Signaling pathway models as biomarkers: Patient-specific simulations of JNK activity predict the survival of neuroblastoma patients

Abstract

Signaling pathways control cell-fate decisions that ultimately determine the behavior of cancer cells.Therefore, the dynamics of pathway activity may contain prognostically relevant information different from that contained in the static nature of other types of biomarkers. To investigate this hypothesis, we characterized the network that regulated stress signaling by the Jun N-terminal kinase (JNK) pathway in neuroblastoma cells. We generated an experimentally calibrated and validated computational model of this network and used the model to extract prognostic information from neuroblastoma patient-specific simulations of JNK activation. Switch-like JNK activation mediates cell death by apoptosis. An inability to initiate switch-like JNK activation in the simulations was significantly associated with poor overall survival for patients with neuroblastoma with or without MYCN amplification, indicating that patientspecific simulations of JNK activation could stratify patients. Furthermore, our analysis demonstrated that extracting information about a signaling pathway to develop a prognostically useful model requires understanding not only components and disease-associated changes in the abundance or activity of the components, but how those changes affect pathway dynamics.

On the personalised modelling of cancer signalling * *Supported by EU FP7 grant “SynSignal” (No. 613879).

“…The advantage of using such personalised dynamic models over static biomarkers, has already been demonstrated in several case studies (Fey et al, 2015;Flanagan et al, 2015;Murphy et al, 2013;Lindner et al, 2013). In contrast to previous approaches, the here proposed methodology is not restricted to models in which the total protein concentrations are time-invariant parameters.…”

“…Recently, multiple studies have shown that patient-specific differences in the dynamic behaviour of these signalling networks underlie individual pathogenetic changes and disease manifestation (Fey et al, 2015;Flanagan et al, 2015;Lindner et al, 2013;Murphy et al, 2013). For example, a dynamic model of the JNK pathway could predict the survival probabilities of cancer patients in neuroblastoma, a common childhood cancer (Fey et al, 2015). These predictions were based on personalised simulations for over 700 patients, and revealed that a high amplitude, switchlike JNK activation was associated with neuroblastoma cell death, and better patient survival.…”

“…In fact, all the personalised models mentioned are based on this simple principle of directly using the measured mRNA or protein concentrations as static parameters in the model (Fey et al, 2015;Flanagan et al, 2015;Lindner et al, 2013;Murphy et al, 2013). Thus, all these personalised models are based on the assumption that the Supported by EU FP7 grant "SynSignal" (No.…”

Community-driven roadmap for integrated disease maps

“…In recent studies, exome transcriptome sequencing data of cancer cell lines have been used to set cell line-specific translation rates [64,65]. In a similar study, the mRNA expression was used to predict the survival of individual neuroblastoma patients [66]. While both approaches were successful in the respective applications, transcription rates and mRNA levels can change in response to treatment.…”

Many are better than one - Next Generation Multi-Variate Biomarkers for Precision Oncology

“…The development of multivariate biomarkers for diagnostic assays could allow more patients to benefit from improved drug regimens. Ultimately, such multivariate markers could be implemented in a dynamic manner, if for example integrated with circulating tumor cell (CTC) techniques so as to monitor patient response and tumor dynamics on treatment [38,39] . Our work described here provides a case study of integrating experimental data generated using multivariate profiling technologies with a variety of computational modeling and simulation methods to interpret such measurements and make clinical predictions on their therapeutic utility.…”