Outcome Assessment of Hemiparesis due to Intracerebral Hemorrhage Using Diffusion Tensor Fractional Anisotropy Tetsuo Koyama, MD, PhD,*† Kohei Marumoto, MD, PhD,† Yuki Uchiyama, MD,† Hiroji Miyake, MD, PhD,‡ and Kazuhisa Domen, MD, PhD†
Background: This study aimed to evaluate the prognostic efficacy of magnetic resonance diffusion tensor fractional anisotropy (FA) for patients with hemiparesis due to intracerebral hemorrhage. Methods: Diffusion tensor FA brain images were acquired 14-21 days after putaminal and/or thalamic hemorrhage. The ratio of FA values within the cerebral peduncles of the affected and unaffected hemispheres (rFA) was calculated for each patient (n 5 40) and assessed for correlation with Brunnstrom stage (BRS, 1-6), motor component of the functional independence measure (FIM-motor, 13-91), and the total length of stay (LOS) until discharge from rehabilitation (P ,.05). Ordinal logistic regression analyses were conducted to determine the relationships between rFA and specific outcomes as measured by BRS range (poor, BRS 1 or 2; moderate, BRS 3 or 4; and good, BRS 5 or 6; P , .05). Results: The rFA values were .571-1.043 (median, .856) and BRS scores were 1-6 (median, 4) for shoulder/elbow/forearm, 1-6 (median, 4) for hand, and 2-6 (median, 4) for lower extremities. FIM-motor scores were 58-86 (median, 78) and LOS ranged from 42 to 225 days (median, 175.5 days). Correlation coefficients were statistically significant between rFA and shoulder/elbow/forearm BRS (.696), hand BRS (.779), lower extremity BRS (.631), FIM-motor (.442), and LOS (2.598). Logistic model fit was moderate for shoulder/elbow/forearm BRS (R2 5 .221) and lower extremity BRS (R2 5 .277), but was much higher for hand BRS (R2 5 .441). Conclusions: Diffusion tensor FA values are predictive of clinical outcome from hemiparesis due to putaminal and/or thalamic hemorrhage, particularly hand function recovery. Key Words: Hematoma—paresis—probability—prognosis—recovery— stroke. Ó 2015 by National Stroke Association
From the *Department of Rehabilitation Medicine, Nishinomiya Kyoritsu Neurosurgical Hospital, Nishinomiya, Hyogo; †Department of Rehabilitation Medicine, Hyogo College of Medicine, Nishinomiya, Hyogo; and ‡Department of Neurosurgery, Nishinomiya Kyoritsu Neurosurgical Hospital, Nishinomiya, Hyogo, Japan. Received October 21, 2014; revision received December 3, 2014; accepted December 5, 2014. Grand-in-Aid for Scientific Research (B), the Japan Society for the Promotion of Science (KAKENHI), 25282168 and the Medical Research Fund of Hyogo Medical Association, MRF-H-08-12. Address correspondence to Tetsuo Koyama, MD, PhD, Department of Rehabilitation Medicine, Nishinomiya Kyoritsu Neurosurgical Hospital, 11-1 Imazu-Yamanaka-cho, Nishinomiya, Hyogo, Japan 663-8211. E-mail: [email protected]. 1052-3057/$ - see front matter Ó 2015 by National Stroke Association http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2014.12.011
Stroke is a leading cause of disability in most advanced countries.1 Stroke due to intracerebral hemorrhage is often associated with severe impairment such as hemiparesis,2 resulting in poor functional outcome.3,4 To facilitate the most effective rehabilitation, the evaluation of brain images in relation to clinical severity is critically important. Magnetic resonance (MR) diffusion tensor imaging (DTI) has recently been applied to assess white matter degeneration after stroke.5 DTI detects the diffusion gradient path of water molecules to reveal preserved axonal fibers, and so, enables clinically useful characterization of Wallerian degeneration after intracerebral hemorrhage. Of the parameters obtained from DTI, fractional anisotropy (FA) has proven to be a useful index of axonal degeneration, and several studies have attempted to use
Journal of Stroke and Cerebrovascular Diseases, Vol. 24, No. 4 (April), 2015: pp 881-889
881
T. KOYAMA ET AL.
882
FA values to characterize the relationships between neural degeneration and clinical manifestations of intracerebral hemorrhage.6-11 Patients often exhibit different outcomes for upper and lower extremity functions after intracerebral hemorrhage, with lower extremity functions generally showing better recovery than upper extremity functions.8,11 Thus, many patients resume independent walking, leading to functional independence in activities of daily living (ADL). However, few previous studies have reported on specific clinical outcomes and their relation to DTI, analyses may be useful for tailoring rehabilitation regimens to target the most severe reversible functional deficits. In this study, we used a variety of analytical procedures to assess the relationships between DTI-FA and multiple clinical outcomes. The results suggest that hand functions are most severely affected by putaminal and/or thalamic intracerebral hemorrhage, whereas lower extremity function and ADL are relatively spared.
Methods Patients The study sampled patients with intracerebral hemorrhage who were admitted to Nishinomiya Kyoritsu Neurosurgical Hospital between December 2009 and March 2014. Patients were typically transferred to our hospital soon after onset and underwent conservative treatment such as medication to reduce hypertension and, when necessary, surgical removal of hematoma. During hospitalization, they also underwent physical therapy, occupational therapy, and speech therapy for a combined daily total of up to 180 minutes. The protocols for these rehabilitative treatments followed the conventional methods stated in the Japanese Guidelines for the Management of Stroke.12 The work presented here extends that of our previously published study11 that included data from 12 patients already entered in the database for the present study. The Ethics Committee of Hyogo College of Medicine approved the study protocol, and patients (or relatives when necessary) provided written informed consent. To minimize the variability arising from differences in prestroke health status and lesion site, the sample population was limited to first-ever stroke patients with thalamic and/or putaminal hemorrhage who were able to walk unaided and had been functionally independent in ADL before stroke. For MR imaging safety, patients with implanted metal items (eg, artificial pacemakers) were excluded. Patients who subsequently required acute medical services (for recurrence of stroke, angina pectoris, or other comorbid conditions) were also excluded. To minimize variability arising from differences in the rehabilitative therapeutic regimen, this study included data only from patients who were transferred to our affiliated
long-term rehabilitation facility (Nishinomiya Kyoritsu Rehabilitation Hospital) to receive inpatient rehabilitative care for at least 1 month.
Computed Tomography Acquisition On arrival at our hospital, patients manifesting hemiparesis or other symptoms of stroke (hemorrhagic or ischemic) underwent head computed tomography with an Aquilion 64SP scanner (Toshiba Medical Systems Corp., Tochigi, Japan). Imaging parameters were 120 kVp and 250 mA, in-plane resolution was .86 mm 3 .86 mm, and slice thickness was 8 mm. The volume of intracerebral hemorrhage was estimated conventionally.13
DTI Acquisition DTI was performed 14-21 days after admission using a 3.0 T MR scanner (Trio; Siemens AG, Erlangen, Germany) with a 32-channel head coil. Details of the DTI acquisition protocol were reported in our previous studies.8,11 In brief, the DTI protocol acquired 12 images with noncollinear diffusion gradients (b 5 1000 seconds/ mm2) and 1 nondiffusion-weighted image (b 5 0 seconds/mm2) using a single-shot echo-planar imaging sequence. A total of 64 axial slices were obtained from each patient. The field of view was 230.4 mm 3 230.4 mm, the acquisition matrix was 128 3 128, and slice thickness was 3 mm without a gap. Echo time was 83 milliseconds and repetition time was 7000 milliseconds.
Outcome Measurements Brunnstrom stage (BRS),14 which is commonly used by Japanese rehabilitation therapists,12 was adopted to assess post–intracerebral hemorrhage motor impairments of the upper and lower extremities on the affected side. In this assessment, recovery of the affected extremities was evaluated by associated reactions and flexion and extension synergy patterns on a 6-point scale from severe (1) to normal (6). Conventionally, BRS is used for separate functional evaluation of the proximal (shoulder/elbow/ forearm) and distal (hand) upper extremity and the entire lower extremity, and its reliability and validity are well established.15,16 Assessments were made by occupational or physical therapists blinded to the purpose of the study. In addition to extremity functions, we obtained scores on the motor component of the functional independence measure (FIM-motor). The FIM is a test battery commonly used for evaluating stroke rehabilitation.17,18 It consists of individual 7-point scales (from total assistance to complete independence) for the following 13 items: eating, grooming, bathing, dressing upper body, dressing lower body, toileting, bladder and bowel management, transfer to bed/chair/wheelchair, transfer to toilet, transfer to tub/shower, walking or wheelchair propulsion, and stair
INTRACEREBRAL HEMORRHAGE OUTCOME ASSESSMENT BY DTI-FA
883
Table 1. Patient profiles Cerebral peduncle Patient
Age (y), M/F
Hemorrhage site, volume (mL)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
64, M 49, M 51, M 68, M 76, M 89, M 55, M 48, F 73, M 53, M 69, M 56, M 73, M 77, F 77, F 69, F 72, M 80, M 62, M 62, F 83, M 50, M 43, F 61, F 50, M 59, M 44, M 71, M 39, F 64, M 69, M 57, M 57, M 61, M 73, M 62, F 74, M 31, F 46, F 55, M
R thal, 7.4 L thal, 13.5 R thal, 17.7 R thal, 7.8 R put, 25.0 L thal/put, 3.8 L thal, 2.8 R thal/put, 4.5 R thal, 6.4 R put, 23.8 L thal, 4.5 L put, 5.6 R thal, 12.6 R thal/put, 2.8 R thal, 4.1 L put, 18.3 L put, 7.6 L thal, 10.5 R thal/put, 3.0 R put, 9.7 L put, 3.9 R put, 20.4 R put, 11.2 L put, 20.1 L put, 62.3 R put, 24.6 R put, 43.5 R thal, 6.4 L put, 27.8 R put, 49.4 L put, 62.8 L put, 6.1 L put, 26.9 L put, 15.2 L put, 47.1 L thal, 8.7 L put, 17.5 L put, 29.7 L thal/put, 53.8 L put, 32.7
BRS
FA (R)
FA (L)
rFA
S/E/F
Hand
L/E
FIM-M
LOS(d)
.587 .509 .519 .577 .562 .574 .538 .595 .571 .561 .579 .576 .462 .472 .526 .552 .587 .545 .475 .489 .569 .472 .509 .551 .527 .419 .442 .439 .595 .433 .535 .567 .579 .558 .595 .536 .594 .585 .537 .563
.562 .508 .530 .592 .579 .556 .520 .617 .598 .604 .537 .515 .524 .535 .599 .484 .510 .472 .555 .571 .487 .554 .607 .458 .437 .523 .555 .566 .460 .562 .394 .414 .421 .396 .422 .360 .382 .367 .315 .321
1.043 .999 .979 .976 .969 .969 .968 .965 .955 .929 .927 .895 .883 .882 .879 .877 .870 .866 .857 .856 .855 .853 .838 .832 .829 .800 .796 .776 .772 .769 .736 .731 .727 .710 .709 .671 .643 .628 .587 .571
5 5 5 4 5 5 6 6 5 6 6 4 4 3 4 5 2 5 4 5 4 5 5 2 2 4 2 5 1 2 1 3 2 4 1 2 3 1 2 3
5 5 5 4 5 5 6 6 5 6 5 5 4 3 6 5 2 5 4 4 4 5 5 2 2 3 2 5 1 2 1 2 2 2 1 1 2 1 2 2
5 5 5 3 5 4 6 6 5 6 5 4 4 5 5 5 3 5 4 6 4 4 6 5 3 5 3 4 3 2 3 4 3 4 2 2 3 3 2 4
86 77 79 77 77 79 85 81 78 84 80 83 75 65 72 81 75 66 79 79 75 81 85 78 75 86 79 80 78 65 65 80 64 78 74 58 64 72 77 77
83 65 153 101 67 165 175 85 147 45 42 165 176 181 183 124 195 203 203 194 191 122 64 116 216 191 172 105 182 198 212 204 166 202 188 204 204 224 225 169
Abbreviations: BRS, Brunnstrom stage; F, female; FA, fractional anisotropy; FIM-M, functional independence measure motor; L/E, lower extremity; LOS, length of total hospital stay from admission to acute medical service to discharge from long-term rehabilitation facility; M, male; put, putamen; rFA, ratio between FA values in affected and unaffected hemispheres; S/E/F, shoulder, elbow, and forearm; thal, thalamus. Patient data are ordered 1-40 according to rFA values (highest to lowest). Eight patients (Nos. 25, 27, 29, 30, 31, 33, 38, and 39) underwent surgical removal of hematoma. 12 patients (Nos. 1, 4, 9, 13, 14, 15, 24, 29, 32, 35, 36, and 38) were included in the database in our previous study.11
climbing. The total scores of these items are often used as indices of independence in ADL (scale range, 13-91). BRS and FIM-motor score were assessed monthly, and data were collected from our long-term rehabilitation at discharge. Discharge from this facility was decided when the increase in FIM-motor score was 1 or lower during the month. The total length of hospital stay (LOS) was also recorded for each patient.
Image Processing The brain image analysis package FSL (Analysis Group, FMRIB, Oxford, UK) was used for image processing.19-21 The details of image analysis were the same as those reported in our previous studies.8,11 In brief, DTI data were first corrected for motion and eddy current distortions by aligning all images to the first image
T. KOYAMA ET AL.
884
Figure 1. Computed tomography images and diffusion tensor imaging-fractional anisotropy maps of the cerebral peduncles in 5 representative cases. Abbreviations: L, left; rFA, ratio of fractional anisotropy; R, right.
(b 5 0 seconds/mm2), and then extracerebral regions were excluded from the images. Regional brain FA values were then calculated to yield an FA brain map for each patient. To assess neural degeneration due to intracerebral hemorrhage, regions of interest were defined in the left and right cerebral peduncles, which were abstracted by referring to the The International Consortium for Brain Mapping DTI-81 standard brain.22 FA values within the left and right regions of interest were calculated, and mean values for single voxels were subsequently estimated. Then, the ratio of FA values in the affected and unaffected hemispheres (rFA) was calculated for each patient as an index of neural degeneration in the corticospinal tracts.6-11
Statistical Analysis Spearman rank correlations were calculated to evaluate the relationships between rFA and clinical parameters (BRS, FIM-motor, and LOS). Ordinal logistic modeling was then used to test the association between rFA and individual outcomes (shoulder/elbow/forearm, hand, and lower extremity) as measured by BRS range (poor, moderate, or good).8 The principle of logistic modeling involves fitting the probability (p) of a dichotomous response (such as ‘‘yes/no’’ or ‘‘dead/alive’’) to a linear model. The probability odds for these types of dichotomous responses [p/(1 2 p)] can take any positive value. The logarithm of these odds is modeled as a simple regression, and parameter estimates are assessed for fit to the model.
log½p=ð12pÞ5a1bX
where a is a constant; b is a coefficient; and X, explanatory variable. To extend the utility to multi-level ordinal responses, modeling the odds to a simple regression cumulative probability is performed at each level.23 For 3-level responses (p1, p2, p3; sum equals 1), the logarithm of the odds is modeled as 2 simple regressions for the 3-level responses, and parameter estimates are assessed for fit to the model (note that single b and 2 levels of a are assessed).
log½p1 =ðp2 1p3 Þ5a1 1b1 X log½ðp1 1p2 Þ=p3 5a2 1b1 X In the present study, to simplify the logistic regression analyses, we defined 3 levels for BRS; poor (BRS 1 or 2), moderate (BRS 3 or 4), and good (BRS 5 or 6). Goodness-of-fit of the logistic modeling was assessed using the Wald chi-square test. In this study, all statistical analysis was performed using JMP software (SAS Institute, Cary, NC), and P less than .05 was considered statistically significant.
Results During the study period, 89 patients underwent DTI, of which 11 were discharged home directly from our hospital, 38 were transferred to other long-term rehabilitation hospitals or nursing homes (20 facilities in total) typically closer to their homes, and the remaining 40 were transferred to our long-term rehabilitation hospital and included in this study.
INTRACEREBRAL HEMORRHAGE OUTCOME ASSESSMENT BY DTI-FA
885
Figure 2. Results from correlation analyses between Brunnstrom stage, FIM-motor scores, length of hospital stay, and rFA. Dashed lines indicate density ellipse (.90). Abbreviations: BRS, Brunnstrom stage; FIM, functional independence measure; rFA, ratio of fractional anisotropy.
Table 1 presents the profiles of the 40 patients transferred to our long-term facility (median age, 62 years; age range, 31-89 years), 18 with right hemisphere lesions and 22 with left hemisphere lesions. rFA values ranged from .571 to 1.043 (median, .856). BRS data ranged from 1 to 6 (median, 4) for shoulder/elbow/forearm, 1 to 6 (median, 4) for hand, and 2 to 6 (median, 4) for lower extremity. FIM-motor scores ranged from 58 to 86 (median, 78) and LOS from 42 to 225 days (median, 175.5 days). Figure 1 shows brain images from 5 represen-
tative patients (numbers 1, 10, 20, 30, and 40), where patients are listed in order of descending rFA. Rank correlation analyses (Fig 2) revealed moderate-tohigh correlations between rFA and extremity function as assessed by BRS: shoulder/elbow/forearm (.696; P , .001), hand (.779; P , .001), and lower extremity (.631; P , .001). In contrast, the correlation between rFA and FIM-motor was weaker (.442; P 5 .004). There was a negative correlation between rFA and LOS (2.598, P , .001).
T. KOYAMA ET AL.
886
Table 2. Parameter estimates for ordinal logistic equations between BRS and rFA N 5 40
Ordinal logistics BRS portion Shoulder/elbow/forearm Estimates Standard error P value Hand Estimates Standard error P value Lower extremity Estimates Standard error P value
b
a, Poor (BRS 1 or 2)
Moderate (BRS 3 or 4)
213.224 3.621 ,.001
9.794 2.937 ,.001
11.689 3.132 ,.001
227.956 7.293 ,.001
22.779 6.107 ,.001
24.428 6.323 ,.001
215.359 4.123 ,.001
9.422 3.003 .002
13.403 3.571 ,.001
R2
.221 ,.001 .441 ,.001 .277 ,.001
Abbreviations: BRS, Brunnstrom stage; rFA, ratio of fractional anisotropy.
Table 2 shows the results obtained from ordinal logistic regression analyses of rFA values and BRS values for shoulder/elbow/forearm, hand, and lower extremity. In these analyses, the fits of ordinal logistic modeling were statistically significant, indicating that the data set results could be validly interpreted by logistic probability. Model fit was much greater for hand (R2 5 .441) than that for shoulder/elbow/forearm (R2 5 .221) or lower extremity (R2 5 .277). Figure 3 shows logistic probability plots of individual rFA values versus BRS scores. In each plot, the leftmost curve defines the probability for poor recovery (BRS 1 or 2) and the rightmost curve the probability of good recovery (BRS 5 or 6) versus rFA value. The distance from the leftmost curve to the bottom of the graph is the probability of poor recovery, the distance between the 2 curves is the probability of moderate recovery (BRS 3 or 4), and the distance from the rightmost curve to the top of the graph is the probability of good recovery at a given rFA. For the shoulder/elbow/forearm, when rFA was around .7, estimated probability for poor recovery was around 60%. When rFA was around .9, the estimated probability for good recovery was close to 55%. The middle panel shows equivalent data for the hand. In this case, when rFA was around .7, estimated probability for poor recovery was nearly 95%, whereas when it was around .9, the estimated probability for good recovery was close to 70%. For the lower extremity (lower panel), even when rFA was around .7, estimated probability for poor recovery was only around 20%. This indicates that compared with the upper extremity, the lower extremity recovered relatively well even when rFA was low.
Discussion Previous DTI studies after intracerebral hemorrhage reported that lower FA values in the corticospinal
tracts were associated with more severe hemiparesis.6-10 Similarly, in our preliminary study (n 5 12) assessing the relationship between DTI-FA and clinical manifestations using simple correlation analysis, we found a strong correlation between rFA and upper extremity function.11 To further extend these results, in the present study, we included additional patients (n 5 40) and conducted ordinal logistic regression to evaluate the correlation between specific functional outcomes and rFA. The results confirmed that DTI-FA values are strongly predictive of upper extremity function. Indeed, rFA was more strongly associated with outcome in upper extremity function than FIM-motor scores. In our preliminary report, we applied only rank correlation to simplify the analysis.11 However, this methodology may not be sensitive enough to identify relationships in cases of uneven distribution patterns. For example, in our present sample (Fig 2 and Table 1), 16 cases (40%) showed good recovery of hand function (BRS 5 or 6), 17 (42.5%) showed poor recovery (BRS 1 or 2), and only 7 (17.5%) showed moderate recovery (BRS 3 or 4). To address this concern, we used ordinal logistic regression analysis and found a strong significant correlation between rFA and hand function outcome. For shoulder/ elbow/forearm and lower extremity, the relationships with rFA were weaker, as indicated by the shallow logistic curves and wide region separating curves for probability of good and poor recovery (Fig 3, upper and lower panels). In contrast, logistic curves for hand function were much steeper with a narrower zone for moderate recovery (Fig 3, middle panel). Thus, DTI-FA was most strongly predictive of hand function outcome. In contrast to the results for the upper extremity, logistic regression for the lower extremity indicated that moderate or good outcome is to be expected even when rFA values are lower (around .7; Fig 3). Similarly, the results from simple correlation analyses revealed that
INTRACEREBRAL HEMORRHAGE OUTCOME ASSESSMENT BY DTI-FA
Figure 3. Logistic probability plots for the relationship between rFA and stratified individual BRS outcomes (good, moderate, and poor). Abbreviations: BRS, Brunnstrom stage; rFA, ratio of fractional anisotropy.
887
the relationship between rFA and FIM-motor scores, although statistically significant, was rather weak (correlation coefficient of .442). This is consistent with previous studies indicating that independence of ADL (of which FIM-motor is a component) is most strongly associated with locomotive ability rather than upper extremity function.24,25 Taken together, these results suggest that some patients with severe neural damage within the corticospinal tracts (as measured by rFA) can still achieve independence with minimal improvement of upper extremity function.11 Indeed, our data for FIMmotor scores ranged from 58 to 85 (Table 1), a range in which most of the patients could walk independently with or without canes.23,26 In this study, we collected data from patients at discharge from our affiliated rehabilitation facility after 42-225 days of therapy. This variation in poststroke treatment duration may confound outcome data because greater recovery is expected among patients discharged earlier (eg, ,90 days).27 Nonetheless, we entered all outcome data in our final analytical database for the following reasons. As shown in Figure 2, analysis detected a significant negative correlation between rFA and LOS, indicating that patients with less-severe neural degeneration tended to be discharged earlier. In parallel, patients with higher rFA showed better outcome for both BRS and FIM-motor parameters. Better recovery is expected for such patients with relatively mild clinical manifestations. However, ceiling effects will minimally affect the statistical significance of the data set. Accordingly, we entered all data for all 40 patients in our final analytical database despite the wide variation in LOS. In this study, of the 89 total samples, we entered 40 (45%) into the final analytical database. Given such a small population, confounds arising from sampling bias may be a critical concern. The aim of this study was to determine the relationship between FA decrease within the cerebral peduncles and the clinical severity such as hemiparesis. Of the 89 patients, 11 were discharged home directly from our hospital. As shown in our previous study, patients with less-severe hemiparesis were associated with less significant decrease of FA values in the cerebral peduncles.8 It is very common for patients who exhibit mild hemiparesis to return home directly from acute care service. Accordingly, in this study, we focused on patients who were transferred to long-term facilities.28 In this study, the assignment of the facilities was typically based on the patients’ areas of residence. Accordingly, the possibility of sampling bias was minimal. This study has several limitations. First, we recruited subjects only with thalamic and/or putaminal hemorrhage to reduce variability, but patients with subcortical and pontine hemorrhage are also commonly encountered in clinical practice. Such patients should be
888
included in larger scale studies to clarify the utility and limitations of DTI-based prognosis for the general population. Second, the sample size was small, partially because of the relatively restrictive inclusion criteria. To minimize variability arising from different rehabilitative regimens, we collected data only from patients transferred to a single rehabilitation hospital. Studies enrolling greater numbers of patients from multiple rehabilitative facilities using similar regimens are required. Third, we sampled data from patients who had no neurologic disease other than stroke. Consequently, patients with commonly observed geriatric comorbidities such as dementia, Parkinsonism, and epilepsy were excluded. Again, careful consideration should be given when applying our findings to the general geriatric population. Fourth, we focused on clinical severity using BRS and FIM-motor assessment, and impairment of higher brain functions such as aphasia and hemineglect may be overlooked by these scoring systems. Despite these limitations, the application of both simple rank correlation and logistic regression proved useful for characterizing the prognostic efficacy of DTI-FA for patient outcome after intracerebral hemorrhage. In summary, we have demonstrated that DTI-FA can be used effectively as a prognostic index of clinical outcome after putaminal and/or thalamic hemorrhage. Our findings suggest that DTI-FA is particularly useful for predicting recovery of hand function. These results may be useful for tailoring rehabilitation accordingly.
References 1. Hankey GJ, Jamrozik K, Broadhurst RJ, et al. Long-term disability after first-ever stroke and related prognostic factors in the Perth Community Stroke Study, 19891990. Stroke 2002;33:1034-1040. 2. Jorgensen HS, Nakayama H, Raaschou HO, et al. Intracerebral hemorrhage versus infarction: stroke severity, risk factors, and prognosis. Ann Neurol 1995;38:45-50. 3. Daverat P, Castel JP, Dartigues JF, et al. Death and functional outcome after spontaneous intracerebral hemorrhage. A prospective study of 166 cases using multivariate analysis. Stroke 1991;22:1-6. 4. Katrak PH, Black D, Peeva V. Do stroke patients with intracerebral hemorrhage have a better functional outcome than patients with cerebral infarction? PM R 2009;1:427-433. 5. Yamada K, Sakai K, Akazawa K, et al. MR tractography: a review of its clinical applications. Magn Reson Med Sci 2009;8:165-174. 6. Yoshioka H, Horikoshi T, Aoki S, et al. Diffusion tensor tractography predicts motor functional outcome in patients with spontaneous intracerebral hemorrhage. Neurosurgery 2008;62:97-103. 7. Kusano Y, Seguchi T, Horiuchi T, et al. Prediction of functional outcome in acute cerebral hemorrhage using diffusion tensor imaging at 3T: a prospective study. AJNR Am J Neuroradiol 2009;30:1561-1565.
T. KOYAMA ET AL. 8. Koyama T, Tsuji M, Miyake H, et al. Motor outcome for patients with acute intracerebral hemorrhage predicted using diffusion tensor imaging: an application of ordinal logistic modeling. J Stroke Cerebrovasc Dis 2012;21: 704-711. 9. Kuzu Y, Inoue T, Kanbara Y, et al. Prediction of motor function outcome after intracerebral hemorrhage using fractional anisotropy calculated from diffusion tensor imaging. Cerebrovasc Dis 2012;33:566-573. 10. Wang DM, Li J, Liu JR, et al. Diffusion tensor imaging predicts long-term motor functional outcome in patients with acute supratentorial intracranial hemorrhage. Cerebrovasc Dis 2012;34:199-205. 11. Koyama T, Marumoto K, Miyake H, et al. Relationship between diffusion-tensor fractional anisotropy and long-term outcome in patients with hemiparesis after intracerebral hemorrhage. Neurorehabilitation 2013;32: 87-94. 12. Shinohara Y, Yanagihara T, Abe K, et al. VII. Rehabilitation. J Stroke Cerebrovasc Dis 2011;20:S145-S180. 13. Broderick JP, Brott TG, Duldner JE, et al. Volume of intracerebral hemorrhage. A powerful and easyto-use predictor of 30-day mortality. Stroke 1993;24: 987-993. 14. Brunnstrom S. Motor testing procedures in hemiplegia: based on sequential recovery stages. Phys Ther 1966;46: 357-375. 15. Safaz I, Yilmaz B, Yasar E, et al. Brunnstrom recovery stage and motricity index for the evaluation of upper extremity in stroke: analysis for correlation and responsiveness. Int J Rehabil Res 2009;32: 228-231. 16. Naghdi S, Ansari NN, Mansouri K, et al. A neurophysiological and clinical study of Brunnstrom recovery stages in the upper limb following stroke. Brain Inj 2010;24: 1372-1378. 17. Heinemann AW, Linacre JM, Wright BD, et al. Relationships between impairment and physical disability as measured by the functional independence measure. Arch Phys Med Rehabil 1993;74:566-573. 18. Stineman MG, Fiedler RC, Granger CV, et al. Functional task benchmarks for stroke rehabilitation. Arch Phys Med Rehabil 1998;79:497-504. 19. Smith SM, Jenkinson M, Woolrich MW, et al. Advances in functional and structural MR image analysis and implementation as FSL. Neuroimage 2004;23(Suppl 1): S208-219. 20. Woolrich MW, Jbabdi S, Patenaude B, et al. Bayesian analysis of neuroimaging data in FSL. Neuroimage 2009; 45:S173-S186. 21. Jenkinson M, Beckmann CF, Behrens TE, et al. FSL. Neuroimage 2012;62:782-790. 22. Mori S, Oishi K, Jiang H, et al. Stereotaxic white matter atlas based on diffusion tensor imaging in an ICBM template. Neuroimage 2008;40:570-582. 23. Koyama T, Matsumoto K, Okuno T, et al. Relationships between independence level of single motor-FIM items and FIM-motor scores in patients with hemiplegia after stroke: an ordinal logistic modelling study. J Rehabil Med 2006;38:280-286. 24. Kwakkel G, Wagenaar RC, Twisk JW, et al. Intensity of leg and arm training after primary middle-cerebralartery stroke: a randomised trial. Lancet 1999;354: 191-196. 25. Sulter G, Steen C, De Keyser J. Use of the Barthel index and modified Rankin scale in acute stroke trials. Stroke 1999;30:1538-1541.
INTRACEREBRAL HEMORRHAGE OUTCOME ASSESSMENT BY DTI-FA 26. Kwon S, Hartzema AG, Duncan PW, et al. Disability measures in stroke: relationship among the Barthel Index, the Functional Independence Measure, and the Modified Rankin Scale. Stroke 2004; 35:918-923. 27. Jorgensen HS, Nakayama H, Raaschou HO, et al. Outcome and time course of recovery in stroke. Part I:
889
outcome. The Copenhagen Stroke Study. Arch Phys Med Rehabil 1995;76:399-405. 28. Koyama T, Marumoto K, Miyake H, et al. Relationship between diffusion tensor fractional anisotropy and long-term motor outcome in patients with hemiparesis after middle cerebral artery infarction. J Stroke Cerebrovasc Dis 2014;23:2397-2404.
Background: This study aimed to evaluate the prognostic efficacy of magnetic resonance diffusion tensor fractional anisotropy (FA) for patients with hemiparesis due to intracerebral hemorrhage. Methods: Diffusion tensor FA brain images were acquired 14-21 days after putaminal and/or thalamic hemorrhage. The ratio of FA values within the cerebral peduncles of the affected and unaffected hemispheres (rFA) was calculated for each patient (n 5 40) and assessed for correlation with Brunnstrom stage (BRS, 1-6), motor component of the functional independence measure (FIM-motor, 13-91), and the total length of stay (LOS) until discharge from rehabilitation (P ,.05). Ordinal logistic regression analyses were conducted to determine the relationships between rFA and specific outcomes as measured by BRS range (poor, BRS 1 or 2; moderate, BRS 3 or 4; and good, BRS 5 or 6; P , .05). Results: The rFA values were .571-1.043 (median, .856) and BRS scores were 1-6 (median, 4) for shoulder/elbow/forearm, 1-6 (median, 4) for hand, and 2-6 (median, 4) for lower extremities. FIM-motor scores were 58-86 (median, 78) and LOS ranged from 42 to 225 days (median, 175.5 days). Correlation coefficients were statistically significant between rFA and shoulder/elbow/forearm BRS (.696), hand BRS (.779), lower extremity BRS (.631), FIM-motor (.442), and LOS (2.598). Logistic model fit was moderate for shoulder/elbow/forearm BRS (R2 5 .221) and lower extremity BRS (R2 5 .277), but was much higher for hand BRS (R2 5 .441). Conclusions: Diffusion tensor FA values are predictive of clinical outcome from hemiparesis due to putaminal and/or thalamic hemorrhage, particularly hand function recovery. Key Words: Hematoma—paresis—probability—prognosis—recovery— stroke. Ó 2015 by National Stroke Association
From the *Department of Rehabilitation Medicine, Nishinomiya Kyoritsu Neurosurgical Hospital, Nishinomiya, Hyogo; †Department of Rehabilitation Medicine, Hyogo College of Medicine, Nishinomiya, Hyogo; and ‡Department of Neurosurgery, Nishinomiya Kyoritsu Neurosurgical Hospital, Nishinomiya, Hyogo, Japan. Received October 21, 2014; revision received December 3, 2014; accepted December 5, 2014. Grand-in-Aid for Scientific Research (B), the Japan Society for the Promotion of Science (KAKENHI), 25282168 and the Medical Research Fund of Hyogo Medical Association, MRF-H-08-12. Address correspondence to Tetsuo Koyama, MD, PhD, Department of Rehabilitation Medicine, Nishinomiya Kyoritsu Neurosurgical Hospital, 11-1 Imazu-Yamanaka-cho, Nishinomiya, Hyogo, Japan 663-8211. E-mail: [email protected]. 1052-3057/$ - see front matter Ó 2015 by National Stroke Association http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2014.12.011
Stroke is a leading cause of disability in most advanced countries.1 Stroke due to intracerebral hemorrhage is often associated with severe impairment such as hemiparesis,2 resulting in poor functional outcome.3,4 To facilitate the most effective rehabilitation, the evaluation of brain images in relation to clinical severity is critically important. Magnetic resonance (MR) diffusion tensor imaging (DTI) has recently been applied to assess white matter degeneration after stroke.5 DTI detects the diffusion gradient path of water molecules to reveal preserved axonal fibers, and so, enables clinically useful characterization of Wallerian degeneration after intracerebral hemorrhage. Of the parameters obtained from DTI, fractional anisotropy (FA) has proven to be a useful index of axonal degeneration, and several studies have attempted to use
Journal of Stroke and Cerebrovascular Diseases, Vol. 24, No. 4 (April), 2015: pp 881-889
881
T. KOYAMA ET AL.
882
FA values to characterize the relationships between neural degeneration and clinical manifestations of intracerebral hemorrhage.6-11 Patients often exhibit different outcomes for upper and lower extremity functions after intracerebral hemorrhage, with lower extremity functions generally showing better recovery than upper extremity functions.8,11 Thus, many patients resume independent walking, leading to functional independence in activities of daily living (ADL). However, few previous studies have reported on specific clinical outcomes and their relation to DTI, analyses may be useful for tailoring rehabilitation regimens to target the most severe reversible functional deficits. In this study, we used a variety of analytical procedures to assess the relationships between DTI-FA and multiple clinical outcomes. The results suggest that hand functions are most severely affected by putaminal and/or thalamic intracerebral hemorrhage, whereas lower extremity function and ADL are relatively spared.
Methods Patients The study sampled patients with intracerebral hemorrhage who were admitted to Nishinomiya Kyoritsu Neurosurgical Hospital between December 2009 and March 2014. Patients were typically transferred to our hospital soon after onset and underwent conservative treatment such as medication to reduce hypertension and, when necessary, surgical removal of hematoma. During hospitalization, they also underwent physical therapy, occupational therapy, and speech therapy for a combined daily total of up to 180 minutes. The protocols for these rehabilitative treatments followed the conventional methods stated in the Japanese Guidelines for the Management of Stroke.12 The work presented here extends that of our previously published study11 that included data from 12 patients already entered in the database for the present study. The Ethics Committee of Hyogo College of Medicine approved the study protocol, and patients (or relatives when necessary) provided written informed consent. To minimize the variability arising from differences in prestroke health status and lesion site, the sample population was limited to first-ever stroke patients with thalamic and/or putaminal hemorrhage who were able to walk unaided and had been functionally independent in ADL before stroke. For MR imaging safety, patients with implanted metal items (eg, artificial pacemakers) were excluded. Patients who subsequently required acute medical services (for recurrence of stroke, angina pectoris, or other comorbid conditions) were also excluded. To minimize variability arising from differences in the rehabilitative therapeutic regimen, this study included data only from patients who were transferred to our affiliated
long-term rehabilitation facility (Nishinomiya Kyoritsu Rehabilitation Hospital) to receive inpatient rehabilitative care for at least 1 month.
Computed Tomography Acquisition On arrival at our hospital, patients manifesting hemiparesis or other symptoms of stroke (hemorrhagic or ischemic) underwent head computed tomography with an Aquilion 64SP scanner (Toshiba Medical Systems Corp., Tochigi, Japan). Imaging parameters were 120 kVp and 250 mA, in-plane resolution was .86 mm 3 .86 mm, and slice thickness was 8 mm. The volume of intracerebral hemorrhage was estimated conventionally.13
DTI Acquisition DTI was performed 14-21 days after admission using a 3.0 T MR scanner (Trio; Siemens AG, Erlangen, Germany) with a 32-channel head coil. Details of the DTI acquisition protocol were reported in our previous studies.8,11 In brief, the DTI protocol acquired 12 images with noncollinear diffusion gradients (b 5 1000 seconds/ mm2) and 1 nondiffusion-weighted image (b 5 0 seconds/mm2) using a single-shot echo-planar imaging sequence. A total of 64 axial slices were obtained from each patient. The field of view was 230.4 mm 3 230.4 mm, the acquisition matrix was 128 3 128, and slice thickness was 3 mm without a gap. Echo time was 83 milliseconds and repetition time was 7000 milliseconds.
Outcome Measurements Brunnstrom stage (BRS),14 which is commonly used by Japanese rehabilitation therapists,12 was adopted to assess post–intracerebral hemorrhage motor impairments of the upper and lower extremities on the affected side. In this assessment, recovery of the affected extremities was evaluated by associated reactions and flexion and extension synergy patterns on a 6-point scale from severe (1) to normal (6). Conventionally, BRS is used for separate functional evaluation of the proximal (shoulder/elbow/ forearm) and distal (hand) upper extremity and the entire lower extremity, and its reliability and validity are well established.15,16 Assessments were made by occupational or physical therapists blinded to the purpose of the study. In addition to extremity functions, we obtained scores on the motor component of the functional independence measure (FIM-motor). The FIM is a test battery commonly used for evaluating stroke rehabilitation.17,18 It consists of individual 7-point scales (from total assistance to complete independence) for the following 13 items: eating, grooming, bathing, dressing upper body, dressing lower body, toileting, bladder and bowel management, transfer to bed/chair/wheelchair, transfer to toilet, transfer to tub/shower, walking or wheelchair propulsion, and stair
INTRACEREBRAL HEMORRHAGE OUTCOME ASSESSMENT BY DTI-FA
883
Table 1. Patient profiles Cerebral peduncle Patient
Age (y), M/F
Hemorrhage site, volume (mL)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
64, M 49, M 51, M 68, M 76, M 89, M 55, M 48, F 73, M 53, M 69, M 56, M 73, M 77, F 77, F 69, F 72, M 80, M 62, M 62, F 83, M 50, M 43, F 61, F 50, M 59, M 44, M 71, M 39, F 64, M 69, M 57, M 57, M 61, M 73, M 62, F 74, M 31, F 46, F 55, M
R thal, 7.4 L thal, 13.5 R thal, 17.7 R thal, 7.8 R put, 25.0 L thal/put, 3.8 L thal, 2.8 R thal/put, 4.5 R thal, 6.4 R put, 23.8 L thal, 4.5 L put, 5.6 R thal, 12.6 R thal/put, 2.8 R thal, 4.1 L put, 18.3 L put, 7.6 L thal, 10.5 R thal/put, 3.0 R put, 9.7 L put, 3.9 R put, 20.4 R put, 11.2 L put, 20.1 L put, 62.3 R put, 24.6 R put, 43.5 R thal, 6.4 L put, 27.8 R put, 49.4 L put, 62.8 L put, 6.1 L put, 26.9 L put, 15.2 L put, 47.1 L thal, 8.7 L put, 17.5 L put, 29.7 L thal/put, 53.8 L put, 32.7
BRS
FA (R)
FA (L)
rFA
S/E/F
Hand
L/E
FIM-M
LOS(d)
.587 .509 .519 .577 .562 .574 .538 .595 .571 .561 .579 .576 .462 .472 .526 .552 .587 .545 .475 .489 .569 .472 .509 .551 .527 .419 .442 .439 .595 .433 .535 .567 .579 .558 .595 .536 .594 .585 .537 .563
.562 .508 .530 .592 .579 .556 .520 .617 .598 .604 .537 .515 .524 .535 .599 .484 .510 .472 .555 .571 .487 .554 .607 .458 .437 .523 .555 .566 .460 .562 .394 .414 .421 .396 .422 .360 .382 .367 .315 .321
1.043 .999 .979 .976 .969 .969 .968 .965 .955 .929 .927 .895 .883 .882 .879 .877 .870 .866 .857 .856 .855 .853 .838 .832 .829 .800 .796 .776 .772 .769 .736 .731 .727 .710 .709 .671 .643 .628 .587 .571
5 5 5 4 5 5 6 6 5 6 6 4 4 3 4 5 2 5 4 5 4 5 5 2 2 4 2 5 1 2 1 3 2 4 1 2 3 1 2 3
5 5 5 4 5 5 6 6 5 6 5 5 4 3 6 5 2 5 4 4 4 5 5 2 2 3 2 5 1 2 1 2 2 2 1 1 2 1 2 2
5 5 5 3 5 4 6 6 5 6 5 4 4 5 5 5 3 5 4 6 4 4 6 5 3 5 3 4 3 2 3 4 3 4 2 2 3 3 2 4
86 77 79 77 77 79 85 81 78 84 80 83 75 65 72 81 75 66 79 79 75 81 85 78 75 86 79 80 78 65 65 80 64 78 74 58 64 72 77 77
83 65 153 101 67 165 175 85 147 45 42 165 176 181 183 124 195 203 203 194 191 122 64 116 216 191 172 105 182 198 212 204 166 202 188 204 204 224 225 169
Abbreviations: BRS, Brunnstrom stage; F, female; FA, fractional anisotropy; FIM-M, functional independence measure motor; L/E, lower extremity; LOS, length of total hospital stay from admission to acute medical service to discharge from long-term rehabilitation facility; M, male; put, putamen; rFA, ratio between FA values in affected and unaffected hemispheres; S/E/F, shoulder, elbow, and forearm; thal, thalamus. Patient data are ordered 1-40 according to rFA values (highest to lowest). Eight patients (Nos. 25, 27, 29, 30, 31, 33, 38, and 39) underwent surgical removal of hematoma. 12 patients (Nos. 1, 4, 9, 13, 14, 15, 24, 29, 32, 35, 36, and 38) were included in the database in our previous study.11
climbing. The total scores of these items are often used as indices of independence in ADL (scale range, 13-91). BRS and FIM-motor score were assessed monthly, and data were collected from our long-term rehabilitation at discharge. Discharge from this facility was decided when the increase in FIM-motor score was 1 or lower during the month. The total length of hospital stay (LOS) was also recorded for each patient.
Image Processing The brain image analysis package FSL (Analysis Group, FMRIB, Oxford, UK) was used for image processing.19-21 The details of image analysis were the same as those reported in our previous studies.8,11 In brief, DTI data were first corrected for motion and eddy current distortions by aligning all images to the first image
T. KOYAMA ET AL.
884
Figure 1. Computed tomography images and diffusion tensor imaging-fractional anisotropy maps of the cerebral peduncles in 5 representative cases. Abbreviations: L, left; rFA, ratio of fractional anisotropy; R, right.
(b 5 0 seconds/mm2), and then extracerebral regions were excluded from the images. Regional brain FA values were then calculated to yield an FA brain map for each patient. To assess neural degeneration due to intracerebral hemorrhage, regions of interest were defined in the left and right cerebral peduncles, which were abstracted by referring to the The International Consortium for Brain Mapping DTI-81 standard brain.22 FA values within the left and right regions of interest were calculated, and mean values for single voxels were subsequently estimated. Then, the ratio of FA values in the affected and unaffected hemispheres (rFA) was calculated for each patient as an index of neural degeneration in the corticospinal tracts.6-11
Statistical Analysis Spearman rank correlations were calculated to evaluate the relationships between rFA and clinical parameters (BRS, FIM-motor, and LOS). Ordinal logistic modeling was then used to test the association between rFA and individual outcomes (shoulder/elbow/forearm, hand, and lower extremity) as measured by BRS range (poor, moderate, or good).8 The principle of logistic modeling involves fitting the probability (p) of a dichotomous response (such as ‘‘yes/no’’ or ‘‘dead/alive’’) to a linear model. The probability odds for these types of dichotomous responses [p/(1 2 p)] can take any positive value. The logarithm of these odds is modeled as a simple regression, and parameter estimates are assessed for fit to the model.
log½p=ð12pÞ5a1bX
where a is a constant; b is a coefficient; and X, explanatory variable. To extend the utility to multi-level ordinal responses, modeling the odds to a simple regression cumulative probability is performed at each level.23 For 3-level responses (p1, p2, p3; sum equals 1), the logarithm of the odds is modeled as 2 simple regressions for the 3-level responses, and parameter estimates are assessed for fit to the model (note that single b and 2 levels of a are assessed).
log½p1 =ðp2 1p3 Þ5a1 1b1 X log½ðp1 1p2 Þ=p3 5a2 1b1 X In the present study, to simplify the logistic regression analyses, we defined 3 levels for BRS; poor (BRS 1 or 2), moderate (BRS 3 or 4), and good (BRS 5 or 6). Goodness-of-fit of the logistic modeling was assessed using the Wald chi-square test. In this study, all statistical analysis was performed using JMP software (SAS Institute, Cary, NC), and P less than .05 was considered statistically significant.
Results During the study period, 89 patients underwent DTI, of which 11 were discharged home directly from our hospital, 38 were transferred to other long-term rehabilitation hospitals or nursing homes (20 facilities in total) typically closer to their homes, and the remaining 40 were transferred to our long-term rehabilitation hospital and included in this study.
INTRACEREBRAL HEMORRHAGE OUTCOME ASSESSMENT BY DTI-FA
885
Figure 2. Results from correlation analyses between Brunnstrom stage, FIM-motor scores, length of hospital stay, and rFA. Dashed lines indicate density ellipse (.90). Abbreviations: BRS, Brunnstrom stage; FIM, functional independence measure; rFA, ratio of fractional anisotropy.
Table 1 presents the profiles of the 40 patients transferred to our long-term facility (median age, 62 years; age range, 31-89 years), 18 with right hemisphere lesions and 22 with left hemisphere lesions. rFA values ranged from .571 to 1.043 (median, .856). BRS data ranged from 1 to 6 (median, 4) for shoulder/elbow/forearm, 1 to 6 (median, 4) for hand, and 2 to 6 (median, 4) for lower extremity. FIM-motor scores ranged from 58 to 86 (median, 78) and LOS from 42 to 225 days (median, 175.5 days). Figure 1 shows brain images from 5 represen-
tative patients (numbers 1, 10, 20, 30, and 40), where patients are listed in order of descending rFA. Rank correlation analyses (Fig 2) revealed moderate-tohigh correlations between rFA and extremity function as assessed by BRS: shoulder/elbow/forearm (.696; P , .001), hand (.779; P , .001), and lower extremity (.631; P , .001). In contrast, the correlation between rFA and FIM-motor was weaker (.442; P 5 .004). There was a negative correlation between rFA and LOS (2.598, P , .001).
T. KOYAMA ET AL.
886
Table 2. Parameter estimates for ordinal logistic equations between BRS and rFA N 5 40
Ordinal logistics BRS portion Shoulder/elbow/forearm Estimates Standard error P value Hand Estimates Standard error P value Lower extremity Estimates Standard error P value
b
a, Poor (BRS 1 or 2)
Moderate (BRS 3 or 4)
213.224 3.621 ,.001
9.794 2.937 ,.001
11.689 3.132 ,.001
227.956 7.293 ,.001
22.779 6.107 ,.001
24.428 6.323 ,.001
215.359 4.123 ,.001
9.422 3.003 .002
13.403 3.571 ,.001
R2
.221 ,.001 .441 ,.001 .277 ,.001
Abbreviations: BRS, Brunnstrom stage; rFA, ratio of fractional anisotropy.
Table 2 shows the results obtained from ordinal logistic regression analyses of rFA values and BRS values for shoulder/elbow/forearm, hand, and lower extremity. In these analyses, the fits of ordinal logistic modeling were statistically significant, indicating that the data set results could be validly interpreted by logistic probability. Model fit was much greater for hand (R2 5 .441) than that for shoulder/elbow/forearm (R2 5 .221) or lower extremity (R2 5 .277). Figure 3 shows logistic probability plots of individual rFA values versus BRS scores. In each plot, the leftmost curve defines the probability for poor recovery (BRS 1 or 2) and the rightmost curve the probability of good recovery (BRS 5 or 6) versus rFA value. The distance from the leftmost curve to the bottom of the graph is the probability of poor recovery, the distance between the 2 curves is the probability of moderate recovery (BRS 3 or 4), and the distance from the rightmost curve to the top of the graph is the probability of good recovery at a given rFA. For the shoulder/elbow/forearm, when rFA was around .7, estimated probability for poor recovery was around 60%. When rFA was around .9, the estimated probability for good recovery was close to 55%. The middle panel shows equivalent data for the hand. In this case, when rFA was around .7, estimated probability for poor recovery was nearly 95%, whereas when it was around .9, the estimated probability for good recovery was close to 70%. For the lower extremity (lower panel), even when rFA was around .7, estimated probability for poor recovery was only around 20%. This indicates that compared with the upper extremity, the lower extremity recovered relatively well even when rFA was low.
Discussion Previous DTI studies after intracerebral hemorrhage reported that lower FA values in the corticospinal
tracts were associated with more severe hemiparesis.6-10 Similarly, in our preliminary study (n 5 12) assessing the relationship between DTI-FA and clinical manifestations using simple correlation analysis, we found a strong correlation between rFA and upper extremity function.11 To further extend these results, in the present study, we included additional patients (n 5 40) and conducted ordinal logistic regression to evaluate the correlation between specific functional outcomes and rFA. The results confirmed that DTI-FA values are strongly predictive of upper extremity function. Indeed, rFA was more strongly associated with outcome in upper extremity function than FIM-motor scores. In our preliminary report, we applied only rank correlation to simplify the analysis.11 However, this methodology may not be sensitive enough to identify relationships in cases of uneven distribution patterns. For example, in our present sample (Fig 2 and Table 1), 16 cases (40%) showed good recovery of hand function (BRS 5 or 6), 17 (42.5%) showed poor recovery (BRS 1 or 2), and only 7 (17.5%) showed moderate recovery (BRS 3 or 4). To address this concern, we used ordinal logistic regression analysis and found a strong significant correlation between rFA and hand function outcome. For shoulder/ elbow/forearm and lower extremity, the relationships with rFA were weaker, as indicated by the shallow logistic curves and wide region separating curves for probability of good and poor recovery (Fig 3, upper and lower panels). In contrast, logistic curves for hand function were much steeper with a narrower zone for moderate recovery (Fig 3, middle panel). Thus, DTI-FA was most strongly predictive of hand function outcome. In contrast to the results for the upper extremity, logistic regression for the lower extremity indicated that moderate or good outcome is to be expected even when rFA values are lower (around .7; Fig 3). Similarly, the results from simple correlation analyses revealed that
INTRACEREBRAL HEMORRHAGE OUTCOME ASSESSMENT BY DTI-FA
Figure 3. Logistic probability plots for the relationship between rFA and stratified individual BRS outcomes (good, moderate, and poor). Abbreviations: BRS, Brunnstrom stage; rFA, ratio of fractional anisotropy.
887
the relationship between rFA and FIM-motor scores, although statistically significant, was rather weak (correlation coefficient of .442). This is consistent with previous studies indicating that independence of ADL (of which FIM-motor is a component) is most strongly associated with locomotive ability rather than upper extremity function.24,25 Taken together, these results suggest that some patients with severe neural damage within the corticospinal tracts (as measured by rFA) can still achieve independence with minimal improvement of upper extremity function.11 Indeed, our data for FIMmotor scores ranged from 58 to 85 (Table 1), a range in which most of the patients could walk independently with or without canes.23,26 In this study, we collected data from patients at discharge from our affiliated rehabilitation facility after 42-225 days of therapy. This variation in poststroke treatment duration may confound outcome data because greater recovery is expected among patients discharged earlier (eg, ,90 days).27 Nonetheless, we entered all outcome data in our final analytical database for the following reasons. As shown in Figure 2, analysis detected a significant negative correlation between rFA and LOS, indicating that patients with less-severe neural degeneration tended to be discharged earlier. In parallel, patients with higher rFA showed better outcome for both BRS and FIM-motor parameters. Better recovery is expected for such patients with relatively mild clinical manifestations. However, ceiling effects will minimally affect the statistical significance of the data set. Accordingly, we entered all data for all 40 patients in our final analytical database despite the wide variation in LOS. In this study, of the 89 total samples, we entered 40 (45%) into the final analytical database. Given such a small population, confounds arising from sampling bias may be a critical concern. The aim of this study was to determine the relationship between FA decrease within the cerebral peduncles and the clinical severity such as hemiparesis. Of the 89 patients, 11 were discharged home directly from our hospital. As shown in our previous study, patients with less-severe hemiparesis were associated with less significant decrease of FA values in the cerebral peduncles.8 It is very common for patients who exhibit mild hemiparesis to return home directly from acute care service. Accordingly, in this study, we focused on patients who were transferred to long-term facilities.28 In this study, the assignment of the facilities was typically based on the patients’ areas of residence. Accordingly, the possibility of sampling bias was minimal. This study has several limitations. First, we recruited subjects only with thalamic and/or putaminal hemorrhage to reduce variability, but patients with subcortical and pontine hemorrhage are also commonly encountered in clinical practice. Such patients should be
888
included in larger scale studies to clarify the utility and limitations of DTI-based prognosis for the general population. Second, the sample size was small, partially because of the relatively restrictive inclusion criteria. To minimize variability arising from different rehabilitative regimens, we collected data only from patients transferred to a single rehabilitation hospital. Studies enrolling greater numbers of patients from multiple rehabilitative facilities using similar regimens are required. Third, we sampled data from patients who had no neurologic disease other than stroke. Consequently, patients with commonly observed geriatric comorbidities such as dementia, Parkinsonism, and epilepsy were excluded. Again, careful consideration should be given when applying our findings to the general geriatric population. Fourth, we focused on clinical severity using BRS and FIM-motor assessment, and impairment of higher brain functions such as aphasia and hemineglect may be overlooked by these scoring systems. Despite these limitations, the application of both simple rank correlation and logistic regression proved useful for characterizing the prognostic efficacy of DTI-FA for patient outcome after intracerebral hemorrhage. In summary, we have demonstrated that DTI-FA can be used effectively as a prognostic index of clinical outcome after putaminal and/or thalamic hemorrhage. Our findings suggest that DTI-FA is particularly useful for predicting recovery of hand function. These results may be useful for tailoring rehabilitation accordingly.
References 1. Hankey GJ, Jamrozik K, Broadhurst RJ, et al. Long-term disability after first-ever stroke and related prognostic factors in the Perth Community Stroke Study, 19891990. Stroke 2002;33:1034-1040. 2. Jorgensen HS, Nakayama H, Raaschou HO, et al. Intracerebral hemorrhage versus infarction: stroke severity, risk factors, and prognosis. Ann Neurol 1995;38:45-50. 3. Daverat P, Castel JP, Dartigues JF, et al. Death and functional outcome after spontaneous intracerebral hemorrhage. A prospective study of 166 cases using multivariate analysis. Stroke 1991;22:1-6. 4. Katrak PH, Black D, Peeva V. Do stroke patients with intracerebral hemorrhage have a better functional outcome than patients with cerebral infarction? PM R 2009;1:427-433. 5. Yamada K, Sakai K, Akazawa K, et al. MR tractography: a review of its clinical applications. Magn Reson Med Sci 2009;8:165-174. 6. Yoshioka H, Horikoshi T, Aoki S, et al. Diffusion tensor tractography predicts motor functional outcome in patients with spontaneous intracerebral hemorrhage. Neurosurgery 2008;62:97-103. 7. Kusano Y, Seguchi T, Horiuchi T, et al. Prediction of functional outcome in acute cerebral hemorrhage using diffusion tensor imaging at 3T: a prospective study. AJNR Am J Neuroradiol 2009;30:1561-1565.
T. KOYAMA ET AL. 8. Koyama T, Tsuji M, Miyake H, et al. Motor outcome for patients with acute intracerebral hemorrhage predicted using diffusion tensor imaging: an application of ordinal logistic modeling. J Stroke Cerebrovasc Dis 2012;21: 704-711. 9. Kuzu Y, Inoue T, Kanbara Y, et al. Prediction of motor function outcome after intracerebral hemorrhage using fractional anisotropy calculated from diffusion tensor imaging. Cerebrovasc Dis 2012;33:566-573. 10. Wang DM, Li J, Liu JR, et al. Diffusion tensor imaging predicts long-term motor functional outcome in patients with acute supratentorial intracranial hemorrhage. Cerebrovasc Dis 2012;34:199-205. 11. Koyama T, Marumoto K, Miyake H, et al. Relationship between diffusion-tensor fractional anisotropy and long-term outcome in patients with hemiparesis after intracerebral hemorrhage. Neurorehabilitation 2013;32: 87-94. 12. Shinohara Y, Yanagihara T, Abe K, et al. VII. Rehabilitation. J Stroke Cerebrovasc Dis 2011;20:S145-S180. 13. Broderick JP, Brott TG, Duldner JE, et al. Volume of intracerebral hemorrhage. A powerful and easyto-use predictor of 30-day mortality. Stroke 1993;24: 987-993. 14. Brunnstrom S. Motor testing procedures in hemiplegia: based on sequential recovery stages. Phys Ther 1966;46: 357-375. 15. Safaz I, Yilmaz B, Yasar E, et al. Brunnstrom recovery stage and motricity index for the evaluation of upper extremity in stroke: analysis for correlation and responsiveness. Int J Rehabil Res 2009;32: 228-231. 16. Naghdi S, Ansari NN, Mansouri K, et al. A neurophysiological and clinical study of Brunnstrom recovery stages in the upper limb following stroke. Brain Inj 2010;24: 1372-1378. 17. Heinemann AW, Linacre JM, Wright BD, et al. Relationships between impairment and physical disability as measured by the functional independence measure. Arch Phys Med Rehabil 1993;74:566-573. 18. Stineman MG, Fiedler RC, Granger CV, et al. Functional task benchmarks for stroke rehabilitation. Arch Phys Med Rehabil 1998;79:497-504. 19. Smith SM, Jenkinson M, Woolrich MW, et al. Advances in functional and structural MR image analysis and implementation as FSL. Neuroimage 2004;23(Suppl 1): S208-219. 20. Woolrich MW, Jbabdi S, Patenaude B, et al. Bayesian analysis of neuroimaging data in FSL. Neuroimage 2009; 45:S173-S186. 21. Jenkinson M, Beckmann CF, Behrens TE, et al. FSL. Neuroimage 2012;62:782-790. 22. Mori S, Oishi K, Jiang H, et al. Stereotaxic white matter atlas based on diffusion tensor imaging in an ICBM template. Neuroimage 2008;40:570-582. 23. Koyama T, Matsumoto K, Okuno T, et al. Relationships between independence level of single motor-FIM items and FIM-motor scores in patients with hemiplegia after stroke: an ordinal logistic modelling study. J Rehabil Med 2006;38:280-286. 24. Kwakkel G, Wagenaar RC, Twisk JW, et al. Intensity of leg and arm training after primary middle-cerebralartery stroke: a randomised trial. Lancet 1999;354: 191-196. 25. Sulter G, Steen C, De Keyser J. Use of the Barthel index and modified Rankin scale in acute stroke trials. Stroke 1999;30:1538-1541.
INTRACEREBRAL HEMORRHAGE OUTCOME ASSESSMENT BY DTI-FA 26. Kwon S, Hartzema AG, Duncan PW, et al. Disability measures in stroke: relationship among the Barthel Index, the Functional Independence Measure, and the Modified Rankin Scale. Stroke 2004; 35:918-923. 27. Jorgensen HS, Nakayama H, Raaschou HO, et al. Outcome and time course of recovery in stroke. Part I:
889
outcome. The Copenhagen Stroke Study. Arch Phys Med Rehabil 1995;76:399-405. 28. Koyama T, Marumoto K, Miyake H, et al. Relationship between diffusion tensor fractional anisotropy and long-term motor outcome in patients with hemiparesis after middle cerebral artery infarction. J Stroke Cerebrovasc Dis 2014;23:2397-2404.
