
MBW:Tumor ModelingFrom MathBioArticle review and some more ideas by Jason Hammond, Michael Kochen, and Sekson Sirisubtawee Kirschner, Denise, and John C. Panetta (1998). Modeling immunotherapy of the tumorimmune interaction. Journal of Mathematical Biology 37, 235252.a
ContentsBackgroundAdvances in the knowledge and treatment of cancer have been numerous in the decade since this paper was published. Even so, conventional chemotherapy and radiation remain widely prescribed by oncologists, and although there are exceptions, they do little for the majority of cancers^{[1]}. Fortunately there are newer treatment options that are either available today, or will be in the coming years. This paper models one of the more promising ones. Adoptive cellular immunotherapy (ACI) is a type of cancer immunotherapy. The treatment begins by collecting immune cells from the patient. Cells that have anticancer properties are then cultured in the lab. The newly grown cells are subsequently injected back into the patient to fight the cancer. Interleukin2 (IL2) is a cytokine, a signaling molecule produced by immune cells that activates, and stimulates growth and differentiation of Tcells, and other immune system cells. The immune system model constructed in this paper first explores the dynamics between immune cells, tumor cells and IL2, and then considers treatment with ACI, and additional IL2.a An extension we make is to use chemotherapy in addition to the immunotherapy in an attempt to find a more efficient technique of erraticating the tumor cells. (See APPM4390:Mathematical Immunology for full review of this article). NonTreatment ModelVariables
Variables The parameters that were used in the paper as well as by us in our analysis can be seen on this table: ODEs
These are the parameters of note. The values for these, and the rest of the parameters are given in table 1. on page 238. Notice the logistic growth term in the second equation indicating limited growth of the tumor. Also note that each equation has a term in MichaelisMenten form indicating limited IL2, immune, and tumor responses. The antigenicity will be varied in the next section, to see what affect it has on the tumor.
The bifurcation diagram of tumor size vs. antigenicity is given on page 240 of the paper. There are two bifurcation points, one at and the other at 0.032. Tumor behavior for the three different regions of c are demonstrated in the following figures. Figure 1. shows the tumor volume as time progresses (in days) for . The tumor grows quickly to near carrying capacity and then remains there. Figure 2. is more interesting. In the region 8 there are cycles of tumor growth and abatement. The amplitude and period depend on where c is in the given region. Its thought that this might explain the recurrence of tumors sometimes years after being treated. Figure 3. shows tumor progression when c > 0.032. Tumor size goes through damped oscillations leading to a relatively small stable tumor that is not cleared from the body. Treatment ModelIn the treatment model, two parameters, and , are added to the system of ODEs. represents treatment with ACI, and with IL2.
Here the authors explore these treatments alone, and in combination. They employ the use of linear stability analysis in which the Jacobian matrix is given by:
ACI only and Through linear stability analysis the authors found that at there is a transcritical bifurcation. To do this they first find the equilibrium with a zero value for the tumor, which is:
The Jacobian for this equilibrium is given by:
and its eigenvalues are We can see that the bifurcation point is then found by setting , which gives us that
and by looking at the eigenvalues we can see that the equilibrium is locally stable if and unstable if . With more analysis the authors created the following bifurcation diagram from page 244 of the paper: In this figure the authors depict five regions. They describe how the solutions behave in each of the five regions. I have provided the next plot to illustrate examples in 4 of the 5 regions.
IL2 only The authors did linear stability analysis again with and To do this they first find the equilibrium with a zero value for the tumor, which is:
The Jacobian for this equilibrium is given by:
and its eigenvalues are We can see that the bifurcation point is then found by setting , which gives us that
and by looking at the eigenvalues we can see that the equilibrium is always unstable since . Below the critical number, we see that if is small, then behavior is similar to the nontreatment model. Above the critical value the IL2 treatment succeeds in destroying the tumor, but effector cell levels grow uncontrollably. This also has harmful effects on the body and is something to be avoided. With more analysis the authors created the following bifurcation diagram from page 244 of the paper: In this figure the authors depict four regions and describe how the solutions behave in each of the regions. I have provided the next plot to illustrate examples in all of the regions.
ACI IL2 Combination The authors did linear stability analysis again with and To do this they first find the equilibrium with a zero value for the tumor, which is:
When they required negative eigenvalues of the Jacobian for this equilibrium they obtained the following conditions for stability:
and
The next figure was created using the two conditions on and In this figure we replicate the plots in figure 6 in the paper on page 248. The solutions in each region are given in the paper. I have provided the next plot to illustrate examples in all of the regions.
Interpretation/Analysis from the paperThe nontreatment model showed that low antigenicity tumors will grow to their carrying capacity. This simply means that the immune system doesn't 'see' the tumor and therefore doesn't respond to it. This suggest an avenue of attack against the tumor, train the immune system to see the cancer for what it is. There are vaccines in development for this, but none currently available. Moderately antigenic tumors show periodic behavior. On the lower end of this scale the amplitude and period of the oscillations are large, with periods sometimes years long. As the antigenicity gets higher the amplitude and period both shorten. Periodic behavior could account for the recurrence of tumors years after being treated. Highly antigenic tumors exibit damped oscillations, eventually becoming small dormant tumors. In this model even these tumors are never cleared from the body without treatment. A possible real world example of this would be a benign tumor. The model for ACI therapy gave impressive results. For moderate to high antigenic tumors This treatment completely cleared the cancer. For low antigenic tumors it was dependent on initial conditions. This would imply that the combination of ACI and early detection would be a powerful way to attack tumors. The IL2 therapy model wasn't so impressive. To clear the tumor with IL2 would take so much of it that it would cause effector cells to grow uncontrolled, causing other problems. The combined ACI/IL2 would work but only when using small amounts of IL2 to amplify the effects of ACI. Our New Model (With Chemotherapy)We now propose a new therapy which combines chemotherapy with the immunotherapy that was described in the paper (for a detailed discussion of optimal chemotherapy treatment see: MBW:Optimal Chemotherapy Strategies). The chemotherapy adds one more equation to the system as well as a few terms. The differential equation for represents the rate at which the chemotherapy drug is administered as well as the rate at which the body naturally clears the drug. The terms added to the Effector cell and Tumor cell equations represent the annihilation of the tumor and effector cells by the drug. We derived our idea by seeing the model given in [3] The system we suggest to model this reaction is the following:
For the purposes of this lab we only analyze the case when , a constant. We realize that this is realistically not feasible since it represents a constant supply of the drug over the whole time. However we hope to obtain results that we could use to determine a good scheme for pulsed dosing( administering chemotherapy for 412hrs biweekly or monthly) Just Chemo: In this case we find the effects of administering chemotherapy while and . In this case the only equilibrium we obtain with no tumor is the trivial equilibrium: Through experimentation we notice an interesting relationship between the behavior of the solutions and the values of and . With more time we would attempt to make a bifucation diagram between the two variables. With we are just at the nontreatment case in which there were two bifurcation points in . One, , at which solutions change from tending to a stable nonzero equilibrium to tending to a stable limit cycle, and the other, , at which solutions go from tending to stable limit cycles to tending to stable spiral nodes. We noticed through experimentation that as we add more to the system the bifurcation point decreases. And there is a bifurcation at at which solutions go from tending to a nontrivial to steady state to tending to the trivial one. The next figure shows some example solutions for various combinations of parameters. Chemo with Immunotherapy:
Here is the Jacobian matrix of the system :
and its eigenvalues are
If then and . The next two figures show the conditions on , , and for stability. The next figure shows sample solutions for each of the regions. Special case with Chemo: The next picture shows the condition on and for stability. The curve was found by setting the eigenvalues of the Jacobian less than zero and obtaining the following condition for stability:
Project categorizationMathematicsThis model uses a system of three ordinary differential equations to model cancer development and treatment and steady state analyses are carried through by studying eigenvalues in different bifurcation regions of the system. Model typeThe system of ODEs are actually modeling three different magnitudes, that is bulk tumor size, number of immune system cells, and IL2 (a protein) concentration. By this, the model is a combined system which looks at input on one parameter (cells) and output of two others (tumor size and IL2 concentration). In general, since tumor size can be measured by the number of the cells, this system can be characterized as a cell population model. Biological systemThe biological system studied is cancer cells and different treatments (cells and chemotherapeutics) in a mammalian body environment. Additional paper discussionThe paper "Modeling subspecies and the tumorimmune system interaction: Steps toward understanding therapy" by S.A. Menchón, R.A Ramos and C.A. Condat (Physica A 2007; 386: 713719) cites the above discussed paper. This paper extands the system from above to simulate cancer development in 2D grid where each grid element contains several healthy, cancerous and dead cells. At the south end of the grid, a constant nutrient supply (a blood vessel) is positioned. The model then simulates cancer cell nutrition uptake and consumption as a function of grid position. Next, a single cell mutation is introduced in the tumor to develop a second independent cancer tumor. The development is then simulated and again analyzed as a function grid position (or distance from the nutrition supply). Finally, diffusible immunesystem cells are introduced in the blood vessel, and impact on the tumor development is simulated and analyzed. This paper is nine years newer then the paper discussed on this page. We se that the models used in the late paper are developments of the models in the early paper, and that the late paper model try to simulate the situation for a tumor in a human body on a more precise level. Still, a 3D simulation would probably give even more accurate results compared to the real situation, but this again would lead to an expansion of the above discussed model. External LinksAmerican Cancer SocietyImmunotherapy National Cancer InstituteBiological Therapy
References
[2]Kirschner, Denise, and John C. Panetta (1998). Modeling immunotherapy of the tumorimmune interaction. Journal of Mathematical Biology 37, 235252. [3] L.G. de Pillis and A.E. Radunskaya, "A Mathematical Tumor Model with Immune Resistance and Drug Therapy: An Optimal Control Approach", Journal of Theoretical Medicine, Vol. 3, pp.79100, 2001. 