Predictive Factors for Surgical Intervention in Prolactinomas

Introduction: Prolactinomas remain pituitary tumors predominantly managed with dopamine agonists (DAs). However, surgery remains an important treatment option for patients suffering from medication side effects, tumor resistance, patient noncompliance, financial costs, and occasionally the psychological effects of long-term therapy. In this review, we analyzed the factors predicting surgical intervention in a retrospective single center patient population.

Methods: We reviewed the charts of 4,660 pituitary patients, based on ICD-10 coding, managed at the University of Colorado, between 2004 and 2019. We identified 448 prolactinomas; 321 treated with DAs alone and 127 managed surgically. We then randomly selected a sample of 160, out of the 321 patients, managed medically. Final analysis included 154 medically and 120 surgically managed patients. We collected demographic, clinical, radiologic, pathologic, biochemical, medical and surgical treatment data. We performed descriptive statistics, Mann–Whitney test for quantitative and contingency tables and Chi-square tests for qualitative variables using SPSS 27. Normal distribution was assessed using Kolmogorov-Smirnov test. Statistical significance was set at p < 0.05.

Results: An increased likelihood of surgical intervention was statistically associated with: a diagnosis made from 2004 to 2009 vs. 2015 to 2019 (p = 0.008), size of adenoma ≥1 cm in any dimension (craniocaudal, transverse, or anteroposterior) (p = 0.0001), suprasellar extension (p = 0.001), Knosp grade >0 (p = 0.0001), mixed tumor density (p = 0.002), sellar floor erosion (p = 0.014), higher maximal cabergoline dose (p = 0.001; 95% CI for surgical group 1.28–1.83; for medical 0.88–1.17), discontinuation of DAs due to side effects (p = 0.001), failure to achieve prolactin <10 (p = 0.001) and failure to decrease prolactin by ≥50% (p = 0.01). Age, gender, race, duration of symptoms, visual acuity, visual field abnormality, bone invasion, middle fossa extension, and peak prolactin level at diagnosis did not predict surgical intervention. Based on our findings, we designed and evaluated seven different scales for predicting surgical intervention. We present the most accurate one below (see [Table 1]). A score ≥2 is 82.35% sensitive and 56.6% specific, for a 47.7% positive predictive and an 87% negative predictive value for the likelihood of surgical intervention.

Conclusion: We identified a number of different factors, in patients with prolactinomas, which may be useful in predicting likelihood to proceed to surgical intervention. We also designed a scale which can be used, as a screening tool, to help in identifying this patient group. This tool may also be useful to help identify patients who will not require surgery. At present, this scale, although useful, should not be used as a stand-alone tool to make decisions regarding treatment intervention. Additional studies are needed to further validate its benefit in individual clinical decision making.

Table 1 Prolactinomas for surgery score

Size ≥1 cm in any dimension (craniocaudal, transverse, or anteroposterior)	1
Suprasellar extension	1
Knosp grade >0	1
Failure to achieve prolactin <10 on dopamine agonist (DA)	1
DA discontinued due to side effects	1

Publication History

Article published online:
15 February 2022