A step forward toward personalised medicine in oncology: Population modelling for the early prediction of disease progression using biomarkers
(1) Núria Buil-Bruna, (2) José-María López-Picazo, (3) Tarjinder Sahota, (4) Marta Moreno-Jiménez, (2) Salvador Martín-Algarra, (5)* Benjamin Ribba and (1)* Iñaki F. Trocóniz
(1) Department of Pharmacy and Pharmaceutical Technology, School of Pharmacy; University of Navarra, Pamplona, Spain (2) Department of Medical Oncology, Clínica Universitaria de Navarra, University of Navarra, Pamplona, Spain (3) Clinical Pharmacology Modelling and Simulation, GSK, UK (4) Department of Radiation Oncology, Clínica Universitaria de Navarra, University of Navarra, Pamplona, Spain (5) Inria, Ecole Normale Supérieure de Lyon, Lyon Cedex 07, France. * These authors share the senior authorship of this work
Objectives: Development of individualised therapies poses a major challenge in oncology. Significant hurdles to overcome include better disease monitoring and early prediction of clinical outcome. We propose a population modelling framework that relates circulating biomarkers in plasma, easily obtained from patients, to tumour progression levels assessed by imaging computed tomography (CT) scans (i.e., Response Evaluation Criteria in Solid Tumors (RECIST) (1) categories). We apply this modelling and prediction strategy to small cell lung cancer (SCLC) patients.
Methods: First, we developed a biomarker model using 369 lactate dehydrogenase (LDH) and 152 neuron specific enolase (NSE) concentrations from medical records of 60 patients. The model comprised two indirect response models for LDH and NSE, where biomarker levels were driven by an underlying latent variable ("disease level") representing (unobserved) tumour size dynamics. Second, based on 215 RECIST categories, we developed a logistic regression model which related the change in the unobserved disease variable to a probability of disease progression [P(DP)]. We then used MCMC Bayesian analysis to obtain the full uncertainty distribution of individual future P(DP). Finally, we determined the minimal number of individual biomarker concentrations (earliest interim time point) required to obtain accurate future predictions of P(DP).
Results: The biomarker model successfully described LDH and NSE dynamics. Predictive checks on the proportion of individuals with disease progression obtained with the combined model showed good model performance. Earlier predictions were associated with less individual data and therefore higher uncertainty. For the end of therapy CT scan, the model achieved accurate prediction of P(DP) in most of the patients before the last scheduled chemotherapy cycle providing a possibility for early transfer to second line treatment in patients where a high P(DP) is predicted. For the follow-up CT scans, accurate predictions were achieved at least 5 weeks prior to the scheduled time.
Conclusions: We have developed a model for LDH and NSE in SCLC patients which allowed estimation of underlying disease levels. We have extended the model to predict P(DP) using the predicted change in latent disease level, showing that the use of the population model enables accurate individual predictions of P(DP) prior to tumour response assessments.
(1) Eisenhauer E, Therasse P, Bogaerts J, Schwartz L, Sargent D, Ford R, et al. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). Eur J Cancer 2009;45(2):228-247.
(2) Buil-Bruna N, López-Picazo JM, Moreno-Jiménez M, Martín-Algarra S, Ribba B, Trocóniz IF. A population pharmacodynamic model for lactate dehydrogenase and neuronal specific enolase to predict tumor progression in small cell lung cancer patients. The AAPS Journal 2014; Advance online publication. doi: 10.1208/s12248-014-9600-
Acknowledgements: The research leading to these results has received support from the Innovative Medicines Initiative Joint Undertaking under grant agreement n° 115156, resources of which are composed of financial contributions from the European Union's Seventh Framework Programme (FP7/2007-2013) and EFPIA companies' in kind contribution. The DDMoRe project is also supported by financial contribution from Academic and SME partners. This work does not necessarily represent the view of all DDMoRe partners.