Predictability of the clinical course of juvenile dermatomyositis based on initial muscle biopsy: A retrospective study of 72 patients




Except when the diagnosis of juvenile dermatomyositis (DM) is in doubt, a case has not been made for routine muscle biopsy (MB). We sought to determine whether MB findings prior to systemic therapy have prognostic value.


We reviewed the hospital records and slides prepared from the initial open MB of 72 patients treated at one center between 1977 and 2002 and followed for a minimum of 2 years. None of the patients had received a course of systemic corticosteroid therapy at the time of MB. Our approach to MB evaluation was based on recent discussions with muscle pathology experts to develop criteria for assessing inflammation, vasculopathy, myofiber atrophy, regeneration, acute and chronic myopathic change, and stromal changes. Using simple and multivariate logistic regression, we tested each MB parameter for ability to predict outcome using 2 published classification systems.


Extensive active myopathic changes (excluding regeneration) and central nuclei without basophilia predicted chronic juvenile DM. Severe arteropathic change, positive arterial direct immunofluorescence, obvious foci of severe capillary loss/endomysial fibrosis, and muscle infarcts predicted chronic juvenile DM, particularly with ulceration. Other MB parameters, regardless of severity, were not significant predictors of chronic juvenile DM versus limited disease.


A scoring system for evaluating pretreatment MB in juvenile DM that focuses on extent of necrotizing myopathy, severity of vasculopathy, and features of established chronicity such as central nucleation of nonbasophilic myofibers may provide a basis for stratification of therapeutic regimens according to risk for chronic disease. The validity of our findings should be prospectively tested.


Juvenile dermatomyositis (DM) commonly involves skin in diagnostic patterns, skeletal muscle, and occasionally the gastrointestinal tract or other organs (1, 2). Although the morbidity and mortality of juvenile DM improved after introduction of corticosteroid therapy, the long-term outcome of treated juvenile DM differs substantially from patient to patient (3). In 1982, Crowe et al (4) defined 3 subsets of juvenile DM: treatment-responsive disease of limited duration, and chronic disease either with or without cutaneous ulceration. Chronic ulcerative juvenile DM had a high incidence of gastrointestinal bleeding and a worse prognosis. Crowe et al suggested that arteriopathic features in muscle biopsy (MB) might predict this form of the disease (4). Two years later, Spencer et al proposed 3 different courses of juvenile DM: limited treatment-responsive disease, chronic polycyclic disease, and chronic unicyclic or continuous disease (5). Their classification did not recognize ulcerative juvenile DM as a subtype.

MB is a useful but underutilized tool for characterizing the pathogenetic changes in juvenile DM, because the diagnosis in many patients may seem clear on clinical grounds. The clinical impression may find additional support in electromyographic and soft tissue magnetic resonance imaging (MRI) findings (6, 7), but MRI alone is more useful for selection of a biopsy site than for primary diagnosis (8). Crowe et al (4) suggested that MB findings might predict chronic ulcerative juvenile DM, but the number of cases in their study was too small to be definitive. Wargula et al (9) and Bukulmez et al (10), in their recent studies of the juvenile DM patient population at Cincinnati Children's Hospital, offered additional support for the predictive value of MB. We hypothesized that a rational basis for therapy could be based on MB if muscle histopathology has predictive value for limited versus chronic disease as a whole or for particular subsets of chronic disease. The objective of this study was to retrospectively assess the correlation of MB findings with clinical outcome based on the Crowe et al (4) or Spencer et al (5) classifications of patients with juvenile DM whose outcomes were known. This study was a more comprehensive morphologic evaluation of archived MB specimens in a larger series of patients than reported previously from Cincinnati Children's Hospital Medical Center (9).


Seventy-two patients (46 girls, 26 boys, mean age 8.1 years) with the clinical diagnosis of juvenile DM who underwent MB at Cincinnati Children's Hospital between 1977 and 2002 were included in this study. Inclusion criteria have been previously reported (8). Patients who had received a course of corticosteroid therapy prior to MB were excluded. The duration of signs and symptoms attributable to juvenile DM prior to biopsy, height, and weight with percentiles at the time of the initial biopsy were extracted from the clinical records of 67, 61, and 64 patients, respectively. MB site selection was based on muscle weakness, and was guided by electromyogram or soft tissue MRI. Using the Crowe and Spencer classifications, the clinical course after treatment and close followup at this institution was characterized in 72 and 67 patients, respectively. In both classification systems, 0 designates limited treatment-responsive disease and 1 or 2 designates subsets of chronic disease. The clinical course was unknown to the review pathologists. Details of processing MB specimens during the study period have been previously reported (9). Cryostat sections stained with hematoxylin and eosin, Gomori's modified trichrome, and myosin ATPase with preincubation at pH 4.6 and 10 were available in departmental files for concurrent review by 2 experienced muscle pathologists (KEB and LM), one of whom (KEB) was the pathologist-of-record for all MB specimens during the study period. Positive direct immunofluorescence (DIF) results in cryostat sections of muscle were available in 66 patients, in whom we had utilized antibodies to IgG, IgM, IgA, C3a, C3b, C5, and fibrinogen. DIF data from original worksheets, MB reports, and photomicrographs of positive reactions were tabulated. Archived slides from each muscle sample were evaluated for pattern and intensity of inflammation, vascular changes, myofiber atrophy, myofiber degeneration and necrosis, myofiber regeneration, central nuclear position without sarcoplasmic basophilia, and stromal changes using criteria that were strongly influenced by discussions carried out in the International Working Group for Biopsy Scoring in Juvenile Dermatomyositis (11, 12), but that differ by including more grades of severity.


Inflammation was graded as follows: 1 = small foci of 4–50 inflammatory cells, 2 = >2 foci containing >50 inflammatory cells, and 3 = large clusters of >200 inflammatory cells. The distribution of inflammation was categorized into perimysial, perivascular, and endomysial. Subtypes of inflammatory cells were not specified because immunostains for this purpose had not been performed.

Vascular changes.

Capillary loss was judged to be present only when a substantial part of a fascicle, typically in the periphery, exhibited endomysial stromal expansion and obvious depletion of capillaries. No immunostains for endothelial reactants were available. Capillary loss was graded based on the number of foci of obvious zonal depletion: 0 = no zones of capillary dropout, 1 = 1 focus, 2 = 2 foci, and 3 = ≥3 foci. Intrafascicular arteriolopathy and perifascicular (PF) arteropathy were graded 0–2 based on the severity of endothelial cell and vascular wall changes and estimated degree of lumen reduction. Capillary and arterial DIF were each categorized as present or absent.

Myofiber atrophy.

PF atrophy was defined as present when more than 20 myofibers with a minimum of 2 layers of myofibers were atrophic compared with myofibers in the center of the same fascicle. PF atrophy was graded as follows: 1 = 1 focus, 2 = multiple foci, 3 = back-to-back foci involving contiguous fascicles. Type II myoatrophy and random myofiber atrophy within individual fascicles were categorized as 0, 1, and 2 based on prevalence.

Active myopathic change.

Included in this category were myofibers with internal myofibrillar disorganization, pallor, vacuolar degeneration, macrophage invasion, and frank necrosis. The prevalence (extent) of myopathic change was rated 0–3, corresponding to absent, mild, moderate, and extensive. Extensive was defined as necrosis involving aggregates of at least 3–5 myofibers in ≥2 fascicles. For statistical purposes, muscle infarcts, defined as demarcated coagulation necrosis, were separately tabulated.

Other myopathic changes.

The prevalence of regenerative myofibers, defined as sarcoplasmic basophilia with or without reactive nuclei, and chronic myopathic fibers, defined as those with small central nuclei without sarcoplasmic basophilia, was assessed using a 0–3 grading system.

Stromal changes.

Perimysial edema and fibrosis were evaluated separately. Each was graded as follows: 0 = none, 1 = mild and focal, 2 = moderate, and 3 = severe and universal.

Statistical analysis.

Means and SDs were calculated for numerical variables and frequencies were calculated for categorical variables. The clinical course variables for Crowe and Spencer (0 versus 1 or 2 in each classification, with 0 designating limited disease) were, respectively, converted into 2 binary variables: a chronicity dummy variable and an ulceration dummy variable. Each morphology variable was evaluated independently; arteropathy, arterial DIF, and infarct were also evaluated in combination as a binary variable. One or more dummy variables were created for each categorical independent variable to evaluate significance of degrees of abnormality (0 and 1 versus 3, 0 and 1 versus 2 and 3, etc.). Simple logistic regression models were constructed to screen for the independent histopathologic variables that were significantly associated with the response variable (limited disease [0] versus chronic disease [1 and 2; 1 or 2]). Those independent variables in which the P values were less than 0.2 for the chi-square statistics were then examined together in a multivariate model. A backward elimination logistic regression procedure was performed with a cutoff P value of 0.10. The obtained final fitted models were presented in a logit formula. All the analyses were conducted using SAS software, version 9.1.3 (SAS Institute, Cary, NC).


Patient data at the time of initial juvenile DM diagnosis and the classification of the clinical course after treatment are summarized in Table 1. The followup period typically greatly exceeded the minimum for inclusion; most patients were followed to adulthood. There was no significant difference in age or sex distribution of patients with different clinical courses (P > 0.3).

Table 1. Clinical data at onset and disease course in juvenile dermatomyositis
VariableCrowe classification (4)Spencer classification (5)
0 (n = 28)1 (n = 26)2 (n = 18)Total (n = 72)0 (n = 29)1 (n = 10)2 (n = 28)Total (n = 67)
Age, mean ± SD years7.9 ± 4.47.7 ± 4.39.0 ± 4.98.1 ± 4.58.0 ± 4.47.4 ± 3.08.6 ± 4.78.1 ± 4.3
Sex, no.        
Height, no. <10th  percentile/total no.5/282/241/118/635/291/82/258/62
Weight, no. <10th  percentile/total no.4/285/253/1412/674/291/95/2610/64
Interval to biopsy        
 Mean ± SD months6.7 ± 7.49.4 ± 10.27.3 ± 8.87.8 ± 8.86.5 ± 7.47.0 ± 7.89.1 ± 10.47.6 ± 8.7
 No. of patients282514672992664

Using the Crowe classification, 28 patients had limited disease (grade 0), 26 patients had chronic disease without cutaneous ulceration (grade 1), and 18 patients had chronic disease and ulceration (grade 2). Using the Spencer classification, 29 patients had limited disease (grade 0), 10 patients had polycyclic chronic disease (grade 1), and 28 had unicyclic chronic disease (grade 2).

Using simple logistic regression for each variable, we determined that most of the histologic features tested, despite proven value for the diagnosis of inflammatory myopathy, had no value as predictors of either limited or chronic disease. The frequency and grade of each variable are presented in Table 2. The results of the simple logistic regression analysis are summarized in Table 3.

Table 2. Summary of biopsy findings in 72 patients with juvenile dermatomyositis*
  • *

    Values are the number (percentage). NA = not applicable; DIF = direct immunofluorescence.

Inflammation8 (11.1)50 (69.4)9 (12.5)5 (6.9)  
 Perimysial    47 (65.3)25 (34.7)
 Endomysial    21 (29.2)51 (70.8)
 Perivascular    55 (76.4)17 (23.6)
Vascular changes      
 Arteropathy31 (43.1)29 (40.3)12 (16.7)NA  
 Capillary dropout/fibrosis36 (50.0)11 (15.3)10 (13.9)15 (20.8)  
 Arterial DIF    25 (37.9)41 (62.1)
 Capillary DIF    26 (39.4)40 (60.6)
 Infarct    14 (19.4)58 (80.6)
Myopathic changes      
 Active myopathic changes24 (33.3)24 (33.3)12 (16.7)12 (16.7)  
  Degeneration    30 (41.7)42 (58.3)
  Vacuolation    17 (23.6)55 (76.4)
  Sarcoplasmic pallor    24 (33.3)48 (66.7)
  Necrosis    34 (47.2)38 (52.8)
 Chronic myopathic changes      
  Central nuclei32 (44.4)23 (31.9)9 (12.5)8 (11.1)  
  Regeneration (basophilia)21 (29.2)31 (43.1)8 (11.1)12 (16.7)  
  Perifascicular atrophy20 (27.8)9 (12.5)8 (11.1)35 (48.6)  
  Type I myoatrophy65 (90.9)3 (4.2)4 (5.6)NA  
  Type II myoatrophy68 (94.4)2 (2.8)2 (2.8)NA  
  Random myoatrophy59 (81.9)9 (12.5)2 (2.8)2 (2.8)  
Stromal changes      
 Perimysial edema/fibrosis15 (20.8)16 (22.2)24 (33.3)17 (23.6)  
 Fibrosis33 (45.8)25 (34.7)12 (16.7)2 (2.8)  
Table 3. Significant histopathologic predictors of clinical course in juvenile dermatomyositis based upon simple logistic regression*
Histopathologic featureCrowe grade 0 vs. 1 and 2Crowe grade 0 and 1 vs. 2Spencer grade 0 vs. 1 and 2Spencer grade 0 and 1 vs. 2
POR (95% CI)POR (95% CI)POR (95% CI)POR (95% CI)
  • *

    Empty cells indicate lack of statistical significance. Features not included did not correlate with clinical course. OR = odds ratio; 95% CI = 95% confidence interval; DIF = direct immunofluorescence; CNWOB = central nuclei without sarcoplasmic basophilia.

Vascular changes        
 Arteropathy, grade 20.049.0 (1.1–74)0.0019.0 (2.52–39.7)  0.042.11 (1.03–4.32)
 Arterial DIF, any  0.0027.66 (2.09–28.03)    
 Capillary dropout/fibrosis,  grade 3  0.0363.66 (1.09–12.26)    
Myopathic changes        
 Active myopathic changes  (necrosis), grade 30.049.0 (1.1–74)0.0076.24 (1.66–23.37)  0.0191.81 (1.10–2.97)
 Infarct  0.0214.27 (1.24–14.71)    
 CNWOB, any (all patients)0.0083.86 (1.0–10.5)  0.0084.02 (1.44–11.21)0.0133.88 (1.34–11.25)
 Central nuclei, any (<9  months from onset)0.016   0.013  0.01

Inflammatory activity.

Neither the distribution (perivascular, perimysial, or endomysial) nor the amount of inflammatory activity had predictive value for chronic juvenile DM (P > 0.3).

Vascular disease.

Mild forms of arteriolopathy and arteropathy, minor focal obvious capillary dropout/fibrosis, and presence or absence of capillary DIF had no predictive value for chronic juvenile DM (P > 0.5). Severe arteropathy (Figure 1) and any positive arterial DIF were predictive of chronic disease (Crowe and Spencer classification), but had no statistical relationship to length of the onset-to-biopsy interval (P < 0.02). Foci of severe obvious capillary loss accompanied by myoatrophy and endomysial fibrosis, typically in PF location (Figure 2), predicted chronicity (Crowe grade 2 only).

Figure 1.

Noninflammatory arteropathic lesions (arrows) in 4 patients with juvenile dermatomyositis (DM) are illustrated. The arteropathy is visible but nonocclusive in panel A and occlusive of the arterial lumens in panels B, C, and D. Occlusive arteropathy predicts chronic juvenile DM. (Hematoxylin and eosin stained; original magnification × 250.)

Figure 2.

Necrotic muscle fibers are observed in all panels. Isolated single necrotic muscle fibers in panel A (arrows) are not predictive of chronic juvenile dermatomyositis (DM). Clusters of necrotic muscle fibers, as seen in multiple fascicles in panels B, C, and D, are predictive of chronic juvenile DM. (Hematoxylin and eosin stain in A and C; modified trichrome stain in B and D. Original magnification × 250 in A, C, and D; × 160 in B.)

Muscle fiber atrophy.

PF atrophy (regardless of severity), random atrophy, and type II myofiber atrophy did not predict chronic disease (P > 0.2).

Muscle fiber regeneration (sarcoplasmic basophilia).

Neither the presence nor the extent of basophilic regeneration predicted a chronic clinical course (P > 0.1).

Myopathic changes.

Focal active myopathic changes involving a few scattered myofibers (grade 1 change) had no predictive value for chronicity. Extensive active myopathic changes (Figure 3) predicted a chronic clinical course in both the Crowe and Spencer classifications (Table 3). There was no relationship to the length of the onset-to-biopsy interval. Muscle infarct predicted chronicity (Crowe grade 2 only).

Figure 3.

Obvious capillary loss is associated with endomysial fibrosis and variable myofiber atrophy (panels A and B). Central nuclei without sarcoplasmic basophilia are shown in panels A, C, and D. Both features predict chronic juvenile dermatomyositis. (Hematoxylin and eosin stain. Original magnification × 250 in A, B, and D; × 160 in C.)

Central nuclei without sarcoplasmic basophilia.

Even if mild, central nuclei without sarcoplasmic basophilia (Figure 3) predicted chronic juvenile DM using both Crowe and Spencer classifications (Table 3) and correlated to the onset-to-biopsy interval (P < 0.007 for Crowe, P < 0.02 for Spencer). This correlation persisted for the subset of cases with intervals <9 months (P < 0.01–0.02). When multivariate linear regression was performed omitting cases with central nuclei without sarcoplasmic basophilia, no additional features with predictive value emerged.

Changes in stroma.

Neither fibrosis nor edema, regardless of severity or distribution, predicted the clinical course (P > 0.3).

Based upon simple logistic regression, morphologic features that predicted a chronic course are listed with P values and odds ratios in Table 3. Severe arteropathy, positive arterial DIF, infarct, and multiple foci of severe capillary loss/endomysial fibrosis were selective predictors of chronic disease with an ulcerative course (Crowe grade 2). Severe arteropathy, severe active myopathic changes (excluding regeneration), and central nuclei without basophilia predicted chronic juvenile DM using the Crowe classification. Only central nuclei without basophilia predicted both forms of chronic disease using the Spencer classification.

In cross-tabulation studies (chi-square analysis) of morphologic features that were predictive of chronic juvenile DM, any degree of muscle fiber degeneration/necrosis correlated with any degree of arteropathy (P < 0.027) and with the combined variable (P < 0.016), but did not correlate with foci of severe capillary dropout/endomysial fibrosis or with central nuclei without basophilia. However, severe focal capillary dropout/endomysial fibrosis correlated with central nuclei without basophilia (P < 0.032); both are features of established chronicity.

The combined variable, a coalition of features consisting of severe arteropathy, positive arterial vascular fluorescence for any of the listed antisera, and muscle infarct, was considered positive when any 2 of these 3 components were present. Analysis demonstrated no improvement in ability to predict chronic juvenile DM compared with the 3 components evaluated separately. The combined variable correlated well with presence of active myopathic change, supporting a close pathogenetic relationship between severity of vascular disease and severity of myopathic changes.

The final model based on multivariate logistic regression identified 4 features that correlated best with clinical course of juvenile DM based on Crowe or Spencer classification. Central nuclei without basophilia was the best predictor for chronic disease using both Crowe and Spencer classifications (P < 0.007 and P < 0.008, respectively); the odds ratio for the Crowe classification (grade 1 or 2) was 4.25 and the odds ratio for the Spencer classification (grade 1 or 2) was 4.02. Using simple linear regression, the greater extent of active myopathic change (moderate and severe degeneration and necrosis) predicted subsets of chronic disease (Crowe grade 2 and Spencer grade 2), but using multivariate logistic regression, extensive active myopathic change strongly predicted only chronic disease with an ulcerative course (Crowe grade 2, P < 0.004, odds ratio 4.2). Similarly, using multivariate logistic regression, foci of severe capillary dropout/endomysial fibrosis was no longer an independent predictor of chronic disease with an ulcerative course. However, severe arteropathy and positive direct arterial fluorescence predicted ulcerative juvenile DM (Crowe grade 2).

The interval between onset of the disease, as determined by clinical history obtained at referral, and the pretreatment MB did not correlate with the disease course based on either Crowe or Spencer classification (P > 0.2). However, when we divided the interval into 2 periods (<9 months and >9 months), we found that chronic disease was more likely if the interval was >9 months (P < 0.03). We found no correlation between low weight and height at the time of referral/biopsy and the ultimate clinical course (P > 0.2). Common, diagnostically useful histopathologic variables in MB specimens from children with juvenile DM such as amount and distribution of inflammation, any pattern or severity of muscle atrophy including PF atrophy, amount of muscle basophilic regeneration, severity of stromal edema and/or fibrosis, mild arteropathy, focal mild capillary loss/endomysial fibrosis, or positive capillary DIF did not predict the clinical course of juvenile DM.


This detailed retrospective analysis of archived slides from MBs performed at one referral center over a 25-year period was based on a sample size of only 72 patients; therefore, the power of this study was limited to detecting big differences. The findings suggest that 4 specific criteria may be used in prospective studies to predict a high probability of chronic disease course based on pretreatment MB findings in a child with juvenile DM. The odds ratios (Table 3), based upon simple linear regression, that a child with one of these findings would fail to achieve remission (i.e., limited disease in both Crowe and Spencer classifications) after a course of corticosteroid therapy were as follows: prevalent acute myopathic changes (odds ratio 9.0 for chronic juvenile DM, Crowe grade 1 or 2), severe intramuscular arteropathy (odds ratio 9.0 for chronic juvenile DM, Crowe and Spencer classifications), presence of nonregenerative muscle fibers with central nuclei (odds ratio 3.86–4.25 for chronic juvenile DM, Crowe and Spencer classifications), and focal severe loss of capillary bed in areas of PF endomysial fibrosis (odds ratio 3.66 for chronic juvenile DM, Crowe grade 2).

The amount and distribution of inflammation was not predictive of clinical course. Because immunomarkers were not used to characterize the inflammatory cells in the infiltrates, this negative result does not preclude the possibility that the proportions or intramuscular distribution of leukocyte subtypes may have predictive value for clinical outcome.

Vascular changes in juvenile DM are well described and presumably result from an immune-mediated injury (13), which as yet is not well understood. Manifestations of active vascular disease include destruction of endothelium of capillaries and small muscular arteries, and lesions that are often accompanied by vascular mural deposits of 1 or more of the following proteins: C3 derivatives, C5, C5–9 attack complex, fibrinogen, and IgM. Capillary obliteration and arterial lumen occlusion may be expected to cause secondary ischemic changes in muscle that are superimposed upon immune-mediated primary myofiber injury. Our analysis indicates that light microscopic evidence for vascular injury is common in juvenile DM. However, the likelihood of chronic juvenile DM is substantially increased only when severe vascular changes, such as lumen compromised by endothelial cell debris and/or myointimal proliferation, and medial hyalinization and smooth muscle cell atrophy are easily observed in cryostat sections, and when obliteration of the capillary bed is obvious in foci of prominent endomysial fibrosis. In this retrospective analysis, no immunomarkers for endothelium were consistently used; such an approach in the future may provide more conclusive documentation of capillary loss and a more refined assessment of its significance, assuming that methodologic problems relating to quantitative analysis of the intramuscular capillary bed can be overcome.

Four patterns of muscle fiber atrophy were assessed: PF atrophy, random intrafascicular atrophy, and selective type I or type II fiber atrophy. PF atrophy is by far the most common pattern observed in juvenile DM, although not universal, because it was only present in 52 of 72 patients. Neither the presence nor the severity of PF atrophy is predictive of outcome, a finding that suggests that PF atrophy is reversible. Reversibility of PF atrophy may depend on the restoration of a healthy capillary network as well as intrinsic myofiber repair.

For the purposes of this analysis, regenerative myofibers were defined as those with sarcoplasmic basophilia in routine stains, regardless of diameter. In these myofibers, reactive nuclear enlargement is common. In the future, refined techniques such as the immunostain for the heavy chain of fetal myosin may be more specific for regeneration and less subjective (14, 15). Regeneration, as defined herein, is common in juvenile DM but is far from universal (51 of 72 patients). Its absence could have many possible meanings such as insufficient myofiber injury to stimulate regeneration, or transience of the regenerative response. Although the term regeneration implies beneficence, no correlation was observed between amount of basophilic regeneration and either good or bad clinical outcome.

Migration of skeletal muscle nuclei from the normal subsarcolemmal position to the center of the myofiber without reactive nuclear enlargement, and without sarcoplasmic basophilia, is a nonspecific response of the myofiber to injury in neuromuscular disorders. It is probably a chronic state. Our results clearly indicate that central nuclei in nonbasophilic fibers are a strong predictor of chronic juvenile DM even when the frequency is low; because of a significant relationship to the length of the onset-to-biopsy interval, it is an indicator of established chronic myopathy. Active/acute myopathic changes include myofiber necrosis involving individual myofibers or, as is often the case in juvenile DM, clusters of necrotic myofibers. The definition used in this study requires unequivocal evidence of myofiber degeneration and encompasses the entire spectrum of changes in the sequence that begins with breakdown of myofibrils and eosinophilic homogenization due to loss of the normal myofibrillar network, pallor and lysis of sarcoplasm, and invasion by macrophages with preservation of the sarcolemma. One-third of biopsy samples from patients with juvenile DM (24 of 72) lack these changes. We found that the presence of multiple foci of myofiber necrosis in clusters is predictive of a chronic and/or an ulcerative course and is unrelated to the interval between onset and MB.

Stromal changes are common in juvenile DM and may be severe, but they are difficult to appreciate except in technically ideal cryostat sections. Particularly in the perimysial connective tissue, edema, presumed to represent increased ground substance, is often very extensive, resulting in abnormally wide spaces between muscle fascicles and in wide separation of unusually coarse collagen fibers. Fibrosis, in the sense of new or replacement connective tissue, is very difficult to appreciate in this context. To overcome this difficulty, we combined edema and fibrosis and only graded and tabulated the more obvious stromal changes as positive. There was no correlation with clinical outcome. Endomysial fibrosis associated with apparent or obvious loss of the intrafascicular capillary network was evaluated separately.

Numerous studies have established that a noninflammatory vasculopathy affecting muscle capillaries and small arteries is prevalent in juvenile DM and that complement components play an important role (12, 15). When severe, the occlusive form of the arteropathy may be appreciated in routine stains such as hematoxylin and eosin or the modified trichrome method, but direct immunofluorescence methods may be more sensitive to mild or lingering arteropathy. Positive capillary fluorescence was observed in 39% of our patients, typically in a focal pattern; positive arterial fluorescence was observed in 38% of muscle specimens. However, only positive arterial fluorescence had predictive value for a chronic course.

In a study of adults and children with inflammatory myopathy, Kissel et al (16) previously observed that vascular labeling with antibodies to the complement membrane attack complex (C5b–9) was inversely related to the duration of the disease prior to biopsy and treatment. They also observed that there was a correlation between the number of fascicles with myofibrillar loss (interpreted as early necrosis) and the percentage of fascicles with C5b–9 deposits in muscle capillaries. They further observed a negative correlation between PF atrophy and deposits of membrane attack complex. Our methods differed in that we used an antibody to C5 rather than C5b–9 complex, and all of our patients were children. In our population, C5 arterial deposits were found in only 4 of 57 specimens studied. Kissel et al (16) inferred that complement membrane attack complex-mediated vasculopathy was early and transient because their patients with negative vascular fluorescence had been ill for a longer period than those in whom vascular deposits were found. In comparison, we found no relationship between positive vascular labeling and duration of disease prior to biopsy. However, the average duration of prebiopsy symptoms in our patients was 8 months, twice as long as in a recent study of 166 children with untreated juvenile DM (17). To reconcile these disparate observations, we hypothesize that positive arterial labeling with antisera to C3 derivatives, IgM, and/or fibrinogen may linger in injured vessels beyond the early phase of injury when C5b–9 complex is most active.

Pachman et al (17) reported that as a group, children with untreated juvenile DM were shorter and weighed less than expected for their age at the time of initial diagnosis. We observed that 30% of our patients at initial examination were less than the tenth percentile for either or both parameters. However, we observed no correlation of short stature or low weight for age at diagnosis with either limited or chronic disease.

A retrospective study cannot definitively identify predictors of clinical course, but can generate hypotheses as to which variables are most important. Our results indicate the feasibility of applying a scoring system, as recently proposed by a consensus of muscle pathologists (11, 12), to pretreatment MB in patients with juvenile DM for the purpose of stratification of therapeutic regimens according to risk of chronic disease. Our data indicate how a score could be weighted toward features with prognostic value. The validity of an approach that focuses on both MB findings and suggested clinical predictors of chronic disease such as severity of muscle weakness at onset (9), onset of cutaneous ulceration shortly after initiation of high-dose corticosteroid therapy (10), and nailfold capillary bed changes (18) must be tested prospectively. A component of such a study should be evaluation of contemporary immunostains to detect major histocompatibility complex expression (19), subsets of inflammatory cells, loss of vascular bed, and myofiber regeneration. The goal is development of a reliable system for classifying juvenile DM into favorable and unfavorable prognostic groups at the time of first evaluation.


Dr. Bove had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Miles, Bove, Lovell.

Acquisition of data. Miles, Bove, Lovell, Wargula, Bukulmez.

Analysis and interpretation of data. Miles, Bove, Lovell, Wargula, Shao, Salisbury, Bean.

Manuscript preparation. Miles, Bove, Lovell, Wargula, Bean.

Statistical analysis. Shao, Salisbury, Bean.

Clinical database. Lovell, Wargula, Bukulmez.


We are indebted to Joseph Levinson, founder of the Division of Pediatric Rheumatology at Cincinnati Children's Hospital, who enthusiastically promoted muscle biopsy and the collection of standardized longitudinally collected data in children with myositis, and to Lucy Wedderburn, the Institute of Child Health, London, UK, who organized the meetings and led the discussions of the International Working Group for Muscle Biopsy Scoring in Juvenile Dermatomyositis, upon which the criteria used in this study are partly based.