Dermoscopy of Lentiginous Melanomas and Equivocal Benign Pigmented Macules of the Scalp: A Case-Control Multicentric Study

Abstract

Introduction

Although the dermoscopic features of facial lentiginous melanomas (LM), including lentigo maligna and lentigo maligna melanoma, have been extensively studied, the literature about those located on the scalp is scarce. This study aims to describe the dermoscopic features of scalp LM and assess the diagnostic accuracy of dermoscopy to discriminate them from equivocal benign pigmented macules.

Methods

Consecutive cases of scalp LM and histopathology-proven benign but clinically equivocal pigmented macules (actinic keratoses, solar lentigos, seborrhoeic keratoses, and lichen planus-like keratoses) from four referral centres were included. Dermoscopic features were analysed by two blinded experts. The diagnostic performance of a predictive model was assessed.

Results

56 LM and 44 controls were included. Multiple features previously described for facial and extrafacial LM were frequently identified in both groups. Expert’s sensitivity to diagnose scalp LM was 76.8% (63.6–87.0) and 78.6% (65.6–88.4), with specificity of 54.5% (38.9–69.6) and 56.8% (41.0–71.7), and fair agreement (kappa coefficient 0.248). The strongest independent predictors of malignancy were (OR, 95% CI) chaos of colour (15.43, 1.48–160.3), pigmented reticular lines (14.96, 1.68–132.9), increased density of vascular network (3.45, 1.09–10.92), and perifollicular grey circles (2.89, 0.96–8.67). The predictive model achieved 85.7% (73.8–93.6) sensitivity, 61.4% (45.5–75.6) specificity, and 81.5 (73.0–90.0) area under curve to discriminate benign and malignant lesions. A diagnostic flowchart was proposed, which should improve the diagnostic performance of dermoscopy.

Conclusion

Both facial and extrafacial dermoscopic patterns can be identified in scalp LM, with considerable overlap with benign pigmented macules, leading to low specificity and interobserver agreement on dermoscopy.

Introduction

Scalp melanomas arise more frequently on bald and chronically sun-damaged scalps of elderly males [1–3], often coexisting with non-melanoma skin cancers, actinic keratoses (AK), and benign pigmented macules, which can obscure the diagnosis [4]. Because scalp location has been consistently associated with poor prognosis [5, 6], the early recognition of melanomas in this area is essential to improve survival.

Lentigo maligna is the most prevalent subtype of melanoma diagnosed on the scalp in some series [2, 7]. Lentigo maligna can progress to lentigo maligna melanoma at an estimated rate of 3.5% per year [8]. Lentigo maligna melanoma accounts for 4–15% of invasive cutaneous melanomas in general [9] and 35.7–39% of invasive scalp melanomas [7, 10]. For the purpose of this study, both lentigo maligna and lentigo maligna melanoma will be referred to as lentiginous melanomas (LM) [11].

The main original studies addressing dermoscopy of LM included exclusively or predominantly facial lesions [12–19], or trunk and limbs extrafacial ones [15, 20, 21], but no subgroups of scalp LM have been analysed by these authors. Although the scalp is considered to be a special site on dermoscopy [4], the literature addressing the features of scalp melanomas and more specifically scalp LM is scarce [3, 22]. The objectives of this study were to describe the dermoscopic features of pigmented LM of the scalp and to assess the diagnostic accuracy of dermoscopy to discriminate scalp LM from benign melanoma mimics.

Materials and Methods

An observational, multicentric, and retrospective case-control study was conducted at four referral centres in Italy and Australia (Sydney Melanoma Diagnostic Centre, Dermatology Clinic of the Melanoma Institute Australia, Dermatology Unit of the University of Campania, and Skin Cancer Unit of Reggio Emilia). All centres comply with a policy of photographing all equivocal lesions prior to biopsy or excision.

Cases were consecutive histopathology-proven pigmented scalp LM. Controls were consecutive scalp AK, solar lentigos (SL), and seborrheic keratoses (SK), and lichen planus-like keratoses (LPLK) with histopathological confirmation, presenting clinically as equivocal pigmented macules. Lesions with low-quality dermoscopic images and amelanotic/hypomelanotic melanomas were excluded.

Dermoscopic images were obtained either with a non-polarised dermatoscope (Heine Delta 20^®, Heine, Germany) or a cross-polarised dermatoscope (Dermlite, 3Gen, USA; or MoleScope II^®, MetaOptima, Canada; or Heine Delta 20T^®, Heine, Germany) attached to a digital camera. Cases and controls were coded and analysed by two dermatologists with internationally recognised expertise in dermoscopy (S.M. and P.P.), blinded for histopathological diagnosis. Both experts provided a presumptive diagnosis for each image.

A checklist of dermoscopic features was recorded by one of the experts (P.P.), including those features of the revised pattern analysis [23]; vascular patterns [24]; classic features described for LM [12]; additional features described for LM [15, 20, 21, 25, 26]; classic features of melanomas in general (pattern analysis [27], 7-point checklist [28]); and features described for AK, SL, SK, and LPLK [13, 17, 27]. The descriptive terminology was prioritised, with metaphoric terms translated into descriptive terms [17]. Table 1 presents the complete dermoscopic checklist and metaphoric synonyms when applicable. Angulated lines were divided in “small” angulated lines, corresponding to the classic description of rhomboidal structures of facial LM [12]; and “large” angulated lines, described for extrafacial LM [20].

Table 1.

Summary of dermoscopic features identified in cases and controls

Dermoscopic features (descriptive terminology)	Dermoscopic features (metaphoric terminology, if applicable)	Cases (N = 56), n (%)	Controlsa (N = 44), n (%)	p value
Chaos of structure/pattern	Asymmetry of pattern	54 (96.4)	33 (75.0)	0.002
Chaos of colour	Asymmetry of colour	55 (98.2)	35 (79.5)	0.005
Grey or blue structures		52 (92.9)	39 (88.6)	0.50
Eccentric structureless area	Blotch	50 (89.3)	34 (77.3)	0.10
Thick lines, reticular or branched	Broadened network	8 (14.3)	7 (15.9)	0.82
Black dots or clods, peripheral	Peripheral black dots or globules	3 (5.4)	7 (15.9)	0.08
Radial lines or pseudopods, segmental	Streaks, pseudopods	9 (16.1)	4 (9.1)	0.30
Polarising-specific white lines	Chrysalis, crystalline structures, shiny white lines	16 (28.6)	14 (31.8)	0.73
Polymorphous vessels	Atypical vessels	9 (16.1)	6 (13.6)	0.74
Small angulated lines/polygons	Pigmented rhomboidal structures	43 (76.8)	18 (40.9)	<0.001
Large angulated lines/polygons		8 (14.3)	4 (9.1)	0.43
Vascular patternsb
Dots		7 (12.5)	7 (15.9)	0.63
Clods	Globules	3 (5.4)	4 (9.1)	0.70
Linear straight		9 (16.1)	8 (18.2)	0.78
Looped	Hairpin	1 (1.8)	7 (15.9)	0.02
Serpentine	Linear irregular	9 (16.1)	3 (6.8)	0.16
Coiled	Glomerular	3 (5.4)	2 (4.5)	0.99
Incomplete perifollicular brown or grey circles	Asymmetric pigmentation of follicular openings	46 (82.1)	25 (56.8)	0.006
Incomplete grey circles		44 (78.6)	24 (54.5)	0.01
Incomplete brown circles		25 (44.6)	12 (27.3)	0.074
Complete perifollicular brown or grey circles		46 (82.1)	24 (54.5)	0.003
Complete grey circles		44 (78.6)	21 (47.7)	0.001
Complete brown circles		25 (44.6)	15 (34.1)	0.29
Dark brown or black structureless area obscuring the hair follicles	Obliteration of follicular openings	13 (23.2)	15 (34.1)	0.23
Perifollicular grey dots	Annular-granular pattern	39 (69.6)	19 (43.2)	0.008
Brown-to-grey structureless area interrupted by follicular openings	Pseudo-network	51 (91.1)	36 (81.8)	0.53
Irregular	Atypical, broken	45 (80.4)	24 (54.5)	0.006
Regular	Typical	6 (10.7)	12 (27.3)	0.03
Circle within a circle or double circle		31 (55.4)	16 (36.4)	0.06
Small triangular or rounded structures		36 (64.3)	17 (38.6)	0.01
Increased density of vascular network		24 (42.9)	10 (22.7)	0.03
Red angulated lines	Red rhomboidal structures	8 (14.3)	9 (20.5)	0.42
White structureless area	Scar-like depigmentation	15 (26.8)	10 (22.7)	0.64
Blue structureless area	Blue-white veil	7 (12.5)	6 (13.6)	0.87
Pigmented reticular lines	Pigmented network	25 (44.6)	9 (20.5)	0.01
Irregular	Atypical	15 (26.8)	7 (15.9)	0.19
Regular	Typical	11 (19.6)	2 (4.5)	0.03
	Atypical vessels	11 (19.6)	7 (15.9)	0.63
Multiple orange/brown clods	Comedo-like openings, pseudocomedones	4 (7.1)	12 (27.3)	0.006
Multiple white clods	Horn pseudocysts, milia-like cysts	3 (5.4)	6 (13.6)	0.18
Sharply demarcated border over total periphery		2 (3.6)	10 (22.7)	0.003
Peripheral concave invaginations	Moth-eaten border	17 (30.4)	15 (34.1)	0.69
Scales		9 (16.1)	11 (25.0)	0.27
Fine brown parallel curved lines	Fingerprint-like structures	26 (46.4)	16 (36.4)	0.31
Prominent (or pale) follicular openings		10 (17.9)	9 (20.5)	0.74

^aIncluding solar lentigo, actinic keratosis, seborrhoeic keratosis, and lichen planus-like keratosis.

^bNo curved (comma-like) or helical (corkscrew) vessels were identified.

Cases were also stratified by location (anterior vs. posterior to the coronal suture) as per macroscopic photographs. Participants consented to the use of their de-identified information and images for research purposes. The study was approved by the Federal University of Sao Paulo Ethics Committee (CAAE: 90241718.6.0000.5505). Both Australian centres also obtained approval from the Sydney Local Health Ethics Committee (X15-0311; 2019/ETH06854).

Patients’ characteristics and dermoscopic features were summarised using descriptive statistics and stratified by histopathological diagnosis (LM vs. controls), and subsite location (anterior vs. posterior, for cases only). Categorical variables were summarised by proportions while continuous variables were described as median and interquartile range (IQR). Differences between groups were tested using Pearson’s χ² test, Fisher’s exact test or t test as appropriate.

Predictive factors associated with malignancy were investigated using multivariable logistic regression model. The final predictive model was determined from backward elimination technique that considered all variables with a significant (p value <0.05) from the univariable analysis as potential predictors. The performance of the model was assessed using the area under the curve (AUC) of the receiver operating characteristic (ROC). To correct for potential overoptimism, internal validation was performed using bootstrap method with 200 replications. The model’s sensitivity and specificity were derived using Youden index as the threshold point for diagnosis decision from the multivariable predictive model. Kappa coefficient was used to calculate the interobserver agreement. Statistical analyses were performed using R (version 3.6.3) and SAS 9.4.

Results

A total of 56 melanomas (47 lentigo maligna and 9 lentigo maligna melanomas) and 44 controls (14 SL, 10 AK, 14 SK and 6 LPLK) were included. There was an absolute predominance of male gender for cases and controls (94.6% and 100%, respectively; p = 0.25). Median age (IQR) was 70 years (78.5–62.5) for cases and 63 years (77.5–63) for controls (p = 0.73).

The dermoscopic features identified in both groups are presented in Table 1. Some examples of cases and controls are presented in Figure 1. Cases located anteriorly to the coronal structure did not differ from those located posteriorly for any of the dermoscopic features (online suppl. Table 1; for all online suppl. material, see https://doi.org/10.1159/000535030).

Fig. 1.

a Lentigo maligna melanoma (Breslow 0.35 mm) showing blue-grey small angulated lines (“facial-like” rhomboidal structures) (white arrowhead), grey circles (blue arrow), and perifollicular grey dots (annular-granular pattern) (white arrows) on the frontal scalp of a female in her 60s (non-polarised dermoscopy). b Chaos of colour and pattern, dark brown and black structureless areas obscuring the hair follicles (obliteration of hair follicles), irregular/atypical brown-to-grey structureless area interrupted by follicular openings all over the lesion (irregular/atypical pseudo-network), perifollicular brown and grey circles (blue arrows), small angulated lines (rhomboidal structures) (white arrowheads), in a challenging solar lentigo on the bald scalp of a male in his 70s, showing striking similarities to lentigo maligna melanoma (polarised dermoscopy). c Large angulated lines (“extrafacial-like” polygons) (white arrowheads) and grey circles (blue arrow) in a lentigo maligna arising on the temporal bald scalp of a male in his 80s (non-polarised dermoscopy). d Brown dots, black dots and clods, grey dots, small angulated lines (incomplete “facial-like” rhomboidal structures) (white arrowheads), incomplete perifollicular brown and grey circles (asymmetric pigmentation of follicular openings) in a pigmented actinic keratosis on the bald scalp of a male in his 90s (polarised dermoscopy). e Lentigo maligna with a “solar lentigo-like” pigment network and lacking other typical positive features for melanoma, arising on the bald scalp of a male in his 70s (polarised dermoscopy). f Chaos of pattern and colour with complete grey circles (blue arrows) and no looped (hairpin) vessels or orange/brown clods (comedo-like openings) in a flat seborrhoeic keratosis on the bald frontal scalp of a male in his 70s (polarised dermoscopy).

Experts reached sensitivity of 76.8–78.6% and specificity of 54.5–56.8% to diagnose melanoma based on pattern analysis (Table 2). There was fair agreement for the specific histopathological diagnosis and to differentiate malignant from benign lesions (kappa coefficient 0.316 and 0.248, respectively).

Table 2.

Diagnostic performance of dermoscopy experts and the multivariable predictive model to differentiate malignant and benign equivocal pigmented macules of the scalp

	Histopathological diagnosis		Sensitivity, % (95% CI)	Specificity, % (95% CI)
	benign	malignant
	n/N (%)	n/N (%)
Multivariable predictive model
Benign	27/44 (61.4)	8/56 (14.3)	85.7 (73.8, 93.6)	61.4 (45.5, 75.6)
Malignant	17/44 (38.6)	48/56 (85.7)
Diagnosis from expert 1
Benign	24/44 (54.5)	13/56 (23.2)	76.8 (63.6, 87.0)	54.5 (38.9, 69.6)
Malignant	20/44 (45.5)	43/56 (76.8)
Diagnosis from expert 2
Benign	25/44 (56.8)	12/56 (21.4)	78.6 (65.6, 88.4)	56.8 (41.0, 71.7)
Malignant	19/44 (43.2)	44/56 (78.6)

On multivariable logistic regression, the main features predicting malignancy were (OR, 95% CI) chaos of colour (15.43, 1.48–160.3); presence of a typical pigmented network (14.96, 1.68–132.9); increased density of vascular network (3.45, 1.09–10.92); and complete perifollicular grey circles (2.89, 0.96–8.67). A benign pigmented macula was favoured in the presence of looped vessels (0.07, 0.01–0.84) and multiple orange/brown clods (0.22, 0.05–1.03) (Table 3).

Table 3.

Univariable and multivariable logistic regression of dermoscopic features identified in cases versus controls

Dermoscopic features	Univariable	p value	Multivariablea	p value
	OR (95% CI)		OR (95% CI)
Chaos of structure/pattern	9.00 (1.88, 43.14)	0.006
Chaos of colour	14.14 (1.72, 116.5)	0.02	15.43 (1.48, 160.3)	0.02
Grey or blue structures	1.67 (0.42, 6.62)	0.47
Eccentric structureless area	2.45 (0.81, 7.38)	0.11
Thick lines, reticular or branched	0.88 (0.29, 2.65)	0.82
Black dots or clods, peripheral	0.30 (0.07, 1.23)	0.1
Radial lines or pseudopods, segmental	1.91 (0.55, 6.69)	0.31
Polarising-specific white lines	0.86 (0.36, 2.02)	0.73
Polymorphous vessels	1.21 (0.40, 3.71)	0.74
Small angulated lines/polygons	4.78 (2.01, 11.33)	<0.001
Large angulated lines/polygons	1.67 (0.47, 5.94)	0.43
Vascular patternsb
Dots	0.76 (0.24, 2.34)	0.67
Clods	0.57 (0.12, 2.67)	0.47
Linear straight	0.86 (0.30, 2.45)	0.78
Looped	0.10 (0.01, 0.81)	0.03	0.07 (0.01, 0.84)	0.04
Serpentine	2.62 (0.66, 10.32)	0.17
Coiled	1.19 (0.19, 7.44)	0.85
Incomplete perifollicular brown or grey circles	3.50 (1.41, 8.66)	0.007
Incomplete grey circles	3.06 (1.28, 7.30)	0.01
Incomplete brown circles	2.15 (0.92, 5.02)	0.08
Complete perifollicular brown or grey circles	3.83 (1.55, 9.48)	0.004
Complete grey circles	4.02 (1.68, 9.59)	0.002	2.89 (0.96, 8.67)	0.06
Complete brown circles	1.56 (0.69, 3.53)	0.29
Dark brown or black structureless area obscuring the hair follicles	0.58 (0.24, 1.41)	0.23
Perifollicular grey dots	3.02 (1.32, 6.89)	0.009
Brown-to-grey structureless area interrupted by follicular openings	1.61 (0.46, 5.67)	0.46
Irregular	3.41 (1.40, 8.28)	0.007
Regular	0.32 (0.11, 0.94)	0.04
Circle within a circle or double circle	2.17 (0.97, 4.87)	0.06
Small triangular or rounded structures	2.86 (1.26, 6.47)	0.01
Increased density of vascular network	2.55 (1.06, 6.16)	0.04	3.45 (1.09, 10.92)	0.04
Red angulated lines	0.65 (0.23, 1.85)	0.42
White structureless area	1.24 (0.50, 3.12)	0.64
Blue structureless area	0.90 (0.28, 2.91)	0.87
Pigmented reticular lines	3.14 (1.27, 7.73)	0.01
Irregular	1.93 (0.71, 5.26)	0.20
Regular	5.13 (1.07, 24.52)	0.04	14.96 (1.68, 132.9)	0.02
Atypical vessels	1.29 (0.46, 3.67)	0.63
Multiple orange/brown clods	0.21 (0.06, 0.69)	0.01	0.22 (0.05, 1.03)	0.05
Multiple white clods	0.36 (0.08, 1.52)	0.16
Sharply demarcated border over total periphery	0.13 (0.03, 0.61)	0.01
Peripheral concave invaginations	0.84 (0.36, 1.96)	0.69
Scales	0.57 (0.21, 1.54)	0.27
Fine brown parallel curved lines	1.52 (0.68, 3.40)	0.31
Prominent (or pale) follicular openings	0.85 (0.31, 2.30)	0.74

^aThe final multivariable predictive model has been determined from backward elimination technique using all the significant (p-value <0.05) variables from the univariable analysis.

^bNo curved (comma-like) or helical (corkscrew) vessels were identified.

A cross-validated multivariable logistic regression predictive model achieved sensitivity of 85.7% and specificity of 61.4% (Table 2). The diagnostic performance of the model is presented in Figure 2. The AUC on 200 bootstrap resamples was 78.0 (70.3–88.6), similar to the AUC of the original model (81.5, 73.0–90.0). Figure 3 presents a diagnostic flowchart based on the logistic regression, which achieved 87.5% sensitivity and 59.1% specificity.

Fig. 2.

ROC curve of the multivariable logistic regression prediction model based on dermoscopic features to predict malignancy (lentiginous melanomas) against controls (equivocal benign pigmented macules of the scalp).

Fig. 3.

Diagnostic flowchart to discriminate scalp lentiginous melanomas from equivocal benign pigmented macules based on the logistic regression results. The flowchart achieved sensitivity of 87.5% and specificity of 59.1% for the sample analysed. *Figure 1e. **Figures 1a, c. ***Figure 1b, d, f.

Discussion

Most patients included were elderly males. This is the typical epidemiological group that presents with equivocal pigmented macules on the scalp during a dermatological examination. These lesions are usually located on bald areas of the scalp, which are more prone to ultraviolet damage [1–3, 22]. Because hair shafts provide photoprotection, females, who are less often affected by androgenic alopecia, develop significantly less melanomas and also benign lesions caused by chronic sun damage on the scalp [29].

The scalp has anatomical and histological particularities that can influence the presentation of melanocytic lesions, such as high concentration of pilosebaceous units, rich vascular supply, and complex lymphatic drainage [4, 30]. The scalp’s skin is expected to be thick and show elongated rete ridges, the same pattern observed on the trunk and limbs [31]. However, part of it is a transitional area that merges into the face, where the epidermis is thinner and devoid of rete ridges, with a flat dermoepidermal junction (DEJ) [12, 27]. Furthermore, cumulative photoaging of the scalp can cause thinning and flattening of the epidermis even in non-transitional subsites [32, 33]. The variability of these factors could explain why dermoscopic features classically associated to both facial and extrafacial sites can be observed on the scalp [3].

Features described for facial LM were highly prevalent in our sample, such as “small” angulated lines (76.8%) and perifollicular grey dots (69.6%). However, other findings expected to be found on the extrafacial skin were also frequently identified, such as the presence of pigmented reticular lines (44.6%) and large angulated lines (14.3%).

Stanganelli et al. [3] identified that both an irregular pseudo-network (correlated histologically with a flat DEJ) and an irregular pigment network (correlated with asymmetrically pigmented epidermal rete ridges) were present in approximately one-third (32.4%) of the 71 melanomas included. This study included 19 LM but did not discriminate the dermoscopic features according to histological subtype. Furthermore, other typical features of LM were not reported, precluding accurate comparisons with our sample.

A study by Garbarino et al. [22] more recently analysed the dermoscopic features of 97 scalp melanomas, including 64 LM. A high prevalence of features associated with both facial and extrafacial skin “architecture,” such as irregular pigment network (34.4%), irregular pseudo-network (59.4%), rhomboidal structures (37.5%), and annular-granular pattern (57.8%) was reported, similarly to our study.

It could be hypothesised that melanomas located more anteriorly on the scalp could manifest different dermoscopic patterns as compared to those located posteriorly. This was, however, proven not to be true in our sample. None of features analysed were able to significantly differentiate melanomas located anteriorly to the coronal suture from those located posteriorly. This finding suggests that there is heterogeneity of the scalp’s skin over the scalp and consequently of dermoscopic patterns, possibly related to variable degrees of photoaging and flattening of the DEJ.

The data presented indicate that differentiating scalp LM from melanoma mimics presenting as equivocal pigmented macules can be challenging. Although many dermoscopic features were statistically more associated with malignancy, there was considerable overlap between cases and controls, with classic melanoma features present with relatively high frequency in benign lesions.

The overlap of dermoscopic features has also been reported by previous authors who compared LM against benign pigmented macules of the face [12–14, 16]. A study by Akay et al. [13] demonstrated that facial pigmented AK can be strikingly similar to LM, presenting multiple dermoscopic features associated with malignancy, except black blotches.

Features associated with regression, such as grey dots, grey or blue structures and white structureless areas, were equally prevalent in cases and controls in our study. Recommendations for performing a biopsy for any scalp lesion with regression [4] could, therefore, lead to unnecessary interventions.

The diagnostic uncertainty on dermoscopy was corroborated by the suboptimal sensitivity (76.8–78.6%) on a background of relatively low specificity (54.5–56.8%) of both experts for this sample and the suboptimal interobserver agreement for the final specific diagnosis and discrimination of benign and malignant lesions. These findings contrast to higher sensitivity and specificity for the dermoscopic diagnosis of melanoma on a systematic review (94% sensitivity, 96% specificity for image-based evaluation of equivocal lesions) [34].

The prediction model was able to improve diagnostic accuracy, as compared to the experts’ assessment. Sensitivity increased to 85.7% (73.8–93.6) and specificity to 61.4% (45.5–75.6) using the six most significant discriminative features (4 positive: presence of chaos of colour, regular pigmented network, increased density of vascular network and complete perifollicular grey circles; 2 negative: absence of looped/hairpin vessels and multiple orange/brown clods). The diagnostic flowchart is useful for clinical use but requires external validation.

Interestingly, the presence of a pigment network, even if regular/typical, was associated with malignancy. This could be explained by the fact that malignant melanocytic lesions were compared to non-melanocytic benign pigmented lesions, and the presence of a pigment network, even if regular/typical, favours a melanocytic nature in this context [27] and helps to differentiate the groups. Although scalp melanocytic naevi in adults typically present as faint pigmented or skin-coloured nodules, with a smooth or papillomatous surface [4], 18.7–34.6% of them can actually show a typical or atypical pigment network on dermoscopy [3, 22]. However, scalp melanocytic naevi were intentionally not included in this study as controls, as previously assessed by other authors [3, 22], and considering our clinical impression that the main mimics of LM, which tend to arise in chronically sun-damaged scalps, are SL, AK, SK, and LPLK, and less often melanocytic naevi [4].

Even with the proposed diagnostic flowchart, not all uncertainty can be eliminated for this group of particularly challenging lesions, as 7 (12.5%) melanomas would have been missed and 18 (40.9%) benign lesions would have been excised. A cost-effective approach to increase diagnostic accuracy in this scenario would be adding a second noninvasive diagnostic step with in vivo reflectance confocal microscopy (RCM) for lesions showing equivocal features on dermoscopy [35, 36]. RCM has been proven to be an accurate tool (93% sensitivity, 82% specificity) to differentiate LM from equivocal pigmented macules of the face [37], and the main positive features for malignancy on RCM were also observed in scalp melanomas [22, 38, 39]. RCM can also provide guidance for partial biopsies of large lesions, reducing unnecessary excisions [37, 39].

The four centres included in this study are known for having high thresholds to recommend excision, which is based not only on a clinical/dermoscopic suspicion, but also on the evidence of a dynamic behaviour (new or changing lesions) and/or suspicious features on confocal microscopy, commonly recommended for large lesions in cosmetically sensitive areas. All the included controls had a histopathological confirmation, which was an inclusion criterion and indicates that these were particularly difficult-to-diagnose lesions, potentially representing a referral centre bias. Most likely, in real-life situations, dermoscopy alone would be enough to discriminate most benign pigmented lesions arising on the scalp from melanoma, which would then not be considered for further diagnostic investigation. The relatively small sample size is an additional limitation, explained by the infrequent nature of scalp LM, added by the necessity of pretreatment photographic documentation as an inclusion criterion.

In conclusion, scalp LM show typical dermoscopic features described for both facial and extrafacial sites. Anterior or posterior location does not influence their dermoscopic presentation. There is considerable overlap of dermoscopic features of LM and benign pigmented lesions arising on the scalp, leading to low diagnostic accuracy and interobserver agreement. A diagnostic flowchart is proposed, which should increase the diagnostic accuracy of dermoscopy to discriminate benign and malignant pigmented macules of the scalp.

Key Message

Facial and extrafacial dermoscopic features were identified in scalp melanomas, overlapping considerably with equivocal benign pigmented lesions.

Statement of Ethics

This study protocol was reviewed and approved by the Federal University of Sao Paulo Ethics Committee (CAAE: 90241718.6.0000.5505). Both Australian participating centres also obtained approval from the Sydney Local Health Ethics Committee (X15-0311; 2019/ETH06854). Participants consented to the use of their de-identified information and images for research purposes by means of written informed consent.

Conflict of Interest Statement

The authors have no conflicts of interest to declare.

Funding Sources

No funding was obtained for the study.

Author Contributions

Amanda Regio Pereira: conceptualisation, methodology, data acquisition, investigation, data analysis, writing – original draft, writing – reviewing and editing, project administration; Sergio Hirata: conceptualisation, methodology, investigation, writing – reviewing and editing, supervision; Pawel Pietkiewicz: methodology, investigation, writing – reviewing and editing; Scott W. Menzies: data analysis, validation, writing – reviewing and editing; Gabriella Brancaccio: data acquisition, writing – reviewing and editing; Helena Collgros: data acquisition, writing – reviewing and editing; Giuseppe Argenziano: data acquisition, writing – reviewing and editing; Serigne Lo: data analysis, writing – reviewing and editing; Tasnia Ahmed: data analysis, writing – reviewing and editing; Riccardo Pampena: data acquisition, writing – reviewing and editing; Caterina Longo: data acquisition, writing – reviewing and editing; and Pascale Guitera: conceptualisation, methodology, writing – reviewing and editing, supervision.

Funding Statement

No funding was obtained for the study.

Data Availability Statement

All data generated or analysed during this study are included in this article and its supplementary material files. Further enquiries can be directed to the corresponding author.