Infantile Haemangioma

“This is almost certainly an infantile haemangioma — a benign vascular tumour, and the commonest tumour of infancy. My structure for any vascular anomaly is the same: is it a tumour or a malformation; if a tumour, is it infantile or congenital; and does this particular lesion need treatment or observation?

The history is characteristic — a pale precursor mark at birth, appearance in the first weeks, and rapid proliferation. On the photograph this is a superficial, focal lesion of the right cheek in the proliferative phase.

The single most useful thing I can do at this visit is risk-stratify. Most infantile haemangiomas need nothing but explanation and photographs. A minority — those that threaten vision, the airway or feeding, those that will ulcerate, and those that will leave permanent disfigurement — need propranolol, and they need it now, because the window of maximum proliferation closes at around five months. She is ten weeks old, so this is the decision point, not a review in three months.

To her mother: ‘The GP is right that most of these fade on their own, and this is not a cancer and not anything you did. What I need to decide today is whether hers is one of the few that needs treatment to stop it growing, because the growing happens in the next few weeks and after that the chance has gone.’”

“It will grow, then stop, then slowly shrink — and the numbers matter because they set the timing of every decision.

• Median age of appearance is about 2 weeks. Growth is front-loaded: the most rapid growth is between 1 and 3 months, and most growth is complete by 5 months — that is the Hemangioma Investigator Group data in Chang’s 2008 Pediatrics paper, and it is the basis of the AAP’s ‘refer by one month’ recommendation.
• Growth plateaus at 9 to 12 months. Involution then runs at roughly 10% a year — about half involuted by 5 years, 70% by 7, 90% by 9.
• The honest counterweight to ‘it will disappear’ is that around half or more leave a residual change — telangiectasia, fibrofatty residuum, anetodermic skin, or a scar where it ulcerated. Bauland’s series in PRS 2011 found residual lesions in the majority of untreated lesions, with superficial epidermal involvement the strongest predictor.

So for her mother: it grows fastest in the next few weeks, stops by about five months, and then fades over years — but roughly half of children are left with some mark, and the way to reduce that is to treat during the growth phase, not after it.”

“It arises by vasculogenesis — de novo vessel formation from a stem or progenitor cell — rather than by angiogenesis, which is sprouting from existing vessels. That is the sentence to say.

The precursor is the haemangioma-derived stem cell — a multipotent CD133-positive cell which, implanted in immunodeficient mice, forms GLUT-1-positive human microvessels that recapitulate the natural history of the lesion. That is Khan and Bischoff’s 2008 paper in the Journal of Clinical Investigation.

The endothelium shares a distinctive immunophenotype with placental microvasculature — GLUT-1, merosin, Lewis Y antigen, FcγRII — none of which are expressed in normal skin vessels or in vascular malformations. That is North’s 2001 Archives of Dermatology paper, and it is the basis of the placental hypothesis: either embolisation of placental cells or differentiation of a primitive precursor along a placental line. Neither is proven, but the shared phenotype explains the epidemiology — placental abnormalities, chorionic villus sampling and pre-eclampsia are all risk factors.

Hypoxia is the central driver of growth. Involution is by apoptosis, beginning before 12 months and peaking at around 24 months, with the stem cells differentiating into adipocytes — which is why the residuum is fibrofatty.

The honest caveat: unlike congenital haemangiomas, which carry GNAQ or GNA11 mutations, and unlike most vascular malformations, infantile haemangioma has no identified somatic driver mutation. That is why there is no targeted therapy and why treatment remains a beta-blocker acting downstream.”

“Usually not. The diagnosis is clinical in over 90% — the history of a herald mark, appearance in the first weeks and rapid proliferation is characteristic. I reserve investigation for three situations: diagnostic uncertainty, assessment of deep extent, and screening for associated anomalies.

• Doppler ultrasound is first line — a well-circumscribed, high-flow soft tissue mass with multidirectional flow, high vessel density and low resistance. Non-ionising, no sedation.
• MRI for deep or segmental lesions, airway or orbital involvement, and PHACE or LUMBAR screening. Proliferating lesions are T1 isointense, T2 hyperintense, enhance avidly and show flow voids; involuting lesions show increasing fatty signal.
• Biopsy is rarely needed but is definitive, and the marker is GLUT-1 — positive in infantile haemangioma at every stage, negative in congenital haemangioma, in every vascular malformation, in pyogenic granuloma, kaposiform haemangioendothelioma and tufted angioma.

In this child I would not image. I would photograph, risk-stratify and decide.”

“I use the four categories in the AAP 2019 clinical practice guideline — Krowchuk and colleagues in Pediatrics. A haemangioma is high risk if there is evidence of, or potential for:

• Life-threatening complications — airway obstruction, or hepatic involvement with high-output cardiac failure.
• Functional impairment or ulceration — vision, feeding, hearing.
• Structural anomalies — PHACE or LUMBAR.
• Permanent disfigurement.

The guideline says high-risk lesions should reach a specialist ideally by one month of age, because the treatment has to run through the proliferative phase to work.

In practice that means a list of sites I do not observe: periocular, nasal tip, lip and perioral, ear, the beard distribution, segmental facial, lumbosacral or perineal, and any child with five or more lesions. And it means one morphology — segmental — which is the single most useful risk marker for everything on that list.

This child’s lesion is focal, on the cheek, and not at any of those sites. So the honest question is the fourth category: will it disfigure? A growing superficial lesion on the face of a girl, with roughly a one-in-two chance of a permanent mark, is a reasonable indication for propranolol on disfigurement grounds — and I would offer it, explain why, and let her mother decide with the facts.”

“That is a large segmental facial haemangioma, and it changes three things: this child needs PHACE screening, she needs ophthalmology today because the upper eyelid is involved, and she needs propranolol — but not until I have imaged her.

PHACE occurs in around 2% of infantile haemangiomas overall, but in about 30% of infants with a large segmental facial lesion — Haggstrom’s 2010 prospective study found 31%. It is 90% female. The diagnosis needs a segmental facial haemangioma, usually over 5 cm, plus one or more extracutaneous anomaly, using the consensus criteria — Metry 2009, updated by Garzon in the Journal of Pediatrics in 2016.

The work-up is MRI and MRA of the head and neck for posterior fossa and cerebrovascular anomalies, echocardiography with aortic arch views, and ophthalmology, with endocrine assessment if there are midline or pituitary anomalies. The arterial imaging is not optional: cervical and cerebral arterial anomalies are present in about 90% of PHACE, and stroke — though rare — is the complication that matters. Narrowing or absence of two or more great cerebral arteries, particularly with coarctation, is the high-risk arteriopathy pattern.

I image before I treat, because the arteriopathy determines how I give the propranolol — not whether.”

“Six lesions means abdominal ultrasound, because the liver is the commonest extracutaneous site. The threshold is five or more: Horii’s prospective Hemangioma Investigator Group study in 2011 found hepatic haemangiomas in 16% of infants with five or more cutaneous lesions and in none with fewer — and, importantly, the large majority of those found did not need treatment.

If the liver is involved I want to know which pattern, because they behave differently:

• Focal — usually a RICH rather than a true infantile haemangioma, GLUT-1 negative, not associated with cutaneous lesions, usually asymptomatic.
• Multifocal — the pattern that goes with cutaneous disease; usually asymptomatic, but can cause high-output cardiac failure from shunting. Propranolol; embolisation if failing.
• Diffuse — near-total hepatic replacement, massive hepatomegaly, respiratory compromise, abdominal compartment syndrome, and consumptive hypothyroidism. The highest-mortality form.

So in a child with diffuse hepatic disease I check thyroid function at diagnosis and monitor it through treatment.”

“Ulceration is the commonest complication — 15 to 25% of referred lesions, and the median age is about 4 months, which is the end of the proliferative phase and exactly where this child is. The lip is one of the highest-risk sites, with the perineum, neck folds and axilla — anywhere moist or under friction.

My management has three parts. First, the wound: gentle cleansing, keep it moist with white soft paraffin, a non-adherent dressing, topical metronidazole for odour or mupirocin if colonised, and a short course of oral antibiotics if it is large or infected. Second, the pain, which is severe and under-treated — regular paracetamol, and topical lidocaine only sparingly, no more than four times a day, because of systemic absorption and methaemoglobinaemia in infants. Bleeding is direct pressure for ten minutes; catastrophic bleeding is very rare. Third — and this is the definitive treatment — systemic propranolol, because ulceration is itself an indication for systemic therapy and propranolol accelerates healing.

And I would say to the parents that once a lesion has ulcerated a scar is inevitable, so we are now treating to limit the scar, not to prevent it.”

“Propranolol, 3 mg/kg/day in divided doses, titrated up over two to three weeks, for at least six months and usually to twelve months of age.

It has been first-line since 2008, when Léauté-Labrèze in Bordeaux reported two infants given propranolol for cardiac indications whose haemangiomas regressed. The trial that licensed it is *Léauté-Labrèze, New England Journal of Medicine 2015 — a multicentre, randomised, double-blind, placebo-controlled adaptive phase 2–3 trial of 456 infants aged 1 to 5 months with proliferating lesions needing systemic therapy. Five arms: placebo, or 1 or 3 mg/kg/day for 3 or 6 months. Primary endpoint complete or near-complete resolution at 24 weeks by blinded central photographic review. 3 mg/kg/day for 6 months achieved that in 60.4% against 3.6% for placebo.* That is the regimen that was licensed.

In the UK the licensed product is Hemangiol — propranolol oral solution 3.75 mg/mL — with a licence for initiation between 5 weeks and 5 months of age. The AAP 2019 guideline says 2 to 3 mg/kg/day with 3 as the target for most infants. There is no NICE guideline; UK practice follows the SmPC, unit protocols such as the GOSH protocol, and the AAP guideline as the principal international reference.

It has displaced corticosteroids, interferon and vincristine almost entirely.”

“Before: full history and examination, heart rate, blood pressure, and a cardiac examination for a murmur. ECG only if there is bradycardia for age, an arrhythmia, a family history of arrhythmia or maternal connective tissue disease. Echocardiography is not routine without clinical concern — that is the Drolet 2013 consensus position.

Contraindications I check for: asthma or bronchospasm, sinus bradycardia or heart block above first degree, cardiogenic shock, uncontrolled heart failure, hypotension, phaeochromocytoma — and, in the UK licence, a corrected age under 5 weeks, Raynaud’s, and an infant prone to hypoglycaemia.

Initiation: start at 1 mg/kg/day in two divided doses, increase to 2 mg/kg/day in week two and to 3 mg/kg/day from week three — the Hemangiol titration — with at least nine hours between doses, given during or immediately after a feed. Heart rate at baseline and at least hourly for two hours after the first dose and after each increase — that is the peak-effect window and it is in the SmPC.

Where: outpatient initiation is appropriate for most. Inpatient for infants under 8 weeks corrected age, those with inadequate social support, comorbidity, PHACE with arterial anomalies, or hypoglycaemia risk.

The instruction that matters: hypoglycaemia. Propranolol blunts the glycogenolytic and adrenergic warning response — the child does not get sweaty and tachycardic, it just goes floppy. So: always with a feed; skip the dose if she is not feeding or is vomiting; stop and contact us during any illness. Given verbally, in writing, and to whoever else gives the feeds.

Duration: at least 6 months, usually to 12 months of age, longer for segmental or deep lesions. Wean rather than stop. And I counsel about rebound before we start.”

“Two alternative beta-blockers, both with randomised evidence, and I would reach for them for tolerance rather than efficacy.

Atenolol — cardioselective, does not cross the blood–brain barrier so less sleep disturbance, less bronchospasm, once-daily. *Ji’s 2021 randomised trial in JAMA Otolaryngology — 377 infants across six Chinese centres — found response at 6 months of 93.7% with propranolol and 92.5% with atenolol, with adverse events significantly more common on propranolol: 70% versus 44%. GOSH has published its own atenolol experience in the British Journal of Dermatology*. It is my choice for the infant with sleep disturbance or reactive airways.

Nadolol — non-selective, long-acting. *Pope’s randomised non-inferiority trial in JAMA Pediatrics 2022 — 71 infants — found nadolol non-inferior, and in fact more effective on size and colour at 24 weeks. The caution is important: nadolol is renally excreted and a large fraction remains in the gut, so it accumulates in constipation — McGillis reported a death in Pediatrics* in 2020 in a 10-week-old who had not stooled for ten days. Bowel habit has to be monitored.

Second line is prednisolone, 2 to 3 mg/kg/day for about a month then tapered. Bennett’s 2001 systematic review gave a response rate of 84% — but that is a pre-propranolol figure, and it comes with Cushingoid facies in about a fifth, growth retardation, hypertension, reflux, immunosuppression and deferred live vaccines. I reserve it for propranolol failure or contraindication.”

“Wrong drug for the lesion. Timolol maleate 0.5%, one drop twice daily, penetrates about a millimetre and is for thin superficial lesions. A 4 cm segmental facial haemangioma needs PHACE screening and systemic propranolol, and every week on timolol is a week of proliferation lost.

I would also be honest about the timolol evidence, because it is weaker than its popularity implies. The randomised trial is *Muñoz-Garza, JAMA Dermatology 2021 — 69 infants under 60 days, timolol against placebo for 24 weeks — and it found no significant difference in complete or near-complete resolution: 42% versus 36%, p=0.37*. There was an improvement in colour at week 4 and nothing else. So my position is: timolol is safe and reasonable for a thin superficial lesion that a parent wants treated, but the randomised evidence for meaningful resolution is limited, and it is never a substitute for systemic therapy in a lesion that threatens function or will disfigure.”

“Surgery has a much smaller role than it did fifteen years ago, and the question is almost always when, not whether. My default is to operate in the involuted phase, at three to four years of age — before school entry and before long-term self-image consolidates — on a smaller, less vascular, better-demarcated lesion.

The indications then are the fibrofatty residuum, redundant or anetodermic skin, reconstruction of destroyed structures — nasal tip, eyelid margin, vermilion, ear, eyebrow — and residual scarring from ulceration.

I operate in the proliferative phase only for a short list: failure of or contraindication to pharmacotherapy in a lesion threatening function; bleeding, painful or non-healing ulceration despite maximal medical treatment; obstruction of vision or airway not responding to propranolol; and a well-localised lesion in a favourable site where the scar of excision now would be identical to the scar of excision later — in which case earlier surgery avoids years of psychosocial burden.

The technique for a round lesion is circular excision with purse-string closure — Mulliken’s technique — which converts a circular defect into a shorter linear scar, minimises tissue sacrifice, and can be staged. Lenticular excision for the lip, eyelid, or where the scar sits in a relaxed skin tension line or an aesthetic unit boundary.”

“Then it is not an infantile haemangioma — it is a congenital haemangioma, and that is a different disease. Fully formed at birth, no postnatal proliferative phase, often seen on antenatal ultrasound, GLUT-1 negative, and driven by somatic activating mutations in GNAQ or GNA11. Equal sex distribution, usually solitary, often on a limb, with a violaceous plaque or exophytic mass, coarse telangiectasia, central pallor or ulceration, and a peripheral pale halo.

There are three behaviours, and they are probably a spectrum rather than three diseases: RICH — rapidly involuting, regressing from birth, half gone by 7 months; NICH — non-involuting, persistent fast flow, grows with the child, resected if symptomatic; and PICH — partially involuting, which starts like a RICH and then arrests.

The practical points: propranolol does not work on a congenital haemangioma; a RICH can cause transient thrombocytopenia and high-output failure that resolve with involution; and a NICH is a surgical lesion.”