Skip main navigation

Early Fiber Number Ratio Is a Surrogate of Corticospinal Tract Integrity and Predicts Motor Recovery After Stroke

Originally published 2016;47:1053–1059


Background and Purpose—

The contribution of imaging metrics to predict poststroke motor recovery needs to be clarified. We tested the added value of early diffusion tensor imaging (DTI) of the corticospinal tract toward predicting long-term motor recovery.


One hundred seventeen patients were prospectively assessed at 24 to 72 hours and 1 year after ischemic stroke with diffusion tensor imaging and motor scores (Fugl-Meyer). The initial fiber number ratio (iFNr) and final fiber number ratio were computed as the number of streamlines along the affected corticospinal tract normalized to the unaffected side and were compared with each other. The prediction of motor recovery (ΔFugl-Meyer) was first modeled using initial Fugl-Meyer and iFNr. Multivariate ordinal logistic regression models were also used to study the association of iFNr, initial Fugl-Meyer, age, and stroke volume with Fugl-Meyer at 1 year.


The iFNr correlated with the final fiber number ratio at 1 year (r=0.70; P<0.0001). The initial Fugl-Meyer strongly predicted motor recovery (≈73% of initial impairment) for all patients except those with initial severe stroke (Fugl-Meyer<50). For these severe patients (n=26), initial Fugl-Meyer was not correlated with motor recovery (R2=0.13; p=ns), whereas iFNr showed strong correlation (R2=0.56; P<0.0001). In multivariate analysis, the iFNr was an independent predictor of motor outcome (β=2.601; 95% confidence interval=0.304–5.110; P=0.031), improving prediction compared with using only initial Fugl-Meyer, age, and stroke volume (P=0.026).


Early measurement of FNr at 24 to 72 hours poststroke is a surrogate marker of corticospinal tract integrity and provides independent prediction of motor outcome at 1 year especially for patients with severe initial impairment.


Cerebral ischemic stroke is a major cause of motor impairment1 with significant impact on the ability of patients to be self-sufficient in activities of daily living. Providing accurate prognosis of long-term motor impairment within the first few days after an insult is highly desirable (1) to correctly inform patients and caregivers; (2) to rapidly anticipate the type, duration, and goals of rehabilitation; and (3) to improve patient selection for clinical trials, such as those focusing on brain repair and reorganization.

Clinically, patient age and initial stroke severity are important predictors of motor disability in the medium to long-term.2,3 Most patients exhibit a proportional recovery of ≈70% of their initial motor impairment,4,5 which, therefore, already conveys strong predictive information. Nevertheless, this 70% rule fails in patients with initial severe deficit,4,5 and the variability in medium to long-term outcome is not adequately explained by initial clinical severity alone for those patients. The interindividual variability in the relation between initial impairment and subsequent recovery hampers accurate individualized prognosis.6 Although infarct volume could be helpful to improve prognostication,7 its power as an independent predictor of clinical outcome is still debated.8

Stroke’s impact on corticospinal tract (CST) integrity might constitute a more relevant determinant of motor impairment and recovery.9 In line with this concept, Byblow et al10 recently demonstrated that CST integrity as assessed by the presence of motor-evoked potentials (MEPs) was a better predictor of the ability to recover compared with initial clinical impairment, thus improving the 70% rule. Furthermore, they also demonstrated the added value of fractional anisotropy (FA) measurements in the CST toward improving prediction of motor function recovery in combination with MEPs.10 Measurement of CST integrity with diffusion tensor imaging (DTI)11 has been proposed both by assessment of decreases in FA distal to the stroke area and by the quantification of a decrease in the number of CST fibers, as an index of fiber degeneration (Wallerian degeneration).12,13 In most previous work, including the one by Byblow et al,10 these measurements were primarily obtained weeks or months after stroke14,15 precluding their use for early prognostication. At the subacute stage, Feng et al16 quantified the overlap of the acute stroke with a canonical CST derived from DTI of healthy controls instead of using DTI in patients. Referring to work by Puig et al17 argue that in acute/subacute stage, DTI measurements along the CST might not accurately reflect CST damage because degeneration takes time to develop and might not be detected by this magnetic resonance imaging (MRI) technique. Nevertheless, a decrease in FA or in the number of CST fibers has already been reported even after only few days after the insult with promising correlations with longer term outcome.1821 Therefore, direct DTI measurements of the CST integrity rapidly after a stroke might be clinically feasible and could obviate the need for relatively complicated image analysis, which might experience image coregistration errors, and necessitates a canonical CST model derived from healthy volunteers. Here, we assessed the relationship between acute and chronic (1 year) tractographic analysis of the CST after stroke, under the hypothesis that DTI in the acute phase might provide a straightforward, quantitative, and objective predictor of CST motor fiber integrity, and, ultimately, of the potential for motor recovery over that provided by the 70% rule for resolution of motor impairment. The primary aim of this work is to investigate the added value of early DTI, compared with established and easy-to-collect variables, such as age and bedside tests for initial motor impairment to improve the prediction of long-term motor recovery.


Study Population

Patients admitted for an ischemic stroke between June 2012 and February 2014 were consecutively enrolled in a prospective cohort study. The study design included initial (between 24 and 72 hours) and chronic (1 year) clinical and MRI poststroke evaluations. The institutional ethics committee approved the study, and written informed consent was obtained for all patients from them or their relatives before inclusion. Primary inclusion criteria were men or women, older than 18 years old, with a suspected clinical diagnosis of minor-to-severe supratentorial cerebral infarct (National Institutes of Health Stroke Scale between 1 and 25) 24 to 72 hours after the insult confirmed on trace diffusion magnetic resonance. The exclusion criteria were a prestroke functional deficit determined as a modified Rankin ≥1 or a previous stroke along the CST documented on imaging, an infratentorial ischemic stroke or hemorrhagic stroke, and patients presenting psychiatric disorders, pregnant women, coma, and contraindications to MRI.

Clinical Assessment

The main outcome was motor impairment at 1 year evaluated by the Fugl-Meyer score. Motor recovery was estimated by subtracting the initial (24–72 hours) Fugl-Meyer from the chronic (1 year) scores (ΔFugl-Meyer=Fugl-Meyer recovery). The Fugl-Meyer score provides a detailed evaluation of motor impairment with inter- and intrarater reliability close to 1.22 The motor function domains of the Fugl-Meyer score measure movement, coordination, and reflex action about upper extremity (shoulder, elbow, forearm, wrist, and hand) and lower extremity (hip, knee, and ankle) and range from 0 (hemiplegia) to a maximum of 100 points (normal motor performance). The patients were followed-up to 1 year to allow recovery to plateau.23 All clinical evaluations were performed by a neurologist assisted by a trained physical therapist, without knowledge of MRI findings.

Magnetic Resonance Imaging

Image Acquisition

MRI scans were acquired at the time of clinical evaluations (24–72 hours and 1 year after stroke) by using a 3T Discovery MR 750w scanner (GE Medical System, Milwaukee, WI) with a 32-channel phased-array head coil. The imaging protocol included a DTI sequence using dual echo-planar imaging (40 axial slices; repetition time: 15 000 ms; echo time set to minimum; slice thickness: 3.5 mm; matrix: 160×160; field of view: 24 cm2; b values: 0 and 1000 s/mm2 applied in 16 noncolinear directions; scan time: 4 minutes and 30 seconds).

Rationale for Early DTI Analysis

Wallerian degeneration distal from the primary lesion (typically within the brain stem) is not expected to be detectable during the first 72 hours after the insult.17 Nevertheless, FA decreases within the stroke itself from this early time point correspondingly with ischemic damage.24 We reasoned that when tracking the CST from primary motor cortex to the brain stem, a significant reduction of FA within a stroke along the CST could interrupt the reconstructed streamlines, even in the absence of Wallerian degeneration. Therefore, although the real number of fibers may not yet be significantly decreased in <3 days MRI as there is not yet anatomic anterograde degeneration, we hypothesized that a measurement of so-called virtual fibers from DTI at an early time point could be a surrogate of the real fiber count at follow-up and, in turn, could be helpful for early prognostication of motor recovery.

Image Processing and Analysis

DTI inherently showed low signal-to-noise ratio because scan time had to be short to be tolerated by most patients, especially in the acute/subacute phase of disease. We consequently used a new DTI denoising filter25 to recover higher signal-to-noise ratio and thus improve image quality and accuracy of diffusion parameters. Distortions induced by eddy currents and head motion were corrected by performing rigid registration of diffusion weighted images to b0 images.26 The direction table was updated with the estimated motion matrices.

Stroke volumes were first calculated by segmenting each slice of the trace DTI from the initial MRI scan using a semiautomatic tool available in 3D slicer (

Then, a diffusion tensor model was fitted at each voxel using Trackvis ( to generate FA maps and to investigate CST integrity. The CST was identified using deterministic tractography between 2 distal seed-regions of interest (ROI): one in the primary motor cortex (M1) and its underlying white matter, the central sulcus serving as the posterior border, and the other in the anterior cerebral peduncles. This approach was applied as implemented in Trackvis and ensures that supratentorial strokes will fall between the 2-seed ROIs. ROIs were placed by a specialized neuroradiologist symmetrically based on color-coded FA maps and trace DTI images using previously published and easy to determine landmarks9,14 (Figure I in the online-only Data Supplement for illustration of ROI positions). Fibers that were not homogeneously blue coded (ie, superior-inferior fiber direction) and whose direction were not corresponding to the CST based on anatomic knowledge and a DTI-derived atlas27 were excluded by slight adjustment of the seed ROIs. Fiber tract propagation was terminated for FA <0.2 and angle <35° based on agreed-upon thresholds.9 The total number of fibers connecting the 2 ROIs ipsilateral to the stroke were normalized by the total number of fibers from the contralateral side to correct for age-related variations or preexisting diffuse microstructural alterations. Whether this strategy perfectly corrects the impact of leukoaraiosis in the ability to accurately quantify the CST integrity transcends scope of work and will be addressed in future analyses. We will refer to initial fiber number ratio (iFNr) for the 24 to 72 hours MRI and to final fiber number ratio (fFNr) for the 1 year follow-up MRI.

Statistical Analysis

To identify whether an early measure of fiber number could be a surrogate for real Wallerian degeneration at the chronic stage, we first assessed the correlation between iFNr and fFNr by using Spearman’s rank order correlation.

Then, we tested the value of iFNr to predict motor recovery compared with the other predictors. We reasoned that the first tests to predict motor recovery should be bedside tests. So we first tested the amount of variance (R2) explained by an initial Fugl-Meyer score with respect to motor recovery at 1 year with linear regression analysis. From this step, we identified a meaningful subpopulation (severely affected patients with initial Fugl-Meyer<50) whose motor recovery at 1 year was not predicted by initial Fugl-Meyer. A subsequent regression was, therefore, conducted in this subpopulation to test whether iFNr can explain more variance than initial Fugl-Meyer with respect to motor recovery at 1 year by comparing R2. For these patients who are severely affected at baseline (initial Fugl-Meyer<50), we also performed receiving operating characteristic analysis to test whether an iFNr cutoff might identify additional patients with a potential to recover by 70% or more despite their initial clinical severity.

The added value of iFNr compared with the other predictors was finally tested with multivariate ordinal logistic regression in the whole population. The dependent variable to predict was motor impairment at 1 year categorized as severe (Fugl-Meyer<50), marked (50≤Fugl-Meyer≤84), and moderate-to-slight (Fugl-Meyer>84) according to previously published thresholds.28 The independent variables introduced in the model were initial Fugl-Meyer, age, and stroke volume (model 1). The model was repeated with all the previous parameters plus the iFNr (model 2), and the overall discrimination of the 2 models was compared using an analysis of deviance table. To test whether prediction of the Fugl-Meyer impairment score would translate to prediction of motor disability, we finally conducted a binary logistic regression analysis including all previous parameters with the aim to predict the modified Rankin scale dichotomized at a threshold of 2 as outcome.

All statistical analyses were performed with the R software package (Version 3.0.1) and the GraphPad Prism version 6.0 software with a type I error set at α=0.05.


Patient Characteristics

Out of the first 230 patients screened between June 2012 and February 2014, we analyzed 117 patients with a complete set of data at 1-year follow-up at the time of analysis (Figure 1). Main characteristics of patients are summarized in Table 1.

Table 1. Characteristics of the Population (n=117)

Sociodemographic factors
 Age, y67 (58–78)
 Sex, % male67.5
 Hypertension, %53.8
 Diabetes mellitus, %12.8
Clinical factors
 Initial NIHSS score4 (2–8)
 Initial NIHSS motor subitems (5a, 5b, 6a, 6b)0 (0–3)
 Initial Fugl-Meyer score88 (55–97.5)
 Fugl-Meyer score at 1 y97 (88.5–99)
 mRS score at 1 y1 (0–3)
Imaging factors
 Stroke side, % left-hemispheric stroke47.9
 Stroke volume, cm313.7 (2.4–38.3)
 iFNr0.76 (0.37–1.03)
  Absolute fiber numbers (ipsilesional/contralesional)102/134
 fFNr at 1 y0.92 (0.44–1.13)
  Absolute fiber numbers (ipsilesional/contralesional)134/146

Values are median (Q1–Q3). iFNr and fFNr indicates initial and final fiber number ratios; mRS, modified Rankin Scale; and NIHSS, National Institutes of Health Stroke Scale.

Figure 1.

Figure 1. Flow chart. MRI indicates magnetic resonance imaging.

Relationship Between iFNr and fFNr

The denoising procedure25 provided improved image quality with 3 minutes of processing per sequence (Figure I in the online-only Data Supplement) and allowed to quantify the fiber number with the same standardized procedure for each patient. The iFNr measured at 24 to 72 hours was strongly correlated to the fFNr measured at 1 year (r=0.70; P<0.0001). Therefore, these results suggest that iFNr could be an early surrogate marker of the integrity of the CST 1 year later.

Prediction of Motor Recovery by Using Initial Clinical Severity

From the initial severity, we calculated the potential for motor recovery as the maximum Fugl-Meyer score (=100) minus the initial Fugl-Meyer.4 In the whole population, the initial severity (and in turn the potential of recovery) was only moderately correlated with Fugl-Meyer recovery (R2=0.35; P<0.001; n=117; Figure 2A). The residuals (difference between the observed recovery and the predicted recovery by using the linear regression model) showed substantial dispersion (heteroscedasticity) especially for the more severely affected patients with initial Fugl-Meyer<50 (and in turn a potential of recovery>50, Figure 2B). To formalize the finding of more variable recovery in severe patients, we repeated the regression by excluding these patients (n=26). The overall fit of this new model was strong (R2=0.65, P<0.0001, n=91, Figure 2C), and the residuals were highly concentrated around 0 (Figure 2D). The estimated regression model without the severe patients (Figure 2C) showed that the remaining mild-to-moderate patients recovered ≈73% of their maximum potential. Therefore, the initial Fugl-Meyer provided a strong estimation of the motor recovery except for the subpopulation with high initial impairment, defined as initial Fugl-Meyer<50, who demonstrated high variance in motor recovery.

Figure 2.

Figure 2. Relationship between motor recovery and initial clinical severity. Fugl-Meyer recovery at 1 year versus potential of Fugl-Meyer recovery (=100−initial Fugl-Meyer) and representation of the residuals (difference between observed recovery and recovery predicted by the model). A and B are plotted with all the patients. C and D are plotted after exclusion of severe patients whose initial Fugl-Meyer was <50. Regression lines and 95% confidence intervals are showed in A and C.

Added Prognostic Value of iFNr Compared With Initial Fugl-Meyer

We investigated the value of the iFNr to predict motor recovery when initial Fugl-Meyer was less informative (initial Fugl-Meyer<50). There was no recurrent stroke on the 1 year magnetic resonance scan that might have contributed to the poor motor recovery of some of these patients. Interestingly, although initial Fugl-Meyer was not correlated with motor recovery for these severe patients (R2=0.13; P=ns; Figure 3, red), the iFNr correlated strongly with motor recovery (R2=0.56; P<0.0001; Figure 3, black). Some patients showed low initial Fugl-Meyer and low iFNr and did not recover (Figures 3 and 4A). On the other hand, patients with low initial Fugl-Meyer, but preservation of some fibers based on the iFNr, showed substantial recovery (Figures 3 and 4B). Such clinical-to-DTI discrepancies are illustrated by the blue arrows in Figure 3. Using receiving operating characteristic analysis, we showed that severe patients with initial Fugl-Meyer <50 can recover by 70% or more of their initial motor impairment if iFNr>0.26 (sensitivity=100%; specificity=83.5%; negative predictive value=100%; positive predictive value=72.7%; area under the curve=0.93±0.05; P=0.006). Figure II in the online-only Data Supplement shows the recovery of these severe patients as a function of initial impairment and iFNr.

Figure 3.

Figure 3. Relationship between motor recovery, iFNr, and initial severity. Fugl-Meyer recovery at 1 year versus initial Fugl-Meyer (red) and iFNr (black) in the subset of severe patients. Regression lines and 95% confidence intervals are showed. The patients with severe initial clinical scores but preserved iFNr are illustrated with the blue arrows and showed substantial motor recovery at 1 year. iFNr indicates initial fiber number ratios.

Added Value of iFNr Compared With Other Predictors

In multivariate ordinal analysis conducted on the whole population, initial Fugl-Meyer was the strongest predictor of motor outcome (model 1), higher initial Fugl-Meyer being associated with better recovery (β=0.083; 95% confidence interval=0.059–0.115; P≤0.001; Table 2). When iFNr was added in the analysis (model 2), the predictive strength toward motor outcome (β=2.601; 95% confidence interval=0.304–5.110; P=0.031; Table 2) was significantly increased compared with model 1 (P=0.026). These results on the overall population probably reflect the contribution of the more severe patients highlighted before. Such prediction of impairment also translated in improving the prediction of disability according to the modified Rankin Scale (Table I in the online-only Data Supplement).

Table 2. Ordinal Logistic Regression Analysis: Predictors of Good Motor Outcome (n=117)

Predictorsβ (95% CI)ZP Value
Model 1
 Initial Fugl-Meyer score0.083 (0.059–0.115)5.947<0.001*
 Age−0.024 (−0.075–0.023)−0.9820.326
 Stroke volume−0.004 (−0.017–0.007)−0.6870.492
Model 2
 Initial Fugl-Meyer score0.074 (0.048–0.107)5.035<0.001*
 Age−0.010 (−0.061–0.038)−0.4060.685
 Stroke volume−0.002 (−0.015–0.011)−0.2460.806
 iFNr2.601 (0.304–5.110)2.1570.031*

Model 2 showed statistically better performances than model 1 (P=0.026). β indicates regression coefficient; CI, confidence interval; and iFNr, initial fiber number ratio.

*Significant P value <0.05.


The iFNr measured by DTI between 24 and 72 hours after the insult provides an early surrogate of chronic CST fiber degeneration measured by the fFNr at 1 year follow-up. This biomarker provides significant independent added value toward predicting long-term motor impairment as measured with Fugl-Meyer compared with detailed clinical testing at baseline, age, and stroke volume. This is especially relevant for the subpopulation of more severely affected patients who demonstrated highly variable motor recovery.

A decrease in FNr is supposed to reflect Wallerian degeneration of the CST, but an accurate physiopathological interpretation early after stroke is difficult. Indeed, Wallerian degeneration takes time to manifest and one might argue that fiber degeneration is not yet present at 24 to 72 hours after stroke onset. T2 hyperintensity along the affected tract is indeed only seen weeks to months after stroke,29 and reduced FA along the CST distal to the stroke area has been shown to be significant at day 30 but not yet at day 3.17 There are 2 alternatives to explain our measurement of reduced DTI fiber measurements at such an early time point. The first is that reduced iFNr reflects the beginning signs of actual histological fiber degradation that could be more or less acute depending on stroke severity.30 In line with this interpretation, histological signs of Wallerian degeneration were already detectable in the brain stem 2 days after experimental stroke in rats.31 In humans, a significant reduction of apparent diffusion coefficient in cerebral peduncles downstream of strokes with severe prognosis was reported on MRI performed 12 hours after insult,18 and early signs of Wallerian degeneration were also observed with DTI.12,19 Alternatively, although destruction of fiber structure is not yet evident histologically, the reduction of FA within a stroke located along the CST might influence the computation of tracking and force streamlines to interrupt. FA values within ischemic strokes progressively diminish, especially after 24 hours (mean delay in our population=58 hours) although they might be elevated hyperacutely (<6 hours).24,32 The severity of cytotoxic edema, and the associated reduction in FA, might represent ischemic damage and determine whether fibers are tracked or not through a stroke along the CST (Figure 4), and ultimately whether fibers will degenerate or not, as well as a patient’s potential for motor function recovery. Further work is needed to shed more light on the specific histological correlates of iFNr, which, nevertheless, seems to be clearly related to the degree of chronic CST damage.

Figure 4.

Figure 4. Illustrative examples of iFNr and fFNr. CST tractography overlaid on denoised diffusion-weighted images in 2 patients with initial severe impairment but different recovery, which was predicted by the iFNr. A, 69-year-old man with a 25 cm3 right-hemispheric stroke. Initial Fugl-Meyer score=12. The iFNr was 0.04 (number of fibers right/left=5/119), and the fFNr was 0 (0/125). No clinical recovery was measured: Fugl-Meyer at 1 year=12. B, 84-year-old man with a 21 cm3 right-hemispheric stroke. Initial Fugl-Meyer=26. The iFNr was 0.4 (number of fibers right/left=81/204) and fFNr was 0.52 (108/207) at 1 year. The patient recovered 53 points: Fugl-Meyer at 1 year=79. CST indicates corticospinal tract; fFNr, finial fiber number ratio; and iFNr, initial fiber number ratio.

Recently Feng et al16 used a metric called weighted-CST lesion load to provide an early quantification of CST integrity similar to our iFNr. The weighted-CST lesion load consists in quantifying stroke voxels that overlap a canonical CST derived from healthy controls.16,33 Although this approach is interesting, it requires projecting the patient’s diffusion weighted images into the space of the canonical CST, thereby facing the challenge of spatial normalization of lesioned brains. In contrast, the iFNr could be calculated in clinical routine with minimal postprocessing from a DTI that can be acquired within an acceptable scan time; denoising procedure25 being available to rapidly recover signal-to-noise ratio if needed. Furthermore, overlap between stroke and the canonical CST with weighted-CST lesion load might not invariably be consistent with fiber degeneration depending on the severity of ischemia-related histological changes, which can range from edema without irreversible destruction to full necrosis. iFNr might provide more specific information relative to fiber degeneration. Comparative work is required to address this question. The iFNr has nevertheless its own limitations. Contralateral alterations can confound the ratio in some patients with preexisting damage, but we chose it to reduce interindividual and interscan variability. Vasogenic edema that peaks within the first week34 can also alter CST tracking in large strokes inducing mass effect and reduced anisotropy on a CST that is nonetheless intact, and that will remain intact at follow-up. As edema is reabsorbed, the number of fibers identified tractographically will be found to increase, as the tissue recovers part of its anisotropic architecture. These confounding factors might contribute to the imperfect relation between true fFNr at 1 year and iFNr during the subacute phase (fFNr being higher than iFNr). However, we consider this ratio as a realistic parameter of Wallerian degeneration as assessed by its capability to predict motor recovery. Functional quantification of CST integrity by measurement of MEPs has also shown to improve predictive models of motor function recovery based on clinical measures,10 but DTI still adds predictive value when MEPs are absent.

Recent strategies to assess motor recovery recommend sequential identification of subsets of patients by combining predictors in a stepwise manner rather than considering them in isolation.10,16,35,36 In line with these strategies10,16,35,36 and in accordance with previous studies,4,5 we found that most stroke patients exhibit a proportional motor recovery of ≈70% of their initial impairment. This means that starting with bedside measures can already provide a strong indication of motor recovery. DTI could be reserved to resolve uncertainty for the subset of patients with more severe initial clinical impairment. Within these severe patients, there were 2 distinct clusters: (1) patients with concordant low iFNr and low clinical score who did not recover and (2) patients with preserved iFNr despite low clinical score who recovered substantially. Residual integrity of ≈30% of the motor fibers (iFNr≈0.26) was associated with motor recovery of 70% or more of the initial impairment despite the initial severity with a positive predictive value of 72.7%. There were still intermediate patients, whose recoveries were difficult to predict accurately (slight recovery despite low fiber ratio and low clinical scores). This might be because of the contribution of alternate motor fiber pathways, such as corticorubrospinal and corticoreticulospinal tracts,14 or to brain reorganization37 and compensation.38 Alternate strategies to capture the residual ability to recover over that provided by the proportional recovery rule include task-related brain activation using functional MRI39,40; MEPs from transcranial magnetic stimulation10; or a combination of methods such as clinical evaluation, transcranial magnetic stimulation, and DTI.10,35,36 These strategies might be helpful to further refine prognostic modeling in combination with the iFNr especially in patients with severe initial deficit.

Our study is not without limitations. Although the initial sample size was relatively substantial, only 117 patients had a complete set of data and an even smaller number presented with severe stroke. Further confirmation by including a more significant proportion of severe patients will be needed. We used clinically available DTI acquisition and deterministic tractography. More advanced image acquisition and analysis models might be helpful, eg, to track fibers at the crossing of the CST with the superior longitudinal fasciculus.41 The reproducibility of the measures was not formally assessed, but the ROI positioning is straightforward with minimal training. Finally, the iFNr might experience a ceiling effect in the sense that it does not correlate linearly with the final outcome in patients with a benign disease course, which can anyway be predicted based on a good initial score.

These limitations notwithstanding, we identified a strategy applicable in clinical routine, before patient discharge from the hospital, which seems to significantly improve prediction of upper/lower limb motor impairment after stroke for patients with initially severe paresis. This strategy has the potential to significantly improve stroke patient management by aiding in determining goals, type, and duration of rehabilitation. Furthermore, applying the proposed iFNr to stratify patients at inclusion in neuroprotective or neuroregenerative treatment trials might help increase statistical power and accelerate discovery and validation of efficacy of novel approaches to stroke treatment.


*Drs Bigourdan and Munsch contributed equally.

The online-only Data Supplement is available with this article at

Correspondence to Thomas Tourdias, MD, PhD, Bordeaux University Hospital, Neuroradiology, CHU de Bordeaux, Place Amélie Raba-Léon, Bordeaux F-33076, France. E-mail


  • 1. Mozaffarian D, Benjamin EJ, Go AS, Arnett DK, Blaha MJ, Cushman M, et alAmerican Heart Association Statistics Committee and Stroke Statistics Subcommittee. Heart disease and stroke statistics–2015 update: a report from the American Heart Association.Circulation. 2015; 131:e29–e322. doi: 10.1161/CIR.0000000000000152.LinkGoogle Scholar
  • 2. Weimar C, König IR, Kraywinkel K, Ziegler A, Diener HCGerman Stroke Study Collaboration. Age and National Institutes of Health Stroke Scale Score within 6 hours after onset are accurate predictors of outcome after cerebral ischemia: development and external validation of prognostic models.Stroke. 2004; 35:158–162. doi: 10.1161/01.STR.0000106761.94985.8B.LinkGoogle Scholar
  • 3. Saposnik G, Guzik AK, Reeves M, Ovbiagele B, Johnston SC.Stroke Prognostication using Age and NIH Stroke Scale: SPAN-100.Neurology. 2013; 80:21–28. doi: 10.1212/WNL.0b013e31827b1ace.CrossrefMedlineGoogle Scholar
  • 4. Prabhakaran S, Zarahn E, Riley C, Speizer A, Chong JY, Lazar RM, et al. Inter-individual variability in the capacity for motor recovery after ischemic stroke.Neurorehabil Neural Repair. 2008; 22:64–71. doi: 10.1177/1545968307305302.CrossrefMedlineGoogle Scholar
  • 5. Winters C, van Wegen EE, Daffertshofer A, Kwakkel G.Generalizability of the Proportional Recovery Model for the Upper Extremity After an Ischemic Stroke.Neurorehabil Neural Repair. 2015; 29:614–622. doi: 10.1177/1545968314562115.CrossrefMedlineGoogle Scholar
  • 6. Stinear C.Prediction of recovery of motor function after stroke.Lancet Neurol. 2010; 9:1228–1232. doi: 10.1016/S1474-4422(10)70247-7.CrossrefMedlineGoogle Scholar
  • 7. Vogt G, Laage R, Shuaib A, Schneider AVISTA Collaboration. Initial lesion volume is an independent predictor of clinical stroke outcome at day 90: an analysis of the Virtual International Stroke Trials Archive (VISTA) database.Stroke. 2012; 43:1266–1272. doi: 10.1161/STROKEAHA.111.646570.LinkGoogle Scholar
  • 8. Johnston KC, Wagner DP, Wang XQ, Newman GC, Thijs V, Sen S, et alGAIN, Citicoline, and ASAP Investigators. Validation of an acute ischemic stroke model: does diffusion-weighted imaging lesion volume offer a clinically significant improvement in prediction of outcome?Stroke. 2007; 38:1820–1825. doi: 10.1161/STROKEAHA.106.479154.LinkGoogle Scholar
  • 9. Puig J, Pedraza S, Blasco G, Daunis-I-Estadella J, Prados F, Remollo S, et al. Acute damage to the posterior limb of the internal capsule on diffusion tensor tractography as an early imaging predictor of motor outcome after stroke.AJNR Am J Neuroradiol. 2011; 32:857–863. doi: 10.3174/ajnr.A2400.CrossrefMedlineGoogle Scholar
  • 10. Byblow WD, Stinear CM, Barber PA, Petoe MA, Ackerley SJ.Proportional recovery after stroke depends on corticomotor integrity.Ann Neurol. 2015; 78:848–859. doi: 10.1002/ana.24472.CrossrefMedlineGoogle Scholar
  • 11. Ciccarelli O, Catani M, Johansen-Berg H, Clark C, Thompson A.Diffusion-based tractography in neurological disorders: concepts, applications, and future developments.Lancet Neurol. 2008; 7:715–727. doi: 10.1016/S1474-4422(08)70163-7.CrossrefMedlineGoogle Scholar
  • 12. Thomalla G, Glauche V, Koch MA, Beaulieu C, Weiller C, Röther J.Diffusion tensor imaging detects early Wallerian degeneration of the pyramidal tract after ischemic stroke.Neuroimage. 2004; 22:1767–1774. doi: 10.1016/j.neuroimage.2004.03.041.CrossrefMedlineGoogle Scholar
  • 13. Liang Z, Zeng J, Liu S, Ling X, Xu A, Yu J, et al. A prospective study of secondary degeneration following subcortical infarction using diffusion tensor imaging.J Neurol Neurosurg Psychiatry. 2007; 78:581–586. doi: 10.1136/jnnp.2006.099077.CrossrefMedlineGoogle Scholar
  • 14. Lindenberg R, Renga V, Zhu LL, Betzler F, Alsop D, Schlaug G.Structural integrity of corticospinal motor fibers predicts motor impairment in chronic stroke.Neurology. 2010; 74:280–287. doi: 10.1212/WNL.0b013e3181ccc6d9.CrossrefMedlineGoogle Scholar
  • 15. Puig J, Blasco G, Daunis-I-Estadella J, Thomalla G, Castellanos M, Figueras J, et al. Decreased corticospinal tract fractional anisotropy predicts long-term motor outcome after stroke.Stroke. 2013; 44:2016–2018. doi: 10.1161/STROKEAHA.111.000382.LinkGoogle Scholar
  • 16. Feng W, Wang J, Chhatbar PY, Doughty C, Landsittel D, Lioutas VA, et al. Corticospinal tract lesion load: an imaging biomarker for stroke motor outcomes.Ann Neurol. 2015; 78:860–870. doi: 10.1002/ana.24510.CrossrefMedlineGoogle Scholar
  • 17. Puig J, Pedraza S, Blasco G, Daunis-I-Estadella J, Prats A, Prados F, et al. Wallerian degeneration in the corticospinal tract evaluated by diffusion tensor imaging correlates with motor deficit 30 days after middle cerebral artery ischemic stroke.AJNR Am J Neuroradiol. 2010; 31:1324–1330. doi: 10.3174/ajnr.A2038.CrossrefMedlineGoogle Scholar
  • 18. DeVetten G, Coutts SB, Hill MD, Goyal M, Eesa M, O’Brien B, et alMONITOR and VISION study groups. Acute corticospinal tract Wallerian degeneration is associated with stroke outcome.Stroke. 2010; 41:751–756. doi: 10.1161/STROKEAHA.109.573287.LinkGoogle Scholar
  • 19. Maraka S, Jiang Q, Jafari-Khouzani K, Li L, Malik S, Hamidian H, et al. Degree of corticospinal tract damage correlates with motor function after stroke.Ann Clin Transl Neurol. 2014; 1:891–899. doi: 10.1002/acn3.132.CrossrefMedlineGoogle Scholar
  • 20. Møller M, Frandsen J, Andersen G, Gjedde A, Vestergaard-Poulsen P, Østergaard L.Dynamic changes in corticospinal tracts after stroke detected by fibretracking.J Neurol Neurosurg Psychiatry. 2007; 78:587–592. doi: 10.1136/jnnp.2006.100248.CrossrefMedlineGoogle Scholar
  • 21. Zhang M, Lin Q, Lu J, Rong D, Zhao Z, Ma Q, et al. Pontine infarction: diffusion-tensor imaging of motor pathways-a longitudinal study.Radiology. 2015; 274:841–850. doi: 10.1148/radiol.14140373.CrossrefMedlineGoogle Scholar
  • 22. Gladstone DJ, Danells CJ, Black SE.The fugl-meyer assessment of motor recovery after stroke: a critical review of its measurement properties.Neurorehabil Neural Repair. 2002; 16:232–240.CrossrefMedlineGoogle Scholar
  • 23. Verheyden G, Nieuwboer A, De Wit L, Thijs V, Dobbelaere J, Devos H, et al. Time course of trunk, arm, leg, and functional recovery after ischemic stroke.Neurorehabil Neural Repair. 2008; 22:173–179. doi: 10.1177/1545968307305456.CrossrefMedlineGoogle Scholar
  • 24. Sorensen AG, Wu O, Copen WA, Davis TL, Gonzalez RG, Koroshetz WJ, et al. Human acute cerebral ischemia: detection of changes in water diffusion anisotropy by using MR imaging.Radiology. 1999; 212:785–792. doi: 10.1148/radiology.212.3.r99se24785.CrossrefMedlineGoogle Scholar
  • 25. Manjón JV, Coupé P, Concha L, Buades A, Collins DL, Robles M.Diffusion weighted image denoising using overcomplete local PCA.PLoS One. 2013; 8:e73021. doi: 10.1371/journal.pone.0073021.CrossrefMedlineGoogle Scholar
  • 26. Avants BB, Tustison NJ, Song G, Cook PA, Klein A, Gee JC.A reproducible evaluation of ANTs similarity metric performance in brain image registration.Neuroimage. 2011; 54:2033–2044. doi: 10.1016/j.neuroimage.2010.09.025.CrossrefMedlineGoogle Scholar
  • 27. Wakana S, Jiang H, Nagae-Poetscher LM, van Zijl PC, Mori S.Fiber tract-based atlas of human white matter anatomy.Radiology. 2004; 230:77–87. doi: 10.1148/radiol.2301021640.CrossrefMedlineGoogle Scholar
  • 28. Sanford J, Moreland J, Swanson LR, Stratford PW, Gowland C.Reliability of the Fugl-Meyer assessment for testing motor performance in patients following stroke.Phys Ther. 1993; 73:447–454.CrossrefMedlineGoogle Scholar
  • 29. Kuhn MJ, Mikulis DJ, Ayoub DM, Kosofsky BE, Davis KR, Taveras JM.Wallerian degeneration after cerebral infarction: evaluation with sequential MR imaging.Radiology. 1989; 172:179–182. doi: 10.1148/radiology.172.1.2740501.CrossrefMedlineGoogle Scholar
  • 30. Thomalla G, Glauche V, Weiller C, Röther J.Time course of wallerian degeneration after ischaemic stroke revealed by diffusion tensor imaging.J Neurol Neurosurg Psychiatry. 2005; 76:266–268. doi: 10.1136/jnnp.2004.046375.CrossrefMedlineGoogle Scholar
  • 31. Iizuka H, Sakatani K, Young W.Corticofugal axonal degeneration in rats after middle cerebral artery occlusion.Stroke. 1989; 20:1396–1402.LinkGoogle Scholar
  • 32. Bhagat YA, Hussain MS, Stobbe RW, Butcher KS, Emery DJ, Shuaib A, et al. Elevations of diffusion anisotropy are associated with hyper-acute stroke: a serial imaging study.Magn Reson Imaging. 2008; 26:683–693. doi: 10.1016/j.mri.2008.01.015.CrossrefMedlineGoogle Scholar
  • 33. Zhu LL, Lindenberg R, Alexander MP, Schlaug G.Lesion load of the corticospinal tract predicts motor impairment in chronic stroke.Stroke. 2010; 41:910–915. doi: 10.1161/STROKEAHA.109.577023.LinkGoogle Scholar
  • 34. Lansberg MG, O’Brien MW, Tong DC, Moseley ME, Albers GW.Evolution of cerebral infarct volume assessed by diffusion-weighted magnetic resonance imaging.Arch Neurol. 2001; 58:613–617.CrossrefMedlineGoogle Scholar
  • 35. Krakauer JW, Marshall RS.The proportional recovery rule for stroke revisited.Ann Neurol. 2015; 78:845–847. doi: 10.1002/ana.24537.CrossrefMedlineGoogle Scholar
  • 36. Stinear CM, Barber PA, Petoe M, Anwar S, Byblow WD.The PREP algorithm predicts potential for upper limb recovery after stroke.Brain. 2012; 135(pt 8):2527–2535. doi: 10.1093/brain/aws146.CrossrefMedlineGoogle Scholar
  • 37. Green JB.Brain reorganization after stroke.Top Stroke Rehabil. 2003; 10:1–20. doi: 10.1310/H65X-23HW-QL1G-KTNQ.CrossrefMedlineGoogle Scholar
  • 38. Zeiler SR, Krakauer JW.The interaction between training and plasticity in the poststroke brain.Curr Opin Neurol. 2013; 26:609–616. doi: 10.1097/WCO.0000000000000025.CrossrefMedlineGoogle Scholar
  • 39. Marshall RS, Zarahn E, Alon L, Minzer B, Lazar RM, Krakauer JW.Early imaging correlates of subsequent motor recovery after stroke.Ann Neurol. 2009; 65:596–602. doi: 10.1002/ana.21636.CrossrefMedlineGoogle Scholar
  • 40. Zarahn E, Alon L, Ryan SL, Lazar RM, Vry MS, Weiller C, et al. Prediction of motor recovery using initial impairment and fMRI 48 h poststroke.Cereb Cortex. 2011; 21:2712–2721. doi: 10.1093/cercor/bhr047.CrossrefMedlineGoogle Scholar
  • 41. Lee DH, Park JW, Park SH, Hong C.Have You Ever Seen the Impact of Crossing Fiber in DTI?: Demonstration of the Corticospinal Tract Pathway.PLoS One. 2015; 10:e0112045. doi: 10.1371/journal.pone.0112045.CrossrefMedlineGoogle Scholar


eLetters should relate to an article recently published in the journal and are not a forum for providing unpublished data. Comments are reviewed for appropriate use of tone and language. Comments are not peer-reviewed. Acceptable comments are posted to the journal website only. Comments are not published in an issue and are not indexed in PubMed. Comments should be no longer than 500 words and will only be posted online. References are limited to 10. Authors of the article cited in the comment will be invited to reply, as appropriate.

Comments and feedback on AHA/ASA Scientific Statements and Guidelines should be directed to the AHA/ASA Manuscript Oversight Committee via its Correspondence page.