Micantonio Tamara, Neri Luca, Longo Caterina, Grassi Simone, Di Stefani Alessandro, Antonini Ambra, Coco Valeria, Fargnoli Maria Concetta, Argenziano Giuseppe, Peris Ketty
Department of Dermatology, University of L'Aquila. L'Aquila, Italy.
Department of Community and Clinical Sciences, University of Milan, Milan, Italy.
Eur J Dermatol. 2018 Apr 1;28(2):162-168. doi: 10.1684/ejd.2018.3246.
The clinical and dermoscopic diagnosis of facial lentigo maligna (LM) and pigmented actinic keratosis (PAK) remains challenging, particularly at the early disease stages. To identify dermoscopic criteria that might be useful to differentiate LM from PAK, and to elaborate and validate an automated diagnostic algorithm for facial LM/PAK. We performed a retrospective multicentre study to evaluate dermoscopic images of histologically-proven LM and PAK, and assess previously described dermoscopic criteria. In the first part of the study, 61 cases of LM and 74 PAK were examined and a parsimonious algorithm was elaborated using stepwise discriminant analysis. The following eight dermoscopic criteria achieved the greatest discriminative power: (1) light brown colour; (2) a structureless zone, varying in colour from brown to brown/tan, to black; (3) in-focus, discontinuous brown lines; (4) incomplete brown or grey circles; (5) a structureless brown or black zone, obscuring the hair follicles; (6) a brown (tan), eccentric, structureless zone; (7) a blue structureless zone; and (8) scales. The newly developed algorithm was subsequently validated using an additional series of 110 LM and 75 PAK cases. Diagnostic accuracy was 86.5% (κ: 0.73, 95% CI: 0.63-0.83). For the diagnosis of LM vs PAK, sensitivity was 82.7% (95% CI: 75.7-89.8%), specificity was 92.0% (95% CI: 85.9-98.1%), positive predictive value was 93.8% (95% CI: 89.0-98.6%), and negative predictive value was 78.4% (95% CI: 68.4-86.5%). This algorithm may represent an additional tool for clinicians to distinguish between facial LM and PAK.
面部恶性雀斑样痣（LM）和色素性日光性角化病（PAK）的临床及皮肤镜诊断仍具有挑战性，尤其是在疾病早期阶段。旨在确定可能有助于区分LM与PAK的皮肤镜标准，并制定和验证一种用于面部LM/PAK的自动诊断算法。我们进行了一项回顾性多中心研究，以评估经组织学证实的LM和PAK的皮肤镜图像，并评估先前描述的皮肤镜标准。在研究的第一部分，检查了61例LM和74例PAK，并使用逐步判别分析制定了一种简约算法。以下八项皮肤镜标准具有最大的判别能力：（1）浅棕色；（2）无结构区，颜色从棕色到棕褐色再到黑色不等；（3）清晰、不连续的棕色线条；（4）不完整的棕色或灰色圆圈；（5）无结构的棕色或黑色区域，遮挡毛囊；（6）棕色（棕褐色）、偏心、无结构区；（7）蓝色无结构区；（8）鳞屑。随后，使用另外110例LM和75例PAK病例对新开发的算法进行了验证。诊断准确性为86.5%（κ：0.73，95%CI：0.63 - 0.83）。对于LM与PAK的诊断，敏感性为82.7%（95%CI：75.7 - 89.8%），特异性为92.0%（95%CI：85.9 - 98.1%），阳性预测值为93.8%（95%CI：89.0 - 98.6%），阴性预测值为78.4%（95%CI：68.4 - 86.5%）。该算法可能为临床医生区分面部LM和PAK提供了一种额外的工具。
