Is Confocal Microscopy a Valuable Tool in Diagnosing Nodular Lesions?

A Study of 140 Cases

C. Longo; F. Farnetani; S. Ciardo; A.M. Cesinaro; E. Moscarella; G. Ponti; I. Zalaudek; G. Argenziano; G. Pellacani

The British Journal of Dermatology. 2013;169(1):58-67.

Abstract and Introduction

Abstract

Background Nodular lesions pose diagnostic challenges because nodular melanoma may simulate all kinds of melanocytic and nonmelanocytic lesions. Reflectance confocal microscopy (RCM) is a novel technique that allows visualization of the skin at nearly histological resolution although limited laser depth penetration hampers visualization of the deep dermis.

Objectives We sought to assess whether the diagnostic accuracy of RCM was comparable to histopathology for the diagnosis of nodular lesions, and to identify possible limitations of this technique.

Methods We retrospectively evaluated 140 nodules by means of RCM while blinded from the histopathological diagnosis. At the end of the study the patient codes were broken and the evaluations were matched with histopathological diagnosis before performing statistical analysis.

Results The study consisted of 140 nodular lesions (23 'pure' nodular melanomas, nine melanoma metastases, 28 BCCs, six invasive SCCs, 32 naevi, 14 seborrhoeic keratoses, 17 dermatofibromas, five vascular lesions and six other lesions). RCM correctly diagnosed 121 of 140 lesions (86·4%); eight of 140 (5·7%) lesions revealed discordance between histopathology and confocal microscopy. Eight of the 140 (5·7%) cases were not evaluable by means of RCM due to the presence of ulceration or hyperkeratosis and three cases showed a nonspecific pattern. Interestingly, confocal microscopy reached a 96·5% sensitivity and 94·1% specificity (area under curve 0·970) (95% CI 0·924–1·015) (P < 0·001) for the diagnosis of melanoma.

Conclusions The study is retrospective and lesions were not included on the basis of their diagnostic difficulty. Despite the limited laser depth penetration of RCM, this imaging tool represents an effective instrument in diagnosing nodular lesions; however, for fully ulcerated lesions or when a marked hyperkeratosis is present, biopsy should always be performed. Prospective studies on difficult-to-diagnose nodules should be performed to analyse further the pros and cons of RCM in skin cancer diagnosis.

Introduction

Reflectance confocal microscopy (RCM) is a relatively new imaging tool that provides horizontal scanning of the skin at nearly histological resolution.^[1,2] It has been applied in several skin diseases and predominantly in skin oncology,^[3–8] where it has been proven to increase the specificity of difficult-to-diagnose lesions.

Reflectance confocal microscopy uses a diode laser as a source of monochromatic and coherent light, which penetrates into the skin and illuminates a small point inside the tissue.^[1] Then, the light is reflected, goes through a small pinhole and generates an image in the detector. The pinhole collects reflected light emanating only from the focal region (confocal); thus, light from out-of-focus planes is rejected at the pinhole. The contrast in RCM images relies on the differences in the refractive index of the tissue; therefore, structures with a higher refractive index (melanin, collagen and keratin) appear brighter in RCM. The deeper a skin section is the more laser power is needed to penetrate the skin. The laser power employed by RCM (near-infrared wavelength of 830 nm) causes no damage to tissue, but limits the imaging depth to 200 μm, which corresponds to the papillary dermis. Higher laser power would provide an increased signal and contrast but would be dangerous for the skin (higher temperature) and the resolution will be poor.

The limited laser penetration depth of the instrument has led to the assumption that nodular lesions with a prevailing dermal component are not suitable for RCM diagnosis, as an evaluation of the 'informative' dermal component is hampered. However, nodular lesions belonging to both melanocytic and nonmelanocytic tumours have been explored by means of RCM,^[2,9,10] albeit extensive studies on this topic are still lacking.

The aim of our study is to elucidate the feasibility of RCM in diagnosing nodular lesions to define the role of this tool in this field in clinical practice.

1 2 3 4

Next Section

Table 1. Reflectance confocal microscopy (RCM) criteria described according to distinct diagnostic categories
Categories	RCM criteria	References
Melanoma and melanoma metastasis		[4, 7–9, 12–15]
Epidermis	Honeycombed or cobblestone pattern
	Regular
	Irregular
	Disarray of epidermis
	Pagetoid spread
	Pagetoid cell shape
	Roundish shape
	Dendritic shape
	Pleomorphic shape
	Pagetoid cell distribution
	Focal
	Localized
	Widespread
Dermo epidermal junction	Nonspecific architecture
Dermo epidermal junction	Cytological atypia (moderate, severe)
Dermis	Sheet-like structures
	Dermal nesting
	Dense nests
	Dense and sparse nests
	Cerebriform nests
	Prominent vascularity: enlarged vessels covering more than 50% of the lesion surface
	Inflammatory infiltrate covering more than 50% of the lesion surface (bright spots and plump bright cells)
Naevi		[2, 18, 19, 21, 22]
Epidermis	Honeycombed or cobblestone pattern
	Regular
	Irregular
	Disarray of epidermis
Dermo epidermal junction	Ringed pattern
	Clod pattern
	Meshwork
	Junctional nests
Dermis	Sheet-like structures
	Dermal nesting
	Dense nests
	Dense and sparse nests
	Cerebriform nests
Basal cell carcinoma		[10, 23–28]
Epidermis	Honeycombed or cobblestone pattern
	Regular
	Irregular
	Disarray of epidermis
	Ulceration or erosions
Dermo epidermal junction	Cauliflower architecture
Dermo epidermal junction	Nonspecific architecture
Dermis	Dark silhouettes
	Tightly packed cells
	Bright filaments within tumour islands
	Prominent vascularity: enlarged vessels covering more than 50% of the lesion surface
	Inflammatory infiltrate covering more than 50% of the lesion surface (bright spots and plump bright cells)
Squamous cell carcinoma		[29–31]
Epidermis	Honeycombed or cobblestone pattern
	Regular
	Irregular
	Disarray of epidermis
	Ulceration or erosions
	Scales
	Keratin inclusion/plugs
Dermis	Prominent vascularity: enlarged vessels covering more than 50% of the lesion surface
Dermis	Inflammatory infiltrate covering more than 50% of the lesion surface (bright spots and plump bright cells)
Seborrhoeic keratosis		[20] (pp. 243–5),[32, 33]
Epidermis	Cobblestone pattern
	Regular
	Irregular
	Scales
	Superficial crypts > 3
	Keratin inclusion/plugs
Dermo epidermal junction	Polycyclic papillary contours
Dermis	Enlarged vessels covering more than 50% of the lesion surface Inflammatory infiltrate covering more than 50% of the lesion surface (bright spots and plump bright cells)
Dermatofibroma		[17]
Dermo epidermal junction	Ringed pattern
Dermis	Presence of bright collagen fibres
Vascular lesions		[20] (pp. 249–50)
Dermis	Vascular lagoon
Xanthogranuloma		[16]
Dermis	Large highly refractive atypical cells some of them with pleomorphic nuclei, corresponding to Touton cells

Table 2. Absolute and relative frequencies of diagnostic categories in histopathology and reflectance confocal microscopy (RCM) analysis
Categories	Histopathological diagnosis	RCM diagnosis concordant with histopathology	RCM diagnosis discordant with histopathology	Kappa value
Nodular melanoma	23/140 (16·4%)	21/23 (91·3%)	2/23 (8·6%) not evaluable	0·955
Melanoma metastasis	9/140 (6·4%)	7/9 (77·7%)	2/9 (22·2%) nonspecific	0·955
Basal cell carcinoma	28/140 (20%)	27/28 (96·4%)	1/28 (3·5%) not evaluable	0·977
Squamous cell carcinoma	6/140 (4·3%)	5/6 (83·3%)	1/6 (16·6%) discordant	0·905
Naevi	32/140 (22·8%)	30/32 (93·7%)	1/32 (3·1%) discordant; 1/32 (3·1%) nonspecific	0·978
Seborrhoeic keratosis	14/140 (10%)	13/14 (92·8%)	1/14 (7·1%) discordant	0·958
Dermatofibromas	17/140 (12·1)	15/17 (88·2%)	1/17 (5·8%) discordant; 1/17 (5·8%) not evaluable	0·963
Vascular lesions	5/140 (3·6%)	3/5 (60%)	1/5 (20%) not evaluable; 1/5 (20%) discordant	0·853
Others	1/140 (0·7%) clear cell acanthoma	–	1/1 (100%) discordant	–
	1/140 (0·7%) angiomyxoma	–	1/1 (100%) not evaluable	–
	2/140 (1·42%) pyogenic granulomas	–	2/2 (100%) not evaluable	–
	1/140 (0·7%) apocrine hidrocystoma	–	1/1 (100%) discordant	–
	1/140 (0·7%) xanthogranuloma	–	1/1 (100%) discordant	–
Total	140	121 (86·4%)	8/140 (5·7%) not evaluable; 8/140 (5·7%) discordant; 3/140 (2·1%) nonspecific

Table 3. Frequencies of significant reflectance confocal microscopy features for each category, as evaluated by univariate analysis
Category	NM+Mets	Others
Category	n = 28	n = 102
Disarray of epidermis	39·3%	7·8%
Pagetoid cells (round + dendritic)	14·3%	2%
Pagetoid distribution (widespread)	25%	1%
Cytological atypia
Moderate	50%	6·9%
Severe	42·9%	3·9%
Dermal nests
Dense and sparse	3·6%	1%
Cerebriform	57·1%	0
Sheet-like structures	42·9%	1%
Prominent vascularity	64·3%	34·3%
Category	BCC	Others
Category	n = 27	n = 103
Irregular honeycombed/cobblestone pattern	74·1%	29·1%
Ulceration	44·4%	13·6%
Cauliflower architecture	59·3%	0
Basaloid islands
Dark silhouettes	44%	1%
Tightly packed aggregates	55·6%	0
With bright filaments	29·6%	0
Collagen	51·9%	24·8%
Category	SCC	Others
Category	n = 6	n = 124
Disarray of epidermis	83·3%	11·3%
Scales	33·3%	3·2%
Prominent vascularity	83·3%	38·7%
Category	Naevi	Others
Category	n = 32	n = 98
Regular honeycombed/cobblestone pattern	75%	37·8%
Meshwork	21·9%	6·1%
Clods	53·1%	8·2%
Category	Seb K	Others
Category	n = 14	n = 116
Regular honeycombed/cobblestone pattern	71·4%	44%
Superficial crypts > 3	21·3%	4·3%
Keratin plugs	71·3%	19·8%
Polycyclic papillary contours	35·7%	0·9%
Category	DF	Others
Category	n = 16	n = 114
Regular honeycombed/cobblestone pattern	81·3%	42·1%
Ringed architecture	81·3%	22·8%
Collagen	75%	24·6%
Category	Vascular lesions	Others
Category	n = 4	n = 126
Regular honeycombed/cobblestone pattern	100%	45·2%
Vascular lagoons	75%	1·6%
Prominent vascularity	75%	39·7%

NM+Mets, malignant melanocytic tumours (including nodular melanomas); DF, dermatofibroma; Seb K, seborrhoeic keratosis; BCC, basal cell carcinoma; SCC, squamous cell carcinoma.

Table 4. Discriminant analysis of NM+Mets vs. all other diagnosis, BCC vs. all other diagnosis and naevi vs. all other diagnosis
	Unstandardized canonical discriminant function coefficients
NM+Mets features
Pagetoid distribution	0·545
Cytological atypia (moderate and severe)	1·346
Dermal nesting	0·872
BCC features
Cauliflower architecture	2·040
BCC islands (dark silhouettes/tightly packed basaloid islands)	3·937
Bright filaments within tumour islands	1·789
Collagen	0·433
SCC features
Irregular honeycombed/cobblestone pattern	0·646
Disarray of the epidermis	3·435
Inflammatory infiltrate	0·874
Naevi features ^a
Meshwork architecture	1·254
Clods architecture	1·610
Junctional nests	2·182
Seb K features
Scales	2·161
Superficial crypts	0·334
Keratin plugs	0·274
Polycyclic papillary contours	4·398
Dermatofibroma features
Ringed pattern	1·495
Collagen	1·713
Vascular lesions features
Vascular lagoons	5·223

NM+Mets, malignant melanocytic tumours (including nodular melanomas); BCC, basal cell carcinoma; SCC, squamous cell carcinoma. ^aNaevi vs. all other lesions including Spitz naevi.