To develop criteria for the classification of systemic sclerosis (SSc) in children (juvenile SSc).
To develop criteria for the classification of systemic sclerosis (SSc) in children (juvenile SSc).
The study consisted of 3 phases: 1) collection of data on the signs and symptoms of actual patients with juvenile SSc that are useful for defining involvement of a particular organ; 2) selection of the parameters essential for the classification of juvenile SSc and preparation of a set of provisional classification criteria (PCC) using 2 Delphi surveys; 3) consensus conference consisting of 2 steps: discussion and rating of clinical profiles of 160 patients with definite juvenile SSc, possible juvenile SSc, or other fibrosing diseases as “having or not having juvenile SSc,” using nominal group technique, and defining those PCC with the best statistical performance and highest face validity by using the clinical profiles of patients with definite juvenile SSc as the gold standard.
In phase 1, 55 centers submitted clinical data on 153 patients with juvenile SSc. A total of 48 signs and symptoms were derived from these patient data and were used to define 9 organ system categories (cutaneous, vascular, gastrointestinal, respiratory, renal, cardiac, neurologic, musculoskeletal, and serologic). During phase 2, these were reduced to 21 criteria (3 major criteria [Raynaud's phenomenon, proximal skin sclerosis/induration of the skin, and sclerodactyly] and 18 minor criteria) and combined to generate 86 different PCC. At the consensus conference, these 86 definitions were tested on the case profiles of 127 patients with juvenile SSc. The PCC with the highest ranking were proximal sclerosis/induration and at least 2 minor criteria.
These provisional classification criteria for juvenile SSc will help standardize the conduct of clinical research, epidemiologic and outcome studies, and therapeutic trials.
Juvenile scleroderma syndromes are a group of rare conditions including juvenile systemic sclerosis (SSc), mixed connective tissue disease (MCTD), and overlap syndromes that result in varying degrees of fibrosis of the skin and internal organs, with onset in children and adolescents younger than age 16 years. At present, there are no universally accepted nomenclature and classification criteria for juvenile SSc.
Classification criteria for SSc in adults were established in 1980 and include the presence of 1 major criterion, namely symmetrical thickening of the skin proximal to the metacarpophalangeal (MCP) or metatarsophalangeal (MTP) joints, or 2 or more minor criteria, including sclerodactyly, digital pitting scars or loss of substance from the finger pad, and bibasilar pulmonary fibrosis (1). More recently, LeRoy and Medsger introduced the concept of early scleroderma, defined as the presence of Raynaud's phenomenon (RP) plus either scleroderma-type nailfold capillary changes or SSc-selective serum autoantibodies (2). Although that concept is still being debated, the group of patients with early scleroderma is quite important in prospective population-based studies, particularly studies in children in whom one of the subtypes of juvenile SSc may develop during extended followup.
In children, the relative rarity SSc, its heterogeneity, and the difficulty in separating juvenile SSc–like conditions from non-juvenile SSc–like conditions such as fasciitis and phenylketonuria point to the need for developing acceptable criteria for the diagnosis and classification of juvenile SSc. The lack of such criteria may in fact lead to inclusion of nonhomogeneous groups of patients in clinical trials, with the associated risk of reporting bias, ambiguous interpretation of results, and inability to compare therapies. We undertook this study, sponsored by the Pediatric Rheumatology European Society, in order to develop uniform nomenclature and criteria for the classification of patients with juvenile SSc, based on their clinical and laboratory features.
The study was conducted by using the Delphi technique and nominal group technique (NGT), both of which are well-recognized consensus formation methodologies specifically designed to combine judgments from a group of experts in a particular field (3, 4). The Delphi technique (used in phases 1 and 2 of the study) utilizes a series of defined questionnaire-based surveys, while NGT (applied in phase 3) is a structured face-to-face meeting designed to facilitate reaching consensus. These techniques have been used to develop diagnostic criteria or outcome measures for several diseases (5–10). Consensus formation methodology was designed so that each step was based on the results of the previous steps. The study was performed in 3 phases.
Information on demographic, clinical, and laboratory features of patients with onset of SSc before age 16 years was retrospectively collected from patients seen at 270 pediatric rheumatology centers in Europe (n = 166), North America (n = 42), South America (n = 28), Asia (n = 30), Australia (n = 2), and Africa (n = 2). These centers were identified from the mailing lists of the Pediatric Rheumatology European Society and the Pediatric Rheumatology International Trials Organization.
A standardized case report form was developed for collection of data. Demographic information (sex, age at the first signs or symptoms of the disease, and age at diagnosis) was collected. Participants were asked to use the following guidelines to define organ involvement at the time of diagnosis (Table 1). Skin involvement was determined by the presence of skin induration proximal or distal to the MCP and MTP joints, edema, calcinosis, or sclerodactyly. Peripheral vascular system involvement was defined as the presence of any 1 of the following: RP, digital ulcers, telangiectasias, abnormal nailfold capillaries, and capillaroscopy findings (such as megacapillaries and avascular areas). Respiratory involvement was defined as the presence of any 1 of the following: dyspnea, lung fibrosis (on chest radiograph or high-resolution computed tomography scan), reduced diffusing capacity for carbon monoxide (DLco) and forced vital capacity, or pulmonary hypertension (assessed by echocardiography). Cardiac disease included any 1 of the following: arrhythmias, pericardial effusion, heart failure, palpitation, chest pain, dizziness, presyncope/syncope, edema, and venous congestion. Musculoskeletal involvement was defined by the presence of myositis, myopathy, arthritis, arthralgia, or tendon friction rubs. Gastrointestinal tract symptoms consisted of dysphagia, gastroesophageal reflux, heartburn, intestinal distention, vomiting, diarrhea, constipation, or weight loss. Renal disease was identified by the presence of a raised creatinine level or urea serum level, abnormal results of urinalysis, renal crisis, or persistent arterial hypertension. Nervous system involvement was determined by the report of seizures, peripheral neuropathy, headache, carpal tunnel syndrome, or neuroimaging abnormalities.
|Organ system||Sign or symptom|
|2. Peripheral vascular||Raynaud's phenomenon|
|Nailfold capillary changes|
|Nailfold capillaroscopy findings|
|4. Cardiac||Congestive heart failure|
|5. Renal||Renal crisis|
|Abnormal serum creatinine value|
|Abnormal serum urea value|
|Abnormal urinalysis results|
|Lung fibrosis (HRCT/radiography)|
|Reduced FVC <80%|
|Reduced DLco <80%|
|Pulmonary arterial hypertension|
|7. Neurologic||Nerve involvement|
|Carpal tunnel syndrome|
|Tendon friction rubs|
|9. Serologic||Antinuclear antibodies|
Data were also collected for serum autoantibodies, specifically antinuclear antibodies (ANA), anticentromere antibodies (ACA), anti–topoisomerase I (Scl-70) antibodies, antifibrillarin, anti–PM-Scl, anti–fibrillin, or anti–RNA polymerase I/III antibodies. The normal ranges of laboratory standards for each participating center were used to determine abnormal values. Since all clinical information was anonymously collected from the patient charts, institutional review board approval was required from only a few centers, mainly in North America. Data were stored electronically on a secure computer network, according to locally applicable guidelines, in the participating centers.
Participants were asked to report data on patients with a diagnosis of SSc based on the American College of Rheumatology (ACR; formerly, the American Rheumatism Association) preliminary criteria (1). Information on patients with overlap syndrome and MCTD was also collected. Before the analysis was completed, the principal investigators at the participating centers were asked to verify the accuracy of the classification of their patients.
Using the data collected in phase 1, questionnaires were developed and mailed to 14 experienced adult and pediatric rheumatologists in 7 countries. Ten were pediatric rheumatologists who had contributed a consistent number of patients to phase 1 and/or had actively participated in the planning of the study, and 4 were adult rheumatologists selected on the basis of their clinical and research expertise in juvenile SSc.
In the first Delphi survey, these experts were asked to define in detail the signs and symptoms that, in their opinion, were most useful in defining involvement of a specific organ system, alone or in combination, and rank them. Since the number of items was different for each organ, the mean score obtained for each organ was divided by the number of variables considered for that organ, in order to make the scores comparable. This weighted score represented the overall judgment of the importance of a sign or symptom in defining involvement of a particular organ.
In the second survey, the experts were asked to select the signs and symptoms that were strongly suggestive of the diagnosis of juvenile SSc and were therefore essential for classification of juvenile SSc. The rate of choice, representing the percentage of experts judging a particular sign or symptom as essential, was multiplied by the weighted score to obtain the final score. Among the variables with the highest final scores, those with a prevalence ≥50% at the time of diagnosis, based on our patient population, were selected as provisional major criteria and the remaining variables were listed as minor criteria (Table 2).
|Signs and symptoms||Weighted score||Rate of choice||Final ranking||Prevalence at diagnosis||Criteria classification|
|Pulmonary fibrosis (radiography, HRCT)||90||0.93||83.7||0.20||Minor|
|Tendon friction rubs||70||0.21||14.7||0.06||Minor|
|Nailfold capillary abnormalities||64.3||0.71||45.7||0.48||Minor|
|Carpal tunnel syndrome||74.3||0.5||37.2||0.01||Minor|
|Digital tip ulcers||62.8||0.36||22.6||0.35||Minor|
In order to mimic as much as possible the diagnostic process used when assessing a patient with juvenile SSc, a set of 86 provisional classification criteria (PCC) definitions consisting of major and minor criteria was generated by the steering committee. These PCC were selected on the basis of a logic combination of major and minor criteria, taking into consideration either the various clinical pictures presented at diagnosis by the patients collected in phase 1 or the personal experience of the experts.
The PCC were grouped under 3 headings: generic definitions, specific definitions, and historic definitions. Generic definitions included the presence of 1, 2, or 3 major criteria, with a combination of 1 or more minor criteria (definitions 1–27). An example of a generic definition is as follows: definition 11 — presence of 2 major criteria and at least 1 minor criterion. Specific definitions included the presence of 1 or 2 specific (e.g., proximal sclerosis/induration) major criteria with a combination of minor criteria (definitions 28–81). An example of a specific definition is as follows: definition 53 — presence of RP and proximal sclerosis/induration and at least 1 minor criterion. A group of 5 historic definitions were obtained from the literature (definitions 82–86) (1, 2).
After the completion of phase 2, a consensus conference was held, with the following objectives: 1) using nominal group technique, rate the profiles of real patients as having or not having juvenile SSc, 2) evaluate the set of PCC using the clinical profiles of patients considered to be the gold standard based on the consensus of experts, and 3) define the PCC that had the best statistical performance and highest face validity.
The profiles of the patients tested at the consensus conference were based on real data collected from patients in phase 1, with emphasis placed on data from patients in whom the diagnosis was more difficult due to incomplete clinical manifestations. Profiles of patients with confounding diseases such as juvenile dermatomyositis, juvenile systemic lupus erythematosus (SLE), overlap syndrome, MCTD, localized scleroderma, and other fibrosing conditions were also included (Table 3). Experts were asked to individually rate each patient as having or not having juvenile SSc, based on the profiles provided. If ≥80% consensus for a particular patient was not achieved, the patient was discussed by the entire group and a second vote was taken. If ≥80% consensus was still not attained, the patient profile was declared indefinable. Patient profiles for which ≥80% physician consensus was achieved were used as the gold standard for the next phase (evaluation of the PCC).
|Disease||Clinical subtype||No. of patients||Clinical features (no. of patients)|
|Systemic sclerosis||100||Skin induration (76), RP (75), sclerodactyly (52), digital tip ulcers (30), pulmonary involvement (17), arthritis (16), myositis (12), dysphagia (13), GER (15), nailfold capillary changes (44), tendon friction rubs (5), cardiac involvement (3), renal involvement (1), carpal tunnel syndrome (1), peripheral neuropathy (1), ANA (80), anti–Scl-70 (20), ACA (5), anti–PM-Scl (3)|
|Overlap syndrome||SLE/juvenile DM||5||Myositis (5), pulmonary involvement (2), calcinosis (1), RP (3), ANA (5)|
|SLE/JIA||2||Arthritis (2), RP (2), ANA (2), RF (1)|
|Juvenile SSc/JIA||3||Arthritis (3), RP (2), ANA (3), sclerodactyly (1)|
|Juvenile SSc/juvenile DM||10||Myositis (10), pulmonary involvement (5), sclerodactyly (2), calcinosis (1), RP (8), ANA (8)|
|Mixed connective tissue disease||15||Arthritis (3), sclerodactyly (3), renal involvement (5), RP (10), pulmonary involvement (3), ANA (15), RNP (15)|
|Localized scleroderma||Linear scleroderma||5||RP (2), GER (3), ANA (5), anti–PM-Scl (1)|
|Pansclerotic morphea||1||Skin sclerosis/induration, ANA|
|Deep morphea||1||Arthritis, RP, ANA|
|Eosinophilic fasciitis||3||Skin sclerosis/induration (3), ANA (2)|
|SLE||5||RP (3), renal involvement (3), pulmonary involvement (1), arthritis (1), ANA (5)|
|Dermatomyositis||5||Myositis (5), pulmonary involvement (2), ANA (1)|
|Miscellaneous||Progeria||1||Skin sclerosis/induration, arthritis|
|Primary pulmonary hypertension||1||Pulmonary involvement, ANA|
|Erythropoietic protoporphyria||1||Sclerodactyly, arthritis|
|Connective tissue nevus||2||Skin sclerosis/induration (2), GER (1), ANA (1)|
Classification of individual patients as having or not having juvenile SSc using the 86 candidate criteria sets was compared with classification by physician consensus (gold standard). For each possible set of criteria, chi-square tests with 1 degree of freedom were performed and the corresponding P values were determined, and sensitivity (ability of the criteria to identify a patient who had been classified by the physicians as having juvenile SSc as having juvenile SSc), specificity (ability of the criteria to identify a patient who was not classified as having juvenile SSc by the physicians as not having juvenile SSc), percent false positive (number of patients falsely identified as having juvenile SSc by the criteria divided by the total number of patients identified as juvenile SSc × 100), and percent false negative (number of patients falsely identified as not having juvenile SSc by the criteria divided by the total number of patients identified as not having juvenile SSc × 100), and the area under the curve (AUC) value were calculated. As a measure of agreement between the physicians' evaluation and the potential criteria, we used the Cohen kappa value (11) with cutoffs as proposed by Landis and Kock (12) (i.e., 0.01–0.2 = slight, 0.2–0.4 = fair, 0.4–0.6 = moderate, 0.6–0.8 = substantial, 0.8–1 = almost perfect agreement). Kappa values ≥0.7 were considered evidence of substantial agreement, and the PCC with those values were selected for further consideration.
The statistical evaluations described above (including the evaluations with low statistical performance) were presented and fully explained to the consensus participants. The consensus conference attendees were asked to decide which of the juvenile SSc PCC definitions that performed best from a statistical point of view in the earlier step were most credible and easiest to use (face validity) in current clinical practice. Attendees were asked to rank the 5 best definitions from 1 (lowest) to 5 (highest). In order to combine the consensus evaluation with the statistical performance, the face validity rankings were multiplied by Cohen's kappa values to obtain the final PCC ranking.
During the period from January 2002 to June 2003, 55 centers (32 European, 8 North American, 11 South American, and 4 Asian) contributed data on 153 patients with juvenile SSc. There were 120 female patients (78.4%) and 33 male patients (21.6%), with a female to male ratio of 3.6:1.
At the time of diagnosis of juvenile SSc, RP was the most frequently reported clinical manifestation (76%), followed by skin induration (66%), sclerodactyly (55%), digital tip ulcers (35%), lung fibrosis (20%), arthritis (18%), myositis (15%), dysphagia (13%), gastroesophageal reflux (13%), tendon friction rubs (6%), pulmonary hypertension (5%), heart failure (3%), arrhythmias (2%), renal crisis (1%), carpal tunnel syndrome (1%), new-onset hypertension (1%), and peripheral neuropathy (1%). Nailfold capillary changes were reported in 48% of the patients, and reduced DLCOwas reported in 17%.
Among the characteristic autoantibodies, ANA were positive in 77% of the patients, SSc-selective autoantibodies were positive in 33% of the patients (anti–Scl-70 24%, ACA 5%, anti–PM-Scl 3%, others 1%). In comparison with the adult form, juvenile SSc appeared less severe, had less internal organ involvement (particularly at the time of diagnosis), and had a less characteristic immunologic profile. A more comprehensive description of the clinical and immunologic features of the juvenile SSc cohort in this study, along with a comparison of clinical features of adult SSc patients, has been reported in a separate article (13).
In the first Delphi survey, the investigators defined 48 signs and symptoms, grouped into 9 organ systems (cutaneous, peripheral vascular, gastrointestinal, cardiac, renal, respiratory, neurologic, musculoskeletal, and serologic) (Table 1). These variables were weighted on the basis of their importance in defining involvement of a particular organ. Specifically, the investigators were asked to rank the parameters by assigning a score of 1 to the variable that is least important in defining involvement of a particular organ and assigning the highest score to the most important variable. Variables that were not considered important were assigned a score of 0.
Because the number of items for each organ system was different, in order to make the scores comparable, the score obtained by the 14 investigators was divided by the maximal hypothetical score obtainable for involvement of a particular organ. This ratio was then multiplied by 100 to obtain a weighted score that represented the overall judgment of the importance of a sign or symptom in defining involvement of a particular organ (Table 2). For example, 5 signs/symptoms were associated with peripheral vascular involvement; therefore, the maximal hypothetical score obtainable would be 70 (5 [maximal score] × 14 [number of experts]). The total score for RP was 69, representing a weighted score of 98.6.
The percentage of experts who considered a particular parameter as essential for classification of juvenile SSc represented the rate of choice. The combination of the weighted score with the rate of choice for each parameter resulted in the final ranking. Recalling the above example, the entire panel of experts considered RP to be essential for the diagnosis of juvenile SSc, and thus the rate of choice for RP was 1 and the final ranking for RP was 98.6. Twenty-one of the initial 48 signs and symptoms were chosen as essential for the diagnosis of juvenile SSc by at least 20% of the investigators. Detailed definitions of the signs and symptoms are presented in Appendix A.
As shown in Table 2, the parameters with the highest final rankings were RP, proximal skin sclerosis/induration, pulmonary fibrosis, sclerodactyly, and renal crisis. Only 3 of these signs/symptoms, RP, proximal skin sclerosis/induration, and sclerodactyly, were present in ≥50% of the patients at the time of diagnosis and were therefore considered to be potential major criteria. The remaining 18 parameters, grouped in 8 categories, were considered to be potential minor criteria (Table 2).
The consensus conference, held in Padua, Italy on June 3–6, 2004, was attended by the same 14 clinicians who participated in phase 2. Two authors (NR, FZ) with expertise in nominal group process and statistical analysis served as facilitators.
At the consensus conference, the clinical profiles of 160 actual patients with a variety of diagnoses were evaluated. One hundred profiles were from patients with juvenile SSc selected from the series of 153 patient profiles collected in phase 1 of the study. The additional 60 patients were controls and included those with conditions overlapping juvenile SSc. The clinical characteristics of the patients evaluated at the consensus conference are summarized in Table 3. Consensus of ≥80% was achieved for 127 of 160 (79%) patients, with 70 of 127 patients (55%) judged as having juvenile SSc, and 57 of 127 patients (45%) judged as not having juvenile SSc. No consensus was reached in 33 of 160 patients (21%).
All 70 patients for whom there was consensus on the diagnosis of juvenile SSc belonged to the original group of 100 patients collected in phase 1 of the study. Among the other 30 original patients with juvenile SSc, 12 were judged by the investigators as not having juvenile SSc, and the diagnoses in 18 patients were undefined. This is understandable, because patients with classic juvenile SSc were defined in phase 1, according to the 1980 ACR criteria (1). These criteria do not necessarily include the presence of proximal skin sclerosis/induration but can be met just by the presence of 2 or more minor criteria, such as sclerodactyly, digital pitting scars, and bibasilar pulmonary fibrosis. In fact, among the 12 children judged as not having juvenile SSc, 5 had sclerodactyly, RP, and arthritis, 5 had digital pitting scars with sclerodactyly, and 2 had pulmonary fibrosis and were ANA-positive. None of the 60 controls were classified as having juvenile SSc (45 were classified as not having the disease, and 15 were undefined).
The 127 patients for whom the investigators reached agreement were then used as the gold standard to test the set of PCC. Ten of the 86 PCC for juvenile SSc showed a Cohen's kappa value >0.7; their corresponding chi-square values, P values, sensitivity, specificity, percent false-positive and false-negative rates, AUC value, and kappa statistics are shown in Table 4. Statistical evaluations for all definitions considered were fully presented to the consensus participants to guide their final face validity choice. The participants were then asked to choose and rank those definitions that, in their opinion, were the more useful in clinical practice, reasonable, and easy to use.
|Criteria||χ2||Sensitivity||Specificity||False negative, %||False positive, %||AUC||K||Initial rank||Final rank|
|37||Proximal skin sclerosis/induration and at least 2 minor criteria||94.1||90||96||11||3||93||0.86||40||34|
|53||RP and proximal skin sclerosis/induration and at least 1 minor criterion||86.9||83||100||17||0||91||0.81||37||30|
|54||RP and proximal skin sclerosis/induration and at least 2 minor criteria||79||79||100||21||0||89||0.77||31||24|
|36||Proximal skin sclerosis/induration and at least 1 minor criterion||96.7||94||93||7||6||94||0.87||15||13|
|11||Two major criteria and at least 1 minor criterion||86.4||94||88||7||10||91||0.82||20||16|
|12||Two major criteria and at least 2 minor criteria||76.7||90||88||12||10||89||0.78||15||12|
|38||Proximal skin sclerosis/induration and at least 3 minor criteria||75.2||79||98||21||2||88||0.75||7||5|
|52||RP and proximal skin sclerosis/induration||86.9||83||100||17||0||91||0.81||5||4|
|80||Proximal skin sclerosis/induration||93.2||94||91||7||7||93||0.86||2||2|
|13||Two major criteria and at least 3 minor criteria||63.9||80||91||21||8||86||0.70||1||1|
Finally, the ranking of the 5 best definitions on face validity was multiplied by Cohen's kappa value to obtain the final ranking. As shown in Table 4, the PCC definition with the highest score was the presence of proximal skin sclerosis/induration and at least 2 minor criteria. On the basis of this result, both RP and sclerodactyly should be considered minor criteria. The definitions that scored second highest (RP and skin sclerosis/induration and at least 1 minor criterion) and third highest (RP and skin sclerosis/induration and at least 2 minor criteria) were very similar to the first choice because they all include skin sclerosis/proximal induration and at least 1 minor criteria, indicating the convergent validity of the process. The final provisional criteria for the classification of juvenile SSc that resulted from this multiphase study are summarized in Table 5.
|Major criterion (required)|
|Proximal skin sclerosis/induration of the skin|
|Minor criteria (at least 2 required)|
|Nailfold capillary abnormalities|
|Digital tip ulcers|
|New-onset arterial hypertension|
|Pulmonary fibrosis (HRCT/radiography)|
|Pulmonary arterial hypertension|
|Carpal tunnel syndrome|
|Tendon friction rubs|
|SSc-selective autoantibodies (anticentromere, anti–topoisomerase I [Scl-70], antifibrillarin, anti–PM-Scl, antifibrillin or anti–RNA polymerase I or III)|
The purpose of our study was to develop uniform nomenclature and criteria for classification of patients with juvenile SSc on the basis of their clinical and laboratory features. The study was conducted by using the Delphi technique and the NGT, both of which are well-recognized consensus formation methodologies specifically designed to combine judgments from a group of experts.
The rigorous procedure followed in this study is consistent with the guidelines recently proposed by the ACR Subcommittee on Classification and Response Criteria (14) and was applied for the first time for the standardization of criteria to evaluate the response to therapy in juvenile SLE (15). With a standardized scientific approach, conducted with the support of evidence-based collection of data from a large multicenter international database, we were able to propose classification criteria for juvenile SSc to be used both in current clinical practice and in research settings to classify patients more homogeneously worldwide. These criteria were essentially defined by the presence of proximal sclerosis/induration of the skin and at least 2 minor criteria chosen from a list of 20 that are grouped in 9 categories of organ system involvement (Table 5).
Until now, pediatricians, rheumatologists, and dermatologists have used adult criteria for SSc, particularly the ACR preliminary classification criteria proposed in 1980 (1), along with the more recent suggestion to add criteria for classification of early or pre-SSc (i.e., before skin changes are evident) (2) systemic sclerosis. The ACR preliminary set of criteria included the presence of a major criterion defined as the symmetric thickening of skin proximal to the MCP and MTP joints or 2 or more minor criteria including sclerodactyly, digital pitting scars or loss of substance from the finger pad, and bibasilar pulmonary fibrosis (1). Recent improvement in our ability to evaluate patients with RP using widefield nailfold microscopy and more precise autoimmune serologic studies has permitted identification of many patients with features of SSc who did not satisfy the ACR preliminary criteria but who, during long-term followup, developed definite SSc (2, 16). Both of these definitions were tested in our study, but their performance in distinguishing children with juvenile SSc from those without juvenile SSc was not satisfactory.
The classification criteria proposed in this study are more restrictive than those for adults. In fact, the major criterion of proximal cutaneous sclerosis, which in adults is enough to define a patient as having SSc, must be accompanied by 2 other minor criteria. This is of particular importance to exclude some pediatric conditions such as eosinophilic fasciitis, progeria, phenylketonuria, or pansclerotic morphea, in which the presence of diffusely thickened skin is the cardinal clinical feature. These conditions may have been incorrectly classified as juvenile SSc according to the 1980 ACR preliminary criteria. Conversely, the minor criteria in our classification system are more numerous and grouped into different categories of organ system involvement. This wider range of signs and symptoms represents a more comprehensive set of clinical and laboratory manifestations of the disease, defines and updates the methods to detect specific organ abnormalities (Appendix A), and introduces serologic results as an important aspect for disease classification.
It has been commonly observed and recently documented (17) that children with RP and capillary changes or autoantibodies do not necessarily develop juvenile SSc, at least in the short-term followup. Therefore, the proposed criteria for pre-SSc may have some limitations that have not yet been adequately confirmed.
Of interest is the limited cutaneous form of SSc (lcSSc), a subtype that is diagnosed more frequently in adults. Our provisional classification criteria include this subtype but do not distinguish it from diffuse cutaneous SSc. This distinction is probably of limited use in children, because lcSSc is reported infrequently before age 16 years (13, 18–20). However, it has now been shown that a substantial number of patients with childhood-onset SSc have their diagnosis confirmed either during adolescence or as young adults (20, 21).
Use of clinical data from real patients in combination with a consensus-based methodology represents an innovative procedure to define classification criteria for juvenile SSc. As mentioned above, these techniques have been used to develop diagnostic criteria or outcome measures for several diseases (5–10). These new classification criteria, which should be validated in prospective studies, will help standardize future research in epidemiology, clinical features, therapy, and outcomes in juvenile SSc.
Dr. Zulian 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. Zulian, Woo, Medsger, Lehman, Matucci Cerinic, Martini, Foeldvari, Ruperto.
Acquisition of data. Zulian, Woo, Athreya, Laxer, Medsger, Lehman, Martini, Ravelli, Russo, Cuttica, de Oliveira, Denton, Cozzi, Foeldvari.
Analysis and interpretation of data. Zulian, Medsger, Lehman, Matucci Cerinic, Ruperto.
Manuscript preparation. Zulian, Woo, Athreya, Laxer, Medsger, Lehman, Martini, Ravelli, Russo, Cuttica, de Oliveira, Denton, Cozzi, Foeldvari, Ruperto, Matucci Cerinic.
Statistical analysis. Ruperto.
Consensus conference participation. Zulian, Woo, Athreya, Laxer, Medsger, Lehman, Martini, Ravelli, Russo, Cuttica, de Oliveira, Denton, Cozzi, Foeldvari, Ruperto, Matucci Cerinic.
We would like to thank Francesca Loro, statistician (University of Padua, Italy) and Marcia Bandeira, MD (Curitiba, Brazil) for the statistical and technical assistance during the consensus conference. We also thank the following investigators for having contributed at various levels to the study: Cristina Vallongo, MD, Maria Teresa Visentin, MD (University of Padua, Padua, Italy), Estevan Castell, MD (Hospital General de Ninos Pedro de Elizalde, Buenos Aires, Argentina), Anne Eberhard, MD, FRACP (Schneider Children's Hospital, Manhasset, New York), Gordana Susic, MD (Institute of Rheumatology, Belgrade, Serbia), Galina Lyskina, MD (Hospital of Childhood Diseases, Moscow, Russia), Dana Nemcova, MD (First Faculty of Medicine and General Faculty Hospital, Praha, Czech Republic), Robert Sundel, MD (Children's Hospital Medical Center, Boston, MA), Fernanda Falcini, MD (Ospedale A. Meyer, Firenze, Italy), Herman Girschick, MD (Kinderklinik der Universitat, Wuerzburg, Germany), Ana Paula Lotito, MD (Children's Institute of the University of Sao Paulo, Sao Paulo, Brazil), Flavio Sztajnbok, MD (Hospital Universitario Pedro Ernesto, Rio de Janeiro, Brazil), S. Al-Mayouf, MD (King Faisal Specialist Hospital and Research Center, Riyadh, Saudi Arabia), Ilonka Orban, MD (National Institute of Rheumatology and Physiotherapy, Budapest, Hungary), Clodoveo Ferri, MD (University of Modena, Modena, Italy).
We also acknowledge the University of Padua, for supporting this study.