The patterns of birthmarks suggest a novel population of melanocyte precursors arising around the time of gastrulation

Summary Systematic work in the mouse and chicken has mapped out two neural crest‐derived pathways of melanocyte precursor migration. With these in mind, this study reappraises the patterns of congenital pigmentary disorders in humans and identifies three recurrent patterns consistent across genetically different diseases. Only two of these are seen in diseases known to be melanocyte cell‐autonomous. The segmental pattern correlates well with the classical dorsolateral population from animal studies, demonstrating respect of the midline, cranio‐caudal axial mixing, unilateral migration and involvement of key epidermally derived structures. Importantly however, the melanocyte precursors responsible for the non‐segmental pattern, which demonstrates circular, bilateral migration centred on the midline, and not involving key epidermally derived structures, have not been identified previously. We propose that this population originates around the time of gastrulation, most likely within the mesoderm, and ultimately resides within the dermis. Whether it contributes to mature melanocytes in non‐disease states is not known; however, parallels with the patterns of acquired vitiligo would suggest that it does. The third pattern, hypo‐ or hyperpigmented fine and whorled Blaschko's lines, is proposed to be non‐cell‐autonomous.

In humans, although there is no reason to believe that melanocytic development should be any different, it is worth noting that dynamic studies of migration in early embryogenesis have not been possible for ethical reasons, and therefore, the small amount of data available has been largely gleaned from static immunohistochemical studies from the age of 6 weeks onwards (equivalent to E11 in mice). From these data, human melanocyte precursors appear to enter the epidermis at week 6 and are clearly distinguishable by HMB-45 positivity (which detects gp100/PMEL) by seven weeks of gestation [E12.5 mouse] (Holbrook, Underwood, Vogel, Gown, & Kimball, 1989). Later, HMB-45-positive cells are abundant in the epidermis and are still present in the developing dermis throughout 3-7 months of gestation [E13.5-E18.5 mouse] (Luciani et al., 2011;Suder & Bruzewicz, 2004). Mature perifollicular and intrafollicular epidermal pigmented melanocyte populations are maintained in post-natal life, located at the basal layer of the epidermis, and mature melanocytes have also been found in sebaceous glands (Jang et al., 2014). There is also evidence for the common presence of dermal melanocytes in children of at least some ethnicities up to the age of 10 years, in the form of the common non-pathogenic pigmentation Mongolian blue spots (also known as dermal melanocytosis), and indeed some evidence of persistence throughout life (Kikuchi & Inoue, 1980). Distinct melanocyte stem cell populations have been reported and characterized in adult mice and humans, with specific hair follicleassociated and skin-derived dermal stem cells both exhibiting capability for differentiation into melanocytes in vitro (Amoh, Li, Katsuoka, Penman, & Hoffman, 2005;Fernandes et al., 2004;Glover et al., 2015; Nishikawa-Torikai, Osawa, & Nishikawa, 2011;Nishimura, 2011;Osawa et al., 2005;Schneider, Dieckmann, Rabe, Simon, & Savkovic, 2014;Yu, Kumar, Kossenkov, Showe, & Xu, 2010;Yu et al., 2006).
There are, in addition, dermal stem cells in glabrous skin, which is devoid of hair follicles (Li et al., 2010), and in sweat glands (Grichnik et al., 1996;Okamoto et al., 2014), both capable of differentiation into melanocytes in culture. The renewal of mature melanocytes is presumed to occur from these stem cell populations, with recent evidence demonstrating that the non-follicle-associated stem cells do indeed contribute to the interfollicular melanocyte population, at least in the mouse (Glover et al., 2015). What is not known, however, is whether these different melanocyte stem cell populations have different embryological origins, and whether they maintain different mature melanocyte populations. Importantly, it is conceivable that an improved understanding of these factors might lead to an improved understanding of acquired melanocytic diseases such as vitiligo or melanoma.
In the absence of an approved ethical method to study human melanocytic development systematically under laboratory conditions, it is possible to use a more traditional tool in the clinical academic's armoury-the study of the patterns of rare congenital diseases of pigmentation (relevant examples of which are outlined in Supplementary Material). Patterns of these diseases have been described phenotypically for many years (Blaschko, 1901;Happle, 1993;Torrelo, Baselga, Nagore, Zambrano, & Happle, 2005); however, many descriptions were undertaken before knowledge of the underlying genetics and of the most recent findings in embryology. In particular, until recently there were limited data on which conditions were definitely due to post-zygotic mutations leading to mosaicism and in which clonality of pigmentary patterns has been proven at the level of the founder mutation. This study therefore set about reappraising patterns and their implications in the light of the latest molecular and embryological data.

| METHODS
Given the number and complexity of patterns of human congenital pigmentary disorders, some premises and hypotheses were established for pattern interpretation.

| Premises and hypotheses on which pattern interpretation is based in this study
Firstly, a small single birthmark can occur in any place due to a late postzygotic mutation of any cell type in the skin. These are therefore not informative for developmental pattern interpretation and are not classifiable on that basis. Interpretation was focused therefore on multiple (two or more of any size or site, where the same causative mutation has been shown for that diagnosis to be present in both birthmarks) or extensive birthmarks (estimated diameter of more than 10 cm at birth).
Secondly, it is proposed that congenital pigmentary patterns may represent persistent embryological or foetal developmental patterns, which do not necessarily correspond to post-natal patterns of cellular organization. This is supported by evidence from systematic studies of other birthmarks, notably infantile haemangiomas (Haggstrom, Lammer, Schneider, Marcucio, & Frieden, 2006;Weitz et al., 2014) and capillary malformations , the patterns of which correspond to the fields of the embryonic facial placodes, rather than to adult vascular supply.
Thirdly, post-zygotic or mosaic clonal disorders can be particularly powerful at revealing the patterns of normal development. This is because the pattern visualized is of necessity limited to the distribution covered by the offspring of a single precursor cell in that disease state. The patterns in these disorders therefore permit mapping of the maximal individual melanocyte precursor fields at a time when that mutation is compatible with survival of the embryo. Smaller areas of abnormal pigmentation should fit within any one of these fields, representing a later (although usually identical) mutation in embryogenesis. Fourthly, where it is possible to identify recurrent congenital patterns of pigmentation in mosaic clonal diseases of known and distinct genetic aetiology, we can reasonably assume that the causal mutations do not alter the final melanocyte precursor "fields"-in other words, the area of the skin covered by melanocytes from a single melanocyte precursor-even if the numbers or morphology of the cells is altered by the disease state. Such diseases therefore allow us to visualize the patterns and extents of normal melanocytic precursor fields, effectively frozen in the embryonic distribution not usually visible in the refined and remodelled phenotype of cutaneous pigmentation seen by the time of birth. Conversely, very rare patterns of pigmentation, which are not yet known to be shared by more than one genetically different disease, may be secondary to a genetic abnormality which interferes with melanocyte precursor migration. These include the sash-like pattern of Ruggieri-Happle syndrome (Ruggieri, 2000), the disordered swirling linearity of Pallister-Killian syndrome and phylloid patterns (Happle, 1993). As a result, these were not considered alongside the recurrent patterns in this work.
Fifthly, we did not assume that all congenital patterns of pigmentation on the skin are melanocyte cell-autonomous. When known melanocytic disorders are not seen in certain patterns, these patterns are more likely to be due to primary abnormalities of other cell types (e.g., keratinocytes) and are visible by the effect they have on the pigmentary system. Lastly, where congenital pigmentary disorders consistently spare or involve structures of known embryological lineage and developmental timing, these data can be used to formulate hypotheses about the origin of melanocytes in those structures, for example, structures which arise in situ from the surface ectoderm or epidermis, such as the palpebral fissures of the eyelids and the eyelashes (from the surface ectoderm in the 10th-11th weeks of gestation; Andersen, Ehlers, & Matthiessen, 1965), the areola of the nipple (from the epidermis in approximately the fifth month of gestation; Hughes, 1950), and the vermilion border of the lips (from the surface ectoderm overlying the component facial placodes after the eighth week; Coslet & Cohen, 1967;Warbrick, 1960). As melanocytes from the neural crest are already within the epidermis from 8 weeks, clonal pigmentary patterns which involve these epidermal structures could be considered to have arisen from these neural crest melanocytes, whereas those which spare these structures could be considered to have arisen independently of the neural crest melanocytes, or at least independently from those which reached the epidermis.

| Resources used
With these premises in mind, a systematic review was undertaken in mosaic disorders were identified. Attention was paid to patterns that might be expected but were never seen, such as fine and whorled Blaschko's lines in known melanocyte cell-autonomous disorders.
The criteria for deciding whether a pattern of pigmentation was melanocyte cell-autonomous were whether it has been seen when the genetic abnormality is proven to affect melanocytes, or where the abnormal cells are highly likely to be melanocytic cells (such as naevus cells). Included in this definition were the post-zygotic mosaic conditions: multiple congenital melanocytic naevi (CMN, usually caused by NRAS mutations; Kinsler et al., 2013), phakomatosis pigmentovascularis and extensive dermal melanocytosis (PPV and EDM, respectively, usually caused by GNA11/GNAQ mutations; Thomas et al., 2016), phakomatosis pigmentokeratotica (PPK, usually caused by HRAS mutations; Groesser et al., 2013), congenital naevus spilus (caused by HRAS or NRAS) Sarin et al., 2013), McCune-Albright syndrome (MAS, caused by GNAS mutations; Weinstein et al., 1991) and mosaic Neurofibromatosis type 1 (NF1, caused by NF1 mutations (Ainsworth, Chakraborty, & Weksberg, 1997); see Supplementary Material for further details of these conditions). Of note, the pigmentary patterns in these mosaic disorders have been found and published to be clonal at the level of the causative mutation in multiple patients with the same diagnosis, even where there is geographical separation of pigmentary lesions on different body parts. In this large group of patients studied here, however, not every patient has been genotyped, although many have. In addition to these mosaic disorders, the patterns of the autosomal dominant condition piebaldism (germline KIT mutations; Fleischman, Saltman, Stastny, & Zneimer, 1991;Giebel & Spritz, 1991, where the phenotype is known to be cell-autonomous; Mayer & Green, 1968), was studied, after definition of the fields using the mosaic disorders.

| RESULTS
Three major recurrent patterns of congenital pigmentary disorders were identified. Where it was not possible to obtain consent for publication of specific features, references have been given which demonstrate these features in photographs. We propose that one of these patterns is melanocyte non-autonomous, corresponding to fine and whorled Blaschko's lines (Blaschko, 1901;Happle, Fuhrmann-Rieger, & Fuhrmann, 1984;Sorlin et al., 2017), and two are melanocyte cell-autonomous. These two are here termed segmental and non-segmental, in an attempt to link them to existing terminology, although their characteristics are redefined here. The segmental and non-segmental patterns are proposed to represent distinct melanocyte precursor populations, on the basis that they are not seen together in clonal mosaic diseases in one individual, they have different defining characteristics (Table 1), and the fields of each pattern come together in a jigsaw-puzzle-like manner to cover more or less the entire integument in two different ways. The segmental pattern corresponds well to the classical neural crest dorsolateral population, whereas the nonsegmental population is currently unstudied in a laboratory setting.

| Two recurrent patterns of congenital melanocyte cell-autonomous pigmentary disorders
The key phenotypic features of the two patterns as defined here are given in Table 1. Where it was not possible to obtain consent for photographs of specific fields, references have been given to photographs in the literature.

| The segmental pattern-"quadrilateral shapes"
This pattern is seen very commonly in naevus spilus (maculosus and papulosus), PPK and MAS, in congenital café-au-lait macules of NF1, regularly in EDM, and rarely in CMN. The pigmentation in this pattern covers nearly the whole integument, including that of epidermal structures such as the vermilion border of the lips, the eyelids includ- Notably, the segmental pattern spectrum does not seem to have areas of overlap between the fields derived from single melanocyte precursors, on the basis of mapping these fields onto an adult human body map. With extensive birthmarks, these segmental patterns include both the dorsal and ventral surfaces, whereas smaller birthmarks within these fields can affect only one or other. The segmental pattern for the scalp skin is not yet clear, as scalp hair obscures many of the photographs studied and further work will be needed to determine this.
T A B L E 1 Comparison of the key phenotypic features of the two melanocyte cell-autonomous patterns of congenital pigmentary disorders and an estimation of melanocyte precursor numbers at a time in development when the pigmentary abnormalities can exist without necessary abnormalities of other organ systems

Segmental-"quadrilateral shapes"
Non-segmental-"round shapes" Pattern edges Straight edges and linear patterns, which translates into parallel transverse borders across trunk and face, and linear edges along length of limbs Convex-outwards edges and therefore circular or ovoid patterns, which translates into rounded edges on trunk and head and transverse edges circumferentially around limbs Pattern types Band (bilateral), flag (unilateral), checkerboard (bilateral) and broad Blaschko-linear (bilateral Figure 1g,h, respectively, as this is a spectrum-numbers below are for 1h) Trunk/Limb 10 dorsal truncal/limb (5 each side of the dorsal midline which are contiguous with the ventral surface) Head Four for face and neck (two each side of midline, contiguous dorsal and ventral; Figure  No, none or very little overlap between fields from single melanocyte precursors, and no areas of paucity of coverage other than palms, soles and digit tips Yes, substantial overlap between fields from single melanocyte precursors, and conversely also areas of paucity of coverage around mouth, eyes, chin, anterior/ ventral trunk, knees, digit tips The crude larger-segmental pattern of pigmentation on the trunk is delimited by sharp transverse edges (Figure 1c,d,g). These are usually also delimited by the posterior/dorsal and/or the anterior/ventral midline, leading to a quadrilateral flag-shaped pattern of unilateral blocks of pigmentary change, or a checkerboard pattern where multiple segments are affected at different craniocaudal levels (Biesecker & Spinner, 2013;Happle, 1993;Tadini et al., 1998). In some cases, the pattern can be bilateral, with transverse blocks either side of the midline at the same cranio-caudal level, leading to a band-like pattern at one or multiple cranio-caudal levels ( Figure 1d; Happle, 2014). In these large-segment patterns, there appears to be four cranio-caudal segments on either side of the midline, constituting eight segments in all ( Figure 1g, Table 1).
This segmental type of pattern is also seen in some of the melanocytic diseases (particularly MAS, and mosaic NF1) with a modest increase in numbers of segments to approximately 14 ( Figure 1h, Table 1) and begins to take on a slightly more complex outline sometimes termed broad Blaschko-linear patterning (Happle, 2014) (Figure 1e,f,h). Notably, this is distinct from fine and whorled Blaschko's lines (Blaschko, 1901;Happle et al., 1984), not having the complexity of the lines and whorls, but are rather a series of broad linear bands which show a midline dip.

| The non-segmental pattern-"round shapes"
This pattern is seen commonly in CMN, EDM and PPV, and rarely in large congenital café-au-lait macules in NF1.  Figure 5). There is however considerable overlap between the fields, based on mapping the fields onto an adult human body map ( Figure 5). There are also areas of paucity with this population, particularly around the eyes, the mouth, to some degree on palms and soles, and at digit tips/nail beds (Figure 5l).

Dorsal non-segmental pattern-trunk and head
There are two dorsal truncal fields, upper and lower, which correspond to what has classically been described in clinical texts as "cape"  (Figures 2a,b,e, 5a, 6d, 7c,d and 8e).
The circumferential anterior/ventral joining of this field is however very variable between individuals in all diseases studied, even with the maximal extent of the field affected on the posterior/dorsal axis, and the anterior/ventral surface of the trunk is often poorly supplied by this field. Posteriorly/dorsally, the lower border is usually convex downwards, with a slight upwards slope from posterior/dorsal to anterior/ventral, resulting in a transverse or concave-upwards lower border on the anterior/ventral trunk (Figures 2a,b, 5a, 6d, 7c, 8e).
The lower truncal field is circumferential on the trunk, extending pos- There is a single dorsal cranial field that is centred bilaterally and symmetrically on the midline of the scalp around the vertex (Figure 3a,e).

Dorsal non-segmental pattern-limbs
The four distal limb fields exhibit a glove-and-stocking-type distribution, from above the fingertips up to the shoulder on the upper limb, and from above the toes up to below the knee (Figures 2g-j,m,n and 5c,d). In both upper and lower limbs, the upper cut-off appears as a transverse line circumferentially across a limb (Figure 2g-j,n) and they appear to develop from dorsal to ventral (Figure 2m). Although these are most commonly seen as unilateral single limb patterns, in the presence of dorsal truncal field involvement, bilateral symmetrical lower limb fields can also be seen in mosaic disorders, with sparing of the mid-limbs (Gupta et al., 2016;Nanda et al., 2016).
The two mid-limb fields are centred on the knees, and extend from the mid-thigh to mid-calf (Figures 4b,c,e and 5k). They also have transverse circumferential upper and lower borders and appear to develop from dorsal to ventral. It is possible there are similar mid-limb fields on the upper limbs, and axillary fields, but there are insufficient data to support this currently. The mid-limb fields appear later than the distal limb fields and appear to be specified unilaterally.
Ventral non-segmental pattern-trunk and head field is currently less clearly defined but appears to be centred on the chin (Figure 3d,f).

The "spray can" effect
The "spray can" effect is found in the non-segmental pattern but not the segmental pattern ( Figure 6) The "spray can" epithet is sug- foci can also be visible within the larger area of pigmentation, depending on the condition (Figure 6c,d,f).
There is sometimes a visible stepwise difference in intensity of pigmentation within the larger focus, from more concentrated in the centre to paler at the edges. This is suggestive of overlaid concentric layers or waves of diffusion, or successive "sprays" from the central focus (Figure 6e-g). The "spray can" effect can also be seen in piebaldism. In this case, the small "sprays"/foci are of normal pigmentation, made visible by the pathogenic lack of background pigmentation in certain areas (Figure 4f-h).

Digit tips and nail beds
The finger and toe tips and the melanocyte population of the nails are not involved by either the segmental or non-segmental pattern 2h,n). Furthermore, in individuals with large pigmented birthmarks and multiple smaller foci elsewhere, the nail bed pigmentation is not usually affected by the smaller foci. This suggests that the nail beds are supplied by their own melanocyte population not addressed here.

Palms and soles
The palms and soles can be involved in the non-segmental pattern, as part of the distal limb field (Figure 2h). In addition, small foci in the spray can pattern can affect palms or soles if the large focus is in any other field. The segmental pattern abnormalities do not involve the palms or soles, despite their ventral/flexor surface coverage of the rest of the limb.

The areola of the nipple
The non-segmental pattern frequently spares the areola (Baykal, Solakoglu, Polat Ekinci, & Yazganoglu, 2015), even usually where it is completely surrounded by the pigmentary abnormality in the truncal fields (Figure 7c,d). However, this should not be taken as an absolute rule, as in some cases, the areola is clearly involved in the pigmentary abnormality (Figure 7b; Gangireddy & Coleman, 2013), and in those cases, it may also not properly formed as a structure. The segmental pattern on the other hand does involve the pigmentation of the areola where the surrounding skin is affected (Chebel et al., 2010;Tadini et al., 1998;Tekin, Yucelten, & Happle, 2016). we propose that the non-segmental "round-shape" pattern represents a hitherto-undescribed population of melanocyte precursors.
Caveats surrounding this proposal are that human early embryonic development cannot be studied directly, that we have not taken into account possible effects of interactions between mutant and wildtype cells in the mosaic clonal disorders, and that there could be an unknown and unsuspected commonality between the genotypically distinct diseases we have studied that is responsible for their identical patterns.
It is not clear when, and from which germ layer, this novel population arises, and exactly where it ends up. There are however important clues which can lead us to a hypothesis. We can observe that it colo- can be involved as well as the caudal truncal field (Gupta et al., 2016;Nanda et al., 2017), or facial fields as well as cranial and caudal truncal fields (Raina & Chaudhuri, 2010;Ramesh et al., 2017), support this hypothesis. The alternative would be that the non-segmental melanocyte precursor for each of these fields arises independently in that new structure.
Fascinatingly, the non-segmental pattern fields are also observable in piebaldism; however, in this case, the abnormal skin is hypo-or depigmented, and it is the normal skin that delineates the upper and lower dorsal truncal fields, and distal limb fields (Figure 4f-h). This germline condition has been shown in mice to cause a reduction in numbers of dorsolateral melanoblasts with an increase in their diffusion rate (Mort et al., 2016). It is also potentially possible that the KIT mutation affects certain melanocyte populations more than others, for example, the ventral truncal and mid-limb fields of the non-segmental population (Figure 4a-d) with variations in phenotypic severity due to the exact mutation and other germline modifiers.
The terminology here is difficult to define from these observations alone. We are proposing that ultimately, a single cell gives rise to a whole population of melanocyte precursors, however hesitate to term it a melanoblast. Diseases causing pigmentary patterns corresponding to the single-cell stage (Figure 2k,l) can also give rise to abnormalities of the melanocytes of the eye, and blood vessels of the skin, eye and brain, but these are not universal. At the stage when we can map the individual fields as detailed in Figures 2a-l, 3 and 5, there can be congenital abnormalities of the brain parenchyma and/or leptomeninges; however, these are only seen in a minority of cases.
The segmental patterns of pigmentation appear to correlate with what is currently known of the dorsolateral melanoblast population.
Alternating mosaic and proven clonal patterns from a single postzygotic mutation, such as the checkerboard pattern in PPK or MAS, directly infer some cranio-caudal axial mixing of the dorsal melanocyte precursors before migration, already documented in this population in mice (Wilkie, Jordan, & Jackson, 2002). Occasionally, these segmental patterns appear truly bilateral at one cranio-caudal level ( Figure 1d). Whether this is truly a representation of a single precursor cell leading to bilateral migration from the neural crest, or simply a version of the checkerboard pattern which by chance occurred at the same cranio-caudal segment is not clear, but the same occasional occurrences were described in early murine chimaeric experiments (Mintz, 1965).
As regards the number of single melanocytic precursors in the segmental population, the estimate in the current study is approximately 12-16 (Table 1, (Mintz, 1967) and later mosaic modelling of melanocyte migration (Huszar et al., 1991). These two studies also concluded that there are very few "founder melanoblasts," although this has been disputed by subsequent studies (Mort et al., 2016;Wilkie et al., 2002). Interestingly, a recent estimate of the number of founder cells for the entire blood lineage is of a similar order, at 10 (Rahbari et al., 2016). We suggest that these apparently conflicting findings in numbers of melanoblasts could in fact be compatible with each other and may be due to differences in definition of the word "melanoblast." There may be a small number of earlier dorsolateral/segmental population founder precursor cells giving rise to a larger number of what are defined molecularly as melanoblasts in the mouse at E11.5. It may also be, however, that the existence of the proposed non-segmental population of melanocyte precursors confounds mathematical modelling of final melanocyte numbers in the skin, particularly if this population only becomes apparent by classical melanoblast labels at a certain embryonic stage.
The absence of known melanocytic disorders in a fine and whorled Blaschko linear suggests that this pattern of pigmentation is non-cellautonomous, the result of mosaicism or chimaerism affecting keratinocytes (either at DNA or RNA level), leading to a pigmentary phenotype.
This would be in line with the basis for the brindle pattern in dogs, proposed to be dictated at the level of keratinocytes not melanocytes (Kerns et al., 2007), and with recent evidence from a mosaic KITLG mutation in one patient (Sorlin et al., 2017). This will be testable by single-cell analysis in patients with this phenotype and in whom the causative mosaic or chimaeric mutation has been established.
The question then arises as to which mature melanocytes the segmental and non-segmental populations relate to at birth. Patterns related to specific structures can potentially be informative in addressing this, and in helping deduce the timing of development. The segmental patterns of pigmentation are demonstrably associated with the epidermis, given the noticeable disregard for the boundaries at the areola, the lips and the eyelids, with the pigmentary abnormality proceeding right through these structures (Figures 1a   and 7f). This suggests that the epidermis had already developed or was developing at the same time as these melanocyte precursors.
An elegant supporting observation is that segmental pattern pigmentary birthmarks take the place of freckles (normal ephelides) in individuals who have them (Figures 1a and 7f). This is in sharp contrast to what is usually seen with the non-segmental pattern where the boundaries of these epidermal structures act as an interruption to pigmentation, strongly suggesting that the epidermal structures developed at a later stage or at least independently of the non-segmental melanocyte population. A correlate of this is that non-segmental disorders take the place of mongolian blue spots where these abut each other.
Perhaps the most important potential sequela of identifying these two melanocyte cell-autonomous patterns is an improved understanding of acquired pigmentary diseases. There are, for example, striking parallels with the patterning of segmental and nonsegmental vitiligo (Ezzedine, Eleftheriadou, Whitton, & Van Geel, 2015;Picardo et al., 2015). In non-segmental vitiligo, depigmentation tends to start in what correlates to the areas of paucity of the non-segmental fields shown in Figure 5i. Furthermore, the melanocyte populations of the eyelashes and eyebrows are typically unaffected in early disease, as are the areolae of the nipples and the lips (Faria, Tarle, Dellatorre, Mira, & Castro, 2014). In contrast, the affected areas in segmental vitiligo closely parallel, and are restricted to, the segmental pattern of melanocytic disorders described here, and in the appropriate area will involve the epidermal structures of the eyelashes, eyebrows, areolae and lips from early in the disease (Faria et al., 2014). These subtypes of vitiligo can therefore be proposed as diseases of different melanocyte populations, independent of the disease mechanism itself.
In conclusion, we propose that a novel population of melanocyte precursors arises around the time of gastrulation within the mesoderm and migrates centrifugally by proliferation and diffusion from dorsal and ventral midline foci to take up residence in the dermis. It is not known whether this population contributes to normal pigmentation, or whether melanocytic differentiation is only seen in congenital genetic disease conditions. However, given the patterns seen in vitiligo, we suggest that this new population does contribute to normal pigmentation of the skin in humans, and most likely in other vertebrates.