Baseline Predictors of Incomplete Biochemical and Radiological Outcomes During Cabergoline Treatment of Prolactinomas: A Retrospective Cohort From a Tertiary Referral Center

PURPOSE: Although cabergoline is the treatment of choice for prolactinomas, a non-negligible proportion of patients show incomplete biochemical and/or tumoral response. Although several baseline factors have been associated with dopamine agonist resistance, their predictive value remains inconsistent. We aimed to evaluate whether baseline clinical, biochemical, and gross tumor burden variables are associated with incomplete biochemical and radiological response to cabergoline in prolactinomas using a predefined composite operational outcome. METHODS: We conducted a retrospective observational cohort study at a tertiary care center in Mexico City. Adult patients with prolactinomas treated with cabergoline were included. The primary outcome was incomplete response to cabergoline, defined using a predefined operational composite endpoint requiring both prolactin normalization and at least 30% estimated tumor volume reduction during follow-up. Follow-up duration and cabergoline dose exposure were summarized descriptively and component analyses evaluated biochemical and radiological response separately. A sensitivity analysis used at least 30% tumor diameter reduction instead of volume reduction. Baseline characteristics were compared between responders and patients with incomplete response. Multivariable models including only baseline variables were constructed to identify independent baseline factors associated with incomplete response. Because maximal tolerated cabergoline dose and the reasons for dose non-escalation were not uniformly available, this endpoint was used exclusively as a research classification; non-attainment was not interpreted as clinical treatment failure or dopamine-agonist resistance. RESULTS: A total of 226 patients were included. Using the primary volume-based definition, 140 patients (61.9%) were classified as responders and 86 (38.1%) as having incomplete response. Follow-up duration from cabergoline initiation to the last clinical assessment was calculable in 225 patients, with a median duration of 107 months (IQR 56.5-153.5). The median cabergoline dose at 12 months was 1.5 mg/week (IQR 1.0-2.0), and 61 patients (27.0%) had received ≥ 2 mg/week at that time. In unadjusted analyses, age, sex, baseline estimated tumor volume, and tumor size category were not associated with incomplete response, whereas baseline prolactin showed only a weak non-significant trend. In the primary multivariable model, none of the baseline variables independently identified patients with incomplete response. In exploratory component analyses, 140 patients (61.9%) achieved both prolactin normalization and ≥ 30% estimated tumor volume reduction, 14 (6.2%) achieved prolactin normalization only, 46 (20.4%) achieved ≥ 30% estimated tumor volume reduction only, and 26 (11.5%) achieved neither endpoint. Higher baseline prolactin was associated with incomplete biochemical response (adjusted OR 1.90 per log10 increase, 95% CI 1.11-3.25; p = 0.019; AUC 0.624), but no baseline variable was independently associated with incomplete radiological response. CONCLUSION: Using a composite operational definition of incomplete response to cabergoline, no routinely available baseline clinical, biochemical, or gross tumor burden variable was independently associated with the composite endpoint or with incomplete radiological response. Baseline prolactin was associated only with incomplete biochemical response, and discrimination remained limited. Conventional baseline variables appear insufficient for clinically useful prediction, highlighting the need for more comprehensive models incorporating dynamic treatment-related, imaging features of tumor invasiveness, and molecular markers.