AUTHOR=Huber Markus , Luedi Markus M. , Schubert Gerrit A. , Musahl Christian , Tortora Angelo , Frey Janine , Beck Jürgen , Mariani Luigi , Christ Emanuel , Andereggen Lukas 

TITLE=Gender-specific prolactin thresholds to determine prolactinoma size: a novel Bayesian approach and its clinical utility

JOURNAL=Frontiers in Surgery

VOLUME=Volume 11 - 2024

YEAR=2024

URL=https://www.frontiersin.org/journals/surgery/articles/10.3389/fsurg.2024.1363431

DOI=10.3389/fsurg.2024.1363431

ISSN=2296-875X

ABSTRACT=Background: 
In clinical practice, adenoma size is critical for guiding prolactinoma treatment decisions. Establishing guidelines for serum prolactin thresholds to assess adenoma size is essential. However, the impact of gender differences in prolactin levels on estimating adenoma size remains incompletely understood.

Objective:
Introduce a novel statistical method for deriving gender-specific prolactin thresholds to differentiate between micro- and macroadenomas and assess their clinical utility.

Methods:
We present a multilevel Bayesian logistic regression approach to compute gender-specific prolactin thresholds in a large prolactinoma patient cohort (N=133) with dichotomized adenoma size. The approach's robustness is examined using ensemble machine learning (a "super learner"), preserving observed gender differences in prolactin and adenoma size and artificially increasing the sample size tenfold.

Results:
The framework yields a global prolactin threshold of 239.4 μg/L (95% credible interval: 44.0 – 451.2 μg/L) for discriminating between micro- and macroadenomas. Gender-specific thresholds are found: 211.6 μg/L (95% credible interval: 29.0 – 426.2 μg/L) for women and 1046.1 μg/L (95% credible interval: 582.2 – 2325.9 μg/L) for men. Applying male-specific thresholds improves specificity (0.99) with moderate sensitivity (0.74). However, male-dependent prolactin thresholds exhibit uncertainty, particularly with small sample sizes. Augmented datasets suggest larger cohorts could reduce threshold uncertainty.

Conclusions:
The framework advances patient-centered care for prolactinoma treatment by introducing gender-specific thresholds. These thresholds facilitate tailored treatment strategies by accounting for gender differences in adenoma size. A universal threshold effectively rules out macroadenomas in men, while a male-specific threshold indicates its presence. However, the clinical utility of a female-specific threshold is limited in our cohort. This adaptable framework offers valuable insights into imbalanced biomarkers and outcomes. Using machine learning to expand datasets while preserving imbalances aids in assessing gender-specific threshold reliability. Nonetheless, external validation is crucial for confirming threshold accuracy.