- Email: [email protected]
Sensitivity and specificity of qualitative muscle ultrasound in assessment of suspected neuromuscular disease in childhood
Neuromuscular Disorders 17 (2007) 517–523 www.elsevier.com/locate/nmd
Sensitivity and specificity of qualitative muscle ultrasound in assessment of suspected neuromuscular disease in childhood Knut Brockmann
a,*
, Peter Becker a, Gudrun Schreiber a, Karin Neubert b, Edgar Brunner b, Carsten Bo¨nnemann c
a
c
Department of Pediatrics and Pediatric Neurology, Georg August University, Go¨ttingen, Germany b Department of Medical Statistics, Georg August University, Go¨ttingen, Germany Division of Neurology, The Children’s Hospital of Philadelphia, University of Pennsylvania School of Medicine, Philadelphia, USA Received 3 January 2007; received in revised form 14 March 2007; accepted 27 March 2007
We dedicate this work to Professor Folker Hanefeld on the occasion of his 70th birthday.
Abstract Muscle ultrasound is considered a useful noninvasive technique for visualizing normal and pathological skeletal muscle. We determined the accuracy of qualitative muscle ultrasound in the discrimination of normal muscle from myopathic, neurogenic, and unspecifically abnormal tissue changes in the evaluation of suspected NMD in childhood. Sensitivity and specificity of muscle ultrasound were assessed by comparing sonographic classification of muscle tissue changes in 134 children with definitive diagnosis as provided by muscle histology or mutation analysis performed subsequently to the sonography. We found a sensitivity of 81% and a specificity of 96% for detection of any abnormal muscle tissue alteration by ultrasound. For detection of neurogenic changes, sensitivity was 77% with even higher specificity (98%). Accuracy was slightly lower for myopathic changes (79%) and clearly lower for unspecific abnormal tissue alterations (70%). Accuracy of ultrasound was lower in younger children. High reliability of muscle sonography justifies a more widespread use of this method in evaluation of suspected NMD in childhood. Ó 2007 Elsevier B.V. All rights reserved. Keywords: Muscle ultrasound; Neuromuscular disorder; Childhood; Sensitivity; Specificity
1. Introduction Ultrasound imaging of the muscle was introduced in the field of neuromuscular disorders (NMD) in 1980 [1,2] and is now considered a useful technique for visualizing normal and pathological skeletal muscle [3–7]. Though ultrasound is successfully used in a
Abbreviations: CMT, Charcot-Marie-Tooth; DMD, Duchenne muscular dystrophy; MD, muscular dystrophy; NMD, neuromuscular disease; OXPHOS, oxidative phosphorylation; SMA, spinal muscular atrophy. * Corresponding author. Tel.: +49 551 39 6210; fax +49 551 39 6252. E-mail address: [email protected] (K. Brockmann). 0960-8966/$ - see front matter Ó 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.nmd.2007.03.015
number of centers it is not widely adopted in the evaluation of neuromuscular disease. Many neurologists still adhere to electromyography (EMG) as the preferred method for discrimination of myopathies vs. neurogenic conditions. A recent review on muscular dystrophies did not mention muscle ultrasound among the laboratory methods for assessment of NMD, which included serum creatine kinase, EMG, muscle histology, immunohistochemistry, and mutation analysis [8]. Nevertheless the use of muscle ultrasound is advantageous especially in children, as it is a painless, quick, relatively cheap, and readily reproducible investigation. Only a limited number of studies have aimed at determining the reliability of muscle ultrasound in
518
K. Brockmann et al. / Neuromuscular Disorders 17 (2007) 517–523
NMD. In adults with idiopathic inflammatory myopathies a robust correlation between histologic and sonographic changes has been reported [9,10]. In a pediatric study including 100 children aged 2 months to 16 years the usefulness of muscle ultrasound in the initial assessment of suspected NMD was investigated. Sensitivity and specificity were 78% and 91% for detection of any type of NMD [11]. In the subgroup of patients older than 3 years muscle ultrasound was even more reliable with a sensitivity of 81% and a specificity of 96%. Few studies aimed at differentiating between myopathies and neuropathies on the basis of muscle ultrasound alone. In a mixed population of children and adults with histologically or genetically confirmed NMD, a combination of conventional muscle ultrasound and computer-assisted texture analysis of sonograms was used. Values for sensitivity from 77% to 94% and specificity from 81% to 98% were found for discrimination between myogenic and neurogenic disorders [12]. Quantitative muscle ultrasound has been successfully used to differentiate between typical myopathies and neuropathies in adults [13]. However, in children quantitative muscle ultrasound was shown to be sensitive for distinguishing between healthy children and children with neuromuscular disorders in general [5] or between neuromuscular and non-neuromuscular diseases, e.g. cerebral palsy [14], but did not allow for a further distinction. A more detailed aspect analysis was presumed to be necessary for further distinguishing between different types of neuromuscular disorders [14]. Recently, a study compared the sensitivity and specificity of visual vs. quantitative evaluation of skeletal muscle ultrasound in children suspected of having a NMD [15]. Quantification of echo intensity was found a more objective and accurate method compared to visual evaluation for screening for NMD in general. However, a distinction between myopathies and neurogenic disorders was not achieved with this approach. Here we report the first study that seeks to determine the reliability of qualitative muscle ultrasound not only for the discrimination of normal and abnormal muscle, but additionally for the distinction between myopathic and neurogenic disorders in the evaluation of suspected NMD in childhood. Sensitivity and specificity of muscle ultrasound were arrived at by comparing sonographic classification of muscle tissue changes with the definitive diagnosis provided by histological or genetic analysis (gold standard). Although our data were collected retrospectively, both the sonologist and the pathologist were blinded as to the corresponding result.
2. Patients and methods In a retrospective review of data in our ultrasound laboratory we identified all patients (n = 826) that had been examined by muscle ultrasound as part of evaluation for suspected NMD during a five-year-period (1997–2001) in our Department of Pediatrics and Pediatric Neurology in a tertiary care University hospital. Only patients fulfilling the following criteria were included in the study: (i) examination by a single sonologist (K.B.) in order to standardize the sonologic criteria (n = 525); (ii) availability of a definitive, final diagnosis provided by muscle histology or molecular-genetic analysis (gold standard) (n = 184); (iii) performance of muscle ultrasound before muscle biopsy or molecular-genetic investigation were completed so that the ultrasound report was generated without any knowledge of the final diagnosis. These criteria were fulfilled by 134 patients (79 boys, aged 21 days to 17 years, mean 5 years 5 months; 55 girls, aged 18 days to 18 years, mean 6 years 4 months), whose ages ranged from 18 days to 18 years (mean 5 years 10 months). Fifty children were younger than 3 years, and of these 14 were younger than 12 months. Ultrasound scans were performed according to a standard protocol. A Siemens Sonoline Prima ultrasound system with 5 and 7.5 MHz linear probes was used. The 5 MHz probe with its greater depth of penetration was used for an overview of thigh and calf in order to detect selective affection of muscles. The 7.5 MHz probe provided depiction of tissue alterations in greater detail and thus allowed for differentiation between neurogenic and myogenic changes [16,17]. A series of standard images of the various muscles in healthy children of various ages had been acquired by the same ultrasonographer on the same machine and served as the normal reference set. Gain and setting parameters were adjusted so that healthy muscle appeared nearly black (Fig. 1A). This setting was conserved in the ultrasound system as a fixed investigation program and not changed during the patients’ investigations, apart from adjustment of the depth setting. Axial scans of mid thigh and upper calf were recorded in all patients in the lying position. The probe was moved along the thigh or calf until the optimal view most closely corresponding to the established standard image for the segment in question was found. Additional scans of upper arm or deltoid muscle were performed when clinically appropriate. Muscle ultrasound findings in muscle disease have been classified on a four-point scale by Heckmatt et al. [3] according to the intensity of echo reflected from the muscle. Heckmatt’s criteria constitute a ‘‘straight-
K. Brockmann et al. / Neuromuscular Disorders 17 (2007) 517–523
519
Fig. 1. Axial muscle ultrasound scans of mid thigh of (A) a healthy 7-year-old girl, (B) a 6-year-old boy with Duchenne muscular dystrophy (DMD), (C) a 12-year-old boy with DMD, (D) a 14-year-old girl with axonal neuropathy, and (E) a 10-year-old boy with unclassified encephalopathy, global retardation, and muscular hypotonia. In a healthy subject (A), muscle appears largely black with few perimysial septa. In myopathies (B), MUS reveals elevated homogeneous fine granular echogenicity. More advanced stages of muscular dystrophy (C) are characterized by marked homogeneous elevation of muscle echogenicity and loss of bone echo. In neurogenic conditions (D), increased echogenicity is accompanied by elevated inhomogeneity. Unspecific abnormal patterns (E) are characterized by elevated echogenicity without clear assignment to myopathic or neurogenic features.
forward and reproducible grading system’’ [11] and provide a semi-quantification of tissue changes. However, no information on the quality of muscle alterations observed is given by these criteria, and they do not help in differentiating between myopathies and neurogenic conditions. From the clinician’s point of view, the major questions posed to the sonography in evaluation of a child with suspected NMD comprise (i) whether the muscle is normal or abnormal, and (ii), in case of muscle abnormality, whether these changes are myogenic or neurogenic. Therefore, in this study muscle ultrasound findings were not classified according to Heckmatt’s criteria, but rather judged as being either (i) normal, (ii) myopathic, (iii) neurogenic, or (iv) unspecifically abnormal. Assignment of ultrasonographic findings to one of the categories mentioned above was based on visual assessment and the pattern recognition approach. Muscle thickness and ratios of muscle thickness to subcutaneous fat thickness were taken into account, but not quantified [18]. In NMD, the increase of fat, fibrosis, and inflammation results in an elevation of the numbers
of reflective interfaces in muscle [7]. Myogenic changes are characterized by a largely homogeneous fine granular elevation of echogenicity within the muscle (Fig. 1B). These alterations are most extensive in muscular dystrophies, resulting in a homogeneous high echogenicity of muscle tissue with loss of visualization of epimysium and bone [4]. Neurogenic alterations depict as inhomogeneous, patchy and stripy echo densities together with interposed hypodense areas (Fig. 1C). Atrophic fibres correspond to hyper-echoic, bright areas, whereas normal and hypertrophic fibres are visualized as hypoechoic, black areas. Recognition of this juxtaposition of fibre types is facilitated by high resolution ultrasound [16]. Fasciculations are readily observable in many cases of neurogenic disease, especially in spinal muscular atrophy (SMA) and axonal forms of CMT neuropathy [19,20]. In longstanding NMD, the differences between myogenic and neurogenic ultrasound patterns often become less distinct and are replaced by an undifferentiated picture with increased echogenicity resulting from loss of muscle mass and concentration of fibrous and fatty tissue in a smaller volume.
520
K. Brockmann et al. / Neuromuscular Disorders 17 (2007) 517–523
others in the whole group and for the age range up to 3 years.
Table 1 Diagnoses and number of patients Number of patients Myopathic disorders Duchenne MD Congenital MD Limb-girdle dystrophy Congenital myopathies Bethlem myopathy Metabolic myopathies Unclassified myopathies Dermatomyositis
47 2 14 1 5 1 3 17 4
Neurogenic disorders SMA total SMA type I SMA type II SMA type III CMT neuropathy Other polyneuropathies
19 9 4 3 2 6 4
Systemic disorders Disorders of OXPHOS Unclassified systemic diseases
68 14 54
Total number
134
MD, muscular dystrophy; SMA, spinal muscular atrophy; CMT, Charcot-Marie-Tooth; OXPHOS, oxidative phosphorylation.
Unspecific abnormal changes show a mixed pattern of myogenic and neurogenic changes or larger, more confluent echo hyperdensities (Fig. 1D). Subsequently to the muscle ultrasound examination, clinical and laboratory findings in all patients included in this study indicated the need for a histologic or molecular-genetic investigation. For the purpose of this study, muscle biopsies of all patients were reviewed by a single, experienced examiner (G.S.). Routine staining as well as immunohistochemical studies were taken into account when appropriate. This histological re-evaluation was performed blind as to the ultrasound findings. Muscle biopsies were classified into the four groups mentioned above (normal, myopathic, neurogenic, unspecifically abnormal). In four patients molecular-genetic investigations resulted in classification of their disorder as SMA, Duchenne muscular dystrophy (DMD) or Charcot-Marie-Tooth (CMT) neuropathy without muscle biopsy being performed. Table 1 displays the diagnoses and number of patients. This design ensured that both the sonologist and the pathologist were blinded as to the results of the others investigators assessment. Of course the physician/ sonologist who performed muscle ultrasound was aware of the patient’s history and clinical features, as it is the case in any ultrasound as well as EMG examination. Sensitivity, specificity, and accuracy (mean of sensitivity and specificity) of muscle ultrasound for detecting children with normal muscle or myogenic, neurogenic, or unspecifically abnormal tissue changes were calculated for each diagnosis in comparison to all
3. Results Tables 2 and 3 show the distribution of sonographic and histologic (or genetic) diagnoses in our 134 patients. Sensitivity, specificity, and accuracy of muscle ultrasound for discrimination of different groups of muscle tissue alterations are displayed in Table 4 for all patients. Highest accuracy was found for neurogenic changes. Among nine patients with spinal muscular atrophy type I, II, or III neurogenic changes were detected correctly by muscle ultrasound in eight patients. A single incorrect sonography result was obtained in a five-weeks-old infant with SMA type I, in whom ultrasound revealed normal appearing muscle. In all six children with CMT neuropathy (three with demyelinating, two with axonal, and one with congenital hypomyelinating neuropathy), neurogenic alterations were discerned accurately by muscle ultrasound. Muscular dystrophies including DMD (2 patients), merosin-deficient and merosin-positive congenital Table 2 Distribution of ultrasound findings Category
Male
Female
Normal Myopathic Neurogenic Unspecific abnormal
25 27 15 12
22 13 10 10
Total number 47 40 25 22
Total number
79
55
134
Table 3 Distribution of definitive histologic or genetic results Category
Male
Female
Total number
Normal Myopathic Neurogenic Unspecific abnormal
13 31 20 15
13 20 10 12
26 51 30 27
Total number
79
55
134
Table 4 Sensitivity, specificity, and accuracy of muscle ultrasound in myopathic, neurogenic, and unspecifically abnormal muscle tissue alterations in all 134 patients Comparison
Sensitivity (%)
Specificity (%)
Accuracy (%)
Abnormal vs. normal Myopathic vs. all others Neurogenic vs. all others Unspecific abnormal vs. all others
81 67 77 48
96 92 98 92
88 79 87 70
K. Brockmann et al. / Neuromuscular Disorders 17 (2007) 517–523 Table 5 Sensitivity, specificity, and accuracy of muscle ultrasound in myopathic, neurogenic, and unspecifically abnormal muscle tissue alterations for 50 patients aged 36 months or less Comparison
Sensitivity (%)
Specificity (%)
Accuracy (%)
Abnormal vs. normal Myopathic vs. all others Neurogenic vs. all others Unspecific abnormal vs. all others
71 67 55 42
92 89 100 92
81 78 77 67
muscular dystrophies (14 patients), and limb-girdle muscular dystrophy (1 patient), were correctly assigned to myogenic disorders in 15 of 17 patients. The only false negative results were a normal sonographic finding in a three-weeks-old newborn with congenital laminin a2-deficient muscular dystrophy (MDC1A) and a normal ultrasound judgement in an 11-years-old boy with unclassified muscular dystrophy. Three of four children with dermatomyositis were accurately detected to have myogenic changes. In a six-year-old girl with clinically remitting dermatomyositis, muscle ultrasound was judged to be normal, whereas histology revealed mild myopathic changes without any signs of inflammation. In general, accuracy of muscle ultrasound was higher in patients older than 3 years. In patients less than 3 years of age we found an accuracy of 81% for discrimination of abnormal vs. normal muscle. Details are given in Table 5. In the subgroup of infants less than 1 year old sonography differentiated between abnormal and normal with an accuracy of 63%. Among the 13 patients with unspecific abnormal findings in both, MUS and histology, the final clinical diagnosis was unclassified encephalopathy in 5, a disorder of oxidative phosphorylation (OXPHOS) in 4, an unclassified leukoencephalopathy in 2, aromatic L-amino acid decarboxylase (AADC) deficiency in 1, and oligosymptomatic triple-A syndrome in 1 patient. Accuracy of muscle ultrasound was relatively low in the subgroup of 14 patients with disorders of OXPHOS. Assessment of tissue alterations by sonography was correct in only 8 patients, including one each with normal and myogenic features, 2 with neurogenic and 4 with unspecific abnormal findings. 4. Discussion We found an overall accuracy of 88% for detection of any abnormal muscle tissue alteration by muscle ultrasound. Accuracy was highest in detection of neurogenic changes, slightly lower for myopathies and clearly lower for unspecific abnormal tissue alterations. Especially high accuracy was found in detection of muscular dystrophies, likely reflecting the more pronounced histopathological changes in that group of conditions.
521
Previous studies aiming at differentiation between normal and pathologic muscle or between myopathies and neuropathies were largely based on quantification of parameters including muscle thickness, density, inhomogeneity, or number of perimysial septa [5,13, 14,18,21,22]. However, several authors stressed the differences of muscle texture patterns observed in myopathies vs. neurogenic conditions or inflammatory muscle disorders [4,12,13]. Evaluation and classification of sonographic findings in our study presented here purposefully did not include quantification of ultrasound parameters, but was based on pattern recognition of different muscle texture changes. A qualitative pattern recognition approach constitutes an established method for diagnostic evaluation not only in other fields of medical imaging [23], but in clinical medicine in general. For example, in clinical dysmorphology, most syndrome diagnoses are suggested instantaneously by the gestalt of the patient, or ‘‘instinctive recognition of features based on past experience’’ [24]. In this approach, reliability of assessment relates to the experience of the investigator. Such an approach with its resulting instant assessment also more accurately reflects the clinical situation in which muscle ultrasound is used as an extension of the clinical neuromuscular examination. Neurogenic disorders produce a characteristic pattern with increased inhomogeneity and juxtaposition of hypo- and hyper-echoic areas, reflecting the grouping of fibre types to atrophic fibres (hyper-echoic, bright areas) and hypertrophic fibres (hypo-echoic, black areas). In addition, fasciculations are clearly observable in many cases. Muscular dystrophies are characterized by an extensive homogeneous echo hyperdensity, resulting in loss of visibility for bone and fascial echoes. These patterns are readily recognizable and hard to overlook, unless muscle ultrasound is performed at a very early stage of the disease, when histological changes are less pronounced. On the other hand, some myopathies produce only subtle tissue changes and are therefore harder to appreciate on ultrasound. The same is true for the less specific mixtures of neurogenic and myogenic appearing alterations which may be difficult to appreciate in either histology and sonography. The ultrasound system used in this study included 5 and 7.5 MHz linear probes. Recent advancements in ultrasound technology have led to the development of probes for imaging at 10 or 12 MHz, thus providing even higher spatial resolution and therefore better visualization of tissue changes. In general, alterations observed on muscle ultrasound are not specific for a single NMD. An exception from this rule is the highly characteristic ‘‘central shadow’’ pattern observed in many, but not all cases of Bethlem myopathy [25]. This feature probably reflects that the myopathic process in Bethlem myopathy is proceeding in an unusual outside-in fashion, from the fascia
522
K. Brockmann et al. / Neuromuscular Disorders 17 (2007) 517–523
inwards. In all other NMD, ultrasound only allows for assignment of muscle tissue changes to a group of disorders. However, sonography may provide valuable information regarding the pattern of muscle involvement or selection of biopsy site. Results for accuracy of muscle ultrasound were clearly less favorable for patients younger than three years, and especially for infants during their first year of life. This confirms findings by other authors [11] and may in part be explained by the lack of substantial structural change or connective tissue and fat proliferation at this early stage of disease [3] as well as by the small size of the muscle and its fibers. Our results show that sonography is not reliable in patients with OXPHOS disorders. This confirms previous findings indicating that muscle ultrasound is not suitable for screening purposes in children with dysfunction in the OXPHOS system [26]. In most clinical units, EMG constitutes the preferred method for detection of NMD and for discrimination of myopathies vs. neurogenic conditions. Only a number of studies have been performed to determine the reliability of EMG in assessment of NMD in children. Most of these investigations are confined to hypotonic infants, some included nerve conduction studies (NCS). In 51 hypotonic infants with suspected NMD, the initial EMG findings were compared with their final clinical diagnoses determined by independent means [27]. The EMG predicted the final diagnosis in 82% of infants less than 4 months of age and 85% of those over 4 months of age. The EMG diagnosis was identical to that obtained by muscle biopsy in 64% of cases in which biopsy was done, and diagnoses obtained by the two methods were inconsistent in 14%. In a retrospective study of 79 hypotonic infants, EMG predicted the final diagnosis in 65% of infants with SMA. In contrast, EMG predicted the final diagnosis in only 10% of infants with myopathy and was normal in 88% of infants with central hypotonia [28]. Data of EMG and NCS were correlated with muscle and nerve biopsies in a study of 41 floppy infants [29]. A positive correlation rate was found between electrophysiological studies and biopsy results in Werdnig–Hoffmann disease (14 of 15, 93%) and congenital infantile polyneuropathy (3/3, 100%). However, only 40% infants with biopsy proven myopathy had an abnormal EMG. In evaluation of 41 floppy infants aged 2–24 months a high concordance (22/24) of ultrasound and EMG findings was reported [30]. The diagnostic value of EMG and NCS was investigated in a retrospective study of 498 pediatric patients with suspected abnormality in the peripheral motor unit or sensory neuron [31]. The overall consistency between EMG results and the final clinical diagnosis in all children examined was 98%. This surprisingly high positive correlation may partially be
explained by the fact that 195 of these patients had generalized or circumscript peripheral neuropathies that were detected largely by NCS. Furthermore, normal electrodiagnostic findings in asymptomatic family members were included in the study as well. In myopathies, the concordance between EMG and clinical findings was lower (80%) [31]. Our results thus provide evidence that the reliability of qualitative muscle ultrasound is at least as high as or even higher than that of EMG in evaluation of NMD in childhood. There is some consensus now that muscle ultrasound appears to be more sensitive in the diagnosis of congenital muscular dystrophies in particular. Clearly, muscle sonography does not substitute for NCS, as both methods address different questions. In clinical suspected generalized or circumscript peripheral neuropathies NCS constitute the diagnostic procedure of first choice. Conversely a neurogenic pattern detected on muscle ultrasound will likely lead to NCS for further evaluation of the process. Furthermore, certain myopathies including myotonias and disorders of neuromuscular transmission will only be properly diagnosed by EMG. However, in a child with more uncharacteristic clinical features including muscular hypotonia, weakness or motor retardation, the diagnostic value of sonography is at least as high as that of EMG. Muscle ultrasound is easy to perform and readily accessible as it is a bedside tool that can be used almost as part of the clinical examination. It is much quicker than EMG, comparable cheap, and readily reproducible, even when a relatively rapid qualitative assessment is used without additional quantification of size and echogenicity of the muscle. Furthermore, a child undergoing sonography will experience clearly less discomfort than in EMG. Therefore we advocate a more widespread use of muscle ultrasound in evaluation of suspected NMD in childhood.
References [1] Heckmatt JZ, Dubowitz V, Leeman S. Detection of pathological change in dystrophic muscle with B-scan ultrasound imaging. Lancet 1980;1(8183):1389–90. [2] Young A, Hughes I, Russell P, Parkers MJ, Nichols PJ. Measurement of quadriceps muscle wasting by ultrasonography. Rheumatol Rehabil 1980;19:141–8. [3] Heckmatt JZ, Leeman S, Dubowitz V. Ultrasound imaging in the diagnosis of muscle disease. J Pediatr 1982;101:656–60. [4] Heckmatt JZ, Pier N, Dubowitz V. Real-time ultrasound imaging of muscles. Muscle Nerve 1988;11:56–65. [5] Maurits NM, Beenakker EA, van Schaik DE, Fock JM, van der Hoeven JH. Muscle ultrasound in children: normal values and application to neuromuscular disorders. Ultrasound Med Biol 2004;30:1017–27. [6] Scott DL, Kingsley GH. Use of imaging to assess patients with muscle disease. Curr Opin Rheumatol 2004;16:678–83.
K. Brockmann et al. / Neuromuscular Disorders 17 (2007) 517–523 [7] Walker FO, Cartwright MS, Wiesler ER, Caress J. Ultrasound of nerve and muscle. Clin Neurophysiol 2004;115:495–507. [8] Emery AE. The muscular dystrophies. Lancet 2002;359:687–95. [9] Reimers CD, Fleckenstein JL, Witt TN, Muller-Felber W, Pongratz DE. Muscular ultrasound in idiopathic inflammatory myopathies of adults. J Neurol Sci 1993;116:82–92. [10] Reimers K, Reimers CD, Wagner S, Paetzke I, Pongratz DE. Skeletal muscle sonography: a correlative study of echogenicity and morphology. J Ultrasound Med 1993;12:73–7. [11] Zuberi SM, Matta N, Nawaz S, Stephenson JB, McWilliam RC, Hollman A. Muscle ultrasound in the assessment of suspected neuromuscular disease in childhood. Neuromuscul Disord 1999;9:203–7. [12] Pohle R, Fischer D, von Rohden L. Computer-supported tissue characterization in ultrasound images of neuromuscular diseases. Ultraschall Med 2000;21:245–52 (in German). [13] Maurits NM, Bollen AE, Windhausen vA, De Jager AE, Van Der Hoeven JH. Muscle ultrasound analysis: normal values and differentiation between myopathies and neuropathies. Ultrasound Med Biol 2003;29:215–25. [14] Pillen S, Scholten RR, Zwarts MJ, Verrips A. Quantitative skeletal muscle ultrasonography in children with suspected neuromuscular disease. Muscle Nerve 2003;27:699–705. [15] Pillen S, van Keimpema M, Nievelstein RA, et al. Skeletal muscle ultrasonography: visual versus quantitative evaluation. Ultrasound Med Biol 2006;32:1315–21. [16] van Baalen A, Stephani U. Muscle fibre type grouping in high resolution ultrasound. Arch Dis Child 2005;90:1189. [17] von Rohden L, Krebs P, Steinbicker V, Wiemann D, Walter A. Ultrasound scanning in differential diagnosis of neuromuscular diseases. J Neurol Sci 1990;98(Suppl.):79–80. [18] Schmidt R, Voit T. Ultrasound measurement of quadriceps muscle in the first year of life. Normal values and application to spinal muscular atrophy. Neuropediatrics 1993;24:36–42. [19] Scheel AK, Toepfer M, Kunkel M, Finkenstaedt M, Reimers CD. Ultrasonographic assessment of the prevalence of fasciculations in
[20]
[21]
[22]
[23]
[24] [25] [26]
[27]
[28]
[29]
[30]
[31]
523
lesions of the peripheral nervous system. J Neuroimaging 1997;7:23–7. Wenzel S, Herrendorf G, Scheel A, Kurth C, Steinhoff BJ, Reimers CD. Surface EMG and myosonography in the detection of fasciculations: a comparative study. J Neuroimaging 1998;8:148–54. Dock W, Happak W, Grabenwoger F, Toifl K, Bittner R, Gruber H. Neuromuscular diseases: evaluation with high-frequency sonography. Radiology 1990;177:825–8. Scholten RR, Pillen S, Verrips A, Zwarts MJ. Quantitative ultrasonography of skeletal muscles in children: normal values. Muscle Nerve 2003;27:693–8. van der Knaap MS, Valk J, de Neeling N, Nauta JJ. Pattern recognition in magnetic resonance imaging of white matter disorders in children and young adults. Neuroradiology 1991;33:478–93. Smithson SF, Winter RM. Diagnosis in dysmorphology: clues from the skin. Br J Dermatol 2004;151:953–60. Bo¨nnemann CG, Brockmann K, Hanefeld F. Muscle ultrasound in Bethlem myopathy. Neuropediatrics 2003;34:335–6. Pillen S, Morava E, Van Keimpema M, et al. Skeletal muscle ultrasonography in children with a dysfunction in the oxidative phosphorylation system. Neuropediatrics 2006;37:142–7. Packer RJ, Brown MJ, Berman PH. The diagnostic value of electromyography in infantile hypotonia. Am J Dis Child 1982;136:1057–9. Russell JW, Afifi AK, Ross MA. Predictive value of electromyography in diagnosis and prognosis of the hypotonic infant. J Child Neurol 1992;7:387–91. David WS, Jones Jr HR. Electromyography and biopsy correlation with suggested protocol for evaluation of the floppy infant. Muscle Nerve 1994;17:424–30. Aydinli N, Baslo B, Caliskan M, Ertas M, Ozmen M. Muscle ultrasonography and electromyography correlation for evaluation of floppy infants. Brain Dev 2003;25:22–4. Hellmann M, von Kleist-Retzow JC, Haupt WF, Herkenrath P, Schauseil-Zipf U. Diagnostic value of electromyography in children and adolescents. J Clin Neurophysiol 2005;22:43–8.
Sensitivity and specificity of qualitative muscle ultrasound in assessment of suspected neuromuscular disease in childhood Knut Brockmann
a,*
, Peter Becker a, Gudrun Schreiber a, Karin Neubert b, Edgar Brunner b, Carsten Bo¨nnemann c
a
c
Department of Pediatrics and Pediatric Neurology, Georg August University, Go¨ttingen, Germany b Department of Medical Statistics, Georg August University, Go¨ttingen, Germany Division of Neurology, The Children’s Hospital of Philadelphia, University of Pennsylvania School of Medicine, Philadelphia, USA Received 3 January 2007; received in revised form 14 March 2007; accepted 27 March 2007
We dedicate this work to Professor Folker Hanefeld on the occasion of his 70th birthday.
Abstract Muscle ultrasound is considered a useful noninvasive technique for visualizing normal and pathological skeletal muscle. We determined the accuracy of qualitative muscle ultrasound in the discrimination of normal muscle from myopathic, neurogenic, and unspecifically abnormal tissue changes in the evaluation of suspected NMD in childhood. Sensitivity and specificity of muscle ultrasound were assessed by comparing sonographic classification of muscle tissue changes in 134 children with definitive diagnosis as provided by muscle histology or mutation analysis performed subsequently to the sonography. We found a sensitivity of 81% and a specificity of 96% for detection of any abnormal muscle tissue alteration by ultrasound. For detection of neurogenic changes, sensitivity was 77% with even higher specificity (98%). Accuracy was slightly lower for myopathic changes (79%) and clearly lower for unspecific abnormal tissue alterations (70%). Accuracy of ultrasound was lower in younger children. High reliability of muscle sonography justifies a more widespread use of this method in evaluation of suspected NMD in childhood. Ó 2007 Elsevier B.V. All rights reserved. Keywords: Muscle ultrasound; Neuromuscular disorder; Childhood; Sensitivity; Specificity
1. Introduction Ultrasound imaging of the muscle was introduced in the field of neuromuscular disorders (NMD) in 1980 [1,2] and is now considered a useful technique for visualizing normal and pathological skeletal muscle [3–7]. Though ultrasound is successfully used in a
Abbreviations: CMT, Charcot-Marie-Tooth; DMD, Duchenne muscular dystrophy; MD, muscular dystrophy; NMD, neuromuscular disease; OXPHOS, oxidative phosphorylation; SMA, spinal muscular atrophy. * Corresponding author. Tel.: +49 551 39 6210; fax +49 551 39 6252. E-mail address: [email protected] (K. Brockmann). 0960-8966/$ - see front matter Ó 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.nmd.2007.03.015
number of centers it is not widely adopted in the evaluation of neuromuscular disease. Many neurologists still adhere to electromyography (EMG) as the preferred method for discrimination of myopathies vs. neurogenic conditions. A recent review on muscular dystrophies did not mention muscle ultrasound among the laboratory methods for assessment of NMD, which included serum creatine kinase, EMG, muscle histology, immunohistochemistry, and mutation analysis [8]. Nevertheless the use of muscle ultrasound is advantageous especially in children, as it is a painless, quick, relatively cheap, and readily reproducible investigation. Only a limited number of studies have aimed at determining the reliability of muscle ultrasound in
518
K. Brockmann et al. / Neuromuscular Disorders 17 (2007) 517–523
NMD. In adults with idiopathic inflammatory myopathies a robust correlation between histologic and sonographic changes has been reported [9,10]. In a pediatric study including 100 children aged 2 months to 16 years the usefulness of muscle ultrasound in the initial assessment of suspected NMD was investigated. Sensitivity and specificity were 78% and 91% for detection of any type of NMD [11]. In the subgroup of patients older than 3 years muscle ultrasound was even more reliable with a sensitivity of 81% and a specificity of 96%. Few studies aimed at differentiating between myopathies and neuropathies on the basis of muscle ultrasound alone. In a mixed population of children and adults with histologically or genetically confirmed NMD, a combination of conventional muscle ultrasound and computer-assisted texture analysis of sonograms was used. Values for sensitivity from 77% to 94% and specificity from 81% to 98% were found for discrimination between myogenic and neurogenic disorders [12]. Quantitative muscle ultrasound has been successfully used to differentiate between typical myopathies and neuropathies in adults [13]. However, in children quantitative muscle ultrasound was shown to be sensitive for distinguishing between healthy children and children with neuromuscular disorders in general [5] or between neuromuscular and non-neuromuscular diseases, e.g. cerebral palsy [14], but did not allow for a further distinction. A more detailed aspect analysis was presumed to be necessary for further distinguishing between different types of neuromuscular disorders [14]. Recently, a study compared the sensitivity and specificity of visual vs. quantitative evaluation of skeletal muscle ultrasound in children suspected of having a NMD [15]. Quantification of echo intensity was found a more objective and accurate method compared to visual evaluation for screening for NMD in general. However, a distinction between myopathies and neurogenic disorders was not achieved with this approach. Here we report the first study that seeks to determine the reliability of qualitative muscle ultrasound not only for the discrimination of normal and abnormal muscle, but additionally for the distinction between myopathic and neurogenic disorders in the evaluation of suspected NMD in childhood. Sensitivity and specificity of muscle ultrasound were arrived at by comparing sonographic classification of muscle tissue changes with the definitive diagnosis provided by histological or genetic analysis (gold standard). Although our data were collected retrospectively, both the sonologist and the pathologist were blinded as to the corresponding result.
2. Patients and methods In a retrospective review of data in our ultrasound laboratory we identified all patients (n = 826) that had been examined by muscle ultrasound as part of evaluation for suspected NMD during a five-year-period (1997–2001) in our Department of Pediatrics and Pediatric Neurology in a tertiary care University hospital. Only patients fulfilling the following criteria were included in the study: (i) examination by a single sonologist (K.B.) in order to standardize the sonologic criteria (n = 525); (ii) availability of a definitive, final diagnosis provided by muscle histology or molecular-genetic analysis (gold standard) (n = 184); (iii) performance of muscle ultrasound before muscle biopsy or molecular-genetic investigation were completed so that the ultrasound report was generated without any knowledge of the final diagnosis. These criteria were fulfilled by 134 patients (79 boys, aged 21 days to 17 years, mean 5 years 5 months; 55 girls, aged 18 days to 18 years, mean 6 years 4 months), whose ages ranged from 18 days to 18 years (mean 5 years 10 months). Fifty children were younger than 3 years, and of these 14 were younger than 12 months. Ultrasound scans were performed according to a standard protocol. A Siemens Sonoline Prima ultrasound system with 5 and 7.5 MHz linear probes was used. The 5 MHz probe with its greater depth of penetration was used for an overview of thigh and calf in order to detect selective affection of muscles. The 7.5 MHz probe provided depiction of tissue alterations in greater detail and thus allowed for differentiation between neurogenic and myogenic changes [16,17]. A series of standard images of the various muscles in healthy children of various ages had been acquired by the same ultrasonographer on the same machine and served as the normal reference set. Gain and setting parameters were adjusted so that healthy muscle appeared nearly black (Fig. 1A). This setting was conserved in the ultrasound system as a fixed investigation program and not changed during the patients’ investigations, apart from adjustment of the depth setting. Axial scans of mid thigh and upper calf were recorded in all patients in the lying position. The probe was moved along the thigh or calf until the optimal view most closely corresponding to the established standard image for the segment in question was found. Additional scans of upper arm or deltoid muscle were performed when clinically appropriate. Muscle ultrasound findings in muscle disease have been classified on a four-point scale by Heckmatt et al. [3] according to the intensity of echo reflected from the muscle. Heckmatt’s criteria constitute a ‘‘straight-
K. Brockmann et al. / Neuromuscular Disorders 17 (2007) 517–523
519
Fig. 1. Axial muscle ultrasound scans of mid thigh of (A) a healthy 7-year-old girl, (B) a 6-year-old boy with Duchenne muscular dystrophy (DMD), (C) a 12-year-old boy with DMD, (D) a 14-year-old girl with axonal neuropathy, and (E) a 10-year-old boy with unclassified encephalopathy, global retardation, and muscular hypotonia. In a healthy subject (A), muscle appears largely black with few perimysial septa. In myopathies (B), MUS reveals elevated homogeneous fine granular echogenicity. More advanced stages of muscular dystrophy (C) are characterized by marked homogeneous elevation of muscle echogenicity and loss of bone echo. In neurogenic conditions (D), increased echogenicity is accompanied by elevated inhomogeneity. Unspecific abnormal patterns (E) are characterized by elevated echogenicity without clear assignment to myopathic or neurogenic features.
forward and reproducible grading system’’ [11] and provide a semi-quantification of tissue changes. However, no information on the quality of muscle alterations observed is given by these criteria, and they do not help in differentiating between myopathies and neurogenic conditions. From the clinician’s point of view, the major questions posed to the sonography in evaluation of a child with suspected NMD comprise (i) whether the muscle is normal or abnormal, and (ii), in case of muscle abnormality, whether these changes are myogenic or neurogenic. Therefore, in this study muscle ultrasound findings were not classified according to Heckmatt’s criteria, but rather judged as being either (i) normal, (ii) myopathic, (iii) neurogenic, or (iv) unspecifically abnormal. Assignment of ultrasonographic findings to one of the categories mentioned above was based on visual assessment and the pattern recognition approach. Muscle thickness and ratios of muscle thickness to subcutaneous fat thickness were taken into account, but not quantified [18]. In NMD, the increase of fat, fibrosis, and inflammation results in an elevation of the numbers
of reflective interfaces in muscle [7]. Myogenic changes are characterized by a largely homogeneous fine granular elevation of echogenicity within the muscle (Fig. 1B). These alterations are most extensive in muscular dystrophies, resulting in a homogeneous high echogenicity of muscle tissue with loss of visualization of epimysium and bone [4]. Neurogenic alterations depict as inhomogeneous, patchy and stripy echo densities together with interposed hypodense areas (Fig. 1C). Atrophic fibres correspond to hyper-echoic, bright areas, whereas normal and hypertrophic fibres are visualized as hypoechoic, black areas. Recognition of this juxtaposition of fibre types is facilitated by high resolution ultrasound [16]. Fasciculations are readily observable in many cases of neurogenic disease, especially in spinal muscular atrophy (SMA) and axonal forms of CMT neuropathy [19,20]. In longstanding NMD, the differences between myogenic and neurogenic ultrasound patterns often become less distinct and are replaced by an undifferentiated picture with increased echogenicity resulting from loss of muscle mass and concentration of fibrous and fatty tissue in a smaller volume.
520
K. Brockmann et al. / Neuromuscular Disorders 17 (2007) 517–523
others in the whole group and for the age range up to 3 years.
Table 1 Diagnoses and number of patients Number of patients Myopathic disorders Duchenne MD Congenital MD Limb-girdle dystrophy Congenital myopathies Bethlem myopathy Metabolic myopathies Unclassified myopathies Dermatomyositis
47 2 14 1 5 1 3 17 4
Neurogenic disorders SMA total SMA type I SMA type II SMA type III CMT neuropathy Other polyneuropathies
19 9 4 3 2 6 4
Systemic disorders Disorders of OXPHOS Unclassified systemic diseases
68 14 54
Total number
134
MD, muscular dystrophy; SMA, spinal muscular atrophy; CMT, Charcot-Marie-Tooth; OXPHOS, oxidative phosphorylation.
Unspecific abnormal changes show a mixed pattern of myogenic and neurogenic changes or larger, more confluent echo hyperdensities (Fig. 1D). Subsequently to the muscle ultrasound examination, clinical and laboratory findings in all patients included in this study indicated the need for a histologic or molecular-genetic investigation. For the purpose of this study, muscle biopsies of all patients were reviewed by a single, experienced examiner (G.S.). Routine staining as well as immunohistochemical studies were taken into account when appropriate. This histological re-evaluation was performed blind as to the ultrasound findings. Muscle biopsies were classified into the four groups mentioned above (normal, myopathic, neurogenic, unspecifically abnormal). In four patients molecular-genetic investigations resulted in classification of their disorder as SMA, Duchenne muscular dystrophy (DMD) or Charcot-Marie-Tooth (CMT) neuropathy without muscle biopsy being performed. Table 1 displays the diagnoses and number of patients. This design ensured that both the sonologist and the pathologist were blinded as to the results of the others investigators assessment. Of course the physician/ sonologist who performed muscle ultrasound was aware of the patient’s history and clinical features, as it is the case in any ultrasound as well as EMG examination. Sensitivity, specificity, and accuracy (mean of sensitivity and specificity) of muscle ultrasound for detecting children with normal muscle or myogenic, neurogenic, or unspecifically abnormal tissue changes were calculated for each diagnosis in comparison to all
3. Results Tables 2 and 3 show the distribution of sonographic and histologic (or genetic) diagnoses in our 134 patients. Sensitivity, specificity, and accuracy of muscle ultrasound for discrimination of different groups of muscle tissue alterations are displayed in Table 4 for all patients. Highest accuracy was found for neurogenic changes. Among nine patients with spinal muscular atrophy type I, II, or III neurogenic changes were detected correctly by muscle ultrasound in eight patients. A single incorrect sonography result was obtained in a five-weeks-old infant with SMA type I, in whom ultrasound revealed normal appearing muscle. In all six children with CMT neuropathy (three with demyelinating, two with axonal, and one with congenital hypomyelinating neuropathy), neurogenic alterations were discerned accurately by muscle ultrasound. Muscular dystrophies including DMD (2 patients), merosin-deficient and merosin-positive congenital Table 2 Distribution of ultrasound findings Category
Male
Female
Normal Myopathic Neurogenic Unspecific abnormal
25 27 15 12
22 13 10 10
Total number 47 40 25 22
Total number
79
55
134
Table 3 Distribution of definitive histologic or genetic results Category
Male
Female
Total number
Normal Myopathic Neurogenic Unspecific abnormal
13 31 20 15
13 20 10 12
26 51 30 27
Total number
79
55
134
Table 4 Sensitivity, specificity, and accuracy of muscle ultrasound in myopathic, neurogenic, and unspecifically abnormal muscle tissue alterations in all 134 patients Comparison
Sensitivity (%)
Specificity (%)
Accuracy (%)
Abnormal vs. normal Myopathic vs. all others Neurogenic vs. all others Unspecific abnormal vs. all others
81 67 77 48
96 92 98 92
88 79 87 70
K. Brockmann et al. / Neuromuscular Disorders 17 (2007) 517–523 Table 5 Sensitivity, specificity, and accuracy of muscle ultrasound in myopathic, neurogenic, and unspecifically abnormal muscle tissue alterations for 50 patients aged 36 months or less Comparison
Sensitivity (%)
Specificity (%)
Accuracy (%)
Abnormal vs. normal Myopathic vs. all others Neurogenic vs. all others Unspecific abnormal vs. all others
71 67 55 42
92 89 100 92
81 78 77 67
muscular dystrophies (14 patients), and limb-girdle muscular dystrophy (1 patient), were correctly assigned to myogenic disorders in 15 of 17 patients. The only false negative results were a normal sonographic finding in a three-weeks-old newborn with congenital laminin a2-deficient muscular dystrophy (MDC1A) and a normal ultrasound judgement in an 11-years-old boy with unclassified muscular dystrophy. Three of four children with dermatomyositis were accurately detected to have myogenic changes. In a six-year-old girl with clinically remitting dermatomyositis, muscle ultrasound was judged to be normal, whereas histology revealed mild myopathic changes without any signs of inflammation. In general, accuracy of muscle ultrasound was higher in patients older than 3 years. In patients less than 3 years of age we found an accuracy of 81% for discrimination of abnormal vs. normal muscle. Details are given in Table 5. In the subgroup of infants less than 1 year old sonography differentiated between abnormal and normal with an accuracy of 63%. Among the 13 patients with unspecific abnormal findings in both, MUS and histology, the final clinical diagnosis was unclassified encephalopathy in 5, a disorder of oxidative phosphorylation (OXPHOS) in 4, an unclassified leukoencephalopathy in 2, aromatic L-amino acid decarboxylase (AADC) deficiency in 1, and oligosymptomatic triple-A syndrome in 1 patient. Accuracy of muscle ultrasound was relatively low in the subgroup of 14 patients with disorders of OXPHOS. Assessment of tissue alterations by sonography was correct in only 8 patients, including one each with normal and myogenic features, 2 with neurogenic and 4 with unspecific abnormal findings. 4. Discussion We found an overall accuracy of 88% for detection of any abnormal muscle tissue alteration by muscle ultrasound. Accuracy was highest in detection of neurogenic changes, slightly lower for myopathies and clearly lower for unspecific abnormal tissue alterations. Especially high accuracy was found in detection of muscular dystrophies, likely reflecting the more pronounced histopathological changes in that group of conditions.
521
Previous studies aiming at differentiation between normal and pathologic muscle or between myopathies and neuropathies were largely based on quantification of parameters including muscle thickness, density, inhomogeneity, or number of perimysial septa [5,13, 14,18,21,22]. However, several authors stressed the differences of muscle texture patterns observed in myopathies vs. neurogenic conditions or inflammatory muscle disorders [4,12,13]. Evaluation and classification of sonographic findings in our study presented here purposefully did not include quantification of ultrasound parameters, but was based on pattern recognition of different muscle texture changes. A qualitative pattern recognition approach constitutes an established method for diagnostic evaluation not only in other fields of medical imaging [23], but in clinical medicine in general. For example, in clinical dysmorphology, most syndrome diagnoses are suggested instantaneously by the gestalt of the patient, or ‘‘instinctive recognition of features based on past experience’’ [24]. In this approach, reliability of assessment relates to the experience of the investigator. Such an approach with its resulting instant assessment also more accurately reflects the clinical situation in which muscle ultrasound is used as an extension of the clinical neuromuscular examination. Neurogenic disorders produce a characteristic pattern with increased inhomogeneity and juxtaposition of hypo- and hyper-echoic areas, reflecting the grouping of fibre types to atrophic fibres (hyper-echoic, bright areas) and hypertrophic fibres (hypo-echoic, black areas). In addition, fasciculations are clearly observable in many cases. Muscular dystrophies are characterized by an extensive homogeneous echo hyperdensity, resulting in loss of visibility for bone and fascial echoes. These patterns are readily recognizable and hard to overlook, unless muscle ultrasound is performed at a very early stage of the disease, when histological changes are less pronounced. On the other hand, some myopathies produce only subtle tissue changes and are therefore harder to appreciate on ultrasound. The same is true for the less specific mixtures of neurogenic and myogenic appearing alterations which may be difficult to appreciate in either histology and sonography. The ultrasound system used in this study included 5 and 7.5 MHz linear probes. Recent advancements in ultrasound technology have led to the development of probes for imaging at 10 or 12 MHz, thus providing even higher spatial resolution and therefore better visualization of tissue changes. In general, alterations observed on muscle ultrasound are not specific for a single NMD. An exception from this rule is the highly characteristic ‘‘central shadow’’ pattern observed in many, but not all cases of Bethlem myopathy [25]. This feature probably reflects that the myopathic process in Bethlem myopathy is proceeding in an unusual outside-in fashion, from the fascia
522
K. Brockmann et al. / Neuromuscular Disorders 17 (2007) 517–523
inwards. In all other NMD, ultrasound only allows for assignment of muscle tissue changes to a group of disorders. However, sonography may provide valuable information regarding the pattern of muscle involvement or selection of biopsy site. Results for accuracy of muscle ultrasound were clearly less favorable for patients younger than three years, and especially for infants during their first year of life. This confirms findings by other authors [11] and may in part be explained by the lack of substantial structural change or connective tissue and fat proliferation at this early stage of disease [3] as well as by the small size of the muscle and its fibers. Our results show that sonography is not reliable in patients with OXPHOS disorders. This confirms previous findings indicating that muscle ultrasound is not suitable for screening purposes in children with dysfunction in the OXPHOS system [26]. In most clinical units, EMG constitutes the preferred method for detection of NMD and for discrimination of myopathies vs. neurogenic conditions. Only a number of studies have been performed to determine the reliability of EMG in assessment of NMD in children. Most of these investigations are confined to hypotonic infants, some included nerve conduction studies (NCS). In 51 hypotonic infants with suspected NMD, the initial EMG findings were compared with their final clinical diagnoses determined by independent means [27]. The EMG predicted the final diagnosis in 82% of infants less than 4 months of age and 85% of those over 4 months of age. The EMG diagnosis was identical to that obtained by muscle biopsy in 64% of cases in which biopsy was done, and diagnoses obtained by the two methods were inconsistent in 14%. In a retrospective study of 79 hypotonic infants, EMG predicted the final diagnosis in 65% of infants with SMA. In contrast, EMG predicted the final diagnosis in only 10% of infants with myopathy and was normal in 88% of infants with central hypotonia [28]. Data of EMG and NCS were correlated with muscle and nerve biopsies in a study of 41 floppy infants [29]. A positive correlation rate was found between electrophysiological studies and biopsy results in Werdnig–Hoffmann disease (14 of 15, 93%) and congenital infantile polyneuropathy (3/3, 100%). However, only 40% infants with biopsy proven myopathy had an abnormal EMG. In evaluation of 41 floppy infants aged 2–24 months a high concordance (22/24) of ultrasound and EMG findings was reported [30]. The diagnostic value of EMG and NCS was investigated in a retrospective study of 498 pediatric patients with suspected abnormality in the peripheral motor unit or sensory neuron [31]. The overall consistency between EMG results and the final clinical diagnosis in all children examined was 98%. This surprisingly high positive correlation may partially be
explained by the fact that 195 of these patients had generalized or circumscript peripheral neuropathies that were detected largely by NCS. Furthermore, normal electrodiagnostic findings in asymptomatic family members were included in the study as well. In myopathies, the concordance between EMG and clinical findings was lower (80%) [31]. Our results thus provide evidence that the reliability of qualitative muscle ultrasound is at least as high as or even higher than that of EMG in evaluation of NMD in childhood. There is some consensus now that muscle ultrasound appears to be more sensitive in the diagnosis of congenital muscular dystrophies in particular. Clearly, muscle sonography does not substitute for NCS, as both methods address different questions. In clinical suspected generalized or circumscript peripheral neuropathies NCS constitute the diagnostic procedure of first choice. Conversely a neurogenic pattern detected on muscle ultrasound will likely lead to NCS for further evaluation of the process. Furthermore, certain myopathies including myotonias and disorders of neuromuscular transmission will only be properly diagnosed by EMG. However, in a child with more uncharacteristic clinical features including muscular hypotonia, weakness or motor retardation, the diagnostic value of sonography is at least as high as that of EMG. Muscle ultrasound is easy to perform and readily accessible as it is a bedside tool that can be used almost as part of the clinical examination. It is much quicker than EMG, comparable cheap, and readily reproducible, even when a relatively rapid qualitative assessment is used without additional quantification of size and echogenicity of the muscle. Furthermore, a child undergoing sonography will experience clearly less discomfort than in EMG. Therefore we advocate a more widespread use of muscle ultrasound in evaluation of suspected NMD in childhood.
References [1] Heckmatt JZ, Dubowitz V, Leeman S. Detection of pathological change in dystrophic muscle with B-scan ultrasound imaging. Lancet 1980;1(8183):1389–90. [2] Young A, Hughes I, Russell P, Parkers MJ, Nichols PJ. Measurement of quadriceps muscle wasting by ultrasonography. Rheumatol Rehabil 1980;19:141–8. [3] Heckmatt JZ, Leeman S, Dubowitz V. Ultrasound imaging in the diagnosis of muscle disease. J Pediatr 1982;101:656–60. [4] Heckmatt JZ, Pier N, Dubowitz V. Real-time ultrasound imaging of muscles. Muscle Nerve 1988;11:56–65. [5] Maurits NM, Beenakker EA, van Schaik DE, Fock JM, van der Hoeven JH. Muscle ultrasound in children: normal values and application to neuromuscular disorders. Ultrasound Med Biol 2004;30:1017–27. [6] Scott DL, Kingsley GH. Use of imaging to assess patients with muscle disease. Curr Opin Rheumatol 2004;16:678–83.
K. Brockmann et al. / Neuromuscular Disorders 17 (2007) 517–523 [7] Walker FO, Cartwright MS, Wiesler ER, Caress J. Ultrasound of nerve and muscle. Clin Neurophysiol 2004;115:495–507. [8] Emery AE. The muscular dystrophies. Lancet 2002;359:687–95. [9] Reimers CD, Fleckenstein JL, Witt TN, Muller-Felber W, Pongratz DE. Muscular ultrasound in idiopathic inflammatory myopathies of adults. J Neurol Sci 1993;116:82–92. [10] Reimers K, Reimers CD, Wagner S, Paetzke I, Pongratz DE. Skeletal muscle sonography: a correlative study of echogenicity and morphology. J Ultrasound Med 1993;12:73–7. [11] Zuberi SM, Matta N, Nawaz S, Stephenson JB, McWilliam RC, Hollman A. Muscle ultrasound in the assessment of suspected neuromuscular disease in childhood. Neuromuscul Disord 1999;9:203–7. [12] Pohle R, Fischer D, von Rohden L. Computer-supported tissue characterization in ultrasound images of neuromuscular diseases. Ultraschall Med 2000;21:245–52 (in German). [13] Maurits NM, Bollen AE, Windhausen vA, De Jager AE, Van Der Hoeven JH. Muscle ultrasound analysis: normal values and differentiation between myopathies and neuropathies. Ultrasound Med Biol 2003;29:215–25. [14] Pillen S, Scholten RR, Zwarts MJ, Verrips A. Quantitative skeletal muscle ultrasonography in children with suspected neuromuscular disease. Muscle Nerve 2003;27:699–705. [15] Pillen S, van Keimpema M, Nievelstein RA, et al. Skeletal muscle ultrasonography: visual versus quantitative evaluation. Ultrasound Med Biol 2006;32:1315–21. [16] van Baalen A, Stephani U. Muscle fibre type grouping in high resolution ultrasound. Arch Dis Child 2005;90:1189. [17] von Rohden L, Krebs P, Steinbicker V, Wiemann D, Walter A. Ultrasound scanning in differential diagnosis of neuromuscular diseases. J Neurol Sci 1990;98(Suppl.):79–80. [18] Schmidt R, Voit T. Ultrasound measurement of quadriceps muscle in the first year of life. Normal values and application to spinal muscular atrophy. Neuropediatrics 1993;24:36–42. [19] Scheel AK, Toepfer M, Kunkel M, Finkenstaedt M, Reimers CD. Ultrasonographic assessment of the prevalence of fasciculations in
[20]
[21]
[22]
[23]
[24] [25] [26]
[27]
[28]
[29]
[30]
[31]
523
lesions of the peripheral nervous system. J Neuroimaging 1997;7:23–7. Wenzel S, Herrendorf G, Scheel A, Kurth C, Steinhoff BJ, Reimers CD. Surface EMG and myosonography in the detection of fasciculations: a comparative study. J Neuroimaging 1998;8:148–54. Dock W, Happak W, Grabenwoger F, Toifl K, Bittner R, Gruber H. Neuromuscular diseases: evaluation with high-frequency sonography. Radiology 1990;177:825–8. Scholten RR, Pillen S, Verrips A, Zwarts MJ. Quantitative ultrasonography of skeletal muscles in children: normal values. Muscle Nerve 2003;27:693–8. van der Knaap MS, Valk J, de Neeling N, Nauta JJ. Pattern recognition in magnetic resonance imaging of white matter disorders in children and young adults. Neuroradiology 1991;33:478–93. Smithson SF, Winter RM. Diagnosis in dysmorphology: clues from the skin. Br J Dermatol 2004;151:953–60. Bo¨nnemann CG, Brockmann K, Hanefeld F. Muscle ultrasound in Bethlem myopathy. Neuropediatrics 2003;34:335–6. Pillen S, Morava E, Van Keimpema M, et al. Skeletal muscle ultrasonography in children with a dysfunction in the oxidative phosphorylation system. Neuropediatrics 2006;37:142–7. Packer RJ, Brown MJ, Berman PH. The diagnostic value of electromyography in infantile hypotonia. Am J Dis Child 1982;136:1057–9. Russell JW, Afifi AK, Ross MA. Predictive value of electromyography in diagnosis and prognosis of the hypotonic infant. J Child Neurol 1992;7:387–91. David WS, Jones Jr HR. Electromyography and biopsy correlation with suggested protocol for evaluation of the floppy infant. Muscle Nerve 1994;17:424–30. Aydinli N, Baslo B, Caliskan M, Ertas M, Ozmen M. Muscle ultrasonography and electromyography correlation for evaluation of floppy infants. Brain Dev 2003;25:22–4. Hellmann M, von Kleist-Retzow JC, Haupt WF, Herkenrath P, Schauseil-Zipf U. Diagnostic value of electromyography in children and adolescents. J Clin Neurophysiol 2005;22:43–8.