- Email: [email protected]
Prognostic value of median and tibial somatosensory evoked potentials in acute stroke
Neuroscience Letters 380 (2005) 99–104
Prognostic value of median and tibial somatosensory evoked potentials in acute stroke Plamen Tzvetanova,∗ , Rossen T. Rousseffb , Penka Atanassovac a
Stroke Unit, Department of Neurology, Medical University of Pleven, Georgi Kochev str. 8A, Pleven 5800, Bulgaria b Department of Clinical Neurophysiology, Medical University of Pleven, Bulgaria c Stroke unit, Medical University of Plovdiv, Bulgaria Received 30 September 2004; received in revised form 27 December 2004; accepted 9 January 2005
Abstract The predictive values of early somatosensory evoked potentials (SSEPs) for the functional outcome after stroke are investigated. Ninety-four stroke patients (mean age: 61.2, S.D.: 11.8) with CT confirmed diagnoses of middle cerebral artery (MCA) infarction in 71 and supratentorial intracerebral hemorrhage in 23. Median and tibial SSEPs were recorded within 3 days of onset. SSEP parameters were compared to motor (MRC) and functional ability (Barthel index) followed up at 1, 3, 6 and 12 months. Upper limb MRC remains the strongest single predictor of functional outcome, determining 54.3% of Barthel index value at 12 months. The highest predictive value among SSEP parameters has N20-P25 amplitude ratio—34.5%. Combined application of upper limb MRC and N20-P25 amplitude ratio provided significantly stronger prognostic information—66%. Combined assessment of SSEP parameters and muscle power in acute stroke considerably improves prediction of functional outcome. © 2005 Elsevier Ireland Ltd. All rights reserved. Keywords: Stroke prognosis; Somatosensory evoked potentials; Functional recovery
Motor function recovery after stroke is one of the most important functional outcomes. An early assessment of the expected recovery level of recovery can help in early decision-making regarding appropriate medical and rehabilitation treatment as well as in research trials for assessment of the real intervention effects [14,18]. The principal clinical predictor was determined to be the muscle power of the affected limbs [14]. However, no single factor can be related strongly enough to allow an accurate prediction as Wade et al. stated back in 1983 [17]. Combined stroke scales have also limitations and only partially predict functional outcome [17,20]. The insufficient predictive strength of the clinical parameters provoked research using neuroimaging and neurophysiological parameters. In the acute phase conventional CT and MRI cannot determine the exact infarction size, which is the main neuroimaging ∗
Corresponding author. Tel.: +359 64 835242; fax: +359 64 890111. E-mail address: [email protected] (P. Tzvetanov).
0304-3940/$ – see front matter © 2005 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.neulet.2005.01.023
predictive parameter. Advanced techniques like diffusionand perfusion-weighted MRI are not universally available; studies on their predictive value are sparse [19]. Motor evoked potentials in the early stage of stroke give controversial results. Their prognostic ability becomes significant only 1 month after onset [4]. The measurement of somatosensory evoked potentials (SSEPs) is an objective method of assessing the integrity of sensory and motor pathways and areas of the central nervous system [1]. The prevailing opinion is that they do contribute to prediction of functional recovery [3,4,7,8]. However, most SSEP studies were performed late, even months after onset of stroke [4,8] and thus they don’t contribute to early functional prognosis and decision-making. The correlation of SSEP changes with clinical parameters and their predictive significance is not clearly defined. The aim of this study is to evaluate the prognostic possibilities of SSEP at onset—alone and in combination with an established clinical predictor as motor strength.
100
P. Tzvetanov et al. / Neuroscience Letters 380 (2005) 99–104
Ninety-four supratentorial stroke patients admitted to the Stroke Unit of an university hospital (43 women and 51 men) were included. Mean patients’ age was 61.2, S.D.: 11.8, years (range 49–81). Informed consent was obtained from the patients or the relatives. The study was approved by the local Ethics Committee. Another eight patients with initially performed SSEPs died within the study period and were excluded from analysis. Excluded were: (1) persons in stupor, coma, acute confusional states or other disorders of consciousness that precluded active cooperation; (2) subjects with severe aphasia making muscle power examination impossible; (3) patients with previous stroke; (4) patients with seizures or history of seizures; and (5) patients with polyneuropathy. All patients presented with either hemiparesis (54) or hemiplegia (40), with hemihypoesthesia in 89 and partial aphasia in 37. Right hemisphere was affected in 43 cases and left in 51. These cardinal stroke features were combined with hemianopia in 19 patients, pronounced muscle tone changes in 25 (early severe spasticity in four patients, hypotonia of upper and lower limb—21). Deep hemihypoesthesia was present in 30 patients. Fifty-five patients were unable to sit independently either because of weakness (40 persons) or because of sitting imbalance (15 persons). Incontinence was observed in 20 patients.
All cases were verified by computed tomography using a Siemens-Somatom ARC. By imaging criteria [6,9] infarctions were divided by region of the MCA yielding: 33 superficial (superior division of the MCA: 28, inferior division of MCA: 5). Eleven patients had infarctions in the deep territory of MCA. In 29 cases infarctions were large. Lacunar lesions were observed in nine patients. Thalamic hemorrhage was found in seven patients, putamenal hemorrhage in six, small capsular in two, massive in two and lobar hemorrhage in four patients. Muscle strength for the upper and lower limb was calculated as the mean of the Medical Research Council (MRC) score for the following movements: shoulder abduction, elbow flexion and wrist extension for the upper limb and hip flexion, knee extension and ankle extension for the lower limb. It was assessed between days 1 and 3 and at 1, 3, 6 and 12 months. Activities of daily living (ADL) were scored using the Barthel index [11]: (0–4, very disabled; 5–9, severely disabled; 10–14, moderately disabled; 15–19, mildly disabled; and 20, physically independent) at 1, 3, 6 and 12 months after stroke. Using Toennies-Multiliner EMB/EP equipment, SSEPs were performed within 3 days of the onset of stroke. In 43 patients SSEPs were completed within hours after stroke onset.
Fig. 1. Severely abnormal median and delayed tibial SSEP over the right hemispere. A 67-year old man with infarction in the superficial superior division of the MCA 3 days after stroke.
P. Tzvetanov et al. / Neuroscience Letters 380 (2005) 99–104
Median nerve SSEPs were recorded after stimulation at the wrist using a frequency of 2.1 Hz, a pulse duration 50 ms and a strength of current enough to produce minimal twitches of the thenar muscles. Two traces of at least 200 averaged responses for each side were recorded. Responses were registered using scalp needle electrodes with impedance below 5 k with the active one over the contralateral C 3 or C 4 after the 10–20 system [1]. The reference was situated at the Fz. The filter band pass was 10–1000 Hz. Tibial nerve SSEPs were recorded after stimulation at the ankle with similar stimulus, recording from active electrode at Cz and reference at Fz [1]. Fig. 1 represents abnormal SSEP recording in one of our patients. The following parameters were measured for both sides: median SSEP, N20 latency, N20-P25 peak-to-peak amplitude, and N20-P25 side-to-side amplitude ratio (affected/sound side). For tibial SSEP, the P40 latency, the P40-N50 amplitude and the P40-N50 affected/sound side amplitude ratio were evaluated. Based on data from the literature [4,12] we accepted the amplitude ratio N20-P25 and P40-N50 as a principal electrophysiological parameter in classifying patient groups. We calculated the prognostic value of SSEP changes independently for the median and tibial SSEPs. Afterwards we assessed the intrasubject correlations for both parameters. According to amplitude ratio (affected/sound site) patients were divided into three cohorts for each SSEP modality [4,12,20]. The first group included patients with normal amplitude ratio (affected/sound site ≥0.5). The third group enlisted persons with absent median or tibial SSEP. The second group included all the other patients—with amplitude ratio <0.5 but >0. Results were compared to reference values in the laboratory obtained from 30 healthy persons, age, sex and height-matched to the study cohort. Statistical methods: The Statgraphics Plus Version 2.1 statistical system was used for data analysis. For categorical data the χ2 and the Kruskal–Wallis tests and one-way ANOVA for dispersion analysis were applied. For numerical data along with descriptive methods we used regression analysis (Pearson’s r correlation coefficient). Multiple regression analysis was applied to calculate the combined predictive value of MRC and SSEP parameters at baseline for the Barthel score at 12 months. Comparing the motor and functional outcome in the three patient groups, we discovered the following tendencies. As expected, patients with normal median SSEP (n = 35) exhibit significantly (p < 0.05) higher MRC score than patients with absent response (n = 30). This trend was present at 1 month (p < 0.05) and increased with time, with little dynamics after the third month (p < 0.01). This is true both for median SSEP and upper limb MRC and for tibial SSEP and lower limb MRC (Figs. 2 and 3). The second group—with elicitable, but abnormal median SSEP (n = 29)—did not differ from patients with normal SSEP (p > 0.05), but also had significantly higher MRC than the group with absent responses. The difference again increased with time (p < 0.01 at 12 months).
101
Fig. 2. Upper limb motor outcome in different median SSEP groups.
Fig. 3. Lower limb motor outcome in different tibial SSEP groups.
Fig. 4. ADL dynamics in different median SSEP groups.
Regarding functional recovery, similar but not same results were observed. At month 1, Barthel results for the three SSEP groups roughly repeat the tendency shown concerning MRC scores (Figs. 4 and 5). However, at 12 months a significant difference is observed between the three median SSEP groups—patients with normal SSEP recover better than patients with abnormal or absent response (p < 0.01); those with obtainable but abnormal SSEP in turn recover significantly better than patients with absent responses (p < 0.01). At the same time, functional recovery of tibial SSEP groups follows the pattern of motor recovery of the lower limb—a significant difference exists only between patients with absent SSEP (n = 19) and the rest (p < 0.05), either with normal (n = 50) or with abnormal but present SSEP (n = 25).
Fig. 5. ADL dynamics in different tibial SSEP groups.
102
P. Tzvetanov et al. / Neuroscience Letters 380 (2005) 99–104
Table 1 Correlation of clinical and SSEP parameters at onset and functional outcome Dependent parameter
MRC 3 month MRC 12 month Barthel 3 month Barthel 12 month ∗ ∗∗
Upper limb parameters at onset
Lower limb parameters at onset
MRC
Amplitude
Amplitude ratio
MRC
Amplitude
Amplitude ratio
0.47**
0.27*
0.26*
0.43**
0.23*
0.63** 0.45** 0.74**
0.33* 0.28* 0.41**
0.31* 0.45** 0.52**
0.58** 0.43** 0.55**
0.30* 0.28* 0.38**
0.20* 0.26* 0.27* 0.33*
p < 0.05. p < 0.01.
To quantify further the rehabilitation capacity we used regression analysis only for patients with obtainable SSEP. The principal variable parameters were MRC score at onset and SSEP responses ratio as defined above. With upper limb we found that the strongest correlations existed between the MRC score measured at baseline and MRC and the Barthel at 12 months. Slightly weaker, but also significant (p < 0.01) is the correlation between N20-P25 ratio at onset and the Barthel scores at 12 months (Table 1). With lower limb MRC and P40-N50 ratio, the correlations, although significant, were much weaker. These ratios reached strongest levels for the initial MRC versus MRC and the Barthel at 12 months and for P40 ratio versus Barthel index at 12 months (Table 1). As evident from the same table, absolute N20-P25 and P40-N50 amplitudes also showed a significant relationship with all clinical parameters. Moderately strong dependence was found only between absolute N20-P25 amplitude and upper limb MRC and Barthel values at 12 months. The combined influence of upper limb muscle power and N20-P25 ratio on Barthel index was 66.04%. The combined P40-N50 ratio and lower limb muscle power predicted 49.7% of Barthel score at 12 month. The absolute N20-P25 amplitude and muscle power for the upper limb in combination determined 60.8% of Barthel index values at 12 months. The combination of P40-N50 amplitude and MRC for the lower limb influenced 44% of the Barthel score for the same period (Table 2). Finally we assessed the correlations between absolute latency of N20 and P40, respectively, and the clinical indices. For both latencies, the correlations were very weak or even missing. To summarize the most important results we must emphasize that muscle power of the upper limb remains the strongest clinical predictor at onset of stroke. SSEP parameters and especially median SSEP amplitude ratio has
independent prognostic value regarding motor and functional outcome. Combined assessment of upper limb muscle power and median SSEP amplitude ratio provides significantly stronger prognostic information than either parameter alone. Variable results for the predictive value of SSEP in acute stroke have been reported in the literature [4,7,8,16]. They concentrate on the different prognostic meaning of the absolute amplitudes, amplitude ratio and latency. The 94 patients in our study constitute one of the largest reported groups. Unlike most series, they were examined in the very early stage of stroke. In this phase different pathophysiological processes aside from direct tissue damage may interfere with some SSEP parameters. Spreading depression, diaschysis, decreased levels of conscience and drug effects may decrease amplitudes and prolong latencies both on the affected and sound side [10]. On the other hand, cortical hyperexcitability may increase the amplitudes and influence SSEP waveforms [12]. We tried to avoid or minimize such effects by excluding patients with disorders of consciousness, seizures and other states of altered cortical excitability. The follow-up period of 12 months seems sufficient, as within 3 months the physiological recovery of the injured tissue is generally completed [17] and further functional improvement is due to reorganization and other positive factors—re-education, availability of medical care. In our series the influence of non-biological factors is limited and this maybe reflects some pitfalls in long-term care after stroke inherent to our health system that now undergoes reform. Thus we have the opportunity to follow the natural history of biological recovery with minimal impact of the social factors. Commenting further on study design we should emphasize that we avoided using the multitude of sophisticated scales reflecting motor aspects of recovery, as our principal objective was to assess the functional recovery and not specifically its motor components. This logically follows the use of SSEP as predictor, as they reflect not only the integrity of motor area
Table 2 Combined predictive value of MRC score, median and tibial SSEP results Dependent parameter Barthel index at 12 month a b
Variable parameters at onset
R2a
Variable parameters at onset
R2a
Upper limb MRC Upper limb MRC + N20 amplitude ratio Upper limb MRC + N20 amplitude
54.3b
Lower limb MRC Lower limb MRC + P40 amplitude ratio Lower limb MRC + P40 amplitude
30.8b 44 49.7
Contribution of the variable parameter to predictive value in percent. p for all values of R2 less than 0.01.
66.04 60.8
P. Tzvetanov et al. / Neuroscience Letters 380 (2005) 99–104
and pathways, but the viability and status of a much larger region where the generators of different SSEP components are situated [20]. Unlike most previous investigators we used as a principal SSEP variable the amplitude ratio. The reason to use the SSEP ratio was the inherent variability of the absolute values of N20-P25 and P40-N50 amplitude [5]. This variance is physiologically determined, but it is still more pronounced in the early stage of stroke [16]. The cause for such increased variability for both sides is sought in the processes of cortical hyperexcitability, depression, diaschysis, mass effect. In regards to this variance amplitude ratio is considered far more stable [5,20]. The difference between N20-P25 amplitude and N20-P25 amplitude ratio is evident in the relationship with Barthel scores. The correlation with absolute N20-P25 amplitude remains moderately strong throughout the study period, while the amplitude ratio exhibits a strong relationship. The positive moderately strong link between muscle power at 12 months and the N20-P25 amplitude was noted also by others [5,8,16]. As such our study results have led us to agree with the dominating opinion that absolute amplitudes reflect more or less directly the number of surviving neurons in the ischemic territory [8]. However, amplitudes do not reach the stronger predictive value of the amplitude ratio. Further evidence of the positive relationship between amplitude ratios and functional recovery of the ischemic territory is supported by the analysis of patient groups when divided according to SSEP changes. The differences in recovery of the three groups at final assessment (12 months) for median responses were most pronounced between absent and normal response groups and to a lesser degree between absent and abnormal or between abnormal and normal. Such relationships of clinical assessment scales and electrodiagnostic parameters are expected and in concert with findings of other authors [4,5,13]. For instance Hendricks et al. [7] reported the significant association between the presence of median evoked potentials early after stroke and observed occurrence of motor recovery. These results suggest strongly that median SSEP predict the occurrence of motor recovery of upper extremity paralysis in patients suffering from first-ever infarction in the territory of the middle cerebral artery. Chu and Wu [2] in the absence of SSEP predicted poor outcome. Feys et al. [4] reported that in the acute phase the combination of motor performance and median SSEPs best-predicted upper limb outcome at 6 and 12 months after stroke. Vredeveld [16] reports similar findings in earlier studies. Unlike median SSEP, tibial responses did not reveal a significant difference between normal and abnormal groups, but only between obtainable and absent SSEP groups. Tibial SSEP absolute amplitude and ratio compared with median ones exhibited weaker relationships with clinical parameters. Also, despite the lower internal variance of the tibial SSEP amplitude ratio, its correlations with clinical indices were weaker than those of the absolute amplitudes. This is
103
in contrast with the findings of Kovala [10] who reported even greater prognostic value of tibial than median SSEP. We explain this difference by the fact that in the cited papers absence of P40 was among the most frequent abnormalities, while in our study it was found in about 25% of cases. The only P40-N50 amplitude ratio correlation exceeding the moderate level of statistical significance was that with the Barthel index at 12 months. This is explained in part by the fact that amplitude ratio is a more complex parameter that is influenced not only by the status of the affected side, but also by that of the sound one. Adequate performance of everyday activities requires bilateral cortical control [8,15,17] and in this we see the reason for a stronger relation with the bilaterally determined amplitude ratio. Latency measurements are disappointing for both median and tibial SSEP. Most authors have reached the same conclusion [8], although there are sparse contradicting data [4,8]. This is expected, having in mind the mechanisms of slowed axonal conduction [16]. We can hardly expect that a vascular lesion would selectively damage the myelin sheets and produce significant latency delay. Indeed Kovala [10] have shown that latency abnormalities are related with lesions of the subcortical white matter of the rolandic region. The MRC at onset in our patients showed a strong correlation with the clinical outcome as already established in previous studies [4,5,14]. We did not find the stronger values cited by others [14]. For this reason we searched in the early stage of our investigation, when one may observe severe motor deficit that resolves rapidly. Other authors usually examined MRC at 14–35 days [4,8,20] however, we examined MRC at 1–3 days after onset. Patient selection may have also influenced results as our study is performed in a stroke unit on minimally selected patients, thus including persons with severe stroke, while many other reports come from rehabilitation units where perspective patients with greater recovery potential are involved [7,8]. With the strong predictive value of muscle power the logical question that follows is whether it is worthwhile to resort to neurophysiological studies requiring special equipment, trained personnel and time. As seen from our results, the combined application of MRC and SSEP parameters had stronger predictive value than MRC alone. This finding justifies the use of additional neurophysiological studies and supports some previous reports [4,8]. Here we agree with Feys et al. [4] when they found that the added prognostic value obtained by SSEPs would clearly be advantageous to measuring MRC alone. However, by contrast, they found a somewhat lower additional predictive value of SSEP (8% compared with 15% in the present study). This discrepancy may be explained with differences in patient selection. For instance plegia of the arm was observed in 40 of our series versus 10 in the series of Feys et al. [4], and the time of SSEP investigation is significantly earlier in our cohort (3 days versus 35 days). Commenting on MRC and SSEP practical importance, one must not forget that in patients with disorders of consciousness as well as in aphasia where MRC cannot be
104
P. Tzvetanov et al. / Neuroscience Letters 380 (2005) 99–104
appropriately studied SSEP can be successfully recorded without patient cooperation. This is an additional argument in support of SSEP use. Besides being of more complex origin than motor output SSEP contribute to prediction of the more complex functional recovery—as compared to purely motor restoration [1,16]. In conclusion our results indicate that in the acute phase of stroke of different type and location both SSEP modalities have significant predictive value regarding short- and longterm functional recovery. Median SSEP parameters have much higher prognostic value than tibial ones. Muscle power at onset is moderately strong predictor of functional recovery. Investigating SSEP in combination with MRC score is reasonable, as the united prognostic value of both is significantly higher than of each one separately.
References [1] K.H. Chiappa (Ed.), Short-Latency Somatosensory Evoked Potentials: Methodology: Evoked Potentials in Clinical Medicine, second ed., Lippincott-Raven, Philadelphia, 1997, pp. 283–339. [2] N. Chu, T. Wu, Motor response patterns and prognostic value of transcranial magnetic stimulation in stroke patients, in: M.A. Lissens (Ed.), Clinical Applications of Magnetic Transcranial Stimulation, Peeters Press, Leuven, 1992, pp. 127–145. [3] S. Eksantivongs, N. Poungvarin, A. Viriyavejakul, The correlation of the short latency somatosensory evoked potentials and the stroke disability, J. Med. Assoc. Thai. 74 (1991) 1–7. [4] H. Feys, J. Van Hees, F. Bruyninckx, R. Mercelis, W. De Weerdt, Value of somatosensory and motor evoked potentials in predicting arm recovery after a stroke, J. Neurol. Neurosurg. Psychiatry 68 (2000) 323–331. [5] B. Fierro, V. La Bua, M. Oliveri, O. Daniele, F. Brighina, Prognostic value of somatosensory evoked potentials in stroke, Electromyogr. Clin. Neurophysiol. 39 (1999) 155–160. [6] T. Heinsius, J. Bogousslavsky, G. Van Melle, Large infarcts in the middle cerebral artery territory. Etiology and outcome patterns, Neurology 50 (1998) 341–350.
[7] H.T. Hendricks, G. Hageman, J. Van Limbeek, Prediction of recovery of upper extremity paralysis after stroke by measuring evoked potentials, Scand. J. Rehabil. Med. 29 (1997) 155–159. [8] O. Keren, H. Ring, P. Solzi, Upper limb somatosensory evoked potentials as a predictor of rehabilitation progress in dominant hemisphere stroke patients, Stroke 24 (1993) 1789–1793. [9] W. Kinkel, Computerized tomography in clinical neurology, in: A.B. Baker, L.B. Baker (Eds.), Clinical Neurology, J.B. Lippincott, Philadelphia, 1988, pp. 37–58. [10] T. Kovala, Prognostic significance of somatosensory potentials evoked by stimulation of the median and posterior tibial nerves: a prospective 1-year follow up study in patients with supratentorial cerebral infarction, Eur. Neurol. 31 (1991) 141–148. [11] F.I. Mahoney, D.W. Barthel, Functional evaluation: Barthel index, MD State Med. J. 14 (1965) 61–65. [12] P.H. Manninen, T.K. Tan, R.M. Sarjeant, Somatosensory evoked potential monitoring during carotid endarterectomy in patients with a stroke, Anesth. Analg. 93 (2001) 39–44. [13] Y. P´er´eon, P. Aubertin, P. Guih´eneuc, Prognostic significance of electrophysiological investigations in stroke patients: somatosensory and motor evoked potentials and sympathetic skin response, Neurophysiol. Clin. 25 (1995) 146–157. [14] R.J. Prescott, W.M. Garraway, A.J. Akhtar, Predicting functional outcome following functional outcome following acute stroke using a standard clinical examination, Stroke 13 (1982) 641–647. [15] P.M. Rossini, F. Tecchio, V. Pizzella, D. Lupoi, E. Cassetta, On the reorganization of the sensory hand areas after mono-hemispheric lesion: a functional (MEG)/anatomical (MRI) integrative study, Brain 78 (1998) 153–166. [16] J.W. Vredeveld, Predictive somatosensory evoked potentials, Electroencephalogr. Clin. Neurophysiol. 52 (1981) 40S. [17] D.T. Wade, E.S. Clive, R.L. Hewer, Predicting Barthel ADL score at 6 month after an acute stroke, Arch. Phys. Med. Rehabil. 64 (1983) 24–28. [18] D.T. Wade, H.R. Langton, V.A. Wood, The hemiplegic arm after stroke: measurement and recovery, J. Neurol. Neurosurg. Psychiatry 46 (1983) 521–524. [19] S. Warach, New imaging strategies for patient selection for thrombolytic and neuroprotective therapies, Neurology 57 (2001) 48– 52. [20] B.D. Zeman, C. Yiannikas, Functional prognosis in stroke: use of somatosensory evoked potentials, J. Neurol. Neurosurg. Psychiatry 52 (1989) 242–247.
Prognostic value of median and tibial somatosensory evoked potentials in acute stroke Plamen Tzvetanova,∗ , Rossen T. Rousseffb , Penka Atanassovac a
Stroke Unit, Department of Neurology, Medical University of Pleven, Georgi Kochev str. 8A, Pleven 5800, Bulgaria b Department of Clinical Neurophysiology, Medical University of Pleven, Bulgaria c Stroke unit, Medical University of Plovdiv, Bulgaria Received 30 September 2004; received in revised form 27 December 2004; accepted 9 January 2005
Abstract The predictive values of early somatosensory evoked potentials (SSEPs) for the functional outcome after stroke are investigated. Ninety-four stroke patients (mean age: 61.2, S.D.: 11.8) with CT confirmed diagnoses of middle cerebral artery (MCA) infarction in 71 and supratentorial intracerebral hemorrhage in 23. Median and tibial SSEPs were recorded within 3 days of onset. SSEP parameters were compared to motor (MRC) and functional ability (Barthel index) followed up at 1, 3, 6 and 12 months. Upper limb MRC remains the strongest single predictor of functional outcome, determining 54.3% of Barthel index value at 12 months. The highest predictive value among SSEP parameters has N20-P25 amplitude ratio—34.5%. Combined application of upper limb MRC and N20-P25 amplitude ratio provided significantly stronger prognostic information—66%. Combined assessment of SSEP parameters and muscle power in acute stroke considerably improves prediction of functional outcome. © 2005 Elsevier Ireland Ltd. All rights reserved. Keywords: Stroke prognosis; Somatosensory evoked potentials; Functional recovery
Motor function recovery after stroke is one of the most important functional outcomes. An early assessment of the expected recovery level of recovery can help in early decision-making regarding appropriate medical and rehabilitation treatment as well as in research trials for assessment of the real intervention effects [14,18]. The principal clinical predictor was determined to be the muscle power of the affected limbs [14]. However, no single factor can be related strongly enough to allow an accurate prediction as Wade et al. stated back in 1983 [17]. Combined stroke scales have also limitations and only partially predict functional outcome [17,20]. The insufficient predictive strength of the clinical parameters provoked research using neuroimaging and neurophysiological parameters. In the acute phase conventional CT and MRI cannot determine the exact infarction size, which is the main neuroimaging ∗
Corresponding author. Tel.: +359 64 835242; fax: +359 64 890111. E-mail address: [email protected] (P. Tzvetanov).
0304-3940/$ – see front matter © 2005 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.neulet.2005.01.023
predictive parameter. Advanced techniques like diffusionand perfusion-weighted MRI are not universally available; studies on their predictive value are sparse [19]. Motor evoked potentials in the early stage of stroke give controversial results. Their prognostic ability becomes significant only 1 month after onset [4]. The measurement of somatosensory evoked potentials (SSEPs) is an objective method of assessing the integrity of sensory and motor pathways and areas of the central nervous system [1]. The prevailing opinion is that they do contribute to prediction of functional recovery [3,4,7,8]. However, most SSEP studies were performed late, even months after onset of stroke [4,8] and thus they don’t contribute to early functional prognosis and decision-making. The correlation of SSEP changes with clinical parameters and their predictive significance is not clearly defined. The aim of this study is to evaluate the prognostic possibilities of SSEP at onset—alone and in combination with an established clinical predictor as motor strength.
100
P. Tzvetanov et al. / Neuroscience Letters 380 (2005) 99–104
Ninety-four supratentorial stroke patients admitted to the Stroke Unit of an university hospital (43 women and 51 men) were included. Mean patients’ age was 61.2, S.D.: 11.8, years (range 49–81). Informed consent was obtained from the patients or the relatives. The study was approved by the local Ethics Committee. Another eight patients with initially performed SSEPs died within the study period and were excluded from analysis. Excluded were: (1) persons in stupor, coma, acute confusional states or other disorders of consciousness that precluded active cooperation; (2) subjects with severe aphasia making muscle power examination impossible; (3) patients with previous stroke; (4) patients with seizures or history of seizures; and (5) patients with polyneuropathy. All patients presented with either hemiparesis (54) or hemiplegia (40), with hemihypoesthesia in 89 and partial aphasia in 37. Right hemisphere was affected in 43 cases and left in 51. These cardinal stroke features were combined with hemianopia in 19 patients, pronounced muscle tone changes in 25 (early severe spasticity in four patients, hypotonia of upper and lower limb—21). Deep hemihypoesthesia was present in 30 patients. Fifty-five patients were unable to sit independently either because of weakness (40 persons) or because of sitting imbalance (15 persons). Incontinence was observed in 20 patients.
All cases were verified by computed tomography using a Siemens-Somatom ARC. By imaging criteria [6,9] infarctions were divided by region of the MCA yielding: 33 superficial (superior division of the MCA: 28, inferior division of MCA: 5). Eleven patients had infarctions in the deep territory of MCA. In 29 cases infarctions were large. Lacunar lesions were observed in nine patients. Thalamic hemorrhage was found in seven patients, putamenal hemorrhage in six, small capsular in two, massive in two and lobar hemorrhage in four patients. Muscle strength for the upper and lower limb was calculated as the mean of the Medical Research Council (MRC) score for the following movements: shoulder abduction, elbow flexion and wrist extension for the upper limb and hip flexion, knee extension and ankle extension for the lower limb. It was assessed between days 1 and 3 and at 1, 3, 6 and 12 months. Activities of daily living (ADL) were scored using the Barthel index [11]: (0–4, very disabled; 5–9, severely disabled; 10–14, moderately disabled; 15–19, mildly disabled; and 20, physically independent) at 1, 3, 6 and 12 months after stroke. Using Toennies-Multiliner EMB/EP equipment, SSEPs were performed within 3 days of the onset of stroke. In 43 patients SSEPs were completed within hours after stroke onset.
Fig. 1. Severely abnormal median and delayed tibial SSEP over the right hemispere. A 67-year old man with infarction in the superficial superior division of the MCA 3 days after stroke.
P. Tzvetanov et al. / Neuroscience Letters 380 (2005) 99–104
Median nerve SSEPs were recorded after stimulation at the wrist using a frequency of 2.1 Hz, a pulse duration 50 ms and a strength of current enough to produce minimal twitches of the thenar muscles. Two traces of at least 200 averaged responses for each side were recorded. Responses were registered using scalp needle electrodes with impedance below 5 k with the active one over the contralateral C 3 or C 4 after the 10–20 system [1]. The reference was situated at the Fz. The filter band pass was 10–1000 Hz. Tibial nerve SSEPs were recorded after stimulation at the ankle with similar stimulus, recording from active electrode at Cz and reference at Fz [1]. Fig. 1 represents abnormal SSEP recording in one of our patients. The following parameters were measured for both sides: median SSEP, N20 latency, N20-P25 peak-to-peak amplitude, and N20-P25 side-to-side amplitude ratio (affected/sound side). For tibial SSEP, the P40 latency, the P40-N50 amplitude and the P40-N50 affected/sound side amplitude ratio were evaluated. Based on data from the literature [4,12] we accepted the amplitude ratio N20-P25 and P40-N50 as a principal electrophysiological parameter in classifying patient groups. We calculated the prognostic value of SSEP changes independently for the median and tibial SSEPs. Afterwards we assessed the intrasubject correlations for both parameters. According to amplitude ratio (affected/sound site) patients were divided into three cohorts for each SSEP modality [4,12,20]. The first group included patients with normal amplitude ratio (affected/sound site ≥0.5). The third group enlisted persons with absent median or tibial SSEP. The second group included all the other patients—with amplitude ratio <0.5 but >0. Results were compared to reference values in the laboratory obtained from 30 healthy persons, age, sex and height-matched to the study cohort. Statistical methods: The Statgraphics Plus Version 2.1 statistical system was used for data analysis. For categorical data the χ2 and the Kruskal–Wallis tests and one-way ANOVA for dispersion analysis were applied. For numerical data along with descriptive methods we used regression analysis (Pearson’s r correlation coefficient). Multiple regression analysis was applied to calculate the combined predictive value of MRC and SSEP parameters at baseline for the Barthel score at 12 months. Comparing the motor and functional outcome in the three patient groups, we discovered the following tendencies. As expected, patients with normal median SSEP (n = 35) exhibit significantly (p < 0.05) higher MRC score than patients with absent response (n = 30). This trend was present at 1 month (p < 0.05) and increased with time, with little dynamics after the third month (p < 0.01). This is true both for median SSEP and upper limb MRC and for tibial SSEP and lower limb MRC (Figs. 2 and 3). The second group—with elicitable, but abnormal median SSEP (n = 29)—did not differ from patients with normal SSEP (p > 0.05), but also had significantly higher MRC than the group with absent responses. The difference again increased with time (p < 0.01 at 12 months).
101
Fig. 2. Upper limb motor outcome in different median SSEP groups.
Fig. 3. Lower limb motor outcome in different tibial SSEP groups.
Fig. 4. ADL dynamics in different median SSEP groups.
Regarding functional recovery, similar but not same results were observed. At month 1, Barthel results for the three SSEP groups roughly repeat the tendency shown concerning MRC scores (Figs. 4 and 5). However, at 12 months a significant difference is observed between the three median SSEP groups—patients with normal SSEP recover better than patients with abnormal or absent response (p < 0.01); those with obtainable but abnormal SSEP in turn recover significantly better than patients with absent responses (p < 0.01). At the same time, functional recovery of tibial SSEP groups follows the pattern of motor recovery of the lower limb—a significant difference exists only between patients with absent SSEP (n = 19) and the rest (p < 0.05), either with normal (n = 50) or with abnormal but present SSEP (n = 25).
Fig. 5. ADL dynamics in different tibial SSEP groups.
102
P. Tzvetanov et al. / Neuroscience Letters 380 (2005) 99–104
Table 1 Correlation of clinical and SSEP parameters at onset and functional outcome Dependent parameter
MRC 3 month MRC 12 month Barthel 3 month Barthel 12 month ∗ ∗∗
Upper limb parameters at onset
Lower limb parameters at onset
MRC
Amplitude
Amplitude ratio
MRC
Amplitude
Amplitude ratio
0.47**
0.27*
0.26*
0.43**
0.23*
0.63** 0.45** 0.74**
0.33* 0.28* 0.41**
0.31* 0.45** 0.52**
0.58** 0.43** 0.55**
0.30* 0.28* 0.38**
0.20* 0.26* 0.27* 0.33*
p < 0.05. p < 0.01.
To quantify further the rehabilitation capacity we used regression analysis only for patients with obtainable SSEP. The principal variable parameters were MRC score at onset and SSEP responses ratio as defined above. With upper limb we found that the strongest correlations existed between the MRC score measured at baseline and MRC and the Barthel at 12 months. Slightly weaker, but also significant (p < 0.01) is the correlation between N20-P25 ratio at onset and the Barthel scores at 12 months (Table 1). With lower limb MRC and P40-N50 ratio, the correlations, although significant, were much weaker. These ratios reached strongest levels for the initial MRC versus MRC and the Barthel at 12 months and for P40 ratio versus Barthel index at 12 months (Table 1). As evident from the same table, absolute N20-P25 and P40-N50 amplitudes also showed a significant relationship with all clinical parameters. Moderately strong dependence was found only between absolute N20-P25 amplitude and upper limb MRC and Barthel values at 12 months. The combined influence of upper limb muscle power and N20-P25 ratio on Barthel index was 66.04%. The combined P40-N50 ratio and lower limb muscle power predicted 49.7% of Barthel score at 12 month. The absolute N20-P25 amplitude and muscle power for the upper limb in combination determined 60.8% of Barthel index values at 12 months. The combination of P40-N50 amplitude and MRC for the lower limb influenced 44% of the Barthel score for the same period (Table 2). Finally we assessed the correlations between absolute latency of N20 and P40, respectively, and the clinical indices. For both latencies, the correlations were very weak or even missing. To summarize the most important results we must emphasize that muscle power of the upper limb remains the strongest clinical predictor at onset of stroke. SSEP parameters and especially median SSEP amplitude ratio has
independent prognostic value regarding motor and functional outcome. Combined assessment of upper limb muscle power and median SSEP amplitude ratio provides significantly stronger prognostic information than either parameter alone. Variable results for the predictive value of SSEP in acute stroke have been reported in the literature [4,7,8,16]. They concentrate on the different prognostic meaning of the absolute amplitudes, amplitude ratio and latency. The 94 patients in our study constitute one of the largest reported groups. Unlike most series, they were examined in the very early stage of stroke. In this phase different pathophysiological processes aside from direct tissue damage may interfere with some SSEP parameters. Spreading depression, diaschysis, decreased levels of conscience and drug effects may decrease amplitudes and prolong latencies both on the affected and sound side [10]. On the other hand, cortical hyperexcitability may increase the amplitudes and influence SSEP waveforms [12]. We tried to avoid or minimize such effects by excluding patients with disorders of consciousness, seizures and other states of altered cortical excitability. The follow-up period of 12 months seems sufficient, as within 3 months the physiological recovery of the injured tissue is generally completed [17] and further functional improvement is due to reorganization and other positive factors—re-education, availability of medical care. In our series the influence of non-biological factors is limited and this maybe reflects some pitfalls in long-term care after stroke inherent to our health system that now undergoes reform. Thus we have the opportunity to follow the natural history of biological recovery with minimal impact of the social factors. Commenting further on study design we should emphasize that we avoided using the multitude of sophisticated scales reflecting motor aspects of recovery, as our principal objective was to assess the functional recovery and not specifically its motor components. This logically follows the use of SSEP as predictor, as they reflect not only the integrity of motor area
Table 2 Combined predictive value of MRC score, median and tibial SSEP results Dependent parameter Barthel index at 12 month a b
Variable parameters at onset
R2a
Variable parameters at onset
R2a
Upper limb MRC Upper limb MRC + N20 amplitude ratio Upper limb MRC + N20 amplitude
54.3b
Lower limb MRC Lower limb MRC + P40 amplitude ratio Lower limb MRC + P40 amplitude
30.8b 44 49.7
Contribution of the variable parameter to predictive value in percent. p for all values of R2 less than 0.01.
66.04 60.8
P. Tzvetanov et al. / Neuroscience Letters 380 (2005) 99–104
and pathways, but the viability and status of a much larger region where the generators of different SSEP components are situated [20]. Unlike most previous investigators we used as a principal SSEP variable the amplitude ratio. The reason to use the SSEP ratio was the inherent variability of the absolute values of N20-P25 and P40-N50 amplitude [5]. This variance is physiologically determined, but it is still more pronounced in the early stage of stroke [16]. The cause for such increased variability for both sides is sought in the processes of cortical hyperexcitability, depression, diaschysis, mass effect. In regards to this variance amplitude ratio is considered far more stable [5,20]. The difference between N20-P25 amplitude and N20-P25 amplitude ratio is evident in the relationship with Barthel scores. The correlation with absolute N20-P25 amplitude remains moderately strong throughout the study period, while the amplitude ratio exhibits a strong relationship. The positive moderately strong link between muscle power at 12 months and the N20-P25 amplitude was noted also by others [5,8,16]. As such our study results have led us to agree with the dominating opinion that absolute amplitudes reflect more or less directly the number of surviving neurons in the ischemic territory [8]. However, amplitudes do not reach the stronger predictive value of the amplitude ratio. Further evidence of the positive relationship between amplitude ratios and functional recovery of the ischemic territory is supported by the analysis of patient groups when divided according to SSEP changes. The differences in recovery of the three groups at final assessment (12 months) for median responses were most pronounced between absent and normal response groups and to a lesser degree between absent and abnormal or between abnormal and normal. Such relationships of clinical assessment scales and electrodiagnostic parameters are expected and in concert with findings of other authors [4,5,13]. For instance Hendricks et al. [7] reported the significant association between the presence of median evoked potentials early after stroke and observed occurrence of motor recovery. These results suggest strongly that median SSEP predict the occurrence of motor recovery of upper extremity paralysis in patients suffering from first-ever infarction in the territory of the middle cerebral artery. Chu and Wu [2] in the absence of SSEP predicted poor outcome. Feys et al. [4] reported that in the acute phase the combination of motor performance and median SSEPs best-predicted upper limb outcome at 6 and 12 months after stroke. Vredeveld [16] reports similar findings in earlier studies. Unlike median SSEP, tibial responses did not reveal a significant difference between normal and abnormal groups, but only between obtainable and absent SSEP groups. Tibial SSEP absolute amplitude and ratio compared with median ones exhibited weaker relationships with clinical parameters. Also, despite the lower internal variance of the tibial SSEP amplitude ratio, its correlations with clinical indices were weaker than those of the absolute amplitudes. This is
103
in contrast with the findings of Kovala [10] who reported even greater prognostic value of tibial than median SSEP. We explain this difference by the fact that in the cited papers absence of P40 was among the most frequent abnormalities, while in our study it was found in about 25% of cases. The only P40-N50 amplitude ratio correlation exceeding the moderate level of statistical significance was that with the Barthel index at 12 months. This is explained in part by the fact that amplitude ratio is a more complex parameter that is influenced not only by the status of the affected side, but also by that of the sound one. Adequate performance of everyday activities requires bilateral cortical control [8,15,17] and in this we see the reason for a stronger relation with the bilaterally determined amplitude ratio. Latency measurements are disappointing for both median and tibial SSEP. Most authors have reached the same conclusion [8], although there are sparse contradicting data [4,8]. This is expected, having in mind the mechanisms of slowed axonal conduction [16]. We can hardly expect that a vascular lesion would selectively damage the myelin sheets and produce significant latency delay. Indeed Kovala [10] have shown that latency abnormalities are related with lesions of the subcortical white matter of the rolandic region. The MRC at onset in our patients showed a strong correlation with the clinical outcome as already established in previous studies [4,5,14]. We did not find the stronger values cited by others [14]. For this reason we searched in the early stage of our investigation, when one may observe severe motor deficit that resolves rapidly. Other authors usually examined MRC at 14–35 days [4,8,20] however, we examined MRC at 1–3 days after onset. Patient selection may have also influenced results as our study is performed in a stroke unit on minimally selected patients, thus including persons with severe stroke, while many other reports come from rehabilitation units where perspective patients with greater recovery potential are involved [7,8]. With the strong predictive value of muscle power the logical question that follows is whether it is worthwhile to resort to neurophysiological studies requiring special equipment, trained personnel and time. As seen from our results, the combined application of MRC and SSEP parameters had stronger predictive value than MRC alone. This finding justifies the use of additional neurophysiological studies and supports some previous reports [4,8]. Here we agree with Feys et al. [4] when they found that the added prognostic value obtained by SSEPs would clearly be advantageous to measuring MRC alone. However, by contrast, they found a somewhat lower additional predictive value of SSEP (8% compared with 15% in the present study). This discrepancy may be explained with differences in patient selection. For instance plegia of the arm was observed in 40 of our series versus 10 in the series of Feys et al. [4], and the time of SSEP investigation is significantly earlier in our cohort (3 days versus 35 days). Commenting on MRC and SSEP practical importance, one must not forget that in patients with disorders of consciousness as well as in aphasia where MRC cannot be
104
P. Tzvetanov et al. / Neuroscience Letters 380 (2005) 99–104
appropriately studied SSEP can be successfully recorded without patient cooperation. This is an additional argument in support of SSEP use. Besides being of more complex origin than motor output SSEP contribute to prediction of the more complex functional recovery—as compared to purely motor restoration [1,16]. In conclusion our results indicate that in the acute phase of stroke of different type and location both SSEP modalities have significant predictive value regarding short- and longterm functional recovery. Median SSEP parameters have much higher prognostic value than tibial ones. Muscle power at onset is moderately strong predictor of functional recovery. Investigating SSEP in combination with MRC score is reasonable, as the united prognostic value of both is significantly higher than of each one separately.
References [1] K.H. Chiappa (Ed.), Short-Latency Somatosensory Evoked Potentials: Methodology: Evoked Potentials in Clinical Medicine, second ed., Lippincott-Raven, Philadelphia, 1997, pp. 283–339. [2] N. Chu, T. Wu, Motor response patterns and prognostic value of transcranial magnetic stimulation in stroke patients, in: M.A. Lissens (Ed.), Clinical Applications of Magnetic Transcranial Stimulation, Peeters Press, Leuven, 1992, pp. 127–145. [3] S. Eksantivongs, N. Poungvarin, A. Viriyavejakul, The correlation of the short latency somatosensory evoked potentials and the stroke disability, J. Med. Assoc. Thai. 74 (1991) 1–7. [4] H. Feys, J. Van Hees, F. Bruyninckx, R. Mercelis, W. De Weerdt, Value of somatosensory and motor evoked potentials in predicting arm recovery after a stroke, J. Neurol. Neurosurg. Psychiatry 68 (2000) 323–331. [5] B. Fierro, V. La Bua, M. Oliveri, O. Daniele, F. Brighina, Prognostic value of somatosensory evoked potentials in stroke, Electromyogr. Clin. Neurophysiol. 39 (1999) 155–160. [6] T. Heinsius, J. Bogousslavsky, G. Van Melle, Large infarcts in the middle cerebral artery territory. Etiology and outcome patterns, Neurology 50 (1998) 341–350.
[7] H.T. Hendricks, G. Hageman, J. Van Limbeek, Prediction of recovery of upper extremity paralysis after stroke by measuring evoked potentials, Scand. J. Rehabil. Med. 29 (1997) 155–159. [8] O. Keren, H. Ring, P. Solzi, Upper limb somatosensory evoked potentials as a predictor of rehabilitation progress in dominant hemisphere stroke patients, Stroke 24 (1993) 1789–1793. [9] W. Kinkel, Computerized tomography in clinical neurology, in: A.B. Baker, L.B. Baker (Eds.), Clinical Neurology, J.B. Lippincott, Philadelphia, 1988, pp. 37–58. [10] T. Kovala, Prognostic significance of somatosensory potentials evoked by stimulation of the median and posterior tibial nerves: a prospective 1-year follow up study in patients with supratentorial cerebral infarction, Eur. Neurol. 31 (1991) 141–148. [11] F.I. Mahoney, D.W. Barthel, Functional evaluation: Barthel index, MD State Med. J. 14 (1965) 61–65. [12] P.H. Manninen, T.K. Tan, R.M. Sarjeant, Somatosensory evoked potential monitoring during carotid endarterectomy in patients with a stroke, Anesth. Analg. 93 (2001) 39–44. [13] Y. P´er´eon, P. Aubertin, P. Guih´eneuc, Prognostic significance of electrophysiological investigations in stroke patients: somatosensory and motor evoked potentials and sympathetic skin response, Neurophysiol. Clin. 25 (1995) 146–157. [14] R.J. Prescott, W.M. Garraway, A.J. Akhtar, Predicting functional outcome following functional outcome following acute stroke using a standard clinical examination, Stroke 13 (1982) 641–647. [15] P.M. Rossini, F. Tecchio, V. Pizzella, D. Lupoi, E. Cassetta, On the reorganization of the sensory hand areas after mono-hemispheric lesion: a functional (MEG)/anatomical (MRI) integrative study, Brain 78 (1998) 153–166. [16] J.W. Vredeveld, Predictive somatosensory evoked potentials, Electroencephalogr. Clin. Neurophysiol. 52 (1981) 40S. [17] D.T. Wade, E.S. Clive, R.L. Hewer, Predicting Barthel ADL score at 6 month after an acute stroke, Arch. Phys. Med. Rehabil. 64 (1983) 24–28. [18] D.T. Wade, H.R. Langton, V.A. Wood, The hemiplegic arm after stroke: measurement and recovery, J. Neurol. Neurosurg. Psychiatry 46 (1983) 521–524. [19] S. Warach, New imaging strategies for patient selection for thrombolytic and neuroprotective therapies, Neurology 57 (2001) 48– 52. [20] B.D. Zeman, C. Yiannikas, Functional prognosis in stroke: use of somatosensory evoked potentials, J. Neurol. Neurosurg. Psychiatry 52 (1989) 242–247.