Wallerian Degeneration in the Corticospinal Tract Evaluated by Diffusion Tensor Imaging Correlates with Motor Deficit 30 Days after Middle Cerebral Artery Ischemic Stroke

Discussion

This prospective study in a relatively large cohort of patients with territorial MCA infarction found significant correlations between DTI-measured WD in the CST and m-NIHSS-measured limb motor deficit 30 days after stroke onset. Decreased anisotropy distal to the infarct on the affected side of the CST, expressed as FA values measured at the rostral pons, correlated significantly with limb motor deficit at day 30.

As far as we know, no DTI data focused on a cutoff value for WD 30 days after stroke onset have been published. Care is needed when applying cutoff anisotropic values because changes in diffusion after WD depend strongly on the preexisting architecture of the white matter.¹⁵ Experimental studies suggest that the orientation of membranes continues to be the major determinant of anisotropy after WD. In fact, changes in diffusion observed in the rostral pons, where the CST intersects the transverse pontine fibers, differ substantially from changes observed in the cerebral peduncle or in the posterior limb of the internal capsule, where the CST consists of isolated and well-defined bundles of parallel fibers.¹⁶ Consequently, after white matter damage, cutoff anisotropic values at different points along the CST might differ, and further studies need to clarify this point.

We propose an rFA cutoff point that discriminates a high proportion of patients with normal motor function from those with motor deficit. We found that the rFA cutoff point discriminates these patients better than the FA cutoff point. Intraindividual evaluation by using rFA can detect damage to the CST ipsilateral to the infarct in patients with high baseline FA (patient 10, Table 2). On the other hand, the trajectory of the transverse pontine fibers can introduce error in calculating CST anisotropy measured at the rostral pons, and using rFA values minimizes this error. Finally, rFA is more applicable across centers and operators than absolute FA.

In the clinical setting, DTI enables evaluation of microstructural changes in the CST better than conventional MR imaging, quantifying the amount of damage specifically within the motor system after stroke. Although WD is associated with motor deficit, motor deficit is determined by loss of function of the motor neurons and/or axons, which may or may not be accompanied by WD.

Our study shows that imaging-based regional markers correlate with motor deficits in MCA stroke. By correlating DTI findings and clinical data, we found that motor deficits 30 days after stroke increased with decreased anisotropy in the pons. The decrease in the FA values in the affected CST becomes progressively smaller during the subacute-to-chronic stages of stroke, and previous studies have demonstrated worse outcome in patients with evident WD on conventional MR imaging techniques.^8–12 Signal-intensity abnormalities related to WD are generally not detected until 4 weeks after stroke, when the main finding is a hyperintensity along the affected tracts on DWI and T2-weighted images. After assessing WD changes qualitatively on conventional images and quantitatively by using DTI, we found differences in anisotropy in patients with motor deficits who had no signal-intensity abnormalities on conventional MR imaging. Most important, most patients with motor deficits at 30 days who presented decreased FA values had no signal-intensity abnormalities on the affected side of the CST; more specifically, fewer than half presented hyperintensities on FLAIR and/or restricted diffusion on DWI. The signal-intensity change in the affected CST was strongly associated with lower FA indexes. These findings suggest that DTI is more sensitive in detecting tissue changes regarded as WD than conventional MR imaging and DWI.

In an experimental model after unilateral MCA occlusion in rats, signs of WD in the CST were detected in the brain stem in histologic stains as early as 2–7 days after the stroke.⁴ Within the second stage, WD decreased gradually up to 6 weeks. Other recent studies have demonstrated early signs of WD in patients with stroke within 2–3 weeks of onset by using DTI.^16,24 Greater FA reduction along the CST on the affected side after cerebral infarction is associated with greater early motor deficit and worse motor recovery at 3 months.^19–22 In our study, we found significant correlations between FA indexes and motor deficit only at 30 days, and the reductions in mean FA, 17% and 32% in patients with m-NIHSS-II and m-NIHSS-III respectively, are similar to other published values measured at the pons.^15,18 Thomalla et al.¹⁷ found a 13% decrease in FA measured at the cerebral peduncle 2–6 months and 2–16 days after stroke.

Severe WD in the CST distal to a supratentorial infarct in the acute stage has been regarded as a predictor of worse motor outcome.^17–25 Liang et al²⁴ conducted a similar longitudinal controlled study by using serial DTI to assess both antegrade and retrograde WD in 12 patients with subcortical ischemic infarctions involving the posterior limb of the internal capsule. They found progressive decrease in FA with time starting from the first week in the region just proximal and distal to the internal capsule lesion. On the other hand, Wanatabe et al²⁵ used serial DTI evaluations to assess WD in the CST of 16 patients with stroke (6 ischemic, 10 hemorrhagic). They found that the good recovery group had no significant change in anisotropy in the rostral pons between 2 and 3 weeks, whereas the poor recovery group had a significant decrease in anisotropy during that time. Although not specifically calculated, Fig 4 of their study shows that the 0.9 ratio is the approximate anisotropy cutoff for distinguishing the separate recovery groups 3 weeks after onset.

Recently, Kusano et al²⁹ reported that FA indexes measured in the cerebral peduncles within 2 days after intracerebral hemorrhage onset may predict functional motor outcome. In this study, a region of interest−based analysis of the FA of the CST in the cerebral peduncles showed that FA measured in the affected side decreased by 11% compared with the unaffected cerebral peduncle. The cutoff point of the rFA for the good (m-NIHSS, 0–2) and poor (m-NIHSS scores ≥ 3) outcomes was 0.85. Despite the differences in the study design, disease, and time of evolution when the cutoff point was determined, our cutoff point of 0.925 at 30 days is very similar to theirs.

We found no earlier changes in DTI that predicted motor outcome. The 3-day follow-up included in our protocol failed to show changes in anisotropy at a point in the CST far from the infarct, specifically at the rostral pons. Placing the region of interest nearer to the infarct might find changes in FA values at day 3 that could predict poor motor outcome.

A valid surrogate functional outcome measure for stroke is needed. Clinical variables such as age and initial stroke severity measured with a neurologic deficit scale such as the NIHSS have consistently been associated with functional outcome after ischemic stroke.³⁰ However, previous studies have demonstrated that infarct volume on conventional MR imaging correlates only modestly with functional outcome.³¹ We found that stroke severity at day 3 correlated negatively with rFA at day 30; however, this correlation was not found during the first 12 hours after onset. In the acute phase (<12 hours after onset), three-quarters of our patients had motor deficits, and they were moderate or severe (m-NIHSS-III) in nearly half of these patients. However, at day 3, only half of our patients had motor deficits and only 20% of these deficits were moderate or severe. These differences in motor deficits between the acute phase and day 3 are probably due to the ischemic penumbra (not evaluated in our study).

On the other hand, stroke volume measured on conventional MR imaging at day 3 and at day 30 correlated weakly with FA indexes at day 30. Volume measures do not take into account the location of the lesion or the functional pathways involved, and 2 different infarcts of the same size in different locations could have very different functional expression. Thus, rather than total stroke lesion volume, it seems much more reliable to use DTI to evaluate the extent of damage specifically within the CST to determine motor deficit and axonal injury. However, we found no evidence that stroke severity at day 3 was expressed as WD. Placing the region of interest in a region of the CST closer to the infarct might make it possible to detect changes in FA indexes earlier and to predict which patients will have worse outcomes in the chronic phase so that rehabilitation interventions can be adjusted to each patient's need from the earliest stages to use health care resources more efficiently.

One new research line aims to improve neurologic outcome through stem cell−based treatment of chronic stroke.³² Although the study of the mechanisms underlying stem cell−based treatment has focused on angiogenesis and neurogenesis, white matter reorganization may contribute to functional recovery after stroke. The earliest stem cell transplantations were carried out 4–9 weeks from onset,³³ and 4 weeks was also the time when our study demonstrated CST damage in the patients with motor deficits. DTI may also be useful in determining the real state of the CST before treatment, quantifying the amount of damage specifically within the motor system after stroke, and FA indexes could be an alternative way of monitoring the changes in cerebral tissue that lead to improved outcome.

Several potential limitations to our study merit comment. First, manual region-of-interest placement is subject to operator bias, especially at day 30, when CST damage is present, and this may result in variability in location, size, and shape. Ideally, all the descending motor pathway areas should have been included to avoid underestimating the degree of DTI change and thus the degree of WD. Further studies by using automated region-of-interest analysis or voxel-based analysis may resolve this issue. Using m-NIHSS categorization does not diminish the significance of the association between decreased FA and motor deficit, though dedicated neuromotor test batteries like the Medical Research Council Scale or Motricity Index might find significant anisotropic differences between different grades of motor deficits. On the other hand, it could be interesting to determine other anisotropic parameters, like mean diffusivity or diffusion tensor eigenvalues, and to analyze whether changes in the affected tract can be detected earlier to enable better and earlier prognosis. Finally, the small number of patients with motor deficits at 30 days could be a limitation; however, our sample size seems sufficient given that the profile of patients defined by the inclusion criteria is similar, the variability in the findings for the unaffected side among all patients is low, and the normal unaffected side served as an internal control that minimized the potential variability among patients. Nevertheless, a larger sample would have increased the power of our findings.