Introduction

Prolactinomas are the most common tumors among functional pituitary adenomas and they constitute 40 % of all pituitary tumors [1]. Galactorrhea, gonadal dysfunction, and the tumoral mass effect are the main symptoms and findings [2]. The primary treatment of prolactinoma is medical treatment with dopamine agonists (DA), whereas surgery and radiotherapy (RT) have a limited role in the treatment of prolactinomas. Commonly used ergo-derived dopamine agonists are bromocriptine (BRC) and cabergoline (CAB). In most cases, prolactin (PRL) levels return to normal, gonadal dysfunction and infertility recover, and significant shrinkage of tumors are seen with DA treatment. The treatment success rates are 70–80 % for BRC and 80–90 % for CAB [3]. It is well known that prolactinomas respond very well to DA treatment, but the optimal duration of DA treatment is not clear. Despite the need for long-term therapy, withdrawal of treatment should be evaluated because of the adverse effects of medical treatment such as cardiac valve fibrosis and treatment cost. The remission rates after withdrawal of BRC are reported as 3–44 %. Although the results were lower than expected, the recurrences presented with hyperprolactinemia and only <10 % showed increase in tumor diameter [4–13]. The remission rates after withdrawal of CAB were expected to be better than BRC withdrawal because CAB is a more effective agent, but the results of early studies did not support this hypothesis (0–31 %) [14–18]. However, subsequent studies showed better remission rates for CAB. Especially in the study of Colao et al. [19], which was the first prospective study, remission rates for CAB were reported as 70 and 64 % for microprolactinomas and macroprolactinomas, respectively. An extension of this study to 8 years documented remission rates of 66 and 47 %, respectively [20].

The aim of our study was to assess the effect of DA therapy (BRC or CAB) withdrawal and the current recurrence rate of hyperprolactinemia, and possible factors that predict recurrence in patients with prolactinoma.

Patients and methods

This was a retrospective observational study. We identified all patients with a confirmed diagnosis of prolactinoma who had attended the pituitary out-patient clinic of Istanbul Faculty of Medicine, Department of Internal Medicine, Division of Endocrinology and Metabolism, during the last 22 years. The diagnosis of prolactinoma was confirmed according to the typical clinical signs and symptoms, radiographic signs (pituitary adenoma confirmed by magnetic resonance imaging-MRI) and laboratory tests (high PRL levels in at least two different blood samples). Serum PRL levels were measured using an electrochemiluminescent immunoassay (ECLIA), with normal ranges for adult males and females of 4.04–15.2 ng/mL and 4.7–23.3 ng/mL, respectively.

We evaluated withdrawal of DA in 67 patients with prolactinoma who had received DA treatment for at least 2 years and had reached normal PRL levels and showed tumor disappearance or ≥50 % tumor shrinkage compared with baseline size. Patients with a history of surgery and/or RT, and also patients whose DA treatment was stopped because of pregnancy or menopause were excluded from the study. Treatment withdrawal was not performed in patients whose tumors were in close proximity to the optic chiasm, cavernous sinus, or any other critical area. The patients of the study were divided into two groups as remission and recurrence, and these two groups were compared to detect factors that predicted recurrence. Also the subgroups of microprolactinoma and macroprolactinoma were evaluated to detect factors that predicted recurrence. Age at diagnosis, sex, maximum tumor diameter, baseline prolactin levels, duration of DA treatment, nadir PRL levels, residual tumor diameter, percentage reduction in levels of PRL, percentage of tumor shrinkage, presence of tumor disappearance before DA withdrawal, and maximum-minimum and cumulative doses of DAs before DA withdrawal were compared in the groups. At least 6 months of follow-up after DA withdrawal was accepted as a required criteria for confirmation of remission. Recurrence was accepted to be an increase in PRL levels above the upper normal levels after cessation of medical treatment. These patients were re-evaluated for the presence of galactorrhea and gonadal dysfunction and underwent renewed pituitary imaging with MRI.

Statistical analyses were performed using SPSS version 21.0. Categorical variables were defined by frequency and percentage rate, and numeric variables by mean ± standard deviation (SD). In dual independent group comparisons, the Student’s t test was used for normally distributed numeric variables and the Mann–Whitney U test for non-normally distributed data. Categorical variables were compared using the Chi square test. A p value of <0.05 was accepted as statistically significant.

Results

Sixty-seven patients (50 female and 17 male) aged 34.3 ± 11.2 years at diagnosis (range 17–66 years) were included in the study. The mean follow-up time was 108.8 ± 55.1 months (range 27–257 months). At initial assessment, the mean size of adenomas was 15.8 ± 11.1 mm (range 5–58 mm) and mean PRL levels were 1493.6 ± 3253.4 ng/dL (range 101–18,500 ng/dL). Twenty-three (34 %) prolactinomas were microadenomas and 44 (66 %) were macroadenomas. Nineteen (43 %) of the 44 macroadenomas presented with extrasellar extension [Class III (n = 14) and IV (n = 5) tumors according to the Hardy classification] [21].

Galactorrhoea (n = 45, 90 %), primary amenorrhea (n = 2, 4 %), secondary amenorrhea (n = 43, 86 %), infertility (n = 8, 16 %), headache (n = 15, 30 %), and visual field defect (n = 6, 12 %) were the symptoms and signs at presentation in the female patients. Impotence and decreased libido (n = 15, 88 %), infertility (n = 2, 12 %), headache (n = 14, 82 %), and visual field defects (n = 13, 76 %) were the symptoms and signs at presentation in the male patients. At the time of presentation, 62 (92 %) patients had central hypogonadism, 13 (19 %) had central hypothyroidism, and 9 (13 %) patients had central hypocortisolism.

Thirty-six of the 67 patients showed recurrence after DA withdrawal and the remission rate was 46 % in the whole group; the remission ratios were 65 % in microprolactinomas and 36 % in macroprolactinomas. The remission rates were 39 % in the BRC withdrawal group (15 of 38 patients) and 55 % in the CAB withdrawal group (16 of 29 patients). The age at diagnosis and sex were not statistically different between the remission and recurrence groups (p = 0.068, p = 0.106, respectively). The maximum tumor diameter and baseline PRL levels were significantly higher in the recurrence group (p = 0.001, p = 0.003, respectively). The nadir PRL levels, percentage reduction in levels of PRL, residual tumor diameter, and percentage of tumor shrinkage before DA withdrawal were not statistically different between the groups (p = 0.064, p = 0.930, p = 0.286, p = 0.959, respectively). Seventeen patients had total tumor disappearance before DA withdrawal, but this was not a predictor of remission (p = 0.882). The mean duration of DA therapy was significantly longer in the remission group than in the recurrence group (88.7 ± 48.1 months; range 29–192 months; and 66.7 ± 30.4 months; range 27–138 months, respectively p = 0.026). The maximum and minimum weekly and cumulative doses of CAB before DA withdrawal were not different in patients with recurrence and remission in the CAB withdrawal group (p = 0.928, p = 0.167, p = 0.444, respectively). The maximum and minimum daily and cumulative doses of BRC before DA withdrawal were not different in patients with recurrence and remission in the BRC withdrawal group (p = 0.337, p = 0.079, p = 0.418, respectively).

The mean duration of follow-up after DA withdrawal was 28.6 ± 19.2 months (range 7–76 months) in the remission group. The mean time to recurrence was 5.3 ± 3.2 months (range 2–17 months) in the recurrence group. The mean PRL levels were 57.5 ± 26.4 ng/mL (range 31–145 ng/mL) at the time of the recurrence and these levels were lower than baseline PRL levels (p = 0.001). Of the 24 female patients with recurrence, 18 reported oligo-amenorrhoea and nine had re-appearance of bilateral galactorrhoea. Of the 12 male patients with recurrence, ten reported impotance and decreased libido, and six also had decreased testosterone levels. Among the patients who had recurrence (n = 36), none had evidence of tumor enlargement or regrowth on MRI.

Data of the remission and recurrence groups for the whole group are shown in Table 1.

Table 1 Data of the remission and recurrence groups for the whole group
Full size table

We also evaluated patients with microprolactinoma (n = 23) and macroprolactinoma (n = 44) separately. Remission rates were 56 % in the BRC withdrawal group (9 of 16 patients) and 86 % in the CAB withdrawal group (6 of 7 patients) of microprolactinomas. In the microprolactinoma subgroup, patients with recurrence were younger (p = 0.033). Similar to the whole group maximum tumor diameter and baseline PRL levels were significantly higher in the recurrence group (p = 0.015, p = 0.048, respectively) of microprolactinomas. The nadir PRL level in the recurrence group of microprolactinomas was significantly high (p = 0.008). The mean duration of follow-up after DA withdrawal was 30.1 ± 19.8 months (range 7–76 months) in the remission group, and the mean time to recurrence was 5.1 ± 1.6 months (range 3–8 months) in the recurrence group of microprolactinomas. The mean PRL levels were 51.5 ± 13.2 ng/mL (range 37–73 ng/mL) at the time of recurrence in microprolactinomas. The data of the microprolactinoma subgroup are shown in Table 2.

Table 2 The data of the macroprolactinoma and microprolactinoma subgroups
Full size table

Remission rates were 27 % in the BRC withdrawal group (6 of 22 patients) and 45 % in the CAB withdrawal group (10 of 22 patients) of macroprolactinomas. In the macroprolactinoma subgroup, the only significant difference between the remission and recurrence groups was the long duration of DA therapy before withdrawal in the remission group (p = 0.018).The mean duration of follow-up after DA withdrawal was 25.9 ± 14.6 months (range 8–50 months) in the remission group, and the mean time to recurrence was 5.4 ± 3.6 months (range 2–17 months) in the recurrence group of macroprolactinomas. The mean PRL levels were 59.3 ± 29.1 ng/mL (range 31–145 ng/mL) at the time of recurrence in macroprolactinomas. The data of the macroprolactinoma subgroup are also shown in Table 2.

Discussion

In this study, we have shown a higher remission rate in patients with prolactinoma after DA withdrawal. The most likely reason for this was thought to be the strict selection criteria of patients for DA withdrawal. These criteria were the presence of significant shrinkage of tumor, gradual tapering of DA dose and a very low maintenance dose of DA at the time of DA withdrawal, and at least 2 years of treatment duration. No randomized controlled studies have compared different withdrawal strategies after successful treatment of prolactinoma. However, some criteria have been determined in current guidelines. The Pituitary Society (2006) guideline suggested that patients who had had normoprolactinemia for at least 3 years and markedly reduced tumor size might safely have their DA doses tapered and treatment withdrawn [22]. Also in 2011, the guideline provided by the Endocrine Society recommended that patients who had had normoprolactinemia for at least 2 years and who had no visible tumor remnant could be candidates for DA withdrawal [23]. Recent studies in which patients with certain selection criteria were included showed higher remission rates in contrast to earlier studies, which were also supported by current meta-analysis [24].

We showed higher remission rates in patients with microprolactinoma than in those with macroprolactinoma (65 vs. 36 %). Associated with this the maximum tumor diameter and baseline PRL level also an indicator of tumor size were significantly higher in the recurrence group. This finding was compatible with the results of other studies [9, 19, 25]. Contrary to the results of our study, some studies found no relation of remission with baseline PRL level and/or maximum tumor diameter [8, 24, 26–28].

CAB is the most effective agent among the DAs with regards normalization of PRL levels, recovery of gonadal dysfunction, and reduction in tumor size [23]. As expected owing to its effectiveness, the remission rates after CAB withdrawal was higher than with the other DA [28, 29]. However, no prospective studies have compared the different DAs with each other, and even retrospective studies showed no difference in remission rates between CAB and BRC [9, 25, 27]. In our study, the remission rate was significantly higher in the CAB group than in the BRC group (55 vs. 39 %). In a recent meta-analysis that evaluated CAB withdrawal strategies, the results of 637 patients from 11 studies were analyzed and the recurrence rate of hyperprolactinemia was found as 65 % [24].

Possible predictive factors of recurrence or remission have been evaluated in the literature. In some studies, as in our study, baseline tumor size was found to be a factor related to recurrence [19, 25]. In two prospective studies, residual tumor size and nadir PRL levels before DA withdrawal were found to predict recurrence; Colao et al. [20] determined cut-off levels for nadir PRL levels and residual tumor size (5.4 µg/L and 3.1 mm, respectively), whereas Kharlip et al. [26] reported that for each additional millimeter reduction in residual tumor size recurrence decreased by 18 %. Barber et al. [25] also found that the nadir PRL level to be the most likely to predict recurrence in their subgroup of microprolactinomas. Similar to this finding, the nadir PRL level was significantly high in the recurrence group of microprolactinomas in our study. Studies have shown that tumor disappearance increases the rate of remission [27, 28]. However, conflicting studies found no effect of tumor disappearance on remission, including our study [9, 26].

Most of the studies in the literature have shown no effect of age and sex on recurrence or remission [8, 9, 27, 28, 30, 31], exceptions include one study that reported that recurrence was higher in male patients [20] and another found that remission rates were higher in young patients [32]. But in our study, patients with microprolactinoma who recurred were significantly younger.

In our study, the duration of DA therapy was longer in the remission group than in the recurrence group; similar results have been reported in the literature [20, 27]. Moreover, in the meta-analysis of Dekkers et al. [29], duration of treatment longer than 24 months was defined to be a factor that increased remission rates. In contrast, another more recent meta-analysis reported that there was no relation of treatment duration with remission rates [24]. It is well known that short-term use of DA causes a reversible decrease in organelles that synthesize PRL, but long-term DA treatment leads to perivascular fibrosis, cytolytic effects, and causes acellular spaces in tumor tissue [10, 33–35]. In view of these reports, we think that duration of treatment may be an effective factor in prevention of recurrence. Despite this, the mean durations of treatment in both the remission and recurrence groups of our study were not much different and as such we could not give a cut-off level for treatment duration.

Tapering of DA dose is one of the criteria of remission determined in guidelines [22, 23]. Colao et al. [20] reported that a higher CAB dose before DA withdrawal was a predictor of recurrence. Furthermore, a meta-analysis showed that gradual tapering of CAB dose and withdrawal of CAB at the lowest dose increased the likelihood of remission [24]. Although the dose was tapered to the lowest dose before withdrawal in the present study, this situation did not cause a difference between groups. We also included an evaluation of the effect of cumulative doses, which has not been previously mentioned. We found the cumulative doses of CAB and BRC before DA withdrawal were higher in the recurrence groups, but it did not reach statistical significance.

Unfortunately, there are also studies showing that none of the possible factors were related to recurrence or remission [8, 31]. Despite all these data, it is still not clear why some prolactinomas recur, just as it is not yet clear why some are aggressive and extensively invasive. This is most probably related to the expression of some tumor associated genes [36]. In another study from our center that included most of the patients in this study, aryl hydrocarbon receptor interacting protein (AIP) gene mutation, which is known to be associated with treatment failure, was not detected in patients with prolactinoma [37]. This could be an explanation for the high remission rate in our study. Medical treatment is the primary choice in the management of prolactinoma; however, surgery is also performed in selected patients because of drug resistance and/or tumor aggressiveness. According to the histopathologic findings obtained from specimens of these few patients, markers such as Ki-67, p53, and proliferating cell nuclear antigen (PCNA) were reported to be associated with recurrence [38]. However, because of the limited number of surgery patients, no studies have compared the histopathologic and genetic findings of these patients with recurrence and remission.

Recurrence has been observed in the first year of follow-up after DA withdrawal, especially in the first 6 months in most cases [9, 19, 25–28]. Similar to the results reported in the literature, the mean recurrence time was 5.3 months in our study. The mean follow-up time of our patients who were accepted to be in remission was about 28 months, which was sufficient to detect early recurrence; however, late recurrence may be seen beyond this time, which is a limitation of our study. The PRL levels at the time of recurrence were lower than the baseline PRL levels, which is in line with the literature [9, 19, 25–27]. None of our patients showed tumoral regrowth with recurrence. These results are also consistent with the literature [19, 25–28]. Only in old studies was tumor regrowth at recurrence after DA withdrawal reported [5, 10]. However, prolactinomas were more susceptible to regrowth after a short course of treatment, which was considered as being <1 year [39–41].

As a result, although the factors that predict remission reported in the literature are variable, patients who have consistently normal PRL levels and no symptoms with lowest dose of DA, significant reduction in tumor size (≥50 %), and received DA for a long time (≥2 years), DA can be withdrawn. However, patients should be checked at frequent intervals during the first year. To determine the optimal DA withdrawal strategy and the most suitable time of treatment cessation, large prospective studies are needed to compare withdrawal strategies in patients selected according to different criteria and treated for different periods of time beyond 2 years.