Bringing histology to the bedside.

In recent decades, enormous strides have been made in non-invasive imaging of deep tissues and viscera with the development and refinement of ultrasonography, computerized axial tomography, magnetic resonance imaging, and positron emission tomography, to name a few. Progress in non-invasive skin imaging, however, has been slower than in other specialties due in part to the ease with which skin is visually examined and biopsied. The pace of progress in skin imaging, however, has picked up significantly during the past decade. Digital skin imaging, standardized wholebody photography, and dermoscopy are now being used routinely in many centers to help guide the decision to biopsy pigmented skin lesions (Marghoob et al, 2003). More recently, high-resolution optical imaging technologies previously confined to the bench top are finding their way to the bedside (Gonzalez et al, 2003). In this issue of the Journal of Investigative Dermatology, an article from the Medical University of Graz, Austria speaks to the potential utility of reflectance confocal scanning laser microscopy (reflectanceCSLM) for the non-invasive bedside evaluation of pigmented lesions. This study and others like it augur a sea change in the way we will view skin processes clinically at the bedside and in small animal translational research studies in the not too distant future. Among the available non-invasive techniques for skin imaging in vivo, confocal microscopy has been the first to deliver single-cell resolution. A currently available commercial reflectance-CSLM utilizes a diode laser, at the nearinfrared wavelength of 830 nm and a 30, 0.9 numerical aperture water immersion objective lens, to illuminate skin with a low power of 2–10 mW, and visualize a microscopic ‘‘optical section’’ within the superficial layers of the skin (http://www.lucid-tech.com). Nuclear and cellular detail within the epidermis, and microcirculation and collagen with the dermis are imaged in en face sections (oriented parallel to the skin surface) with a measured lateral resolution of 0.7 mm and an axial resolution (i.e., optical section thickness) of 3 mm, to a maximum depth of 100–200 mm (depth of the papillary dermis and superficial reticular dermis) (Rajadhyaksha et al, 1995, 1999). Note that the confocal resolution is similar to that of histology; in particular, the confocal section thickness is comparable to the typically 5 mm thin sections that are prepared for routine histology. Back-scattering of the incident laser light due to variations in the refractive indices of organelles and ultrastructures within the tissue provides contrast in the confocal image (Rajadhyaksha et al, 2004). At a wavelength of 830 nm, melanosomes—due to their size being on the same order as the illumination wavelength, and melanins—due to their high refractive index relative to that of the surrounding skin, produce strong back-scatter. Therefore, melanocytes and basal keratinocytes rich in melanosomes and melanin appear bright and with high contrast in reflectance confocal images (Rajadhyaksha et al, 1995, 1999, 2004). In this issue of the Journal of Investigative Dermatology, Gerger et al (2004) describe excellent sensitivity and specificity achieved for the diagnosis of melanoma using reflectance-CSLM, based on morphologic features that were defined a priori from earlier studies (Langley et al, 2001). In a series of 117 melanocytic skin tumors, consisting of 90 benign and 27 malignant (five in situ, 22 invasive) melanomas, four of the five observers had diagnostic sensitivities and specificities ranging from 93% to 100%. Statistical analysis showed excellent to perfect intra-observer and inter-observer agreement among all five observers for the ten confocal morphologic attributes studied. Of note, none of the five observers (which included two dermatology residents, one dermatologist, and two dermatopathologists) had prior experience in the interpretation of reflectance confocal images. They received a 30 min presentation that instructed them in the confocal morphologic features of benign nevi versus melanomas, using six images of each. None of the images in the instruction set were used in the analysis, and the diagnoses were subsequently made strictly on the basis of the images without knowledge of the clinical setting, clinical appearance, or dermoscopic appearance of the lesions. The authors point out the limitations of their study, which include the pre-selection of the confocal images by a single observer who was not blinded to the clinical/dermoscopic appearance of the lesions and the limited generalizability of the results, given the number of lesions and observers in the data set. A further critique of the study stems from the lack of histology for a subset of the benign melanocytic tumors. Nonetheless, the results provide a ringing endorsement for further assessment of the utility of reflectance-CSLM for the bedside quasi-histologic evaluation of pigmented lesions (Pellacani et al, 2004). Right now, reflectance-CSLM is too expensive and laborious to be considered as a screening tool for melanoma, even if the excellent sensitivity and specificity data presented by Gerger et al (2004) stand up to further scrutiny. With minor advances, however, this technology may become practical for assessing pigmented lesions that have been selected as suspicious on the basis of clinical Abbreviations: reflectance-CSLM, reflectance confocal scanning laser microscopy