Results
Of the initial 476 identified articles, 23 full text articles were assessed for eligibility. Eventually, 15 case reports were included in this review (figure S1) . Baseline patient characteristics, clinical presentation, mode of diagnosis, pregnancy course and the maternal and neonatal outcomes in the 15 pregnancies described in these articles are given in table S1 and S2.
Of the 15 pregnancies, two (2/15, 13.3%) resulted in a stillbirth possibly due to the uterine hemangioma. The first one occurred at 28 weeks after massive thrombosis of the uterine and placental vessels, the latter following acute rupture of the uterine hemangioma at 36 weeks. Another two pregnancies were delivered preterm (2/15, 13.3%). One patient delivered at 35 weeks after preterm prelabour rupture of the membranes (PPROM) at 26 weeks, and a second patient had a cesarean section at 30 weeks due to progressive abdominal discomfort and the assumed bleeding risk of the hemangioma. Forty percent of women (6/15, 40.0%) went into labour spontaneously, which resulted into a vaginal delivery in four cases. Most patients were delivered by cesarean section (10/15, 66.7%), of which seven (7/10, 70%) were unplanned/in an emergency setting. The postpartum period was complicated by a hemorrhage in eight women (8/15, 53.3%), which necessitated a hysterectomy in four cases (4/8, 50%). Two women developed progressive hypovolemic shock (2/8, 25%). Of those for whom data was provided (5/8, 62.5%), all had an estimated blood loss of ≥ 1000 mL. Furthermore, another two patients developed a pulmonary embolism in the postpartum period, ultimately fatal in one of them. Perinatal outcomes were mentioned in only six reports (6/15, 40%). Only half of these cases (3/6, 50%) had an uneventful outcome. One case of respiratory problems due to preterm birth at 30 weeks was described beyond the two stillbirths mentioned above.
The hemangioma was diagnosed before delivery in the majority of cases (10/15, 66.7%), generally during the second trimester of pregnancy. Most often, symptoms such as abdominal discomfort, dyspnea, or vaginal bleeding and/or the finding of an enlarged uterus led to a work-up. Four patients were referred for abnormal ultrasound findings, which in all but one case were suspicious for a partial mole. The antenatal diagnosis was generally based on ultrasound findings, MRI was used to confirm the hemangioma in only one patient.
Half of the patients (5/10, 50%) with an established diagnosis of hemangioma before delivery had a non-planned cesarean section. In two of these patients (2/5, 40%), this was for reasons related to the hemangioma: one patient had a cesarean section for fetal death at 28 weeks and another patient was delivered at 30 weeks due to progressive abdominal discomfort as mentioned earlier. Five of the 10 antenatal diagnosed patients (5/10, 50%) developed a postpartum hemorrhage, which necessitated a hysterectomy in only one of them. Among the patients with an antenatal diagnosis of a uterine hemangioma, one more patient underwent a hysterectomy at 17 weeks because of recurrent syncopes and the presumed risk of uterine rupture.
These numbers contrast with the undiagnosed group in which three out of the five patients (3/5, 60%) had a postpartum hemorrhage and a hysterectomy was necessary in all of them (3/3, 100%). There were no cases of maternal mortality in the group of antenatal diagnosed patients.