Prospective Study of Infantile Hemangiomas: Clinical Characteristics Predicting Complications and Treatment
OBJECTIVES. Infantile hemangiomas are the most common tumor of infancy. Risk factors for complications and need for treatment have not been studied previously in a large prospective study. This study aims to identify clinical characteristics associated with complications and the need for therapeutic intervention.
PATIENTS AND METHODS. We conducted a prospective cohort study at 7 US pediatric dermatology clinics with a consecutive sample of 1058 children, aged ≤12 years, with infantile hemangiomas enrolled between September 2002 and October 2003. A standardized questionnaire was used to collect data on each patient and each hemangioma, including clinical characteristics, complications, and treatment.
RESULTS. Twenty-four percent of patients experienced complications related to their hemangioma(s), and 38% of our patients received some form of treatment during the study period. Hemangiomas that had complications and required treatment were larger and more likely to be located on the face. Segmental hemangiomas were 11 times more likely to experience complications and 8 times more likely to receive treatment than localized hemangiomas, even when controlled for size.
CONCLUSIONS. Large size, facial location, and/or segmental morphology are the most important predictors of poor short-term outcomes as measured by complication and treatment rates.
Infantile hemangiomas (IHs) are classically considered “birthmarks,” but unlike most birthmarks, they are uniquely dynamic. At birth they are either absent or barely evident, but they proliferate in the first few weeks to months of life, followed by an involution phase over several months to years. Most IHs are uncomplicated, but a significant minority involve complications, including ulceration, threat to vision, airway obstruction, and congestive heart failure. Hemangiomas can also leave residual scarring and/or permanent distortion of facial anatomic landmarks, which can be truly life altering.1–3 Given the wide spectrum of disease and the natural tendency for involution, the greatest challenge in caring for infants with hemangiomas is predicting which infants need treatment or are at highest risk for complications. A primary aim of this study was to identify specific clinical features that are most predictive of complications and/or need for treatment.
The Hemangioma Investigator Group prospectively enrolled patients with IH over a 13-month period, between September 2002 and October 2003. Patients/guardians provided signed informed consent, and institutional review boards at each participating site approved the study protocol. Eligible patients included those who were ≤12 years at the time of enrollment and who had ≥1 IH in any stage of evolution. A consecutive sample was enrolled by each investigator. A total of 1096 patients, of whom 1058 were US patients, were enrolled during this 13-month period, and clinical follow-up continued until June 2004. For the purposes of this report, data were analyzed from US sites only. Using standardized computer-scannable forms, investigators collected information about patient gender, ethnicity, prematurity, birth weight, maternal/paternal age, prenatal testing procedures, placental abnormalities, maternal illnesses, smoking, recreational drug use and medication history, and family history. The location, size, and morphologic subtype for ≤4 hemangiomas per patient were noted. Morphologic subtypes were localized, segmental, indeterminate, and multifocal (Fig 1). 4 “Segmental hemangiomas” were those hemangiomas or clusters of hemangiomas with a configuration corresponding to a recognizable and/or significant portion of a developmental segment. “Localized hemangiomas” were defined as those hemangiomas that seem to grow from a single focal point or were localized to an area without any apparent linear or developmental configuration. “Indeterminate hemangiomas” were those that were not readily classified as either localized or segmental, and “multifocal hemangiomas” were defined as ≥10 cutaneous hemangiomas. Two investigator training sessions and a study manual helped insure uniformity of subtype classification, and interrater reliability was then tested and found to be excellent using a κ statistic (data not shown).
The size of each hemangioma was measured using a “hemispherical measurement.” Using a flexible tape measure, the surface of the hemangioma was measured in the 2 perpendicular directions using the maximum diameter in each to obtain a final measurement (cubic centimeters). Treatment and complications incurred before and during the study period were noted (Table 1). The study period extended for a minimum of 8 months and up to 23 months after enrollment to maximally capture morbidities during hemangioma proliferation. Follow-up visits were scheduled according to clinical need, and follow-up forms recording hemangioma characteristics, treatment, and complications were completed by study investigators. Treatment decisions were determined by physicians caring for the patient and not altered for purposes of this study.
Data were processed by the National Outcomes Center, where forms were scanned into a computer database. Information was manually verified. Data analysis was performed in conjunction with the National Outcomes Center and the University of California San Francisco Biostatistics Department. χ2 tests and t tests were used to compare categorical variables and continuous variables, respectively. Spearman rank correlation, Mann-Whitney rank sum, and Kruskal-Wallis statistical methods were used when appropriate. Random-effects logistic regression models were used to address potential confounders.
A total of 1915 hemangiomas were observed in 1058 patients (mean: 1.8; median: 1.0). The majority (68.8%) of patients had a solitary hemangioma, and 97% of patients had ≤6 hemangiomas. Detailed clinical information was recorded for each hemangioma, except when patients had >4 hemangiomas, in which case information was only recorded for the 4 most clinically significant hemangiomas.
The detailed clinical characteristics of a total of 1530 hemangiomas were documented using hemangioma-specific intake forms. Approximately 41% (630; 41.2%) were located on the face. Involvement of other sites was as follows: head and neck, excluding the face in 322 (21.0%), trunk in 357 (23.3%), extremities in 281 (18.4%), and perineum in 94 (6.1%). Uncommonly, some hemangiomas extended into >1 site (for example: face, neck, and chest). The majority of hemangiomas were classified as localized (1022; 66.8%), whereas 200 (13.1%) were segmental, 253 (16.5%) were indeterminate, and 55 (3.6%) were multifocal. The mean size of an individual hemangioma was 18.9 cm2 (SD ±66.7) with a median of 3.6 cm2 (95% confidence interval [CI]: 3.3–4.0 cm2). Hemangiomas ranged from pinpoint lesions (<0.004 cm2) to 1320 cm2. Facial hemangiomas averaged 19.9 cm2 (95% CI: 14.7–25.0 cm2). Twenty-two percent (136 of 630) of facial hemangiomas were segmental.
Complications that occurred before enrollment (based on parental recall and/or medical charts) and during the study period were recorded (Table 1). Before enrollment, 299 patients (28.3%) had complications; 255 patients (24.1%) experienced complications during the study period. During the study period, ulceration was the most common complication noted in 168 patients (16.0%); threat to vision (59; 5.6%), airway obstruction (15; 1.4%), auditory canal obstruction (6; 0.6%), and cardiac compromise (4; 0.4%) were seen less commonly. Hepatic hemangiomas were detected in 10 patients (0.1%), but routine abdominal imaging was not performed as a part of this study protocol.
Treatment before and during enrollment was recorded. Some form of treatment was administered in 269 patients (25%) before enrollment, and 402 (38%) received therapy during the study period. During the study period, patients received systemic steroids (130; 12.3%), intralesional steroids (43; 4.1%), topical steroids (103; 9.8%), wound care for ulceration (145; 13.7%), oral antibiotics (21; 2.0%), pulsed dye laser (84; 8.0%), and excisional surgery (60; 5.7%). Rarely, interferon (1 patient) and vincristine (3 patients) were used. Some patients received >1 mode of therapy. The most common indications for treatment were disfigurement (33.1%), ulceration (19.8%), and rapid growth (16.5%).
Factors Predicting Complications and/or Treatment
Size, location, and subtype were major factors that predicted complications and/or need for treatment. The mean size of hemangiomas experiencing any complication (including ulceration and bleeding, visual compromise, auditory canal obstruction, cardiac compromise, and airway obstruction) was 37.3 cm2 (95% CI: 30.36–44.2 cm2) compared with 19.1 cm2 (95% CI: 14.4–23.9 cm2) for hemangiomas without complications. Complicated hemangiomas were, on average, 18.1 cm2 larger than uncomplicated hemangiomas (P < .0001; 95% CI: 11.5–24.7 cm2). Hemangiomas complicated by surface ulceration or bleeding had a mean size of 40.4 cm2; the observed difference between ulcerated and nonulcerated hemangiomas was 20.1 cm2 (P < .0001; 95% CI: 12.3–28.1 cm2). For every 10-cm2 increase in hemangioma size, there was a 5% increase in the likelihood of experiencing a complication (odds ratio [OR]: 1.051; P < .05). Hemangiomas that received treatment of any type (including systemic therapy, wound care, laser, ad surgical excision) had a mean size of 30.4 cm2, which was 11.1 cm2 larger than those that did not receive treatment (P < .0001; 95% CI: 5.7–16.5 cm2). Hemangiomas requiring oral corticosteroid treatment were 21.2 cm2 larger than those that did not receive treatment (P < .0001; 95% CI: 12.0–30.4 cm2). For each 10-cm2 increase in hemangioma size, there was a 3.9% increase in the likelihood of treatment (OR: 1.039; P < .05).
Morphologic subtype was the best single predictor of complications and need for treatment. The influence of hemangioma size and subtype on the need for treatment and development of complications was studied, using a random-effects logistic model. Controlling for hemangioma size, hemangioma subtype was a strong predictor for development of complications and need for treatment. Segmental hemangiomas were 8 times more likely to receive treatment compared with localized hemangiomas after controlling for hemangioma size (OR: 8.4; 95% CI: 5.8–12.2). Similarly, segmental hemangiomas were 11 times more likely to develop complications (including ulceration, bleeding, visual compromise, auditory compromise, cardiac compromise, or airway obstruction) than localized hemangiomas after controlling for size (OR: 11.5; 95% CI: 7.8–17.0). When comparing outcomes between subtypes, segmental hemangiomas consistently had higher rates of complications and treatment compared with indeterminate and localized hemangiomas (Table 2). Indeterminate hemangiomas had rates of complication and treatment higher than localized but lower than segmental hemangiomas. The effect of subtype was even more important on the face. Complication and treatment rates varied by location of hemangioma (Table 3). Facial hemangiomas experienced complications 1.7 times as often as nonfacial hemangiomas (OR: 1.7; 95% CI: 1.3-2.3; P = .0002). Hemangiomas located on the face were 3.3 times more likely to receive some form of treatment, including systemic therapy, wound care, and pulsed dye laser, when compared with nonfacial hemangiomas (OR: 3.3; 95% CI: 2.6-4.3; P < .0001). Demographic and perinatal factors, including gender, ethnicity, prematurity, birth weight, family history, and maternal chronic illness, did not predict increases in complications or the need for treatment.5
Physicians evaluating infants with hemangiomas face a challenge. Virtually all IHs involute spontaneously, and most do so without sequelae, but a sizeable minority have complications or need treatment. In this study, 24% had complications during the study period, and 38% received some form of therapeutic intervention. A major goal of this, the largest prospective study of IH, was to determine which clinical features are most predictive of morbidity and/or need treatment. In attempt to capture the broadest spectrum of the disease, enrollment included all of the children with hemangiomas seen at enrolling sites, even if they were found incidentally on children being seen for other dermatologic conditions. Because enrollment was exclusively in pediatric dermatology practices, with most patients being referred specifically for IH, we assume that the complication and treatment rates are higher than would be seen by primary care physicians, but they are lower than those reported by surgical practices, where an estimated 40% to 60% experienced complications or are anticipated to require corrective surgery.6,7 We would not expect referral bias to impact the actual clinical features in our cohort that predicted poor outcome. The relatively short-term follow-up duration (minimum 8 to maximum 21 months) of the study precluded a true estimate of rates of scarring and/or anatomic distortion leading to permanent disfigurement, but this concern is highlighted by the fact that perceived risk of disfigurement was the leading indication for treatment (33%). A prospective study lasting 4 to 5 years or longer would be best to fully capture this risk. In this study, patients who were determined to have uncomplicated hemangiomas by the study investigators were not routinely seen in follow-up, which could potentially result in the underestimation of complication rates if unexpected complications were not reported to study investigators.
Morphologic Subtype as a Predictor of Complication and/or Treatment
Although both size and location are important predictors, the greatest single predictor of prognosis is morphologic subtype. Segmental hemangiomas were 11 times more likely to develop complications than localized hemangiomas, even after controlling for size, and they had a much greater need for treatment. This result affirms the findings of 2 previous retrospective studies.4,8 Although not the focus of this article, facial segmental hemangiomas can also be associated with extracutaneous malformations, especially the PHACE association9 and visceral hemangiomas.10 Their propensity for ulceration is clearly established in this study (33.5% vs 7.2% for localized), but the reasons for this are not clear. Segmental hemangiomas are believed to arise from an error in development occurring early in gestation11–13 and conceivably have more pervasive tissue dysregulation, making them more prone to tissue breakdown, but this is purely conjecture.
Unfortunately, the classification of hemangiomas by morphologic subtype may be difficult for physicians who do not see large numbers of infants with hemangiomas, and some may feel unsure about their ability to recognize what actually constitutes a segmental hemangioma (Fig 1).13 The concept of an anatomic territory defining a risk, however, is not new to physicians. Examples include dermatomal distribution maps (eg, for port-wine stains and herpes zoster) and maps defining the lines of Blaschko (mapping epidermal nevi, incontinentia pigmenti, and other mosaic disorders). Such patterns, although initially unfamiliar, become much easier to recognize once the viewer is cognizant of their existence and appearance, even in cases where they are only partially manifest, as is the case with “indeterminate” hemangiomas. These hemangiomas have been shown recently to occur within the boundaries of segmental hemangiomas, suggesting that they are really a forme fruste of segmental hemangiomas.13 Therefore, it is not surprising that their rates of morbidity are intermediate between localized and segmental subtypes. Together, segmental and indeterminate hemangiomas comprised less than one third (29.6%) of hemangiomas in the study further emphasizing the large number of lower-risk hemangiomas
Size as a Predictor of Complications and/or Treatment
In addition to morphologic subtype, hemangioma size and location are both highly associated with complications and need for treatment. The median size of hemangiomas in the study was 3.6 cm2, a size remarkably smaller than the mean of 18.9 cm2. This discrepancy reflects the contribution of very large hemangiomas, many of which were 50 to 100 cm2 or larger, which increased the mean size. It also emphasizes the large number of small, rather innocuous hemangiomas in the study. In contrast, the mean size of hemangiomas with complications was 37.3 cm2, and the mean of those receiving treatment was 30.4 cm2, numbers which contrast significantly with the mean size of those with neither complications (19.2 cm2) nor treatment (19.3 cm2).
Anatomic Location as a Predictor of Complications and/or Treatment
Complication rates and need for treatment varied according to hemangioma location (Table 3). Perineal hemangiomas had the highest rate of complications (52%) in our study, which reflects their propensity for ulceration. The face was the single most common anatomic area of involvement (41%), and 43% of facial hemangiomas received treatment of some kind, a rate significantly higher than hemangiomas in other locations. This finding is not surprising given the higher risk for facial hemangiomas impinging on vital structures, such as the eye or nose. Facial hemangiomas also carry a much higher risk of disfigurement than in other sites, because scars are readily visible. Furthermore, soft tissue distortion of the unique three-dimensional structures of the face, even after involution, can lead to permanent deformities.14–19 This fact is underscored by the finding that in fully one third of patients, risk of disfigurement was either the sole indication or one of the indications for treatment. At the same time, it is important to note that more than half of facial hemangiomas did not have complications or receive treatment. Just as important as recognizing which hemangiomas need treatment is recognizing which do not. Although many small, localized hemangiomas on the face, especially those on the lateral areas of the face, may be candidates for observation without treatment, there are still situations when even small and/or localized hemangiomas (especially those on the central face) may need intervention to prevent anticipated permanent soft-tissue distortion. Our study was not designed to capture long-term outcomes, and there certainly may have been patients in the cohort who did not receive treatment, yet ended up with some permanent sequelae. Further prospective studies are needed to refine our understanding of which smaller hemangiomas have the greatest potential for causing scars and disfigurement.
Indications for Treatment
Potentially life-threatening morbidities of hemangiomas, such as airway hemangiomas causing obstruction and liver hemangiomas causing high-output cardiac failure, are well-recognized and justifiably feared complications, but these were very uncommon, seen in 1.4% and 0.4% of our patients, respectively. Sight-threatening hemangiomas were noted in 5.6% of patients. The most frequent complication was ulceration, which occurred in 16% of infants during the study period. Ulceration can be very painful and virtually always results in scarring. Treatment of ulceration was the second-most common indication for treatment (19.8%).
Although the older medical literature has emphasized the benign nature of hemangiomas (except those causing specific medical morbidities), there has been a growing recognition that scarring and disfigurement are real and important morbidities that can be life altering even if not life threatening.20–22 This potential is greatest with facial hemangiomas, which, for unknown reasons, are an anatomic site that is disproportionately affected (41% in this study cohort). Information disseminated via the Internet has greatly heightened parental awareness of the potential risk for scarring and can result in undue anxiety even for those hemangiomas that are unlikely to leave scarring. Physicians caring for children with hemangiomas can expect questions from parents regarding prognosis and, therefore, must understand which hemangiomas have the greatest risks of complication and need for treatment. Because hemangiomas proliferate rapidly in the first few weeks to months of life, there may be a window of opportunity to intervene in high-risk hemangiomas in an attempt to prevent complications, including permanent scarring. This study should help guide physicians in determining which hemangiomas may be most appropriate for referral and/or treatment.
We gratefully acknowledge the Dermatology Foundation and American Skin Association for their funding support.
We thank Dr Nancy Esterly for her inspiration, support, and guidance in developing this study. We also thank our colleagues and staff in the Vascular Anomalies Centers at each institution for their hard work and caring for our patients and Alan Bostrom and Charles McCulloch at Biostatistics, University of California, San Francisco, and the National Outcomes Center for their database and statistical expertise.
- Accepted April 10, 2006.
- Address correspondence to Ilona J. Frieden, MD, Department of Dermatology, 3rd Floor, 1701 Divisadero St, San Francisco, CA 94143. E-mail:
All authors are members of the Hemangioma Investigator Group.
The authors have indicated they have no financial relationships relevant to this article to disclose.
- ↵Haggstrom AN, Drolet BA, Baselga E, et al. A prospective study of infantile hemangiomas, part I: demographic, prenatal and perinatal characteristics. J Pediatr. In press
- ↵Waner M, Suen JY. The natural history of hemangiomas. In: Waner M, Suen JY, eds. Hemangiomas and Vascular Malformations of the Head and Neck. New York, NY: Wiley-Liss;1999:13– 46
- Weon YC, Chung JI, Kim HJ, Byun HS. Agenesis of bilateral internal carotid arteries and posterior fossa abnormality in a patient with facial capillary hemangioma: presumed incomplete phenotypic expression of PHACE syndrome. AJNR Am J Neuroradiol.2005;26 :2635– 2639
- ↵Haggstrom AN, Lammer EJ, Schneider RA, Marcucio R, Frieden IJ. Patterns of infantile hemangiomas: New clues to hemangioma pathogenesis and embryonic facial development. Pediatrics.2006; 117: 698–703
- ↵Tanner JL, Dechert MP, Frieden IJ. Growing up with a facial hemangioma: parent and child coping and adaptation. Pediatrics.1998; 101: 446–452
- Copyright © 2006 by the American Academy of Pediatrics