A preliminary disease severity score for juvenile systemic sclerosis

Authors


Abstract

Objective

To develop a preliminary disease severity score for juvenile systemic sclerosis (SSc).

Methods

We conducted an evidence- and consensus-based study that included the following 5 phases: 1) prospective collection of data regarding the demographic and clinical characteristics of patients with diffuse juvenile SSc who were followed up for at least 4 years or until death; 2) blinded evaluation of the disease course profiles of these patients by experts in juvenile SSc, so that patient profiles with a defined clinical course could be used as the gold standard for the score validation phase; 3) definition of candidate severity indices to be included in potential scores; 4) selection of the pediatric severity score with the best statistical performance, as determined by its ability to classify individual patients as having improvement or worsening of disease compared with baseline values or the previous evaluation; 5) validation of the efficiency of the selected score in patients with a mild, moderate, or severe disease course and comparison with the Medsger severity score for adults with SSc.

Results

Thirty-five patients classified as having a mild (n = 17), moderate (n = 10), or severe (n = 8) disease course entered the study. The selected pediatric SSc score, defined as the Juvenile Systemic Sclerosis Severity Score (J4S), included indices of 9 organ systems each scored on a scale of 0–4. To weight the importance of the involvement of different organ systems, a coefficient of severity was introduced. Compared with the modified Medsger severity score, the J4S performed significantly better in detecting change in severity, both in patients with a moderate disease course (0.89 versus 0.52) and in patients with a severe disease course (0.82 versus 0.75).

Conclusion

The J4S is a valid instrument to assess disease severity in juvenile SSc.

Although juvenile systemic sclerosis (SSc; scleroderma) is rare in children, it represents one of the most severe conditions in pediatric rheumatology (1). It is therefore necessary to assess the severity of this disease early in its course in order to determine the most effective therapeutic regimen. During the last few years, several systems for scoring disease severity in adults with SSc have been proposed (2–5), but none has been validated in pediatric patients with SSc. This may be attributable to the rarity of SSc in children (6) and, until recently, to the lack of pediatric patient–specific classification criteria (7). Indeed, existing disease severity scores for adults are not fully applicable in pediatric patients due to their specific symptoms at disease presentation, the different manner of assessing the function of various organs, and the presence of growth-related issues. The aim of this study was to develop a preliminary disease severity score for juvenile SSc.

PATIENTS AND METHODS

The study was conducted using an evidence- and consensus-based methodology that included 5 phases, as follows: 1) prospective collection of data regarding the characteristics of patients classified as having juvenile SSc on the basis of the Pediatric Rheumatology European Society/American College of Rheumatology/European League Against Rheumatism classification criteria (7); 2) evaluation of the disease course profiles of these patients, performed in a blinded manner by 3 experts in juvenile SSc, with patient profiles with a defined clinical course used as the gold standard during phase 5; 3) selection of pediatric disease severity indices according to the clinical evidence–based experience of the physicians participating in the study and to the existing data in the literature on adult SSc (Delphi method 1) for the construction of hypothetical multidimensional scores (Delphi method 2); 4) selection of the pediatric SSc disease severity score (from the list developed during the previous phase) with the best statistical performance according to its ability to classify individual patients as having improvement or worsening of disease when compared with baseline values or the previous evaluation; 5) evaluation of the statistical efficiency of the selected score in patients with 3 different disease courses (mild, moderate, or severe) by using the gold standard defined in phase 2 and comparison with the existing score for adult SSc (Figure 1).

Figure 1.

Flow diagram showing the consecutive phases of the study. MSS = Medsger severity score.

Data collection (phase 1).

We performed prospective data collection of the demographic, clinical, laboratory, and treatment characteristics of patients with juvenile SSc who were followed up at 5 pediatric rheumatology centers (4 in Italy and 1 in Argentina) for at least 4 years or until death. Demographic characteristics included sex, age, race, disease duration since the first non–Raynaud's phenomenon symptom, age at the onset of the first non–Raynaud's phenomenon symptom, and age at diagnosis.

The clinical features used to define possible disease severity parameters included the following:

  • 1General indices, such as weight, height, body surface area (BSA); body mass index (BMI), defined as centile value by age and sex (8); and laboratory indices, such as erythrocyte sedimentation rate, C-reactive protein, alanine aminotransferase, aspartate aminotransferase, white blood cell count, hemoglobin, platelet count, creatine kinase, aldolase, and lactate dehydrogenase. These parameters were compared with reference values for the group ages 5–16 years (9).
  • 2Vascular involvement, such as Raynaud's phenomenon, number of ulcerations, and number of digital scars.
  • 3Cutaneous involvement, defined according to the modified Rodnan skin thickness score (2).
  • 4Osteoarticular involvement, defined as the presence of tendon friction rubs, limited joint range of motion (decreased joint range of motion >25% of normal), or active arthritis.
  • 5Muscular involvement, quantified by the Childhood Myositis Assessment Scale (CMAS), which assesses muscle strength, endurance, and function (10). The CMAS (scale of 0–52) is based on the ability of the child to perform specific tasks scored by an observer.
  • 6Gastrointestinal involvement, including symptoms referred to gastroesophageal reflux (pyrosis and dysphagia), malabsorption (serum folate and/or vitamin B12 levels, stool trypsin, carbohydrate permeability test), and abnormal esophageal transit (assessed by barium swallowing, 24-hour pH-metry, esophageal scintiscan, or manometry).
  • 7Respiratory involvement, defined as forced vital capacity (FVC), diffusing capacity for carbon monoxide (DLCO), standard chest radiography, high-resolution computed tomography (HRCT), and systolic pulmonary arterial pressure (PAP) measured by echocardiography.
  • 8Cardiac involvement, evaluated by electrocardiography and echocardiography.
  • 9Renal involvement, defined by the degree of functional impairment in the presence or absence of scleroderma renal crisis. Creatinine clearance (glomerular filtration rate [GFR]) was calculated according to reference values for the age group 0–18 years (11).
  • 10Neurologic involvement, defined by the presence of convulsions, neuropathy, or paralysis.

Autoantibody profiles (antinuclear antibodies and SSc-specific autoantibodies [anti–Scl-70, centromere, anti–PM-Scl, fibrillin, fibrillarin]) were used only to classify the patients and not as severity indices. Treatment was also recorded and included the type of drug, dosage, and duration. Data were recorded at least every 6 months, when clinical worsening occurred, or at the onset of a major complication such as renal crisis, acute cardiac insufficiency, and/or respiratory insufficiency.

Patient profiles (phase 2).

The clinical and laboratory features of each patient, which were summarized in an Excel sheet, were independently evaluated by 3 experts in juvenile SSc (FLT, GM, FZ). Abnormal clinical or laboratory parameters were represented, in an Excel sheet, with different colors (green = normal, yellow = slightly abnormal, red = abnormal). According to these features, each patient was judged as having a mild, moderate, or severe disease course. Agreement was considered to be identical judgment among the 3 experts, and disagreement occurred when the judgment of at least 1 of the experts was different from that of the other 2 experts.

In case of discordance after the first individual evaluation, the doubtful cases were collegially discussed in order to obtain a consensus on the type of disease course. In case of persistent disagreement at the second run, the patient was designated as having an “indefinable” course and was excluded from the following analysis. Patient profiles with a defined clinical course were used as the gold standard for the following score validation phase.

Severity score definition (phase 3).

On the basis of the first Delphi survey, involving 10 physicians participating in the study (12), we selected a group of parameters that theoretically better defined the severity of the disease (Delphi method 1). Consensus was obtained when a parameter was selected by at least 8 of 10 physicians (consensus >80%).

On the basis of clinical importance and sensitivity to change, a group of 8 different severity scores was developed with various combinations of clinical and laboratory indices. Considering the distinctive age-related clinical, laboratory, and instrumental parameters of pediatric patients with juvenile SSc as well as the growth issue, we adapted the scale developed by Medsger et al and the international SSc study group (3, 5) to the pediatric age group by including different combinations of clinical and laboratory indices while maintaining the 0–4-point scores for involvement of each organ. The pediatric scores were proposed and evaluated by the physicians participating in the study, by a second Delphi procedure (Delphi method 2). The majority of evaluators agreed that for cutaneous, vascular, and cardiac involvement, the same indices and scores described in the Medsger severity score (MSS) should be included (3).

For the growth parameters (weight, height, BSA, and BMI), reductions in centile lines were classified as class 1 (1 centile line lower), class 2 (2 centile lines lower), class 3 (3 centile lines lower), and class 4 (4 centile lines lower) with respect to the baseline visit.

Hemoglobin was the only hematologic parameter that was considered sensitive enough to be used as an index of both inflammation and malnutrition. The score was assigned on the basis of a range of arbitrarily defined hemoglobin values, as follows: 1 = 10–11.4 gm/dl, 2 = 9–9.9 gm/dl, 3 = 7–8.9 gm/dl, and 4 = <7 gm/dl.

Osteoarticular involvement was scored as follows: 2 = presence of functional limitation (defined as limited range of motion of >25% of the normal range), and 4 = presence of active arthritis or tendon friction rubs in at least 1 joint. Muscular involvement was objectively quantified by the CMAS (10), and the scoring classes were defined as 39–51 (score 1), 26–38 (score 2), 13–25 (score 3), and 0–12 (score 4).

Gastrointestinal involvement was weighted on the basis of clinical and instrumental parameters. According to the scoring system, 1 = presence of at least 1 of the following: symptoms referred to gastrointestinal reflux (dysphagia, burning, etc.), distal esophageal hypomotility, or gastroesophageal reflux; 2 = presence of mid and/or upper esophageal hypomotility, as demonstrated by barium swallowing or manometry; 3 = demonstration of the presence of malabsorption; and 4 = need for hyperalimentation.

Respiratory involvement was scored on the basis of abnormal changes in at least 1 of the following 4 domains: pulmonary function tests, FVC and DLCO, respiratory tract imaging (chest radiography and HRCT), and systolic PAP. FVC and DLCO, defined as the percent predicted, included the following 4 pathologic scoring classes: 70–79%, 50–69%, <50%, and O2 dependence. The radiologic severity score was based on the presence of alveolitis on HRCT (score 1), fibrosis on HRCT (score 2), and fibrosis on standard radiography (score 3). Systolic PAP, which was evaluated indirectly by color Doppler echocardiography, was referred to the pediatric age group normal values and scored as follows: 1 = systolic PAP <30 mm Hg, 2 = systolic PAP 31–45 mm Hg, 3 = systolic PAP 46–75 mm Hg, and 4 = systolic PAP ≥75 mm Hg (13). The presence of O2 dependence was assigned a score of 4, irrespective of pulmonary function test results or imaging abnormalities.

The degree of renal function impairment was scored as follows: 1 = GFR of 75–89 ml/minute, 2 = GFR of 50–74 ml/minute, 3 = GFR of 10–49 ml/minute, and 4 = end-stage renal failure.

Selection of the pediatric severity score (phase 4).

The pediatric severity scores that obtained the highest preference using Delphi method 1 (indices defined by the clinical experience of the participating physicians and existing data in the literature on adult SSc) were applied to the profiles of patients with juvenile SSc in order to select the single score that showed the best statistical performance in terms of face validity and sensitivity to clinical change. Spearman's rank correlation coefficient was used to verify the correlation between the assigned clinical opinion and the mean score of each patient.

The mean score assigned to each patient, the mean change between each visit and the previous one (X̄TN − X̄TN-1), and the mean of the difference between the mean score during followup and the score at baseline ([X̄T1…TN] − X̄T0) were calculated for each of the 3 categories of severity. Moreover, the distribution of patients with respect to the score according to the clinical course (mild, moderate, severe) was evaluated.

Finally, as a reference index for the sensitivity to change of the proposed scores, we used the standardized response mean (SRM), which was calculated by dividing the mean change by the standard deviation of the change scores, according to the following formula: SRM = X̄change/ SDmath image. Ninety-five percent confidence intervals (95% CIs) for these indices were also calculated.

High SRM values indicate high sensitivity in identifying changes in the clinical status of the patients. In particular, a value of ≥0.80 corresponds to high responsiveness or sensitivity to change, a value of 0.5–0.79 indicates moderate responsiveness, and a value of 0.2–0.49 or <0.2 corresponds to low or very low responsiveness, respectively (14).

Severity score validation in various clinical settings (phase 5).

Finally, we validated the efficiency of the selected score in various clinical settings (mild, moderate, or severe disease course) and compared the statistical efficiency of the pediatric score with that of the modified MSS (which is actually the most-used tool for the assessment of disease severity in adult patients with SSc) (3, 5), by using the gold standard profiles defined in phase 2.

RESULTS

Forty-two patients with diffuse juvenile SSc who were recruited from 5 pediatric rheumatology centers (4 in Italy and 1 in Argentina) entered into the study. A validation analysis was conducted on 35 patients (32 females and 3 males) who were classified as having a mild (n = 17), moderate (n = 10), or severe (n = 8) disease course. In 7 patients, the disease course was “indefinable,” and these patients were excluded from the analysis. The demographic and clinical characteristics of the patients are summarized in Table 1.

Table 1. Demographic and clinical characteristics of the patients with juvenile systemic sclerosis, according to the disease course
CharacteristicDisease course
Mild (n = 17)Moderate (n = 10)Severe (n = 8)Overall (n = 35)
  1. * Except where indicated otherwise, values are the mean.

Female sex, %94.110075.091.4
Age at disease onset, years8.58.610.28.9
Age at diagnosis, years10.210.410.910.4
Diagnostic delay, years1.61.80.81.5
Followup, years5.64.83.64.7
Autoantibodies, %    
 Antinuclear antibodies88.28087.585.7
 Antitopoisomerase5.84037.522.9
 Anticentromere11.6005.7
Treatment, %    
 D-penicillamine5.82037.517.6
 Corticosteroids58.89010077.1
 Methotrexate58.8607562.9
 Cyclophoshamide03062.522.9
 Calcium channel blockers88.2705074.3
 Prostanoids35.3402534.3
 Other (mycophenolate  mofetil, imatinib, etc.)17.6405031.4

On the basis of the Delphi evaluation among physicians participating in the study, the clinical/laboratory indices that obtained the highest scores, as far as sensitivity to change, and were suggested to be included in multidimensional scores are summarized in Table 2. For neurologic involvement, consensus was <80%, and therefore it was discarded from further consideration.

Table 2. Clinical and laboratory indices that obtained the highest scores in phase 3
Organ systemComponents of the severity index
GeneralBody mass index
 Hemoglobin
VascularRaynaud's phenomenon requiring vasodilators
 Number of digital scars
 Number of ulcers/gangrene
CutaneousModified Rodnan skin thickness score
OsteoarticularArthritis
 Limited range of motion
 Tendon friction rubs
MuscularChildhood Myositis Assessment Scale
GastrointestinalSymptoms of gastroesophageal reflux
 Abnormal esophageal transit
 Malabsorption
RespiratoryForced vital capacity
 Diffusing capacity for carbon monoxide
 Pulmonary artery systolic pressure by Doppler  echocardiography
 Standard chest radiography
 High-resolution computed tomography of the chest
CardiacElectrocardiogram
 Echocardiogram
 Clinical signs of congestive heart failure
RenalCreatinine clearance (glomerular filtration rate)

According to the judgment of the study investigators, consensus of >80% was reached for 2 of the 8 proposed pediatric scores. In order to give greater weight to those organs whose involvement indicates greater disease severity, we modified these 2 scores by introducing coefficients based on the seriousness of involvement of a particular organ. The following coefficients were proposed: 2 for cardiac and pulmonary involvement; 1 for general, cutaneous, vascular, gastrointestinal, and renal parameters; and 0.5 for muscular and osteoarticular parameters.

Among the 4 final pediatric severity scores, the one that had the best statistical profile and face validity after testing in the 35 patient profiles was named the Juvenile Systemic Sclerosis Severity Score (J4S) (Table 3). The J4S is composed of 16 indices grouped into 9 sections corresponding to 1 general and 8 specific organ systems, with a possible overall score ranging from 0 to 40. The J4S was comparable, as far as correlation between the assigned clinical status and the mean score value for the patient, to the MSS, with a high Spearman's rank correlation coefficient for both scores (for J4S, rs = 0.80 [95% CI 0.64–0.89]; for MSS, rs = 0.86 [95% CI 0.73–0.92] [P < 0.001]). We performed a total of 210 measurements (mean of 6.0 per patient), using both the J4S and the MSS.

Table 3. Juvenile Systemic Sclerosis Severity Score*
 0 (normal)1 (mild)2 (moderate)3 (severe)4 (end-stage)Maximum possible score
  • *

    BMI = body mass index; Hgb = hemoglobin; MRSS = modified Rodnan skin thickness score; CMAS = Childhood Myositis Assessment Scale; GI = gastrointestinal; GERD = gastroesophageal reflux disease; DLCO = diffusing capacity for carbon monoxide; FVC = forced vital capacity; HRCT = high-resolution computed tomography; sPAP = estimated systolic pulmonary artery pressure by Doppler echocardiography; EKG = electrocardiogram; LVEF = left ventricular ejection fraction.

  • Any of the parameters defines the score.

  • Each score should be multiplied by 0.5 to obtain the final score.

  • §

    Any of the parameters defines the score. Each score should be multiplied by 2 to obtain the final score.

  • The glomerular filtration rate (GFR) was used to measure creatinine clearance. Values are calculated as follows: for males ages 0–12 years, 0.55 × height (cm)/creatinine (mg/dl); for males ages 12–18 years, 0.7 × height (cm)/creatinine (mg/dl); for females ages 0–18 years, 0.55 × height (cm)/creatinine (mg/dl); for males older than age 18 years, (140 − age) × weight (kg)/(72 × creatinine [mg/dl]); for females older than age 18 years, (140 − age) × weight (kg)/(72 × creatinine [mg/dl]) × 0.85.

GeneralBMI ≥ baseline Hgb >11.5 gm/dlBMI < 1st centile Hgb 10–11.4 gm/dlBMI < 2nd centile Hgb 9–9.9 gm/dlBMI < 3rd centile Hgb 7–8.9 gm/dlBMI < 4th centile Hgb <7 gm/dl4
VascularNo Raynaud'sRaynaud's requiring vasodilatorsDigital tip scarsDigital tip ulcerationsDigital gangrene4
SkinMRSS 0MRSS 1–14MRSS 15–29MRSS 30–39MRSS >404
OsteoarticularNo articular involvement Presence of limited range of motion Presence of arthritis and/or tendon friction rub2
MuscleNormal proximal muscle strengthCMAS 39–51CMAS 26–38CMAS 13–25CMAS 0–122
GINormal results of proximal GI tract investigationsGI symptoms Distal esophageal hypomotility GERD on 24-hour Ph-metry or scintiscanMedium and/or high esophageal hypomotilityMalabsorption syndromeHyperalimentation4
Respiratory§DLCO >80%DLCO 70–79%DLCO 50–69%DLCO <50%O2 dependence8
 FVC >80%FVC 70–79%FVC 50–69%FVC <50%  
 Normal HRCTGround glass changes on HRCTHoneycomb changes on HRCTFibrosis on radiography  
 sPAP <30 mm HgsPAP 31–45 mm HgsPAP 46–75 mm HgsPAP >75 mm Hg  
Cardiac§Normal EKGEKG conduction defectArrhythmiaArrhythmia requiring treatmentCongestive heart failure8
 LVEF >50%LVEF 45–49%LVEF 40–44%LVEF 30–39%LVEF <30% 
RenalGFR >90 ml/minuteGFR 75–89 ml/minuteGFR 50–74 ml/minuteGFR 10–49 ml/minuteEnd-stage renal failure4

We performed 119 assessments in the group of patients with a mild disease course (mean of 7.0 per patient), 52 assessments in the group with a moderate disease course (mean of 5.2 per patient), and 39 assessments in the group with a severe disease course (mean of 4.9 per patient). The distribution of the scores assigned to the patients, according to the 3 clinical courses, showed a lower dispersion of the MSS with respect to the distribution of the J4S (Figure 2).

Figure 2.

Distribution of scores assigned to patients with juvenile systemic sclerosis, according to the 3 clinical disease courses. A, Mild disease course. B, Moderate disease course. C, Severe disease course. MSS = Medsger severity score; J4S = Juvenile Systemic Sclerosis Severity Score.

The means of scores assigned to the patients at the last evaluation, according to the 3 clinical disease courses, are shown in Figure 3A. To evaluate the sensitivity to change of the 2 scores, we calculated the median of the differences between subsequent visits (Tn − Tn-1), which resulted in a value of 0.0 in patients with a mild disease course (MSS = −0.13), 0.55 in patients with a moderate disease course (MSS = 0.23), and 1.02 in patients with a severe disease course (MSS = 0.81) (Figure 3B). For a mild disease course, the differences between the 2 scores were minimal. For moderate and severe disease courses, the J4S detected higher variations compared with the MMS.

Figure 3.

Scores assigned to patients with juvenile systemic sclerosis according to the 3 clinical disease courses. A, Scores assigned at the last evaluation. Values are the mean ± 95% confidence interval. B, Median of the differences in scores between 2 visits. MSS = Medsger severity score; J4S = Juvenile Systemic Sclerosis Severity Score.

By applying the SRM to quantify the sensitivity to change of the 2 scores, we obtained low negative results for both scores in the group of patients with a mild disease course (MMS = −0.33, J4S = −0.12), and obtained high SRM values, especially for the J4S, in the group of patients with a moderate disease course (J4S = 0.89, MSS = 0.52) and those with a severe disease course (J4S = 0.82, MSS = 0.75). Thus, the J4S showed better statistical performance than the MSS.

DISCUSSION

Systemic sclerosis is uncommon in children (6); pediatric cases account for only 3% of all cases of SSc (1). Therefore, other than 2 recent large studies (15, 16), only small case series describing the clinical features of the juvenile form of SSc have been published (17, 18). The rarity of childhood SSc has restricted our ability to develop criteria to assess disease activity and severity, which will have great usefulness in future therapeutic trials (19–21).

Disease activity is usually defined as the aspect of disease that varies over time and is potentially reversible spontaneously or following treatment. Assessment of disease severity is appropriate in children and adolescents with juvenile SSc, because the effect of disease on organ function includes a combination of both reversible (activity) and irreversible (damage) components (3). The need to define a disease severity score comes from the difficulty in judging the disease course of patients with juvenile SSc in daily clinical practice, because the disease often has an indolent course without clear signs of inflammation. Indeed, the clinical features attributable to vascular and connective tissue fibrosis are more difficult to appreciate and quantify than are those attributable to inflammation (21).

The aim of this study was to explore the efficiency of a multidimensional score (including growth parameters, skin, and internal organ involvement) that may express clinical judgment in a single score, taking into account the great variability of the clinical manifestations of the disease. Therefore, we devised a severity score by adapting the scale developed by Medsger et al (3, 5) to the pediatric age group, following an evidence- and consensus-based methodology. The result was the Juvenile Systemic Sclerosis Severity Score. The J4S was shown to be a simple and reliable instrument to assess disease severity in juvenile SSc. Compared with the MSS, which includes 9 separate severity scores (1 score for each organ assessed), the J4S weights the involvement of various organs on the basis of their clinical importance and arrives at a single score.

A significant correlation between the average score of patients at the time of the last visit and the assigned clinical course was observed for both the MSS and the J4S. To better test the statistical performance of the J4S and the MSS, we compared the average score assigned to each individual patient, grouped according to the clinical disease course. The J4S reached higher values than the MSS, demonstrating better sensitivity for quantifying disease severity.

It is clear that a high score that does not vary in several evaluations is less important, as far as clinical decision-making, than a lower score that increases during 2 subsequent visits. To evaluate the sensitivity to change of the score, we considered both the mean of the difference between one score and the subsequent one and the mean of the difference between each score during followup and the initial one. The J4S demonstrated a greater sensitivity to change for both moderate and severe disease courses compared with the MSS. This finding confirmed the pediatric specificity of the J4S as well as its construct validity, which parallels, in many ways, that of the MSS. Indeed, the J4S performs better than the MSS, especially in patients with a moderate-to-severe disease course, for whom clinical decision-making is more important.

The J4S may have several applications for the pediatric age group. In observational studies, it can be used to compare study populations and identify potentially reversible aspects of SSc. Indeed, it has the potential to be used in daily clinical practice to guide decision-making and for regulatory documentation. Finally, given that research into new treatments for juvenile SSc is rapidly advancing, the J4S could be used to determine eligibility, as a measure of outcome, for upcoming clinical trials.

In conclusion, the J4S is a reliable and sensitive instrument to assess SSc in childhood. The use of variables adapted to the pediatric age group and the inclusion of coefficients to appropriately weight the importance of the involvement of different organs represent innovative features.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. 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 conception and design. La Torre, Martini, Russo, Vittadello, Zulian.

Acquisition of data. Martini, Russo, Katsicas, Corona, Calcagno, Falcini, Zulian.

Analysis and interpretation of data. La Torre, Martini, Russo, Vittadello, Zulian.

Acknowledgements

We acknowledge Professor Balu H. Athreya (Jefferson University, Philadelphia) for the critical review of the manuscript and Marta Balzarin, MD (University of Padua) for helping with data collection and analysis.

Ancillary