Topographic diagnosis of internuclear ophthalmoparesis: evidence from a lesion-behavior mapping study

Abstract

The syndrome of internuclear ophthalmoparesis (INO) is very common and almost always associated with a homolateral lesion localized to the medial longitudinal fasciculus (MLF) making INO among the most eloquent syndromes in neurologic practice [1–5]. However, to the best of our knowledge, there is no statistical anatomical approach comparing patients with INO vs. without INO identifying the brainstem structures involved. Thus, the present voxel based lesion-mapping (VLBM) approach allows for the first time precise statistical lesion localization and the direct comparison of the voxels affected in patients showing INO vs. the voxels affected in patients without INO [6]. Forty-four patients with magnetic resonance imaging (MRI) documented brainstem lesions (mean age 64 years; standard deviation (SD) 16.9 years; right-sided lesions 52 %; left-sided lesions 48 %) and a control group of 30 healthy age-matched controls (mean age 58 years; SD 10.4 years (independent t test p = 0.117)) participated in the study. Fifteen patients presented with an INO (four patients each with left-sided and bilateral INO and seven patients with right-sided INO) and 29 patients without. The INO was detected clinically by an experienced neuroophthalmologist and confirmed by electro-oculography [1]. To verify the INO, the versional disconjugacy index (VDI), defined as the ratio of the abducting and the adducting eye velocity and its corresponding z-scores which represents a standardized number in relation to the present control group, was applied [7, 8]. With respect to the z-score and the VDI, a difference was seen between the patient group without INO, the INO group and the controls (one-way ANOVA, p = 0) (see Table 1). Thirty-seven patients (84 %) had a lesion due to ischemic stroke, one patient (2 %) due to a bleeding, and one patient (14 %) due to multiple sclerosis (MS) demonstrated by MRI (1.5T scanner Vision, Siemens, Erlangen, Germany). The MRI T2-weighted fluid-attenuated inversionrecovery sequences, diffusion-weighted sequences, T2 and Gadolinium T1-sequences were applied. The MRI dataset was first prepared using the isolation algorithm of the SUIT toolbox [9, 10]. Afterwards the automated normalization algorithm provided by SPM8 and the SUIT toolbox implemented in SPM8 were applied [9, 10]. The lesions were then delineated directly on the normalized lesions of the individual normalized MRI scans with MRIcroN software [6] (http://www.mccauslandcenter.sc.edu/mricro/ mricron/install.html). To evaluate the relationship between lesion location and INO, i.e., to analyze whether patients with INO differ with respect to their lesion location from patients without INO, a subtraction analysis was conducted. Subtraction analysis reflects the relative frequency of damage. Statistical VLBM analysis was performed by using the dichotomic Liebermeister test implemented in the B. Baier (&) F. Thomke Department of Neurology, University of Mainz, Langenbeckstr. 1, 55131 Mainz, Germany e-mail: baierb@uni-mainz.de