Mapping of Segmental and Partial Segmental Infantile Hemangiomas of the Face and Scalp

This cohort study aims to refine previous maps defining the distribution of facial segment hemangiomas and determine whether patterns of indeterminate hemangiomas are related to their segmental counterparts.

Key Points

Question

What are the anatomic patterns of segmental and indeterminate infantile hemangiomas (IHs) on the face and scalp?

Findings

In this cohort study, 549 segmental and indeterminate IHs were mapped onto standardized templates. Borders of the maxillary and mandibular segments of facial hemangiomas in the preauricular region differed from previous maps; indeterminate hemangiomas fell within defined facial segments, suggesting that they are partial segmental hemangiomas; and a new pattern for segmental hemangiomas involving the lateral scalp was defined.

Meaning

Enhanced understanding of segmental IH patterning can provide insights into pathogenesis during embryonic development and improve risk prediction for associated extracutaneous anomalies.

Abstract

Importance

Recognizing segmental infantile hemangioma (IH) patterns is important for risk stratification and provides clues to pathogenesis. Previously, segmental hemangiomas were mapped to 4 facial regions, 3 corresponding to known facial metameres.

Objectives

To refine existing maps of facial segmental IHs, examine so-called indeterminate hemangiomas as they relate to known segmental patterns, and define a novel pattern of segmental scalp hemangiomas.

Design, Setting, and Participants

This retrospective cohort study was conducted at 4 pediatric dermatology centers (University of California, San Francisco; Indiana University; Medical College of Wisconsin; and Northwestern University/Ann & Robert H. Lurie Children’s Hospital of Chicago) using photographic archives of patients younger than 12 years with segmental and indeterminate hemangiomas on the face and scalp. Clinical images were used to map hemangioma distribution onto standardized facial templates. Heat map densiometry identified recurrent patterns that were compared with previously published patterns of facial segmental hemangiomas. Patterns of indeterminate hemangiomas were compared with those of segmental hemangiomas. Data collection took place in 2017, and analysis took place from 2017 to 2019.

Main Outcomes and Measures

Distribution and patterning of segmental and indeterminate IHs of the face and scalp.

Results

A total of 549 IHs were mapped. The borders of the frontotemporal (S1) and frontonasal (S4) segments agreed with previous segmental maps; however, the maxillary (S2) and mandibular (S3) segment borders differed with respect to the preauricular skin. In contrast with previous reports, preauricular skin segregated with the mandibular (S3) rather than the maxillary (S2) segment. Indeterminate hemangiomas occurred within and respected the same borders as segmental hemangiomas. Hemangiomas on the lateral scalp commonly occurred in a C shape extending from the posterior auricular region.

Conclusions and Relevance

This cohort study provides an updated map of facial segmental IHs with redefined maxillary (S2) and mandibular (S3) segment borders. It provides evidence that indeterminate hemangiomas are partial segmental hemangiomas respecting anatomic boundaries of their larger segmental counterparts. A newly recognized C-shaped pattern of segmental scalp hemangioma is reported.

Introduction

The distinction between localized infantile hemangiomas (IHs) and segmental IHs has helped in both risk stratification and understanding of hemangioma pathogenesis. Localized IHs are spatially confined, often surrounding a central focal point, whereas segmental IHs involve an anatomic territory, with many corresponding to known embryologic developmental units. Some IHs have been more difficult to classify and have been referred to as indeterminate.^1,2 Indeterminate hemangiomas lack the round or ovoid shape of many localized hemangiomas yet are smaller in size than classic segmental lesions and fail to encompass the majority of a segmental unit as defined by IH segment maps.¹

Infants with segmental IHs have an increased risk of complications, including association with underlying structural anomalies (eg, PHACE [posterior fossa malformations, hemangioma, arterial anomalies, cardiac defects, eye anomalies] syndrome or LUMBAR [lower body hemangioma, urogenital anomalies and ulceration, myelopathy, bony deformities, anorectal malformations, arterial anomalies, and renal anomalies] syndrome), functional compromise, or ulceration, and are more likely to require treatment.³ Segmental scalp IHs can also be associated with PHACE syndrome,⁴ but their patterns have not been clearly delineated. Having reliable maps to define hemangioma patterns is important for proper recognition and risk stratification. This cohort study aims to refine previous maps defining the distribution of facial segments and determine whether patterns of indeterminate hemangiomas are related to their segmental counterparts.

Methods

Investigators at 4 pediatric dermatology centers (University of California, San Francisco; Indiana University; Medical College of Wisconsin; and Northwestern University/Ann & Robert H. Lurie Children’s Hospital of Chicago) used secure photograph archives to identify all patients younger than 12 years who had head and neck segmental and indeterminate hemangiomas. For photographs to be included in the study, the anatomic borders of the hemangiomas had to be clearly visible. Institutional review board approval was obtained from each participating institution with need for patient informed consent waived based on exempt status of the study design.

Segmental hemangiomas were defined as hemangiomas that covered a broad, specific anatomic territory. Indeterminate hemangiomas were typically larger and more plaquelike than localized hemangiomas, did not encompass an entire segment, and were not round or oval but did respect anatomic boundaries similar to segmental hemangiomas. Using a detailed study manual, the investigators were trained to use standard definitions to categorize hemangiomas and provided detailed instructions regarding mapping techniques. Inclusion criteria in terms of anatomic locations involved and segment boundaries were clearly defined for each segment.

Anatomic mapping of hemangiomas was performed by representing the distribution of the hemangioma with a series of points plotted on a standard template that were connected to form a unique polygon. Specially designed heat map software allowed investigators to overlap the polygons to visualize the data. Qualitative analysis of heat maps identified the borders of segmental hemangiomas on the face and compared them with previously published maps of IH patterning. Distribution patterns for indeterminate hemangiomas were identified and compared with segmental maps. Segmental scalp hemangioma distributions were also mapped to identify their patterns.

Results

A total of 549 IHs on the face and scalp were mapped, including 181 segmental and 297 indeterminate IHs on the face, 27 IHs on the ear, and 44 IHs on the scalp (Table). The distribution of frontotemporal (S1) and frontonasal (S4) hemangiomas was remarkably similar to previously reported diagrams of segmental hemangiomas (Figure 1A).¹ The S1 segment encompassed the lateral forehead, anterior temporal scalp, and lateral frontal scalp, and frequently also involved all or part of the upper eyelid, sparing the glabella and central forehead; the S4 segment encompassed the medial frontal scalp, medial forehead, nasal bridge, nasal tip, ala, and philtrum (Figure 1B). The maxillary (S2) segment included the ipsilateral upper cutaneous lip, respecting the lateral border of the philtrum, and the medial cheek without lateral extension to the preauricular region; the mandibular (S3) segment included the preauricular region, mandible, and chin, and often extended to the midline if unilateral, or frequently was bilateral in the “beard” distribution (Figure 1C).

Table. Infantile Hemangioma Subtypes Based on Location and Anatomic Distribution.

Distribution	Hemangiomas, No. (%)
Total	549
S1 (Frontotemporal)	41 (7.5)
S2 (Maxillary)	56 (10.2)
S3 (Mandibular)	63 (11.5)
S4 (Frontonasal)	21 (3.8)
Indeterminate S1	35 (6.4)
Indeterminate S2	87 (15.8)
Indeterminate S3	53 (9.7)
Indeterminate S4	74 (13.5)
Indeterminate periorbital	48 (8.7)
Scalp	44 (8.0)
Ear	27 (4.9)

Figure 1. Facial Segmental Infantile Hemangiomas.

A, Previous facial segmental hemangioma map reproduced with permission from Haggstrom et al.¹ B, The overlay mapping of frontotemporal (S1) and frontonasal (S4) segments confirmed previously reported anatomic mapping. C, The maxillary (S2) segment spares the preauricular region in contrast with previous segmental infantile hemangiomas maps, whereas the mandibular (S3) segment includes the preauricular skin.

Mapping of segmental scalp hemangiomas revealed a pattern adjacent to the posterior auricular sulcus, extending superiorly and posteriorly to varying degrees. Most scalp hemangiomas in this C-shaped pattern did not cross the midline (Figure 2). Based on the more precisely defined borders of S2 and S3 segments and the newly defined scalp pattern, a revised map of facial hemangiomas was created (Figure 3).

Figure 2. Scalp Segmental Infantile Hemangiomas.

Patterns of segmental scalp hemangiomas depicted via overlay mapping. The lesions occurred in C shapes of varied sizes extending from the back of the ear and sparing the midline.

Figure 3. Revised Map of Facial Segmental Infantile Hemangiomas.

This new map reflects the preauricular region segregating with the mandibular (S3) segment and depicts the novel C-shaped pattern of scalp segmental infantile hemangiomas.

The 297 indeterminate hemangiomas were also mapped. They were observed to have borders that respected those of their larger segmental counterparts. Although indeterminate hemangiomas were smaller in size, their distribution nearly always resided within the already established 4 facial segments. An example of indeterminate S3 (partial S3 hemangiomas) is depicted in Figure 4A. Indeterminate hemangiomas in the periorbital location that did not seem to collocate with 1 of the 4 defined segments were mapped separately (Figure 4B). These hemangiomas were restricted to the region confined by the borders of the orbital rim and encompassed either the lower lid only, the upper lid only, or both the upper and lower lids.

Figure 4. Indeterminate Infantile Hemangiomas.

A, Indeterminate hemangioma involving the vermilion lip and chin within boundaries of the mandibular (S3) segment (as shown in Figure 3). B, Patterns of indeterminate preorbital hemangiomas depicted via overlay mapping reveal several distinct patterns.

Discussion

The results of this cohort study demonstrate that a revised map of the facial segmental IHs published in 2006 by Haggstrom et al¹ is needed to more accurately define borders of the maxillary (S2) and mandibular (S3) segments. In contrast with the previously published map, we found that the preauricular skin corresponds to the mandibular (S3) segment rather than the maxillary (S2) segment. To our knowledge, this study is also the first to delineate a C-shaped pattern of scalp segmental IHs, which appears to be an additional area of risk for PHACE syndrome.⁴ Results of this study also provide evidence that indeterminate IHs are, in fact, partial segmental IHs. Refining the understanding of the patterning of segmental hemangiomas provides insights into pathogenesis as it relates to embryonic development and helps determine risk for underlying developmental anomalies such as PHACE syndrome.

The revised map of facial segmental IHs more closely mimics the boundaries of embryonic facial primordia. It has been observed in prior studies that the patterning of segmental hemangiomas does not correspond with facial dermatomes or branches of the trigeminal nerve, but instead follows the territories of embryological facial prominences.^1,5 During craniofacial development, head mesenchyme is derived from migration of neural crest cells that segregate into 5 facial prominences: the paired maxillary and mandibular prominences and the unpaired frontonasal prominence.⁶ Diagrams from embryology textbooks often depict the mandibular prominence as encompassing the preauricular skin in addition to the mandible and lower lip, while the maxillary prominence includes the infraorbital cheek and upper lip, and ends at the lateral border of the zygomatic bone.⁷ The present inclusion of the preauricular area in the mandibular (S3) segment of the revised facial hemangioma map better reflects the territories of these developmental primordia.

As was previously noted by Haggstrom et al,¹ there is a key difference between segmental hemangioma patterns and traditional depictions of embryonic facial placodes with respect to the upper half of the face. Whereas diagrams of facial primordia depict 1 single frontonasal segment encompassing the entire forehead, both the previous and current studies of IH patterning suggest the presence of 1 central frontonasal (S4) segment involving the medial forehead and 2 frontotemporal (S1) segments encompassing the upper eyelid, lateral forehead, temple, and lateral frontal scalp. A similar frontotemporal segmental pattern is also observed in facial port-wine stains.⁸ Haggstrom et al¹ hypothesized that the discrete frontotemporal (S1) domain may be explained by an embryonic boundary between the neural crest–derived and mesodermally derived bones of the lateral margins of the skull.

This study is unique in mapping the distribution of indeterminate hemangiomas. The term indeterminate IH was first proposed by Chiller et al⁹ to describe hemangiomas that were not clearly segmental nor localized. The current study now demonstrates that indeterminate IHs fall within known facial segments and respect their boundaries. From this we conclude that indeterminate IHs are partial segmental hemangiomas. Many birthmarks have similar patterns, with some encompassing a large territory and others, even if known to have the same genomic cause, being smaller. This has been attributed to the timing of birthmark development in utero, with those greater in extent occurring earlier in gestation. While there is no recognized somatic mutation yet proven to cause IHs, this same concept likely explains these patterns in IHs.

Recognition of indeterminate hemangiomas as partial segmental hemangiomas also has implications for possible extracutaneous associations. For example, small bilateral preauricular IHs, even in the absence of more extensive S3 involvement, may confer a risk of airway IHs or PHACE syndrome. Other partial segmental IHs in the forehead region might also confer some risk (the degree yet undefined) of PHACE syndrome. Current screening recommendations for PHACE syndrome include a large segmental hemangioma of more than 5 cm on the face or scalp.¹⁰ Future studies may want to reexamine these criteria with the knowledge that indeterminate IHs can follow the same embryonic patterns as segmental IHs and may thus confer some (arguably smaller) risk for developmental anomalies. While this might not alter current screening recommendations, this awareness could allow for further evaluations should specific signs or symptoms arise in individual patients. As an example, a recent report documented that small indeterminate IHs or focal mixed perioral hemangiomas on the upper lip were associated with enamel hypoplasia, as seen in patients with PHACE syndrome.¹¹ Thus, indeterminate hemangiomas in certain high-risk locations can be associated with underlying anomalies and, depending on clinical context, may warrant evaluation for associated developmental anomalies such as those seen in PHACE syndrome. Additionally, IHs in high-risk locations that may appear at first glance to be focal could actually be indeterminate or partial segmental and carry an increased risk for complications.

Although not addressed specifically in the current study, periorbital IHs deserve special comment. We identified 48 indeterminate periorbital IHs that demonstrated partial or complete circumferential involvement of the periorbital area (eg, upper lid, lower lid, medial canthus, or lateral canthus) without extension onto any of the 4 facial segments. Knowledge of the embryology of eye and orbital development makes consideration of a unique orbital segment worth considering. The orbital structures begin to develop as a pair of optic vesicles on each side of the forebrain at the end of the fourth week of pregnancy.⁶ These outgrowths of the brain make contact with the surface ectoderm, inducing changes that lead to further development of the eye. Several embryologic layers, including the neural tube, neural crest, surface ectoderm, and mesoderm, contribute to the development of the eye. Knowledge of this embryology together with case reports documenting PHACE syndrome in individuals with orbital IHs without significant cutaneous involvement of other segments suggests that IHs encompassing both upper and lower areas of the eye (either involving the skin or in a retro-orbital distribution) might be considered as part of a distinct orbital segment (perhaps referred to as S5).^12,13 The distinction between isolated orbital IHs vs collocation with segmental S1 or S2 hemangiomas may indicate timing of the developmental aberration, with contiguous involvement of the orbital area and adjacent segments suggesting an earlier error in the developmental pathway.¹⁴

There has been an increased recognition that extrafacial segmental hemangiomas on the posterior scalp and upper torso can also be associated with PHACE syndrome.¹⁵ As a result, the diagnostic criteria for PHACE syndrome were modified in 2016 to indicate that infants with IHs of the face or scalp should be screened for PHACE syndrome.¹⁰ However, until now, there has been little information describing the patterning of segmental scalp IHs in the literature. The present study helps to fill this gap, but additional research is still needed to better understand the magnitude of risk of segmental scalp IHs and implications regarding pathogenesis.

Haggstrom et al¹⁶ mapped 463 localized hemangiomas on the scalp. They found that localized scalp IHs were more likely to cluster on the midline or central portions of the scalp rather than the lateral scalp. In contrast, the pattern of segmental scalp IHs found in the current study is quite different. A lateral C-shaped pattern extends from the postauricular area to involve the temporal, parietal, and occipital scalp to varying degrees and does not cross the midline (Figure 2). This pattern occurs on the posterolateral scalp without contiguous involvement of any of the facial segments, whereas the frontal scalp often segregates with frontotemporal (S1) or frontonasal (S4) segments. The C-shaped distribution seems to be associated with the territory supplied by branches of the external carotid artery. One possible explanation for this may be that neural crest cells exhibiting a specific regional identity act through signaling pathways to cause segmental restriction of the primitive vasculature to embryonic primordia.¹ A similar pattern was proposed for acral segmental IHs, which also appear to follow the distribution of embryonic vasculature.¹⁷ Reimer et al¹⁸ postulated that acral hemangioma proliferation may be stimulated by hypoxia within a vascular distribution in the affected segments. We hypothesize that the “developmental hit” resulting in the C-shaped pattern of scalp IHs may occur in the segmentally restricted neural crest cells that guide formation of the external carotid branches in this region of the scalp.

A similar hypothesis was proposed regarding facial port-wine stains and risk of Sturge-Weber syndrome.^19,20 Waelchli et al²⁰ observed that facial port-wine stains follow the embryological vasculature, as opposed to the trigeminal nerve distribution, and using a vascular classification system improved prediction for Sturge-Weber syndrome. Port-wine stains involving the forehead, an area corresponding to the embryologic frontonasal prominence, were at highest risk for associated Sturge-Weber syndrome. A similar model based on embryonic vasculature to better understand PHACE syndrome risk warrants further exploration.

Limitations

This study is limited by potential errors when mapping segmental hemangioma boundaries onto a heat map. Because maps were created using clinical photographs, the exact extent of any deeper hemangioma components may be underestimated owing to poor visualization via photographs alone. Additionally, exact anatomic landmarks for each patient may not be accurately recapitulated on standardized 3-dimensional facial images. Finally, because lesions were mapped by different individuals at each academic institution, there may have been interoperator variability in data collection.

Conclusions

This cohort study refines the previous map of facial segment IHs and describes a new segmental scalp IH pattern. It highlights that indeterminate IHs are actually partial segmental IHs because they respect anatomic boundaries of facial segments. Future studies are needed to clarify the prevalence of PHACE syndrome or other developmental anomalies in scalp and indeterminate IHs. Additional investigation is also warranted to better characterize clinical predictors of high risk in indeterminate hemangiomas in terms of size, location, and morphology. We believe that hemangioma development is likely related to embryologic neural crest influence on local vasculogenesis. The exact cellular and molecular mechanisms implicated in IH pathogenesis remain unclear, but continued investigation of hemangioma patterning can provide insights into mechanisms underlying craniofacial development.