Validation of the predictive value of modeled hCG decline profiles in low-risk gestational trophoblastic neoplasia (GTN) treated with methotrexate (MTX).

Abstract

5099 Background: In low-risk GTN, MTX treatment is changed according to serum hCG levels. We previously showed that population modeling of hCG levels using kinetic equations provided parameters that could be used as independent predictors of MTX resistance (You et al; Ann Oncol 2010). In the present study we applied this approach to a large independent patient cohort in which hCG measurements were made using the same immunoassay. Methods: The hCG kinetic profiles of 563 patients with low-risk GTN (281 MTX Resistant: 282 MTX Sensitive) treated at Charing Cross Hospital were retrospectively analyzed in a 3-step validation study. Initially a 189 patient “Model Set” was used to derive kinetic parameters of the model using the NONMEM program. Patient hCG kinetics were described from day0 to day50 using: “[hCG(time)] = hCG0i * exp(–k*time)+ P”, where P is residual hCG tumor production, hCG0i is the initial hCG level, and k is the rate constant. The predictive values of different kinetic parameters (clearance, area under the curve, AUC and P) were assessed using ROC curves and logistic regression. Subsequently, 2 independent datasets comprising 149 and 225 patients with blinded-resistance status (Validation1 and Validation2 sets) were used to assess the validity of the modeling and predictions. Results: Despite large inter-individual variability, individual hCG decline profiles of the Model Set patients were well fitted by the model. The best discriminator of MTX resistant vs. sensitive patients was median P (cut-off = 23: ROC AUC = 77.3). P was the only significant predictor of resistance against WHO score items. The model was then applied to the Validation1 and 2 patient cohorts. Again, the decreasing hCG values were well fitted by the kinetic model. The predictive value of P for MTX resistance was confirmed: Validation1 ROC AUC = 82.1% (Sensitivity = 75%; Specificity = 89%) and Validation2 ROC AUC = 92.0% (Sensitivity = 94%; Specificity = 89%). Conclusions: We have validated population kinetic modeling of hCG decline profiles in a large cohort of low-risk GTN patients treated with MTX and confirmed its clinical utility for early prediction of MTX resistance.