Content » Vol 42, Issue 1

Original report

Prediction of good functional recovery after stroke based on combined motor and somatosensory evoked potential findings

Sang Yoon Lee, MD1, Jong Youb Lim, MD1, 2, Eun Kyoung Kang, MD1, Moon-Ku Han, MD, PhD2, Hee-Joon Bae, MD, PhD2 and Nam-Jong Paik, MD, PhD1, 2

From the 1Department of Rehabilitation Medicine and 2Department of Neurology, Seoul National University College of Medicine, Stroke Center, Seoul National University Bundang Hospital, Seongnam, South Korea

OBJECTIVE: To delineate whether functional recovery after stroke, determined by the modified Rankin Scale during the neurologically stable chronic stage, is associated with the presence or absence of motor evoked potential or somatosensory evoked potential measured during the sub-acute stage at the commencement of rehabilitation.

DESIGN: Retrospective medical records review.

Patients: Consecutive 105 first-ever unilateral patients after stroke.

METHODS: Patients underwent motor evoked potential and somatosensory evoked potential studies at the commencement of rehabilitation (i.e. approximately 1 month post-onset), and functional recovery was measured using the modified Rankin Scale at 3 months post-onset. The independent abilities of motor evoked potentials and somatosensory evoked potentials for predicting good functional recovery (modified Rankin Scale 2) were determined by multivariable logistic regression analysis adjusted for age, laterality of lesion, and National Institute of Health Stroke Scale scores at onset of rehabilitation.

RESULTS: The adjusted logistic regression model revealed that patients with negative motor evoked potential or somatosensory evoked potential responses in the lower limb were less likely to achieve good functional recovery (odds ratio = 0.057~0.099, p < 0.05) relative to positive motor evoked potential and somatosensory evoked potential responses in the lower limb.

CONCLUSION: Evoked potential studies measured at the commencement of rehabilitation could be used in a complementary manner to predict functional recovery after stroke.

Key words: stroke; recovery; function; somatosensory evoked potential; motor evoked potential.

J Rehabil Med 2010; 42: 16–20

Correspondence address: Nam-Jong Paik, Department of Rehabilitation Medicine, Seoul National University College of Medicine, Seoul National University Bundang Hospital 166 Gumi-ro, Bundang-gu, Seongnam-si, Gyeonggi-do, 463-707, South Korea. E-mail: njpaik@snu.ac.kr

INTRODUCTION

The prediction of post-stroke functional recovery is important in stroke rehabilitation. Several clinical and demographic factors have been suggested previously as predictors of functional recovery after stroke (1).

Evoked potential (EP) studies provide a useful, objective means of assessing the integrities of cortico-spinal and somatosensory pathways, and therefore, EP findings have prognostic value in terms of predicting post-stroke functional recovery (2, 3). However, previous reports either involved small number of subjects or did not evaluate motor EP (MEP) and somatosensory EP (SEP) simultaneously.

In the present study, we studied whether functional recovery after stroke, determined by the modified Rankin Scale (mRS) during the neurologically stable chronic stage, is associated with the presence or absence of MEP or SEP measured during the sub-acute stage at the commencement of rehabilitation.

PATIENTS AND METHODS

Patients

We reviewed the medical records of 191 consecutive patients with first-ever unilateral stroke who commenced rehabilitative training approximately one month (mean 25.9 standard deviation (SD) 14.3 days) post-onset onset at a university hospital during the 14-month period from January 2007 to February 2008. Stroke was diagnosed based on clinical history, a physical examination (by neurologists), and confirmation by brain magnetic resonance imaging (MRI).

The exclusion criteria applied were: (i) a recurrent stroke history; (ii) the presence of nonvascular brain lesions, such as, lesions due to traumatic injury, a brain tumour, or Parkinsonism; (iii) no functional deficit (mRS = 0); (iv) a bed-ridden status and an inability to tolerate rehabilitative training (mRS = 5); (v) a co-morbidity, such as peripheral polyneuropathy; (vi) a concomitant severe medical illness that could affect the prognosis; and (vii) those in whom an MEP study was contraindicated, for example, those with an implanted metallic device, such as aneurysmal clip or cardiac pacemaker, and those with a history of epilepsy or craniotomy.

Measurement of functional recovery after stroke

Motor function was measured at the commencement of rehabilitation at mean 25.9 (SD 14.3) days post-onset and during the neurologically stable chronic stage (mean 96.0 (SD 18.1) days post-onset) using the mRS. Rehabilitative training was routinely administered during admission to the rehabilitation unit, consisting of 1–1.5 h of physical therapy (mobility and gait training), 1–1.5 h of occupational therapy (task-oriented training and activities of daily living training), and/or 1 h of speech therapy on each therapy day, which was divided into 2 daily sessions, 5 days a week, according to the individual patient’s needs.

The mRS is a clinician-reported measure of global disability with high validity and reliability (4) and is widely applied for evaluating stroke outcomes, degree of disability, and dependence with respect to the daily activities (5). We defined a good functional recovery as mRS 1 or 2 (6) out of 6 grades (0–5), where 0 corresponds to no symptoms and 5 to severe disability or a bedridden status (7). The protocol used was approved by the institutional review board of (our) institution.

Evoked potential studies

Neurophysiological measurements were carried out bilaterally on upper and lower limbs on the same day as mRS measurements. MEPs were measured in all 4 extremities (abductor pollicis brevis in upper limbs and the adductor hallucis in lower limbs) using a standard protocol (8). Reproducible responses with minimal peak to peak amplitude of 200 µV in at least 5 of 10 consecutive trials using a figure-of-eight coil were defined as a positive response (“MEP(+)”). SEPs were measured by stimulating median nerves at wrists and tibial nerves at ankles using a standard protocol (9). In cases with an amplitude (N1 to P1) difference of < 50% and a latency (N1) difference of < 10% compared with unaffected sides, the response was regarded as positive (“SEP(+)”).

Statistical analysis

Baseline demographics, stroke factors, and the presence of an EP response were compared with respect to functional recovery status at 3 months post-onset (good recovery mRS 2 and poor recovery mRS > 3). The independent effects of MEP and SEP on functional recovery were assessed using a multivariate logistic regression model, unadjusted or adjusted for 8 key prognostic factors, namely, sex, age, lesion location, lesion laterality, stroke type (ischaemic or haemorrhagic), the presence of diabetes mellitus, National Institute of Health Stroke Scale (NIHSS) at onset of rehabilitation, and for MEP and SEP responses. The adjusted model was developed by backward elimination using a significance level of 0.2 to enter and 0.05 to stay. Sex, lesion location, stroke type, and presence of diabetes mellitus were eliminated from the final adjusted regression model. The sensitivities and specificities of MEP and SEP for predicting a good functional recovery (mRS 2) were calculated. In addition, we also calculated the positive and negative predictive values of MEP and SEP.

All analyses were performed using SPSS 15.0 for Windows. Significance was accepted for < 0.05, and 95% confidence intervals (CI) are provided.

RESULTS

Of the 191 patients considered initially, 86 were excluded due to; recurrent stroke (n = 38), no or a severe functional deficit (n = 28), non-vascular brain lesions (n = 13), and diabetes mellitus requiring insulin therapy (n = 7). Thus, 105 patients constituted the study cohort. Baseline characteristics of the 105 subjects are shown in Table I.

Table I. Baseline characteristics of subjects

Characteristics

Mean age, years, n (SD)

65.7 (14.3)

Sex, male/female, n

59/46

Stroke localization, n

Cortical

7

Cortico-subcortical

16

Subcortical

44

Brain stem

19

Cerebellum

3

Multiple

16

Stroke type, n

Ischaemic

85

Haemorrhagic

20

Stroke laterality, n

Right hemisphere

54

Left hemisphere

51

mRS at onset of rehabilitation, mean (SD)

NIHSS at onset of rehabilitation, mean (SD)

3.32 ( 0.88)

6.16 (4.95)

mRS: modified Rankin Scale; NIHSS: National Institute of Health Stroke Scale; SD: standard deviation.

The mean mRS (from 3.32 (SD 0.88) to 2.47 (SD 1.19), p < 0.001 by paired t-test) and mean NIHSS (from 6.16 (SD 4.95) to 3.05 (SD 3.47), p < 0.001) significantly improved 3 months post-stroke from the rehabilitation start. Of the 105 patients, 53 patients (50.5%) achieved a good functional state at 3 months post-onset. By unadjusted analysis, patients older than 80 years were found to have a significantly lower odds ratio of achieving a good functional recovery than those younger than 50 years (unadjusted OR = 0.21, 95% CI = 0.05–0.93; p = 0.024). An NIHSS at onset of rehabilitation of 5 was found to be a significant predictor of a poor functional recovery (unadjusted OR = 0.11, 95% CI = 0.04–0.29; p < 0.001). Adjusted analysis also showed that age 80 years and an NIHSS at onset of rehabilitation 5 were significant predictors of a poor functional recovery (adjusted OR = 0.04, 0.06; p = 0.020, p < 0.001, respectively). Other demographic and stroke variables were not found to be significantly associated with recovery.

By unadjusted EP analysis, MEP(+) in a lower limb, SEP(+) in an upper and in a lower limb were found to be significantly associated with a better recovery than the corresponding negative responses (unadjusted OR = 0.34, 0.23, 0.17; p = 0.008, p = 0.001, p < 0.001, respectively). However, the presence of MEP and SEP responses were not found to be significantly associated with better recovery when the adjusted model was used.

We then applied cross-matched EP analysis. By unadjusted cross-matched EP analysis, the upper limb MEP(–)SEP(–) group had a significantly lower OR for a good recovery (unadjusted OR = 0.20, 95% CI 0.07–0.56; p = 0.002) than the MEP(+)SEP(+) group. Furthermore, the lower limb MEP(+)SEP(–), MEP(–)SEP(+), and MEP(–)SEP(–) groups had significant lower OR for a good recovery than the MEP(+)SEP(+) group (unadjusted OR = 0.19, 0.07, 0.07; p = 0.016, p = 0.001, p < 0.001, respectively).

By adjusted EP analysis, only cross-matched EP findings in a lower limb were found to be significantly associated with a good recovery. The lower limb MEP(+)SEP(–), MEP(–)SEP(+), MEP(–)SEP(–) groups had significantly lower OR than the MEP(+)SEP(+) group (adjusted OR = 0.09, 0.06, 0.10; p = 0.033, p = 0.013, p = 0.040, respectively) (Table II).

Table II. Unadjusted and adjusted odds ratios (OR) of good functional recovery (mRS ≤ 2) at 3 months after stroke in 105 patients

n

Unadjusted OR* (95% CI)

p-value

Adjusted OR† (95% CI)

p-value

Sex

Male

56

1.000

ni

Female

49

0.657 (0.304–1.421)

0.286

Age, years

< 50

15

1.000

1.000

50–59

11

1.333 (0.242–7.348)

0.741

0.683 (0.067–6.973)

0.748

60–69

29

0.818 (0.221–3.031)

0.764

0.458 (0.050–4.158)

0.487

70–79

33

0.286 (0.079–1.034)

0.056

0.239 (0.035–1.613)

0.142

80

17

0.208 (0.047–0.931)

0.024

0.039 (0.030–0.603)

0.020

Lesion

Cortical

7

1.000

Cortico-subcortical

16

0.341 (0.055–2.131)

0.250

Subcortical

44

0.750 (0.150–3.750)

0.726

ni

Brain stem

19

2.812 (0.438–18.056)

0.276

Cerebellum

3

0.375 (0.022–6.348)

0.497

Multiple

16

0.450 (0.074–2.741)

0.386

Laterality of lesion

Right

54

1.000

1.000

Left

51

1.718 (0.769–3.839)

0.187

2.179 (0.574–8.269)

0.252

Stroke type

Ischaemic

85

1.000

ni

Haemorrhagic

20

0.977 (0.369–2.587)

0.962

DM

(–)

77

1.000

ni

(+)

28

1.184 (0.498–2.817)

0.702

NIHSS at onset of rehabilitation

< 5

42

1.000

1.000

5

49

0.109 (0.042–0.286)

< 0.001

0.058 (0.012–0.287)

< 0.001

MEP – Upper

(+)

34

1.000

1.000

(–)

71

0.508 (0.220–1.171)

0.112

2.411 (0.517–11.246)

0.263

MEP – Lower

(+)

46

1.000

1.000

(–)

59

0.341 (0.153–0.759)

0.008

0.366 (0.076–1.767)

0.211

SEP – Upper‡

(+)

58

1.000

1.000

(–)

44

0.227 (0.098–0.526)

0.001

0.584 (0.133–2.576)

0.478

SEP – Lower‡

(+)

51

1.000

1.000

(–)

49

0.167 (0.070–0.396)

< 0.001

0.329 (0.094–1.159)

0.084

EP – Upper

MEP (+) SEP (+)

28

1.000

1.000

MEP (+) SEP (–)

3

0.947 (0.316–2.840)

0.923

2.755 (0.392–19.375)

0.309

MEP (–) SEP (+)

30

0.947 (0.076–11.870)

0.967

0.377 (0.007–20.126)

0.631

MEP (–) SEP (–)

41

0.196 (0.069–0.555)

0.002

0.925 (0.110–7.796)

0.943

EP – Lower

MEP (+) SEP (+)

29

1.000

1.000

MEP (+) SEP (–)

22

0.192 (0.050–0.739)

0.016

0.091 (0.010–0.820)

0.033

MEP (–) SEP (+)

13

0.071 (0.015–0.346)

0.001

0.057 (0.006–0.552)

0.013

MEP (–) SEP (–)

36

0.070 (0.020–0.251)

< 0.001

0.099 (0.011–0.900)

0.040

*Unadjusted odds ratios by logistic regression analysis.

†Adjusted odds ratios by multivariate logistic regression analysis; adjusted for all other variables in model.

‡Three and 5 patients (upper limb and lower limb SEP study, respectively) were excluded because SEP study were impossible by their poor compliance.

CI: confidence interval; DM: diabetes mellitus; EP: evoked potentials; MEP: motor evoked potentials; ni: not included in the final adjusted model; NIHSS: National Institute of Health Stroke Scale; SEP: somatosensory evoked potentials.

The relationship between a good functional recovery at 3 months post-onset and the results of EP studies at rehabilitation commencement are shown in Table III. Sensitivity and specificity for predicting a good functional recovery were 39.6% and 75.0% for upper limb MEP, 56.6% and 69.2% for lower limb MEP, 73.6% and 61.2% for upper limb SEP, and 71.2% and 70.8% for lower limb SEP, respectively.

Table III. Contingency table for good functional recovery in relation to the presence of evoked responses

Cut-off point of good functional recovery

Results of EPs

Functional recovery

Sensitivity

(%)

Specificity

(%)

Positive predictive value (%)

Negative predictive value (%)

Good, n

Poor, n

mRS 2

MEP – Upper (+)

(–)

21

32

13

39

39.6

75.0

61.8

54.9

MEP – Lower (+)

(–)

30

23

16

36

56.6

69.2

65.2

61.0

SEP – Upper (+)

(–)

39

14

19

30

73.6

61.2

67.2

68.2

SEP – Lower (+)

(–)

37

15

14

34

71.2

70.8

72.5

69.4

mRS 3

MEP – Upper (+)

(–)

29

47

5

24

38.1

82.8

85.3

33.8

MEP – Lower (+)

(–)

37

39

9

20

48.7

69.0

80.4

33.9

SEP – Upper (+)

(–)

50

25

8

19

66.7

70.4

86.2

43.2

SEP – Lower (+)

(–)

46

27

5

22

63.0

81.5

90.2

44.9

EP: evoked potentials; mRS: modified Rankin Scale; MEP: motor evoked potentials; SEP: somatosensory evoked potentials.

DISCUSSION

The main finding of this study is that patients with a negative lower limb MEP or SEP response, and patients with negative MEP and SEP responses in the lower limbs during the sub-acute stage are less likely to achieve a good functional recovery during the chronic stage than those with positive MEP and SEP responses in the lower limbs.

To the best of our knowledge, this is the first study to use the cross-matched analysis of MEP and SEP to predict functional recovery after a stroke. Although MEP and SEP assess anatomically different pathways, both motor and sensory recoveries after stroke are important, because they are integral aspects of a single functional activity. Therefore, considering both MEP and SEP would better predict functional recovery after a stroke than considering MEP or SEP alone. However, no previous study has considered MEP and SEP simultaneously.

In the present study, several demographic and neurophysiological variables were found to be significantly associated with functional recovery by unadjusted analysis. However, only age < 50 years, NIHSS at onset of rehabilitation < 5, and a positive response in lower limbs for MEP and SEP retained significance after adjusted analysis. The adjusted ORs for prediction of good functional recovery in Table II provide information on the relative effects of NIHSS and age vs EPs, showing that the relative contribution of EPs in predicting good functional recovery 3 months after a stroke is less than age or NIHSS. The adjusted ORs were age (0.039; ≥ 80 years vs < 50 years), NIHSS (0.058; ≥ 5 vs < 5), and EPs (0.099; lower MEP(–)SEP(–) vs lower MEP(+)SEP(+)).

Many studies have investigated the predictive value of MEP in terms of post-stroke functional recovery. Heald et al. (10) reported that the presence of MEP within 72 h of stroke onset can discriminate patients with high probabilities of survival and good functional recovery and, conversely, that its absence indicated poor recovery and an increased risk of death.

The predictive value of SEP for post-stroke functional recovery has also been well studied. Tzvetanov et al. (3) found that upper limb SEP amplitude recorded within 3 days of stroke onset provided significant prognostic information.

In the present study, only positive MEP and SEP responses in the lower limbs were found to be significantly related to a good functional recovery. Our explanation is that the dichotomous outcome measure (good or poor functional recovery) based on the mRS is probably too crude to reflect the effects of upper limb MEP and SEP findings. On the other hand, it is also possible that had we applied different mRS cut-off points for good and poor functional recovery our findings might have been different.

When the predictive value of EP study was calculated in terms of sensitivities and specificities, the specificity of MEP was found to be better than its sensitivity. However, this higher specificity was not significant in our SEP study. If the cut-off point for good functional recovery had been adjusted from mRS 2 to mRS 3, the positive predictive value of EP study would have increased to > 80%, but the negative predictive value would have decreased to < 50% (Table III). In this case, stroke patients with a positive response by EP study could be predicted to achieve a good functional recovery, but post-stroke prognosis could not be predicted in patients with a negative response. As a result of low negative predictive value of EP studies, the prediction of post-stroke functional recovery based on neurophysiological findings alone has its limitations. Therefore, we advise that other findings, both clinical and radiological, should also be considered, and that MEP and SEP should be considered complementary tests in terms of predicting functional recovery after stroke.

In conclusion, MEP and SEP studies performed during the sub-acute stage of stroke could be useful for predicting functional status during the chronic stage.

REFERENCES

Comments

Do you want to comment on this paper? The comments will show up here and if appropriate the comments will also separately be forwarded to the authors. You need to login/create an account to comment on articles. Click here to login/create an account.
Advertisement