hCGres

hCGres calculates residual human chorionic gonadotropin (hCGres) concentration using pharmacokinetic modeling to predict methotrexate (MTX) treatment failure in patients with low-risk gestational trophoblastic neoplasia (GTN) during the initial 50 days of therapy.

Key Features:

• Modeling hCGres: Calculates residual hCG concentration during the first 50 days of MTX therapy using pharmacokinetic modeling.
• NONMEM implementation: Implements the hCGres model via a specialized algorithm executed with the NONMEM program® (NM).
• Consistent estimates: Produces consistent hCGres estimates that are used to assess MTX resistance risk.
• Validation with simulated and real data: Validated using 60 fictive patients with simulated hCG profiles and an independent database of 531 actual patients, reporting median relative prediction errors (RPE) of -0.059 and 9.9×10^-7, respectively.
• Predictive accuracy: Demonstrates robust prediction of MTX failure-risk with Receiver Operating Characteristic area under the curve (ROC AUC) of 0.90 (95% CI 0.87, 0.93) in simulated and real patient datasets.
• Clinical interpretation: Interprets hCG decline curves to enable early detection of potential MTX treatment failure and prediction of MTX resistance.
• Large-scale validation: Validated across more than 1,690 patients from four independent datasets as an early predictor of MTX resistance.

Scientific Applications:

• Early prediction of MTX failure: Predicts methotrexate treatment failure risk in low-risk GTN patients within the initial 50-day treatment window.
• Therapeutic decision support: Informs timely adjustments in treatment strategy by identifying patients at high risk of MTX resistance.
• Clinical interpretation of hCG kinetics: Supports interpretation of hCG decline curves in MTX-treated GTN for clinical monitoring.
• Research and protocol optimization: Provides validated quantitative endpoints for studies aimed at optimizing GTN treatment protocols.

Methodology:

Pharmacokinetic modeling implemented via a specialized algorithm (https://www.biomarker-kinetics.org/hCG) using the NONMEM program® (NM); validation used 60 simulated (fictive) patient hCG profiles and an independent database of 531 patients, reporting median RPEs of -0.059 and 9.9×10^-7 and ROC AUC 0.90 (95% CI 0.87, 0.93), with aggregated validation across >1,690 patients from four independent datasets.