Loss of up to one-half of the visual field (hemianopia) as result of post-chiasmatic stroke in one hemisphere occurs in about 30% of all stroke patients. Following a period of spontaneous recovery in the first 3–6 months (1, 2), the patient enters the chronic phase of hemianopia.
Rehabilitation treatment most often involves eye movement training to compensate for the visual field defect (3) rather than visual restitution training, which reduces the defect itself. The latter has long been controversial (4). However, a recent series of investigations (5–12) have argued for the more balanced view that visual training of the defect may provide an additional and valuable approach to rehabilitation of occipital stroke patients.
Brain plasticity is believed to be greater in the acute stage after stroke when there is a window for relatively quick and extensive synaptic reorganization (13). Recommendations that rehabilitation should begin “as soon as possible” or “early” are therefore common in clinical guidelines (14, 15). However, many of these recommendations are based on limited data (16), and there are no agreed definitions of what constitutes early rehabilitation (17). Thus far, visual restitution training is generally applied in the chronic phase after stroke, so that spontaneous recovery can be excluded, and changes in the visual field can be attributed to training. In this way, one can obtain an accurate estimate of the effect of the training itself (8–11). Yet, we wondered if visual restitution training would profit from an early start as suggested in the rehabilitation literature.
The effect of visual perceptual learning in normally sighted subjects is often restricted to the trained region of the visual field (18–20) and specific to the trained task (21, 22). This raises the question whether the visual recovery that is induced by visual restitution training is also limited to just the trained region and task. Several studies have shown that recovered vision after restitution training transfers to untrained visual tasks (10, 11) but only to a limited extent to untrained regions. For example, the defect reduction induced by training of the intact visual hemifield was significantly smaller than the reduction induced by training the affected hemifield itself, and it was not significantly different from the defect reduction following a non-intervention period (11). Because spontaneous recovery could be excluded in that study, any improvement during intact training could point to a spillover effect of training between the two hemispheres. That is, the defect reduces—albeit to limited extent—even when another part of the visual field is trained.
Following the practice of general rehabilitation medicine, one would preferably train patients in the early phase of stroke. To do so, we applied a method that builds on the observation that visual training carries over to neighboring areas only to a limited extent. That is, we used two training rounds, which targeted complementary parts of the defect [regions of interest (ROIs)], while monitoring in both training rounds the trained and the untrained half of the defect. The untrained half of the defect, which serves as an internal control for the trained half, will show spontaneous recovery and a potential spillover from the neighboring trained region. To assess that spill over, we used data from seven patients who were trained in the chronic phase of stroke using the same method. The differences between the defect reductions for the subacute phase of stroke and the chronic phase of stroke in the trained and untrained parts of the defect should allow us to distinguish between spillover, spontaneous recovery and training-induced recovery. This allows us to test the hypothesis that training in the early phase leads to a larger defect reduction than training in the chronic phase.
Materials and Methods
The study was approved by the ethical committee CMO Arnhem–Nijmegen in correspondence with the 1964 Declaration of Helsinki.
20 Subacute stroke patients and 10 chronic stroke patients with visual field defects due to post-geniculate damage were included following written informed consent. Subacute stroke patients were screened for participation in four neurology departments of Dutch hospitals: UMC in Utrecht, St. Elisabeth Hospital in Tilburg, CWZ in Nijmegen and St. Antonius Hospital in Nieuwegein (screening; eight patients). Patients could also sign up for the study by filling out a form on our website (www.hemianopsie.nl; 12 patients), to be screened at a regional office by the first author. Chronic stroke patients all applied through the website.
Patients inclusion criteria as follows:
∗ age between 18 and 75 years;
∗ presence of homonymous visual field defect.
Patient exclusion criteria as follows:
∗ visual neglect (as assessed by line bisection test);
∗ cardiac or other implants (for the chronic patients only: MRI scans were made; to be presented elsewhere).
The intake procedure included a Goldmann perimetry measurement. Patient demographics can be found in Table S2 in Supplementary Material.
For the 30 included patients, we had to exclude the data of 3 subacute and 3 chronic patients from further analysis. In the three subacute patients, the training was not applied as intended because the defect was not divided in two equal halves (n = 2), or for unequal duration of the training rounds (n = 1). In the chronic patients, absence of an absolute defect (n = 1), inability to cope with training demands (n = 1), and anxiety for fMRI scanner measurements (n = 1) were reasons to exclude their data. Thus, in total, we analyzed 17 subacute and 7 chronic data sets.
The 20 subacute patients were trained by DB for this study, the 10 chronic patients were trained by JE in a parallel study using the same training paradigm.
Before the training, baseline values were established for visual field size (Goldmann perimetry), reading speed and Goal Attainment Scaling (GAS: personally customized and realistic goals).
Following these baseline measurements, the visual field defect was divided in equal halves using the following procedure. First, meridional angles through the defect were established that were farthest apart. Then, the average of these two outer meridional angles formed the border between the two training regions (in the case of SA15, the division was along the vertical midline). One-half of the visual field defect was trained for 8 weeks, while the other half was untrained. After this period, intermediate measurements were carried out (perimetry and reading speed tests) during the course of one week. Then, a second training period of 8 weeks was started, in which the training was applied to the other half of the defect, while the first half received no further training. Post-measurements were carried out as during baseline measurements (Figure 1). Finally, we collected follow-up perimetry data in the subacute group. The period without training, in-between the final training session and the follow-up perimetry of the subacute group, is denominated “No Intervention.”