**2. Methods**

#### **2.1 Subjects**

A total number of fourteen patients were included in the study. Patients (characteristics: see Table 1), all with first-ever stroke and a hemiparesis of the lower extremity, entered the study on average 3.5 months after stroke and 2 months after admission in the rehabilitation centre (t=0). Data on muscle function in relation to functional performance of these patients at t=0 are reported elsewhere (Horstman et al., 2008). They were invited for measurements again 3 (n=8), 6 (n=5) and 12 (n=3) months after the first measurement (follow-up (F)1, F2 and F3 respectively). Because of drop-out of a number of subjects, data will be largely descriptive. Before participation, each subject was thoroughly informed about the procedures, completed a health questionnaire and signed an informed consent.

The exclusion criteria were medical complications (such as unstable cardiovascular problems), severe cognitive and/or communicative problems preventing understanding verbal instructions or limiting performance of the requested tasks (e.g. aphasia, hemineglect) and contra-indications for electrical stimulation (unstable epilepsy, cancer, skin abnormalities and pacemaker). The project carried the approval of the institutional review board (Medical Ethical Committee) of the VU University Medical Centre, Amsterdam, The Netherlands.


Table 1. Subject characteristics.

### **2.2 Experimental set-up**

68 Rehabilitation Medicine

recent study showed a significant association between paretic lower limb strength and balance both *cross-sectionally* in *acute* patients with stroke as well as *longitudinally* in *post-*

Besides a reduction in maximal muscle strength, the ability to generate torque as fast as possible, is also impaired after stroke (Horstman et al., 2010; Bohannon & Walsh, 1992; Gerrits et al., 2009). Rate of torque development is an important determinant of e.g. risk of falling and (again) for controlling balance (Shigematsu et al., 2006; Pijnappels et al., 2008). Recent work from our group has shown lower maximal rates of torque development during electrically stimulated (Horstman et al., 2010; Gerrits et al., 2009) as well as during voluntary (Horstman et al., 2010) contractions of the paretic and non-paretic knee-extensors. Decreased ability to rapidly develop knee extension torque contributes more to lower walking speed

In summary, there is clear evidence that difficulties in executing daily tasks in patients with stroke are related to both impaired strength and speed of paretic and/or non-paretic muscles. Nevertheless, most studies are performed at one point in time (Kim & Eng, 2003; Mercier & Bourbonnais, 2004; Ada et al., 2006; Bohannon, 2007b; Patterson et al., 2007; Horstman et al., 2008). It is not fully elucidated whether the improvements in functional performance at the activity level of patients with stroke during rehabilitation relate to changes in specific contractile function of the thigh muscles. Therefore, the present study reports on *longitudinal* changes in functional performance in a group of patients with stroke during the first year after stroke. Furthermore, it is determined whether these changes relate to alterations in strength and speed characteristics of the paretic and non-paretic thigh

A total number of fourteen patients were included in the study. Patients (characteristics: see Table 1), all with first-ever stroke and a hemiparesis of the lower extremity, entered the study on average 3.5 months after stroke and 2 months after admission in the rehabilitation centre (t=0). Data on muscle function in relation to functional performance of these patients at t=0 are reported elsewhere (Horstman et al., 2008). They were invited for measurements again 3 (n=8), 6 (n=5) and 12 (n=3) months after the first measurement (follow-up (F)1, F2 and F3 respectively). Because of drop-out of a number of subjects, data will be largely descriptive. Before participation, each subject was thoroughly informed about the

The exclusion criteria were medical complications (such as unstable cardiovascular problems), severe cognitive and/or communicative problems preventing understanding verbal instructions or limiting performance of the requested tasks (e.g. aphasia, hemineglect) and contra-indications for electrical stimulation (unstable epilepsy, cancer, skin abnormalities and pacemaker). The project carried the approval of the institutional review board (Medical Ethical Committee) of the VU University Medical Centre, Amsterdam, The

procedures, completed a health questionnaire and signed an informed consent.

*acute* patients (van Nes et al., 2009).

**2. Methods 2.1 Subjects** 

Netherlands.

after stroke than does maximal strength (Pohl et al., 2002).

muscles and voluntary activation capacity of patients with stroke.

Body function and activity-participation level were assessed with different clinical tests ('functional performance' tests). In addition, muscle function characteristics of the kneeextensors and -flexors were assessed in both limbs. The measurements were spread over four different days with at least one day of rest in between.

#### **2.3 Experimental procedures**

#### **2.3.1 Functional performance tests**

The following tests were performed by the subjects under supervision of a physiotherapist (except for the Rivermead Mobility Index, which was carried out by one of the researchers):


Functional Recovery and Muscle Properties After Stroke: A Preliminary Longitudinal Study 71

A modified super-imposed stimulation technique was used in which electrically evoked triplets (pulse train of three rectangular 200 s pulses applied at 300 Hz) were used to establish the subjects' capacity to voluntarily activate their muscles (Kooistra et al*.*, 2005). Measurements started with PL in a knee angle of 60 knee flexion. First, stimulation current was increased until supramaximal stimulation was ensured. Next, subjects underwent measurements consisting of a triplet superimposed on the plateau of the force signal of the

*MVC torque* (Nm) was determined as the peak force from the force plateau multiplied by the external moment arm. MVCe and MVCf were assessed. *Maximal rate of torque development* was defined as the steepest slope of torque development during fast voluntary contractions (MRTDvol) (de Ruiter et al., 2004) and during a pulse train of 80 ms, 300 Hz (MRTDstim). MRTDvol was normalized to MVCe torque in order to correct for the number of parallel muscle fibers ('muscle thickness') to get a fair comparison of contractile speed of muscles between different subjects independent of absolute maximal torques. MRTDstim was therefore expressed as a percentage of 150 Hz torque (obtained at the same stimulation

*Voluntary activation* is defined as the completeness of skeletal muscle activation during voluntary contractions and was calculated by means of a modified interpolated twitch technique (Kooistra et al., 2005). *Voluntary activation (%) = [1 – (superimposed triplet/rest triplet)] \* 100.* Here the superimposed triplet is the force increment during a maximal contraction at the time of stimulation and the control triplet is that evoked in the relaxed muscle (Shield & Zhou, 2004). Supramaximal *triplet torque* of the relaxed muscle is used as a measure for the maximal (intrinsic) torque capacity of the knee extensors, independent of

Spearman rank correlations were calculated between changes (Δ, between t=0 and F2) in scores at the tests of functional performance and changes in the 6 muscle variables (MVCe, MVCf, MRTDvol, MRTDstim, triplet and voluntary activation). For the differences over

Not all subjects participated in follow-up sessions. The most important reason given was that travelling to the testing site was too time-consuming. Some experienced the measurements (especially the electrical stimulation and the duration of the experiments) as too uncomfortable. Other patients missed the follow-up measurements due to severe illness. One patient decided to spend the winter abroad. Moreover, data were not complete for some of the patients due to unreliability, e.g. concentration problems (one subject dozed off a few times during the measurements), no force plateau during the MVCs or subjects did not reach 90% of their MVC during the familiarization session. Therefore, the data will be

mainly descriptive and hardly any statistical analyses were performed.

**2.3.5 Triplet stimulation and voluntary activation** 

**2.4 Data analysis of muscle function** 

intensity as the 300 Hz pulse train).

time, shown in Table 2, a Friedman test was used.

voluntary activation.

**3. Results** 

MVC. Subsequently, these measurements were performed with NL.


## **2.3.2 Force measurements**

The procedures for the measurements as well as the calculation of variables are described in detail elsewhere (Horstman et al., 2008). Briefly, maximal voluntary and electrically evoked isometric torques of the knee extensors and maximal voluntary isometric torques of the knee flexors were measured while subjects were seated on a custom built Lower EXtremity System (LEXS) (Horstman et al., 2008). The lower leg was strapped tightly to a force transducer just above the ankle by means of a cuff at a knee flexion angle of 60 (0 = full extension). Electrical stimulation, used for the knee extensors only, was applied via two surface electrodes placed over the quadriceps muscles with a computer-controlled constant current stimulator (Digitimer DSH7, Digitimer Ltd., Welwyn Garden City, UK).

#### **2.3.3 Familiarization session**

During the familiarization session, measurements were performed with the non-paretic lower limb to check whether the instructions were understood by the subject. After a warming-up (existing of 5 submaximal contractions) subjects were trained to perform maximal isometric knee flexion (MVCf) and extension (MVCe) contractions and fast voluntary knee extensions. Subsequently, the subjects were familiarized with electrical stimulation. During the follow-up measurements, no familiarization session was performed.

#### **2.3.4 Muscle strength and speed**

Subjects were asked to maximally generate isometric knee extensions for 3-4 s to determine MVCe. Alternately, MVCfs were performed, as described in Horstman et al. (2008). Thereafter, subjects were asked to perform knee extensions as fast as possible (Horstman et al., 2008) with the command: 3, 2, 1 GO! They were encouraged to reach a peak force of at least 70% of their MVC and were not allowed to make a countermovement (flexion) or have pretension before the fast extension (de Ruiter et al., 2004). The same measurements as performed with the paretic lower limb (PL) were repeated with the non-paretic lower limb (NL), carried out on a separate day. Control subjects just performed one session, with the right leg only.

#### **2.3.5 Triplet stimulation and voluntary activation**

A modified super-imposed stimulation technique was used in which electrically evoked triplets (pulse train of three rectangular 200 s pulses applied at 300 Hz) were used to establish the subjects' capacity to voluntarily activate their muscles (Kooistra et al*.*, 2005). Measurements started with PL in a knee angle of 60 knee flexion. First, stimulation current was increased until supramaximal stimulation was ensured. Next, subjects underwent measurements consisting of a triplet superimposed on the plateau of the force signal of the MVC. Subsequently, these measurements were performed with NL.

#### **2.4 Data analysis of muscle function**

70 Rehabilitation Medicine

 *10 meter walk test (10m)* is performed at comfortable (self selected) walking speed by patients who are able to walk independent with or without mobility aid and/or orthesis. Time to walk 10m is measured and averaged over three trials (Smith & Baer, 1999). Then, the speed is calculated (10m divided by the average time to walk that

 *Motricity Index (MI)* evaluates the arbitrary movement activity and maximum isometric muscle force. Possible scores are 0-9-14-19-25-33 at each of the three parts of the test for lower extremities (Demeurisse et al., 1980; Collin & Wade, 1990; Cameron & Bohannon,

 *Functional Ambulation Categories-score* (*FAC*) evaluates the measure of independence of walking of the patient. Categories are scored on a six-point scale (0-5) (Holden *et al.*,

The procedures for the measurements as well as the calculation of variables are described in detail elsewhere (Horstman et al., 2008). Briefly, maximal voluntary and electrically evoked isometric torques of the knee extensors and maximal voluntary isometric torques of the knee flexors were measured while subjects were seated on a custom built Lower EXtremity System (LEXS) (Horstman et al., 2008). The lower leg was strapped tightly to a force transducer just above the ankle by means of a cuff at a knee flexion angle of 60 (0 = full extension). Electrical stimulation, used for the knee extensors only, was applied via two surface electrodes placed over the quadriceps muscles with a computer-controlled constant

During the familiarization session, measurements were performed with the non-paretic lower limb to check whether the instructions were understood by the subject. After a warming-up (existing of 5 submaximal contractions) subjects were trained to perform maximal isometric knee flexion (MVCf) and extension (MVCe) contractions and fast voluntary knee extensions. Subsequently, the subjects were familiarized with electrical stimulation. During the follow-up measurements, no familiarization session was

Subjects were asked to maximally generate isometric knee extensions for 3-4 s to determine MVCe. Alternately, MVCfs were performed, as described in Horstman et al. (2008). Thereafter, subjects were asked to perform knee extensions as fast as possible (Horstman et al., 2008) with the command: 3, 2, 1 GO! They were encouraged to reach a peak force of at least 70% of their MVC and were not allowed to make a countermovement (flexion) or have pretension before the fast extension (de Ruiter et al., 2004). The same measurements as performed with the paretic lower limb (PL) were repeated with the non-paretic lower limb (NL), carried out on a separate day. Control subjects just performed one session, with the

current stimulator (Digitimer DSH7, Digitimer Ltd., Welwyn Garden City, UK).

10m).

2000).

1984, 1986).

**2.3.2 Force measurements** 

**2.3.3 Familiarization session** 

**2.3.4 Muscle strength and speed** 

performed.

right leg only.

*MVC torque* (Nm) was determined as the peak force from the force plateau multiplied by the external moment arm. MVCe and MVCf were assessed. *Maximal rate of torque development* was defined as the steepest slope of torque development during fast voluntary contractions (MRTDvol) (de Ruiter et al., 2004) and during a pulse train of 80 ms, 300 Hz (MRTDstim). MRTDvol was normalized to MVCe torque in order to correct for the number of parallel muscle fibers ('muscle thickness') to get a fair comparison of contractile speed of muscles between different subjects independent of absolute maximal torques. MRTDstim was therefore expressed as a percentage of 150 Hz torque (obtained at the same stimulation intensity as the 300 Hz pulse train).

*Voluntary activation* is defined as the completeness of skeletal muscle activation during voluntary contractions and was calculated by means of a modified interpolated twitch technique (Kooistra et al., 2005). *Voluntary activation (%) = [1 – (superimposed triplet/rest triplet)] \* 100.* Here the superimposed triplet is the force increment during a maximal contraction at the time of stimulation and the control triplet is that evoked in the relaxed muscle (Shield & Zhou, 2004). Supramaximal *triplet torque* of the relaxed muscle is used as a measure for the maximal (intrinsic) torque capacity of the knee extensors, independent of voluntary activation.

Spearman rank correlations were calculated between changes (Δ, between t=0 and F2) in scores at the tests of functional performance and changes in the 6 muscle variables (MVCe, MVCf, MRTDvol, MRTDstim, triplet and voluntary activation). For the differences over time, shown in Table 2, a Friedman test was used.

## **3. Results**

Not all subjects participated in follow-up sessions. The most important reason given was that travelling to the testing site was too time-consuming. Some experienced the measurements (especially the electrical stimulation and the duration of the experiments) as too uncomfortable. Other patients missed the follow-up measurements due to severe illness. One patient decided to spend the winter abroad. Moreover, data were not complete for some of the patients due to unreliability, e.g. concentration problems (one subject dozed off a few times during the measurements), no force plateau during the MVCs or subjects did not reach 90% of their MVC during the familiarization session. Therefore, the data will be mainly descriptive and hardly any statistical analyses were performed.

Functional Recovery and Muscle Properties After Stroke: A Preliminary Longitudinal Study 73

Table 2. Median values and 1st and 3rd quartiles for the Berg Balance Scale (BBS), Motricity Index (MI), Functional Ambulation Categories-score (FAC), Rivermead Mobility Index (RMI), Brunnstrom Fugl-Meyer (FM), 10 meter walk test (10m) and Timed ''get-up-and-go'' test (TUG) at the 3 measurement points 3.5 months after stroke (t=0) and during follow-up (F)1 and F2 for the five subjects who participated at these three measurement points (n= 5).

Data were not complete for reasons explained earlier. Furthermore, it is a general experience that subjects, knowing that a superimposed stimulation will be performed, anticipate upon stimulation and perform less when compared with MVC without stimulation. To minimize this effect, which influences the activation results, only data were used when MVCs with superimposed stimulation were more than 90% of their highest attempt. Therefore, variables of muscle functions in Table 3 show missing values. The zero values are values from subjects who did perform the measurements but were not able to generate force with their

There was a substantial variation between subjects with respect to the outcome of the muscle variables at the start of the study (t=0) as well as with respect to the changes in these

Table 4 shows the potential relevant correlations (n=4 or 5) with |ρ|>0.7 between changes (Δ, i.e., MVCe at F2 minus MVCe at t=0) in scores at tests of functional performance and changes in muscle variables. Figure 2 shows the five significant correlations namely the correlations between ΔMVCf of PL and Δ10m (ρ=0.99, p=0.002, n=5), between ΔMRTDstim of PL and ΔFM (ρ=0.99, p=0.01, n=4), between Δtriplet of PL and ΔFM (ρ=0.90, p=0.04, n=5), between Δtriplet of PL and ΔRMI (ρ=0.92, p=0.03, n=5), and between ΔMVCf of NL and

A considerable improvement was found in de scores at the tests of functional performance, especially up to F2. In general also the muscle variables improved over time. Changes in muscle variables of the 5 subjects that were measured at t=0, F1 and F2 were shown to correlate with improvements in tests of functional performance during the first 9 months

**3.3 Correlations between changes in functional performance and muscle function** 

 **t=0 F1 F2**

\* significant improvement (p<0.05) compared with t=0

^ trend (p<0.1) compared with t=0

**3.2 Muscle variables** 

paretic lower limb.

variables over time (Table 3).

ΔBBS (ρ=0.92, p=0.03, n=5).

**4. Discussion** 

after stroke.

**BBS** 43 (36-48) 50 (46-51)\* 51 (49-52)\* **MI** 42 (39-47) 61 (42-64) 61 (53-61) **FAC** 4 (3-4) 5 (4-5)\* 5 (5-5)\* **RMI** 8 (7-8) 13 (12-13)\* 13 (12-13)\* **FM** 13 (13-15) 17(14-19)^ 17 (16-10)\* **10m** (m/s) 0.28 (0.23-0.39)) 0.36 (0.35-0.47)^ 0.35 (0.30-0.53)\* **TUG** (s) 43.5 (34.7-49.0) 34.0 (34.0-41.0) 30.1(20.8-38.0)^

#### **3.1 Functional performance**

Figure 1 shows the course of the scores of the subjects with stroke at two important (see below, Table 4) tests of functional performance, namely the BBS, a measure of abilityactivity level, and at the FM, an impairment (bodily functions) measure developed to assess physical recovery after stroke (Sanford et al*.*, 1993). Because the outcome of most tests of functional performance seemed to plateau at F2 and because F3 values could only be obtained in three subjects, mean values of the five subjects assessed until F2 are presented in Table 2. Data in this table are expressed as median and 1st and 3rd quartile. Overall, the data of the functional performance tests show improvement (except for MI) during the follow-up period.

Fig. 1. Course of all individual scores at the Berg Balance Scale (BBS) and Brunnstrom Fugl-Meyer, lower extremity (FM) at 4 measuring times (t=0, F1, F2, F3). Note that t=0 is on average 3.5 months after stroke.


\* significant improvement (p<0.05) compared with t=0

^ trend (p<0.1) compared with t=0

Table 2. Median values and 1st and 3rd quartiles for the Berg Balance Scale (BBS), Motricity Index (MI), Functional Ambulation Categories-score (FAC), Rivermead Mobility Index (RMI), Brunnstrom Fugl-Meyer (FM), 10 meter walk test (10m) and Timed ''get-up-and-go'' test (TUG) at the 3 measurement points 3.5 months after stroke (t=0) and during follow-up (F)1 and F2 for the five subjects who participated at these three measurement points (n= 5).

#### **3.2 Muscle variables**

72 Rehabilitation Medicine

Figure 1 shows the course of the scores of the subjects with stroke at two important (see below, Table 4) tests of functional performance, namely the BBS, a measure of abilityactivity level, and at the FM, an impairment (bodily functions) measure developed to assess physical recovery after stroke (Sanford et al*.*, 1993). Because the outcome of most tests of functional performance seemed to plateau at F2 and because F3 values could only be obtained in three subjects, mean values of the five subjects assessed until F2 are presented in Table 2. Data in this table are expressed as median and 1st and 3rd quartile. Overall, the data of the functional performance tests show improvement (except for MI) during the follow-up

Fig. 1. Course of all individual scores at the Berg Balance Scale (BBS) and Brunnstrom Fugl-Meyer, lower extremity (FM) at 4 measuring times (t=0, F1, F2, F3). Note that t=0 is on

**3.1 Functional performance** 

average 3.5 months after stroke.

period.

Data were not complete for reasons explained earlier. Furthermore, it is a general experience that subjects, knowing that a superimposed stimulation will be performed, anticipate upon stimulation and perform less when compared with MVC without stimulation. To minimize this effect, which influences the activation results, only data were used when MVCs with superimposed stimulation were more than 90% of their highest attempt. Therefore, variables of muscle functions in Table 3 show missing values. The zero values are values from subjects who did perform the measurements but were not able to generate force with their paretic lower limb.

There was a substantial variation between subjects with respect to the outcome of the muscle variables at the start of the study (t=0) as well as with respect to the changes in these variables over time (Table 3).

#### **3.3 Correlations between changes in functional performance and muscle function**

Table 4 shows the potential relevant correlations (n=4 or 5) with |ρ|>0.7 between changes (Δ, i.e., MVCe at F2 minus MVCe at t=0) in scores at tests of functional performance and changes in muscle variables. Figure 2 shows the five significant correlations namely the correlations between ΔMVCf of PL and Δ10m (ρ=0.99, p=0.002, n=5), between ΔMRTDstim of PL and ΔFM (ρ=0.99, p=0.01, n=4), between Δtriplet of PL and ΔFM (ρ=0.90, p=0.04, n=5), between Δtriplet of PL and ΔRMI (ρ=0.92, p=0.03, n=5), and between ΔMVCf of NL and ΔBBS (ρ=0.92, p=0.03, n=5).
