Sensitivity and specificity of reflectance-mode confocal microscopy for in vivo diagnosis of basal cell carcinoma: A multicenter study - 24/08/11

Doi : 10.1016/j.jaad.2004.06.028

Sarita Nori, MD ^a,^{^{Current affiliation for Dr Nori is the Department of Dermatology and Skin Surgery, Roger Williams Medical Center, Providence, Rhode Island. Dr González holds a joint appointment at the Dermatology Service, Memorial Sloan Kettering Cancer Center, New York, New York.}}, Francisca Rius-Díaz, PhD ^a, Jesus Cuevas, MD ^b, Mark Goldgeier, MD ^e, Pedro Jaen, MD ^c, Abel Torres, MD ^d, Salvador González, MD, PhD ^a,^⁎
^a From the Wellman Laboratories of Photomedicine at Massachusetts General Hospital
^b Pathology
^c Dermatology
^d Services at Hospital Guadalajara at Alcala University in Madrid; Division of Dermatology at Loma Linda University Medical School
^e Dermatology and Dermatologic Surgery, Rochester

Reprint requests: Salvador González, MD, PhD, Wellman Laboratories of Photomedicine, Department of Dermatology, Massachusetts General Hospital, 50 Blossom St, Bartlett 3, Boston, MA 02114.

Boston, Massachusetts; Madrid, Spain; Loma Linda, California; and Rochester, New York

Abstract

Background

The current standard diagnostic procedure for basal cell carcinoma (BCC) is histologic examination after invasive biopsy. Reflectance-mode confocal microscopy (RCM) offers noninvasive high-resolution imaging of human skin in vivo.

Objective

The objective of this study was to explore the sensitivity and specificity of RCM for diagnosis of BCC.

Methods

This was a retrospective study of RCM images from 4 institutions of 152 skin lesions representing a variety of benign and malignant diagnoses. These 152 lesions were examined clinically, with biopsies recorded for all the 83 BCCs detected. Based on a previous study, a set of 5 histologically correlated confocal imaging criteria for diagnosing BCC was established, eg, the presence of elongated monomorphic nuclei. Blinded retrospective analysis of the images from the 152 lesions was carried out by a single novice reviewer to determine the sensitivity and specificity of these 5 RCM criteria for diagnosing BCC. The accuracy of combining the probability of BCC based on examination of clinical photographs with the predicted probability of BCC based on confocal criteria was also evaluated.

Results

The presence of two or more criteria is 100% sensitive for the diagnosis of BCC, and with 4 or more RCM criteria present the specificity was 95.7% and sensitivity was 82.9%. These results were found to have little variability across study sites and across BCC subtypes. The combination of RCM with photography-based predictions of clinical probability of BCC significantly improved the accuracy for noninvasive diagnosis of BCC.

Conclusion

RCM offers a sensitive and specific tool for the noninvasive diagnosis of BCC in vivo.

Le texte complet de cet article est disponible en PDF.

Plan

Methods and materials

Patients

Confocal imaging

Data analysis

Combining clinical assessment with confocal microscopy

Results

Characteristics of the data set

Sensitivity and specificity of confocal microscopy criteria

Sensitivity and specificity by study site

Sensitivity and specificity by BCC subtype

Combining clinical probability with confocal microscopy

Discussion

Supported in part by National Institutes of Health National Institute for Occupational Safety and Health grant RO1 OH04029.
Disclosure: Dr Goldgeier was a member of the Scientific Advisory Board of Lucid Inc, the company that commercialized the imaging system used for real-time reflectance-mode confocal laser scanning microscopy. In addition, the imaging system used in some of the centers was originally provided by this company.