Clinical and Laboratory Studies Received: December 27, 2007 Accepted: February 19, 2008 Published online: July 25, 2008
Dermatology 2008;217:231–234 DOI: 10.1159/000148249
Reproducibility of Dermoscopic Features of Congenital Melanocytic Nevi Vito Ingordo a Silvia S. Iannazzone b Francesco Cusano b Luigi Naldi c a
Department of Dermatology, Hospital Military Centre ‘G. Venticinque’, Taranto, b Department of Dermatology, ‘G. Rummo’ Hospital, Benevento, and c Coordinating Center, Italian Group for Epidemiological Research in Dermatology, Department of Dermatology, Ospedali Riuniti, Bergamo, Italy
Key Words Congenital melanocytic nevi ⴢ Dermoscopy ⴢ Dermoscopic features ⴢ Reproducibility ⴢ Interobserver agreement ⴢ Intraobserver agreement
Abstract Background: The dermoscopic features of congenital melanocytic nevi (CMN) have recently been described, and some structures have been proposed as distinctive of these moles. Objective: The aim of our study was to evaluate the agreement between observers, i.e. the reproducibility, in the identification of these features. Method: Fifty-two CMN, randomly selected from 127 male subjects carrying at least 1 CMN, were examined by a 1st medium-experienced observer. The image recording was performed according to the dermoscopic patterns of the Consensus Net Meeting on Dermoscopy and to the features proposed for CMN by Seidenari et al. The presence/absence of these features was recorded in a specifically developed patient’s card, and the most typical dermoscopic features of the lesions were also photographed and observed again in this form by the same investigator and by a 2nd basic-experienced observer. The reproducibility of the dermoscopic features assessed directly and on photographs by the 1st observer (intraobserver agreement) and the agreement between the 1st and the 2nd observers judging nevi on photographs (interobserver agreement) were estimated by Cohen’s kappa statistics. Re-
© 2008 S. Karger AG, Basel 1018–8665/08/2173–0231$24.50/0 Fax +41 61 306 12 34 E-Mail
[email protected] www.karger.com
Accessible online at: www.karger.com/drm
sults: Intraobserver agreement was highly satisfactory for all the dermoscopic features, with the exception of focal thickening of network lines and follicles (satisfactory), and hyperpigmented areas and target vessels (fairly satisfactory), which were more frequently identified on Dermaphot pictures. Interobserver agreement was highly satisfactory for network, globules/dots, blotches, focal hypopigmentation, skin furrow hypopigmentation and follicles; it was satisfactory for perifollicular hypopigmentation, vessels and target vessels, and barely satisfactory for focal thickening of network lines, target globules and hyperpigmented areas; finally, it was fairly satisfactory for target network. Conclusion: The identification of dermoscopic features of CMN seems to show a good reproducibility, with a satisfactory intra- and interobserver agreement. Other studies involving a higher number of experienced and less experienced observers are requested to confirm these results. Copyright © 2008 S. Karger AG, Basel
Introduction
The dermoscopic features of congenital melanocytic nevi (CMN) have recently been described, and some structures have been proposed as distinctive of these moles [1–4]. The aim of our study was to evaluate the agreement between the observers, i.e. the reproducibility, in the identification of these features. Capt (I) Vito Ingordo, MD Chief of Department of Dermatology Italian Navy Main Hospital ‘G. Venticinque’ IT–74100 Taranto (Italy) Tel. +39 099 775 0209, Fax +39 099 775 0344, E-Mail
[email protected]
Table 1. Frequency (%) of dermoscopic features identified on 52 CMN and intraobserver/interobserver agreement (Cohen’s ) Dermoscopic features
1st observer by Intraobserver direct dermoscopy, % agreement
Network Target network Focal thickening of network lines/network1 Globules/dots Target globules2 Hyperpigmented areas Blotches Focal hypopigmentation Skin furrow hypopigmentation Follicles Perifollicular hypopigmentation/follicles3 Vessels Target vessels
88.5 67.3 43.4 50 57.6 5.8 23.1 53.8 46.1 73.1 55.3 65.3 15.4
0.83 0.95 0.74 0.81 0.97 0.41 0.87 0.98 0.92 0.72 1 0.81 0.52
1st observer by Dermaphot, %
84.6 69.2 56.8 59.6 58.1 17.3 19.1 51.9 42.3 100 40.4 73.1 34.6
Interobserver agreement
0.86 0.36 0.50 0.83 0.59 0.46 0.93 0.83 0.95 1 0.68 0.78 0.79
2nd observer by Dermaphot, %
80.7 30.3 33.3 67.3 37.1 11.5 17.3 59.6 44.2 100 56.8 80.7 44.2
: ≤0.20 = unsatisfactory; 0.21–0.40 = barely satisfactory; 0.41–0.60 = fairly satisfactory; 0.61–0.80 = satisfactory; 0.81–1 = highly satisfactory. 1 The number of observations of focal network thickening was compared to the number of lesions with network. 2 The number of observations of target globules was compared to the number of lesions with globules/dots. 3 The number of observations of perifollicular hypopigmentation was compared to the number of observed follicles.
Subjects and Methods The study subjects were enrolled from 23,354 young men assessed at the Draft Council’s Medical Unit of the Italian Navy in Taranto for evaluating their psychophysical fitness to compulsory recruitment and referred to the Department of Dermatology of the Italian Navy Hospital for dermatological examination [5]. In this population, 157 CMN were diagnosed, with a prevalence of 1:148 or 0.67% (Bayesian 95% confidence interval 0.57–0.79) [6]. The dermoscopic features of 127 moles examined among this group had previously been assessed in another survey [4]. The sample considered in the present study consisted of 52 subjects, aged 18–22 years (median age 19), with 52 CMN, randomly selected among the aforesaid 127 male subjects each carrying 1 CMN. The lesions were medium-sized (diameter from 61.5 to ^19.9 cm) in 41 cases (78.8%) and large (diameter 620 cm) in 11 cases (21.2%). CMN were located on the chest (13 cases), on the abdomen (10 cases), on the upper back (12 cases), on the lower back (15 cases), on the upper limb (2 cases), on the lower limb (7 cases) and on the shoulder (5 cases). In 11 cases, the lesions were located on several contiguous anatomical sites (e.g. shoulder and upper limb, or lower back and lower limb). In 26 moles (50%), terminal hairs were present. Five CMN (9.6%) showed a zosteriform (i.e. segmental) feature. Lesions were examined by a 1st medium-experienced observer (V.I.), using the Heine Delta 10 Dermatoscope (Heine Optotechnik, Herrsching, Germany) with a 10-fold magnification. The image recording was performed according to the dermoscopic patterns of the Consensus Net Meeting on Dermoscopy [7] and to the features proposed for CMN by Seidenari and Pellacani [1] and Seidenari et al. [2, 3]. Because of the lower magnification of our instrumentation as regards vid-
232
Dermatology 2008;217:231–234
eodermoscopy employed by the aforesaid authors, only some of the dermoscopic features which had proved suitable for the description of CMN were considered. The presence/absence of these features was described and recorded in a specifically developed patient’s card. The most typical features of the lesions were also photographed using Dermaphot (Heine Optotechnik; min. 2/max. 4 pictures) and were observed again in this form by the same investigator and by a 2nd basic-experienced observer (S.S.I.). The dermoscopic features detected in this way were also recorded. The reproducibility of the dermoscopic features assessed directly and on photographs by the 1st observer (intraobserver agreement) was evaluated by estimating Cohen’s kappa statistics. In the same way, we assessed the agreement between the 1st and the 2nd observers judging nevi on photographs (interobserver agreement).
Results
The morphological descriptors employed for image evaluation are listed in table 1. The definition of the patterns refers to the Consensus Net Meeting on Dermoscopy [7]. Among the other features which had proved suitable for the description of CMN [1–4], target globules (light brown globules with a central dot), target network (network meshes centered by a dot) and target vessels (a network mesh centered by a vessel) were employed. Focal thickening of network lines, sometimes also showing Ingordo /Iannazzone /Cusano /Naldi
small dot-like eversions, and skin furrow and perifollicular hypopigmentation were also recorded. Because of the difficulty in measuring the diameter of globules with our instrumentation, no subdivision in small globules, large globules and dots was provided. Table 1 shows the frequency of dermoscopic features recorded in the three sets of observations and the kappa values (intra- and interobserver agreement). No streaks, regression structures and blue-whitish veil were identified by the observers. Intraobserver agreement was highly satisfactory for all the dermoscopic features, with the exception of focal thickening of network lines and follicles (satisfactory), and hyperpigmented areas and target vessels (fairly satisfactory), which were more frequently identified on Dermaphot pictures. Interobserver agreement was highly satisfactory for network, globules/dots, blotches, focal hypopigmentation, skin furrow hypopigmentation and follicles; it was satisfactory for perifollicular hypopigmentation, vessels and target vessels, and barely satisfactory for focal thickening of network lines, target globules and hyperpigmented areas; finally, it was fairly satisfactory for target network.
Discussion
The reproducibility of dermoscopic features of pigmentary lesions has been evaluated firstly in the literature with regard to the diagnosis of melanocytic and nonmelanocytic pigmented skin lesions in the Consensus Net Meeting on Dermoscopy [7]. In this study, the intraobserver and interobserver agreement among 40 experienced clinicians who used both the pattern analysis and the main diagnostic algorithms was pointed out. The interobserver reproducibility among the observers was assessed according to the method of Fleiss [8, 9] and Fleiss et al. [10] to calculate the kappa statistics for multiple ratings for a patient. The participating clinicians were able to classify pigmented skin lesions with fair to good interobserver agreement (kappa values ranging from 0.47 to 0.55) and nearly excellent to perfect intraobserver agreement (kappa values ranging from 0.72 to 1). Other studies have also evaluated the reproducibility of dermoscopic features of pigmentary lesions with regard to the clinical diagnosis of melanoma. These studies, based on pattern analysis and diagnostic algorithms, provide kappa values for interobserver agreement ranging from 0.52 to 0.55 in a small group of experienced and nonexperienced dermatologists [11], a good interobserver reproducibility ( = 0.53) among 150 clinicians with different expertise in dermoscopy [12] and kappa values ranging Reproducibility of Dermoscopic Features of Congenital Melanocytic Nevi
from 0.27 to 0.33 in residents [13]. In our study, the kappa values were altogether higher, showing a good intra- and interobserver reproducibility in the dermoscopic examination of CMN. With regard to intraobserver agreement, only hyperpigmented areas and target vessels showed a less than satisfactory reproducibility in our study, while, in interobserver assessment, target network showed a limited agreement. In a recently published work, Changchien et al. [14] observed additional dermoscopic features in CMN which they added to those proposed by Seidenari and Pellacani [1] and Seidenari et al. [2, 3]: the milia-like cysts (characteristic dermoscopic finding of seborrheic keratosis) and the haloed globules, which are globules surrounded by a hypopigmented halo. The authors structurally differentiated haloed globule from target globule, which was initially described by Seidenari et al. as a clear globule containing a central dot. They proposed also to redefine the term ‘target network’, previously described as a reticular network centered by a dot, encompassing in this term the reticular network that contains dots, globules or blood vessels. Because the design of our study was originally outlined on the grounds of the dermoscopic features proposed by Seidenari et al., we did not use these new structures in our evaluation. We conclude that the identification of dermoscopic features of CMN seems to show a good reproducibility, with a satisfactory intra- and interobserver agreement. Other studies involving a higher number of experienced and less experienced observers, also using the dermoscopic morphological descriptors introduced in the meanwhile, are requested to confirm these results.
References
1 Seidenari S, Pellacani G: Surface microscopy features of congenital nevi. Clin Dermatol 2002;20:263–267. 2 Seidenari S, Martella A, Pellacani G: Polarized light-surface microscopy for description and classification of small and mediumsized congenital melanocytic naevi. Acta Derm Venereol 2003; 83:1–6. 3 Seidenari S, Pellacani G, Martella A, Giusti F, Argenziano G, Buccini P, Carli P, Catricalà C, De Giorgi V, Ferrari A, Ingordo V, Manganoni AM, Peris K, Piccolo D, Pizzichetta MG: Instrument-, age-, and site-dependent variations of dermoscopic patterns of congenital melanocytic nevi: a multicenter study. Br J Dermatol 2006; 155:56–61.
Dermatology 2008;217:231–234
233
4 Ingordo V, Iannazzone SS, Cusano F, Naldi L: Dermoscopic features of congenital melanocytic nevus and Becker nevus in an adult male population: an analysis with a 10-fold magnification. Dermatology 2006; 212: 354– 360. 5 Ingordo V, Gentile C, Iannazzone SS, Cusano F, Naldi L: The ‘EpiEnlist’ project: a dermo-epidemiologic study on a representative sample of young Italian males. Prevalence of selected pigmentary lesions. J Eur Acad Dermatol Venereol 2007;21:1091–1096. 6 Ingordo V, Gentile C, Iannazzone SS, Cusano F, Naldi L: Congenital melanocytic nevus: an epidemiologic study in Italy. Dermatology 2007;214:227–230. 7 Argenziano G, Soyer HP, Chimenti S, Talamini R, Corona R, Sera F, Binder M, Cerroni L, De Rosa G, Ferrara G, Hofmann-Wellenhof R, Landthaler M, Menzies SW, Pehamberger H, Piccolo D, Rabinovitz HS, Schiffner R, Staibano S, Stolz W, Bartenjev I, Blum A, Braun R, Cabo H, Carli P, De Giorgi V,
234
View publication stats
8 9 10
11
Fleming MG, Katz B, Kenet RO, Kittler H, Kreusch J, Malvehy J, Mazzocchetti G, Oliviero M, Ozdemir F, Peris K, Perotti R, Perusquia A, Pizzichetta MA, Puig S, Rao B, Rubegni P, Saida T, Scalvenzi M, Seidenari S, Stanganelli I, Tanaka M, Westerhoff K, Wolf IH, Braun-Falco O, Kerl H, Nishikawa T, Wolff K, Kopf AW: Dermoscopy of pigmented skin lesions: results of a consensus meeting via the Internet. J Am Acad Dermatol 2003;48:679–693. Fleiss JL: Measuring nominal scale agreement among many raters. Psychol Bull 1971; 76:378–382. Fleiss JL: Statistical Methods for Rate and Proportions. New York, Wiley & Sons, 1981. Fleiss JL, Nee JCM, Landis JR: The large sample variance of kappa in the case of different sets of raters. Psychol Bull 1979; 86: 974–976. Soyer HP, Argenziano G, Zalaudek I, Corona R, Sera F, Talamini R, Barbato F, Baroni A, Cicale L, Di Stefani A, Farro P, Rossiello L,
Dermatology 2008;217:231–234
Rocco E, Chimenti S: Three-point checklist of dermoscopy: a new screening method for early detection of melanoma. Dermatology 2004;208:27–31. 12 Zalaudek I, Argenziano G, Soyer HP, Corona R, Sera F, Blum A, Braun RP, Cabo H, Ferrara G, Kopf AW, Langford D, Menzies SW, Pellacani G, Peris K, Seidenari S, Dermoscopy Working Group: Three-point checklist of dermoscopy: an open Internet study. Br J Dermatol 2006;154:431–437. 13 Carli P, Quercioli E, Sestini S, Stante M, Ricci L, Brunasso G, De Giorgi V: Pattern analysis, not simplified algorithms, is the most reliable method for teaching dermoscopy for melanoma diagnosis to residents in dermatology. Br J Dermatol 2003; 148:981–984. 14 Changchien L, Dusza SW, Agero ALC, Korzenko AJ, Braun RP, Sachs D, Usman MHU, Halpern AC, Marghoob AA: Age- and sitespecific variation in the dermoscopic patterns of congenital melanocytic nevi. Arch Dermatol 2007;143:1007–1014.
Ingordo /Iannazzone /Cusano /Naldi
