Rheumatology Advance Access originally published online on January 30, 2007
Rheumatology 2007 46(5):856-860; doi:10.1093/rheumatology/kel446
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
A new computerized method for the assessment of skin lesions in localized scleroderma
1Department of Pediatrics, 2Department of Information Engineering, 3Department of Dermatology and 4Department of Environmental Medicine and Public Health, University of Padua, Italy.
Correspondence to: F. Zulian, MD, Dipartimento di Pediatria, Università di Padova, Via Giustiniani 3, 35128 Padova, Italy. E-mail: zulian{at}pediatria.unipd.it
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionAknowledgementsReferences |
|---|
Objective. Up to now, no validated tools are in use for the assessment of the skin lesions in localized scleroderma (LS). The aim of this study is to evaluate the performance of a new computerized skin score (CSS) method for the measurement of circumscribed lesions in LS.
Methods. The study consisted of three phases: set up of the CSS technique, measurement of target lesions of LS patients, assessment of intra- and inter-rater reliability. The CSS technique consists in delimitating the indurate lesion on an adhesive transparent film, transferring it over a cardboard and then calculating the affected area with a specifically created software.
The technique was explained to a panel of 10 physicians with different expertise in LS (three paediatric rheumatologists, two dermatologists, five paediatric residents). All participants, singularly and blindly to the others, examined 10 consecutive patients twice after a time interval of at least 6 h. The intra-observer variability was evaluated by the repeatability coefficient and the inter-rater reliability by the intra-class correlation coefficient (ICC).
Results. The repeatability coefficients were good, ranging between 1.90 and 7.03. The mean values of skin scores were not significantly different among the examiners. The ICC for indurate area calculation were high in both the experts (0.97) and the residents (0.91–0.94).
Conclusions. CSS has shown to be a reliable method to assess the skin lesions in patients with LS. It is reproducible, easy to use and, with the support of the CSS software, applicable worldwide.
KEY WORDS: Localized scleroderma, Morphoea, Scleroderma en coup de sabre, Computerized skin score, Outcome measure
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAknowledgementsReferences |
|---|
Localized scleroderma (LS) is a chronic disease characterized by an abnormal accumulation of collagen in the skin and subdermal tissues. Up to now, LS lesions have been assessed by empirical methods biased by significant subjective variability such as simple clinical judgment or serial pictures. The lack of objective and validated measures to determine the disease progression significantly hampered the development of therapeutic trials and outcome studies. The aim of the present study is to introduce a new computerized method for the measurement of circumscribed lesions in LS and to assess its performance.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAknowledgementsReferences |
|---|
The study was designed in three different steps:
- set up of the computerized skin score (CSS) technique,
- measurement of target lesions of LS patients and
- assessment of intra- and inter-rater reliability.
Set up of the CSS technique
The setting up of CSS consisted in the assessment and quality control of the technique for the precise measurement of the indurate skin area without shape approximation. Briefly, an adhesive transparent film (TegadermTM, 3M Health Care, St Paul, MN, USA) was applied over the skin of volunteer patients and the border of indurate lesions was marked with permanent colour. The film was then transferred over a cardboard (A4 paper dimension) and finally digitized with a scanner and recorded as a JPEG image. The scanning resolution (dpi) and the acquired image were recorded.
A dedicated software able to recognize irregular shapes and to calculate exactly their area was developed. Briefly, the algorithm was able to identify the contours of the areas of interest in digitized image was based on an analysis of the hue and saturation contents of the image. Noise suppression ensured the connectedness of the identified contours and the elimination of spurious areas. The internal area was then computed and converted into square centimetres on the basis of the resolution by which the image was previously scanned.
The accuracy of the software was verified by measuring simple geometric shapes such as squares, triangles, ovals and circles with definite known size and then also by measuring irregular shapes.
Measurement of target lesions of LS patients
Ten volunteer patients with different subtypes of LS, defined according to the Mayo clinic classification [1] were scored. Two patients had LS of the face, scleroderma en coup de sabre, one patient had linear scleroderma of the limbs, four plaque morphoea, two generalized morphoea and one patient had a mixed subtype consisting of both linear and plaque lesions. The patients were consecutively and independently assessed by the physicians.
Before applying the CSS to real patients, the technique on how to take the drawings and the possible conventional rules were carefully explained to the panel of physicians taking part in the study. All participants, singularly and blindly to the others, examined the same lesion of the same patients in the same day after a time interval of at least 6 h.
The examination steps, summarized in Fig. 1, were:
- choice of the ‘target lesion’ of skin induration of a volunteer patient,
- application of the adhesive transparent film (TegadermTM) over the skin,
- delimitation of the indurate lesion by palpation,
- drawing of the borders of palpable indurate area with a permanent colour and
- transfer of the film over a cardboard and preparation for scanning.
|
Some conventional rules have been established ‘a priori’:
- For the lesions on the face extending up to the scalp, as in some patients with the en coup de sabre variety of LS, the hairline was chosen as upper border reference.
- For lesions exceeding the dimension of the cardboard, as, for example, in linear scleroderma of the limbs, the limits of the measurement was arbitrarily defined by standardized anatomic benchmarks, such as the upper or lower border of the patella, the tibial protuberance, the malleoli, the radial styloid, the olecran, etc. These limits were recorded to be maintained during consecutive measurements.
- For lesions on body regions, such as the abdomen, where the size of a lesion can change depending on the position of adjacent body segments, standardized positions should be taken.
For example, when measuring lesions on the abdomen, the patient, in supine position, was advised to keep both arms behind his/her head.
Assessment of intra- and inter-rater reliability
The intra- and inter-rater reliability was assessed in a group of 10 physicians evaluating 10 LS patients. Five physicians (three paediatric rheumatologists and two dermatologists) had consolidate experience in examining LS patients. Five physicians were first- or second-year paediatric residents with no previous experience in scleroderma.
Each patient was evaluated by the same examiner twice in the same day after an interval time of at least 6 h.
A comparison between the data obtained by the experienced and inexperienced physicians was then performed.
Statistical methods
To describe the magnitude and direction of the change in the lesions’ size from the first to the second measurement of the same patients at the same session, summary statistics were calculated. To assess intra-rater reliability, according to Bland and Altman [2], the repeatability coefficient and 95% limits of agreement were calculated.
To asses inter-rater reliability we used the intra-class correlation coefficient (ICC) two-way random effect model [3]. The ICC was analysed by calculating and pooling the S.D.s of the skin scores recorded on each patient. The mean S.D. within patient and variance of all measurements on a single patient were calculated providing a quantitative assessment of inter-observer reliability. In addition, as relative measure of dispersion, the coefficient of variation was calculated for each patient and for each factor.
ICC values ranging between 0.75 and 1 were considered as signs of excellent reliability and values ranging between 0.4 and 0.74 as signs of good reliability [4]. The 95% confidence interval (CI) defined the degree of significance and the accuracy of the estimate [3].
Ten patients were involved in the study and, since each examiner evaluated each patient twice, a total of 20 measurements per subject were taken. This allows to estimate the within-subject S.D. with a 95% CI width of 0.2 [5].
All analyses were performed by using the StatsDirect (Version 2.4.5, Stats Direct Ltd, UK) and SPSS (Version 10.1.4, SPSS Inc., Chicago, IL, USA) statistical software.
Patients' and parents' judgement
Patients, if older than 7 years, or parents were asked to give their judgement on the scoring method as far as discomfort and duration. The degree of acceptance of the scoring method was evaluated on a visual analogue scale (VAS) ranging from 0 to 100 mm (0 = no pain at all or very rapid, 100 = very painful or very long lasting).
Physicians’ judgement
Physicians participating to the study were asked to give their judgement on the scoring method as far as feasibility, duration and face validity. These three items were singularly evaluated by VAS (0–100 with 0 = poor and 100 = very good).
Ethical considerations
The study was approved by the Ethical Committee of the Pediatrics Department at the University of Padua. Parents and/or patients signed to give their informed consent.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAknowledgementsReferences |
|---|
Ten patients, 3 males and 7 females, mean age 8.6 years, range 4–22 years, entered the study. The disease duration ranged from 6 months to 2.5 years (mean 1.5 years). All lesions were active and indurate. Inactive atrophic lesions were not considered. As each patient was independently evaluated twice by each examiner, a set of 20 measures (10 in the first section and 10 in the second) was obtained for each patient for a total of 200 measures.
The mean size of the lesions (scores) ranged from 4.48 to 52.85 cm2. The coefficient of variation (CV), defined as the ratio between S.D. and mean value, ranged between 0.08 and 0.20 (Table 1). There was no correlation between the CV and the mean size of the lesions (data not shown).
|
Intra-rater reliability
As shown in Table 2, the repeatability coefficient ranged between 1.90 and 7.03. The limits of agreement, that define the range within which 95% of the differences between two measurements are expected to fall, were acceptable in both groups of raters. However, the variability of repeated measures was mildly higher for the residents (range 3.23–7.03) than for the experts (coefficient of repeatability range 1.90–4.22) (Table 2). This was probably due to the different expertise of the two groups as residents utilized this method for the first time.
|
In no examiner the mean value of the first evaluation was significantly different from the second one, being 0 included in all 95% CI mean differences.
Inter-rater reliability
The agreement among the examiners’ scores was high being the ICC 0.95 (95% CI = 0.88–0.98) in the first set of measures and 0.96 (95% CI = 0.90–0.99) in the second set (Table 3).
|
Interestingly, the mean ICC of the residents went from 0.91, in the first set of measures, up to 0.94 in the second set, showing an improved accuracy probably due to a ‘learning-by-doing’ effect.
The ICC values obtained by the experts (0.97 in both sets of measures) was higher than from those obtained by the residents but not significantly, confirming the face validity of the method.
Procedure duration
The average time needed to conduct the CSS ranged from 2 to 11 min with an average of 5.35 min and a median of 5 min. We did not find any significant variation over time as the experience with the method increased.
Patients and parents judgement
Patients, seemed to accept well this scoring method. The mean degree of discomfort was 13.3 (median 12, range 3–30). The duration of the procedure was judged acceptable: mean VAS 24.7 (median 21, range 7–55).
Physicians’ judgement
The global physician judgement on CSS was good. In particular, the method was considered quite feasible with mean VAS of 90.8 (median 91, range 75–100). Mean VAS for face validity was 87.7 mm (median 88, range 76–100). The time requested to complete the procedure was judged appropriate (mean VAS 90.5, median 92, range 77–100).
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAknowledgementsReferences |
|---|
Up to now no validated tools are in use for the assessment of the skin lesions in LS. As a first effort toward validating an outcome instrument for LS, we assessed a new scoring method, the CSS, for inter- and intra-rater reliability and face validity.
The score performed excellently in all these aspects, which is especially relevant for multicentre studies that are essential for any therapeutic evaluation for such a rare disease.
Most of the lesions in LS consist in circumscribed areas of indurate and thickened skin surrounded by an hyperemic halo. The hyperaemic borders are often well defined and induration can be easily appreciated on palpation. For this reason, CSS performed well in most of the LS subtypes, such as linear scleroderma, plaque morphoea and generalized morphoea. Unfortunately, CSS cannot be used for pansclerotic morphoea or eosinophilic fasciitis, and for atrophic ‘burned-out’ lesions where the borders of the involved skin are not well demarcated.
The intra- and inter-rater reliability were very good both in the group of experts and in the group of residents, meaning that the method is quite simple and easy to implement. Indeed, it is applicable to different lesion sizes and shapes.
The CSS does not need a sophisticated technology but just a plastic adhesive film, a scanner and a personal computer. For this reason, it can be utilized worldwide, which is especially relevant for international collaborative studies.
An important application, included in the software, is the possibility of comparing CSS to the body surface area (BSA), giving the so called standardized-CSS (s-CSS), which is the ratio CSS/BSA in cm2. This parameter is particularly useful when we intend to evaluate the disease progression during childhood. In fact, the observation of an enlarged lesion, over the time, does not mean necessarily disease progression but can be related just to the natural growth process of the child. The s-CSS, calculated in different times, allows to establish if a skin lesion is enlarging or not: in the first instance s-CSS increases, in the latter it remains unchanged.
Over the years, several procedures have been tried to assess the skin lesions in LS. Unfortunately, very few were objective enough and none has been validated prospectively. This has limited the interpretation of most therapeutic studies and up to now represents a bar for the clinical research in this field.
One of the most used scoring system in LS is the modified skin score (MSS) [6, 7]. According to this score the body surface is divided into seven regions: head-neck, trunk, arms, hands, fingers, legs and feet. The skin thickness is assessed on a 0–3-point scale (0, normal; 1, thickened skin; 2, decreased ability; 3, unable to pinch or move skin). The degree of involvement for each body area is also assessed on a 0–3-point scale (0, no involvement; 1, < 33%; 2, 33-67%; 3, > 67%). The sum of numerical units for skin thickness and extension involvement equals the MSS with a maximum score of 42. This method, partially validated only in adult patients, does not evaluate the real size of the lesions, is not reliable for a growing body as in childhood and can be applied only for extended lesions as for the small ones a ceiling effect is possible. In fact, if a 10-year-old boy has a lesion on the thigh that increases from 150 to 300 cm2 in 1 month, the MSS score does not change (score 1) because in both instance it involves <33% of the leg surface area that is around 1500 cm2. It is evident that, although the lesion increases 100% in size, MSS remains unchanged and this is a relevant bias for the clinical judgement.
Other outcome measures are based on the patient or physician global judgement on lesions’ progression/regression by visual analogue scale (VAS) for tightness and itching (0–100) [7–9]. Also in this case, the method is partially objective, does not evaluate the real size of the lesions and can be heavily influenced by psychological influence.
More recently, a follow-up telephone questionnaire was utilized as one of the outcome measures to evaluate the efficacy of methotrexate treatment in children with LS [10]. The bias on the parts of patients and families and the reliance on subjective interpretations of the lesions are evident limitations of this scoring method.
One of the latest proposed outcome measures is the so called DIET, an acronyms that stands for Dyspigmentation, Induration, Erythema and Telangiectasia. Each of these four items is rated on a scale of 0 (none) to 3 (severe) [11]. The innovative aspect of this scoring method is that it includes a wider range of skin manifestations present in different phases of LS. Unfortunately, it does not include the exact extension of the lesions and the body surface area of a growing patient. Indeed, telangiectasias, quite commonly observed in SSc, are rare in LS and particularly in children.
Some authors defined an active disease by the appearance of new lesions, by the increased size of previous lesions or by direct signs of active inflammation such as erythema or warmth [12].
Additional objective measures such as thermography, ultrasound and imaging with MRI are also in use but not validated prospectively, yet [13–17].
Thermography, when associated with clinical examination, is a useful tool in discriminating disease activity. This technique has been shown to have a very high reproducibility but it remains to be seen whether it will truly predict outcome and sub clinical areas likely to progress to serious disease [13, 14].
The application of newer imaging techniques such as MRI and ultrasound also seems promising in supporting clinical management and greater understanding of disease characteristics. MRI is clearly more useful in LS when CNS or eye involvement is suspected but is able to demonstrate the true depth of soft tissue lesions and the degree to which different tissue are involved also in other sites [15].
High frequency ultrasound is also a promising tool. Unfortunately, the operator-dependency, the lack of standardization and the cost of the new probes represent some limitations on its use as an outcome measure for LS [16, 17].
The CSS, is focused on one parameter, the skin induration. It represents the first method to measure the real extension of sclerotic lesions and is easily applicable in growing body as in childhood. Dyspigmentation and erythema can be adjunctive parameters possibly measurable by CSS but this was out of the scope of this study. A computerized scoring method, similar to CSS, was used in 1997 by Huzelmann et al. [18] in a double-blind placebo-controlled trial. Unfortunately, the reliability of the method was not evaluated and no further applications have been tried later on.
In conclusion, the CSS represents a simple, reliable and widely applicable method to assess disease progression in LS. Once prospectively validated, this outcome instrument might be used worldwide and significantly affect the development of multicentre therapeutic trials and outcome studies in this rare and often disabling condition.
|
Aknowledgements |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAknowledgementsReferences |
|---|
We thank Roberto Mancin, informatics engineer and Rosaria Cuomo, administrative secretary for their technical assistance and IL VOLO, Association for childhood rheumatic diseases for having partially supported this project.
The authors have declared no conflicts of interest.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAknowledgementsReferences |
|---|
- Peterson LS, Nelson AM, Su WPD. (1995) Subspecialty clinics: rheumatology and dermatology. Classification of morphea (localized scleroderma). Mayo Clin Proc 70:1068–76.[Abstract]
- Bland MJ and Altman DG. (1999) Measurement agreement in method comparison studies. Stat Methods Med Res 8:135–60.
[Abstract/Free Full Text] - McGraw KO and Wong SP. (1996) Forming inferences about some intraclass correlation coefficients. Psychol Methods 1:30–46.[CrossRef][Web of Science]
- Fleiss J. (1986) Reliability of measurements. In Fleiss J (Ed.). The design and analysis of clinical experiments(John Wiley & Sons, New York) pp. 2–31.
- Bland MJ and Altman DG. (1996) Statistics notes: measurement error. Br Med J 313:744.
[Free Full Text] - Zachariae H, Bjerring P, Halkier Sorensen L, Heickendorff L, Sondergaard K. (1994) Skin scoring in systemic sclerosis: a modification-ralations to subtypes and the aminoterminal propeptide of type III procollagen. Acta Derm Venereol 74:444–6.[Web of Science][Medline]
- Seyger MMB, van den Hoogen FHJ, de Boo T, De Jong EMGJ. (1997) Reliability of two methods to assess morphea: skin scoring and the use of a durometer. J Am Acad Dermatol 37:793–6.[CrossRef][Web of Science][Medline]
- Kreuter A, Gambichler T, Breuckmann F, et al. (2005) Pulsed high-dose corticosteroids combined with low-dose methotrexate in severe localized scleroderma. Arch Dermatol 141:847–52.
[Abstract/Free Full Text] - Kreuter A, Hyun J, Stucker M, et al. (2006) A randomized controlled study of low-dose UVA1, medium-dose UVA1, and narrowband UVB phototherapy in the treatment of localized scleroderma. J Am Acad Dermatol 54:440–7.[CrossRef][Web of Science][Medline]
- Fitch PG, Retting P, Burnham JM, et al. (2006) Treatment of pediatric localized scleroderma with methotrexate. J Rheumatol 33:609–14.
[Abstract/Free Full Text] - Dytoc M, Ting PT, Man J, et al. (2005) First case series on the use of imiquimod for morphoea. Br J Dermatol 153:815–20.[CrossRef][Web of Science][Medline]
- Uziel Y, Feldman BM, Krafchik BR, et al. (2000) Methotrexate and corticosteroid therapy for pediatric localized scleroderma. J Pediatr 136:91–5.[CrossRef][Web of Science][Medline]
- Martini G, Murray KJ, Howell KJ, et al. (2002) Juvenile-onset localized scleroderma activity detection by infrared thermography. Rheumatology 41:1178–82.
[Abstract/Free Full Text] - Birdi N, Shore A, Rush P, Laxer RM, et al. (1992) Childhood linear scleroderma: a possible role of thermography for evaluation. J Rheumatol 19:968–73.[Web of Science][Medline]
- Liu P, Uziel Y, Chuang S, et al. (1994) Localized scleroderma: imaging features. Pediatr Radiol 24:207–9.[CrossRef][Web of Science][Medline]
- Seidenari S, Conti A, Pepe P, et al. (1995) Quantitative description of echographic images of morphea plaques as assessed by computerized image analysis on 20 MHz B-scan recordings. Acta Derm Venereol 75:442–5.[Web of Science][Medline]
- Cosnes A, Anglade MC, Revux J, Radier C. (2003) Thirteen-megahertz ultrasound probe: its role in diagnosing localized scleroderma. Br J Dermatol 148:724–9.[CrossRef][Web of Science][Medline]
- Hunzelmann N, Anders S, Fierlbeck G, et al. (1997) Double-blind, placebo-controlled study of intralesional interferon gamma for the treatment of localized scleroderma. J Am Acad Dermatol 36:433–5.[CrossRef][Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||