Preferential deficit of fat-free soft tissue in the appendicular region of children with cerebral palsy and proposed statistical models to capture the deficit

Abstract

Background:

Cerebral palsy (CP) is a neurological disorder characterized by a profound skeletal muscle deficit. However, whether there is a regional-specific skeletal muscle deficit in children with CP is unknown. The purpose of this study was to determine whether fat-free soft tissue mass (FFST), a commonly used surrogate for skeletal muscle mass, is more compromised in the limbs than in the trunk in children with CP. A second purpose was to determine whether physical characteristics can be used to accurately estimate appendicular FFST (AFFST) in children with CP.

Methods:

Forty-two children with CP (4 – 13 y) and 42 typically developing children matched to children with CP for sex, age and race were studied. Whole body FFST (FFST_whole), FFST in the upper limbs (FFST_upper), FFST in the lower limbs (FFST_lower), the ratio of AFFST to height (AFFST/ht), the ratio of AFFST to height² (AFFST/ht²) and non-appendicular FFST were estimated from dual-energy X-ray absorptiometry. Statistical models were developed to estimate AFFST, AFFST/ht and AFFST/ht² in both groups of children, and the leave-one-out method was used to validate the models.

Results:

Children with CP had 21 % lower FFST_whole, 30 % lower AFFST, 34 % lower FFST_lower, 14 % lower non-appendicular FFST, 23 % lower AFFST/ht, 19 % lower AFFST/ht² and 9 % lower AFFST/FFST_whole (all p < 0.05). Statistical models developed using data from typically developing children overestimated AFFST, AFFST/ht and AFFST/ht² by 35 %, 30 % and 21 % (all p < 0.05), respectively, in children with CP. Separate models developed using data from children with CP yielded better accuracy, with the estimated results highly correlated (r² = 0.78, 0.66 and 0.50, respectively; all p < 0.001) and not different from calculated AFFST, AFFST/ht and AFFST/ht² (all p > 0.99). However, when the difference in estimated values and measured values of AFFST, AFFST/ht and AFFST/ht² were plotted against measured values, there was an inverse relationship (r = −0.38, −0.47 and −0.61, respectively, all p < 0.05).

Conclusion:

Children with CP have a remarkable deficit in FFST that is more pronounced in appendicular than the non-appendicular regions and more pronounced in the lower than the upper limbs. Preliminary models developed using data from children with CP can provide reasonable estimates of AFFST and indexes of AFFST relative to height, but further development of the models may be needed.

Introduction

Cerebral palsy (CP) is a neurological disorder of movement and posture that onsets before, during or shortly after birth and lasts across the lifespan. It currently affects nearly 1 million people in the United States [1]. It is well established that children with CP have skeletal muscle deficits compared to their typically developing peers. The scope of such compromise includes, but is not limited to, decreased muscle size [2-5], decreased force generation capacity [6-8], and increased inter- and intramuscular fat infiltration [2, 3]. However, the evaluation of skeletal muscle deficits in children with CP has primarily focused on the lower limbs and studies that have evaluated the muscle development in all limbs in children with CP are lacking.

Appendicular fat-free soft tissue (AFFST), as measured by dual-energy X-ray absorptiometry (DXA), has been used as an indicator of sarcopenia in the aging population [9], and cachexia in clinical populations [10]. In healthy adults, the upper and lower limbs contain the largest portion of total body skeletal muscle mass [11]. Appendicular FFST is calculated by summing the FFST in the upper limbs (FFST_upper) and the lower limbs (FFST_lower), to provide a summative index that serves as a surrogate for skeletal muscle mass in the limbs. Because muscles at the appendicular sites are frequently involved in daily movement and physical activity, deficits in AFFST may severely impair one’s mobility and affect quality of life.

To our knowledge, the degree and distribution of the AFFST deficit among children with CP have not been elucidated. It has been reported that the lean tissue compromise in individuals with spinal cord injury is more pronounced in the arms and legs than in the trunk when compared to their sex-, age-, height-, and weight-matched controls [12, 13]. Spinal cord injury is a condition that results in a low level of mobility due to an injury to the central nervous system. Because of the similar etiology, spinal cord injury is sometimes used to infer the longterm consequences of having CP [14, 15]. However, whether there is a regional preference for the FFST deficit in children with CP (i.e., whether the FFST deficit is more profound at the appendicular regions than at the trunk in children with CP) is unknown.

The purpose of this study was to determine whether FFST is more compromised at the appendicular sites than at the trunk in children with CP, and whether we can use simple physical characteristics data to accurately estimate AFFST and its derived indexes that account for the lower heights of children with CP. It was hypothesized that 1) AFFST and non-appendicular (trunk) FFST would be lower in children with CP compared to their typically developing peers, but the extent of the deficit would be more pronounced at the appendicular regions, and 2) AFFST and AFFST indexes in children with CP would be overestimated when derived from models developed using data from typically developing children, but would be more accurately estimated using models based on data from children with CP.

Methods

Participants

Forty-two children with spastic CP, between the ages of 4-13 were recruited from CP programs at local hospitals to participate in this study. Forty two typically developing children who had no known neurological disorders, had height, body mass and BMI between the 5th and 95th percentile for their respective age and matched to children with CP for age (± 1.5 y), sex and race were also invited to participate via flyers and word of mouth. This study was approved by the Institutional Review Board. Prior to any testing, consent and assent were obtained from the legal guardian and the child participant (when able), respectively.

Anthropometrics

Height and body mass for all children were assessed with minimal clothing on and without shoes or braces. Height was assessed using a stadiometer (Seca 217; Seca GmbH & Co. KG., Hamburg, Germany) in typically developing children and ambulatory children with CP to the nearest 0.1 cm. For nonambulatory children with CP, height was estimated using forearm length and methods described by Miller et al [16]. Forearm length was measured by a research assistant using a specifically designed ruler with the participant’s elbow flexed at 95 to 110 degrees and the forearm pronated. The measurement was from the elbow to the tip of the middle finger. The reliability of height estimated from forearm length was assessed in 20 children with CP (5 - 14 years) by measuring them twice on the same day after repositioning. The reliability was excellent, as indicated by an intraclass correlation > 0.99. Body mass for all children was assessed using a digital weight scale to the nearest 0.2 kg (Detecto, 6550, Cardinal Scale, Webb City, MO). Body mass index was calculated based on the acquired height and body mass values. Height, body mass and BMI percentiles were determined based on the growth charts published by the Centers for Disease Control and Prevention [17].

Gross motor function

Gross motor function in children with CP was assessed by a physician or physician assistant according to the Gross Motor Function Classification System (GMFCS). The GMFCS is a 5 point system with a larger number indicating a higher level of functional deficiency [18]. Specifically, level I indicates the ability to walk and run but at a reduced speed while level II indicates limited walking ability and minimal ability for running independently. Children with GMFCS levels of I and II were considered ambulatory. Level III indicates the ability to walk only with an assistive device or the help of others. Level IV and V reflect minimal or lack of independent motor function and the need for a wheelchair. Children with GMFCS levels III -V were considered nonambulatory because they were unable to ambulate without an assistive device.

Body composition

Whole body DXA scans for all participants were acquired using a Delphi-W (Hologic, Bedford, MA) DXA densitometer. Children with CP were secured from the waist down using a modified version of the BodyFix (Medical Intelligence, Inc., Schwabműnchen, GER) during scanning acquisition to reduce potential involuntary contractions. The procedure has been shown to have no effect on body composition measurements [19]. Whole body FFST (FFST_whole) and fat mass were first determined (excluding the head). Upper and lower limb regions of the whole body scan were isolated, and the FFST in each limb were added together to calculate AFFST. In addition, FFST in each upper limb were added together to calculate upper body FFST (FFST_upper), and FFST in each lower limb were added together to calculate lower body FFST (FFST_lower). The remaining FFST (excluding the head) was considered non-appendicular FFST. The reliability of FFST_whole, AFFST, FFST_upper and FFST_lower measurement was assessed in 26 typically developing children (5 - 14 years) by testing them twice on the same day after repositioning. Reliability was excellent, as indicated by an intraclass correlation coefficient > 0.99 for all measures. To account for the shorter stature that children with CP usually exhibit, height-adjusted indexes were calculated as follows:

$$\begin{gathered} \text{AFFST/ht}=\text{AFFST}(\mathrm{kg})∕\text{height (m)} \\ {\text{AFFST/ht}}^2=\text{AFFST}(\mathrm{kg})∕{\text{height}}^2(m^2) \end{gathered}$$

Statistical analysis

Data were analyzed using SPSS version 24.0 (IBM Corp, Armonk, NY). All data were checked for normality first, and mean group differences between CP and typically developing children were compared using an independent t-test or a Mann-Whitney U test accordingly. To compare data from nonambulatory children with CP, ambulatory children with CP and typically developing children, subgroup analyses were performed using ANOVA if data were normally distributed and using the Kruskal-Wallis test if data were nonnormally distributed. To control for type I error, a Bonferroni adjustment was applied for post-hoc comparisons if the variances were equal and the Games-Howell test was applied for post-hoc comparisons if the variances were unequal. Effect size was estimated using Cohen's d (d), with 0.2, 0.5 and 0.8 indicating small, medium and large effect sizes, respectively [20] when applicable. An alpha level of 0.05 was used for all significance tests.

Multiple regression analyses were first performed in typically developing children to determine whether simple physical characteristics data (i.e., age, sex, height, body mass and BMI) could be used to accurately estimate AFFST, AFFST/ht and AFFST/ht². All independent predictors were examined for interactions and if they were significant contributors to the model. Models either included BMI, or body mass and height, depending on which explained the most variance in the dependent variable. Models did not include all three because inclusion of all three did not improve any of the models. The final models were cross-validated using the leave one-out technique [21]. Models were also applied to children with CP to determine whether accurate estimates could be obtained. Separate models using data from children with CP were created and checked for validity. Paired t-tests were used to determine if AFFST, AFFST/ht and AFFST/ht² estimated from predictive models were significantly different from their respective values measured using DXA. The validity of the models was also assessed using Bland-Altman plots [22] and scatter plots.

Results

Physical characteristics

Eighteen children with CP were classified as nonambulatory and 24 were classified as ambulatory. Physical characteristics for all participants were summarized in Table 1. No significant differences were detected for age, BMI or BMI percentile between typically developing children and children with CP. However, children with CP had lower height and height percentile (d = 0.690 and 1.358, respectively, bothp < 0.05), and lower body mass and body mass percentile (d = 0.413 and 0.842, respectively, bothp < 0.05) as compared to the typically developing children.

Table 1.

Physical characteristics of children with cerebral palsy (CP) and typically developing children (Con).

	CP (n = 42)	NACP (n =18)	ACP (n =24)	Con (n = 42)
Age (y)	9.1 ± 2.5	9.9 ± 2.1	8.5 ± 2.6	9.2 ± 2.3
Height (m)	1.26 ± 0.15a	1.27 ± 0.15	1.25 ± 0.16a	1.36 ± 0.14
Height (%)	21 ± 26a	15 ± 22a	27 ± 28a	57 ± 27
Body mass (kg)	28.1 ± 11.2a	28.7 ± 12.2	27.6 ± 10.7	32.6 ± 10.6
Body mass (%)	31 ± 33a	27 ± 33a	35 ± 33a	56 ± 26
BMI (kg/m2)	17.2 ± 4.3	17.3 ± 4.9	17.1 ± 3.9	17.3 ± 2.9
BMI (%)	47 ± 36	44 ± 39	50 ± 34	53 ± 30
GMFCS (I/II/III/IV/V)	14/10/7/1/10	0/0/7/1/10	14/10/0/0/0	–

Values are mean ± SD. NACP = nonambulatory CP; ACP = ambulatory CP; BMI = body mass index. % for height, body mass and BMI reflects the percentile relative to age- and sex- based norms; GMFCS = gross motor function classification system.

^aDifferent from Con, p < 0.05.

^bDifferent from ACP, p < 0.05.

When children with CP were separated based on ambulatory status, nonambulatory children with CP had lower height percentile and body mass percentile (d = 1.705 and 0.976, respectively, bothp < 0.05) compared to typically developing children. Ambulatory children with CP had lower height, height percentile and body mass percentile (d = 0.732, 1.091 and 0.707, respectively, all p < 0.05) compared to the typically developing children.

Body composition

Body composition estimates from DXA are summarized in Table 2 and representative DXA scans from a child with CP and a typically developing child are presented in Figure 1. There was no group difference in fat mass, but compared to controls, children with CP had 21% lower FFST_whole (d = 0.661, p = 0.002). Appendicular and non-appendicular FFST were both lower in children with CP than in controls, but the degree of the difference was greater for AFFST (30 % lower, d = 0.909, p < 0.001) than non-appendicular FFST (14 % lower, d = 0.444, p = 0.019). Compared to controls, children with CP also had 23% lower AFFST/ht, 19% lower AFFST/ht², 9% lower AFFST/FFST_whole, 34 % lower FFST_lower, and 31% higher FFST_upper/FFST_lower (d = 0.909, 0.994, 0.907, 1.034, and 1.403, respectively, all p < 0.05).

Table 2.

Body composition estimates of children with cerebral palsy (CP) and typically developing children (Con) using dual-energy X-ray absorptiometry.

	CP (n = 42)	NACP (n = 18)	ACP (n = 24)	Con (n = 42)
Fat mass (kg)	8.4 ± 5.7	9.4 ± 7.1	7.6 ± 4.5	7.4 ± 4.3
FFSTwhole (kg)	15.3 ± 5.8a	14.8 ± 5.2a	15.7 ± 6.2	19.3 ± 6.3
AFFST (kg)	6.4 ± 2.6a	5.7 ± 2.2a	6.9 ± 2.9a	9.1 ± 3.3
non-appendicular FFST (kg)	8.9 ± 3.3a	9.1 ± 3.2	8.8 ± 3.4	10.3 ± 3.0
AFFST/ht (kg/m)	5.0 ± 1.6a	4.4 ± 1.2a	5.4 ± 1.7a	6.5 ± 1.7
AFFST/ht2 (kg/m2)	3.9 ± 1.0a	3.4 ± 0.7a	4.2 ± 1.0	4.8 ± 0.8
AFFST/FFSTwhole	0.42 ± 0.05a	0.39 ± 0.04a	0.44 ± 0.03a	0.46 ± 0.03
FFSTUpper (kg)	1.7 ± 0.7	1.6 ± 0.6a	1.8 ± 0.7	2.0 ± 0.7
FFSTlower (kg)	4.7 ± 2.0a	4.1 ± 1.6a	5.1 ± 2.2a	7.1 ± 2.6
FFSTupper/FFSTlower	0.37 ± 0.08a	0.40 ± 0.09a	0.36 ± 0.07a	0.29 ± 0.04

Values are mean ± SD. NACP = nonambulatory CP (Gross Motor Function Classification System (GMFCS) level III, IV and V); ACP = ambulatory CP (GMFCS level I and II); FFST = fat-free soft tissue mass; FFST_whole = whole body FFST; AFFST = FFST in the upper and lower appendages; AFFST/ht = ratio of AFFST to height; AFFST/ht² = ratio of AFFST to height²; FFST_upper = FFST in the upper limbs; FFST_lower = FFST in the lower limbs.

^aDifferent from Con, p < 0.05.

^bDifferent from ACP, p < 0.05.

Figure 1.

Whole body dual-energy X-ray absorptiometry (DXA) scan for a 9.4 year old child with cerebral palsy (A; CP) at GMFCS level II, and a typically developing child (B; Con) with the same age, sex and race. The white arrows point to the soft tissue, a combination of fat mass and fat free soft tissue at the appendicular regions. Children with CP have much less soft tissue compared to Con.

When children with CP were separated based on ambulatory status, all measures of FFST were significantly lower in nonambulatory children with CP compared to typically developing children (d range = 0.597 to 2.121, all p < 0.05), except for FFST_upper/FFST_lower which was significantly higher (d = 2.104, p < 0.001), and non-appendicular FFST which was not significantly different (d = 0.392, p = 0.530). The ambulatory children with CP had lower FFST_whole compared to typically developing children, although it was marginally insignificant (d = 0.576, p = 0.060). The ambulatory children with CP also had lower AFFST, AFFST/ht, AFFST/FFST_whole and FFST_lower compared to the typically developing children (d = 0.708, 0.647, 0.667 and 0.830, respectively, all p < 0.05). In addition, the ambulatory children with CP had higher FFST_upper/FFST_lower compared to typically developing children (d = 1.229, p < 0.001), but there was no significant difference between the two groups in non-appendicular FFST or FFST_upper (d = 0.478 and 0.286, respectively, both p > 0.14). When compared to ambulatory children with CP, nonambulatory children with CP had lower AFFST/ht² (d = 0.898, p = 0.012) and AFFST/FFST_whole (d = 1.458, p < 0.001). The group differences in AFFST/FFST_whole are demonstrated in Figure 2.

Figure 2.

Appendicular fat-free soft tissue mass (AFFST) to whole body fat-free soft tissue mass (FFST_whole) ratios for all participants. NACP = non-ambulatory children with cerebral palsy; ACP = ambulatory children with cerebral palsy; Con = typically developing children. ^aDifferent from Con, p < 0.05. ^bDifferent from ACP, p < 0.05.

Statistical models developed in typically developing children

Multiple regression analysis was performed using physical characteristics data from typically developing children to estimate AFFST, AFFST/ht and AFFST/ht². The resulting models yielded good estimates in typically developing children, as indicated by the high amount of variance explained (R² = 0.92, 0.88 and 0.73, respectively, all p < 0.001; Table 3). When the models were cross-validated using the leave-one-out technique, the estimates were strongly related and not different from measured AFFST, AFFST/ht and AFFST/ht² (r² = 0.89, 0.84 and 0.65, respectively; all p > 0.91; Figure 3A-C). However, when the difference in estimated values and measured values of AFFST, AFFST/ht and AFFST/ht² were plotted against measured values, there was an inverse relationship (r = −0.31, −0.39 and −0.56, respectively, all p < 0.05; Figure 3D-F). This indicated a trend for an overestimation of AFFST, AFFST/ht and AFFST/ht² for typically developing children with lower values and an underestimation for children with higher values.

Table 3.

Statistical models developed to estimate appendicular fat-free soft tissue (AFFST) mass and AFFST indexes in typically developing children.

Model	Outcome measure	Coefficients	β	t-Value	SE	p-Value	Model R2	Model adjusted R2
1	AFFST (kg)						0.917	0.908
		Intercept	−14.106	−4.685	3.011	0.000
		Sex	−0.416	−1.306	0.318	0.199
		Age (year)	−0.048	−0.273	0.177	0.786
		Height (m)	14.485	4.112	3.523	0.000
		Body mass (kg)	0.128	4.512	0.028	0.000
2	AFFST/ht (kg/m)						0.877	0.864
		Intercept	−3.565	−1.875	1.901	0.069
		Sex	−0.411	−2.044	0.201	0.048
		Age (year)	0.527	0.219	0.112	0.828
		Height (m)	5.487	2.468	2.224	0.018
		Body mass (kg)	0.08	4.479	0.018	0.000
3	AFFST/ht2 (kg/m2)						0.732	0.703
		Intercept	2.238	1.672	1.339	0.103
		Sex	−0.352	−2.488	0.142	0.017
		Age (year)	0.048	0.611	0.079	0.545
		Height (m)	0.386	0.246	1.566	0.807
		Body mass (kg)	0.052	4.148	0.013	0.000

AFFST/ht = ratio of AFFST to height; AFFST/ht² = ratio of AFFST to height²; male = 0 and female = 1 for sex; SEE = Standard error of estimation. All models p < 0.001.

Figure 3.

Scatter plots demonstrating the cross-validation of models for appendicular fat-free soft tissue mass from dual-energy X-ray absorptiometry (AFFST), the ratio of AFFST to height (AFFST/ht), and the ratio of AFFST to height squared (AFFST/ht²) developed using physical characteristics data from typically developing children (Con; A-C). The models were cross-validated using the leave-one out method and data from Con. The models were also cross-validated using data from children with cerebral palsy (CP). Estimated values are on the x-axis. Measured values are on the y-axis. The dotted lines represent the lines of identity. The thick solid lines represent the regression lines for the children with CP. The thin solid lines represent the regression lines for Con. The cross-validation was also evaluated using Bland-Altman plots (D-F). The solid horizontal lines with dotted lines above and below them represent estimated values minus measured values ± SD. The thick solid lines represent the regression lines for children with CP, and the thin solid lines represent the regression lines for Con.

Although the models developed using data from typically developing children yielded estimates of AFFST, AFFST/ht and AFFST/ht² that were moderately-to-strongly related to the measured values during the cross-validation in children with CP, they overestimated AFFST, AFFST/ht and AFFST/ht² by 11, 14 and 15 %, respectively (all p < 0. 01; Figure 3). The overestimation was demonstrated by most data points and the regression line residing below the lines of identity in the scatter plots (Figure 3A-C), and by most data points above the no difference line in the Bland-Altman plots (Figure 3D-F). When the difference in estimated values and measured values of AFFST, AFFST/ht and AFFST/ht² were plotted against measured values, there was a significant inverse relationship for AFFST/ht² (r = −0.66, p < 0.001; Figure 3F). This indicated a trend for a greater overestimation AFFST/ht² for children with CP with lower values. The differences between the estimated values and the measured values of AFFST, AFFST/ht and AFFST/ht² were significantly correlated with the low AFFST/FFST_whole in children with CP (r = −0.52, −0.66 and −0.74, respectively; all p < 0.001), as shown in Figure 4.

Figure 4.

Scatter plots of the ratio of appendicular fat-free soft tissue mass to whole body fat-free soft tissue mass (AFFST/FFST_whole) compared to (A) the difference in appendicular fat-free soft tissue mass (AFFST), (B) the ratio of AFFST to height (AFFST/ht), and (C) the ratio of AFFST to height² (AFFST/ht²) and their estimates in children with cerebral palsy (CP) by statistical models developed using data from typically developing children. AFFST_Estimatee, AFFST/ht_Estimate and AFFST/ht²_Estimate represent the estimates by the models developed using data from typically developing children. The horizontal dotted line represents the point where the difference between AFFST_Estimate and AFFST measured is zero. The vertical dotted line represents the average AFFST/FFST_whole for typically developing children. The thick solid lines represent the regression lines for children with CP.

Statistical models developed in children with CP

Multiple regression analysis was performed using physical characteristics data from children with CP to estimate AFFST, AFFST/ht and AFFST/ht². The resulting models yielded good estimates, as indicated by the high amount of variance explained (R² = 0.85, 0.77 and 0.64, respectively, all p < 0.001; Table 4). When the models were cross-validated using the leave-one out method, the results were strongly related to (r² = 0.78, 0.66 and 0.50, respectively, all p < 0.001; Figure 5) and not different from measured AFFST, AFFST/ht and AFFST/ht² (all p > 0.95). Good estimation is indicated visually by most data points and the regression lines near the lines of identity in the scatter plots (Figure 5A-C), and by most data points and the mean difference lines near the no difference lines in the Bland-Altman plots (Figure 5D-F). However, when the difference in estimated values and measured values of AFFST, AFFST/ht and AFFST/ht² were plotted against measured values, there was an inverse relationship (r = −0.38, −0.47 and −0.61, respectively, all p < 0.05; Figure 5D-F). This indicated a trend for an overestimation of AFFST, AFFST/ht and AFFST/ht² for children with CP with lower values and an underestimation for children with higher values.

Table 4.

Statistical models developed to estimate appendicular fat-free soft tissue mass (AFFST) and AFFST indexes in children with cerebral palsy.

Model	Outcome measure	Coefficients	β	t-Value	SE	p-Value	Model R2	Model adjusted R2
1	AFFST (kg)						0.854	0.833
		Intercept	−3.512	−1.529	2.296	0.135
		Sex	−0.595	−1.602	0.371	0.118
		Age (year)	0.220	1.422	0.155	0.164
		Height (m)	4.342	1.493	2.908	0.144
		Body mass (kg)	0.118	4.721	0.025	0.000
		Ambulatory status	−1.484	−3.696	0.402	0.001
2	AFFST/ht (kg/m)						0.770	0.738
		Intercept	2.277	1.323	1.721	0.194
		Sex	−0.449	−1.612	0.278	0.116
		Age (year)	0.179	1.547	0.116	0.131
		Height (m)	−0.666	−0.306	2.179	0.762
		Body mass (kg)	0.091	4.853	0.019	0.000
		Ambulatory status	−1.134	−3.769	0.301	0.001
3	AFFST/ht2 (kg/m2)						0.637	0.597
		Intercept	1.360	2.744	0.496	0.009
		Sex	−0.334	−1.561	0.214	0.127
		Age (year)	0.106	2.500	0.042	0.017
		BMI (kg/m2)	0.120	5.062	0.024	0.000
		Ambulatory status	−0.839	−3.914	0.214	0.000

AFFST/ht= ratio of AFFST to height; AFFST/ht² = ratio of AFFST to height²; male = 0 and female = 1 for sex; ambulatory = 0, and nonambulatory = 1 for ambulatory status; SEE = Standard error of estimation. All models p < 0.001.

Figure 5.

Scatter plots demonstrating the cross-validation of models for appendicular fat-free soft tissue mass from dual-energy X-ray absorptiometry (AFFST), the ratio of AFFST to height (AFFST/ht), and the ratio of AFFST to height squared (AFFST/ht²) developed using physical characteristics data from children with cerebral palsy (CP; A-C). The models were also cross validated using data from children with CP and the leave-one-out method. Estimated values are on the x-axis. Measured values are on the y-axis. The dotted lines represent the lines of identity. The cross-validation was also evaluated using Bland-Altman plots (D-F). The dotted lines represent AFFST estimated minus AFFST measured ± SD. The thin solid lines represent the regression lines for children with CP.

Discussion

To our knowledge, this is the first study to demonstrate that the FFST deficit in children with CP is greater at the appendicular regions than at the non-appendicular region. In addition, we demonstrated that AFFST, along with its height-adjusted indexes (AFFST/ht and AFFST/ht²), can be estimated with reasonable accuracy in children with CP using models based on physical characteristics. However, the models must be generated using data from children with CP. Although models developed using data from typically developing children yield reasonably accurate estimates of AFFST and its indexes in typically developing children, significant overestimates result when the models are applied to children with CP. Low AFFST is a concern because it has been linked to multiple comorbidities in adults and older people, including functional impairment [23], physical disabilities [23], impaired bone structure [24, 25], impaired balance [24] and cardiovascular risks [26, 27]. The low AFFST in children with CP may be an early indicator of these complications, which are all present at an accelerated rate in individuals with CP [2, 28-32].

In the present study, AFFST and non-appendicular FFST were both lower in children with CP, but there was a greater compromise in AFFST as demonstrated by the greater percentage difference (30 % vs 14 %) and greater effect size (d = 0.909 vs. d = 0.444) when compared to typically developing children. These deficits are also reflected by the lower AFFST/FFST_whole in children with CP. The observation that the deficit in FFST is more pronounced in the appendicular than the non-appendicular regions is consistent with previous studies that show a proportionately smaller FFST in the arms and legs than in the trunk of individuals with spinal cord injury [12, 13], a group that experiences exceptional loss of muscle and physical activity after injury. The preferential deficit of FFST in the limbs versus the trunk may be attributed to the larger proportion of muscle in the FFST of the limbs and the larger proportion of non-muscular tissue (e.g. internal organs) in the FFST of the trunk [11]. It is likely that skeletal muscle compared to non-muscle is more affected by mobility-related issues.

The reason for the diminished AFFST in children with CP compared to their typically developing peers is likely multi-factorial. Physical activity levels in children with CP are very low, with some reports of 70 to 80 % lower levels in nonambulatory children with CP [3, 33] and ~40% reduction in ambulatory children with CP [2]. Such a low level of physical activity may contribute to a muscular deficit simply by disuse; specifically, there is insufficient stimulus for muscle growth. This problem may be exacerbated as children with CP grow older because there is evidence that gait function declines as children with CP age, which may contribute to further sedentary behavior in this population [34]. Another potential contributing factor is malnutrition, which is common in children with CP [35]. It was reported that the prevalence of malnutrition is much higher in women with sarcopenia, as defined by AFFST/ht², than in those without sarcopenia [36]. Whether a similar phenomenon is present in children with CP is yet to be determined. Catabolic hormones like myostatin may also play a role in this process. There is some evidence that myostatin is upregulated in children with CP [37]. It was reported that myostatin RNA expression tends to be higher (29%; p = 0.09) in older adults with sarcopenia than in those without sarcopenia [38]. However, to our knowledge, the relationship between myostatin and AFFST has not been examined in children with CP. More research is required to identify specific mechanisms underlying the lower AFFST in children with CP compared to typically developing children.

Another novel finding in the current study was the disproportionate FFST compromise in the lower limbs compared to the upper limbs in children with CP. Although FFST was lower in the upper and lower limbs of the nonambulatory children with CP when compared to typically developing children, the deficit was smaller in the upper limbs. This was confirmed by the higher FFST_upper/FFST_lower in the nonambulatory children with CP compared to typically developing children. In the ambulatory children with CP, a detectable compromise in FFST was found in the lower limbs, but not in the upper limbs. The detectable FFST deficit in the upper limbs of nonambulatory children, but not the ambulatory children with CP is probably due to a much lower usage of the arms by the nonambulatory children. Most of the nonambulatory children with CP (14 out of 18) were quadriplegic, where all four limbs are affected. On the other hand, all of the ambulatory children with CP were either diplegic, where the two lower limbs are affected, or hemiplegic, where the upper limb and lower limb on one side are affected. It is likely that a deficit in FFST is present in children with milder forms of CP, such as those who are ambulatory, but larger samples and/or more sensitive measures that specifically assess muscle are needed.

The finding that models developed using data from typically developing children and simple characteristics, such as sex, age, height, body mass and BMI, overestimate AFFST and indexes of AFFST when applied to children with CP is consistent with previous studies. For example, DXA-based models developed using data from typically developing children overestimated midthigh muscle mass by 12-15 % [39] and midleg muscle mass by 13-22 % [40] in children with CP. The finding that AFFST and AFFST indexes can be estimated in children with CP by models developed using data from children with CP is encouraging. However, the models tended to overestimate AFFST in children with CP with lower values and underestimate AFFST children with higher values. Therefore, further efforts are needed to develop models to estimate AFFST and associated indexes.

In addition to the novel findings already reviewed, there are strengths of this study that should be discussed. First, children with CP and typically developing children were matched for age, sex and race. Moreover, the height, body mass and BMI for the typically developing children were not different from the 50^th sex- and age-based percentiles. Therefore, the differences in body composition observed in the current study likely reflect the population difference between children with CP and their typically developing peers. Second, FFST at appendicular sites was expressed not only as an absolute value, but also as indexes that corrected for the shorter stature of the children with CP (i.e., AFFST/ht and AFFST/ht²). A marked reduction was still present with the relative measures of AFFST indicating that the level of compromise in children with CP is striking.

The limitations of this study must also be addressed. First, although commonly used to represent skeletal muscle mass at the appendicular regions, AFFST is not a perfect surrogate for skeletal muscle mass. Apart from skeletal muscle, FFST also contains tendons, ligaments, vessels and other connective tissues. Therefore, the actual appendicular skeletal muscle mass is only a portion of the AFFST measured by DXA. In addition, children with CP have a lower proportion of muscle in the FFST as reflected by a lower ratio of skeletal muscle mass to FFST [39, 40]. Therefore, the actual deficit in muscle is probably even greater in this population than is reflected by AFFST. Second, regression models were developed in the present study using limited physical characteristic data, such as age, sex and height. In a previous study of adults, it was found that AFFST is accurately estimated by models that include sex, height and weight, as well as other predictors, such as hip circumference and grip strength [9]. Whether adding other anthropometric measures and muscle strength can increase predictability for models specific to children with CP is unknown and needs further investigation. Despite the limitation, the models developed in the present study provided moderate-to-strong estimates of AFFST and indexes of AFFST in children with CP. Lastly, it is unknown if the statistical models developed in the present study can be used to monitor changes in children with CP due to growth, maturation, surgery or alterations in physical activity, nutrition or rehabilitation. Future studies are needed to assess the usefulness of these models for scientists studying and clinicians monitoring the muscle development and health in children with CP. If proven valid, the models developed in the current study may provide a convenient method for researchers and clinicians to assess the longitudinal muscle growth pattern, as well as the effectiveness of intervention on muscle development, in children with CP.

Conclusion

Children with CP have a remarkable deficit in FFST that is more pronounced in appendicular than non-appendicular (trunk) regions, with greater deficits noted in the lower limbs than in the upper limbs. Although the unique FFST profile is more pronounced in nonambulatory children, it is also present in ambulatory children with CP. Preliminary models developed in the current study using data from children with CP can provide good estimates of AFFST, but further development of the models to estimate AFFST indexes may be needed.