Introduction Dermatoscopy, which allows detection of atypical structures before they become clinically patent, has shown its effectiveness in early melanoma diagnosis. Four cases of thin melanomas with an Inversed Network (IN) have been recently reported. The objective of our study was to establish the diagnostic value of IN-a non pigmented reticulated network on a brown or red background-for the early detection of melanoma (image1).
Methods Numeric photos of the pigmented lesions included in this study come from ambulatory patients seen in our consultation of melanoma screening and follow-up of the University Hospital of Geneva. We retrospectively screened all the pigmented lesions of our dermatoscopic database (Fotofinder), looking for the IN sign independently of other signs of atypia. Eighty seven lesions presented an inversed network (IN+) in the whole data base containing approximately 25 000 lesions. Seven lesions IN+ without available histology were excluded. The histological analyses of the selected lesions were reviewed by the same pathologist. To establish the sensibility of this sign for melanoma diagnosis, we reviewed the dermatoscopic images of every melanoma excised in our consultation over the same period (60). We did the same for dysplastic naevi, taking a large number of them (1053) to obtain a representative control group (IN−).
Results On 80 lesions IN+, 47.5%, (in situ: 16.3%, invasive: 31.3%) were melanomas. 48.8% were dysplastic nevi, with severe (16.3%), marked (18.8%) and moderate atypia (13.8%) respectively. 3.8% nevi were without atypia (image2). Sensibility of the IN for the diagnosis of melanoma, independently of other signs of atypia, is 38.8%. Specificity of IN for melanoma is 96.5%. Invasive melanomas have a median Breslow of 0.35 mm. The positive predictive value (VPP) for melanomas and dysplasic nevus is 96.3% and for melanoma only 47.5%. If we add to melanomas, dysplastic nevus with severe atypia, histologically difficult to distinguish from in situ melanomas, VPP is 63.8%. Predictive negative value of IN for melanoma is >94.7%. Positive likelihood ratio is >11.0 and negative <0.6. Finally, histological analysis revealed that IN is due to an inversion of the usual distribution of the dermo-epidermic pigment (image3), as other authors have initially observed .
Conclusion This study confirms the diagnostic value of IN and clarifies its histological substrate. Practically, IN rationalizes to excise the lesion, especially when sequential digital dermoscopy is not available.