Visual field defect (VFD) commonly manifests as an impairment in visuospatial function after occipital stroke, hampering overall function [1]. Unveiling the brain mechanisms underlying spontaneous visual field recovery in acute stroke would facilitate the development of effective treatment strategies for stroke-induced VFD [2]. White matter connectivity of the visual cortex may be crucial for visuospatial function and visual field recovery [2-4]. In contrast to functional connectivity, structural connectivity in the visual cortex underlying spontaneous visual field recovery after stroke has received little attention [5]. Therefore, we hypothesized and assessed whether structural connectivity of the visual cortex is related to spontaneous recovery from VFD in acute ischemic stroke.

The Institutional Review Board of Asan Medical Center approved this study (IRB No. 2011-0538 and 2012-0097). Participants included VFD patients with acute ischemic stroke (n=21), who were selected from a previous study as those having consistent diffusion tensor imaging (DTI) acquisition parameters across three visits and with those of other patients [5]. Written informed consent was obtained from each participant. Within four time points (1 week, 1 month, 3 months, and 6 months after stroke onset), Humphrey Visual Field tests (Swedish Interactive Threshold Algorithm [SITA]-Fast 30-2) were used to measure the mean total deviation (MTD) scores in the affected visual hemifield-light detection performance averaged for two eyes relative to the age-matched normal values. T1-weighted images and DTI were acquired using a brain magnetic resonance imaging scanner (3.0 T Philips Achieva; Philips Medical Systems, Amsterdam, the Netherlands) and used to measure fractional anisotropy (FA) and mean diffusivity (MD) values in the nine white matter tracts connecting the visual cortex at three time points (1 week, 1 month, and 3 months after stroke onset) (Supplementary Figure 1). Changes in the MTD, FA, and MD values over time and their correlations were assessed using statistical analyses. The detailed acquisition and analysis are provided in the Supplementary Methods.

Demographic and clinical characteristics of patients with VFD are presented in Supplementary Table 1. MTD scores in the affected hemifield changed significantly between the four time points. At Bonferroni-corrected P value, the MTD scores improved between 1 week and 1 month (P<0.001; change=4.13±4.81 dB), 1 week and 3 months (P<0.001; change=5.37±5.91 dB), and 1 week and 6 months after stroke (P<0.001; change=5.34±4.97 dB). However, the changes between 1 and 3 months (P=0.029), 1 and 6 months (P=0.031), and 3 and 6 months (P=0.950) did not survive Bonferroni correction (Figure 1, Supplementary Table 2).

Among the nine white matter tracts connecting the visual cortex, the FA values significantly changed in seven white matter tracts between the three time points (Supplementary Table 3). The FA values significantly decreased between 1 week and 1 month and between 1 week and 3 months after stroke onset for the following six white matter tracts: forceps major, contralesional inferior fronto-occipital fasciculus, contralesional inferior longitudinal fasciculus (ILF), ipsilesional ILF, contralesional superior longitudinal fasciculus (SLF), and ipsilesional SLF (temporal part) (Figure 2).

The MD values changed significantly in the four tracts among the three time points (Supplementary Table 4). The MD values of the four tracts (forceps major, ipsilesional inferior fronto-occipital fasciculus, ipsilesional ILF, and ipsilesional SLF) significantly increased between the initial (1 week) and final (3 months) assessment time points. Among these, MD values significantly increased between 1 week and 1 month after onset in three tracts (forceps major, ipsilesional inferior fronto-occipital fasciculus, and ipsilesional ILF) (Figure 3).

Changed FA of the contralesional ILF (β=0.586, uncorrected P=0.006) and ipsilesional SLF, temporal part (β=0.469, uncorrected P=0.045) between 1 week and 3 months was positively correlated with changed MTD scores in the affected hemifield between 1 week and 6 months (Supplementary Table 5). Changed MD of the ipsilesional ILF (β=-0.433, uncorrected P=0.034) between 1 week and 1 month was negatively correlated with changed MTD scores between 1 week and 6 months (Supplementary Table 6). Changed MD of the ipsilesional ILF (between 1 week and 1 month) was negatively correlated with changed FA of the contralesional ILF (between 1 week and 3 months) (β=-0.350, Bonferroni-corrected P=0.014) (Supplementary Table 7).

We evaluated the natural trajectory of spontaneous recovery from VFD and the brain structural connectivity in patients with acute ischemic stroke. There was considerable visual field recovery within the first month, with gradually decreasing magnitude until 6 months after stroke [2,6]. Nevertheless, MD increase in the ipsilesional tracts and FA decrease in the overall tracts of the visual cortex were observed mostly within the first month, reflecting edema related to acute ischemic stroke [7,8]. Despite prominent visual field recovery within the first month, edema may still occur, as shown by decreased FA and increased MD [7,9]. Our negative correlation between 1-month ipsilesional MD changes and 3-month contralesional FA changes of the ILF is consistent with timing of acute stroke-related diffusivity changes: acute intracellular edema (changes in MD), followed by subacute extracellular edema (changes in FA) [7,8]. Importantly, less MD increase in the ipsilesional occipitotemporal tract (ILF) during the first month and less FA decrease in the bilateral occipitotemporal tracts (ILF, SLF) up to 3 months were correlated with greater visual field recovery up to 6 months after onset [7,9]. The occipitotemporal tracts involved in visual perception [3,10], connecting the visual cortex with the temporal regions, may recover from intracellular edema during the first month and from extracellular edema during 3 months, leading to visual field recovery during 6 months after onset [2,7,9].

The current study had a few limitations, including the small number of patients, lack of a control group, lack of consideration of patient confounding variables, and predefined analysis of the visual cortex. Future larger studies adjusting for the potential confounding effects of patient characteristics may distinguish structural changes underlying visual field recovery from changes due to natural aging and stroke pathology [2,9]. Nevertheless, our findings suggest that early FA or MD changes in the white matter tracts between the visual cortex and temporal regions, indicating recovery from stroke-related edema, may be related to later spontaneous recovery from VFD.