Introduction
Lymphatic malformations (LMs) are rare slow-flow vascular malformations with an estimated prevalence of 1:4000 births, which commonly occur in the head and neck (48%) and other lymphatic rich areas [
3]. The majority (60%) of LMs is observed at birth and 90% before the age of 2 years. Consequently, patients are predominantly young children with both genders being equally affected [
8]. LMs can cause significant morbidity in 70% of cases with intermittent swelling (44%), pain (36%), intralesional bleeding (23%), (recurrent) secondary infection (20%), airway compromise (11%), lymphorrhea (6%), cellulitis, cosmetic disfigurement and lymphocytopenia [
2].
Infection or intralesional hemorrhage, which is present in 35% of cases, can lead to acute complications [
9]. An overall mortality rate of 3.4–5.7% has been reported with LMs [
11]. Peri-operative tracheotomy (8.3%) or prolonged endotracheal intubation can be required to ensure safe treatment [
16]. LMs in the head and neck can be subdivided in macrocystic (21%), microcystic (24%) and mixed (49%) subtypes, in which occurrence varies along different subsites of the head and neck region. Midline and oral lesions tend to be more microcystic, whereas parotid and submandibular lesions are more often mixed and cervical lesions are predominantly macrocystic and mixed [
17]. The term cystic hygroma and lymphangioma refer to macrocystic LM and microcystic LM respectively and should be abandoned as they insinuate a neoplastic origin [
18].
LMs may be the result from an aberrant bud arising from a primordial lymph sac and are associated with genetic disorders including trisomies 13, 18 and 21, Noonan syndrome, Turner syndrome, CLOVES syndrome and Klippel-Trenaunay syndrome [
8]. In a large number of patients a mutation in the PIK3CA gene, known to play a role in cell growth, isolated in the lymphatic endothelium is found [
1]. However the exact pathogenesis of the condition is still unknown.
The management of LMs in the head and neck region is still challenging and requires a multi-disciplinary team of head and neck surgeons, interventional radiologists and maxillofacial surgeons. The principal goal of LM management is restoration or preservation of functional and aesthetic integrity. Spontaneous involution can occur in approximately (3%) but is reported with varying frequency (0–41%) [
21].
Various treatment modalities are used to treat this condition, however no consensus exists regarding optimal treatment. Currently, the main treatment options are surgical resection and sclerotherapy or a combination of these, though newer systemic regimens such as the administration of sildenafil are also being ventured [
23]. Initially, surgery was the mainstay of treatment. It’s effectiveness lies in removing the totality or subtotality of the cyst(s) and its lining [
22]. This can be hampered by the extent of the lesion and trans-spatial growth, which occasionally involves vital structures. As concerns for damage to nervous and vascular structures during surgery rose, sclerotherapy gained popularity among healthcare professionals due to its uncompromising nature of these structures [
7]. The effect of sclerotherapy relies firstly on the collapse of the cyst on itself by aspiration and then partly refilling the cavity with a sclerosing agent to initiate an inflammation response of the endothelial lining to ensure fibrosis of the lining on itself, thus eliminating the cavity [
22]. Recently, multiple sclerosing agents were trialed in a search for the optimal safety-efficacy profile [
24]. Additionally, the potential for recurrent infections, airway compromise, feeding difficulties, interference with the development of normal speech and concerns for aesthetics, complicate treatment planning. De Serres et al. developed a staging system based on location and extension of the LM that helped to predict outcome of surgery. De Serres noted a positive correlation between higher stages and higher complication rate in operated patients, with a 100% risk of complication in with de Serres stage 5 LM [
15]. However, this staging system does not take into account the different lesion configurations such as microcystic, mixed, macrocystic, diffuse and focal lesions for each of which management and outcome can differ vastly [
25]. The de Serres classification can be found in Table
1. The de Serres staging system combined with cyst typing has been increasingly used. Improvements in imaging possibilities and capabilities aided in more accurate diagnoses of the type and extent of the LM [
7]. With reports using improved imaging and more universal staging systems, a better understanding of treatment response can be obtained. Trends become visible showing that, analogous to surgery, macrocystic LMs respond better to sclerotherapy than mixed or microcystic LMs [
24].
Table 1
De Serres staging
1 | Unilateral infrahyoid |
2 | Unilateral suprahyoid |
3 | Unilateral supra- en infrahyoid |
4 | Bilateral infrahyoid |
5 | Bilateral supra- en infrahyoid |
The most recent systematic review regarding the treatment of LMs in the head and neck region dates from 2012 [
26]. Since 2012 combined therapies of sclerotherapy and surgery have been introduced and new research data with often better description of the pathology has appeared [
28]. In this systematic review and meta-analysis we compare the effectiveness of three different treatment modalities, sclerotherapy, surgery and sclerotherapy combined with surgery along the cyst type, in an attempt to aid decision-making in treatment for the different types of LMs in the head and neck area.
Methods
Selection of studies
A review protocol based on the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) statement (
http://www.prisma-statement.org) was used. No ethics committee approval was required. A systematic search on PubMed, Embase.com and Clarivate Analytics/Web of Science Core Collection from inception up to November 2021 (by JLAE, SCH and JCFK). The following terms were used (including synonyms and closely related words) as index terms or free-text words: ‘lymphatic malformation’ and ‘head and neck’. The full search strategy for all databases is available in the supplementary information appendix A. Duplicate articles were excluded. Two authors (SCH, JLAE) independently screened titles and abstracts. We applied the following exclusion criteria: duplicate studies or use of the same dataset, studies not discussing LM, LM not located in the head and neck area (intra-orbital LMs were excluded), absence of cyst type description, treatment modalities other than sclerotherapy or surgery or their combination, venoLMs, animal studies, studies with fewer than 5 cases, cases without intention to treat, studies with a follow-up of less than 0.5 years, and studies discussing EXIT procedures. In cases where abstracts were missing, a full-text review was carried out. After exclusion based on abstracts, both authors independently screened the remaining articles’ full texts. Risk of bias assessment was carried out using the Newcastle–Ottawa quality assessment scale for cohort studies. The scale was modified excluding the comparability follow-up question. Studies with less than 4 points were excluded in addition to studies lacking data on treatment outcome and when treatment results were not traceable to a specific treatment. Studies in English or French were screened by someone fluent in that language, articles in Spanish or Portugese were translated with DeepL version 3.7.277083. Articles in other languages were excluded.
Data pertaining type of study, location of the institution, treatment goal, population demographics, LM type (microcystic, macrocystic or mixed) and staging (de Serres stage), mediastinal involvement, treatment modality, different sclerosing agents, number of treatments, tracheostomy, adverse events, recurrence rate, follow-up period, treatment outcomes (degree of volumetric reduction, complications, number of treatments needed, average number of treatment) and adjuvant therapy was extracted from the studies.
When available, individual patient level data were extracted for a separate analysis. The de Serres stage and lesion type were noted by the reviewer or inferred from the description whenever possible. Volume reduction was categorised as follows: 1 = 0% reduction, 2 = 1–49% reduction, 3 = 50–74% reduction, 4 = 75–90% reduction, 5 = 100% reduction. In instances where studies described volume reduction qualitatively, we assigned the minimum percentage decrease in volume according to Appendix B. Grading of adverse events was done according to the Clavien-Dindo classification of adverse events [
29]. Recurrences were reported when the lesion increased in size after an initial decrease in size post-intervention. In cases where the de Serres staging was reported but treatment outcomes were assigned along different cyst types, the de Serres staging was allocated to the different cyst type subgroups according to their weighted distribution.
Data analysis
Zotero (version 6.0.26) was used for reference managing. Microsoft Excel 2013 (version 16.72, Microsoft Corp, Redmont, Washington) was used for creating the dataset. R version 4.3.2 (2023-10-31) was used for descriptive statistics and analysis of the data. ANOVA,
t test, Kruskal–Wallis and Chi-square tests were used to test for significant differences. Confidence intervals will be reported between parentheses throughout the article. We used the meta package for R to conduct a meta-analysis with a random effects model and tested for between-study heterogeneity with the Higgins and Thompson’s
I2 statistic [
30]. Meta-analysis were performed on the different cyst types and events of more or equal than fifty percent or hundred percent volume reduction. Volume reduction is quantified as the percentual decrease from the initial volume. To compare the estimates of the different treatments, a fixed-effects meta-regression model was used. Using individual patient data, we estimated associations between patient characteristics and treatment outcome using a one-stage multiple linear regression model to forecast initial lesion volume reduction using the predictors treatment modality, cyst phenotype, and de Serres staging. The reference standards were sclerotherapy, microcystic lesion and de Serres stage one.
Discussion
This meta-analysis compared different treatment modalities of head and neck LMs. Upon evaluation, there was no significant difference in volume reduction between one of the three therapies (surgery, sclerotherapy or combined) in attaining either hundred percent or more than fifty percent volume reduction. Notably, macrocystic lesions showed the most significant amount of volume-reduction post-treatment, succeeded by mixed and microcystic lesions. These findings are confirmed in previous literature [
37]. A lower mean amount of treatment sessions were necessary in the surgical group compared with the sclerotherapy group (
p < 0.001). Three studies reported on combined treatment, which demonstrated a superior response to therapy in complete as well as partial response in mixed lesions [
27]. However this did not significantly differ from other treatment modalities and the number of studies that reported on combined therapy was low. The cases presented in the combined treatment group had an higher mean de Serres stage than the other treatment groups. Previous trends tend to show a less successful outcome with a higher stage [
7]. The de Serres stages does not account mediastinal expansion which can influence treatment planning. In a cohort of Ghaffarpour et al. patients requiring emergency surgery presenred with involvement of the mediastinum and/or the abdominal or retroperitoneal cavities [
38]. The same author suggested that surgery is the treatment of choice in LM with mediastinal expansion [
28]. Respiratory failure and obstruction were observed exclusively after sclerotherapy. This discrepancy might be attributed to surgeons frequently anticipating respiratory complications and consequently employing tracheostomies. Rates of adverse events in surgery such as infection (in 1.5%) [
35] conformed as well. Rates of nerve injuries 3.5% and recurrences 7.6% in surgery groups were more prevalent in a surgery group [
35] but in other studies non-existent [
40]. Fever was reported as an Claviend-Dindo class 1 adverse event, however in sclerotherapy such an event might be expected due to the injection of an inflammatory agent in sclerotherapy.
The meta-analysis conducted in this study carries limitations that merit discussion. First, the principal goal of LM management is restoration or preservation of functional and aesthetic integrity and elimination of objective and subjective symptoms related to the abnormality. In the current literature on head and neck LM, treatment objectives are infrequently documented and solely evaluating volume reduction might not adequately describe treatment success. The majority of the studies lacked a systematic presentation of patient-reported outcomes. Earlier a patient satisfaction of approximately 50–60% was reported after sclerotherapy [
41,
42]. Sclerotherapy generally leads to moderate patient-reported improvement in health and quality of life in about half of treated patients, irrespective of the type, size and location of the lesion [
43]. Second, the quality of evidence varied among the different studies. No studies featured randomisation between the different treatment modalities and the predominance of the data consisted of restrospectively collected case series. Only 9.1% of included studies were prospective studies, the remaining were retrospective case series. Third, the data was collected from various medical resources and outcomes were measured heterogeneously. Surgery and combined therapy showed the most dramatic response in terms of volume reduction. However, treatment goals between surgery and sclerotherapy might differ. Surgery often aims towards complete excision of the lesion, whereas sclerotherapy is more directed towards volume reduction of the cysts enough to restore anatomical function and relieve subjective symptoms. Contrastingly, in most studies reduction of volume after sclerotherapy was assessed using imaging while surgical outcomes were more often evaluated clinically. Hence, the treatment effect of the surgical groups on quality of life could be overestimated. In the absence of reporting on cyst type, de Serres staging or degree of volume reduction, the outcomes were deduced when possible from the case descriptions in a conservative manner. i.e. when distinction between two de Serres stages was unclear, the lower de Serres stage was reported in our series. Consequently, this could lead to a underestimation of treatment results. In several articles recurrences were reported after sclerotherapy. A clear definition of recurrence was most often not reported and can arguably also be defined as residual lesion. We defined recurrence as an increase in size of the lesions after an initial decrease post-intervention, which followed how this was defined in other articles. In numerous studies, instances of recurrence following sclerotherapy were documented [
44]. However, these articles frequently lacked a clear, standardized definition of ‘recurrence’, with some equating it to a ‘residual lesion’. This ambiguity necessitates a clear operational definition, since after sclerotherpy there is always a residual lesion on MR imaging after therapy. Whether there are clinical symptoms or not. For our research purposes, we have characterized ‘recurrence’ as an increase in lesion size following an initial reduction post-intervention, aligning with definitions presented in certain previous studies [
46]. Fourth, outcomes could not always be distinguished between the different treatment subgroups and were calculated as weighted means. Consequently, the affected outcomes regress towards the average outcome value of the subgroups combined, potentially obscuring distinct insights. Fifth, the choice of treatment could be influenced by anatomical location and type of the lesion as well as local experience among specialist resulting in a selection bias. It is more likely that lesions most suitable for one of the treatment options (as judged by the treating phsyician) will be treated accordingly. For surgery extension of the lesion is important in decision making, whereas for sclerotherapy it is the cyst type, e.g. macrocystic LM. The aim of the chosen treatment may also differ, complete resection or tumor reduction only. Sixth, the de Serres classification of LM’s was used for subgroup analysis. The de Serres classification does not take into account mediastinal expansion or separate oral cavity analysis which could negatively impact treatment outcome as oral cavity lesions are often micocystic and more prone to obstruct the oral airway. In the individual group most microcystic lesions treated by surgery were located in the tongue.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.