Observation as a Therapeutic Option for Head and Neck Lymphatic Malformations: A Survival Analysis of Spontaneous Regression

Abstract

Importance: Head and neck lymphatic malformations (HNLM) demonstrate significant variability in their natural history. While some malformations cause chronic severe functional impairment, others are asymptomatic and spontaneously regress. Understanding the frequency of spontaneous regression and the characteristics associated with regression will aid clinicians and families in making informed treatment choices and avoid unnecessary risks of intervention for a subset of patients. Objective: To determine the incidence of spontaneous regression and identify factors associated with spontaneous regression. Design: Non-concurrent prospective cohort study. Setting: Single regional pediatric tertiary care academic center. Participants: Patients with HNLM who were seen in the Vascular Anomalies Clinic and prospectively enrolled in an institutional quality improvement database between 2003 to 2022. Exposure(s): Age of HNLM onset, primary location, distribution, cystic structure type, grade and De Serres staging. Main Outcome and Measure(s): The incidence of complete spontaneous regression was estimated for the study population. A Kaplan-Meier curve was used to estimate the probability of spontaneous regression occurring over time. The two-sided log-rank test was used to test the association between exposure variables and survival. The effect of the exposures on the likelihood of developing spontaneous regression were assessed using Cox proportional hazards regression models. Results: Of 368 patients with HNLM, 56% were male, and 55% were white. Most HNLMs were diagnosed prenatally or during the first year of life (53.5%), had a focal distribution (78.5%), were De Serres I or II (69%), and macrocystic (47.1%). Among the study population, 6.2% experienced complete spontaneous regression, at a median time of 23.3 months from HNLM onset. Factors significantly associated with complete spontaneous regression included cystic structure, HNLM distribution, primary location and De Serres staging (p<0.05). Conclusions and Relevance: Macrocystic, localized HNLM are more likely to experience spontaneous regression. Future studies will examine interaction among these factors to better understand the drivers of regression. This work contributes to a deeper understanding of the natural history of HNLM that can directly inform clinical decision-making, decrease treatment risk, and optimize patient outcomes.