Our search yielded 75 articles from PubMed and 101 from Embase and Medline published between the start of 2000 and the end of 2020. ClinicalTrials.gov did not contain articles about rAML and pregnancy. After exclusions, we included 45 case reports, with 48 cases described, and 1 retrospective study (Fig. 1).

The 45 case reports included data from 48 cases (Additional file 1, 2, 3 and 4). Each case involved a patient with at least one pregnancy. The group consisted of women with an average age of 30.3 (SD 5.5; n = 48) years old. The diagnosis TSC was established, based on clinical or genetic diagnostic criteria, in nine of the 48 cases (18.8%) (Table 1, Additional file 1, 2, 3 and 4). From these nine cases, seven women had the TSC diagnosis before the described pregnancy. The other 39 patients were not investigated for TSC, had sporadic rAML or no mention of TSC in their case report. The average rAML size was 11.0 cm (SD 4.6; n = 40). From the 48 cases, 8.3% of the women received rAML treatment (embolization (n = 4) or/and everolimus (n = 1)) before pregnancy.

We divided the 48 cases into four different groups. One group with an evidently established TSC diagnosis (n = 9, Table 1, Additional file 1). The other patients (n = 39), without TSC diagnosis (not tested, not mentioned, or proven sporadic rAML), were divided into 3 groups based on the moment the presence of rAML became apparent: rAML diagnosis before pregnancy (n = 8, Additional file 2), rAML diagnosis during pregnancy (n = 28, Additional file 3) and rAML diagnosis after pregnancy (n = 3, Additional file 4). The majority of the women (n = 41) came to the hospital with complaints related to rAML in the 2nd (43.9%) or 3rd (34.1%) trimester. The complaints were predominantly flank pain and hematuria.

In order for us to determine whether measured rAML sizes were reliable for interpretation we set up two criteria. Firstly, rAML size measurements had to be collected during pregnancy or one year before or after the pregnancy. This way we limited the change in size caused by time. Secondly, rAML sizes measured after individuals were treated for rAML by embolization or mTOR inhibitor therapy were excluded. Results from computed tomography (CT) scans were chosen over ultrasounds when made on the same day. The 37 cases that met these criteria and had one or more rAML measurements available are illustrated (Fig. 2). Of the 37 cases, seven cases had more than one measurement available, none of which involved a patient with a known TSC diagnosis. The rAML size increased in three cases, decreased in one case, stayed consistent in one case and was variable over time in two cases. From the 37 cases, case number 25 (Fig. 2) depicts a rAML measurement of 3.0 cm before pregnancy. Lopater et al. [25] described the formation of a tumorthrombus in this case. The other 97.3% had a rAML size of 4.0 cm or lager. Renal AMLs of women who suffered a haemorrhage were significantly larger (12.1 ± 4.6 cm) than rAMLs of women who did not suffer a haemorrhage (8.3 ± 3.2 cm; p = 0.018). For patients with TSC, rAML sizes were 12 cm [20], 21 cm [17] and 21.70 cm [19] in patients with haemorrhage and 9 cm [23] and 14 cm [22] without haemorrhage.

Overview of all rAML measurements before, during and after pregnancy

Only measurements that met the reliability criteria described in the results section were used. We placed the rAML measurement in chronological order on the x-axis when there was more than one measurement available in one-time frame. Location of a single rAML measurement was placed randomly in the time frame. This allowed us to make each rAML size visible, as there was overlap between rAML sizes among different cases. In case rAML was bilaterally we chose the value that had a follow up rAML measurement. When this was not available, we chose the largest rAML measurement. Case numbers 1–5 display patients in which TSC has been established.

In Fig. 3, we illustrate the differences in complication rates between patients with and without TSC diagnosis in the 48 cases. Complications (defined as haemorrhage, rapid rAML growth, or tumorthrombus) occurred in 84.7% of patients without TSC diagnosis and in 55.5% of patients with established TSC (p = 0.074). Haemorrhage occurred in 44.4% (n = 4) of the patients with established TSC compared to the 74.4% (n = 29) in the group without diagnosis of TSC (p = 0.115). Three patients developed a tumorthrombus, none were diagnosed with TSC (two sporadic rAML, one TSC diagnosis not mentioned in article). Though being a benign entity, rAML can be locoregionally aggressive and show vascular invasion resulting in tumorthrombus formation. This is a rare but known complication of rAML, occurring mostly in large rAMLs (> 4 cm) [26]. The tumorthrombus can grow and extend into the renal vein, possibly all the way to the inferior vena cava (IVC) and right atrium. This is associated with increased risk of (large) pulmonary embolism and also potentially results in IVC obstruction, which can be life threatening conditions for both mother and fetus in pregnancy. Imaging is important for identification of the rAML, determination of the extent of tumorthrombus formation and pre-operative planning, with magnetic resonance imaging (MRI) being the modality of choice during pregnancy. All three patients with tumorthrombi included in this review were treated surgically during pregnancy with either nephrectomy or thrombectomy.

Renal complication occurrences, described in the included case reports, in percentages (%) in patients with (n = 9) and without (n = 39) an established TSC diagnosis. No patients with TSC developed a tumorthrombus. Rapid rAML growth was only noted when explicitly stated in the article

We used the study from Mitchell et al. [27] only for analysis of complication risk since data about rAML growth was not provided. The quality assessment score of this study using the NOS was three out of nine, meaning a high risk of bias when translated to the Agency for Health Research and Quality (AHRQ) standards. The study consisted of 145 patients with TSC, 115 with a pregnancy history and 30 without a pregnancy history. This single center study was retrospective and used self-reported data from surveys (response rate was 30%) and data from electronical charts. The average age at which information was obtained was 41.1 (SD = 11.1) years old in the pregnant group and 35.1 (SD = 9.3) years old in the group who had never been pregnant. Renal involvement (rAML or renal cysts) was around 70% in both groups (pregnant and non-pregnant TSC women). Severity or type of renal involvement was not described. Of the women with renal involvement (18 in the non-pregnant group and 67 in the pregnant group) an analysis showed that renal complications were similar in the non-pregnant group, 67%, in comparison to the pregnant group, 57% (p = 0.62). Renal complications were defined in this study as either rAML-related hypertension, pain, rupture, haemorrhage, renal failure or rAML-related treatments. Haemorrhage occurred in 30% (n = 20) of the pregnant group and in 11% (n = 2) of the never-pregnant group (significance not reported). Explanation for omitting the p-value could not be found. The rates of haemorrhage in the pregnant group only includes patients with haemorrhage that occurred during or after pregnancy. During pregnancy there were eight cases of haemorrhage which occurred during pregnancy or within five weeks of post-partum. Overall, Mitchell et al. [27] concluded that “pregnancy did not appear to increase either the prevalence of renal involvement or the risk of a renal complication in the women we studied”.