Musculoskeletal growth in the upper arm in infants after obstetric brachial plexus lesions and its relation with residual muscle function


Dr Johannes A van der Sluijs, Department of Orthopaedic Surgery, VU University Medical Centre, Boelelaan 1117, PO Box 7057, 1007 MB Amsterdam, the Netherlands. E-mail:


Aim  Denervation after obstetric brachial plexus lesion (OBPL) is associated with reduced musculoskeletal growth in the upper arm. The aim of this study was to investigate whether reduced growth of upper arm flexor and extensor muscles is related to active elbow function and humeral length.

Method  In this study, 31 infants age less than 6 months (mean age 4.3mo; range 2.1–5.9mo; 17 males; 14 females;) with unilateral OBPL (Narakas class I, 19; II, 3; III, 2; and IV, 7) treated at the VU medical centre, in whom neurosurgical reconstruction was considered were prospectively studied using magnetic resonance imaging of both arms at a mean age of 4.3 months. Humeral length and the cross-sectional area (CSA) of elbow flexor and extensor muscles were measured in both upper arms. Paresis of elbow function was estimated when the infants were a mean age of 4.5 months using the Gilbert score.

Results  Both flexor and extensor CSAs were significantly smaller on the affected side than on the unaffected side (88% [SD 32%], p=0.020, and 88% [SD 24%], p=0.001 respectively), as was humeral length (96% [SD 7%], p=0.005) (unaffected side 100% in all cases). There was no relation between the reduction in flexor and extensor CSA and residual muscle function. In 17 out of 31 patients, hypertrophy of flexor and/or extensor muscles was observed. Humeral length was not related to muscle parameters.

Interpretation  Denervation has different effects on muscle growth and function as well as bone growth. In young infants with an OBPL, muscle size is not a predictor of muscle function. Flexion contractures of the elbow later in childhood may not be explained by a dominance of flexor muscle mass in infants.


Cross-sectional area


Obstetric brachial plexus lesion

What this paper adds

  •  Muscle CSA and humeral length are reduced in the upper arm after plexus lesions in infants aged <6 months.
  •  There is a lack of correlation between function of denervated upper arm muscles and muscle CSA on MRI.
  •  In about half the infants, hypertrophy of (partially) denervated muscles is found.

Obstetric brachial plexus lesion (OBPL), also known as birth brachial plexus injury, occurs in 1 to 4.6 per 1000 live births. In most cases the upper trunk of the brachial plexus is involved. Approximately 70% to 90% of affected infants recover without noticeable sequelae.1,2 Those with impaired arm function exhibit varying degrees of paresis, contractures, and growth reduction in the upper extremity.3,4 The contractures are generally considered to be related to abnormal muscle development secondary to neurological lesions.3,4 Shoulder paresis and contractures have been studied extensively;3 however, little is known about the early changes in the upper arm, the mechanisms underlying the development of contractures around the elbow, and the relation between contractures and muscle growth and function. In children with residual signs after OBPL, the prevalence of elbow contractures has been reported to be as high as 70%.5,6 Prevention of elbow contractures requires knowledge of the underlying mechanisms through which (partial) denervation of muscles causes these contractures.

In general, muscle denervation or spinal cord injury in animals, as well as in humans, results in fast and severe atrophy of skeletal muscle.7,8 In both children and infants with OBPL, a reduction in the size (atrophy) of shoulder muscles has been shown.3,9–12 Partial denervation of a muscle or group of muscles around a joint may cause an imbalance in muscle forces exerted around the joint. The imbalance in exerted muscle force may be a causative factor in the development of contracture around the elbow and differences in growth rates of agonist and antagonist muscles.

The relation between size and residual function of elbow flexor and extensor muscles after (partial) denervation in infants is relatively unknown. In children with OBPL, in addition to adaptation of muscles after partial denervation, bone growth has also been shown to be affected.3 Based on clinical measurements in children in the age range of 2 months to 14 years 1 month, the length of the upper arm on the affected side has been shown to be significantly smaller.13 One of the various factors related to bone mass is muscle mass, although this mechanism is unclear as bone mass growth precedes muscle mass growth in puberty.14,15 Based on this relation, we expected that elbow flexion and extension and bone growth would be to some extent related in children with OBPL.

The aim of this study was to investigate whether reduced growth of upper arm flexor and extensor muscles is related to active elbow function and humeral length. We hypothesized that soon after denervation the growth of flexor and extensor muscles of the upper arm is reduced and that this reduction correlates with reduced muscle function. We also expected to find a relation between muscle size, muscle function, and humeral length growth. These hypotheses were tested in a prospective study in infants under the age of 6 months with OBPL.



This prospective study included 31 infants (17 males, 14 females) younger than 6 months (mean age 4.3mo; range 2.1–5.9mo) with unilateral OBPL treated in the VU medical centre, Amsterdam. These infants belong to a subgroup of an ongoing study of infants and children with sequelae of OBPL. In these 31 infants considered for neurosurgery between 1999 and 2007, magnetic resonance imaging (MRI) was used as part of the preoperative screening for neurosurgery. The study was approved by the Medical Ethical Examination Committee of the Vanderbilt University Medical Center and informed consent was obtained from the infants’ parents. All but three children went on to have a neurosurgical correction. At the time the measurements were performed none of the children had been operated on.

Children with unilateral OBPL, Narakas class I–IV,16 in whom MRI was adequate and included the entire upper arm were eligible for the study. Infants who had bilateral OBPL (n=1) or in whom MRI was of low-quality or protocols were incomplete (n=50) were excluded.

Infants were sedated for the MRI while their posture was standardized with both hands on their abdomen. After visualization of the neural structures, the upper arms were visualized with a three-dimensional, fast-imaging, steady-state precession pulse-acquisition sequence imager (repetition time 25ms, echo time 10ms, flip angle 40°). The partitions used ranged from 0.8 to 1.5 mm. The protocol included imaging of the affected and unaffected arms to enable comparison with the unaffected anatomy. Software from Centricity RA 600 (General Electric Health Care, Slough, UK) was used to measure lengths and cross-sectional areas (CSAs) in the MRI scans. MRI measurements of length and CSA were determined by one of the authors (JMR); the MRI measurements were obtained without knowledge of the muscle function scores.

Parameters were related to (1) extent of neurological lesion, (2) humerus length, (3) elbow flexor and extensor muscle size, and (4) muscle function.

Extent of obstetric brachial plexus lesion

OBPL was classified according to Narakas: class I, deltoid and biceps paresis, C5, C6 injured; class II, deltoid and biceps paresis plus paresis of the extensors of the elbow, hand, and fingers, C5, C6, C7 injured; class III, almost complete paresis, C5, C6, C7, C8; class IV, total arm paresis, Horner syndrome, C5, C6, C7, C8, T1.

Measurements of humerus dimension

Measurements of humeral dimensions of affected and unaffected arms were made on transversal images. To measure the length of the humerus, we determined the most proximal MRI slice showing the top of the cartilaginous humeral head and the most distal slice showing the cartilaginous distal humerus (at the level of the capitulum/trochlea). Subtracting the two slice numbers and considering slice thickness yielded the humeral length in centimetres. To correct for age and interindividual differences, the humeral length of the affected side was also expressed as a ratio of the unaffected side.

Measurements of cross-sectional area of upper arm muscles

The CSAs of the flexors and extensors were measured at the level of the most distal humeral insertion of the deltoid muscle (Fig. 1a,b). Flexors are biceps and brachialis and extensors are triceps. Using MRI it is not possible to distinguish between the two flexor muscles at this stage and both flexors were therefore measured together. To correct for age and interindividual differences, we expressed the values for the muscles of the affected side relative to the unaffected side.

Figure 1.

 Examples of magnetic resonance images of both upper arms of an infant with obstetric brachial plexus lesion showing measurements of cross-sectional areas (CSA) of the muscles; (a) affected arm, (b) unaffected arm. Flexor and extensor muscle groups of affected and unaffected arm are outlined and the enclosed areas (i.e. CSA of the muscles) were measured. Note the typical difference in muscle CSA measured in the affected and unaffected arms.

Muscle function measurements according to Gilbert score

Elbow flexion and extension were measured using the Gilbert and Tassin muscle grading score.17 In this system, four categories are classified: M0, no contraction; M1, contraction without movement; M2, slight or complete movement with weight eliminated; and M3, complete movement against the weight of the corresponding segment of extremity. The Gilbert score was measured by the neurosurgeon (WJO) and was performed every month. We used the elbow function score of the month the MRI was carried out, at mean 4.5 months (SD 1mo). The MRI was carried out at mean 4.3 months (SD 1mo). Muscle function and size were measured in the same month.

Statistical analysis

All data were collected and analysed using SPSS (version 15.0; SPSS Inc., Chicago, IL, USA). Results are presented as means (SD).

Differences between the affected and unaffected side were tested using paired t-tests. Relations between parameters were tested using Pearson’s correlation coefficient test.

To correct for age and interindividual differences, humeral length and CSA measurements are also expressed as the ratio of the value on the affected side to the value on the unaffected side. The purpose of the ratio is to reduce the differences between infants, since infants differ in size. By comparing the affected side with the unaffected side of the same infant, the effect of the OBPL on growth can be measured. This is expressed as a percentage of the unaffected side, or ratio. Note that a ratio over 1 means that the measurement at the affected side was larger than at the unaffected side.

The Wilcoxon test was used for non-parametric paired data (Narakas class and Gilbert score). Tests were two-tailed and a p value of < 0.05 was considered significant.


For an overview of the characteristics of the 31 infants, see Table I. There were no complications related to the MRI protocol. The measured parameters did not differ between females and males.

Table I.   Overview of infant characteristics (n=31)
Sex, nSide of OBPLGestational age, wks
Mean (SD)
Birthweight, kg
Mean (SD)
Narakas class16 (n)
  1. OBPL, obstetric brachial plexus lesion.

Males 17, females 14Left 16, right 1739.4 (2.5)4.1 (0.78)I (19), II (3), III (2), IV (7)

The passive range of motion in these infants was normal. There were no flexion contractures of the elbow, and because of this, no formal measurements were taken.

Clinically, in Narakas I the upper arm is internally rotated with the elbow slightly flexed. In Narakas II, besides internal rotation of the upper arm, the elbow is more extended and the wrist/hand is in pronation/flexion. In Narakas III, hand paresis is more pronounced, while in Narakas IV there is complete paresis. At this stage the biceps bulge is present; however, in some, particularly the more severe cases, it is smaller than usual.

Extent of obstetric brachial plexus lesion

Narakas class was related to the combined Gilbert score of elbow flexion and extension (r = −0.604; p<0.001) but was not related to humeral length, muscle size, or function.

Humeral length

Table II shows means of humeral length measures. Humeral length on the affected side was 0.4 cm (i.e. 4.0%) smaller (p=0.005; 95% confidence interval of the difference −0.64 to −0.12) than that measured on the unaffected side.

Table II.   Effects of obstetric brachial plexus lesion on humeral length, cross-sectional area of elbow flexor and extensor muscles, and Gilbert score. Relation to Narakas class is shown
 Unaffected mean (SD)Affected mean (SD)Ratio mean (SD)95% CI of the difference
  1. Values of the unaffected and affected side are represented by the mean (SD). Narakas class:16 I, n=19; II, n=3, III; n=2, IV; n=7. Ratios were estimated by dividing the value for the affected side by value for the unaffected side. aDifferences between the unaffected and affected sides were significant (p<0.05, paired t–test; for Gilbert score, Wilcoxon test). CI, confidence interval; CSA, cross-sectional area.

Humeral length, cm9.9 (0.9)9.5 (0.1)0.96 (0.07)−0.64 to −0.12 (p=0.005)a
 Narakas class I9.99.40.95
 Narakas class II10.810.70.99
 Narakas class III9.09.81.08
 Narakas class IV9.69.20.96
Flexor CSA, cm²2.24 (0.73)1.89 (0.59)0.88 (0.32)−0.64 to −0.06 (p=0.02)a
 Narakas class I2.281.970.92
 Narakas class II1.881.630.90
 Narakas class III2.331.920.83
 Narakas class IV2.241.760.80
Extensor CSA, cm²3.69 (0.71)3.17 (0.75)0.88 (0.24)−0.80 to −0.24 (p=0.01)a
 Narakas class I3.763.400.94
 Narakas class II3.252.420.74
 Narakas class III3.743.110.84
 Narakas class IV3.682.880.79
Ratio flexor CSA/extensor CSA0.62 (0.03)0.61 (0.04) −0.06 to 0.07 (p=0.97)
 Narakas class I0.620.60 
 Narakas class II0.580.68 
 Narakas class III0.630.62 
 Narakas class IV0.610.63 
Flexor function (Gilbert score)30.970.32 (0.44)0.48–s1.46 (p<0.001)a
 Narakas class I31.420.47
 Narakas class II30.430.14
 Narakas class III300
 Narakas class IV300
Extensor function (Gilbert score)31.810.6 (0.44)1.32–2.29 (p<0.001)a
 Narakas class I32.210.74
 Narakas class II32.670.89
 Narakas class III31.500.50
 Narakas class IV300

Muscle size measurements

Flexor muscle cross-sectional area

Compared with the unaffected side, the mean flexor CSA of the affected side was significantly smaller (p=0.02; t-test), with a mean ratio of affected to unaffected sides of 0.88 (SD 0.32), i.e. flexor CSA on the affected side was 88% of that on the unaffected side. This ratio was not related to the humeral size or to the Gilbert function score of flexors.

In nine children the ratio was above 1, which means that flexor CSA was higher on the affected side than the unaffected side. These high ratios were not related to any other parameter.

Extensor muscle cross-sectional area

The extensor CSA was also significantly smaller on the affected side, (p=0.001; t-test) and the mean ratio of extensor CSA on the affected side to that on the unaffected side was also 0.88 (SD 0.24), i.e. extensor CSA on the affected side was 88% of that on the unaffected side. Eight infants showed an extensor CSA ratio above 1, meaning that the measurement on the affected side was larger than on the unaffected side. This did not coincide with larger flexor muscle CSA.

Muscle function in the affected side

Both extensors and flexors were significantly weaker in the affected arm, but mean elbow flexion, as measured by the Gilbert score, was reduced more than mean elbow extension (0.97 vs 1.81 [unaffected side 3]; Wilcoxon test; p=0.019).

Relation between upper arm muscle sizes and Gilbert score

In the affected upper arm, there was no relation between flexor or extensor muscle CSA and elbow flexion or extension function (Fig. 2). There was also no relation between function and CSA in antagonist muscles.

Figure 2.

 Gilbert score of the elbow muscles is not related to muscle cross-sectional area (CSA). (a) Individual values of the ratio of the flexor CSA (i.e. affected/unaffected) expressed as a function of the Gilbert score of the elbow flexion. (b) Individual values of the ratio of the extensor CS (affected/unaffected) as a function of the Gilbert score of the elbow extension. Means are indicated by horizontal lines showing that, on average, both are about 90% of the unaffected side (ratio=0.9).

There was no relation between the ratio of the CSA of the affected flexor (biceps and brachialis) muscles to the CSA of affected extensor (triceps) arm muscles and the Gilbert scores for elbow flexion and extension (Fig. 3). The mean of the ratio of the CSA of the flexor muscles to the CSA of the extensor muscles on the affected side was 0.61 (SD 0.20); the corresponding mean on the unaffected side was 0.62 (SD 0.18).

Figure 3.

 Gilbert score of the elbow muscles is not related to the extent to which muscle are selectively affected on either the flexor or extensor side. The ratio of the cross-sectional areas (CSAs) of affected elbow flexor and affected extensor muscles fluctuates around 1, meaning that flexor and extensor muscles are, on average, affected equally (values in each case about 90% of that on the unaffected side). Individual values of the ratio of the CSA of flexor and extensor muscles on the pathological side are expressed as a function of (a) elbow flexion and (b) elbow extension. Means are indicated by horizontal lines; no relation is present.

Relations between muscle cross-sectional area, muscle function and bone length

On both the affected and unaffected side, there was no relation between humeral length and CSAs of flexor and extensor muscles. On the affected side there was no relation between humeral length reduction and residual muscle force (as measured by the Gilbert score) of flexors or extensors.


Our hypotheses that affected muscle size and residual function are related and that reduced bone growth is related to residual muscle size and muscle functions were not confirmed.

Several limitations of this study should be considered. First, measurements of muscle CSA were related to muscle function using the ordinal Gilbert score. This score is a widely accepted practical test for clinical muscle function measurement in this age group.18 The score tests the ability to make particular joint movements on a 4-point ordinal scale. Although only ordinal, it is a standard clinical test in this young age group. Second, differences in elbow flexion during MRI may have affected the estimates of muscle CSA. During MRI, both of the child’s hands were placed on the abdomen. In this position, some transversal scans were possibly taken at an angle with respect to the line of pull of the muscle fibres and hence muscle CSA may have been slightly overestimated. In addition, on both sides, the elbow was positioned in a similar degree of flexion; however, small differences in elbow flexion (<15º) during the measurements may have affected muscle length and thus muscle CSA. Given an assumed biceps moment arm of maximum 1cm, a difference of 15° elbow flexion will result in a change of sign 15°=0.25×1cm, which is 2.5 mm. Assuming biceps length to be similar to humerus length (i.e. 7.0–12.0cm), the effects on muscle CSA are presumably less than 5% and not systematic. A final limitation is that this study involved infants with a severe OBPL, in whom neurosurgical reconstruction was considered because of absent natural recovery. It might be that in infants with less neurological lesions and more recovery, the relation between muscle size and function is different.

On average, OBPL in the infants caused a reduced growth of both flexor and extensor muscle groups. A lack of growth was expected, as denervation of mature muscle causes severe atrophy.7,8 Generally, in healthy muscle, the physiological muscle CSA is proportionally related to its optimum muscle force.19 As mentioned before, (partial) denervation is expected to reduce the growth rate of muscles as it reduces the rate of protein synthesis and stimulates the rate of protein degradation.20 On average, muscle size and function in our patients were reduced; however, a correlation between flexor and extensor CSA and the Gilbert score for the affected side was absent.

Several mechanisms may underlie the lack of correlation: (1) both selective denervation or reinnervation of motor units and fibre type-dependent susceptibility to denervation-induced atrophy7 affect the relation between muscle size and function; (2) the combination of normal/hypertrophic CSA with absent function might be due to an increased proportion of fatty tissue in the muscles, although this was not observed on MRI; and (3) neural lesions might affect the transmission of action potential and hence the recruitment of the number of active motor units and force output. However, local neurotrophic ‘growth’ factors released by functional motor neurons and/or endocrine anabolic factors remain present. Such factors have been suggested to continuously stimulate muscle fibre hypertrophy during development.21,22 Taken together, the results suggest that (partial) neurological impairment predominantly affects muscle function rather than muscle growth and that after early partial motor denervation other yet unknown factors are involved in regulating muscle growth. Besides residual motor innervation and residual function, local and systemic growth factors as well as local mechanical interactions between muscle fibres may affect muscle growth.23 Although in the first 6 months there is no relation between size and function, this does not mean that, after regeneration of neural structures (later during childhood), such a relation may not arise.

Hypertrophy of denervated muscles was found in a substantial proportion of our patients. Although the mean CSA of the flexors was significantly smaller on the affected side than on the unaffected side, in nine patients flexor CSA was larger on the affected side than on the unaffected side. In eight out of the nine patients the difference was over 5%. In eight infants, extensor CSA was larger on the affected side, and in five of these the difference was more than 5%. The frequency and size of the differences make it unlikely that measurement errors were the only cause. Hypertrophy of muscles of the lower extremity after denervation has been reported previously24 and in shoulder muscles in infants with OBPL.11

Also, in some patients, humeral length was increased compared to the unaffected side. Although mean humeral length was significantly lower on the affected side than on the unaffected side, in seven children the humerus was longer on the affected side. Although, again, the lack of precise measurements may partly explain this discrepancy, it seems unlikely that this is the only reason as the finding was noted in almost one-quarter of the tested population, and in five of the seven infants the magnitude of the difference was more than 5%. Humeral hypertrophy in infants with OBPL is presumably an early and temporary phenomenon after denervation.

The premise that muscle denervation generally leads to atrophy may not always be valid in infants with OBPL. Nerve injuries (or partial nerve injury) may result in either hypotrophy or hypertrophy of growing muscles and bones.

In later stages of development, flexion contractures of the elbows are often reported.5,6 In such cases, one expects that this contracture is caused by the increased strength or activity of flexors compared with extensors or an imbalance in muscle size, or the potential to exert force. This is known to have occurred in children with a spastic cerebral paresis, causing shortening and/or stiffening of the affected (spastic) muscles compared with their antagonists, and consequently a limited range of joint motion.25,26 Based on our data, imbalances in muscle force or activity, as in cerebral palsy, do not seem to be present in the arm muscles of children with OBPL during the early stages after birth. Instead, because of the mechanism of the injury and the anatomy of the brachial plexus following OBPL, elbow flexors are more often affected than extensors4,16 and the latter are likely to dominate movement around the elbow into extension. Such dominance would strain the flexor muscles. In mice, a high muscle strain applied to a muscle in vivo has been shown to be a stimulus for longitudinal growth of mature muscle fibres by the addition of sarcomeres in series,27 but not in 1-week-old mice. Furthermore, ex vivo culture of mature muscle fibres has shown that strain per se is not sufficient to stimulate addition of sarcomeres in series.28 In addition to a high strain, contractile activity and/or anabolic growth factors may be necessary to induce an increase in the number of sarcomeres in series and hence longitudinal growth of the muscle.23 An alternative explanation for the development of contracture is that the flexor muscles become stiffer. Biopsies of subscapularis muscle obtained from children with OBPL have shown that the muscle fascicles become stiffer and that sarcomere slack length is reduced.29 These observations suggest that changes in muscle morphology during later development predominantly affect flexor muscle growth, with possibly a slower increase in the number of sarcomeres in series and/or increased stiffness of the muscle fibres or connective tissue. This confirms a recent study in mice which found decreased growth of the mouse biceps, brachialis, and subscapularis muscles after experimental surgical brachial plexus lesions.30

The variation in the degree of atrophy in the early stage during development implies large differences in muscle size at the time of neurosurgical reconstruction. Whether the long-term effects of neurosurgical reconstruction in this group of patients are co-determined by the initial effects on muscle size needs to be assessed in longitudinal studies on muscle function and size.

In conclusion, we found that, among infants with OBPL who experience little natural recovery in the first months of life and/or severe extension of the neural lesion, mean CSA of elbow flexor and extensor muscles on the affected side is reduced shortly after birth (<6mo) compared with the muscles of the unaffected side; however, hypertrophy was also observed in some patients. Contrary to our hypothesis, in the first 6 months after partial denervation, we were not able to find a relation between muscle size and residual muscle function, nor between muscle size or function and bone growth. The common occurrence of flexion contractures of the elbow during later childhood may not be explained by a dominance of biceps CSA in the first 6 months in infants with OBPL. Our findings support the experimentally confirmed notion that reduced or absent muscular length growth is a factor in the aetiology of contractures.30 Insight into the structural changes in muscle as well as in the degree of denervation is required to understand the development of contractures. Further investigation is needed to determine whether early estimates of the effects of the partial denervation are a predictor of the success of neurosurgical reconstruction.