Growth models with stochastic differential equations. An example from tumor immunology (Q1083391)

The author considers a stochastic differential equation to describe the growth of a tumor under influence of internal and external random fluctuations. Let X be the number of tumor cells per unit volume, suitably normalized, with \(0\leq X\leq 1/\vartheta\), \(\alpha\) the rate of transformation of normal into neoplastic cells, \(\beta\) the degree of immunological interaction. Furthermore, let \(\omega^ 2\) and \(\tau^ 2\) be the measures of the external (environmental) and internal (survival ability, fertility) fluctuations, respectively. The proposed equation \((B=s\tan dard\) Wiener process) is \[ dX=((1-\vartheta X)(\alpha +X)- \beta (1+y)^{-1}y)dt+\sqrt{\omega^ 2X^ 2(1-\vartheta X)^ 2+\tau^ 2X(1-\vartheta X)}dB. \] This equation is derived as diffusion limit of a branching process with random environment. The properties of the limit distributions \[ F(X)=\lim_{t\to \infty}P(X(t)\leq x)\quad and\quad G(x)=\lim_{T\to \infty}T^{-1}\int^{T}_{0}P(X(t)\leq x)dt, \] resp. their densities f and g in relationship to the parameters are the following: 1) There is a \(\vartheta_ c(\approx 1)\), such that f has a maximum in \(0<x<\vartheta^{-1}\) for \(\vartheta \geq \vartheta_ c\) and two maxima for \(\vartheta <\vartheta_ c\) and suitable \(\beta\). 2) For small internal variations \((0<\tau^ 2<2\alpha)\) there is an interval \((\beta_ 1,\beta_ 2)\), such that the local maximum of f is in the first half of \((0,\vartheta^{-1})\) for \(\beta >\beta_ 2\), in the second for \(\beta <\beta_ 1\) and in both (two maxima) for \(\beta_ 1<\beta <\beta_ 2\). Furthermore, \(f(\vartheta^{-1})=\infty\) for \(\beta <\beta_ c\leq \beta_ 1.\) 3) For large internal variations, \(\tau^ 2>2\alpha\), we have \(G(0+)>0\), \(g(0+)=\infty\); \(g(\vartheta^{-1})=0\) for \(\beta >\beta_ c\); \(g(\vartheta^{-1})=\infty\) for \(\beta <\beta_ c\); g has one (no) local maximum, if \(\beta\in (\not\in) (\beta_ c,\beta_ 0)\) for some \(\beta_ 0.\) These results may explain the role of random fluctuations for an early elimination or standstill of a tumor.