Research article

Analysis of a free boundary problem for vascularized tumor growth with time delays and almost periodic nutrient supply

  • Received: 01 December 2023 Revised: 08 March 2024 Accepted: 19 March 2024 Published: 10 April 2024
  • MSC : 34K13, 35Q92

  • In this research, we have proposed and investigated a time-delayed free boundary problem concerning tumor growth in the presence of almost periodic nutrient supply with angiogenesis. This study primarily focused on examining the impact of almost periodic nutrient supply, angiogenesis, and time delay on tumor growth dynamics. We analyzed the existence, uniqueness, and exponential stability of almost periodic solutions. Furthermore, we established conditions for the disappearance of almost periodic oscillations in tumors. The existence and uniqueness of almost periodic solutions were proven, while sufficient conditions for the exponential stability of the unique solution were established. Finally, computer simulations were employed to illustrate our results.

    Citation: Shihe Xu, Zuxing Xuan, Fangwei Zhang. Analysis of a free boundary problem for vascularized tumor growth with time delays and almost periodic nutrient supply[J]. AIMS Mathematics, 2024, 9(5): 13291-13312. doi: 10.3934/math.2024648

    Related Papers:

    [1] Zijing Ye, Shihe Xu, Xuemei Wei . Analysis of a free boundary problem for vascularized tumor growth with a time delay in the process of tumor regulating apoptosis. AIMS Mathematics, 2022, 7(10): 19440-19457. doi: 10.3934/math.20221067
    [2] Ping Zhu . Dynamics of the positive almost periodic solution to a class of recruitment delayed model on time scales. AIMS Mathematics, 2023, 8(3): 7292-7309. doi: 10.3934/math.2023367
    [3] Huiyan Peng, Xuemei Wei . Qualitative analysis of a time-delayed free boundary problem for tumor growth with Gibbs-Thomson relation in the presence of inhibitors. AIMS Mathematics, 2023, 8(9): 22354-22370. doi: 10.3934/math.20231140
    [4] Qi Shao, Yongkun Li . Almost periodic solutions for Clifford-valued stochastic shunting inhibitory cellular neural networks with mixed delays. AIMS Mathematics, 2024, 9(5): 13439-13461. doi: 10.3934/math.2024655
    [5] Yongkun Li, Xiaoli Huang, Xiaohui Wang . Weyl almost periodic solutions for quaternion-valued shunting inhibitory cellular neural networks with time-varying delays. AIMS Mathematics, 2022, 7(4): 4861-4886. doi: 10.3934/math.2022271
    [6] Ramazan Yazgan . An analysis for a special class of solution of a Duffing system with variable delays. AIMS Mathematics, 2021, 6(10): 11187-11199. doi: 10.3934/math.2021649
    [7] Xingxiao Wu, Lidong Huang, Shan Zhang, Wenjie Qin . Dynamics analysis and optimal control of a fractional-order lung cancer model. AIMS Mathematics, 2024, 9(12): 35759-35799. doi: 10.3934/math.20241697
    [8] Yanshou Dong, Junfang Zhao, Xu Miao, Ming Kang . Piecewise pseudo almost periodic solutions of interval general BAM neural networks with mixed time-varying delays and impulsive perturbations. AIMS Mathematics, 2023, 8(9): 21828-21855. doi: 10.3934/math.20231113
    [9] Ardak Kashkynbayev, Moldir Koptileuova, Alfarabi Issakhanov, Jinde Cao . Almost periodic solutions of fuzzy shunting inhibitory CNNs with delays. AIMS Mathematics, 2022, 7(7): 11813-11828. doi: 10.3934/math.2022659
    [10] Xin Liu, Yan Wang . Averaging principle on infinite intervals for stochastic ordinary differential equations with Lévy noise. AIMS Mathematics, 2021, 6(5): 5316-5350. doi: 10.3934/math.2021314
  • In this research, we have proposed and investigated a time-delayed free boundary problem concerning tumor growth in the presence of almost periodic nutrient supply with angiogenesis. This study primarily focused on examining the impact of almost periodic nutrient supply, angiogenesis, and time delay on tumor growth dynamics. We analyzed the existence, uniqueness, and exponential stability of almost periodic solutions. Furthermore, we established conditions for the disappearance of almost periodic oscillations in tumors. The existence and uniqueness of almost periodic solutions were proven, while sufficient conditions for the exponential stability of the unique solution were established. Finally, computer simulations were employed to illustrate our results.



    Delay differential equations (DDEs), or functional differential equations, arise in models representing biological phenomena when considering the time-delays occurring in these phenomena. Mathematical modeling using such DDEs is widely applied for analysis and predictions in various areas of life sciences, including population dynamics, epidemiology, immunology, tumor growth, physiology, and neural networks. The memory or time-delays in these models are associated with the duration of hidden processes such as life cycle stages, the time between cell infection and new virus production, the infection period, the time between cell division and new cell production, and the immune period [3,6,10,15,17,18,20,23,24,25]. Reference [17] covers important topics related to DDEs including numerical methods, stability analysis, inverse problems, parameter estimation, sensitivity analysis, optimal control, and time-delayed biological systems. In this paper, we investigate a free boundary problem for vascularized tumor growth with time delays and almost periodic nutrient supply. The mathematical model describing the tumor growth process considers cell division and death along with external almost periodic nutrient supply. Compared to the apoptosis process of tumor cells, the proliferation process exhibits a time delay. In the model, two unknown functions σ(r,t) and R(t) represent nutrient concentration and tumor radius, respectively. The mathematical model is given by:

    cσt=1r2r(r2σr)Γσ,0<r<R(t),t>0, (1.1)
    σr+α(σψ(t))=0,r=R(t),t>0, (1.2)
    ddt(4πR3(t)3)=4π(R(tτ)0sσ(r,tτ)r2drR(t)0s˜σr2dr),t>0, (1.3)

    where Γσ represents the consumption rate of nutrients, and α is a constant denoting the density of blood vessels. The external concentration of nutrients is denoted by ψ(t), while τ represents the time delay. Equation (1.3) originates from the law of conservation of mass. The term 4πR(tτ)0sσ(r,tτ)r2dr corresponds to the volume increase induced by cell proliferation, where sσ denotes the proliferation rate. On the other hand, 4πR(t)0s˜σr2dr accounts for the volume decrease caused by natural death, assuming a natural death rate of s˜σ. Additionally, we consider that c=Td/Tg1minute1day1, which represents the ratio between the nutrient diffusion timescale and tumor growth timescale (see [10,11] for further details). In this study, we will discuss the aforementioned model with respect to its initial condition given in Eq (1.4):

    R(t)=φ(t),τt0. (1.4)

    The proposed model is based on the framework presented in [12] with two modifications. First, we consider the provision of external nutrients as an almost periodic function, which is a more realistic assumption compared to the constant nutrient supply assumed in [12]. Second, we incorporate the impact of time delay in tumor cell proliferation, as observed in [3]. It is important to analyze the stability of tumor growth models with time-delay terms, and several methods such as Lyapunov exponents, the comparison principle, and stability theorems have been proposed by scholars (see for instance [6,10,14,18,22]). In particular, reference [6] investigates a special case where α= and ψ is a positive constant using functional differential equations theory to establish existence, uniqueness, and asymptotic stability of steady-state solutions. Furthermore, researchers have also studied bifurcation phenomena in mathematical models for tumor growth with time-delay terms (e.g., [15,16,24,26]), which are crucial for understanding tumor development mechanisms and predicting future trends. By considering almost periodic functions instead of exact periodicity due to their robustness under perturbations, our model provides a more realistic representation of actual tumor growth dynamics. This paper focuses on investigating the impact of almost periodic nutrient supply along with angiogenesis and time delay.

    Several studies have investigated mathematical models for tumor growth in angiogenesis, including those by Friedman and Lam [12], Ye et al. [25], and Zhou et al.[26]. However, these studies assumed a constant provision of external nutrients. In this paper, we consider the external concentration of nutrients as a bounded almost periodic function and also incorporate time delays in tumor cell division. While our previous work [25] also considered angiogenesis and time delays, it assumed a constant concentration of external nutrients. While we have previously proven the asymptotic stability of constant steady-state solutions [25], this paper investigates the case where the provision of external nutrients is almost periodic and establishes the existence, uniqueness, and stability of almost periodic solutions using a different methodology.

    Noticing that c1, this paper focuses on the limiting case where c=0. By employing spatial scale transformation, we can assume Γ=1. Consequently, Eq (1.1) simplifies to the following form:

    1r2r(r2σr)=σ,0<r<R(t),t>0. (1.5)

    Using the solution of (1.5), (1.2) is given by

    σ(r,t)=αα+Rp(R)l(r)l(R)ψ(t), (1.6)

    where

    p(x)=xcothx1x2,l(x)=sinhxx.

    Substituting (1.6) into (1.3), we obtian

    R(t)=sR(t)[αψ(tτ)α+R(tτ)p(R(tτ))R3(tτ)p(R(tτ))R3(t)˜σ3], (1.7)

    where p(x)=xcothx1x2. Denoting x=R3, after rescaling coefficients of ψ(t),˜σ as follows

    ˆψ=sψ,ˆ˜σ=s˜σ,

    and dropping the hat notation, Eq (1.7) takes the form

    x(t)=3αψ(tτ)F(x(tτ))˜σx(t), (1.8)

    where

    F(x)=xp(3x)α+3xp(3x).

    Accordingly,

    x0(t)=φ3(t),τt0. (1.9)

    The remaining part of the paper is arranged as follows. In Section 2, some preliminaries are given. In Section 3, we prove the existence and uniqueness of the almost periodic solution to Eq (1.8). Section 4 is devoted to the stability of the unique positive almost periodic solution. In the last section, computer simulations and conclusions are given.

    Let

    q(x)=xp(x)=xcothx1x,k(x)=x3p(x),g(x)=p(x)α+q(x),G(x)=x3g(x)

    and

    D(x)=p(x)p(θx),S(x)=θp(θx)(α+xp(x))p(x)(α+θxp(θx))=α+q(x)αD(x)/θ+q(x),

    where α and θ are positive constants.

    Lemma 2.1. (1) p(x)<0 for x>0, limx0+p(x)=1/3,limxp(x)=0.

    (2) q(x)>0 for x0. limx0q(x)=0, limxq(x)=1, and q(0)=1/3.

    (3) k(x)>0 and k(x)>0 for x>0.

    (4) (x3g(x))>0 and (x3g(x))>0 for x>0.

    (5) For any θ(0,1), D(x)<0 for x>0 and limx0+D(x)=1,limxD(x)=θ.

    (6) S(x)>0 for x>0. Moreover, for any θ(0,1), limx0+S(x)=θ,limxS(x)=1.

    (7) F(x)<0 for x>0.

    Proof. For the proof of (1) and (2), please see Lemmas 2.1 and 2.2 in [12]. For the proof of (3), please see [6]. Now, we prove (4)–(7).

    (3) The fact that k(x)=(x3p(x))>0 for x>0 can be found in [6]. Next, we aim to prove that k(x)>0 for x>0. Since

    (x3p(x))=2xcothxx2sinh2x1,

    it follows that, by noticing cosh2xsinh2x=1,

    (x3p(x))=2coshxsinh2x+2x2coshx4xsinhxsinh3x=2sinhx(coshxsinhxx)+2x(xcoshxsinhx)sinh3x>0.

    This result is derived from the facts:

    coshxsinhxx>0,xcoshxsinhx>0,

    for x>0.

    (4) From [23], noticing k(x)>0, it is known that:

    (x3g(x))=(k(x)α+q(x))=k(x)α+k(x)q(x)q(x)k(x)(α+q(x))2=αk(x)(α+q(x))2+(k(x)q(x))(q(x)α+q(x))2=αk(x)(α+q(x))2+2x(q(x)α+q(x))2>0.

    Then,

    (x3g(x))=(k(x)α+q(x))=(αk(x)(α+q(x))2)+(2xq2(x)(α+q(x))2)=αk(x)(α+q(x))2q(x)k(x)(α+q(x))3+2q(x)(q(x)+2xq(x))(α+q(x))3.

    By utilizing equation

    q(x)+2xq(x)=xp(x)+2x(p(x)+xp(x))=xp(x)+2xq(x)>0, (2.1)

    for x>0, we can deduce that (x3g(x))>0 for x>0.

    (5) For 0<θ<1, from [21], we know that D(x)=(p(x)p(θx))<0 for any x>0. From [5], we know that p(x)p(θx) is strictly monotone increasing if 0<θ<1, and

    limx0+p(x)p(θx)=1θ,limxp(x)p(θx)=θ2.

    Noting (1), it follows that

    limx0+p(x)p(θx)=1,limxp(x)p(θx)=limxp(x)θp(θx)=θ.

    Thus, limx0+D(x)=1 and limxD(x)=θ follows.

    (6) For 0<θ<1, from [21], we know that D(x)=(p(x)p(θx))>0 for any x>0. By direct computation, one can get

    S(x)=(α+q(x)αD(x)/θ+q(x))=αq(x)(D(x)/θ1)αD(x)(α+q(x))/θ(αD(x)/θ+q(x))2>0,

    where the facts D(x)>θ, q(x)>0, and D(x)<0 have been used. For θ(0,1), the facts that limx0+S(x)=θ,limxS(x)=1 follow from (5).

    (7) Direct computation yields

    F(x)=19y2[4g(y)+yg(y)]|y=3x,

    and

    4g(y)+yg(y)=J(y)(α+q(x))3,

    where

    J(y)=α2[4p(y)+yp(y)]+α[4p2(y)+y2p(y)p(y)2y2(p(y))2]+V(y),

    and

    V(y)=4p2q2yp(y)q(y)p(y)+2yp2q(y).

    From [20], we know that 4p(y)+yp(y)<0. Since

    4p2(y)+y2p(y)p(y)2y2(p(y))2=yp(yp+2p)2p(yp+p)2(p2+y2(p)2)<0

    and

    V(y)=4p2q+2yp(qppq)=4p2q+2yp3=p2(4q+2yp)=2yp3<0,

    it follows that J(y)<0 for y>0. Thus 4g(y)+yg(y)<0, then F(x)<0 follows. This completes the proof.

    To discuss the existence and uniqueness of almost periodic solutions, let's recall some basic introductions about the symbols and results of almost periodic functions (see [2,4,9,13,19] for more details).

    Definition 2.2. (see [4,9]) A function gC(R) is called almost periodic if for all ε>0, there exists l(ε)>0 such that any interval Ⅰ of length l(ε) contains a number A with the property that

    suptR|g(t+A)g(t)|<ε.

    The space of all the almost periodic functions is denoted by CAP(R).

    Recall that AP(X) is a Banach space with the sup norm.

    Definition 2.3. (see [9]) Let M() be an n×n continuous matrix defined on R. The linear system

    Y(t)=M(t)Y(t), (2.2)

    is said to admit an exponential dichotomy on R if there exists positive constants k,ω and a projection P such that

    Y(t)PY1(s)keω(ts),ts,
    Y(t)(IP)Y1(s)keω(st),ts,

    for a fundamental solution matrix Y(t) of (2.2).

    Lemma 2.4. (see [9]) If the linear system (2.2) admits an exponential dichotomy with a projection P, then the almost periodic system

    Y(t)=M(t)Y(t)+g(t),

    has a unique almost periodic solution Y(t) given by

    Y(t)=tY(t)PY1(s)g(s)ds+tY(t)(IP)Y1(s)g(s)ds.

    Theorem 2.5. (see [7,8,9]) Suppose that P is a normal and solid cone of a real Banach space X. Let P0 be the interior of P. Suppose further that the operator A from P0 to P0 is a nondecreasing operator. Assume that there exists a function ϕ:(0,1)×P0(0,+) such that for any ϑ(0,1) and xP0, then the following holds

    (1) ϕ(ϑ,x)>ϑ.

    (2) ϕ(ϑ,) is nondecreasing in P0.

    (3) A(ϑx)ϕ(ϑ,x)A(x).

    Assume further that there exists zP0 such that A(z)z. Then A has a unique fixed point x in P0. Moreover, for any initial x0P0, the iterative sequence defined by

    xn=A(xn1),nN, (2.3)

    satisfies

    xnx0(n). (2.4)

    Rewrite the problems (1.8) and (1.9) in the following form:

    x(t)=x0(0)e˜σt+3αt0ψ(sτ)e˜σ(ts)F(x(sτ))ds.

    Then, by the method of steps, the problems (1.8) and (1.9) have a unique solution x(t) which exists for all t0. From Lemma 2.1, it follows that F(x)0 for all x0. Then, by Theorem 1.1 in [1], it is easy to get get that the solution to problems (1.8) and (1.9) is nonnegative.

    For the remainder of the paper, we always assume that ψ(t) is a positive almost periodic function and denote

    ψ=suptRψ(t),ψ=inftRψ(t).

    By Definition 2.3 and Lemma 2.4, it is not hard to get:

    Lemma 3.1. There exists a nonnegative almost periodic solution to Eq (1.8) given by

    x(t)=3αtψ(sτ)F(x(sτ))e˜σ(ts)ds,tR. (3.1)

    Actually, Eq (1.8) is equivalent to (3.1) in the sense of nonnegative almost periodic solutions, i.e., every nonnegative almost periodic solution of Eq (1.8) is also a nonnegative almost periodic solution of (3.1), and vice versa.

    Theorem 3.2. (1) If ψ>˜σ, there exists one unique positive almost periodic solution x to Eq (1.8). Moreover, for any initial value function x0CAP(R) with positive infimum, the iterative sequence

    xk(t)=3αtψ(sτ)f(xk1(sτ))e˜σ(ts)ds,k=1,2,3... (3.2)

    satisfies

    xkx0,k. (3.3)

    (2) If ψ<˜σ, then Eq (1.8) has exactly one unique almost periodic solution which equals zero. Moreover, for any nonnegative initial value function x0CAP(R), the iterative sequence

    xk(t)=3αtψ(sτ)F(xk1(sτ))e˜σ(ts)ds,k=1,2,3... (3.4)

    satisfies

    xk0,k. (3.5)

    Proof. (1). Let

    P={xCAP(R):x(t)0,tR}.

    Then, P is a normal and solid cone in CAP(R) and its interior

    P0={xCAP(R):ε>0,suchthatx(t)>ε,tR}.

    Define an operator A on P0 in the following way:

    A(x)(t)=3αtψ(sτ)F(x(sτ))e˜σ(ts)ds. (3.6)

    The fact

    F(x)=1y2G(y)|y=3x>0, (3.7)

    implies that F is monotone increasing for x>0. It follows that A is a nondecreasing operator.

    Next, let us prove that A is from P0 to P0. Since

    limx0g(3x)=limx0p(3x)α+q(3x)=13α,

    and ψ>˜σ, noticing g is decreasing (see Lemma 2.1), there exists ϵ>0 such that

    g(3ϵ)=p(3ϵ)α+q(3ϵ)>˜σ3αψ,

    which implies

    3αg(3ϵ)ψ˜σ>1.

    If x0P0, there exists ϵ0>0 such that x0(t)ϵ0 for all tR. It follows that

    A(x0)(t)3αtψF(ϵ0)e˜σ(ts)ds=3αψ˜σϵ0g(ϵ0)>3αψ˜σϵ0g(3ϵ0)>ϵ0,

    for all tR, which means that A(x0)P0. And for ϵ2(0,ϵ1), we obtain

    A(ϵ2)(t)3αtψF(ϵ2)e˜σ(ts)ds=3ψ˜σϵ2g(3ϵ2)>3ψ˜σϵ2g(3ϵ1)>ϵ2.

    It is easy to get that

    F(ϑx)=ζ(ϑ,x)F(x),

    for all 0<ϑ<1 and x(0,+), where ζ(ϑ,x)=ϑS(y)|θ=ϑ,y=x. Let

    ϕ(ϑ,x)=ζ(ϑ,inftRx(t)),xP0.

    By Lemma 2.1 (6), one can get that ζ(ϑ,.) is strictly monotone increasing in (0,+) and limx0ζ(ϑ,x)=ϑ, which implies ζ(ϑ,x)>ϑ for ϑ(0,1) and x(0,+). Therefore,

    ϕ(ϑ,x)>ϑ,ϑ(0,1),xP0.

    Also, by the fact that ζ(ϑ,) is strictly monotone increasing in (0,+), one can get that ϕ(ϑ,) is nondecreasing in P0. It follows that

    A(ϑx)(t)=3αtψ(sτ)F(ϑx(sτ))e˜σ(ts)ds=3αtψ(sτ)x(sτ)g(3x(sτ))ζ(ϑ,x(tτ))e˜σ(ts)ds3αtψ(sτ)x(sτ)g(3x(sτ))ϕ(ϑ,x)e˜σ(ts)ds3αϕ(ϑ,x)tψ(sτ)x(sτ)g(3x(sτ))e˜σ(ts)ds=ϕ(ϑ,x)A(x)(t).

    By Theorem 2.5 (see (2.3) and (2.4)), Eq (3.1) has exactly one positive almost periodic solution xP0. Then, by Lemma 2.4, x is just the unique almost periodic solution with a positive infimum to Eq (1.8). Moreover, (3.2) and (3.3) follow from (2.3) and (2.4).

    (2). By Lemma 3.1, Eq (1.8) has a nonnegative almost periodic solution

    x(t)=3αtψ(sτ)F(x(sτ))e˜σ(ts)ds,tR. (3.8)

    Define operator A:CAP(R)CAP(R) as follows:

    A(x)(t)=3αtψ(sτ)F(x(sτ))e˜σ(ts)ds. (3.9)

    Next, we show that A is a contraction operator. For any x,yCAP(R),

    A(x)(t)A(y)(t)=3αtψ(sτ)[F(x(sτ))F(y(sτ))]e˜σ(ts)ds3αtψ|F(ξ(t))|e˜σ(ts)dsxy,

    where ξ(t) lies between x(t) and y(t). For any x>0, since

    F(x)=13y2G(y)|y=3x>0,

    and g(3x)<0, it follows that |F(x)|g(3x)1/3. Then,

    |A(x)(t)A(y)(t)|ψ˜σxy,

    which implies that A is a contraction operator since ψ<˜σ. Therefore, Eq (1.8) has exactly one nonnegative almost periodic solution x(t). If we define p(0)=1/3, then p is continuous on R. Therefore, zero is also an almost periodic solution of Eq (1.8). By the uniqueness, we have x(t)0. Since

    xk(t)=3αtψ(sτ)f(xk1(sτ))e˜σ(ts)dsψ˜σxk1(ψ˜σ)2xk2(ψ˜σ)kx0,

    and ψ˜σ<1, we can get xk0,k. This completes the proof of Theorem 3.2.

    Remark. Theorem 3.2 (2) implies that if ψ<˜σ, Eq (1.8) has no positive almost periodic solution.

    Lemma 4.1. Assume that the function F(x,y) is defined on R+×R+ and continuously differentiable. Suppose Fy>0 for (x,y)R+×R+. For any T>0, if z1,z2C[τ,T)C1(0,T) satisfy the following inequalities:

    z1(t)F(z1(t),z1(tτ))fort>0, (4.1)
    z2(t)F(z2(t),z2(tτ))fort>0, (4.2)
    z1(t)z2(t)forτt0, (4.3)

    then, z1(t)z2(t) for tτ.

    Proof. Please see Lemma 3.1 in [6].

    Lemma 4.2. Consider the following problem

    x(t)=F(x(t),x(tτ))fort>0, (4.4)
    x(t)=x0(t)forτt0. (4.5)

    Assume that the function F is defined on R+×R+ and continuously differentiable. Suppose Fy>0 for (x,y)R+×R+. Let xs be a positive solution of equation F(x,x)=0 such that

    (xxs)F(x,x)<0forxxs.

    If x(t) is the solution of the problem of (4.1), (4.5), and x0(t)C[τ,0] for τt0, then,

    limtx(t)=xs.

    Proof. Please see Lemma 3.4 in [6].

    By Eq (1.7), we can get

    3αψF(x(tτ))˜σx(t)x(t)3αψF(x(tτ))˜σx(t).

    Consider the following two initial value problems

    z(t)=3αψF(z(tτ))˜σz(t), (4.6)
    z0(t)=φ(t),τt0, (4.7)

    and

    y(t)=3αψF(y(tτ))˜σy(t), (4.8)
    y0(t)=φ(t),τt0. (4.9)

    Define

    F1(x,y)=3αψF(y)˜σx,F2(x,y)=3αψF(y)˜σx.

    From (3.7), we know that F1 and F2 are monotone increasing in y. Since ψ>˜σ, by Lemma 2.1, one can get

    0<αp(x)α+q(x)=αG(x)x3=αg(x)<1/3,

    for all x>0. Then, when ψ>˜σ (i.e., 0<˜σ3ψ<˜σ3ψ<13), it follows that the equations

    F1(x,x)=3αxψ[g(3x)˜σ3ψ]=0

    and

    F2(x,x)=3αxψ[g(3x)˜σ3ψ]=0,

    have a unique positive constant solution x1 and x2, respectively, and x1<x2, where the fact that g(x)<0 for x>0 is used.

    Lemma 4.3. If ψ>˜σ, then the following assertion holds:

    x1=limtz(t)x_=lim inftx(t)lim suptx(t)=ˉxlimty(t)=x2.

    Moreover, there exists T>0 such that

    x(t)>x1/2>0, (4.10)

    for t>T.

    Proof. Since g(x)<0, we have (xx1)F(x,x)<0 for xx1. By Lemma 4.2, for any nonnegative initial value function x0(t), one can get

    limtz(t)=x1, (4.11)

    where x(t) is the solution of (4.6) and (4.7). Similarly, it is easy to get that for any nonnegative initial value function x0(t), one can get

    limty(t)=x2, (4.12)

    where y(t) is the solution of (4.8) and (4.9). Using Lemmas 4.1 and 4.2, one can get

    x1=limtz(t)x_=lim inftx(t)lim suptx(t)=ˉxlimty(t)=x2.

    Thus, (4.10) follows. This completes the proof.

    The solution to Eq (1.8) is related to α. Denote x(t)=x(t,α). Assume α1α2. Consider the following two problems

    y1(t)=3α1ψ(t)F(y1(tτ))˜σy1(t), (4.13)
    y1(t)=φ(t),τt0 (4.14)

    and

    y2(t)=3α2ψ(t)F(y2(tτ))˜σy2(t), (4.15)
    y2(t)=φ(t),τt0. (4.16)

    By Lemma 4.1, it is easy to get that x(t,α1)x(t,α2). Then,

    Lemma 4.4. The solution to Eq (1.8) is monotone increasing in α.

    Theorem 4.5. (I) If ψ>˜σ and ˜σψ(˜σψ+3A0)>0 hold, where A0=1/33x1g(3x1), then there exists τ0>0 such that for all τ(0,τ0), the unique almost periodic positive solution to Eq (1.8) is exponentially stable.

    (II) If ψ<˜σ, then there exists τ1>0 such that for all τ(0,τ1), every solution to Eq (1.8) exponentially asymptotically tends to 0.

    Remark. By selecting appropriate parameters, the condition ˜σψ(˜σψ+3A0)>0 in Theorem 4.5 (Ⅰ) can be satisfied. Actually, let

    l(y)=˜σ(1y1ψ)+3A0ψ.

    Then, l(y)<0 and limy0+l(y)=+. By Lemmas 2.1 (1) and (3), g is decreasing, thus A0<0. Then l(ψ)=3A0ψ<0. Thus, there exists a positive constant l0 such that l(y)<0 for l0<y<ψ. Since

    ˜σψ(˜σψ+3A0)>0˜σ(1ψ1ψ)+3A0ψ<0,

    the conditions in Theorem 4.5 (Ⅰ) will be satisfied if we choose the almost function ψ(t) satisfying ψ(l0,ψ) and ˜σ satisfying ˜σ<ψ.

    Proof. (Ⅰ) Since ˜σψ(˜σψ+3A0)>0, due to the sign preserving property of continuous functions, there exists η>0 which is small enough such that

    ˜σψ(˜σψ+3(A0+η))>0.

    Let

    ϑ(τ)=˜σψ(˜σψ+3(A0+η))e˜στ,

    where A0=133x1g(3x1)<0. Then

    ϑ(0)=˜σψ(˜σψ+3(A0+η))>0, (4.17)

    which implies that there exists a constant τ0>0 such that

    θ(τ)>θ(0)2>0,

    for all τ(0,τ0).

    Let x(t) be an arbitrary solution of (1.8) and x(t) is the unique almost periodic solution of (1.8). Then,

    x(t)=x0e˜σt+3αt0e˜σ(ts)ψ(sτ)F(x(sτ))ds,

    and

    x(t)=x0e˜σt+3αt0e˜σ(ts)ψ(sτ)F(x(sτ))ds,

    for all t0. Then we can get

    x(t)x(t)=(x0(0)x0(0))e˜σt+3αt0e˜σ(ts)ψ(sτ)(F(x(sτ))F(x(sτ)))ds.

    It follows that

    |x(t)x(t)|(x0(0)x0(0))e˜σt+3αt0e˜σ(ts)ψ(sτ)|F(x(sτ))F(x(sτ))|ds.

    Because of the continuity of F, for any η>0, there exists δ>0 such that when |z(t)x1|<δ, there holds

    |F(z(t))F(x1)|<η. (4.18)

    Since limtz(t)=x2, for the above δ, there exists T>τ>0 such that when t>Tτ, there holds

    |z(t)x1|<δ.

    Thus, there exists T>0 such that when t>Tτ, (4.18) holds. It follows that for t>Tτ, there holds

    F(z(t))F(x1)+η. (4.19)

    Let u(t)=|x(t)x(t)|e˜σt. We can get for t>T,

    u(t)˜M+3αtTe˜σsψ(sτ)|F(ξ)|.|x(sτ)x(sτ)|ds,

    where ξ=ϑx(tτ)+(1ϑ)x(tτ), ϑ(0,1), and

    ˜M=|x0(0)x0(0)|+3αT0e˜σsψ(sτ)|F(ξ)|.|x(sτ)x(sτ)|ds.

    By the fact that

    |F(ξ)|=g(3ξ)+1/33ξg(3ξ)<g(3ξ)<1/3,

    we have

    ˜M=|x0(0)x0(0)|+3αT0e˜σsψ(sτ)|F(ξ)|.|x(sτ)x(sτ)|ds|x0(0)x0(0)|+αT0e˜σsψ(sτ).|x(sτ)x(sτ)|ds=:M.

    It follows that

    u(t)M+3αtTe˜σsψ(sτ)|F(ξ)|.|x(sτ)x(sτ)|ds,

    for t>T. Notice that z(t) is a solution of (4.6) and (4.7), then by Lemma 4.1, one can get

    ξ=ϑx(tτ)+(1ϑ)x(tτ)z(tτ).

    Since F(x)<0 for all x>0, then there exists T>0 such that for t>T,

    u(t)M+3αtTe˜σsψ(sτ)|F(ξ)|.|x(sτ)x(sτ)|dsM+3αtTe˜σsψ(sτ)F(z(sτ)).|x(sτ)x(sτ)|dsM+3αtTe˜σsψ(sτ)(F(x1)+η)|x(sτ)x(sτ)|dsM+3αtTe˜σsψ(˜σ3αψ+A0+η)u(sτ)dsM+3αt0e˜στψ(˜σ3αψ+A0+η)u(s)dsM+3αtττe˜στψ(˜σ3αψ+A0+η)u(s)ds=M+3α0τe˜στψ(˜σ3αψ+A0+η)u(s)ds+3αtτ0e˜στψ(˜σ3αψ+A0+η)u(s)dsM1+αt0e˜στψ(˜σαψ+3(A0+η))u(s)ds=M1+t0e˜στ(ψ˜σψ+3αψ(A0+η))u(s)ds,

    where

    M=|x0(0)x0(0)|+αT0e˜σsψ(sτ)|x(sτ)x(sτ)|ds,

    and M1=M+3α0τe˜στψ(˜σ3αψ+A0+η)u(s)ds and Lemma 2.1 has been used. By the Gronwall inequality, one can get

    u(t)M1eκt,

    where κ=e˜στψ(˜σψ+3(A0+η)). It follows that

    |x(t)x(t)|M1e(κ˜σ)t=M1e(˜σκ)tM1eγt,

    where

    γ=˜σψ(˜σψ+3(A0+η))e˜στ>0,

    for τ(0,τ0) and η is sufficiently small, which means x(t) is exponentially stable. The proof of Theorem 4.5 (Ⅰ) is complete.

    (Ⅱ) Let

    L(τ)=˜σψe˜στ.

    Then L(0)=˜σψ>0, thus there exists τ1>0 such that L(τ)>0 for any τ(0,τ1). Let v(t)=|x(t)|e˜σt. We can get

    v(t)C+3t0e˜σsψ(sτ)|x(sτ)|g(3x(sτ))|dsC+t0e˜σsψ|x(sτ)|dsC+tττe˜στψv(s)ds=C+0τe˜στψv(s)ds+tτ0e˜στψv(s)dsC1+t0e˜στψv(s)ds,

    where C=x0(0),C1=C+0τe˜στψv(s)ds. By the Gronwall inequality, one can get

    u(t)C1eκt.

    It follows that

    |x(t)|C1e(κ˜σ)t=C1e(˜σκ)tC1e(˜σψe˜στ)t,

    for τ(0,τ1), which means x(t) is exponentially stable since ˜σψe˜στ>0. This completes the proof.

    In this section, the results of computer simulations are presented. By using MATLAB R2016a, we present some examples of solutions of Eq (1.8) for different parameter values (see Figures 16). The model parameter values used in the simulations are given with the figures' captions.

    Figure 1.  The corresponding curves of the solutions to Eq (1.8) for ψ(t)=5+1/2cos(t)+1/2sin(2t),α=1,˜σ=1,τ=0.1, and x0=260,360, respectively.
    Figure 2.  The corresponding curves of the solutions to Eq (1.8) for ψ(t)=5+cos(t)+sin(2t),α=1,˜σ=20,τ=1, and x0=3,1, respectively.
    Figure 3.  The corresponding curves of the solutions to Eq (1.8) for ψ(t)=5+12[cos(t)+sin(2t)],˜σ=1,τ=0.1, x0=260, and α=1,3,5, respectively.
    Figure 4.  The corresponding curves of the solutions to Eq (1.8) for ψ(t)=5+cos(t)+sin(2t),˜σ=20,τ=1, x0=3, and α=1,3,5, respectively.
    Figure 5.  The corresponding curves of the solutions to Eq (1.8) for ψ(t)=5+cos(t)+sin(2t),˜σ=1,τ=2, x0=260, and α=1,3,5, respectively.
    Figure 6.  The corresponding curves of the solutions to Eq (1.8) for ψ(t)=5+cos(t)+sin(2t),˜σ=10,τ=2, x0=60, and α=1,3,5, respectively.

    Let ψ(t)=5+12[cos(t)+sin(2t)]. Then ψ=6,ψ=4. Assume α=1,τ=0.1, and ˜σ=1. The solution to

    g(3x)=˜σ3ψα=16

    is x10.16. It follows that

    A0=133xg(3x)0.339.
    γ=˜σψ(˜σαψ+3A0)e˜στ0.166>0.

    The conditions of Theorem 4.5 (Ⅰ) are satisfied.

    Let ψ(t)=5+cos(t)+sin(2t). Then ψ=7,ψ=3. Assume α=1,τ=1, and ˜σ=20. Then

    ˜σψe˜στ=207e>0.

    The conditions of Theorem 4.5 (Ⅱ) are satisfied.

    Let ψ(t)=5+cos(t)+sin(2t). Then ψ=7,ψ=3. Assume α=1,τ=2, and ˜σ=1. The solution to

    g(3x)=˜σ3ψα=19,

    is x56.0247. It follows that

    A0=133xg(3x)0.0297.
    γ=˜σψ(˜σαψ+3A0)e˜στ11.6326<0.

    The conditions of Theorem 4.5 (Ⅰ) are not satisfied. Assume α=1,τ=2, and ˜σ=10. Then

    ˜σψe˜στ=107e2<0.

    So these parameter values do not meet the conditions of Theorem 4.5 (Ⅱ).

    In Figure 1, the behavior of the solutions covered by Theorem 4.5 (Ⅰ) is presented. Numerical simulation results indicate that, for certain parameter values satisfying Theorem 4.5 (Ⅰ) and different constant initial values, the tumor will asymptotically tend towards an almost periodic solution. Figure 2 illustrates the behavior of the solutions covered by Theorem 4.5 (Ⅱ). It is observed that, for specific parameter values and different constant initial values satisfying Theorem 4.5 (Ⅱ), the tumor will disappear. In Figures 3 and 4, the behavior of the solutions is presented for varying α representing blood vessel density. Numerical simulation results demonstrate that when other parameters remain unchanged, the solution increases with increasing α without affecting its final trend of the solution.

    Figures 5 and 6 indicate that even if the conditions of Theorem 4.5 are not met, the solution of the problem may exponentially asymptotically tend to the unique almost periodic solution or exponentially asymptotically tend to 0. This indicates that the conditions for exponential asymptotic stability in Theorem 4.5 are only sufficient conditions.

    The focus of this study lied in investigating the impact of almost periodic nutrient supply, angiogenesis, and time delay on tumor growth. Our results demonstrated that an almost periodic nutrient supply led to a unique almost periodic solution for this problem (refer to Theorem 3.2). Furthermore, we established that this unique solution was exponentially asymptotically stable under certain parameter conditions, while the presence of time delay did not affect the final growth trend of the tumor (see Theorem 4.5). Additionally, when keeping other parameters constant, our findings indicated that the solution increased with an increase in α, which represents the intensity of angiogenesis; however, it should be noted that the magnitude of α did not affect the final trend of the solution (refer to Lemma 4.4 and Theorem 4.5).

    The authors declare that they have not used Artificial Intelligence (AI) tools in the creation of this article.

    This work is partially supported by the National Science Foundation of Guangdong Province (2023A1515011911) and the Academic Research Projects of Beijing Union University (ZKZD202306).

    The authors are thankful to the editor and the three anonymous reviewers for their constructive comments for improving this paper.

    The authors declare that they have no conflicts of interest.



    [1] M. Bodnar, The nonnegativity of solutions of delay differential equations, Appl. Math. Lett., 13 (2000), 91–95. https://doi.org/10.1016/S0893-9659(00)00061-6 doi: 10.1016/S0893-9659(00)00061-6
    [2] H. Bohr, Almost periodic functions, New York, 1947.
    [3] H. M. Byrne, The effect of time delays on the dynamics of avascular tumor growth, Math. Biosci., 144 (1997), 83–117. https://doi.org/10.1016/S0025-5564(97)00023-0 doi: 10.1016/S0025-5564(97)00023-0
    [4] C. Corduneanu, Almost periodic functions, New York, 1989.
    [5] S. B. Cui, A. Friedman, Analysis of a mathematical model of the effect of inhibitors on the growth of tumors, Math. Biosci., 164 (2000), 103–137. https://doi.org/10.1016/S0025-5564(99)00063-2 doi: 10.1016/S0025-5564(99)00063-2
    [6] S. B. Cui, S. H. Xu, Analysis of mathematical models for the growth of tumors with time delays in cell proliferation, J. Math. Anal. Appl., 336 (2007), 523–541. https://doi.org/10.1016/j.jmaa.2007.02.047 doi: 10.1016/j.jmaa.2007.02.047
    [7] H. S. Ding, J. Nieto, A new approach for positive almost periodic solutions to a class of Nicholson's blowflies model, J. Comput. Appl. Math., 253 (2013), 249–254. https://doi.org/10.1016/j.cam.2013.04.028 doi: 10.1016/j.cam.2013.04.028
    [8] H. S. Ding, T. J. Xiao, J. Liang, Existence of positive almost automorphic solutions to nonlinear delay integral equations, Nonlinear Anal. Theor., 70 (2009), 2216–2231. https://doi.org/10.1016/j.na.2008.03.001 doi: 10.1016/j.na.2008.03.001
    [9] A. M. Fink, Almost periodic differential equations, Berlin: Springer, 1974. https://doi.org/10.1007/BFb0070324
    [10] U. Forýs, M. Bodnar, Time delays in proliferation process for solid avascular tumour, Math. Comput. Model., 37 (2003), 1201–1209. https://doi.org/10.1016/S0895-7177(03)80019-5 doi: 10.1016/S0895-7177(03)80019-5
    [11] A. Friedman, F. Reitich, Analysis of a mathematical model for the growth of tumors, J. Math. Biol., 38 (1999), 262–284. https://doi.org/10.1007/s002850050149 doi: 10.1007/s002850050149
    [12] A. Friedman, K. Lam, Analysis of a free-boundary tumor model with angiogenesis, J. Differ. Equations, 259 (2015), 7636–7661. https://doi.org/10.1016/j.jde.2015.08.032 doi: 10.1016/j.jde.2015.08.032
    [13] D. Guo, V. Lakshmikantham, Nonlinear problems in abstract cones, New York: Academic Press, 2014.
    [14] G. Huang, A. P. Liu, U. Foryś, Global stability analysis of some nonlinear delay differential equations in population dynamics, J. Nonlinear Sci., 26 (2016), 27–41. https://doi.org/10.1007/s00332-015-9267-4 doi: 10.1007/s00332-015-9267-4
    [15] M. J. Piotrowska, Hopf bifurcation in a solid avascular tumor growth model with two discrete delays, Math. Comput. Model., 47 (2008), 597–603. https://doi.org/10.1016/j.mcm.2007.02.030 doi: 10.1016/j.mcm.2007.02.030
    [16] M. J. Piotrowska, A remark on the ODE with two discrete delays, J. Math. Anal. Appl., 329 (2007), 664–676. https://doi.org/10.1016/j.jmaa.2006.06.078 doi: 10.1016/j.jmaa.2006.06.078
    [17] F. A. Rihan, Delay differential equations and applications to biology, Singapore: Springer, 2021. https://doi.org/10.1007/978-981-16-0626-7
    [18] R. R. Sarkar, S. Banerjee, A time delay model for control of malignant tumor growth, National Conference on Nonlinear Systems and Dynamics, 1 (2006), 1–5.
    [19] D. R. Smart, Fixed points theorems, Cambridge: Cambridge University Press, 1980.
    [20] S. H. Xu, Analysis of a delayed free boundary problem for tumor growth, Discrete Cont. Dyn. B., 18 (2011), 293–308. https://doi.org/10.3934/dcdsb.2011.15.293 doi: 10.3934/dcdsb.2011.15.293
    [21] S. H. Xu, Analysis of a free boundary problem for tumor growth in a periodic external environment, Bound. Value Probl., 2015 (2015), 1–12. https://doi.org/10.1186/s13661-015-0399-0 doi: 10.1186/s13661-015-0399-0
    [22] S. H. Xu, M. Bai, X. Q. Zhao, Analysis of a solid avascular tumor growth model with time delays in proliferation process, J. Math. Anal. Appl., 391 (2012), 38–47. https://doi.org/10.1016/j.jmaa.2012.02.034 doi: 10.1016/j.jmaa.2012.02.034
    [23] S. H. Xu, Analysis of a free boundary problem for tumor growth with angiogenesis and time delays in proliferation, Nonlinear Anal. Real World Appl., 51 (2020), 103005. https://doi.org/10.1016/j.nonrwa.2019.103005 doi: 10.1016/j.nonrwa.2019.103005
    [24] S. H. Xu, Hopf bifurcation of a free boundary problem modeling tumor growth with two time delays, Chao Soliton. Fract., 41 (2009), 2491–2494. https://doi.org/10.1016/j.chaos.2008.09.029 doi: 10.1016/j.chaos.2008.09.029
    [25] Z. J. Ye, S. H. Xu, X. M. Wei, Analysis of a free boundary problem for vascularized tumor growth with a time delay in the process of tumor regulating apoptosis, AIMS Math., 7 (2022), 19440–19457. https://doi.org/10.3934/math.20221067 doi: 10.3934/math.20221067
    [26] H. H. Zhou, Z. J. Wang, D. M. Yuan, H. J. Song, Hopf bifurcation of a free boundary problem modeling tumor growth with angiogenesis and two time delays, Chaos Soliton. Fract., 153 (2021), 111578. https://doi.org/10.1016/j.chaos.2021.111578 doi: 10.1016/j.chaos.2021.111578
  • Reader Comments
  • © 2024 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(862) PDF downloads(39) Cited by(0)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog