Enhancing Skin Cancer Diagnosis with Dermoscopy

Synopsis

Dermoscopy increases the sensitivity for skin cancer detection, decreases the number of benign lesions biopsied for each malignant diagnosis, and enables the diagnosis of thinner melanomas compared to naked eye examination. Three meta-analyses have all identified that dermoscopy improved diagnostic accuracy for melanoma when compared to naked eye exam. In addition, studies have established that dermoscopy can aid in the detection of keratinocyte carcinomas. Dermoscopy triage algorithms have been developed to help novices decide when a biopsy or a referral is most appropriate. In this chapter we illustrate the dermoscopic features that assist in identifying melanoma and keratinocyte carcinomas.

Introduction

Dermoscopy has been shown to increase sensitivity for skin cancer detection, decrease the benign to malignant biopsy ratio, and allow for the diagnosis of thinner melanomas compared to naked eye exam (NEE).^1–3 In 2009, a survey of academic dermatologists and chief residents in US dermatology training programs demonstrated that 84% of attending dermatologists used dermoscopy in daily practice and 90.2% of chief dermatology residents received dermoscopy training as part of their curricula.⁴ These findings represented a significant increase in training and use of dermoscopy compared to a similar survey performed ten years prior.⁵ The use of dermoscopy has also increased among nondermatologist physicians who actively participate in skin cancer management, such as family physicians.^6–8

Despite the growing number of practitioners incorporating dermoscopy into their daily practices, there remain significant barriers, like lack of training resources, preventing its widespread adoption.⁴ Dermoscopic teaching methodologies, which include pattern analysis, diagnostic algorithms, and simplified triage algorithms, continue to emerge. They aim to provide an entry point in educating beginners in dermoscopy. Here we review the diagnostic utility of dermoscopy for detection of melanoma and keratinocyte carcinoma (KC; also known as nonmelanoma skin cancers) and outline the specific dermoscopic features that can help discriminate these cancers from benign lesions. A review of dermoscopic teaching methodologies, including triage algorithms, is also provided.

Dermoscopy for the diagnosis and management of cutaneous melanoma

In 2001, the first meta-analysis of the diagnostic power of dermoscopy found dermoscopy to be more accurate than NEE alone for the diagnosis of cutaneous melanoma.⁹ Two additional meta-analyses have since reinforced these findings.^2,10 The most recent, by Vestergaard et al, included only prospective studies performed in a clinical setting, and thus more accurately reflected everyday dermoscopy use.² In total, 8 487 suspicious pigmented and nonpigmented lesions were included and melanoma prevalence ranged from 0.5% to 21.1% with a Breslow thickness that ranged from 0.35 to 0.95 mm.² Dermoscopic and clinical accuracy were evaluated through the diagnostic odds ratio (DOR) which considers both sensitivity and specificity and their respective trade-offs.² The DOR for dermoscopy was 15.6 (CI: 2.9–83.7, P=0.016) times higher for dermoscopy than NEE. A summary estimate of sensitivity was higher with dermoscopy (0.90, CI: 0.80–0.95) than for NEE (0.71, CI: 0.59–82, P=0.002);² although specificity was higher for dermoscopy (0.90, CI: 0.57–98) than NEE (0.81, CI: 0.48–0.95, P=0.18), it was not statistically significant.²

In all three meta-analyses, diagnostic accuracy was dependent upon the experience of the examiner.^2,9,10 However, inexperienced users, after a two-hour course, had an improved sensitivity without compromising their specificity.² Terushkin et al found, through assessment of the benign-to-malignant biopsy ratio (BMR), that a single dermatologist newly adopting dermoscopy experienced a learning curve. Initially, the BMR of the dermatologist increased compared to NEE, but with time and experience, the BMR dropped below the baseline value with NEE alone and approached the level of pigmented lesion specialists.¹¹ Several studies have demonstrated that short training modules can improve the diagnostic performance of inexperienced dermatologists, general practitioners, and even medical students.^6,12,13 However, the training modalities have varied widely among studies and the ideal teaching method for beginners remains to be standardized.

The BMR (which is directly related to positive predictive value), while not a surrogate, is impacted by the sensitivity and specificity of dermoscopy. Improvements in this ratio suggest that the increased sensitivity seen with dermoscopy does not entail an increase in the number of unnecessary biopsies and thus an increase in morbidity. Carli et al retrospectively examined two users before and after the introduction of dermoscopy and four nonusers. Analysis demonstrated a significant improvement in the BMR over a four-year study period in the dermoscopy arm (18:1 to 4.3:1, P = 0.037).¹ The BMR for nonusers had no significant difference at the beginning and the end of the study (11.8:1 to 14.8:1).¹ In a randomized controlled trial comparing one-time evaluations of equivocal pigmented lesions with dermoscopy or NEE, 9% of patients followed with dermoscopy were referred for biopsy or excision compared to 15.6% with NEE (P=0.013). The reduction in surgical morbidity was not hindered by a decreased ability to diagnose melanoma.¹⁴ Finally, a multicenter survey over ten years showed that the BMR at sites dedicated to skin cancer treatment improved from 12.8 to 6.8 (p<0.001) and remained unchanged in sites not dedicated to screening for skin cancer. The authors of that study argue that the introduction of dermoscopy was largely responsible for the observed improvement in the BMR.¹⁵

Dermoscopy and dermoscopic screening allow for the earlier detection of melanoma and improved clinical management. Several studies have shown that dermoscopic monitoring of lesions enables the detection of thin, featureless melanomas.^3,16 In a meta-analysis with a mean follow-up of 30 months, Salerni et al showed dermoscopy users to detect a greater number of thinner melanomas compared to NEE (mean Breslow depth 0.77 mm vs 1.43 mm, P = <0.05).³ Haenssle et al found that participation in specialized dermoscopic screening programs and dermoscopic examinations at the time of diagnosis were also significantly associated with thinner melanomas (P ≤ 0.01).¹⁶ Dermoscopy users have identified melanoma specific dermoscopic features that have enabled them to recognize melanoma with the sensitivity and specificity described. The following sections will expand upon these melanoma specific dermoscopic features and other features for basal cell carcinoma and keratinizing carcinomas.

Dermoscopic features in melanoma

By allowing for the in vivo assessment of subsurface skin structures, dermoscopy offers a window into the histological diagnosis of skin cancers.¹⁷ Many of the described dermoscopic features have known histological correlates. The presence of these features can provide insight into the cellular nature of skin neoplasms and allow for more precise clinical diagnoses. When the dermoscopic diagnosis is unclear, structures can hint at malignant potential. While the dermoscopic features discussed in Table 1 can be applied to any lesion on the skin, there are additional criteria in Table 2 that are specific to lesions on the face, mucosal, nails and volar surfaces, which are outlined below. Examples of dermoscopic features found in melanoma are shown in Figure 1.

Table 1.

Dermoscopic features of melanoma

Dermoscopic structures	Schematic illustration	Definition/Significance	Diagnostic Value
Atypical pigment network and angulated lines		Atypical pigment network has a disorganized and asymmetric distribution, areas of disruption, and/or significant variability in line thickness and color. In addition, the holes of the network are of differing sizes and shapes. An atypical network can appear “out of focus” with smudging of the network lines and/or with gray color.26 An atypical network is associated with superficial spreading melanoma.55 Angulated lines create a zigzag pattern with lines merging to create polygonal structures such as rhomboids. Angulated lines usually have a gray color and are associated with melanoma on sun damaged skin which includes lentigo maligna and lentigo maligna melanoma.30	Sensitivity	21%56; 52%–77.1%57–60; 100%61
			Specificity	46%61; 64.7%–88.556–60,62
			Odds ratio	2.0–2.818,43,56,60, 938
Negative network		Negative network consists of elongated and curvilinear globular structures that are surrounded by hypopigmentation. The hypopigmentation takes on the appearance of serpiginous lines meandering between the hyperpigmented elongated structures.26 A negative netork is asociated with melanomas arising in nevi.63	Sensitivity	22%–34.6%53,61
			Specificity	77.2%–95%53,61
			Odds ratio	1.443; 1.853
Atypical dots/globules		Dots and globules consist of round to oval structures of differing size, color and distribution. In melanoma the dots tend to be black-brown in color, are not located on the lines of the network and are distributed in a disorganized fashion. Globules in melanoma tend to vary in color from blue-black to brown, have increased variability in size and shape and are distributed in a disorganized manner.26	Sensitivity	13–39.6%56,57,59,60
			Specificity	74.3–92%56,57,59,60
			Odds ratio	1.7–4.838,56,57
Irregular streaks		Streaks, which encompass pseudopods and radial streaming, are linear projections located at the perimeter of the lesion that radiate from the tumor towards normal skin.26 Pseudopods are linear projections with small knobs at their tips. Radial streaming are the same structures without the knobs. Streaks are a manifestation of radial growth and are associated with superficial spreading melanoma, especially when distributed focally at the periphery.26,61 Pseudopods appear to have a higher odds ratio than streaming.43	Sensitivity	4.8%–2356,59–61 3257; 58%58
			Specificity	62%58; 77%57; 90.2%–98.7%56,59–62
			Odds ratio	1.5–5.838,43,56,57,60
Regression structures		Regression structures consist of nonpalpable areas revealing peppering/granularity and scar-like depigmentation. Peppering/granularity and scarlike depigmentation often occur together but can also present independently. Peppering/granularity consist of fine to coarse dots with a blue-grey color. Scar-like depigmentation consists of white areas that are lighter in color than the adjacent normal skin.26 In areas of scarlike depigmentation, one will not see shiny white lines/crystalline structures or blood vessels.	Sensitivity	11.4%–41.756,57,59–61; 79%58
			Specificity	63%58; 83.5%–99%56,57,59–61
			Odds ratio	2–5.738,43,60; 18.357
Blue-white veil		Blue-white veil consists of palpable areas having a blue-black color with an overlying whitish ground-glass haze.26 The blue white veil corresponds with melanin/melanocytes in the dermis in association with compact orthokeratosis.64	Sensitivity	11.4%–28%56,59,60; 5161; 92%58
			Specificity	74%–99.0%56,58– 62
			Odds ratio	1.74–2.543,57,60; 1338
Shiny white lines		Shiny white lines (formerly known as crystalline structures) are only visible under polarized light and consist of short bright white line/s that, when in groups, often have a parallel or orthogonal orientation to each other. The presence of these structures is suggestive of melanoma or Spitz nevus and when present in melanoma are associated with invasion.52,61	Sensitivity	70.0%52
			Specificity	80.6%52
			Odds ratio	2.543; 9.752
Atypical blotch		Blotches are an area of heavy pigmentation that obscures the ability to see any other structures in that area. Atypical or irregular blotches are defined as the presence of more than one blotch or the presence of an off center blotch located towards the periphery of the lesion.26	Sensitivity	18%–37.5%56,57,59; 71.3%60
			Specificity	30.5%60; 88.2%–92.6%56,57,59
			Odds ratio	1.88–4.138,43,56,60
Polymorphous vessels		Polymorphous vessels are defined as multiple types of vessels in a single lesion.26 Except for comma shaped vessels seen in intradermal nevi, any vessel or vascular blush (milky red areas) seen in a melanocytic lesion should raise concern for melanoma. Melanomas can reveal dotted vessels, serpentine vessels, tortuous vessels and milky red areas. The vascular structures with the strongest association with melanoma include multiple shades of pink and/or polymorphous vessels. The most predictive vasculature for melanoma is the combination of dotted and serpentine vessels.56	Sensitivity	9.4%59; 18%60; 62.9%56
			Specificity	53.8%56; 91.2%60; 96.1%59
			Odds ratio	2.0–3.0443,56,60

Courtesy of Natalia Jaimes, MD, Miami, FL.

Table 2.

Dermoscopic features of melanoma arising in special locations

	Dermoscopic structures	Schematic illustration	Definition/Significance	Diagnostic Value
Melanoma of the face	Blotches with obliteration of follicles		Blotches with obliteration of follicles are defined by the loss of visible adnexal openings forming the pseudonetwork typically seen in facial melanocytic tumors.21 Blotches may be present with preservation of the follicle.	Sensitivity	-
				Specificity	-
				Odds ratio	3.8221
	Concentric circles		Concentric circles (also known as circles within circles) are pigmented rings each surrounding an additional circle.43 The pigmented ring can be seen within the adnexal opening.	Sensitivity	4.2%18
				Specificity	98.8%18
				Odds ratio	2.018
	Grey circles		Grey circles are small rings that appear within follicular openings.21	Sensitivity	54.2%18
				Specificity	83.3%18
				Odds ratio	2.8621; 4.618
	Asymmetric follicular opening (incomplete circles)		Asymmetric follicular openings (incomplete circles) are a pigment rings that do not uniformly surround an adnexal opening.26	Sensitivity	58.3%18
				Specificity	71.3%18
				Odds ratio	3.0018
	Polygonal lines coalescing to form rhomboidal structures		Polygonal lines coalescing to form rhomboidal structures typically surrounding adnexal openings.21	Sensitivity	16.7%18
				Specificity	91.7%18
				Odds ratio	218; 6.1821
Melanoma of the mucosa	Blue, gray, or white color and a structureless zone		Structureless areas contain none of the basic elements such as dots, globules, circles, or lines. In addition, a lesion with a structureless zone that had blue, gray, or white color was most predictive of melanoma.23	Sensitivity	100%23
				Specificity	82.2%23
				Odds ratio	-
Melanoma of volar surfaces	Parallel Ridge Pattern		Parallel ridge pattern consists of thick lines of pigmentation on the ridges or cristae superficialis of volar skin.26	Sensitivity	86.5%65
				Specificity	99.0%65
				Odds ratio	-
Melanoma of the nail apparatus	Irregular brown band		Brown bands suggest melanocyte mediated pigmentation. Irregularity is present in the form of multiple colors (i.e. black, grey), varied thickness and spacing of lines, and loss of parallelism.22

Courtesy of Natalia Jaimes, MD, Miami, FL.

Fig. 1.

Dermoscopic features of melanoma. (A) Melanoma presenting with atypical globules and dots of different sizes and shapes (yellow arrows), patches of atypical network (blue arrowhead) and a blue-white veil (blue arrow). (B) Melanoma with diffuse polymorphous vasculature, consisting of serpentine, dotted, and glomerular vessels, can be found throughout the lesion (yellow arrowheads). In addition, patches of atypical network (blue arrowheads) are seen. (C) Superficial spreading melanoma with pseudopods distributed asymmetrically around the lesion (black arrowheads). (D) Melanoma with the regression structure blue-grey peppering (black star). Shiny white lines are also seen throughout the entire lesion (red arrows) along with a central blue-white veil (red arrowhead).

Lentigo maligna and lentigo maligna melanoma both arise on sun damaged skin, most commonly the face and scalp (both lentigo maligna and lentigo maligna melanoma will be represented by the abbreviation LMM). The anatomic predilection for the face and sun damaged skin entails a broad differential diagnosis including pigmented actinic keratoses, lichen planus-like keratoses, and early-flat seborrheic keratoses.¹⁸ Dermoscopy can help the user sort through the numerous diagnoses. Facial skin has a flattened dermoepidermal junction resulting in absence of traditional reticular patterns commonly seen in melanocytic lesions. Instead, the pigmentation is punctuated by adnexal structures creating a pseudonetwork.¹⁹ LMM specific structures include: asymmetrical pigmentation of follicular openings, dots aggregated around adnexal openings, polygonal lines and rhomboid structures, and dark blotches with or without obliteration of the adnexal openings (Figure 2A).^18,20,21

Fig. 2.

Dermoscopic features of melanoma in special locations. (A) Lentigo maligna melanoma on the face with concentric circles, also know as circle within a circle (blue arrowhead), grey circles (blue arrow), incomplete circles (black arrowhead), and angulated lines (black star. Within the melanoma, there is a seborrheic keratosis (black arrow) with comedo-like openings. (B) Melanoma of the nail matrix with brown lines that vary in thickness and have a disruption in parallelism. The patient has distal onycholysis. (C) Melanoma of the vulva with a linear pattern with features resembling a negative network (red asterisk). Yet, shiny white structures (red arrow) and multiple shades of brown with asymmetric distribution of pigment are indicative melanoma. (D) Acral lentiginous melanoma on the volar skin of the heel with pigment on the ridges (red arrowheads).

Additional dermoscopic features are also needed to evaluate for melanoma arising on the nail apparatus, mucosa or volar skin. In melanoma of the nail apparatus, dermoscopy of the pigmented band will show multiple brown to black lines with irregular arrangement and thickness and possible areas of pigment interruption (Figure 2B).²² The mucosa comprises a wide range of anatomic sites such as the lip, glans penis, anogenital orifice, labia, and praeputium. On dermoscopy, an early sign of mucosal melanoma is the presence of structureless areas and a gray color. More advanced melanoma can present with multiple patterns and colors, particularly white, blue, and/or gray (Figure 2C).^23,24 The dermoscopic appearance of acral lentiginous melanoma (ALM) is shaped by the unique furrow and ridge pattern of acral volar skin surfaces. Melanocytic features specific to ALM are the parallel ridge pattern and irregular diffuse pigmentation (Figure 2D).²⁵ Table 1 and 2 provide an overview of melanoma specific criteria, including definitions, pictorial examples, and reported sensitivities, specificities, and odds ratios. This list was compiled using the most recent definitions of dermoscopic criteria published by the International Dermoscopy Society in 2016.²⁶

Dermoscopy for the diagnosis and management of keratinocyte carcinomas

In addition to differentiating nevi from melanoma and detecting early melanoma, dermoscopy can also be used to detect KC. Rosendahl et al examined 217 consecutive pigmented nonmelanocytic lesions, of which 138 were malignant, with both dermoscopy and NEE. They reported a diagnostic improvement with dermoscopy of 0.89 (area under Receiver Operating Curve) compared to 0.83 with NEE alone (P<0.001).²⁷ In 2016 a two part study of optional then mandatory dermoscopy use, the overall diagnostic sensitivity of basal cell carcinoma (BCC) was 93.3% (91.9–94.5) and specificity was 91.8% (90.6–93.0).²⁸ Furthermore, dermoscopy when compared to NEE was shown to help predict BCC histopathology subtypes.²⁸ In the following sections, we will detail further the specific diagnostic features of KC.

Dermoscopic features in basal cell carcinoma

BCCs are broadly classified as pigmented or nonpigmented. While most BCCs on NEE are nonpigmented, 29.8% of these lesions actually reveal pigment structures when viewed with dermoscopy.²⁹ The pigmented structures seen in BCC include blue-grey globules or ovoid nests, spoke-wheel structures and leaflike areas. The structures seen in nonpigmented or pigmented BCC include arborizing vessels, shiny white blotches and strands (seen with polarized dermoscopy) and ulcers/erosions (Table 3).^30–32 Studies have demonstrated that certain dermoscopic structures and patterns are associated with different subtypes of BCC.^28,33 The presence of spoke-wheel and leaflike structures, multiple small erosions, short fine vessels and shiny white blotches and strands in a nonpalpable lesion are associated with superficial BCC. In contrast, the presence of blue-grey globules, ovoid nest, arborizing vessels and ulceration are associated with nodular BCC.^28,33 Examples of dermoscopic features found in BCC are shown in Figure 3.

Table 3.

Classic features of basal cell carcinomas

Dermoscopic structures	Schematic illustration	Definition	Diagnostic Value
Spoke-wheel areas, concentric structures (clod within a clod) and leaflike areas		Spoke-wheel areas are brown to blue/gray well- circumscribed radial projections that originate from a central darker hub.26 If individual radial projections are not visible then this structure will appear as a concentric globular structure where the center of the globule will appear darker Leaflike areas, considered to be a similar structure to spoke wheel areas, are brown to blue/gray bulbous projections that coalesce into a darker off-center base. This structure often resembles the shape of a leaf.26	Sensitivity	10–17%30
			Specificity	100%30
			Odds ratio	-
Multiple blue-grey nonaggregated dots and globules and large blue-grey ovoid nests		Blue-grey dots in basal cell carcinoma have a buck-shot distribution pattern. Blue-grey globules are well-circumscribed oval structures that are distributed in a nonaggregated patern.66 Blue-grey ovoid nests are larger than globules and encompass at least 10% of the surface area of the lesion. They are well-circumscribed structures with a confluent to nearly confluent color.26,66	Sensitivity	27%30; 55%30
			Specificity	97%30; 99%30
			Odds ratio	-
Arborizing (branched) vessels		Large caliber vessels that branch into thinner vessels. They are sharply in focus and bright-red in color.26	Sensitivity	20%35; 52%30; 72%67
			Specificity	92–100%30,35,67
			Odds ratio	6.735
Ulceration		A structureless area with a red-orange color that may have a serous crust.66 Smaller ulcers grouped together are known as multiple small erosions.	Sensitivity	27%30
			Specificity	97%30
			Odds ratio	-
Short fine vessels		Short (<1mm), thin, linear vessels with little to no branching points.66 These vessels are more frequently found in superficial basal cell carcinomas when compared to nodular basal cell carcinomas.68,69
Shiny white blotches and strands		Shiny white blotches and strands appear shiny white under polarized light, a feature common across all shiny white structures. Unique to basal cell carcinoma are shiny white structures in the shape of blotches and strands.26,32

Data from Menzies SW, Westerhoff K, Rabinovitz H, et al. Surface microscopy of pigmented basal cell carcinoma. Arch Dermatol. Aug 2000;136(8):1012–1016.

Courtesy of Natalia Jaimes, MD, Miami, FL.

Fig. 3.

Dermoscopic features of basal cell carcinoma. (A) Pigmented basal cell carcinoma with leaf-like structures (black arrow) and blue-grey ovoid nests, globules, and dots (blue arrows). (B) Nodular basal cell carcinoma with an arborizing vessel (yellow arrow) and blue-grey dots and globules (blue arrows). (C) Nonpigmented superficial basal cell carcinoma with shiny white blotches and strands (yellow arrowheads) and short fine vessels (blue arrowhead). (D) Superficial basal cell carcinoma with multiple erosions/ulcerations (white arrows) over an erythematous background with polymorphous vessels.

Dermoscopic features in actinic keratoses, squamous cell carcinoma in situ/Bowen disease, invasive squamous cell carcinoma, and keratoacanthoma

Dermoscopic features have been identified that can assist in diagnosing actinic keratoses (AK), Bowen disease (BD) or in situ squamous cell carcinoma (full thickness atypia) (IEC), invasive squamous cell carcinoma (SCC), and keratoacanthoma (KA). Dermoscopically, these lesions exist on a spectrum that we have grouped under the category of keratinizing skin cancer (KSC), thus several features, such as scale, vascular structures, and erythema, are ubiquitous among them. While more research is needed to determine the sensitivity and specificity of these features, it appears we can tease out the full spectrum of KSCs. Dermoscopy users can differentiate AKs and in situ SCC and SCC.

There are several notes that need to be made concerning the data presented in Tables 4, 5, and 6. Many of the lesions examined occur in higher frequency on the head and neck, thus making generalizability to KSCs on other body parts difficult.^18,34 In addition, several studies look at specific comparisons of the dermoscopic features of AK, for example, to those of SCC, but not all possible diagnoses.^21,35 When a definition was absent from the International Society of Dermoscopy 2016 consensus meeting²⁶, we cited definitions provided from other sources. Finally, while purported to be significant, some feature have yet to be rigorously studied to determine their diagnostic significance. The table below represents our best representation of available data and, hopefully, makes clear when little analytic information is known. In addition, examples of dermoscopic features found in KSC are shown in Figure 4.

Table 4.

Actinic keratoses

Pigmented Tumors
Dermoscopic structures	Schematic illustration	Definition/Significance	Diagnostic Value
Strawberry Pattern		Strawberry pattern is a red pseudonetwork (background erythema and fine wavy vessels) punctuated by adnexal openings accentuated by a white halo.26 A study comparing actinic keratoses to lentigo maligna demonstrated the preservation of follicle integrity is 12.45 times more likely of being pigmented actinic keratosis.21	Sensitivity	95.6%70
			Specificity	95.0%70
			Odds ratio	3.6; 13.5221
Surface Scale		Surface scale consists of homogenous opaque adherently keratotic structures with yellowish to brown or white coloration.70,71 It is a common feature of actinic keratosis and squamous cell carcinoma.34	Sensitivity	93.7%70; 37.7%18
			Specificity	35%70; 94.2%18
			Odds ratio	7.6721; 3.4 (RR)18

Courtesy of Natalia Jaimes, MD, Miami, FL.

Table 5.

Intraepidemal carcinoma and squamous cell carcinoma

Pigmented Tumors
Dermoscopic structures	Schematic illustration	Definition/Significance	Diagnostic Value
Globules/Dots or Round circles		Globules/dots or Round circles are brown to gray structures that, when arranged in a linear fashion, were frequently found in pigmented intraepidermal carcinoma.72
Nonpigmented Tumors
Ulceration		Ulceration includes large structureless area with dark red to brown coloration with a serous crust. Classically attributed and identified in squamous cell carcinoma, but no studies have determined sensitivity or specificity.34,71 “Blood spots” have been identified as independent positive predictors of keratoacanthomas and squamous cell carcinomas, yet an association with ulceration remains unevaluated.73
Rosettes		Rosettes consist of four bright white clods or dots arranged in a 2 by 2 pattern, also called a 4- leaf clover pattern. This structure can only be seen with polarized light and corresponds to keratin filled adnexal openings that are most commonly seen in actinically damaged skin, actinic keratosis and squamous cell carcinoma.26,32,74
Glomerular Vessels		Glomerular or coiled vessels are often seen in squamous cell carcinoma.26,75 The presence of clustering of glomerular, dotted and irregular vessels is highly associated with intraepidermal carcinoma.35	Sensitivity	60%35; 27%67
			Specificity	94%35; 100%67
			Odds ratio	21.935
Polymorphous vasculature		Polymorphous vasculature is an array of multiple (looped, dotted, and/or serpentine) vessels distributed irregularly throughout the lesion.26 Classically this pattern is associated with squamous cell carcinoma but no studies have evaluated sensitivity/specificity of the polymorphous pattern.34,76 Increasing vascularity, higher vessels caliber, and visible bleeding were significantly associated with poorly differentiated squamous cell carcinoma.77

Courtesy of Natalia Jaimes, MD, Miami, FL.

Table 6.

Keratoacanthoma and/or well differentiated squamous cell carcinoma

Dermoscopic structures	Schematic illustration	Definition/Significance	Diagnostic Value
White Circles		White circles are bright white circles surrounding a dilated infundibulum filled with a yellow/orange keratin plug.73 This structure is associated with well differentiated squamous cell carcinoma and often seen in keratoacanthomas.73	Sensitivity	31.9%18
			Specificity	94.2%18
			Odds ratio	3.0 (RR)18
Looped Vessels		Looped vessels are vascular loops, that depending on the dermoscopy user’s angle of view, can appear twisted.26 Looped vessels have been observed in higher frequencies in nodular squamous cell carcinoma and keratoacanthomas.34,73,78

Courtesy of Natalia Jaimes, MD, Miami, FL.

Fig. 4.

Dermoscopic features of keratinizing skin cancers. (A) Squamous cell carcinoma showing diffuse polymorphous vasculature consisting of numerous serpentine (black arrow), dotted (blue arrow), and looped vessels (yellow arrow). A triangular ulceration (yellow arrowhead) with associated scale (blue arrowhead) is also seen. (B) Keratoacanthoma presenting with a central keratin plug and a peripheral rim of irregular looped (red arrowheads) and serpentine vessels (black arrowhead). (C) Pigmented squamous cell carcinoma with numerous rosettes (white arrows). (D) Actinic keratosis with a strawberry pattern consisting of a structureless red area interrupted by the follicular openings.

Dermoscopy algorithms

Pehamberger et al first described pattern analysis in dermoscopy as a diagnostic aid to supplement clinical exam of pigmented skin lesions.^36,37 Pattern analysis is the assessment of the heterogeneity in a pigmented lesion’s overall structure, colors, patterns, and margins.³⁶ Pattern recognition, when used by experts, has been shown to have the best diagnostic performance when compared to other methods;³⁸ yet, this technique is developed through years of experience. For new users, algorithms are often needed to sift through the diverse array of presentations of any given type of neoplasm. With the number of new dermoscopy users ever on the rise⁴ along with the inclusion of dermoscopy in national management guidelines^39–41, algorithms that address skin cancer management as well as diagnosis has become essential.

Although there are multiple algorithms used to teach dermoscopy the authors rely on the 2-step algorithm.⁴² The first step differentiates melanocytic from nonmelanocytic lesions. The second step separates benign from malignant melanocytic lesions.³⁸ Numerous algorithms have been developed to help new users approach the second step of diagnosing melanoma with high sensitivity and specificity.⁴³ Some of these algorithms include the ABCD Rule, the 7-point checklist, the Menzies’ Method, the CASH algorithm, and the “Chaos and Clues” algorithm.^7,44–47 The latter emphasizes the importance of a disorganized distribution of dermoscopic structures in malignant lesions.⁷ While on rare occasions skin cancers lack disorganization, they can usually be identified based on the presence of starburst like pattern, negative networks, shiny white structures, and polymorphous vessels. While the aforementioned algorithms have served us well, there is still a need for simpler triage algorithms that can aid beginners in effectively screening for malignancies. Multiple attempts have been made to create a simplified triage algorithm.^7,44–47 The triage amalgamated dermoscopy algorithm (TADA) is a new algorithm that attempts to provide a comprehensive approach to dermoscopy guided management of pigmented and nonpigmented skin cancers through a combination of previously validated criteria and features (Figure 5).^{19,31,48–53} A novel aspect of TADA is that it first asks users to determine if a lesion is an unequivocal example of a dermatofibroma, angioma, and seborrheic keratosis. If the user is confident that the lesions are one of these three, it is excluded from further evaluation with the algorithm. The TADA algorithm is not meant to evaluate lesions on the palms, soles, or nails.

Fig. 5.

Triage amalgamated dermoscopic algorithm (TADA) is a multistep process to guide dermoscopic management of skin lesions. First, unequivocal angiomas, dermatofibromas, and seborrheic keratoses are identified and excluded from further evaluation. Next the lesion is evaluated for the presence of architectural disorder or order. If architectural disorder is present, the lesion should be biopsied or referred for further management. On rare occasion a malignancy can present in an organized fashion, thus all organized lesions are then evaluated for blue-black or gray color, white structures, negative network, ulcer/erosion, starburst pattern and/or vessels. The presence of any one of these features would raise concern for malignancy. (*) Lesions of the volar surfaces, nails, mucosal surfaces, and face should not be evaluated with this algorithm.

Courtesy of Natalia Jaimes, MD, Miami, FL; Zachary J. Wolner, BA, Oriol Yelamos, MD, Konstantinos Liopyris, MD, Tova Rogers, MFA, Michael A. Marchetti, MD, Ashfaq A. Marghoob, MD, New York, NY.

The second step of TADA asks user to evaluate architectural disorder. A study led by the International Dermoscopy Society found architectural disorder, with an OR of 6.6, the most powerful discriminator of melanoma.⁴³ Furthermore, the subjective interpretation of architectural disorder had the highest interobserver agreement among participants (intraclass correlation coefficient (ICC) of 0.43, where an ICC of 0 is an agreement by chance and an ICC of 1 is perfect agreement).⁴³ The additional criteria included in TADA (starburst pattern, blue-black or gray color, shiny white structures, negative network, ulcer/erosion, and/or vessels) are needed to identify organized-appearing malignancies, such as spitzoid, nodular, and amelanotic melanoma. The presence of any one feature indicates the need for a biopsy or specialist referral.

In a pilot study, novice dermoscopy users were able to achieve a 93.3% sensitivity and 74.1% specificity using TADA for the identification of malignant study lesions (melanoma, BCC, SCC) after one day of dermoscopy training.⁵⁴ The high sensitivity seen with TADA was at the expense of specificity, however, the value seen in this study were still equivalent or better to those reported for existing triage algorithms.³⁸ Thus, with a minor increase in biopsies and referrals for unequivocal lesions, fewer malignant lesions will be missed. TADA has the remarkable potential for enhancing skin cancer screening and detection among individuals with limited dermoscopy training and experience.

Summary

Dermoscopy has been shown to increase the sensitivity and specificity for diagnosing melanoma. It also enables the diagnosis of melanoma at an earlier stage. In addition, several studies have established dermoscopy as a diagnostic aid for BCC and more recently in clinically determining its histopathological subtype. AK, BD, IEC, SCC, and KA share common dermoscopic features, although multiple specific features have been described allowing accurate diagnosis of specific tumors. Diagnostic algorithms are often used to aid in the dermoscopic evaluation and diagnosis of skin lesions. The advent of triage algorithms can assist beginners with determining if a lesion requires a biopsy or specialist referral without having to arrive at a specific diagnosis. TADA is one such algorithm that includes a limited set of dermoscopic features and can thus be easily learned and implemented by new users. The features included in TADA were selected for their abilities to discriminate a wider range of common pigmented and nonpigmented skin cancers.

Key Points.

• Dermoscopy increases diagnostic accuracy and sensitivity for melanoma, which translates into detection of thinner tumors.

• Multiple studies have established that dermoscopy is an aid for diagnosing basal cell carcinoma; the presence of dermoscopic criteria can predict the histopathological subtype of basal cell carcinoma.

• While actinic keratosis, Bowen’s disease/squamous cell carcinoma in situ, invasive squamous cell carcinoma, and keratoacanthoma share common dermoscopic features, specific criteria can permit diagnostic discrimination.

• Dermoscopy novices can benefit from using diagnostic or triage algorithms to improve their diagnostic abilities and management decisions.