Trans-Sphenoidal Surgery in Prolactin Secreting Pituitary Tumors: A Long-Term Referral Center Experience.

Backgrounds: Prolactin secreting pituitary tumors (PRL-omas) represent a unique challenge for endocrinologists and neurosurgeons. Considering recent innovations in surgical, we aimed to investigate the best management for PRL-omas. Methods: A retrospective, cross-sectional and monocentric study was designed. Consecutive patients affected for PRL-omas were enrolled if treated with rst line treatment with dopamine-agonist (DA) or trans-sphenoidal surgery (TSS). Patients carried giant PRL-omas and those with a follow-up < 12 months were excluded. Results: 259 patients were enrolled. The rst treatment was DA for 140 patients and TSS for 119 cases. 146 of 249 patients (58.6%) needed a second therapy. Mean follow up was 102.2 months (12-438 months). Surgery highly impacted on the cure rate, in particular in females (p=0.0021) and in micro- PRL-omas (p=0.0020). Considering multivariate analysis, female gender and surgical treatment in the course of clinical history were the only independent positive predictors of cure at the end of 5 years follow-up (p=0.0016, p=0.0005). The evaluation of serum prolactin (24 hours after TSS) revealed that 86.4% of patients with post-operative PRL ≤ 10 ng/ml resulted cured at the end of follow-up (p< 0.0001). Conclusion: According to our experience, surgery allows a higher cure rate of PRL-omas, in particular in females with micro-PRL-omas, with a good safety prole. TSS for PRL-oma should be considered as rst line therapy, in the management of patients, during the multidisciplinary evaluation, in center of reference for pituitary diseases, in order guarantee a valid surgical outcome and a safety prole.


Introduction
Pituitary tumours account for 10-15% of primary intracranial neoplasms. Among secreting pituitary tumors, PRL-omas are the most common (40-50% of total). 1 PRL-omas, due to their biological and clinical features, represent a challenge for endocrinologists and neurosurgeons. Dopamine agonists (DA) are the gold standard as rst-line management of PRL-omas due to their e cacy and safety pro le. 2,3 Data on cure after withdrawn DA are less clear. 2,4 Trans-sphenoidal surgery (TSS) is con ned to failure of medical therapy, pituitary apoplexy with neurological worsening, and cystic PRL-oma. 4 Nonetheless, various papers also report good results with TSS as rst-line therapy for PRL-omas. [5][6][7] Moreover, the recent technical innovations introduced in TSS, such as high-de nition surgical endoscopes and extremely reliable neuronavigation systems, have made possible to further expand the surgical possibilities and to obtain better results as complete tumor resection with lower complication rates. 8 Thus, in the neuro-endocrinology community, a debate on the possibility to expand the traditional indications of TSS as rst-line treatment for PRL-omas is still alive. 9,10 The aim of the present study was to ascertain the role of surgery in the treatment of PRL-omas, though the analysis of our surgical and medical experience of over 25 years in the treatment of PRL-omas.

Study Design
We conducted a retrospective and cross-sectional study, reviewing the clinical, radiological and surgical charts of enrolled patients.
Patients were consecutively enrolled according the following inclusion criteria: 1) Diagnosis of PRL-omas; 2) treatment with with rst line therapy with DA (cabergoline or bromocriptine) or with rst line neurosurgical operation via TSS (either endonasal endoscopic or sublabial microsurgical); 3) Diagnosis and treatment of PRL-omas conducted at our Institution between January 1st, 1992 and December 31, 2016; Were excluded from the study: 1) patients carried a giant PRL-omas (diameter > 4 cm); 2) patients with a follow-up shorter than 12 months .
All patients provided informed consent according to research principles of Institutional Ethics Committee.
Clinical management of patients PRL-omas were diagnosed according to guidelines, that had been approved at the time of diagnosis for each patient. With regard of therapeutic management, over the years, at our institution, all patients affected by PRL-omas had underwent a multidisciplinary evaluation by the endocrinologist and the neurosurgeon. According to Klibanski ( 11 ), TSS was offered as rst line treatment in: -pituitary apoplexy -cystic PRL-omas -macro-PRL-omas in a patient with a psychiatric condition for which dopamine agonists are contraindicated -after a joint discussion between endocrinologists, neurosurgeons and patient, concerning risks and bene ts, with the nal choice based upon the patient's preference.
During the follow-up, all patients underwent: -prolactin dosage, one and three months after pituitary surgery/start of DA therapy and then every six months; -Pituitary MR three and six months after pituitary surgery/start of DA therapy and then every year.

Follow-up
Outcome of patients at the end of follow was classi ed as: 1) cured: in cases of regression of clinical symptoms, stable normalization (below the gender-speci c normal upper limit) of basal prolactin levels and absence of neuroradiological evidence of residual/recurrent tumor, at least 12 months after neuro-surgery or DA discontinuation; 2) controlled disease: in cases with normal serum PRL concentration and stable neuroimaging, during DA therapy; 3) uncontrolled disease: in cases with high serum PRL concentration (above the gender-speci c normal upper limit) and/or tumor progression on neuroimaging, during DA therapy.

Statistical analysis
Continuous variables were expressed as mean (range), categorical variables as absolute and relative frequency. Comparison of continuous variables between groups was performed using the Mann-Whitney U test. Comparison of categorical variables was performed by chi-square statistic using the Fisher exact test when appropriate. A multivariate analysis model was built using logistic regression to calculate the Odds Ratio (OR) of healing by adjusting for the following parameters: size of the adenoma, age at diagnosis, gender, surgical treatment in the course of clinical history, type of rst treatment (TSS/DA). A logistic regression model was also built to calculate the OR of healing depending on treatment. A p < .05 was considered signi cant. A ROC curve was built to assess the diagnostic accuracy for cure of postoperative serum prolactin; the value with the highest Youden index [sensitivity-(1-speci city)] was designed as best cut-off. StatViewver 5.0 software was used (SAS Institute, Cary, NC).

Patient Population
Among all patients treated for PRL-oma at our Institution in the study period, 259 (164 women and 95 men) ful lled the criteria for enrollment in the present study. Baseline characteristics of patients are detailed in Table 1. The 64% of female patients were affected by micro-PRL-omas, whereas the majority of male patients harbored a macro-PRL-omas (78.9%; p < .0001, Fisher exact test). Mean age was 35.2 years; females were signi cantly younger than males (p < .0001, Mann-Whitney U test). Mean follow-up was 102.2 months (range, 12-438 months) and did not differ signi cantly between genders.  Table 2 Univariate analysis of predictive factors for rate of cure in whole court and subgroup of patients with follow-up >=5 years.
We conducted a subgroup analysis according to the chosen rst-line treatment. Among the group of patients underwent surgery as rst-line treatment, we found that the cure rate was higher in female patients (cure rate 67.9% vs 45.4% for rst line treatment with DA, p = .0080), also independently from the dimension of the PRL-omas. In fact, the rate of cure was signi cantly higher in female patients treated with rst-line surgery as compared to those treated with rst-line DA, both in cases of micro-PRL-omas (cure rate 69% vs 48.7% for DA, p = .0080) both in cases of macro-PRL-omas (cure rate 66.7% vs 37.5% for DA, p = .0037) (  Table 3 Rate of cure depending on rst treatment and surgery in the whole cohort In addition, we tried to investigate if the surgery conducted at any time during the patients' clinical history may play a role in the outcome. We found that patients underwent pituitary surgery had a higher rate of cure, both in cases of micro-PRL-omas (surgical treated patients: 78.8% vs not-surgical treated patients: 21.2% p < 0.0001) and both in cases of macro-PRL-omas (surgical treated patients: 38.8% vs not-surgical treated patients: 11.1% p = 0.03 Table 3).
The multivariate analysis conducted on the whole study cohort con rmed that female sex, the surgery as rst/second-line treatment and micro-PRL-omas were found to be independent predictive factors for a cure at the end of follow-up (odds ratio for no cure 0.323, 0.196 and 0.469, respectively; p = .0009, p < .0001 and p = .0182, respectively) (  Table 4 Multivariate analysis of predictive factors for a cure in the whole cohort and subgroup of patients with follow-up >=5 years.

Cure in long follow-up
In order to validate the results on predictive factors for a cure, we focused on those patients with long follow-up, of at least 5 years from the diagnosis of PRL-omas.
One hundred sixty-four patients met this criterion, 128 of which had undergone at least one surgical treatment in the course of their clinical history. At univariate analysis for cure rate, female sex and surgical adenoma removal as rst or second choice con rmed their positive predictive value for cure (p = .0003 and p = .0042, respectively; Fisher exact test), whereas tumor size showed only a nonsigni cant trend to an improved cure rate (Table 2).
Patients treated only with surgery had the best chance of cure. Importantly, neurosurgical adenoma removal, whenever during the clinical course, persistently determined a higher cure rate in the whole cohort, in female patients, in micro-PRL-omas, and in the subgroup of female patients harboring micro-PRL-omas (p = .0042, p = .0021, p = .0020 and p = .0041, respectively; Fisher exact test; Table 5).  Table 5 Rate of cure depending on rst treatment and surgery in patients with follow-up >= 5 years. Finally, female sex and surgical treatment were the only independent positive predictors of cure at the end of follow-up in patients with follow-up > 5 years (odds ratio for no cure 0.257 and 0.158, respectively; p = .0016 and p = .0005, respectively; Table 4).
1st Day post-operative serum PRL as biomarker for cure The dosage of PRL conducted in fasting condition, in the morning of the rst day after pituitary surgery was signi cantly lower in cured vs not cured patients (p < 0.0001, Mann-Whitney U test; Fig. 1). The best cut-off was set at 37.5 ng/ml. Moreover, by applying a clinically relevant cut-off value of 10 ng/ml, 86.4% of patients with lower values vs 27.3% of patients with higher values resulted to be cured at the end of follow-up (p < .0001, Fisher exact test).

Discussion
In this study, we investigated the role of the trans-sphenoidal pituitary surgery in the treatment of PRLomas, reviewing our cohort of patients that had been managed with surgical and/or medical therapies, over a period of 25 years.
Our nding demonstrated that patients underwent pituitary surgery had the high probability of reaching the cure for PRL-omas. In fact, in this cohort, for the rst time, we proved that the TSS acts as a positive prognostic factor for the cure of PRL-omas, together with the female gender and the small (lower than 10 mm) dimension of the pituitary tumor.
In addition, in the group of patients with a follow-up longer than 5 years, surgery alone was linked to 91% of chance of cure and carried a 57% rate of cure also when performed after DA failure. Similarly, in patients with follow-up longer than 5 years, the only independent positive prognosticators for cure remained female sex and surgery, independently from the dimension of the pituitary tumors.
Our results allow us to describe a group of patients that may strongly bene t by surgical removal of PRLomas. In this study in fact we demonstrated that female patients treated with rst-line surgery had a high rate of cure, as compared to patients treated with DA (as rst line therapy), also independently from the pituitary tumor dimension.
According to current guidelines ( 4 ), however, the rst-line therapeutic choice for prolactinoma is the medical therapy with DA, instead surgery is usually con ned to a complementary therapy for patients resistant to DA.
However, recent studies had proved that a recurrence of prolactinoma may be observed in 20-77% of cases after the withdrawal of DA ( 25 ), according to tumoral dimension, invasion of cavernous sinus, nadir prolactin value reached during DA treatment ( 12 ), persistence of tumoral residual disease and duration of treatment with DA ( 26 ).
Our evidences deserve special consideration because, in high-volume centers with low surgical complications, the positive in uence of surgery in long-term remission/cure could decrease the concerns that obviously accompany any surgical procedure. In fact, in the most recent years, it was suggested that the trans-sphenoidal pituitary surgery should be considered an important part of the patients counselling for the decision of the initial treatment, according to the low morbidity of this procedure, if performed by an experienced neurosurgeon ( 27 ). In addition, the available of tumoral tissue may allow a detailed pathological analysis, for investigating both biomarkers of aggressiveness, as Ki67, p53, mitotic count, minichromosome maintenance 7 (MCM7) and estrogen receptors and both biomarkers of treatment response, as dopamine and somatostatin receptors ( 28-31 ). In fact, a very detailed pathological analysis may facilitate the identi cation of cases with high risk of recurrence and may orient towards a personalized therapy, in particular in cases of di cult and aggressive PRL-omas.
In addition, our results suggested an additional bene t of the surgical management of PRL-omas: the duration of adjuvant treatment with DA (after rst line pituitary surgery) is signi cantly shorter as compared to those reported in patients treatment with DA as rst line therapy. Our nding is in-line with a previous paper of Andereggen et al. ( 32 ) that showed that at 10 years follow-up, the control of hyperprolactinaemia required DA-agonist therapy in 32% of patients who underwent primary surgical therapy and in 64% of patients who had primary medical therapy and that the primary surgical therapy is a protective factor for long-term treatment with DA.
The main limitation of our study are its retrospective design and the lack of randomization. However, this study described a real life scenario of a large and monocentric series of patients, managed at a historical pituitary unit, that had make a strong effort in the research on prolactin in physiological and pathological conditions ( 33-41 ).
Our study has the privilege to analyze a homogeneous case series: though collected in a 25-year timeframe, indications, treatment, and follow-up evaluation were conducted by the same medical staff.
To further reduce the confounding factors of our analysis, giant PRL-omas have been excluded, because in such invasive cases a multimodality treatment, including radiotherapy and life-long DA, is often required. In addition, in this study lacks a detailed analysis of the various aspects that characterize the clinical and endocrine syndrome of PRL-omas and that can be differently controlled/cured by the chosen treatment. On the other hand, focusing on a few selected serological and neuro-radiological aspects of the illness was essential in order to analyze a considerable number of patients with a long follow-up.
In conclusion, surgery allows a higher cure rate of PRL-omas: tran-sphenoidal surgery may represents a valid alternative to DA therapy, particularly in females with micro-PRL-omas, as provides the highest chance to cure in the long-term follow-up. According to our experience and data, surgical removal of PRLomas should be considered also as rst line treatment in the management of affected patients, during the multidisciplinary evaluation, in center of reference for pituitary diseases, that may guarantee a valid surgical outcome and a safety pro le.

Declarations
Con ict of Interest: None.
Disclosure of Funding: None. Figure 1 Box plot showing post-operatory serum PRL in the subgroup of patients cured vs not cured at the end of 5-year minimum follow up