Prolactin-secreting tumors, dopamine agonists and pregnancy: a longitudinal experience of a tertiary neuroendocrine center

Prolactinomas represent the most common cause of hyperprolactinemia and account for 40% of all pituitary adenomas [1]. Prolactinoma has a prevalence of 60–100 cases/million, and it is more common in women, especially those aged 20 to 50 years [2]. In women, hyperprolactinemia is a frequent cause of anovulatory cycles and infertility, but fertility can often be restored by treatment with dopamine agonists (DAs) [3]. DA represents the first-line therapy since it regulates menses and can reduce pituitary lesion size, restoring normal PRL levels [4]. During DA therapy, 85% of women desire offspring. However, there are some significant concerns regarding the treatment of prolactinomas with DA during conception and pregnancy. The literature has extensively demonstrated that, compared with the general population, bromocriptine (BRM) does not increase the number of miscarriages, premature births, multiple pregnancies or congenital malformations [3, 5, 6]; moreover, available information on the safety of cabergoline (CAB) is reassuring, although specific data remain scant [3, 7‐9]. Despite the demonstration that DAs used previously and during pregnancy are safe drugs, the Endocrine Society recommends the immediate suspension of medical therapy once pregnancy is confirmed as a precaution [4].

The latest Pituitary Society guidelines also recommend withdrawal of DAs for patients with microprolactinomas and intrasellar macroprolactinomas as soon as pregnancy is confirmed. However, they allow for the possibility of continuing DA therapy in the case of expansive/invasive macroadenoma [10].

As regards the Italian regulation, cabergoline should even be interrupted once ovulatory cycles have been re-established, at least one month before the supposed conception, or immediately, at the time of ascertained pregnancy. The possibility of stopping treatment once pregnancy is established derives from a limited series of 329 pregnancies, suggesting that foetal exposure to cabergoline through early pregnancy does not induce any increase in the risk of miscarriage or foetal malformation [11].

Another relevant aspect is the possible impact of pregnancy and breastfeeding on the progression of prolactinoma. In fact, some studies have shown that 10–68% of women with prolactinoma or idiopathic hyperprolactinemia experience disease remission following pregnancy. Breastfeeding also does not seem to increase the risk of persistence or worsening of the disease [12‐15].

Based on these data, the purposes of the present study were 1) to evaluate the prevalence of short- and long-term maternal or foetal complications in patients with prolactinoma treated with CAB or BRM during pregnancy induction or during pregnancy itself, afferent to the Neuroendocrinology Centre of A.O.U. City of Health and Science of Turin; and 2) to evaluate the effect of pregnancy on the remission of hyperprolactinemia and on the progression of prolactinoma.

This was a retrospective single-center longitudinal observational study involving a cohort of women affected by prolactinoma who were followed up at the Neuroendocrinology Centre of University Hospital City of Health and Science of Turin (Italy) and who became pregnant during treatment with CAB or BRM. Women with hyperprolactinemia caused by causes other than prolactinoma were not included in the present study. All pregnancies that occurred from 1981 to 2012 were taken into consideration to allow a postpartum follow-up of at least ten years to assess any late side effects. Clinical and hormonal data are shown in Table 1.

The diagnosis of prolactinoma was made according to the criteria of the International Guidelines [4, 10], while disease remission was defined as persistence of biochemical and clinical normalization at least 1 year after therapy withdrawal.

For each patient, computerized medical records were analysed; data were also integrated through an outpatient visit or, in the absence of this possibility, by telephone interview. The salient clinical and pregnancy-related data (outcome, type of delivery, complications, breastfeeding), drug exposure (duration of exposure to CAB/BRM, starting dose, maximum dose) and development of the child (birth complications, weight at birth, APGAR score, malformations, developmental deficiencies) were evaluated. The data regarding the course of pregnancy, delivery, and the child's health conditions were collected through interviews with the study patients and not through direct access to the gynecological or pediatric clinical records (patient interview script available on Supplemental Material).

Data on PRL levels at diagnosis, the nadir of PRL levels before pregnancy and at the last follow-up were included as adenoma features at the same time. The study was conducted according to the principles of the Declaration of Helsinki. All patients signed an informed consent form for study participation. The study was approved by the Internal Ethical Committee of our institution.

Forty-three patients were studied (age at diagnosis 27 ± 6.0 years and age at pregnancy 32.4 ± 5.4 years) for a total of 58 pregnancies (31 patients underwent one pregnancy; two pregnancies were recorded in nine patients, and three were recorded in three patients). Among the 43 women studied, 10 (23.3%) had macroadenomas, and 33 (76.7%) had microadenomas. At diagnosis, the median tumor diameter was 6 mm (IQR 5–9 mm), and the median PRL was 112.3 ng/ml (IQR 50–149.7 ng/ml). None of the patients underwent pituitary neurosurgery (NS) or radiotherapy (RT) before pregnancy, while a single patient underwent NS during the postpregnancy follow-up (six years later). By the time of conception, 18 patients were receiving BRM therapy (dose 6.16 ± 6.6 mg/day; IQR 2.5–20 mg/day), and 40 were receiving CAB (dose 1.6 ± 3.3 mg/week, IQR 0.25–17.5 mg/week). The two populations analysed were comparable in terms of age at diagnosis, age at pregnancy, prevalence of macro/microadenoma, adenoma size at diagnosis and PRL levels at diagnosis (Table 1). The estimated cumulative dose of BRM taken during pregnancy was 274.8 ± 399 mg (IQR 2.5–1260 mg), the median foetal exposure time was 35 days (IQR 28–35 days), and the estimated cumulative dose of CAB was 4.6 ± 4.3 mg (IQR 1–18 mg), with a median foetal exposure time of 28 days (IQR 28–35 mg) (Table 1). Although all patients were instructed to interrupt DA therapy once pregnancy was confirmed, in two cases, the patients continued BRM therapy throughout the course of gestation (in one there was no medical indication; in the other, there was severe headache). The patients were followed up for at least 10 years, with a mean of 16.4 ± 8.4 years (CAB 12.7 ± 6.3 vs BRM 24.7 ± 6.3; p < 0.01).

Our study confirmed the safety of administering DA to women affected by PRL-secreting tumours during the first weeks of pregnancy for maternal and foetal outcomes and for pituitary adenoma. In fact, in our series, the incidence of abortion was lower than that in the general population (6–9%) [16] or compared to that in other studies conducted on a similar population (9.1%, 9.8%, 10.2%, respectively) [11, 17, 18]. Furthermore, no cases of malformations have been reported in children born to women treated with BRM or CAB, in agreement with the findings of previous studies [18]. Additionally, the percentage of preterm births was comparable to that of the general population (13.7% vs 12.7%), as was the percentage of low-birth-weight children (8.6% vs 8% in the general population) [19]. Additionally, in the case of a woman who was taking BRM during the whole pregnancy, no complications were detected.

No statistically significant differences were found in terms of obstetric or neonatal outcomes between CAB and BRM treatment. However, a higher rate of child complications (six speech disorders, one walking delay, two atopic dermatitis) in infants born was found in the CAB group. Specifically, among them, developmental disorders (walking and speech delay) constituted the majority. All were mild and transient, and they were not influenced by the mother's age, birth weight, or breastfeeding. Although language disorders did not significantly differ between the two groups, in our case study they occurred in 10.7% of the whole children population (all in the CAB group). Also Lebbe et al. investigated this aspect in a series of 100 pregnancies and found no difference in verbal fluency [15]. Furthermore, the reported prevalence of specific language impairments in the general population varies widely from 2 to 12% [20]. In our study, a 15% (6/40) prevalence was documented in the CAB group,10.7% when considering the whole population studied, which is consistent with the findings of the literature. Very little is known about the etiology, and multiple causes are probable. In her work, Diepeveen et al. concluded that children with a speech disorder tended to have lower APGAR scores 5 min after birth; in addition, the difference in APGAR scores was greater for females than for males [20]. In our study, no difference emerged in the APGAR score between the two groups.

Our data demonstrate the safety of DAs in pregnancy, in contrast with the data from the EFEMERIS pharmaco-epidemiological database study, which showed an increase in the rates of abortion and premature birth [21]. The data from this specific study were evaluated by the authors of the recent Pituitary Society guidelines [10]. In addition, most women in the exposed groups were treated with BRM and other DAs (quinagolide, lisuride, ropinirole, and piribedil) but not CAB. Conversely, a review of the literature for the European Society of Endocrinology Clinical Practice Guidelines, in accordance with our data, showed normal rates of miscarriage, preterm delivery, and congenital malformations with the use of CAB in prolactinoma [22].

The second endpoint of our study was to evaluate the impact of pregnancy and breastfeeding on pituitary tumours. Our results are less optimistic than previous ones. Indeed, the postpregnancy remission rate was 42.9%, which was significantly lower than the 68% reported by Auriemma et al. [14]. A longer medical treatment before pregnancy and the inclusion of women with functional and nontumour hyperprolactinemia are possible causes of this discrepancy. In fact, in our study, we included only women with pituitary lesions and excluded patients with idiopathic hyperprolactinemia. On the other hand, the remission rate reported in our study was higher than that reported in other previous works, in which remission was documented in 15 to 35% of hyperprolactinemic women [23‐25]. Notably, in our study, we found no significant differences between women taking BRM or CAB. A higher probability of disease remission is related to lower PRL nadir obtained before pregnancy, which probably implies that the response to DA therapy plays a greater role in disease remission than pregnancy. We did not observe any correlation between the size of the adenoma at diagnosis and postpregnancy remission, unlike what was described in the study by Domingue et al., which reported a twofold greater remission rate in patients with microadenomas than in patients with macroadenomas (46% vs 26%) [26]. Contrary to what was previously described in the literature, our data showed a negative correlation between the number of pregnancies and the probability of disease remission; in fact, the women with the highest number of pregnancies had a lower healing rate. However, other studies have shown a similar percentage of patients in remission (18–33%) after a first, second or third pregnancy [26].

Finally, in women with prolactinoma, breastfeeding is generally considered safe, but few studies are available in the literature. In our study, no significant difference in remission rate was shown between women who breastfed and those who did not. Furthermore, breastfeeding duration was not related to remission status. These results, in agreement with the literature data, confirm that breastfeeding is not a risk factor for worsening of the disease [15]. Therefore, breastfeeding should not be discouraged in these patients.

Finally, another limitation is the absence of a specific matched control group; therefore, the data concerning the occurrence of maternal-foetal and development outcomes were compared with those available for the general population.

Foetal exposure to DA during the first weeks of embryogenesis is not associated with an increased risk of abortion, malformation or childhood disease. In our case series, a greater presence of mild and transient language disorders by maternal report was observed in the cabergoline group. However, it should be noted that this prevalence was comparable to that observed in the general population. Finally, our data did not currently allow us to consider pregnancy and breastfeeding as predictive factors for persistence or recovery from the disease.