Phenotypes | Hawk | Dove |
Hawk | v-c | 2v |
Dove | 0 | v |
Citation: A. Swierniak, M. Krzeslak, D. Borys, M. Kimmel. The role of interventions in the cancer evolution–an evolutionary games approach[J]. Mathematical Biosciences and Engineering, 2019, 16(1): 265-291. doi: 10.3934/mbe.2019014
[1] | Yuanyuan Huang, Yiping Hao, Min Wang, Wen Zhou, Zhijun Wu . Optimality and stability of symmetric evolutionary games with applications in genetic selection. Mathematical Biosciences and Engineering, 2015, 12(3): 503-523. doi: 10.3934/mbe.2015.12.503 |
[2] | Andrzej Swierniak, Michal Krzeslak . Application of evolutionary games to modeling carcinogenesis. Mathematical Biosciences and Engineering, 2013, 10(3): 873-911. doi: 10.3934/mbe.2013.10.873 |
[3] | Bo Lan, Lei Zhuang, Qin Zhou . An evolutionary game analysis of digital currency innovation and regulatory coordination. Mathematical Biosciences and Engineering, 2023, 20(5): 9018-9040. doi: 10.3934/mbe.2023396 |
[4] | Dario Madeo, Chiara Mocenni, Jean Carlo Moraes, Jorge P. Zubelli . The role of self-loops and link removal in evolutionary games on networks. Mathematical Biosciences and Engineering, 2019, 16(5): 5287-5306. doi: 10.3934/mbe.2019264 |
[5] | Jim M. Cushing . The evolutionarydynamics of a population model with a strong Allee effect. Mathematical Biosciences and Engineering, 2015, 12(4): 643-660. doi: 10.3934/mbe.2015.12.643 |
[6] | Zheng Liu, Lingling Lang, Lingling Li, Yuanjun Zhao, Lihua Shi . Evolutionary game analysis on the recycling strategy of household medical device enterprises under government dynamic rewards and punishments. Mathematical Biosciences and Engineering, 2021, 18(5): 6434-6451. doi: 10.3934/mbe.2021320 |
[7] | Shahab Shamshirband, Javad Hassannataj Joloudari, Sahar Khanjani Shirkharkolaie, Sanaz Mojrian, Fatemeh Rahmani, Seyedakbar Mostafavi, Zulkefli Mansor . Game theory and evolutionary optimization approaches applied to resource allocation problems in computing environments: A survey. Mathematical Biosciences and Engineering, 2021, 18(6): 9190-9232. doi: 10.3934/mbe.2021453 |
[8] | Jingyu Liu, Weidong Meng, Yuyu Li, Bo Huang, Bixi Zhang . Effective guide for behaviour of farmers in the withdrawal of rural homesteads: An evolutionary game-based study. Mathematical Biosciences and Engineering, 2022, 19(8): 7805-7825. doi: 10.3934/mbe.2022365 |
[9] | Huan Zhao, Xi Chen . Study on knowledge cooperation of interdisciplinary research team based on evolutionary game theory. Mathematical Biosciences and Engineering, 2023, 20(5): 8782-8799. doi: 10.3934/mbe.2023386 |
[10] | Xiaochun Chen, Jie Zhao, Yingying Ma, Bo Lv, Xuanjin Du . Tripartite evolutionary game study on coordination information security in prescription circulation. Mathematical Biosciences and Engineering, 2023, 20(12): 21120-21146. doi: 10.3934/mbe.2023934 |
The objective of this study is to develop and apply a new mathematical model to address the important questions for basic cancer research and practice of oncology: What is the role of different kinds of intervention (exposures or treatments) in the evolution of cancer? Can proliferation acceleration make carcinogenesis more likely? Can cell killing create conditions that enable the evolution of heterogeneity and therefore increase the viability of the cancer cell population? An essential part of the problem we address is whether the genetic heterogeneity in cancer cells exists ab initio (before diagnosis and intervention), or does it evolve as a result of the intervention. We believe, basing on evidence, that the former is the case, as stated in the following hypothesis: positive (stimulating) or negative (cell-killing) treatment modifies pre-cancer and tumor cell heterogeneity through the promotion of selection of preexisting cell clones, their phenotypic plasticity and self-organization. We are far from believing that using the new model we prepared will prove completely this hypothesis. Nevertheless, we hope that the new approach based on the theory of evolutionary games gives an efficient tool for investigation of different aspects related to the hypothesis. The evolutionary game theory (EGT) combines machinery of the theory of games with present knowledge of population biology and evolution [28,30]. EGT differs from standard game theory in deviating from the rational approach of the competitive players, in considering treatment of strategies as phenotypes of individuals acquired through the evolution. Moreover, the players are members of a population consisting of individuals with different phenotypes (strategies), who can cooperate or compete for resources. As a result of various adaptations to the environment and following games in time (generations) the population can tend to stabilize its structure at the same time gaining stable monomorphism or polymorphism of population's phenotypes. Such state is called evolutionary stable. Whereas evolutionary stable strategy (ESS) is defined as a phenotype that, if adopted by the vast majority of a population, will not be displaced by any other phenotype [29]. However, opposite situation is very probable to occur.
Classical models of cancer development assume that mutations, which promote the growth of cancer cells affect only the units in which they occur [13,23,35] but recent studies point out that tumor cells are able to adopt various genetic strategies which may influence the rate of their own development. Moreover the mutation can also affect neighboring cells [37], and combined with cooperative behaviours, competition for resources such as space, oxygen and nutrition, occurs between different subspecies within the same tumor [1,26] in which internal communication between tumor cells, and between tumor and normal cells, their competition for resources, hierarchical subordination, and collaboration, play an important role in cancer development and differentiation or disease transmission and reaction to stress factors including therapy [4,15]. The evolutionary game is performed between cells with different phenotypes (both normal and cancer cells). The main aim of these game theoretic models is to study the possibility of coexistence or even domination of newly formed tumor cells, which have acquired new strategies (phenotypes) by mutations. To our knowledge, Tomlinson and Bodmer [37] first proposed such a model describing inter-cellular interactions including avoidance of apoptosis and production of angiogenic factors. The models that followed their research described phenomena such as: production of the cytotoxic substances [36], production of growth factors [3], invasion and metastasis [25], tumor-environment interactions [14], radiation bystander effect [21], resistance to chemotherapy and p53 vaccine [7], interaction between osteoclasts and osteoblasts [11], tumor-stroma interaction [15], interaction between different tumors [5] and others (see [4,10,17,32] for survey). Usually, the analysis is limited to two and three phenotypes. Our previous paper [20] makes for the exception in which interaction between four different phenotypes of cells are illustrated using three-dimensional simplexes and time courses. The work of Basanta et al.[8] is the only other paper in which interaction of four phenotypes was discussed. The authors, however, were interested in changes in subpopulations of chosen phenotypes with respect to changes of cost parameters rather than in studying equilibrium between all phenotypes and dynamics of their evolution. The reason for that is related to difficulties with analysis of simplexes of dimension higher than three and complexities in dynamics represented by nonlinear replicator equations increasing with their dimension. It is important to notice that dimension of replicator dynamics equations in the case of three phenotypes is equal to two which means that complex dynamical behaviours typical for nonlinear dynamics should be absent. In our opinion, it is one of the major disadvantages of the small number of considered strategies. Using EGT, we are able to predict whether the population has a tendency to become heterogeneous or rather only one phenotype will survive dominating the overall population. Moreover, using the so-called replicator dynamics [16] we may track the rate of population in time starting from the initial state to the equilibrium defined by ESS. Nevertheless, because of assumptions that perfect mixing occurs it gives only mean field results without possibility to take the effect of local arrangements on intercellular interactions into account. This drawback could be overcome by the methods of spatial evolutionary games (SEGT) based on cellular automata which are the next step in the discovery of new behaviors among cells and give different results than mean-field (averaged, "well mixed", non-spatial and giving values for the population without spatial information [22]) models. Nowadays, spatial games are quickly gaining popularity. Nevertheless, it should be remembered that the use of cellular automata [27] in conjunction with the classical theory of games originated spatial games. The line of reasoning presented by Bach et al. [2], where spatial tools used in modeling of carcinogenesis is best suited to our expectations. Similarly to non-spatial games, the spatial ones are also iterated. We proceed transient generations according to the following steps [2]: payoff updating - a sum of local fitness of neighboring cells, cell mortality - removing a certain number of players, reproduction by competition - defining which of the cells (or strategies) will be on an empty place. The game is played on the lattice forming torus, and all the competition results giving tie are settled randomly. The authors [2] present three ways of cell mortality: synchronous updating - all the cells die simultaneously, and they replacement depends on the strategy of their neighbors before dying, asynchronous updating - in every generation a single cell, chosen at random, dies and is replaced, semi-synchronous updating - probability of individual cellular mortality is equal to 0.1. In the last type 10% of players are deleted from the lattice in every generation. In our studies we are using mainly semi-synchronous updating, since this method allows for the biologically realistic situation. Furthermore simulations show that synchronous updating assumes a global controller of the system, while asynchronous updating implies that the vanishing of small cell clusters is impossible. The reproduction of the removed players (killed cells) is the next stage of the algorithm. Bach et al. have suggested two kinds of reproduction: deterministic - in competition for the empty place the winner is the strongest player (with highest local adaptation - sum of eight scores from cell-cell interaction), probabilistic - values of adaptation (or sum of values from payoff matrix) for each player are divided by total score in their neighborhood. Such local competition, with appropriate fitness and location, allows for strategies of cells with lower fitness, but in better location and locally superior in numbers to dominate in the population. In our opinion [34] deterministic reproduction describes a response to direct exposure of cells to an external stress factor while the probabilistic one is more adequate in modeling a bystander effect in cell populations. The main assumption of the spatial games presented is that each cell on the lattice is represented by a player following only one strategy. The local payoff for each player is the sum of payoffs due to interactions (according to the payoff matrix) with cells in the neighbourhood. We will refer to this approach as the standard, classical one or SEGT. Cells on the spatial lattice can also be considered as heterogeneous (instead of homogeneous), so that each particular player may contain mixed phenotypes. Spatial games of the type proposed in [19,20,33] are called mixed spatial evolutionary games (MSEG). It is important to mention the definition of the phenotype, which is the set of traits or characteristics of an organism [24]. Hence, in MSEG different degrees of playing a particular strategy are treated as different characteristics that define different phenotypes. It may happen that within the population, all of the players have diverse phenotypes (which probably better describes biological phenomena). For the sake of simplicity and following the way of reasoning from SEGT, those strategies and traits still correspond to the phenotypes and a general, collective point of view is defined as a players' phenotypic composition. In fact, the game is performed on a multidimensional lattice (dependent on the number of defined phenotypes in the model), where each layer represents a particular phenotype (as the frequency of occurrence) of the player. For the computation of the local adaptation, the sum of the payoffs between each phenotype (within two players) multiplied by their rate of occurrence is calculated first. The second step is summing of these values for each player in the neighbourhood. Another difference between SEGT and MSEG is that the tie (when payoffs are equal) for the former is settled randomly, while for the latter the average of phenotypic compositions is computed. The payoff matrix does not have to remain constant during the game; the parameters may change and may depend on other factors (e.g. changes in the environment or any other external influence). The model may be extended by yet another parameter
Hawk and Dove game is one of the first evolutionary models proposed by John Maynard Smith [30]. It includes two kinds of behaviour, fight or avoid, within a population of one species which is a symbolic representation of the ritual conflicts of two different strategies that arose in the process of evolution.
This game is an example of a two-player game, where each player can choose a strategy from a finite set of strategies. This game has two players, participants of the game and every player have its own decision set, which we call here a strategy. Every pair of players strategies will result in some game outcome for each player, which we call a payoff. Those are the values, written in the matrix form, that can be treated as a profit or cost of selecting a particular strategy. This values can also model, for example, Darwinian fitness. The payoff matrix (Table 1) contains two parameters:
Phenotypes | Hawk | Dove |
Hawk | v-c | 2v |
Dove | 0 | v |
For non-zero sum game, in a bimatrix game where payoff matrices are respectively
(1) |
(2) |
for all
Extending the notion of repeating game for the whole population, where we consider many players, that play the two-player game, we will reach the idea of evolutionary games. Here, the role of the individual player is negligible and the resulting payoff depends on the frequency of strategy played by the members of the population. This value characterizes the population and each individual interacts with other using some phenotype rather than choosing any strategy. For those games we can determine the evolutionarily stable strategy (ESS)
(3) |
where:
In this particular case for the Hawk and Dove game we obtain:
(4) |
For Hawk and Dove game stable polymorphism (coexistence between all phenotypes) is defined by the following ESS:
Now we introduce a new parameter
● Hawks, being more agammaessive, tend to be better adjusted when
● Doves do better (or at least not worse than in a classical game) when resources are greater, providing opportunities, for instance, to "steal" food or to find it without interaction with another individual in the vast area.
● Even if
The parameter
Thus the payoffs for particular interactions are now functions dependent on
Phenotypes | Hawk | Dove |
Hawk | v-c | 2v |
Dove | r(v/2) | (1+r)v |
r=0 | r=1 | |
H, H | v-c | v-c |
D, D | v | 2v |
D, H | 0 | v/2 |
H, D | 2v | 2v |
Using the RDE (3), we can formulate the dynamics of this new model including parameter
(5) |
which for
Spatial games give an opportunity to change
First we start with the mean-field model where parameter
For the MSEG model we introduce additional rules:
1. Individuals who are more hawks take an amount of resources from those who are less agammaessive, which amount results from the difference between hawks frequency of occurrence.
2. A dove shares its resources with those doves that have less resources according to the dove's frequency of occurrence.
3. The resources are limited to the (0, 1) interval.
An example describes this idea in a clearer way. The players are:
Player B: the first rule does not have an application here, since it is dependent on the frequency of occurrence of hawks. Due to the fact that player A has got more resources, rule 2 can be applied. The difference between resources shall be multiplied by 70%, which gives 0.14 to be added to player B (
For the set of parameters
One of the first models based on the evolutionary game theory describing results of interactions between tumor cells was proposed by Tomlinson and Bodmer [37] and described the paracrine production of growth factors including angiogenic promoters. The two phenotypes considered in this model represent cells producing growth factors (in paracrine fashion) and cells which do not produce them, and model parameters are related to costs of proangiogenic factor production (parameter
Phenotypes | A+ | A- |
A+ | 1-i+j | 1-i+j |
A- | 1+j | 1 |
Phenotypes | A+ | A- |
A+ | 1-i+j+r/2 | 1-i+j+r/2 |
A- | 1+j+r/2 | 1+r/2 |
The results qualitatively are similar to those for the hawk-dove game. An interesting, but quite intuitive finding is that additional external growth factors result in changes in final distribution and transient dynamics especially in the case of the deterministic reproduction. Production of angiogenic factors becomes less profitable than in the case when there is no external intervention. Moreover, the spatial distribution of the use of resources by the cells remarkably depends on the initial distribution of phenotypes on the lattice.
Similarly to the previous model we can write the replicator dynamics equations (Eq. 6), in the following form for the original angiogenic model as well as for the model with resources:
(6) |
The introduction of resources in this model will change the baseline for the payoff matrix, which, in model without resources, was equal to 1 in our case. Nevertheless, since
To illustrate advantages of our approach to the analysis of increasing number of strategies let us consider the model which combines two classical models of Tomlinson [36,37], and one of them-the angiogenic game-was described in the previous section. The model contains four different strategies/phenotypes of cells:
● A cell produces a growth factor and the benefit impacts all the neighbors and the cell itself (phenotype A);
● A cell produces a cytotoxic substance against nearby cells (phenotype P);
● A cell is resistant to the cytotoxic substance (phenotype Q);
● Strategy which shall be considered as a baseline (no production of the cytotoxic substance, no resistance to it, no growth factor) (phenotype R).
In this case, the proposed pay-off table is given by Table 6 and its counterpart with resources by Table 7. We assume the existence of two different resources one of which represents an effect of growth factors (
Phenotypes | A | P | Q | R |
A | 1-i+j | 1-i+j-f | 1-i+j | 1-i+j |
P | 1+j-e+g | 1-f-e+g | 1-e | 1-e+g |
Q | 1+j-h | 1-h | 1-h | 1-h |
R | 1+j | 1-f | 1 | 1 |
Phenotypes | A | P | Q | R |
A | 1-i+j+r/2-c/2 | 1-i+j-f+r/2-c/2 | 1-i+j+r/2-c/2 | 1-i+j+r/2-c/2 |
P | 1+j-e+g+r/2-c/2 | 1-f-e+g+r/2-c/2 | 1-e+r/2-c/2 | 1-e+g+r/2-c/2 |
Q | 1+j-h+r/2 | 1-h+r/2 | 1-h+r/2 | 1-h+r/2 |
R | 1+j+r/2-c/2 | 1-f+r/2-c/2 | 1+r/2-c/2 | 1+r/2-c/2 |
In the first case, the external growth factor can have a positive (to recruit cells from the dormant or quiescent phase) or negative effect (as angiogenesis inhibitors, for example, Sunitibib in antiangiogenic therapy). The way they appeared in the payoff matrix results from their actions. Hence the similarity to the cytotoxic and growth effects of the cells that produce them, but without bearing the cost of their production. Therefore, it is so important to include in the matrix both impacts produced by the cells itself and by the external factors. Thus, we have an image of the effect of combination therapy on the evolution of cancer cells, which can be a cell killing with recruitment from quiescent (G0) phase or chemotherapy with antiangiotherapy [12,31].
In this model the baseline fitness is set to 1, and other parameters used to define the measure of fitness in Tables 6 and 7 are given by:
e cost of producing the cytotoxin
f disadvantage of being affected by the cytotoxin
g benefit of harming other cells
h cost of resistance to the cytotoxin
i cost of proangiogenic factor production
j beneficial effect of receiving the growth factor
r external resources stimulating growth (e.g proangiogenic growth factors)
c external cytotoxic resources (e.g. cytotoxic drugs).
Once more, introduction of resource
To achieve quadruple equilibrium following relations should be satisfied:
Therefore, for a polymorphism (coexistence) of all strategies, each frequency should be contained within the interval (0, 1). It has to be added that calculated formulas for frequencies could be applied only when the above-mentioned conditions are satisfied. In other cases the results could lead to an equilibrium point which may be either an attractor or a repeller, to any other than quadrupled stable polymorphism, to monomorphism or even to unstable populations. To track the evolution of different phenotypes in the population, it is feasible to simulate equations for replicator dynamics [16]. They show how frequencies of different strategies change over time, thereby influencing the composition of the studied population. Some examples of the phase portraits (since
For inference analysis in this game, the result when all phenotypes coexist is taken as a reference (Figure 10). Relatively to this result the cost of cytotoxic production has been increased by 0.1 and equals to the benefits from harming the neighbors. Similarly, the adaptation of
An important finding is that the four-phenotype model implies third-order dynamics of replication which enables the existence of complex dynamical behaviours including strange attractors. This may be an important hallmark of the evolutionary game theory analysis. A similar model could be used to describe one of the important processes related to communication within cell population resulting from therapy. One very example of such processes analysed by us is the so-called radiation-induced bystander effect. Despite complex analysis of this model due to numerous parameters and relations between cells, the model gives a finite number of diverse results. One of them is the possibility of stable coexistence between different tumor cells within the population. In a similar way, other strategies could be added. However, the graphical representation of the results might be a limitation here. This constraint results from a maximum number of strategies equal to four. What is more, any additional strategies (phenotypes), added to the model, would increase internal dependency between parameters, strategies and results. The machinery of EGT supported by the replication dynamics enables analysis of the evolution of phenotype structure in time within cell populations; nevertheless, it gives no information about the spatial distribution of these phenotypes in tumors. Such possibilities are created by the methodology of spatial evolutionary games theory (SEGT) which enables a study of players' allocation. The lack of perfect mixing is a crucial difference between non-spatial and spatial models. Figure 13 is a spatial counterpart of phase portrait presented on Figure 10 for different reproduction schemes. Within the reference results the neutral phenotype
The idea of multilayer space evolutionary games which has been presented above allows to take into account heterogeneity of cancer cells. It leads to the conclusion that cancer cells should be considered as representing various phenotypes at the same time described by the frequency of occurrences. This entails more results in the meaning of different spatial structures; however, the analysis could be a bit more difficult. This is why we propose an additional way of graphical representation which takes the average value of players' phenotypes from entire lattice (similar to time charts from the non-spatial game).
Introduction of external resources may result in qualitative changes in the phenotype evolution. In these examples we assume
We have proposed a new approach to modelling the effect of interventions on the cancer population evolution based on the evolutionary games with resources. More precisely we apply the multilayer spatial evolutionary games with additional layers representing the evolution of external resources which may describe both the stimulating and the cell-killing treatments. We have posed essentially a question if the genetic heterogeneity in cancer cells exist ab initio or as a result of interventions. Our results, although still far from being complete, indicate that genetic heterogeneity in cancer cells evolve as a result of interventions. However, our response is only partial. More precisely, we can only state that it depends on the character of the intervention. Presented analysis shows that external interventions, depending on its nature are influencing the heterogeneity. One possibility is to change through the change in the baseline of payoff matrix, it means the change of baseline cost of the meeting. The other is an introduction of new phenotypes into the population. This can lead to the conclusion that treatment can generate new phenotypes of cancer cells. This fact means that the initial heterogeneity, that was observed before intervention, in the result of treatment can change producing for example a new, possibly more malignant, cancer subtype. The multilayer approach enables modelling cancer cells as representing different phenotypes at the same time described by a frequency of occurrences. It corresponds to the biologically observed effect of heterogeneity of cancer cells not only at the population level but also at the level of single cells. Moreover, by comparing simulation results for deterministic and probabilistic reproductions, we can study differences in the case of a direct exposure of the cancer cells to external actions and an indirect exposure caused by bystander effect. Mostly the probabilistic reproduction reflects behaviour of the mean field model, as this type of reproduction replace the bystander effect in population individuals. Meanwhile, the deterministic reproduction reflects direct impact of specific subpopulation and results in different model behaviour that the mean field.
To our knowledge, in literature, such discussion is almost absent. An additional key factor (for game theory applications) has been studied by [6], being the impact of the ecosystem or the interactions between tumour cells and their environment. They have already modelled changes in the cancer ecosystem in the context of different anti-cancer therapeutic strategies. They indicate that elimination of as many cancer cells as possible may not be essentially the best strategy and have found that destroying only some fraction of the cancer cells (with a particular phenotype) may be far more efficient. Kaznatchev et al. [18] is yet another paper in which anticancer treatment scheduling and its effect on the evolution of cancer heterogeneity is considered. The authors focus on the inter-dependence of growth-factor production and acidification and study the resulting mean-field dynamics using the double goods game with three phenotypic strategies. In our study, we are mainly interested in spatial effects, and the number of phenotypes is not an essential constraint. Nevertheless, we also analyze the mean-field case including four phenotypic strategies. Of course, it is an open question as to how to determine a kind of game that cancer cells play and how to experimentally find parameters for game models.
This work was supported by the Polish National Science Centre Grants no. DEC-2016/21/B/ST7/02241 (AS), UMO-2015/19/B/ST7/02984 (MKi) the Institute of Automatic Control, Silesian University of Technology under Grant no. BK-204/RAU1/2017, t.3 (MKr) and the National Centre for Research and Development Grant no. STRATEGMED2/267398/4/NCBR/2015 (MILESTONE - Molecular diagnostics and imaging in individualized therapy for breast, thyroid and prostate cancer) (DB).
Calculations were performed on the Ziemowit computer cluster in the Laboratory of Bioinformatics and Computational Biology, created in the EU Innovative Economy Programme POIG.02.01.00-00-166/08 and expanded in the POIG.02.03.01-00-040/13 project.
The authors declare there is no conflict of interest.
[1] | A. R. A. Anderson, K. A. Rejniak, P. Gerlee and V. Quaranta, Microenvironment driven invasion: a multiscale multimodel investigation, J. Math. Biol., 58 (2009), 579–624. |
[2] | L. A. Bach, D. J. T. Sumpter, J. Alsner and V. Loeschcke, Spatial evolutionary games of interaction among generic cancer cells, J. Theor. Med., 5 (2003), 47–58. |
[3] | L. Bach, S. Bentzen, J. Alsner and F. Christiansen, An evolutionary-game model of tumour-cell interactions: possible relevance to gene therapy, Eur. J. Cancer., 37 (2001), 2116–2120. |
[4] | D. Basanta, H. Hatzikirou and A. Deutsch, Studying the emergence of invasiveness in tumours using game theory, Eur. Phys. J. B, 63 (2008), 393–397. |
[5] | D. Basanta, J. G. Scott, M. N. Fishman, G. Ayala, S. W. Hayward and A. R. A. Anderson, Investigating prostate cancer tumour-stroma interactions: clinical and biological insights from an evolutionary game, Br. J. Cancer., 106 (2012), 174–181. |
[6] | D. Basanta and A. R. A. Anderson, Exploiting ecological principles to better understand cancer progression and treatment, Interface focus, 3 (2013), 20130020. |
[7] | D. Basanta, R. A. Gatenby and A. R. A. Anderson, Exploiting evolution to treat drug resistance: combination therapy and the double bind, Mol. Pharm., 9 (2012), 914–921. |
[8] | D. Basanta, J. G. Scott, R. Rockne, K. R. Swanson and A. R. A. Anderson, The role of IDH1 mutated tumour cells in secondary glioblastomas: an evolutionary game theoretical view, Phys. Biol., 8 (2011), 015016. |
[9] | D. T. Bishop and C. Cannings, A generalized war of attrition, J. Theor. Biol., 70 (1978), 85–124. |
[10] | K. Bohl, S. Hummert, S. Werner, D. Basanta, A. Deutsch, S. Schuster, G. Theissen and A. Schroeter, Evolutionary game theory: molecules as players, Mol. Biosyst., 10 (2014), 3066–3074. |
[11] | D. Dingli, F. A. C. C. Chalub, F. C. Santos, S. Van Segbroeck and J. M. Pacheco, Cancer phenotype as the outcome of an evolutionary game between normal and malignant cells, Br. J. Cancer., 101 (2009), 1130–1136. |
[12] | M. Dolbniak, M. Kardynska and J. Smieja, Sensitivity of combined chemo-and antiangiogenic therapy results in different models describing cancer growth, Discrete Cont. Dyn. - B, 23 (2018), 145–160. |
[13] | J. C. Fisher, Multiple-mutation theory of carcinogenesis, Nature, 181 (1958), 651–652. |
[14] | R. A. Gatenby and T. L. Vincent, An evolutionary model of carcinogenesis, Cancer Res., 63 (2003), 6212–6220. |
[15] | M. Gerstung, N. Eriksson, J. Lin, B. Vogelstein and N. Beerenwinkel, The temporal order of genetic and pathway alterations in tumorigenesis, PloS one, 6 (2011), e27136. |
[16] | J. Hofbauer, P. Schuster and K. Sigmund, A note on evolutionary stable strategies and game dynamics, J. Theor. Biol., 81 (1979), 609–612. |
[17] | S. Hummert, K. Bohl, D. Basanta, A. Deutsch, S. Werner, G. Theissen, A. Schroeter and S. Schuster, Evolutionary game theory: cells as players, Mol. BioSyst., 10 (2014), 3044–3065. |
[18] | A. Kaznatcheev, R. Vander Velde, J. G. Scott and D. Basanta, Cancer treatment scheduling and dynamic heterogeneity in social dilemmas of tumour acidity and vasculature, Br. J. Cancer., 116 (2017), 785–792. |
[19] | M. Krzeslak, D. Borys and A. Swierniak, Angiogenic switch - mixed spatial evolutionary game approach, Intell. Inf. Database Syst., 9621 (2016), 420–429. |
[20] | M. Krzeslak and A. Swierniak, Multidimensional extended spatial evolutionary games, Comput. Biol. Med., 69 (2016), 315–327. |
[21] | M. Krzeslak and A. Swierniak, Spatial evolutionary games and radiation induced bystander effect, Arch. Control Sci., 21. |
[22] | J. M. Lasry and P. L. Lions, Mean field games, Jpn J. Math., 2 (2007), 229–260. |
[23] | L. A. Loeb, Mutator phenotype may be required for multistage carcinogenesis, Cancer Res., 51 (1991), 3075–3079. |
[24] | M. Mahner and M. Kary, What exactly are genomes, genotypes and phenotypes? And what about phenomes? J. Theor. Biol., 186 (1997), 55–63. |
[25] | Y. Mansury, M. Diggory and T. S. Deisboeck, Evolutionary game theory in an agent-based brain tumor model: exploring the 'genotype-phenotype' link, J. Theor. Biol., 238 (2006), 146–156. |
[26] | L. M. F. Merlo, J.W. Pepper, B. J. Reid and C. C. Maley, Cancer as an evolutionary and ecological process, Nat. Rev. Cancer, 6 (2006), 924–935. |
[27] | J. V. Neumann, Theory of Self-Reproducing Automata, University of Illinois Press, Champaign, IL, USA, 1966. |
[28] | K. Sigmund and M. A. Nowak, Evolutionary game theory, Curr. Biol., 9 (1999), R503–R505. |
[29] | J. M. Smith and G. R. Price, The logic of animal conflict, Nature, 246 (1973), 15–18. |
[30] | J. M. Smith, Evolution and the theory of games. |
[31] | A. Swierniak, M. Kimmel, J. Smieja, K. Puszynski and K. Psiuk-Maksymowicz, System Engineering Approach to Planning Anticancer Therapies, Springer International Publishing, 2016. |
[32] | A. Swierniak and M. Krzeslak, Application of evolutionary games to modeling carcinogenesis, Math. Biosci. Eng., 10 (2013), 873–911. |
[33] | A. Swierniak and M. Krzeslak, Cancer heterogeneity and multilayer spatial evolutionary games., Biol. Direct, 11 (2016), 53. |
[34] | A. Swierniak, M. Krzeslak, S. Student and J. Rzeszowska-Wolny, Development of a population of cancer cells: Observation and modeling by a mixed spatial evolutionary games approach, J. Theor. Biol., 405 (2016), 94–103. |
[35] | I. P. Tomlinson and W. F. Bodmer, Failure of programmed cell death and differentiation as causes of tumors: some simple mathematical models., Proceedings of the National Academy of Sciences of the United States of America, 92 (1995), 11130–11134. |
[36] | I. Tomlinson, Game-theory models of interactions between tumour cells, Eur. J. Cancer., 33 (1997), 1495–1500. |
[37] | I. Tomlinson and W. Bodmer, Modelling the consequences of interactions between tumour cells, Br. J. Cancer., 75 (1997), 157–160. |
1. | Andrzej Swierniak, Michal Krzeslak, Damian Borys, 2019, Games with resources and their use in modeling control processes in heterogeneous populations, 978-1-7281-0933-6, 530, 10.1109/MMAR.2019.8864644 | |
2. | Jose M. Sanz Nogales, S. Zazo, An evolutionary dynamics model for metastatic tumour growth based on public goods games, 2021, 99, 10075704, 105783, 10.1016/j.cnsns.2021.105783 | |
3. | Jessica Cunningham, Frank Thuijsman, Ralf Peeters, Yannick Viossat, Joel Brown, Robert Gatenby, Kateřina Staňková, Paul K. Newton, Optimal control to reach eco-evolutionary stability in metastatic castrate-resistant prostate cancer, 2020, 15, 1932-6203, e0243386, 10.1371/journal.pone.0243386 | |
4. | Arin Nam, Atish Mohanty, Supriyo Bhattacharya, Sourabh Kotnala, Srisairam Achuthan, Kishore Hari, Saumya Srivastava, Linlin Guo, Anusha Nathan, Rishov Chatterjee, Maneesh Jain, Mohd W. Nasser, Surinder Kumar Batra, Govindan Rangarajan, Erminia Massarelli, Herbert Levine, Mohit Kumar Jolly, Prakash Kulkarni, Ravi Salgia, Dynamic Phenotypic Switching and Group Behavior Help Non-Small Cell Lung Cancer Cells Evade Chemotherapy, 2021, 12, 2218-273X, 8, 10.3390/biom12010008 | |
5. | Guim Aguadé-Gorgorió, Stuart Kauffman, Ricard Solé, Transition Therapy: Tackling the Ecology of Tumor Phenotypic Plasticity, 2022, 84, 0092-8240, 10.1007/s11538-021-00970-9 | |
6. | Chixiang Chen, Libo Jiang, Biyi Shen, Ming Wang, Christopher H. Griffin, Vernon M. Chinchilli, Rongling Wu, A Computational Atlas of Tissue-specific Regulatory Networks, 2021, 1, 2674-0702, 10.3389/fsysb.2021.764161 | |
7. | Annick Laruelle, André Rocha, Claudia Manini, José I. López, Elena Inarra, Effects of Heterogeneity on Cancer: A Game Theory Perspective, 2023, 85, 0092-8240, 10.1007/s11538-023-01178-9 |
Phenotypes | Hawk | Dove |
Hawk | v-c | 2v |
Dove | 0 | v |
Phenotypes | Hawk | Dove |
Hawk | v-c | 2v |
Dove | r(v/2) | (1+r)v |
r=0 | r=1 | |
H, H | v-c | v-c |
D, D | v | 2v |
D, H | 0 | v/2 |
H, D | 2v | 2v |
Phenotypes | A+ | A- |
A+ | 1-i+j | 1-i+j |
A- | 1+j | 1 |
Phenotypes | A+ | A- |
A+ | 1-i+j+r/2 | 1-i+j+r/2 |
A- | 1+j+r/2 | 1+r/2 |
Phenotypes | A | P | Q | R |
A | 1-i+j | 1-i+j-f | 1-i+j | 1-i+j |
P | 1+j-e+g | 1-f-e+g | 1-e | 1-e+g |
Q | 1+j-h | 1-h | 1-h | 1-h |
R | 1+j | 1-f | 1 | 1 |
Phenotypes | A | P | Q | R |
A | 1-i+j+r/2-c/2 | 1-i+j-f+r/2-c/2 | 1-i+j+r/2-c/2 | 1-i+j+r/2-c/2 |
P | 1+j-e+g+r/2-c/2 | 1-f-e+g+r/2-c/2 | 1-e+r/2-c/2 | 1-e+g+r/2-c/2 |
Q | 1+j-h+r/2 | 1-h+r/2 | 1-h+r/2 | 1-h+r/2 |
R | 1+j+r/2-c/2 | 1-f+r/2-c/2 | 1+r/2-c/2 | 1+r/2-c/2 |
Phenotypes | Hawk | Dove |
Hawk | v-c | 2v |
Dove | 0 | v |
Phenotypes | Hawk | Dove |
Hawk | v-c | 2v |
Dove | r(v/2) | (1+r)v |
r=0 | r=1 | |
H, H | v-c | v-c |
D, D | v | 2v |
D, H | 0 | v/2 |
H, D | 2v | 2v |
Phenotypes | A+ | A- |
A+ | 1-i+j | 1-i+j |
A- | 1+j | 1 |
Phenotypes | A+ | A- |
A+ | 1-i+j+r/2 | 1-i+j+r/2 |
A- | 1+j+r/2 | 1+r/2 |
Phenotypes | A | P | Q | R |
A | 1-i+j | 1-i+j-f | 1-i+j | 1-i+j |
P | 1+j-e+g | 1-f-e+g | 1-e | 1-e+g |
Q | 1+j-h | 1-h | 1-h | 1-h |
R | 1+j | 1-f | 1 | 1 |
Phenotypes | A | P | Q | R |
A | 1-i+j+r/2-c/2 | 1-i+j-f+r/2-c/2 | 1-i+j+r/2-c/2 | 1-i+j+r/2-c/2 |
P | 1+j-e+g+r/2-c/2 | 1-f-e+g+r/2-c/2 | 1-e+r/2-c/2 | 1-e+g+r/2-c/2 |
Q | 1+j-h+r/2 | 1-h+r/2 | 1-h+r/2 | 1-h+r/2 |
R | 1+j+r/2-c/2 | 1-f+r/2-c/2 | 1+r/2-c/2 | 1+r/2-c/2 |