Risk for PHACE Syndrome in Infants With Large Facial Hemangiomas
OBJECTIVES: This study was conducted to determine the prevalence of posterior fossae of the brain, arterial anomalies, cardiac anomalies, and eye anomalies (PHACE) in infants with large facial hemangiomas. The extracutaneous manifestations of PHACE may be associated with significant morbidity, and the prevalence of PHACE in patients with facial hemangiomas has not previously been reported.
METHODS: A multicenter prospective study was conducted with 108 infants who had large facial hemangiomas and were systematically evaluated for manifestations of PHACE. The prevalence of PHACE and its extracutaneous manifestations in this cohort was calculated. The relationship between hemangioma distribution and the manifestations of PHACE was analyzed.
RESULTS: Thirty-three (31%) of 108 had PHACE. Thirty of the 33 patients with PHACE had >1 extracutaneous finding. The risk for PHACE syndrome was higher in infants with larger hemangiomas and in those with hemangiomas that encompassed >1 facial segment. The most common extracutaneous anomalies observed in infants with PHACE were of the arteries of the cerebrovasculature (91%) and cardiac anomalies (67%). Upper face (frontotemporal and frontonasal) hemangiomas were commonly observed in infants with PHACE; isolated maxillary hemangiomas were rarely associated with PHACE.
CONCLUSIONS: In infants with large facial hemangiomas, one-third have extracutaneous manifestations consistent with the diagnosis of PHACE syndrome, most commonly cerebrovascular and cardiovascular anomalies. The high prevalence of arterial anomalies in this cohort has implications for clinical management and future research regarding the pathophysiology of PHACE.
WHAT'S KNOWN ON THIS SUBJECT:
Previously, estimates of PHACE syndrome in patients with hemangiomas were unknown. PHACE is a neurocutaneous syndrome that may have potential life-threatening and function-threatening consequences.
WHAT THIS STUDY ADDS:
In this, the first prospective study of patients with facial hemangiomas the authors provide a more-accurate estimate of the incidence of PHACE in children with large facial hemangiomas and expand our understanding of the PHACE Spectrum.
PHACE is the acronym that refers to the association of large facial hemangiomas and anomalies of the posterior fossae of the brain, arterial anomalies, cardiac anomalies (most commonly coarctation of the aorta), and eye anomalies. The association of ventral sternal and supraumbilical defects has led some authors to modify this acronym (PHACES; OMIM 606519). Although the term PHACE was coined in 1996, scattered cases had been reported in the medical literature before this, and subsequently numerous single case reports and series have been reported.1,2 Overall, >200 cases of PHACE have been reported, and more recent evidence suggests that its incidence is as least as high as the widely known neurocutaneous syndrome Sturge-Weber.3 Previously, a broadly inclusive definition of facial hemangioma plus ≥1 extracutaneous features had been proposed for PHACE1,3; however facial hemangiomas occur frequently, and the significance of their association with other common extracutaneous anomalies such as an arachnoid cyst, a cerebral vascular “variant,” or a patent foramen ovale is uncertain. Recently, diagnostic criteria were established by a multidisciplinary group of physicians to identify more accurately patients with PHACE.4
The prevalence of PHACE among asymptomatic infants with large hemangiomas of the head or neck is unknown. Guidelines for evaluating infants who are at risk have been suggested, but no prospective studies have established the prevalence of PHACE in the population at risk.5 This study prospectively determined the prevalence of PHACE in a cohort of infants with large facial hemangiomas on the basis of standardized investigation that included MRI and magnetic resonance angiography of the head and neck, echocardiography, and ophthalmologic examination. We report the frequency of specific extracutaneous manifestations and discuss the relationship between hemangioma distribution and specific extracutaneous anomalies.
Early identification of PHACE affects clinical care. For example, coarctation of the aorta or other critical arterial anomalies may require surgical intervention to prevent cardiovascular compromise or cerebrovascular complications, including stroke or seizure. Even in the absence of correctable anomalies, recognition of PHACE influences management and therapeutic decisions, including pharmacologic treatment for the hemangioma.
Each site obtained approval by its institutional review board. Eight academic pediatric dermatology centers in the United States, Canada, and Spain consecutively recruited 108 infants (<1 year of age) with head and neck hemangiomas that measured ≥22 cm2 from 2005 to 2008. All patients who met inclusion criteria in the participating pediatric dermatology clinics were offered enrollment into the study. All investigators are members of the Hemangioma Investigator Group. Demographic and clinical information was recorded by study investigators. Hemangiomas were measured with a soft, flexible measuring tape. Clinical photographs were used to compare hemangioma distribution to previously described facial segments6 (Fig 1). Hemangiomas were classified as segmental when their distribution resembled developmental units and were broader than localized hemangiomas, which seem to arise from a single focal point.7 Hemangiomas were classified as bilateral when they involved both the left and the right frontotemporal (S1), maxillary (S2), or mandibular (S3) segments. Frontonasal (S4) hemangiomas were not classified as bilateral. Each child was evaluated with a complete physical examination, MRI and magnetic resonance angiography of the head and neck, echocardiography, and ophthalmologic examination. The diagnosis of PHACE was based on newly accepted criteria, summarized in Table 1. 4
Data were analyzed by Drs Haggstrom and Perkins at Indiana University. Prevalence of PHACE was estimated by using an exact binomial confidence interval. Infants with PHACE were compared with infants without PHACE by using Fisher's exact tests for dichotomous variables, Mantel-Haenszel exact tests for ordinal variables, 2-sample t tests (equal or unequal variance, as appropriate) for normally distributed numerical variables, and exact Wilcoxon rank-sum tests for nonnormal numerical variables. Comparisons of extracutaneous anomalies by segmental involvement within the infants with PHACE were explored with Fisher's exact tests.
Thirty-three (31%) of 108 (95% confidence interval: 22%–40%) had “definite” PHACE. One patient had “possible” PHACE and was not included in analyses when infants with and without PHACE were compared. The female-to-male ratio was 3.1:1 for infants without PHACE and 5.6:1 in those with PHACE (P = .32). The majority of patients were white (74%) with a minority that were Hispanic (14%), African-American (4%), Asian (4%), and unknown or other races/ethnicities (4%). Ethnicity and gender did not vary significantly between infants with and without PHACE (P = .16 and P = .32, respectively). The first visit to pediatric dermatology for patients with PHACE occurred at a mean of 13 weeks (SD: 9 weeks; range: 2–35 weeks), compared with those without PHACE. who were seen an average of 6 weeks later (mean: 19 weeks; SD: 13 weeks; range: 3–52 weeks; P = .005). Mean gestational age was 39 weeks (SD: 1.8 weeks; range: 30–42 weeks), and mean birth weight was 3345 g (SD: 652 g; range: 1588–4706 g); there were no statistical differences between infants with and without PHACE (P = .67 and P = .33, respectively).
No significant differences were detected in the average maternal age, number of miscarriages, rates of assisted conception, or multiple gestations between infants with and without PHACE (P = .24 or higher in all cases). No significant differences between infants with and without PHACE were found in patients with history of maternal preeclampsia (n = 8) or multiple gestation (n = 7; P = .69 and P = .67, respectively).
The mean size of hemangioma was 86 cm2 (SD: 65 cm2; range: 22–360 cm2) in the cohort. Infants with PHACE had significantly larger hemangiomas than those without PHACE (mean: 120 cm2 [SD: 68 cm2] vs 72 cm2 [SD: 60 cm2]; P = .002). The risk for PHACE increased with increasing size (Fig 2). Fifteen of 33 patients with PHACE had bilateral hemangiomas. For those with unilateral hemangiomas (n = 18), the left face was more frequently involved, although this was not statistically significant (P = .09).
Analysis of hemangioma distribution revealed the frontotemporal segment (S1) to be frequently involved in patients with PHACE (Fig 3). Twenty-eight of the 33 patients had involvement of S1 with or without other segments; 5 showed isolated S1 involvement. Four of 33 had involvement of the mandibular segment (S3) only, and 24 of 33 had involvement of S3 with or without other segments. Only 1 patient had hemangioma confined to the maxillary (S2) segment. The number of facial segments involved ranged from 1 to all 4. The most common combination of segments was S1/S3 seen in 11 (30%) of 33 of those with PHACE (Fig 4). The probability of having PHACE was significantly higher in infants with >1 segment involved (P < .0001; Fig 5).
Thirty of the 33 patients withPHACE had >1 extracutaneous finding. Twenty (61%) patients had both cardiac and CNS manifestations (Table 2).
Cerebrovascular and Structural Brain Anomalies
CNS anomalies were most common and were seen in 31 (94%) of 33 patients. CNS anomalies included structural abnormalities (17 of 33) and cerebrovascular anomalies (30 of 33). Structural brain anomalies included unilateral/ipsilateral posterior fossae anomalies (cerebellar hypoplasia/dysgenesis) in 7 patients. Common cerebrovascular anomalies included anomalous carotid arteries in 22 patients, anomalous trigeminal arteries in 5 patients, and anomalous vertebral arteries in 4 patients. Three patients had cerebrovascular anomalies as their sole extracutaneous manifestations of PHACE.
Cardiac anomalies were seen in 22 (67%) of 33 patients. Ten had coarctation of the aorta, 1 had an aortic aneurysm, and 1 had a significantly narrowed transverse arch and tortuous distal segment. Other, less specific cardiac anomalies included ventricular septal defect, atrial septal defect, tricuspid atresia, and patent foramen ovale. No patients had cardiac anomalies as their only manifestation of PHACE.
Ocular and Ventral Anomalies
Eight infants had ocular anomalies, including persistent retinal vessels, morning glory deformity, Mittendorf deformity, suspected pupillary membrane, microphthalmos, and iris hypoplasia. Five infants had ventral anomalies, including sternal pits and supraumbilical raphe (Fig 6).
Relationship to Facial Segments
As previously noted, the majority (28 [85%] of 33) of patients with PHACE had hemangiomas involving the S1 segment, with or without involvement of other segments. This distribution was associated with significantly higher risk for structural CNS anomalies (P = .02) and cerebrovascular anomalies (P = .05). All 28 patients with PHACE and hemangiomas involving the S1 segment had CNS involvement with 27 (96%) having cerebrovascular anomalies and 17 (60%) having structural anomalies. Of those with S1 involvement, cardiac anomalies were detected in 19 (68%), ocular anomalies in 7 (25%), and ventral anomalies in 3 (11%; Table 3).
Of 24 patients with S3 involvement, with or without other segment involvement, 22 (92%) had CNS anomalies, 18 (75%) had cardiac anomalies, 5 (22%) had ocular anomalies, and 5 (21%) had ventral defects. Of 10 patients with S4 involvement, all had CNS findings, including 80% with structural anomalies and 100% with cerebrovascular anomalies; however, these rates were not significantly different compared with those without S4 involvement (39% with structural anomalies [P = .06] and 87% with cerebrovascular anomalies [P = .54]).
Only 1 patient who met criteria for PHACE had an S2 hemangioma; however, this patient also had hemangioma in the periocular region, or what is referred to as the “watershed” area, which obtains vascular supply variably from the upper and lower face. When S2 only was involved, patients were significantly likely not to have PHACE (P = .02).
This prospective study provides new information about the prevalence of PHACE in an at-risk population with large facial infantile hemangiomas. In this fully evaluated cohort, PHACE was diagnosed in 31% of infants with facial hemangiomas that measured ≥22 cm2. Despite using stricter diagnostic criteria, this reported prevalence is higher than previous estimates in previous studies (20%).3 This higher prevalence is likely attributable to comprehensive screening in our cohort. This study also highlights the high prevalence of aortic coarctation in PHACE (9.3%), a prevalence that approaches that observed in Turner syndrome.8
The female-to-male ratio for infants with PHACE was 5.6:1, lower than other retrospective series, in which female predominance of 9:1 has been reported.3 Previous reports of infantile hemangiomas (not limited to those with PHACE) have indicated that infants tend to be preterm; however, our infants were more likely to be term. Segmental hemangiomas are hypothesized to arise as an error in neural crest development, as early as 4 to 8 weeks of gestational age, and thus their occurrence may be less likely to be affected by preterm delivery.6
PHACE was more frequently observed in infants with larger hemangiomas and those with >1 facial segment involved. The overall incidence of PHACE was 31%; however, subgroup analysis on the basis of hemangioma size revealed that size was a predictor of risk. The incidence of PHACE ranged from 15% in infants with hemangiomas that measured 22 to 50 cm2 to 50% in those with hemangiomas that measured >100 cm2. The study did not include infants with hemangiomas that measured <22 cm2, so the risk for PHACE in these infants remains unknown. In addition to larger size, the number of facial segments affected is an important risk factor. Infants with only 1 segment involved, particularly S2, are significantly less likely to have PHACE than those with >1 segment involved, an observation that is helpful for clinical assessment of risk.
Before this study, it was estimated that 70% of patients with PHACE had only 1 extracutaneous manifestation,1,3 but in our study, the vast majority (91%) of infants had >1 extracutaneous finding. As the first study to perform prospective and standardized screening for infants at risk, this study provides an opportunity to uncover common or essential features of this syndrome that may provide clues to pathogenesis. The most common extracutaneous anomalies observed were of the arteries of the cerebrovasculature. Of note, commonly described anatomic “variants” of the Circle of Willis, such as missing A1 segment, were excluded. The prevalence of cerebrovascular anomalies among patients with PHACE was 91% compared with 35% to 51% reported in the literature.3 Although there are isolated reports of venous and cerebral sinus abnormalities, predominant involvement of the arterial system is thought to distinguish PHACE from other neurocutaneous syndromes. The presence of the CNS vascular anomalies suggests that factors that lead to their development might represent the primary defect in this disorder.
There were no reported anomalies of the venous system in our cohort. Furthermore, all but 1 infant with PHACE had arterial anomalies, indicating that the pathogenesis may be intimately linked with arterial development. Arterial anomalies generally manifest as (1) dysplasia, (2) narrowing, (3) aberrant course or origin, and (4) persistence of embryonic anastomoses (arterial segments that are present but normally involute during arterial development, such as persistent trigeminal artery). The incidence of structural brain abnormalities is comparable to previous reports (52% vs 42%–45%).9
Detection of cerebrovascular anomalies, particularly large-vessel anomalies, is important given their potential association with neurologic sequelae including stroke and seizures.10,–,12 The carotid arteries were frequently anomalous (22 [67%] of 33), highlighting the importance of extending imaging to include the neck to detect areas of critical stenosis. Anomalous cerebrovasculature in patients with PHACE has been associated with progressive vasculopathy, which is noted on serial imaging in some patients10,12; however, this study was not designed to follow patients longitudinally and therefore cannot address the risk for ischemic cerebral events secondary to progressive vascular occlusion or stenosis. Depending on the extent of vascular stenosis and potential compromise, consultation with neurology and neurosurgery is indicated for consideration of aspirin therapy or surgical interventions.
The results of this study further delineate clinical clues that can help in identifying patients who are at particularly high risk for neurologic complications. Previous studies suggested that infants with S1 hemangiomas are at higher risk for CNS anomalies3,9 and those with S3 hemangiomas are at higher risk for cardiac anomalies.3 The results herein found that all patients with CNS involvement had either S1 or S4 involvement. S1 involvement itself conferred a particularly high risk in that those who had S1 hemangioma were more likely to have CNS involvement compared with those who had no S1 involvement (P = .02). The specific cerebrovascular anomalies, especially the presence of fetal vessels, supports that the error in development occurs early in gestation (6–8 weeks), and the coexistence of the cutaneous and CNS findings may support the theory of a developmental “field defect.”
The common origin of the great arteries and the derivatives of the first brachial arch suggests a potential relationship between mandibular hemangiomas and cardiac and aortic anomalies. In our study, 75% of those with PHACE and mandibular (S3) involvement had cardiac anomalies compared with 44% of those without S3 involvement, but this was not statistically significant. Of note, 4 patients with cardiac anomalies had no involvement of S3.
It is interesting that 16 (15%) of 108 patients had hemangiomas restricted to the maxillary (S2) segment, and only 1 of these had PHACE; however, this patient had an S2 hemangioma with extensive periocular hemangioma (both upper and lower lids), which could be considered to represent partial S1 involvement. Thus, no patients with PHACE had a purely maxillary hemangioma, without extension into the periorbital region, suggesting that the S2 segment alone poses a lower risk for PHACE.
Because there were only 33 infants with PHACE, analyses of specific anomalies relative to specific segments involved have limited power; however, because of the unique nature of this sample, we believed that it was relevant to report findings and their statistical significance by segment group.
The finding that 31% of infants with large facial hemangiomas had PHACE supports comprehensive screening of this population. In addition to treating extracutaneous anomalies when necessary, the diagnosis of PHACE may influence therapeutic decisions for the cutaneous hemangioma itself. Systemic steroids, β-blockers, interferon, and vincristine all are therapeutic options for complicated and large cutaneous hemangiomas. The presence of severe arterial anomalies of the CNS or cardiovasculature could potentially preclude or restrict the use of β-blockers in this population, because potential hypotension and reduced perfusion through already-stenosed or tortuous vessels could have catastrophic consequences in a subset of patients. Although PHACE is not an absolute contraindication for the use of β-blockers, we advocate proceeding cautiously in this high-risk group of patients, with a thorough investigation for significant internal vascular anomalies and consultation with a neurologist and a cardiologist as appropriate.
Extracutaneous manifestations, of which the vast majority are arterial anomalies, are seen in one-third of patients with large facial hemangiomas. The clinical significance and long-term consequences of specific extracutaneous anomalies have not been well-studied. Longitudinal studies of children with PHACE are warranted for better understanding of the implications and importance of the detected extracutaneous anomalies on long-term physiologic and neurodevelopmental outcomes.
- Accepted May 7, 2010.
- Address correspondence to Anita N. Haggstrom, MD, 550 N University Blvd, UH 3240, Indianapolis, IN 46220. E-mail:
FINANCIAL DISCLOSURE: Dr Frieden has served as a consultant for Pierre-Fabre; the other authors have no financial relationships relevant to this article to disclose.
- PHACE =
- posterior fossae of the brain, hemangiomas, arterial anomalies, cardiac anomalies, and eye anomalies
- Metry DW,
- Heyer G,
- Hess C,
- et al
- Haggstrom AN,
- Lammer EJ,
- Schneider RA,
- Marcucio R,
- Frieden IJ
- Ho VB,
- Bakalov VK,
- Cooley M,
- et al
- Oza VS,
- Wang E,
- Berenstein A,
- et al
- Drolet BA,
- Dohil M,
- Golomb MR,
- et al
- Copyright © 2010 by the American Academy of Pediatrics