Introduction

Despite the fact that consensus guidelines recommend long-term dopamine agonist (DA) therapy as the first-line approach to the treatment of small prolactinomas, some patients continue to prefer a primary surgical approach because of concern over potential adverse effects of long-term medical therapy and/or the desire to avoid long-term medication [1]. Dopamine agonists (DA), especially cabergoline, is highly efficient for the treatment of small prolactinomas [2–5]. However, following the withdrawal of cabergoline treatment relapse of hyperprolactinemia has been shown in a number of studies [6–10]. Furthermore, relevant systemic side effects, including nausea, vomiting, headache, dizziness and vertigo have been reported in about one third of patients treated with DAs [1]. Recently, the increased risk of cardiac valve disease in patients treated for Parkinson’s disease with cabergoline has raised concerns about the safety of treatment with dopamine agonists in patients with prolactinoma [11–13].

In carefully selected patients with microprolactinomas and PRL levels less than 200 μg/l, long-term remission rates higher than 80% can be achieved with pituitary surgery [14–17]. After complete surgical resection, PRL levels return to normal, gonadal function is restored in both sexes, and the decrease in bone mineral density can be halted and reversed. A low postoperative PRL level has been shown to be the most reliable predictor of long term surgical cure [14]. The surgical techniques for pituitary surgery have been markedly improved over the last 20 years [18, 19]. Complications from pituitary surgery for small prolactinomas are infrequent in specialized pituitary centers [14–17]. This study was performed in order to assess surgical outcomes in patients with small prolactinomas who opted for primary pituitary surgery instead of long-term therapy with DA.

Patients and methods

This retrospective study included 34 patients who underwent primary pituitary surgery without prior DA treatment for small intrasellar prolactinomas at the Department of Neurosurgery, University of Bern, between January 2003 and December 2006. The indication for first-line pituitary surgery in this study was patient’s preference for surgery rather than long-term DA therapy for all patients. A microprolactinoma was defined as a tumor of 1–10 mm; a macroprolactinoma was a lesion of 10–20 mm in diameter. Preoperative data included history and physical examination, PRL levels, and type and results of imaging studies. Other hormonal levels were not systematically evaluated in all patients and were hence not part of the analysis. The highest documented preoperative value was used as the preoperative serum level of PRL. Neuroradiological images were reviewed to exclude extrasellar extension and/or invasion of the cavernous sinuses [20]. All operations were performed by the single neurosurgeon (R.W.S.), using the transseptal, transsphenoidal microsurgical approach while the patients were in a semi sitting position [19, 21–23]. After a medial nasal incision, the septal mucosa was unilaterally detached from the cartilaginous and osseous nasal septae, respectively. A nasal speculum was inserted to keep the mucosal tunnel open. At this stage, the operating microscope was introduced and the x-ray image intensifier, also called fluoroscopy system, served for navigation and intra-operative imaging. Using the operating microscope, the vomer was exposed and opened with a drill. The septations of the sphenoid sinus were resected. Through the sphenoid sinus, i.e. transsphenoidally, the sellar floor was removed and the basal dura of the pituitary fossa was incised. Variously shaped cuvettes and micro forceps were applied to loosen and resect the intrasellar micro- or macroprolactinoma. The normal anterior pituitary gland was characterized by its yellowish color, firmer consistency and vascular surface structure and was preserved. After the microsurgical removal of a micro- or macroprolactinoma, the cavity was filled with gelatin foam soaked in fibrin glue, and the sellar opening was covered with a vicryl patch fixed with the same material. A second patch was fixed to the opening in the sella and held in place with gelatin foam and glue. Cartilage or bone was not used for reconstruction, because the package is held in place and sealed by fibrin glue. The duration of a routine pituitary operation was between 60 and 120 min. Antibiotic treatment was started prophylactically before the operation and administered for 24 h. Body weight, fluid intake and output, serum electrolytes, serum and urine osmolarity were monitored daily after surgery. Nasal tampons were removed on the 2nd or 3rd postoperative day, and the patients were discharged 3–5 days after the operation. Histopathological diagnosis was based on conventional light microscopy and specific immunohistochemistry detecting prolactin in accordance with the WHO classification of tumours of the endocrine system [24, 25]. The postoperative PRL level was determined on the morning following the surgery. Remission was defined as a normalized PRL level below 20 μg/l on the morning after surgery. A postoperative clinical, neuroradiological and biochemical evaluation was performed at 6 weeks by the operating neurosurgeon. Long-term follow-up PRL levels which were performed at the discretion of the referring physician and clinical symptoms of hyperprolactinemia were obtained by the patient’s referring physician. Recurrence was defined as hyperprolactinemia (>20 μg/l) irrespective of the presence of symptoms and neuroradiological findings. PRL measurements were performed at different laboratories in Switzerland. Assay kits for ELISA or EIA differed between referral sources. At our institution, an electrochemiluminescence sandwich-immunoassay (Roche Modular E170, Roche Diagnostics, Mannheim, Germany) was used. The upper limit of normal for PRL was defined as 20 μg/l. Preoperative and postoperative PRL measurements were made with the same assay kit during the study period. Data was entered into a computerized database, after which they were analyzed using SPSS 12.0 (SPSS, Chicago, IL, USA). If uniformly distributed, the data was subjected to Student’s t test, otherwise a Mann–Whitney test was used. The cut-off level for significance was set at P < 0.05. Data are presented as medians, including the range and percentage, unless otherwise specified.

Results

Clinical characteristics

Thirty four consecutive patients (30 female, 4 male) underwent transsphenoidal microscopic removal of a small prolactinoma at the Department of Neurosurgery (Table 1). Mean age at surgery was 30.4 years for women, 38.5 years for men. On the basis of magnetic resonance imaging (MRI) findings, 24 patients (70%) had a microprolactinoma, and the remaining 10 (30%) had a macroprolactinoma. At the time of diagnosis, 31 of 34 patients (91%) had hypogonadism. Twenty eight out of 30 female patients (93%) showed amenorrhea and/or galactorrhea. Three out of 4 male patients (75%) complained about decreased libido and/or impotence. Two female patients (7%) and one male patient (25%) were asymptomatic and the adenomas were discovered by head MRI for other medical reasons. The median preoperative PRL concentration in the total cohort was 203.4 μg/l (range 40–1,046 μg/l). Patients with microprolactinomas had significantly lower preoperative PRL levels compared to patients with macroprolactinomas (106.5 μg/l vs. 273 μg/l; P = 0.0004, Fig. 1). Eleven of 24 patients (45%) with microprolactinomas and 2 of 10 patients (20%) with macroprolactinomas had preoperative PRL levels <100 μg/l.

Table 1 Characteristics of patients with micro- and macroprolactinomas
Full size table
Fig. 1
figure 1

Box plots (a and b) show preoperative and postoperative PRL levels in the serum for patients with micro- (n = 24) and macroprolactinomas (n = 10). The lower and upper bars in the box plots delineate the lowest and highest measured PRL level for both groups, respectively. The box contains all PRL levels from the 25th to 75th percentile. The median PRL level is shown as a line in the box. The P-values for the preoperative (a) and postoperative PRL levels (b) between the two groups were *P = 0.0004 and P = 0.88, respectively

Full size image

Early postoperative results

Histologically all tumours were classified as prolactin cell adenomas (WHO grade I) of the sparsely granulated type. There was no evidence of invasive growth in those specimens which included fragments of dura mater. The median postoperative PRL concentration among all patients was 3.45 μg/l (range 0.5–23.4 μg/l). This value was significantly lower than the preoperative one of 203.4 μg/l (range 40–1,046 μg/l). Remission, defined according to the criteria outlined in Patients and Methods was achieved in 32 (94%) out of 34 patients. All 10 patients with macroprolactinomas achieved remission. Patients with microprolactinomas did not have significantly lower postoperative PRL levels compared to patients with macroprolactinomas (2.8 μg/l vs. 7 μg/l; P = 0.56, Fig. 1). No operative or perioperative complications were recorded. The postoperative MRI imaging after 6 weeks showed no tumour remnants in any patients. Postoperatively, one patient had transient SIADH which necessitated rehospitalization (Table 2). One patient underwent lysis of postoperative nasal synechia. There was no mortality and no major morbidities such as new endocrinological deficits, visual loss, vascular injury or meningitis/abscess apart from one patient suffering from postoperative hypogonadotropic hypogonadism despite postoperative normal PRL levels.

Table 2 Neurosurgical outcomes for patients with micro- and macroprolactinomas
Full size table

Long-term outcome

Most patients experienced significant improvement of their symptoms. Twenty seven of the 30 (90%) premenopausal women with preoperative amenorrhea resumed normal and regular menses after surgery. Two of three premenopausal women who did not resume normal menses had a recurrence of mildly elevated PRL levels (26.8 μg/l, 46.8 μg/l) during the follow-up. One premenopausal woman is suffering from a persistent hypogonadotropic hypogonadism despite a normal prolactin level which accounts for a new pituitary deficiency. In men, restoration of normal sexual function was reported in two out of three cases (66%). Apart from one patient with hypogonadotropic hypogonadism, no additional patient was diagnosed with a new pituitary deficiency at postoperative or long-term evaluation. The median long-term PRL concentration in all patients was 10.95 μg/l (range 1.6–60.1 μg/l). Patients with microprolactinomas and those with macroprolactinomas had similar median long-term PRL levels (10.75 μg/l vs. 10.95 μg/l, P = 0.36, Fig. 2). The median follow-up time period was 33.5 months for all patients (range 6–77 months). Two patients (8%) with microprolactinomas did not achieve long-term PRL levels below 20 μg/l (26.8, 38.7 μg/l). The one patient with a long-term PRL level of 26.8 μg/l had a preoperative PRL level that was above 400 μg/l. For the group of 10 patients with macroprolactinomas, 2 patients (20%) had long-term PRL levels of 46.8 μg/l and 60.1 μg/l. Those two patients had preoperative PRL levels above 200 μg/l, i.e. 360 and 466 μg/l, respectively. Summarizing, three out of four patients with recurring hyperprolactinemia had preoperative PRL levels >200 μg/l and were medically re-evaluated. Thirty patients with small sized prolactinomas (88%) achieved long-term remission through surgical resection. Twenty two of 24 patients (91%) with microprolactinomas and 8 of 10 patients (80%) with macroprolactinomas are in long-term remission by pituitary surgery.

Fig. 2
figure 2

Box-plots for long-term PRL levels in the serum for patients with micro- (n = 24) and macroprolactinomas (n = 10) were obtained by the patient’s referring physician after a median follow-up period of 33.5 months. The lower and upper bars show the lowest and highest measured long-term PRL levels for both groups, respectively. The box contains all long-term PRL levels from the 25th to 75th percentile range. The line in the box represents the median. The P-value for the long-term PRL levels between the two groups was 0.11

Full size image

Discussion

The current management guidelines for prolactinomas [2] accept pituitary surgery as a second line treatment in non-responders to medical therapy or in those suffering intolerance to or complications of DA. However, some patients continue to prefer a primary surgical approach because of concern over potential adverse effects of long-term medical therapy and/or the desire to avoid long-term medication use. At our institution, patients can opt for surgical resection of a small prolactinoma versus long-term DA treatment. Our overall remission rates of 88% for small prolactinomas are similar to previously reported surgical remission rates of >90 and 87.2% for microprolactinomas [14, 15] and compare favourably with the best medical remission rates after withdrawal of cabergoline in microprolactinomas [9]. Specific selection criteria for primary pituitary surgery have not been defined. A preoperative PRL level <200 μg/l was a strong predictor of a surgically achieved long-term remission of hyperprolactinemia in our study. Patients with microprolactinomas and lower preoperative PRL levels than patients with macroprolactinomas (109 μg/l vs. 340 μg/l) had better remission rates (91% vs. 80%). A correlation of surgical high remission rate with preoperative prolactin levels was previously demonstrated in a study by Tyrell et al. 92% of the patients with preoperative PRL levels of <100 μg/l and 75% of the patients with preoperative PRL levels between 101 and 200 μg/l experienced initial remission. However, only 37% of patients with preoperative PRL levels above >200 μg/l achieved successful surgical outcomes [17]. It remains unknown, why higher PRL levels appear to be a risk factor for surgical failure for small prolactinomas [3]. Another predictor of long term surgical cure was the postoperative PRL level as demonstrated in a previous study of Amar et al. [14]. Consistent with prior findings, we found that all 22 patients (100%) with microprolactinomas and 7 out of 8 patients (87%) with macroprolactinomas and postoperative PRL levels lower than 10 μg/l achieved long-term surgical remission. The difference in neurosurgical long-term remission rates between micro- and macroprolactinomas (91% vs. 80%) might suggest that complete tumour removal is more difficult to achieve with macro- than microprolactinomas.

Complications from pituitary surgery for microadenomas are uncommon. The mortality rate is typically reported as less than 0.5%, while the rate of major morbidity has been cited 3.4% and includes most frequently transient diabetes insipidus, cerebrospinal fluid rhinorrhea and rarely oculomotor palsy, visual loss, stroke/vascular injury and meningitis/abscess [26, 27]. In our cohort, one patient had transient SIADH and another patient had to undergo lysis of postoperative nasal synechia. One patient was diagnosed with a new pituitary deficiency, in this case with hypogonadotropic hypogonadism after surgery which is probably explained by resecting a critical amount of pituitary gonadotropic cells surrounding the small prolactinoma. The endonasal endoscopic approach has similar complication rates compared to the traditional microsurgical technique, but less nasal traumatism, no nasal packing, less post-operative pain and usually quicker recovery have been reported for endoscopic pituitary surgery [28–30]. High surgical long-term remission rates with very low major morbidity suggest that pituitary surgery, using the traditional operating microscope in this study, can be considered an alternative and safe option in the treatment of small prolactinomas.

The surgical techniques for pituitary surgery have been refined over the last 20 years and have resulted in reductions in complication rates, particularly if the surgeon and his or her institution have sufficient experience in the field [17, 31, 32]. Many studies reporting lower surgical remission rates for prolactinomas were published more than 10 years ago. An extensive table on surgical outcomes is provided in Gillam’s et al. [49] review. Recent surgical studies with similar remission and complication rates to ours were published over the last 15 years and probably demonstrate a refinement of the surgical techniques [15–17]. Recently, no study has been published including only patients preferably opting for primary surgical removal of their small prolactinoma as a first-line therapy. In the study by Kreutzer, there were 23 patients out of 212 patients analyzed who elected on a primary surgical approach. However, those patients were harboring different sizes of prolactinomas, encompassing microprolactinoma, macroprolactinoma and also giant prolactinomas and were not separately discussed in the mentioned study [15]. The study by Losa et al. also reports 21 patients (35.6%) with microprolactinomas and 14 patients (22.9%) with macroprolactinomas who opted for a primary surgical approach. Data for patients with small macroprolactinoma (11–20 mm) have not been published separately so far. Typically, those patients are included into the category of macroprolactinomas which have a much lower reported initial remission rate of 35% compared to our initial remission rate of 100% in patients with macroprolactinomas [3]. To further determine the optimal management of small prolactinomas, a large number of patients with a longer follow up, preferentially in the setting of a randomized study, need to be studied.

The recurrence rate of 8.9% in this study for patients with microprolactinomas is comparable with the recurrence rate of 5.6% for microprolactinoma reported in the study by Kreutzer et al. [15]. Losa et al. reported a recurrence rate of 16.9% in their study which was similar to most other recent surgical series [17, 32–34]. Given the median follow-up period of 33.5 months in this study, the possibility of a later recurrence of hyperprolactinemia needs to be reassessed. As shown in Fig. 3, Panel A and B the long-term PRL levels in selected patients might rise over 20 ng/μl over a longer follow-up period. The relapse of hyperprolactinema varied widely among different surgical series. In some studies almost all relapses occurred within the first 3 years after surgery [35–37], whereas in other studies the median time to relapse was longer than 5 years [17, 38]. The median follow-up period of 33.5 months (range 6–77 months) in this study is therefore an important limitation, especially for patients in the shorter range of the follow-up period. The relapse of hyperprolactinemia might indicate microscopic remnants after pituitary surgery which might grow back to a tumor entity over an extended period of time. Those microscopic remnants might not be detected by postoperative MRI imaging and regular PRL levels assays over a longer-follow up period, because most small prolactinomas have very low proliferative cell indexes [39]. The study by Delgrange et al. reports only occasional Ki-67- or PCNA-positive nuclei (<1%) in a series of 14 female microprolactinomas. In the cohort of Thomson’s et al. [38] who reported 44 patients with confirmed microprolactinoma, 8 patients (18.2%) experienced recurrent hyperprolactinemia at 2–10 years, but the elevation was permanent in only two patients. Thomson et al. concluded that the recurrence of hyperprolactinemia after pituitary surgery for microprolactinomas is not necessarily a permanent feature and does not inevitably indicate operative failure. However, we would suggest DA treatment for patients with recurrent hyperprolactinemia after a primary surgical approach if the hyperprolactinemia persists over 2 months.

Fig. 3
figure 3

Postoperative and long-term PRL levels in the serum for individual patients with small prolactinomas. a Twenty four individual postoperative and long-term PRL levels for patients with microprolactinomas. The values are shown by a black dots connected by a line. b Postoperative and long-term PRL values connected by a line for all 10 individual patients with microprolactinomas

Full size image

Dopamine agonists have been efficient overall in the treatment of hyperprolactinemia. Cabergoline, the most effective compound, is a D2 selective agonist and induces fewer side effects than bromocriptine [40, 41]. In a high percentage of patients, PRL normalization and tumor reduction have been shown upon treatment with cabergoline in several prospective studies [7–9, 42]. For cabergoline, the most common side effects are nausea or vomiting (~35%), followed by headache (~30%) and dizziness or vertigo (~25%) [1]. After withdrawal of cabergoline, 30% of carefully selected patients had recurrent hyperprolactinemia in a study reported by Colao et al. [9]. Patients needed to meet following criteria for discontinuation of cabergoline: normal serum prolactin levels, no visible tumor on MRI scan, or a 50% decrease or more from baseline with a tumor mass at least 5 mm from the optic chiasm and no evidence of cavernous sinus invasion. In the study by Kharlip et al. [10] more than half of the patients (54%) meeting similar discontinuation criteria experienced recurrence of hyperprolactinemia within a year. In both studies, there was a strong correlation between the size of the tumor remnant and recurrence of hyperprolactinemia. In several other studies on unselected patients, even higher recurrence rates of hyperprolactinemia have been reported [6–8].

Although the side effects of dopamine agonist therapy such as nausea, headache, and dizziness are well known, the concern of increased cardiac valve abnormalities has recently arisen in patients with long-term DA treatment. Increased risk of valvular heart disease was first reported in patients with Parkinson’s disease (PD) who were treated with very high doses of cabergoline or pergolide, showing a cumulative dose and duration effect [11, 12]. Cabergoline or pergolide are potent serotonin 2B receptor (5-HT2B) agonists and stimulate 5-HT2B receptors on normally quiescent valvular cells [43, 44]. The stimulation of 5-HT2B receptors, which is known to activate several mitogenic pathways, appears to be a key step in initiating cabergoline-or pergolide induced valvular heart disease in patients with PD [44]. In several studies on the association between PD and valvular heart disease, the cumulative dose ranged between 2,600 and 6,700 mg, whereas it ranged between 200 and 443 mg in similar studies on hyperprolactinemic patients [45]. Seven of eight observational studies on the effects of cabergoline in hyperprolactinemic patients do not show an association between cabergoline therapy and clinically relevant, i.e. moderate or severe valvular disease, whereas one study found an increased rate of moderate tricuspid regurgitation [46–51]. Within these eight observational studies, two showed an increased prevalence of mild tricuspid regurgitation [52, 53]. A recent meta-analysis by Bogazzi et al. [49] which pooled data from six clinical studies on hyperprolactinemic patients treated with cabergoline showed an increased risk of regurgitation of the tricuspid valve. One study also found increased thickening and calcifications of cardiac valves in patients treated with cabergoline compared with controls [51]. Though onset of cardiac valve disease is not associated with cabergoline treatment in patients with prolactinoma, subtle changes in cardiac valves such as calcifications, thickening and increased mitral tenting area have been observed in a subset of studies [50–52] and might predispose to more severe and clinically relevant changes in patients to whom cabergoline treatment is given long-term or in high doses.

The management of small prolactinomas in pregnant women can be complex [54, 55]. In our series no patient was pregnant, but some of the patients who opted for a primary surgical approach in this study wished to become pregnant and avoid long-term DA medication. Only a small percentage (2.2–4.5%) of pregnant women with microprolactinomas experience a symptomatic tumor enlargement (headaches and/or visual disturbances) during gestation [56]. However, 40% of pregnant women with microprolactinomas demonstrate asymptomatic tumour growth [57]. During pregnancy, rising estrogen levels stimulates PRL synthesis and promotes lactotrope cell hyperplysia [58]. Most patients will require treatment of hyperprolactinemia to restore fertility. With a primary surgical approach, DA therapy can be avoided in a patient to restore ovulation. Therefore, the foetus is not exposed to either bromocriptine or cabergoline for at least three to 4 weeks of gestation, until a pregnancy test can verify conception and the DA medication can be discontinued. DAs are not approved for use during pregnancy. Cabergoline and bromocriptine treatment at the time of conception appears to be safe for both the pregnant patient and the neonate, showing rates of spontaneous abortions (11%) and neonatal malformations (3.5%) similar to those rates in normal population [57, 59–61]. However, reinstitution of the DA or surgical debunking/removal is indicated if symptomatic enlargement of a prolactinoma occurs.

Conclusion

Prolactinomas are the most common neoplasm of the anterior pituitary gland. Treatment of hyperprolactinemia is indicated because of resulting infertility, gonadal dysfunction and osteoporosis. The data presented here shows high long-term remission rates for primarily surgically treated small prolactinomas (88%) with very low major morbidity rates in a specialized pituitary centre. Patients with small prolactinomas can therefore consider pituitary surgery as an alternative to treatment with DA. The likelihood of achieving long-term remission is high, especially if preoperative PRL levels are below 200 μg/l. However, the possible recurrence of hyperprolactinemia after pituitary surgery needs to be addressed. Long-term DA medication and potential long-term side effects can be avoided in carefully selected patients opting for primary pituitary surgery.