Quantitative In Vivo Imaging to Enable Tumour Forecasting and Treatment Optimization

Mathematical characterization of population dynamics in breast cancer cells treated with doxorubicin

“…Fifth, we only analyzed a constant versus an adaptive parameterization for the surviving fraction ( ) and the death delay rate ( ) because we hypothesized that these would suffice to account for the development of chemoresistance. While this choice was supported by the results presented herein, an uncertainty quantification approach could also be exploited to conduct a model selection study aiming to investigate the optimal combination of constant and adaptive parameters in Experiments 2 and 3 ( Lorenzo et al, 2022 ), which may provide new insights in the development of chemoresistance to doxorubicin. Furthermore, while experimental observations and modeling results in our study support the adoption of a fixed value of after treatment, the aforementioned modeling selection analysis could also be extended to investigate whether the change in the carrying capacity after treatment (i.e., from to ) is permanent or temporal; although this analysis most likely requires additional experiments and data types to investigate the biological mechanisms underlying either of these two modeling alternatives (e.g., doxorubicin-induced changes in cell size or genetic alterations).…”

“…Fourth, given that our model requires a moderate number of parameters that may increase with the number of delivered doses of doxorubicin, their estimation from specific experimental data may exhibit a certain degree of uncertainty (see Supplementary Appendix C ). Thus, future studies should investigate whether and how the levels of uncertainty obtained for the parameters of our models affect the description of the therapeutic action of doxorubicin on tumor cells, for example, by leveraging a robust Bayesian framework ( Lorenzo et al, 2022 ). Fifth, we only analyzed a constant versus an adaptive parameterization for the surviving fraction ( ) and the death delay rate ( ) because we hypothesized that these would suffice to account for the development of chemoresistance.…”

“…We (and others) ( Zhang et al, 2015 ; Strobl et al, 2021 ; Howard et al, 2022 ; Pisco et al, 2013 ; Álvarez-Arenas et al, 2019 ; McKenna et al, 2017 ; Kazerouni et al, 2020 ) posit that these time-resolved datasets can be integrated in mathematical models of tumor cell population dynamics to systematically investigate the effects of drugs on tumor cells across a much broader variety of regimens than are possible to test in vivo . Then, our ultimate goal is to exploit the knowledge gained from a model constructed and validated in a data-rich in vitro preclinical environment (e.g., where hundreds of data points are available for each replicate) to refine mathematical models and their predictions of therapeutic response in a data-poor in vivo clinical environment (i.e., where less than five data points may be available for each patient) ( Lorenzo et al, 2022 ; Weis et al, 2015 ; Jarrett et al, 2018 ; Jarrett et al, 2020a ). In particular, we think that the mechanistic insights provided by the models and empirical formulas proposed in this study could be leveraged to identify the minimal dose range required to effectively inhibit breast cancer growth in vivo and achieve optimal tumor control, both of which are of great clinical interest ( Carvalho et al, 2009 ; Jarrett et al, 2020b ; Harahap et al, 2020 ; Chan et al, 1999 ).…”

“…Indeed, to refine the description of these doxorubicin effects, our models could also be extended to account for the dynamics of doxorubicin uptake and binding (McKenna et al, 2017); although this would require additional data reporting on those phenomena. From a modeling point of view, previous studies have leveraged other formulations to represent cytotoxic drug action (e.g., an exponential decay, alternative transition terms from proliferation to drug-induced death) (Lorenzo et al, 2022;Howard et al, 2022;Colli et al, 2021) which could be explored with our modeling framework in a model selection study (Lorenzo et al, 2022) to assess the optimal approach to capture the cytotoxic action of doxorubicin. Experiment 2 involved the delivery of two doses of doxorubicin to each replicate of MCF-7 breast cancer cells at varying-inter-treatment intervals ranging from 0 to 16 days (Table 1).…”

“…To this end, experimentally-validated mathematical models of chemoresistance mechanisms could be a potent tool in understanding the dynamics of overall tumor drug response. The description of cancer growth and therapeutic response by leveraging mechanistic mathematical models is a rich field known as mathematical oncology ( Yankeelov et al, 2013 ; Rockne et al, 2019 ; Lorenzo et al, 2022 ). This approach has already shown promise in characterizing breast cancer growth and treatment response in both the preclinical and clinical settings ( Atuegwu et al, 2013 ; Pascal et al, 2013 ; Weis et al, 2015 ; Zhang et al, 2015 ; Geng et al, 2017 ; Palmer and Sorger, 2017 ; Jarrett et al, 2018 ; Jarrett et al, 2020a ).…”