Management of Kaposiform Haemangioendothelioma—A Retrospective Case Series Examining the Use and Effectiveness of Vincristine

ABSTRACT

Aim

Kaposiform haemangioendothelioma is a rare vascular tumour. Management has evolved over the last two decades, with most cases receiving vincristine or sirolimus. We aim to describe our experience with managing this tumour, with a focus on the safety and efficacy of vincristine.

Methods

A retrospective cohort study of all cases treated at Sydney Children's Hospital over a 19‐year period was conducted. Explanatory variables included age, tumour location, imaging, the presence of Kasabach‐Merrit Phenomenon, and treatment modality. The primary outcome was disease recurrence and the secondary outcome, complications.

Results

Twenty‐one cases were identified with 13 (60%) presenting as neonates. The most common site was limb; and imaging included ultrasound (7/21, 33%) and MRI (12/21, 57%). Almost all had consumptive coagulopathy (20/21, 95%) with 6/21 (29%) having severe Kasabach‐Merrit Phenomenon. Six tumours were excised and 16 (76%) were treated with vincristine, with three (14%) progressing to treatment with sirolimus. There were no recurrences. Complications of vincristine therapy were common but mild in most cases, with only one proven blood stream infection and one requiring central line replacement.

Conclusion

Vincristine remains a highly effective first‐line treatment for kaposiform haemangioendothelioma with an acceptable safety profile. Further well‐designed multicentre trials are needed to objectively compare its effectiveness and safety with sirolimus.

Key Points

• What is known?

  • ○
    Kaposiform haemangioendothelioma (KHE) is a rare vascular tumour which is life threatening due to its association with severe consumptive coagulopathy.
  • ○
    While excision is curative, this is not always possible.
  • ○
    Pharmacological therapy has evolved over the last two decades. Vincristine was widely and successfully introduced in the 1990's, and sirolimus has emerged in the last 10–15 years as an alternative therapy.
• What this paper adds?

  • ○
    As sirolimus gains favour, this paper serves as a reminder of the effectiveness of vincristine in treating KHE.
  • ○
    Where primary excision is not advised or possible, vincristine is well tolerated and achieves an excellent and sustained response in the management of KHE. It is particularly effective in achieving rapid resolution of Kasabach‐Merritt Phenomenon.
  • ○
    Vincristine should still be considered when selecting first‐line therapy, particularly in severe cases.

1. Introduction

Kaposiform haemangioendotheliomas (KHE) are rare vascular tumours with an incidence of less than 1/100000 [1]. They are usually diagnosed in the first months of life based on a combination of clinical examination, imaging, blood tests and biopsy [2]. Definitive diagnosis may be delayed and misdiagnosis is historically common, especially where the lesion is deep. Differentials include haemangioma of infancy, congenital haemangioma, infantile fibrosarcoma, congenital rhabdomyosarcoma and, rarely, non‐accidental injury [2, 3].

Clinically, KHE occurs mostly in soft tissues of the face, trunk and limbs, and may involve all tissue layers, including skin, subcutaneous tissues, muscle and, less commonly, bone [4, 5, 6]. They range from small vascular lumps to large, firm brawny‐purple tumours with surrounding bruising and oedema [2, 7]. It may enter organ spaces, but involvement of organs is rare. The mass effect and tissue infiltration are often associated with pain and impaired function [6].

Ultrasound reveals a high flow lesion with indistinct borders. Anatomical detail is better with MRI, which confirms its poorly circumscribed nature, the crossing of tissue planes and associated oedema [4]. Blood tests may reveal anaemia, thrombocytopenia and deranged coagulation profile with a raised D‐dimer [8].

The tumour is associated with anaemia and consumptive coagulopathy due to trapped red cells and platelets within the tumour vessels. The mechanism has been researched in detail by Ji et al. [9]. Simplified, haphazard lymphatic endothelial cell proliferation within the tumour leads to dysplastic angiogenesis. These vessels have poor endothelial integrity, exposing the extravascular matrix to platelets, to which they adhere. This causes the platelets to activate, secreting vascular growth factors, which add to further deranged endothelial growth and further platelet aggregation. Platelet activation also activates the clotting cascade leading to consumptive coagulopathy and hypofibrinogenemia with elevated D‐dimer (coagulation markers). This positive feedback cycle is known as the Kasabach‐Merritt Phenomenon (KMP), characterised by thrombocytopenia, microangiopathic haemolytic anaemia, intralesional haemorrhage and consumptive coagulopathy.

The incidence of KMP is approximately 70% [1]. Although no particular low platelet count and degree of coagulation derangement denotes KMP, the mortality risk of KHE is related to the bleeding risk of severe KMP. This is quoted to be as high as 30% [7], however, a more recent series found a much lower death rate of 1.3% [10].

Treatment includes management of the KMP, resection and/or pharmacological management. In managing KMP, routine transfusion of blood products is not advised as this can exacerbate further platelet aggregation, consumptive coagulopathy and tumour growth [11]. Complete resection is curative [12], although is often not anatomically possible. Historically, pharmacological options have included steroids and interferon alpha‐2 [13]. Vincristine gained favour in the 1990's, and sirolimus has emerged in the last 10–15 years as an alternative oral therapy.

Vincristine and sirolimus have different mechanisms of action. Vincristine inhibits proliferation of the actively growing tumour cells, including the tumour endothelial cells, reducing vascular dysplasia, extravascular matrix exposure to platelets and thus reducing platelet activation [14]. Sirolimus, an M‐TOR inhibitor, is anti‐inflammatory and antiangiogenic. The latter includes inhibition of lymphangiogenesis [15] and suppression of fibrosis [16].

Although sirolimus is now often a favoured first‐line therapy [17, 18], with or without steroids [19], it may not be effective as a single‐agent therapy for KMP, nor give sustained resolution of the tumour [9, 10, 20].

The problem with assessing KHE treatment response in the literature is the heterogeneity of included cases and therapeutic approach, small numbers and lack of long term follow up. As such, this study aims to:

• Describe the clinical, imaging and haematological features of a defined cohort of children with KHE.

• Assess the efficacy and safety of vincristine in the treatment of KHE.

We hypothesise that, where excision is not possible, vincristine is a safe and effective first‐line single‐agent treatment.

2. Methods

2.1. Study Design and Setting

A single centre, retrospective cohort study of all children presenting with KHE was conducted at Sydney Children's Hospital, Randwick, NSW, Australia, from January 1995 to December 2023. This is a tertiary referral children's hospital that established a multidisciplinary Vascular Tumours and Anomalies Centre (V‐TAC) in 1995. Ethics approval was obtained from the Sydney Children's Hospitals Network Human Research and Ethics Committee (2022/ETH01486). The STROBE guideline for reporting cohort studies was used [21].

2.2. Participants

Children with KHE were identified in the electronic medical record (EMR) and V‐TAC patient files. The EMR search used ICD‐10‐AM codes: ‘D18’, ‘M9120/0’, ‘M9120/3’, ‘M9121/0’, ‘M9125/0’, ‘M9130/0’, ‘M9130/1’, ‘M9130/3’, ‘M9132/0’, ‘M9133/1’, ‘M9133/3’ and both the EMR and patient files searches used the key words ‘kaposiform’, ‘haemangioendothelioma’, ‘kaposiform haemangioendothelioma’, or ‘KHE’. Retrieved records were manually reviewed and patients were included if they had a diagnosis of KHE based on clinical, haematological and/or biopsy features, with or without KMP. Patients were excluded if the final diagnosis after biopsy was not KHE, or if they were diagnosed with KHE but managed elsewhere.

2.3. Outcomes

The primary outcome was disease recurrence requiring further treatment during the follow up period. Secondary outcomes included treatment duration, survival, need for adjuvant pharmacotherapy and vincristine treatment complications. Complications included infection, vincristine‐specific side effects, bone marrow suppression and central line complications (Table 3). They were graded using Common Terminology Criteria for Adverse Events (CTCAE) for pharmacological complications, (Common Terminology Criteria For Adverse Events (CTCAE)—Cancer Science) and the Clavien‐Dindo classification for surgical complications (The Clavien‐Dindo Classification | AssesSurgery GmbH).

TABLE 3.

Vincristine treatment complications.

Complication	N (%)	Details/management	Grading
Febrile illness (blood cultures negative)	13 (81) a	Admission until cultures negative a	CTCEA III
Blood stream infection	1 (5)	Admission for IV antibiotics	CTCEA III
Bone marrow suppression	8 (50)	Iron supplementation	CTCEA II
CVL site infection	7 (33)	1 (5%) required early removal, not reinserted	1: CD III 6: CD II
Neuropathy (reversible)	2 (13)		CTCEA I
CVL early removal (not replaced)	2 (13)	1× due to repeat blockage of CVL and recurrent site infection 1× due to recurrent CVL site infections	CD II
Sepsis (non‐CVL)	1 (6)		CTCEA III
CVL replacement	1 (6)	Due to displacement early in treatment course	CD III
Poor weight gain/growth	6 (38)	1 (5%) required admission for NGT feeding	5: CTCEA I 1: CTCEA II
Constipation	5 (31)	Aperients	CTCEA II
CVL dermatitis	5 (31)	Topical steroids	CD II
CVL blocked	3 (19)	2 required alteplase, 1 just ran fluids	CD II
Nausea	1 (6)		CTCEA I
Jaw pain	1 (6)		CTCEA I

Abbreviations: CTCEA, common terminology criteria for reporting adverse events; CD, Clavien Dindo classification.

^a 13 total patients had at least one febrile illness with negative blood cultures, 3 of those had one febrile illness with associated neutropenia.

2.4. Explanatory Variables

Explanatory variables are summarised in Table 1. They included patient and lesion characteristics, presentation details, blood test results, imaging modality, biopsy and treatment. The lesion was described according to body region location and tissue involvement. Blood tests included full blood count, D‐dimer and fibrinogen. Coagulopathy was defined as D‐dimer > 2× normal and Kasabach‐Merritt Phenomenon (KMP) was defined as present if the platelet count was < 150 × 10⁹/L and severe if < 50 × 10⁹/L [10].

TABLE 1.

Patient characteristics.

Sex
Female (n(%))	7 (33%)
Male (n(%))	14 (67)
Presentation
First week of life (n(%))	8 (38)
1–4 weeks of life (n(%))	5 (24)
4–52 weeks of life (n(%))	7 (33)
> 1 year of life (n(%))	1 (5)
Age (days) at definitive diagnosis (Median (IQR))	27 (7–161)
Time (days) from presentation to definitive diagnosis (Median (IQR))	5 (3–11)
Location
Head and neck (n(%))	4 (19)
Thoracic wall (n(%))	3 (14)
Abdominal wall (n(%))	4 (19)
Pelvis (n(%))	1 (5)
Upper limb (n(%))	3 (14)
Lower limb (n(%))	6 (29)
Clinical characteristics
Mass (n(%))	19 (90)
Discolouration (n(%))	16 (76)
Pain (n(%))	6 (29)
Loss of function (n(%))	2 (10)
Blood tests at presentation
Platelet count < 150 × 109/L (n(%))	9 (43)
D‐Dimer > 0.5 (n(%))	20 (95)
D‐Dimer > 1 (n(%))	17 (81)
Fibrinogen < 1.0 g/L (n(%))	4 (21) (missing n = 2)
Severe KMP (Platelet count < 50 × 109/L) (n(%))	6 (29)
Imaging at presentation
MRI (n(%))	11 (52)
US (n(%))	11 (52)
PET CT (n(%))	3 (14)
Tissue involved
Skin (n(%))	13 (62)
Subcutaneous tissue (n(%))	19 (90)
Muscle (n(%))	6 (29)
Bone (n(%))	3 (14)
Organ space (n(%))	1 (5)
Treatment
Excision alone (n(%))	4 (19)
Excision plus vincristine (n(%))	2 (10)
Vincristine alone (n(%))	11 (52)
Vincristine followed by sirolimus (n(%))	3 (14)
Steroids (n(%))	2 (10)
Length of follow‐up (years) (Median (IQR))	6.1 (2.8–9.2)

Regarding treatment, up‐front excision was undertaken when feasible. Otherwise, vincristine was first‐line. A central line was inserted (port or tunnelled single lumen line) and the protocol included weekly vincristine for 6–12 weeks, followed by fortnightly then monthly until response was deemed adequate or a change to sirolimus considered. Pneumocystis prophylaxis was given until the course was complete and 3 months thereafter. Response to vincristine and decisions about duration of therapy were based on clinical findings, progress imaging, and D‐dimer and platelet measurements. Sirolimus 0.8 mg/m² BD was given when response to vincristine was slow, vincristine was not tolerated, or prolonged treatment considered necessary. Blood products were given for severe anaemia and for KMP if life‐threatening bleeding was expected, particularly in the context of biopsy or excision.

2.5. Analysis

Data was entered into REDCap. Analysis was performed with Excel.

The study size was dictated by the number of cases over the study period. Categorical variables were described in terms of frequency and percentage and continuous variables in terms of mean (standard deviation [SD]) where normally distributed, otherwise as median (interquartile range [IQR]).

3. Results

The EMR search identified 435 records, and search of the V‐TAC files identified a further six cases. After manual review and excluding duplicates, 21 cases met inclusion criteria over the 19‐year study period.

Table 1 outlines patient and lesion characteristics. Male to female ratio was 2:1. Over 60% (n = 13) presented as neonates: eight within a week of birth and five within the first month of life. Only one child presented after age one. The most common site was limb (9/21, 43%) followed by head and neck (4/21, 19%); thorax (3/21, 14%) and abdominal wall (4/21, 19%). One was located in a deep pelvic position (Figure 1). In most, presentation was due to a mass (19/21, 90%) with or without discolouration (16/21, 76%), with only six (29%) presenting with pain and two (10%) with loss of function of the associated limb.

FIGURE 1.

Lesion location.

The lesion was characterised on ultrasound alone in seven (33%), while the rest had cross‐sectional imaging with MRI (n = 12, 57%), CT scan (n = 1, 5%) or both (n = 1, 5%). Three patients (14%) also had PET CT, as the original working diagnosis was sarcoma. The majority (n = 14, 67%) were confined to the skin and/or subcutaneous tissue alone. Six (29%) extended to involve the underlying muscle, with two of these (10%) also having bony involvement. One head and neck case without muscular involvement had bony destruction and extension intracranially.

Consumptive coagulopathy at diagnosis was present in 20/21 patients (95%): low platelets (< 150 × 10⁹/L) were present in 9/21 (43%) and 20/21 had an elevated D‐dimer (95%). Two patients (10%) with low platelets had spontaneous resolution of this prior to treatment commencement.

Six patients (29%) met the criteria for severe KMP. Five required blood products: four in the context of tumour biopsy or excision and one due to life‐threatening KMP. Two required subsequent transfusions for management of persistent KMP. One child died in the neonatal period due to complications from KMP, sepsis and multiorgan failure, without definitive management of their KHE.

All patients underwent biopsy which was informative in 20/21 cases. The exception was a newborn with an unequivocal clinical diagnosis of KHE and severe KMP. They had involvement of an entire lower limb but biopsy failed to sample lesional tissue due to considerable oedema.

Treatment is summarised in Tables 1 and 2. Sixteen (76%) were managed with vincristine, with 11/16 (69%, 52% overall) not requiring any other treatment. Three (19%, 14% overall) progressed from vincristine to sirolimus after a median of 18 doses (range 11–31): one as a vincristine sparing manoeuvre in an older child with persistent tumour bulk; and two due to a plateau in vincristine response based on D‐dimer. Two patients received steroids, both early in the series: one with severe KMP who was too unwell to receive vincristine and ultimately died from sepsis, and another in combination with vincristine.

TABLE 2.

Vincristine treatment protocol and outcomes.

Total number treated with vincristine (n)	16
Platelets < 150 × 109/L at start of treatment (n(%)) a	4 (27) (missing n = 1)
Severe KMP at start of treatment (n(%)) b	3 (19)
Line type
Port (n(%))	7 (44)
Single lumen cuffed tunnelled central line (n(%))	9 (56)
Adjuvant therapy
Pretreatment excision	1 (6)
Post treatment excision	1 (6)
Post treatment sirolimus	3 (19)
Treatment details
Number of weekly doses (median (IQR))	12 (7.75–12)
Number of fortnightly doses (median (IQR))	4 (3.75–6)
Number of monthly doses (median (IQR))	0 (0–2.75)
Total number of doses (median (IQR))	17 (12.75–20.5)
Duration of treatment (days) (median (IQR))	185 (126.5–254)
Outcome
Doses until platelets normalised (median (range))	2 (1–2)
Recurrence	0

^a Two further patients had platelets < 150 × 10⁹/L at presentation but were normal on the day of the first vincristine dose. One further patient had complete platelet correction following pre‐treatment excision.

^b One further patient with severe KMP had complete platelet correction following pre‐treatment excision.

Six cases (29%) underwent excision, two with severe KMP which resolved within 24 h of the operation. One patient had preoperative vincristine, one postoperative vincristine and subsequent wider excision, and four (19%) underwent excision alone.

Regarding severe KMP (n = 6), one died without definitive treatment. Two resolved completely after early excision, one of which had adjuvant vincristine and re‐excision due to extension to margins and histological findings of a high mitotic rate. The other had no further treatment. The other three commenced vincristine, with KMP resolving after two doses in all. One case later transitioned to sirolimus due to a plateau in vincristine response (Table 2). There were no cases of recurrent KMP after initial resolution.

Complications of vincristine are summarised in Table 3. Fifteen (94%) children treated with vincristine had at least one complication, although most were CTCAE grade III or less and none persisted beyond the end of therapy. Thirteen (81%) had at least one febrile illness, three of which were neutropenic, and all prompted the protocolised response of admission, blood cultures and intravenous antibiotics. Only one had a positive blood culture—Streptococcus viridans. Eight (50%) had documented bone marrow suppression based on the haemoglobin of < 100 g/L, all managed with iron supplementation and none requiring blood product transfusions outside of initial management of their KMP.

There were three (19%) central line complications requiring earlier than planned removal: one due to blockage; one due to displacement; and the other due to recurrent entry site cellulitis. Only the displaced line was replaced. Two (13%) developed possible vincristine‐related neuropathy: both a transient ptosis which recovered after increasing dosing intervals; one of whom also had sensory axonal neuropathy on nerve conduction studies.

Median vincristine treatment duration was 185 days (IQR 126.5–254 days), or 16 doses (IQR 12.75–20.5 doses). The median follow‐up was 6.1 years (range 0.9–15.8 years; IQR 2.9–9.22 years). There have been no recurrences.

4. Discussion

Vincristine has previously been established as first‐line therapy for KHE, although recently superseded by sirolimus in many centres. This single centre series describes the clinical and haematological features of KHE and response to treatment, with a focus on the efficacy and safety of vincristine. It demonstrates that, where primary excision is not advised or possible, vincristine is effective, well tolerated and achieves an excellent and sustained response in the management of KHE. There were no recurrences and no deaths in those who were able to commence treatment. Vincristine was particularly effective at achieving a rapid resolution of KMP, with platelets normalising after two doses in all cases. Complications were common, but most required supportive treatment only, with only one proven blood stream infection and one replacement of a central line.

Vincristine was first used in this centre at a time when steroids were still first‐line therapy in many places. Over time, a consistent approach and well‐defined protocol have allowed good assessment of treatment response in a small but well‐defined cohort. This is in contrast to some other series where the included cases and interventions are varied [10]. While others have suggested that monotherapy is not advised in the presence of severe KMP [22, 23], our experience is that vincristine alone is rapidly effective alongside supportive measures, with no rebound recorded.

The common use of vincristine for other tumours means protocols are well established and side effects understood. Review by the haematologist/oncologist at each visit ensures close monitoring of treatment response, with tailoring of dose frequency and treatment duration. A median treatment duration of 6 months is considerably shorter than that recommended for treatment with sirolimus which often extends well beyond 12 months [11, 24].

Both vincristine and sirolimus have resource utilisation and side effect profiles that need to be considered when selecting therapy.

Regarding resource utilisation, initial diagnosis and instigation of management are likely to occur as an inpatient. Children receiving vincristine require a central line insertion and removal—necessitating two general anaesthetics. However, insertion can be coordinated with biopsy, also under general anaesthetic, and the line can be used for blood tests as well as drug administration during treatment, improving the patient experience. Once discharged from hospital, vincristine is administered as a day admission, while sirolimus and steroids are given as an outpatient. However, both require close monitoring with frequent review and blood tests [18]. The recommended duration of treatment is longer for sirolimus [17]. Admission to hospital for fever is recommended for infants treated with either drug as they are immunosuppressive, although most episodes would be considered low risk, with discharge once cultures are negative and the child clinically stable. We had one patient requiring a longer admission for confirmed bacteraemia.

Side‐effects of vincristine include bone marrow suppression and infection risk; constipation; jaw pain; and peripheral neuropathy [10]. Haematological and systemic side effects of vincristine in this series were mild. Vincristine‐associated anaemia was noted, but no child required transfusion for this. Infection is always a concern in any immunocompromised patient. Oncology protocols mandate hospitalisation for fever, which happened for around half of our cases. However, only one patient had subsequent positive blood cultures, with other febrile episodes either proven or likely to be viral in aetiology. There were a number of central line issues, but these were mostly managed on an outpatient basis, with only three requiring early removal and only one replacement. In infants, peripheral neuropathy is difficult to detect. The two cases of ptosis resolved on completion of treatment.

Sirolimus is largely well tolerated, but side effects include mucositis, headaches, poor wound healing and immunosuppression with infection risk [10, 25]. As it is recommended that sirolimus be given in combination with steroids in the presence of KMP [2, 10], steroid side effects also require consideration in that context. They include the risk of hypertension, effects on growth, sleep disturbance and hyperglycaemia [26, 27]. Response onset is slower with sirolimus than vincristine [10, 19]. Thus, we recommend caution with severe KMP or other concerning symptoms. By 6 months, response is similar to that of vincristine [10], but recommended treatment duration is definitely longer, with a median of 15.2 months [17]. In addition, Borst's 2024 systematic review found the incidence of persistent disease to be higher compared with vincristine, but could not draw conclusions as some patients were commenced on sirolimus due to plateau in vincristine response, as was the case in two of our patients [10]. This impression, however, is corroborated by a 2016 meta‐analysis looking at all pharmacological treatments, which found vincristine to be the most effective [28]. Similarly, Zhou reported a 17.3% rebound rate after cessation [20], with treatment duration of less than 18 months a risk factor for rebound.

5. Limitations

As a retrospective cohort study, this research has limitations. An institutional preference for treating KHE with vincristine during the study period is acknowledged. Although all patients had a confirmed diagnosis of KHE, the tumours and presentations were heterogeneous, with some missing information. Similarly, KMP was difficult to define and fibrinogen levels were not routinely done early in the series and therefore not included. Vincristine treatment protocols and monitoring were uniform, but assessment of treatment response and subsequent protocol adjustment or cessation were subjective, not always well documented and evolved over time.

6. Conclusion

Where complete excision is not possible, vincristine remains effective first‐line therapy for children with a definitive diagnosis of KHE, with rapid resolution of potentially life‐threatening KMP. Accurate case selection based on imaging and biopsy where possible is important, and careful management of haematological and other complications in a multidisciplinary team is critical. While treatment adjustment, the addition of sirolimus and end point were determined on a case‐by‐case basis, treatment was successful; steroids were not necessary and no child experienced recurrence. Further prospective research into the true definition and significance of KMP, the ideal duration of treatment, the definition of adequate and sustained treatment response and objective comparison with sirolimus and/or surgery, incorporating resource utilisation, are warranted. This will help clinicians to continue to evolve their management approach.

Funding

The authors have nothing to report.

Ethics Statement

Ethics approval was obtained from the Sydney Children's Hospitals Network Human Research and Ethics Committee (2022/ETH01486).

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.