpreted only spin-echo T2W images because they are the most sensitive for MS. The presence, number, size and location of AIS were recorded for all scans. We calculated the sensitivity for the four different sets of criteria (Paty's A and B criteria, Fazekas' criteria and Barkhof's criteria). The presence of four AIS greater than 3 mm was designated as Paty's A criteria, and presence of three AIS, one periventricular, greater than 3 mm was designated as Paty's B criteria. Fazekas' criteria require the presence of at last three AIS and two of the following three features: abutting body of lateral ventricles, infratentorial lesion location, and size = 6 mm. Barkhof's criteria were considered fulfilled if at least eight AIS were present, one infratentorial.

Results and discussion: Brain MRI revealed normal finding in four (8%) of 49 studied patients with clinically definite MS. Number, size and location of AIS detected in the remaining patients is presented in Table. Characteristic patterns of MRI lesions on T2W images fulfilling different sets of criteria for the diagnosis of MS are presented in Figures 1-3. Using Paty's B criteria, sensitivity of MRI in patients with clinically definite MS was the highest, reaching 92%. Applying Paty's A criteria sensitivity slightly decreased to 88%. Using Fazeka's and Barkhof's criteria led to a further, significant decrease in sensitivity (71% and 57%, respectively; p = 0.045). In 1993, Offenbacher et al. have reviewed 1500 consecutive brain MRI scans for the presence, number, size and location of AIS, and calculated the sensitivity and specificity of Paty's and Fazekas' criteria with respect to clinical evidence for MS. In this study, similar to our results, using Paty's criteria resulted in higher sensitivity (87% for Paty A and 90% for Paty B) than Fazekas' criteria which led to decrease in sensitivity (81%). However, when Fazekas' criteria were used, decrease in sensitivity was associated with highly significant improvement of specificity. In conclusion, according to our and previously published results, Paty's criteria based on the presence of three AIS, one periventricular, or on the presence of four AIS, greater than 3 mm, are the most sensitive for MRI interpretation in MS. Other criteria (greater number of AIS, greater AIS diameter, infratentorial location and gadolinium enhancement) should be used, in order to improve the specificity, only to the MRIs of elderly patients (> 50 years) with suspected MS or patients suspected of MS in whom alternative explanations seem equally similar.