Analysis of an HIV infection model incorporating latency age and infection age

  • Received: 06 February 2017 Accepted: 25 May 2017 Published: 01 June 2018
  • MSC : Primary: 92D25, 92D30; Secondary: 37G99

  • There is a growing interest to understand impacts of latent infection age and infection age on viral infection dynamics by using ordinary and partial differential equations. On one hand, activation of latently infected cells needs specificity antigen, and latently infected CD4+ T cells are often heterogeneous, which depends on how frequently they encountered antigens, how much time they need to be preferentially activated and quickly removed from the reservoir. On the other hand, infection age plays an important role in modeling the death rate and virus production rate of infected cells. By rigorous analysis for the model, this paper is devoted to the global dynamics of an HIV infection model subject to latency age and infection age from theoretical point of view, where the model formulation, basic reproduction number computation, and rigorous mathematical analysis, such as relative compactness and persistence of the solution semiflow, and existence of a global attractor are involved. By constructing Lyapunov functions, the global dynamics of a threshold type is established. The method developed here is applicable to broader contexts of investigating viral infection subject to age structure.

    Citation: Jinliang Wang, Xiu Dong. Analysis of an HIV infection model incorporating latency age and infection age[J]. Mathematical Biosciences and Engineering, 2018, 15(3): 569-594. doi: 10.3934/mbe.2018026

    Related Papers:

    [1] Xiaohong Tian, Rui Xu, Jiazhe Lin . Mathematical analysis of an age-structured HIV-1 infection model with CTL immune response. Mathematical Biosciences and Engineering, 2019, 16(6): 7850-7882. doi: 10.3934/mbe.2019395
    [2] Yan Wang, Minmin Lu, Daqing Jiang . Viral dynamics of a latent HIV infection model with Beddington-DeAngelis incidence function, B-cell immune response and multiple delays. Mathematical Biosciences and Engineering, 2021, 18(1): 274-299. doi: 10.3934/mbe.2021014
    [3] Patrick W. Nelson, Michael A. Gilchrist, Daniel Coombs, James M. Hyman, Alan S. Perelson . An Age-Structured Model of HIV Infection that Allows for Variations in the Production Rate of Viral Particles and the Death Rate of Productively Infected Cells. Mathematical Biosciences and Engineering, 2004, 1(2): 267-288. doi: 10.3934/mbe.2004.1.267
    [4] A. M. Elaiw, N. H. AlShamrani . Stability of HTLV/HIV dual infection model with mitosis and latency. Mathematical Biosciences and Engineering, 2021, 18(2): 1077-1120. doi: 10.3934/mbe.2021059
    [5] A. M. Elaiw, N. H. AlShamrani . Analysis of an HTLV/HIV dual infection model with diffusion. Mathematical Biosciences and Engineering, 2021, 18(6): 9430-9473. doi: 10.3934/mbe.2021464
    [6] Jinliang Wang, Ran Zhang, Toshikazu Kuniya . A note on dynamics of an age-of-infection cholera model. Mathematical Biosciences and Engineering, 2016, 13(1): 227-247. doi: 10.3934/mbe.2016.13.227
    [7] Jinliang Wang, Jingmei Pang, Toshikazu Kuniya . A note on global stability for malaria infections model with latencies. Mathematical Biosciences and Engineering, 2014, 11(4): 995-1001. doi: 10.3934/mbe.2014.11.995
    [8] Yicang Zhou, Yiming Shao, Yuhua Ruan, Jianqing Xu, Zhien Ma, Changlin Mei, Jianhong Wu . Modeling and prediction of HIV in China: transmission rates structured by infection ages. Mathematical Biosciences and Engineering, 2008, 5(2): 403-418. doi: 10.3934/mbe.2008.5.403
    [9] Lu Gao, Yuanshun Tan, Jin Yang, Changcheng Xiang . Dynamic analysis of an age structure model for oncolytic virus therapy. Mathematical Biosciences and Engineering, 2023, 20(2): 3301-3323. doi: 10.3934/mbe.2023155
    [10] Yu Yang, Shigui Ruan, Dongmei Xiao . Global stability of an age-structured virus dynamics model with Beddington-DeAngelis infection function. Mathematical Biosciences and Engineering, 2015, 12(4): 859-877. doi: 10.3934/mbe.2015.12.859
  • There is a growing interest to understand impacts of latent infection age and infection age on viral infection dynamics by using ordinary and partial differential equations. On one hand, activation of latently infected cells needs specificity antigen, and latently infected CD4+ T cells are often heterogeneous, which depends on how frequently they encountered antigens, how much time they need to be preferentially activated and quickly removed from the reservoir. On the other hand, infection age plays an important role in modeling the death rate and virus production rate of infected cells. By rigorous analysis for the model, this paper is devoted to the global dynamics of an HIV infection model subject to latency age and infection age from theoretical point of view, where the model formulation, basic reproduction number computation, and rigorous mathematical analysis, such as relative compactness and persistence of the solution semiflow, and existence of a global attractor are involved. By constructing Lyapunov functions, the global dynamics of a threshold type is established. The method developed here is applicable to broader contexts of investigating viral infection subject to age structure.


    1. Introduction

    Determining the threshold dynamics of infection-free and infection equilibrium in viral infection model has made great progress in the last decades [1,2,4,8,9,10,16,17,24,31,36,32,33,34,35,40,44]. A key insight in this progress is that if threshold value (named, the basic reproduction number) is less than one then the infection-free equilibrium is globally asymptotically stable otherwise the endemic equilibrium attracts all solutions (is globally asymptotically stable) whenever it exists. One method adopted here is due to the classical Volterra type Lyapunov function, which was discovered by Volterra [30]. These confirmed global stability properties of steady states for within host virus model establish our understanding the virus dynamical behaviors, that is, whether the viruses die out or not.

    Even large discrete and continuous delay differential equations of viral infection models have been successfully treated by Volterra type Lyapunov function, (ⅰ) nonlinear incidence rate functions [8,24,31]; (ⅱ) discrete delays [10,16,35] and finite distributed delays [17,31,36], and infinite distributed delays [8,24,34]; (ⅲ) immune responses [24,35,44,39,41]; and (ⅳ) additional infection processes [31]. It is still a hot topic in in-host model to determine how these factors affect the virus dynamical behaviors. We also refer the reader to see these citations for more references.

    Recently, age-structured viral infection model has attracted much attentions of researchers. HIV latency remains a major obstacle to viral elimination. Although HIV-1 replication can be controlled by antiretroviral therapy in suppress the plasma viral load to below the detection limit, the time spent in this progress may last half life of months or years [21]. Virus persisting in reservoirs, such as latency infected CD4+ T cells, may the reason that long-term low viral load persistence in patients on antiretroviral therapy and keeping the virus from being eliminated. These latency infected CD4+ T cells are not affected by immune responses but can produce virus once activated by relevant antigens.

    Some recent studies reveals the decay dynamics of the latent reservoir. For example, a model has been developed by Muller et al. [11] to describe the heterogeneity of latent cell activation. An ordinary differential equations (ODEs) model has been studied by Kim and Perelson [7] to include decreasing activation of latently infected cells. Activation of these latently infected cells needs specificity antigen. A recent study by Strain et al. [22] reveals that the dynamics of latently infected CD4+ T cells are often heterogeneous. They argued that cells specific to frequently encountered antigens are activated soon while cells specific to rare antigens need more time to be activated. Thus, the activation rate depends on the time spent since the cell is latently infected (that is the time elapsed since the establishment of latency), which we refer as latency age for short. A recent paper by Alshorman et al. [1] introduced a latency age model to mathematically analyze the dynamics of the latent reservoir under combination therapy. They give an affirmed answers that the long-term activation rate of latently infected cells plays an important role in determining the dynamics.

    Taking into account the picture that the mortality rate and viral production rate of infected cells may depend on the infection age of cells, Nelson et al. [15], Huang et al. [4] and Wang et al. [37] have studied age-structured model of HIV infection by considering age to be a continuous variable rather than be constant in ODEs models. These assumptions lead to a hybrid system of ODEs and partial differential equations (PDEs) formulation and allow us to have a good understanding on productively infected cells. Together with the infinite-dimensional nature of system, this formulation creates some mathematical difficulties in establishing the existence of a global compact attractor, even in other epidemic models (see some relevant references for our discussion on age-structured models, [6,43,42,25,27,26,14]).

    Denote by T(t),e(a,t),i(a,t),V(t) the concentration of uninfected CD4+ T cells at time t, the concentration of latently infected T cells with latency age a at time t, the concentration of productively infected cells, and the concentration of virions in plasma at t, respectively. The parameter h is the production rate of uninfected CD4+ T cells, d is the per capita death rate of uninfected cells, and β is the infection rate of the target cell by virus. c is the viral clearance rate.

    The following assumptions are a compromise between generality and simplicity.

    Assumption 1.1 (ⅰ) There is a small fraction (f) of infected cells lead to latency and that the remaining become productively infected cells [1,40].

    (ⅱ) When latently infected cells are activated to become productively infected cells, an age-dependent remove rate θ1(a) is used to illustrate the decreasing effect of the pool size of latent reservoir. The integral term 0ξ(a)e(a,t)da describes he total number of productively infected cells gained per unit time from the activation of latently infected cells, where ξ(a) denotes the activation rate of latently infected T cells with latency age a. Biologically, we omit the proliferation rate [25] and the death rate of latently infected cells, which is assumed to be included in the removal rate.

    (ⅲ) We assume that production rate of viral particles p(b) and the death rate of productively infected cells θ2(b) wiht infection age b are two continuous functions of age (the time passed since infection), see, e.g., [15,4,32].

    Biologically, (ⅰ) of Assumption 1.1 comes from the evidences that a very small fraction of CD4+ T cell infection leads to HIV latency. They don't produce new virus unless activated by antigens, please see [18,19,20]. (ⅱ) of Assumption 1.1 based on the fact the latently infected cell population is very likely to be heterogeneous [22,23]. Cells specific to frequently encountered antigens may be preferentially activated and quickly removed from the reservoir. It may depends on the time elapsed since latent infection and affect the activation rate, that is the reason why we are interested in the latent infection age. (ⅲ) of Assumption 1.1 it is known that viral proteins and unspliced viral RNA accumulate within the cytoplasm of an infected cell, and thus, they actually ramps up [3,12,29]. Therefore, infection age should be incorporated into the model.

    In this paper, we introduce the following HIV infection model with latency and infection age,

    {dT(t)dt=hdT(t)βT(t)V(t),(t+a)e(a,t)=θ1(a)e(a,t),(t+b)i(b,t)=θ2(b)i(b,t),dV(t)dt=0p(b)i(b,t)dbcV(t), (1)

    with boundary and initial conditions

    {e(0,t)=fβT(t)V(t),i(0,t)=(1f)βT(t)V(t)+0ξ(a)e(a,t)da,T(0)=T00, e(a,0)=e0(a)L1+(0,),i(b,0)=i0(b)L1+(0,), V(0)=V00,

    where L1+(0,) is the set of all integrable nonnegative functions on R+:=[0,).

    Mathematically, for the ease of simplicity, we make the following assumptions.

    Assumption 1.2 (ⅰ) h, d, β, c>0;

    (ⅱ) For 1=1,2, θi(),p(),ξ()L+(0,) satisfy the conditions:

    ¯θi:=esssupa[0,)θi(a)<,  ˉp:=esssupa[0,)p(a)<,  ˉξ:=esssupa[0,)ξ(a)<,

    (ⅲ) p(), ξ() are Lipschitz continuous on R+ with Lipschitz constants Mp, Mξ respectively;

    (ⅳ) There exists μ0(c,d] such that θ1(a),θ2(b)μ0 for all a,b0;

    (ⅴ) There exists a maximum age b+>0 for the viral production such that p(b)>0 for b(0,b+) and p(b)=0 for b>b+.

    Our goal of the present paper is to adopt previous model in [32,33] by incorporating the latency age for infected cells as discussed in [1], and to study the threshold dynamics of infection-free and infection equilibrium in viral infection model subject to latently age and infection age. We will show the existence of a compact attractor of all compact sets of nonnegative initial data and use the Lyapunov function to show that this attractor is the singleton set containing the endemic equilibrium. Roughly speaking, if the basic reproduction number is less than one then the infection-free equilibrium is globally asymptotically stable otherwise the endemic equilibrium attracts all solutions with active infection at some time.

    The remaining part of this paper is organized as follows. In Section 2, we present some preliminary results including model formulation (equivalent integrated semigroup formulation and Volterra formulation), properties of solutions and existence of equilibria. Then we show the asymptotic smoothness of Φ(t,X0) of orbits in Section 3, where we arrive at a key result on the existence of global compact attractor. In section 4, we prove that system (1) is the uniformly persistent. The Section 5 is devoted to local stability analysis of the infection-free equilibrium and the infection equilibrium. Then we establish their global attractivity in Section 6 by constructing Lyapunov functions.


    2. Preliminary

    For ease of notations, we introduce the following notations:

    Ω(a)=ea0θ1(τ)dτ  and  Γ(b)=eb0θ2(τ)dτ  for  a,b0.

    Biologically, Ω(a) is the probability of an infected cell staying in the latent state until age a. Γ(b) is typically interpreted as the probability that an infected cell can survive to age b.

    Then, by (ⅱ) and (ⅴ) of Assumption 1.2, we have that for all a,b0,

     0Ω(a)eμ0a  and  0Γ(b)eμ0b, dΩ(a)da=θ1(a)Ω(a)  and  dΓ(b)db=θ2(b)Γ(b). (2)

    2.1. Integrated semigroup formulation

    Following the line of [27], we reformulate the model (1) as a semilinear Cauchy problem. Taking into account the boundary conditions, we consider the following state space,

    X=R×R×L1((0,),R)×R×L1((0,),R)×R,
    X+=R+×R+×L1+((0,),R)×R+×L1+((0,),R)×R+,

    endowed with the usual product norm, and set

    X0=R×{0}×L1((0,),R)×{0}×L1((0,),R)×R,
    X0+=X0X+.

    We consider the linear operator A:Dom(A)XX defined by

    A(T(0e)(0i)V)=(dT(e(0)eθ1(a)e)(i(0)iθ2(a)i)cV)

    with

    Dom(A)=R×{0}×W1,1((0,),R)×{0}×W1,1((0,),R)×R,

    where W1,1 is a Sobolev space. Note that ¯Dom(A)=X0 is not dense in X.

    Define nonlinear operator F:¯Dom(A)X by

    F(T(0e)(0i)V)=(hβTV(fβTV0L1)((1f)βTV+M0L1)N).

    where

    M(t)=0ξ(a)e(a,t)da,  N(t)=0p(b)i(b,t)db. (3)

    Then by setting u(t)=(T(t),(0e(,t)),(0i(,t)),V(t))T, we can reformulate system (1) with the boundary and initial conditions as the following abstract Cauchy problem

    du(t)dt=Au(t)+F(u(t))  for  t0  and  u(0)X0+.

    If any initial value X0=(T0,e0(),i0(),V0)Y satisfies the coupling equations

    e(0,0)=fβT0V0

    and

    i(0,0)=(1f)βT0V0+0ξ(a)e0(a)da,

    then (1) is well-posed under Assumption 1.2 due to Iannelli [6] and Magal [14]. Denote

    Y=R+×L1+(0,)×L1+(0,)×R+

    with the norm

    (x,φ,ψ,y)Y=|x|+φL1+ψL1+|y|  for (x,φ,ψ,y)Y.

    In fact, for such solutions, it is not difficult to show that (T(t),e(,t),i(,t),V(t))Y for each t0. In the sequel, we always assume that the initial values satisfy the coupling equations.

    Using the results presented in [14,27], thus we can get a continuous solution semi-flow Φ:R+×YY defined by

    Φ(t,X0)=Φt(X0):=(T(t),e(,t),i(,t),V(t)),t0,  X0Y.

    The precise result is the following proposition.

    Proposition 1. For system (1), there exists a unique strongly continuous semiflow Φ: X0+X0+t0 such that for each x0X0+, the operator xC([0,),X0+) defined by x=Φ(t)x0 is a mild solution of (1), that is, it satisfies

    t0x(s)dsDom(A), and x(t)=x0+At0x(s)ds+t0F(x(s))ds,t0.

    2.2. Volterra formulation

    According to the Volterra formulation (see Webb [42] and Iannelli [6]), integrating the second and third equations of (1) along the characteristic lines ta=const. and tb=const. respectively yields

    e(a,t)={fβT(ta)V(ta)Ω(a)=e(0,ta)Ω(a),ift>a,e0(at)Ω(a)Ω(at),ifta; (4)

    and

    i(b,t)={[(1f)βT(tb)V(tb)+M(tb)]Γ(b)=i(0,tb)Γ(b),ift>b,i0(bt)Γ(b)Γ(bt),iftb. (5)

    Thus system (1) can be rewritten as the following Volterra-type equations,

    {dT(t)dt=hdT(t)βT(t)V(t),dV(t)dt=t0p(b)Γ(b)((1f)βT(tb)V(tb)+M(tb))db+tp(b)i0(bt)Γ(b)Γ(bt)dbcV(t),

    where M(tb)=tb0ξ(a)e(0,tba)da+tbξ(a)e0(atb)Ω(a)Ω(atb)da.


    2.3. Boundedness of solutions

    Proposition 2. Define

    Ξ:={X0=(T0,e0,i0,V0)Y|T0+e0(a)L1hμ0,T0+e0(a)L1+i0(b)L1hμ1,V0ˉphcμ0+hˉpˉξcμ20,X0Yh˜μ0},

    where ˜μ0 := μ01+ˉξμ0+ˉpc+ˉpˉξcμ0, μ1 := μ01+ˉξμ0. Then Ξ is a positively invariant subset for Φ, that is,

    Φ(t,X0)Ξforallt0andX0Ξ.

    Moreover, Φ is point dissipative and Ξ attracts all points in Y.

    Proof. By (4) and changes of variables, we have

    e(,t)L1= t0e(0,ta)Ω(a)da+te0(at)Ω(a)Ω(at)da= t0e(0,σ)Ω(tσ)dσ+0e0(τ)Ω(t+τ)Ω(τ)dτ.

    We derivative this equality,

    de(,t)L1dt=e(0,t)Ω(0)+t0e(0,σ)dΩ(tσ)dtdσ+0e0(τ)Ω(τ)dΩ(t+τ)dtdτ.

    By (2) and changing of variables, we have

    de(,t)L1dt= e(0,t)Ω(0)t0e(0,σ)θ1(tσ)Ω(tσ)dσ 0e0(τ)Ω(τ)θ1(t+τ)Ω(t+τ)dτ= e(0,t)Ω(0)0θ1(a)e(a,t)da.

    By the first equation in (1), (ⅳ) of Assumption 1.2, and use f<1,

    d(T(t)+e(,t)L1)dt=hdT(t)βT(t)V(t)+fβT(t)V(t)0θ1(a)e(a,t)dahμ0(T(t)+e(,t)L1).

    We integrate this differential inequality and obtain the a priori estimate,

    T(t)+e(,t)L1hμ0eμ0t{hμ0(T0+e(,t)L1)},t0. (6)

    This implies that

    T(t)+e(,t)L1hμ0.

    Similarly, we have

    di(,t)L1dt=i(0,t)Γ(0)0θ2(b)i(b,t)db.

    We add the two equations,

    d(T(t)+e(,t)L1)+i(,t)L1)dt=hdT+0ξ(a)e(a,t)da0θ1(a)e(a,t)da0θ2(b)i(b,t)db.

    and since (ⅱ) and (ⅳ) of Assumption 1.2, obtain the estimates

    d(T(t)+e(,t)L1)+i(,t)L1)dth+ˉξhμ0μ0(T(t)+e(,t)L1+i(,t)L1).

    We integrate this differential inequality and obtain the a priori estimate,

     T(t)+e(,t)L1)+i(,t)L1 h+ˉξhμ0μ0eμ0t{h+ˉξhμ0μ0(T(t)+e(,t)L1+i(,t)L1)},  t0. (7)

    This implies i(,t)L1hμ0+ˉξhμ20. Further, since (ⅱ) of Assumption 1.2, we have

    dV(t)dtˉpi(,t)L1cV(t)ˉp(hμ0+ˉξhμ20)cV(t).

    It follows (ⅳ) of Assumption 1.2, we have that

    V(t) ˉp(hμ0+ˉξhμ20)cect{ˉphcμ0+hˉpˉξcμ20V0} ˉphcμ0+hˉpˉξcμ20eμ0t{¯phcμ0+hˉpˉξcμ20V0} (8)

    Consequently, from (6), (7) and (8), we conclude that if X0Ξ, then for t0,

    Φt(X0)Y(1+ˉξμ0+ˉpc+ˉpˉξcμ0)hμ0 (9)
    eμ0t{(1+ˉξμ0+ˉpc+ˉpˉξcμ0)hμ0X0Y}=h˜μ0eμ0t{h˜μ0X0Y}h˜μ0. (10)

    In summary, we have shown that Ξ is positively invariant with respect to Φ. Lastly, it follows from (9) that lim suptΦt(X0)Yh˜μ0 for any X0Y, that is, Φ is point dissipative and Ξ attracts all points in Y. This completes the proof.

    As a consequence of Proposition 2, we have the following result.

    Proposition 3. Let Ah˜μ0 be given. If X0Y satisfying X0YA, then the following statements hold for all t0.

    (ⅰ) T(t), e(,t)L1, i(,t)L1, V(t)A;

    (ⅱ) M(t)ˉξA and N(t)ˉpA;

    (ⅲ) e(0,t)fβA2 and i(0,t)(1f)βA2+ˉξA.


    2.4. Existence of equilibria

    System (1) always has an infection-free equilibrium

    P0=(T0,e0(a),i0(b),V0):=(hd,0,0,0).

    The equations for an equilibrium are obtained from (1) by setting the time derivatives equal to 0 with boundary conditions, that is, infection equilibrium P=(T,e(),i(),V)Y of (1) satisfies

    {hdTβTV=0,ddae(a)=θ1(a)e(a),ddai(b)=θ2(b)i(b),0p(b)i(b)db=cV,e(0)=fβTV,i(0)=(1f)βTV+0ξ(a)e(a)da. (11)

    Denote

    K=0ξ(a)Ω(a)da,  J=0p(b)Γ(b)db.

    Biologically, K is the total number of infected cells activated by latency infected cells. J accounts for the total number of virus particles produced by an infected cell during its life-span, i.e., the burst size.

    We define basic reproduction number, 0 of (1) as

    0=fβT0KJc+(1f)βT0Jc.

    which accounts for the total number of virons resulted from a single viron through the virus-to-cell infection mod. 1f is the fraction of productive infection that leads to viral production, and fK represents the contribution to productively infected cells from activation of latently infected cells. 0 will serves as threshold value for (1), which completely determines the global behaviors of equilibria of (1).

    Direct calculation yields that if 0>1, then (1) admits a unique infection equilibrium P=(T,e(a),i(b),V) with

     T=T00, e(a)=fh(110)Ω(a), i(b)=(1f+fK)h(110)Γ(b), V=1c0p(b)i(b)db. (12)

    In summary, we have shown the following result.

    Proposition 4. (ⅰ) System (1) always has an infection-free equilibrium P0.

    (ⅱ) If 0>1, then (1) admits a unique infection equilibrium P, which is defined by (12).


    3. Asymptotic smoothness of Φ(t,X0)

    By Proposition 2 and 3, the semiflow is point-dissipative and Φ(R+×B) is bounded for every bounded subset B of Y. By Theorem 3.4.6 in [5], the semiflow has a compact attractor of bounded sets if it is asymptotically smooth. To give the existence of compact attractor, we follow the approach in [43,Theorem 4.2 of Chapter IV].

    Definition 3.1. [28] A set A in Y is called a compact attractor of a set BX if A is compact, invariant, and non-empty and Φt(B)A as t. The last means that, for every open subset U of Y with AU, there is some r>0 such that Φt(B)U for all tr (i.e. Φ([r,)×B)U).

    Recall that M(t) and N(t) are defined by (3). The following Proposition is devoted to prove basic properties of the functions M(t) and N(t) using Proposition 2, Assumption 1.2 and [38,Proposition 4.1].

    Proposition 5. For any solution of (1), the associated functions M(t) and N(t) are Lipschitz continuous on R+.

    Proof. Let t0 and h>0. We can check that

    M(t+h)M(t)= 0ξ(a)e(a,t+h)da0ξ(a)e(a,t)da h0ξ(a)e(a,t+h)da+hξ(a)e(a,t+h)da0ξ(a)e(a,t)da h0ξ(a)e(0,t+ha)Ω(a)da +hξ(a)e(a,t+h)da0ξ(a)e(a,t)da. (13)

    By applying ξ(a)ˉξ, e(0,t)fβA2 and Ω(a)1 for the first integral, and making the substitution σ=ah for the second integral to (13), we get

    M(t+h)M(t)fβA2ˉξh+0ξ(σ+h)e(σ+h,t+h)dσ0ξ(a)e(a,t)da

    It follows from (4) that

    e(σ+h,t+h)=e(σ,t)Ω(σ+h)Ω(σ).

    Thus,

    M(t+h)M(t) fβA2ˉξh+0(ξ(a+h)Ω(a+h)Ω(a)ξ(a))e(a,t)da= fβA2ˉξh+0(ξ(a+h)ea+haθ1(s)dsξ(a))e(a,t)da= fβA2ˉξh+0ξ(a+h)(ea+haθ1(s)ds1)e(a,t)da +0(ξ(a+h)ξ(a))e(a,t)da.

    From (ⅱ) of Assumption 1.2, we obtain θ1ha+haθ1(s)ds0. It follows that 1ea+haθ1(s)dse¯θ1h1¯θ1h. Therefore,

    0ξ(a+h)|ea+haθ1(s)ds1|ˉξ¯θ1h.

    Recall that 0e(a,t)daΦt(X0)YA. From (ⅲ) of Assumption 1.2, we obtain the following estimate,

    M(t+h)M(t)fβA2ˉξh+ˉξ¯θ1Ah+MξAh.

    Hence, M(t) is Lipschitz continuous with Lipschitz coefficients LM=(ˉξfβA+ˉξ¯θ1+Mξ)A. Similarly, it is easy to check that N(t) is Lipschitz continuous with Lipschitz coefficients LN=[ˉp(1f)βA+ˉpˉξ+ˉp¯θ2+Mp]A.

    Next we divide Φ: R+×YY into the following two operators Θ, Ψ: R+×YY:

    Θ(t,X0):=(0,˜φe(,t),˜φi(,t),0),Ψ(t,X0):=(T(t),˜e(,t),˜i(,t),V(t)),

    where

    ˜φe(a,t)={0,   if  t>a0,e(a,t),   if  at0; ˜φi(b,t)={0,  if  t>b0,i(b,t),  if  bt0;
    ˜e(a,t)={e(a,t),ift>a0,0,ifat0;˜i(b,t)={i(b,t),ift>b0,0,ifbt0.

    Then Φ(t,X0)=Θ(t,X0)+Ψ(t,X0) for t0. Following the proof of [42,Proposition 3.13], we can arrive at the following main result of this section.

    Theorem 3.2. For X0Ξ, the orbit {Φ(t,X0) | t0} has a compact closure in Y if the following two conditions hold,

    (ⅰ) There exists a function Δ: R+×R+R+ such that, for any r>0, limtΔ(t,r)=0 and if X0Ω with X0Yr then Θ(t,X0)YΔ(t,r) for t0;

    (ⅱ) For t0, Ψ(t,) maps any bounded sets of Ξ into sets with compact closure in Y.

    proof Proof of (ⅰ) of Theorem 3.2. Let Δ(t,r)=eμ0tr, then limtΔ(t,r)=0. By (4) and (5),

    ˜φe(a,t)={0,ift>a0,e0(at)Ω(a)Ω(at),ifat0;

    and

    ˜φi(b,t)={0,ift>b0,i0(bt)Γ(b)Γ(bt),ifbt0.

    Then, for X0Ξ satisfying X0Yr and for t0, we have

    Θ(t,X0)Y= |0|+˜φe(,t)L1+˜φi(,t)L1+|0|= t|e0(at)Ω(a)Ω(at)|da+t|i0(bt)Γ(b)Γ(bt)|db= 0|e0(σ)Ω(σ+t)Ω(σ)|dσ+0|i0(σ)Γ(σ+t)Γ(σ)|dσ= 0|e0(σ)eσ+tσθ1(τ)dτ|dσ+0|i0(σ)eσ+tσθ2(τ)dτ|dσ eμ0te0L1+eμ0ti0L1 eμ0tX0Y.

    Proof of (ⅱ) of Theorem 3.2. It is sufficient to show that Ψ(t,) maps any bounded sets of Ξ into sets with compact closure in Y. From Proposition 2, T(t) and V(t) remains in the compact set [0,h/˜μ0][0,A]. Thus it remains unknown that whether ˜e(a,t) and ˜i(b,t) remain in a precompact subset of L1+(0,), which is independent of X0Ξ. To this end, we next to verify the following conditions for ˜e(a,t) and similar ones for ˜i(b,t) (see, for example, [25,Theorem B.2]).

    (ⅰ) The supremum of ˜e(,t)L1 with respect to X0Ξ is finite;

    (ⅱ) limhh˜e(a,t)da=0 uniformly with respect to X0Ξ;

    (ⅲ) limh0+0|˜e(a+h,t)˜e(a,t)|da=0 uniformly with respect to X0Ξ;

    (ⅳ) limh0+h0˜e(a,t)da=0 uniformly with respect to X0Ξ.

    It follows from (4), (5), Proposition 3 and (2) that ˜e(a,t)fβA2eμ0a, ˜i(b,t)[(1f)βA2+ˉξA]eμ0b. Thus, (ⅰ), (ⅱ) and (ⅳ) are directly satisfied.

    Next we verify condition (ⅲ). For sufficiently small h(0,t), we have

    0|˜e(a+h,t)˜e(a,t)|da=th0|e(a+h,t)e(a,t)|da+tth|0e(a,t)|da=th0|e(0,tah)Ω(a+h)e(0,ta)Ω(a)|da+tth|e(0,ta)Ω(a)|daΔ1+Δ2+fβA2h,

    where

    Δ1=th0e(0,tah)|Ω(a+h)Ω(a)|da

    and

    Δ2=th0|e(0,tah)e(0,ta)|Ω(a)da.

    We first get an estimate of Δ1. Since

    th0|Ω(a+h)Ω(a)|da= th0(Ω(a)Ω(a+h))da= th0Ω(a)dathΩ(a)da= th0Ω(a)dathhΩ(a)datthΩ(a)da= h0Ω(a)datthΩ(a)da h,

    it follows from Proposition 3 that

    Δ1fβA2h.

    Next we estimate Δ2. We rewrite Δ2 as

    Δ2= th0|fβT(tah)V(tah)fβT(ta)V(ta)|Ω(a)da.

    It is easy to see that T(t) and V(t) are both Lipschitz continuous on R+ with Lipschitz constants MT=h+dA+βA2 and MV=(ˉp+c)A, respectively. According to [13,Proposition 6], we conclude that T(t)V(t)is Lipschitz continuous with Lipschitz constants MTV=AMV+AMT. Denote that G=fβMTV. This estimate immediately yields

    Δ2Ghth0eμ0adaGhμ0.

    Hence

    0|˜e(a+h,t)˜e(a,t)|da (2fβA2+Gμ0)h,

    and condition (ⅲ) directly follows.

    As to ˜i(b,t), we have

    0|˜i(b+h,t)˜i(b,t)|db=th0|i(b+h,t)i(b,t)|db+tth|0i(b,t)|db=th0|i(0,tbh)Γ(b+h)i(0,tb)Γ(b)|db+tth|i(0,tb)Γ(b)|dbΥ1+Υ2+[(1f)βA2+ˉξA]h,

    where

    Υ1=th0i(0,tbh)|Γ(b+h)Γ(b)|db

    and

    Υ2=th0|i(0,tbh)i(0,tb)|Γ(b)db.

    Similarly, we have th0|Γ(b+h)Γ(b)|dbh. Hence from Proposition 3, we can conclude that

    Υ1[(1f)βA2+ˉξA]h.

    Next we estimate Υ2. Firstly, we have

    Υ2=th0|(1f)βT(tah)V(tah)+M(tah)(1f)βT(ta)V(ta)M(ta)|Γ(b)da(1f)βth0|T(tah)V(tah)T(ta)V(ta)|Γ(b)da+th0|M(tah)M(ta)|Γ(b)da.

    As before, MTV=AMV+AMT. Recall that M(t) is Lipschitz continuous on R+ with Lipschitz constants LM=(ˉξfβA+ˉξ¯θ1+Mξ)A. Set H=(1f)βMTV+LM, By a zero-trick, then we have

    Υ2M1hth0eμ0bdbHhμ0.

    Finally, we have

    0|˜i(b+h,t)˜i(b,t)|db{2[(1f)βA2+ˉξA]+Hμ0}h,

    thus condition (ⅲ) directly follows. This completes the proof.

    Consequently, we have the following theorem for the semi-flow {Φ(t)}t0, which establish the existence of global attractors by Smith and Thieme [25].

    Theorem 3.3. The semi-flow {Φ(t)}t0 has a global attractor A in Y, which attracts any bounded subset of Y.


    4. The uniform persistence

    This section is spent on proving that (1) is uniformly persistent under the condition 0>1, which indicates that 0>1 is a threshold index for infection persistence.

    Let ˆe(t):=e(0,t) and ˆi(t):=i(0,t). We rewrite the first three equations of (1) as

    {dT(t)dt=hdT(t)1fˆe(t),e(a,t)={ˆe(ta)Ω(a),ifta0,e0(at)Ω(a)Ω(at),ifat0;i(b,t)={ˆi(tb)Γ(b),iftb0,i0(bt)Γ(b)Γ(bt),ifbt0, (14)

    where

    ˆe(t)= fβT(t)V(t) (15)

    and

    ˆi(t)=(1f)βT(t)V(t)+t0ξ(a)Ω(a)ˆe(ta)da+tξ(a)Ω(a)Ω(at)e0(at)da. (16)

    Lemma 4.1. If 0>1, then there exists a positive constant ϵ0>0 such that

    lim suptˆe(t)>ϵ0. (17)

    Proof. We first get an estimate on ˆi(t) as follows. By (16), we have

    ˆi(t)(1f)βT(t)V(t)+t0ξ(a)Ω(a)ˆe(ta)da. (18)

    Solving the fourth equation of (1) with initial condition V(0)=V0, we have that

    V(t)=V0ect+t00p(b)i(b,τ)dbec(tτ)dτ

    Then

    V(t)t0ec(tτ)τ0p(b)i(b,τ)dbdτ=t0ec(tτ)τ0p(b)Γ(b)ˆi(τb)dbdτ.

    This, combined with (18), gives us

    ˆi(t)(1f)βT(t)t0ec(tτ)τ0p(b)Γ(b)ˆi(τb)dbdτ+fβt0ξ(a)Ω(a)T(ta)ta0ec(taτ)τ0p(b)Γ(b)ˆi(τb)dbdτda. (19)

    Since 0>1, there exists a sufficiently small ϵ1>0(ϵ1=1fϵ0) such that

    (1f)βchϵ1d0p(b)Γ(b)db+fβchϵ1d0p(b)Γ(b)db0ξ(a)Ω(a)da> 1.

    We claim that (17) holds for this ε0.Suppose that there exists a T>0 such that

    ˆe(t)ϵ0  for all  tT.

    Then it follows from (14) that dT(t)dthdT(t)ϵ1 for tT. This implies that lim inftT(t)hε1d. Thus there exists ˆT>T such that T(t)hϵ1d for all tˆT and hence (19) becomes

    ˆi(t) (1f)βhϵ1dt0ec(tτ)τ0p(b)Γ(b)ˆi(τb)dbdτ +fβhϵ1dt0ξ(a)Ω(a)ta0ec(taτ)τ0p(b)Γ(b)ˆi(τb)dbdτda (20)

    for all tˆT. Without loss of generality, we can assume that (20) holds for all t0 (just replace X0 by Φ(ˆT,X0)). Then taking the Laplace transforms of both sides of (20), we obtain

    L[ˆi](1f)βh1d0eλtt0ec(tτ)τ0p(b)Γ(b)ˆi(τb)dbdτdt+fβh1d0eλtt0ξ(a)Ω(a)ta0ec(taτ)τ0p(b)Γ(b)ˆi(τb)dbdτdadt
    = (1f)βhϵ1d1c+λ0eλbp(b)Γ(b)dbL[ˆi] +fβhϵ1d1c+λ0eλbp(b)Γ(b)db0eλaξ(a)Ω(a)daL[ˆi].

    Here L[ˆi] denotes the Laplace transform of ˆi, which is strictly positive because of (15) and Assumption 1.2. Dividing both sides of the above inequality by L[ˆi] and letting λ0 give us

    1(1f)βchϵ1d0p(b)Γ(b)db+fβchϵ1d0p(b)Γ(b)db0ξ(a)Ω(a)da,

    which yields a contradiction.

    In order to apply a technique used by Smith and Thieme [25,Chapter 9] (see also McCluskey [13,Section 8]), we consider a total Φ-trajectory of (1) in space Y. A total trajectory of Φ is a function X:RY such that Φs(X(t))=X(t+s) for all tR and all s0. For a non-empty compact set ˜A, it is said to be a compact attractor of a class C of set if ˜A is invariant and d(Φt(C),˜A)0 for each CC. For each X0˜A, there exists a total trajectory X such that X(0)=X0 and X(t)˜A for all tR.

    Let ϕ:RY be a total Φ-trajectory such that ϕ(r)=(T(r),e(,r),i(,r),V(r)), rR. Then

    ϕ(r+t)= Φ(t,ϕ(r))  for  t0  and  rR,e(a,r)= e(0,ra)Ω(a)=ˆe(ra)Ω(a)  for  rR and  a0,i(b,r)= i(0,rb)Γ(b)=ˆi(rb)Γ(b)  for  rR  and  b0.

    So it follows from (14)-(15) that

    {dT(r)dr=hdT(r)1fˆe(r),ˆe(r)=fβT(r)V(r),ˆi(r)=(1f)βT(r)V(r)+0ξ(a)Ω(a)ˆe(ra)da,dV(r)dr=0p(b)Γ(b)ˆi(rb)dbcV(r),  for  rR.

    By the similar arguments as in McCluskey [13,Section 8] and Wang et al. [38,Section 5] and a slight modification of the proof in [17, Lemma 4.1], actually, a total Φ-trajectory ϕ enjoys the following nice properties.

    Thus Lemma 4.1 tells us that if 0>1 then the semi-flow Φ is uniformly weakly ρ-persistent. Moreover, with the help of Theorem 3.3 and the Lipschitz continuity of ˆi (which immediately follows from Proposition 5), we can apply Theorem 5.2 of Smith and Thieme [25] to conclude that the uniform weak ρ-persistence of the semi-flow Φ implies its uniform (strong) ρ-persistence, that is, we have obtained the following result.

    Theorem 4.2. If 0>1, then the semi-flow Φ is uniformly (strongly) ρ-persistent.

    When 0>1, the uniform persistence of (1) immediately follows from Theorem 4.2. In fact, it follows from (14) that e(,t)L1t0ˆe(ta)Ω(a)da and hence from a variation of the Lebesgue-Fatou lemma [26,Section B.2] we get

    lim infte(,t)L1ˆe0Ω(a)da,

    where ˆe=lim inftˆe(t). Under Theorem 4.2, there exists a positive constant ϵ>0 such that ˆe>ϵ if 0>1 and hence the persistence of e(a,t) with respect to L1 follows. By a similar argument, we can prove that T(t) and V(t) are persistent with respect to || and i(a,t) is persistent with respect to L1. In a summary, we get the following result.

    Theorem 4.3. If 0>1, then the semiflow {Φ(t)}t0 is uniformly persistent in Y, that is, there exists a constant ϵ>0 such that, for each X0Y,

    lim inftT(t)ϵ,lim infte(,t)L1ϵ,lim infti(,t)L1ϵ,lim inftV(t)ϵ.

    5. The local stability of equilibria

    This section is devoted to investigate the local stability of equilibria of (1).

    Theorem 5.1. (ⅰ) If 0<1, the infection-free equilibrium P0 of (1) is locally asymptotically stable while it is unstable if 0>1.

    (ⅱ) If 0>1, the infection equilibrium P of (1) is locally asymptotically stable.

    Proof. Proof. of (ⅰ) of Theorem 5.1. Linearizing (1) around the infection-free equilibrium P0 by using

    x1(t)=T(t)hd, x2(a,t)=e(a,t), x3(b,t)=i(b,t), x4(t)=V(t),

    we get

    {dx1(t)dt=dx1(t)βhdx4(t),(t+a)x2(a,t)=θ1(a)x2(a,t),(t+b)x3(b,t)=θ2(b)x3(b,t),dx4(t)dt=0p(b)x3(b,t)dbcx4(t),x2(0,t)=fβhdx4(t),x3(0,t)=(1f)βhdx4(t)+0ξ(a)x2(a,t)da. (21)

    Set

    x1(t)=x01eλt,  x2(a,t)=x02(a)eλt,  x3(b,t)=x03(b)eλt,  x4(t)=x04eλt, (22)

    where x01,x02(a),x03(b),x04 are to be determined later. Substituting (22) into (21), we have

    λx01=dx01hβdx04,
    {λx02(a)+dx02(a)da=θ1(a)x02(a),x02(0)=fβhdx04, (23)
    {λx03(b)+dx03(b)db=θ2(b)x03(b),x03(0)=(1f)hβdx04+0ξ(a)x02(a)da, (24)
    λx04=0p(b)x03(b)dbcx04. (25)

    We integrate the first equation of Equ (23) and Equ (24) from 0 to a,

    x02(a)= x02(0)eλaa0θ1(s)ds, (26)

    and

    x03(b)= x03(0)eλbb0θ2(s)ds= [(1f)hβdx04+0ξ(a)x02(a)da]eλbb0θ2(s)ds. (27)

    From (25), (26) and (27),

    x04= 0p(b)x03(b)dbλ+c= 1fλ+chβdx040p(b)eλbb0θ2(s)dsdb +x02(0)λ+chd0ξ(a)eλaa0θ1(s)dsda0p(b)eλbb0θ2(s)dsdb. (28)

    Combining (23) into (28), it follows the equation that

    W(λ)=1, (29)

    where

    W(λ)= fβλ+chd0ξ(a)eλaa0θ1(s)dsda0p(b)eλbb0θ2(s)dsdb +1fλ+chβd0p(b)eλbb0θ2(s)dsdb.

    Since W have the properties that

    limλW(λ)=0, limλW(λ)=, W(λ)<0,

    Thus (29) admits a unique real root, λ. Recall that W(0)=0, it follows that, λ<0 if 0<1 and λ>0 if 0>1, that is, P0 is unstable if 0>1. Suppose that 0<1. Let λ=μ+νi be an arbitrary complex root of (29). It is easy to see that

    1=|W(λ)|=|W(μ+νi)|W(μ),

    which implies that 0>λμ. Hence all roots of (29) have negative real parts, that is P0 is locally asymptotically stable if 0<1.

    Proof of (ⅱ) of Theorem 5.1. Linearizing the system (1) at P by using

     y1(t)=T(t)T, y2(a,t)=e(a,t)e(a), y3(b,t)=i(b,t)i(b), y4(t)=V(t)V,

    we get

    {dy1(t)dt=d0y1(t)βTy4(t),(t+a)y2(a,t)=θ1(a)y2(a,t),(t+b)y3(b,t)=θ2(b)y3(b,t),dy4(t)dt=0p(b)y3(b,t)dbcy4(t),y2(0,t)=fd(01)y1(t)+fβTy4(t),y3(0,t)=(1f)d(01)y1(t)+(1f)βTy4(t)+0ξ(a)y2(a,t)da, (30)

    Set

    y1(t)=y01eλt,  y2(a,t)=y02(a)eλt,  y3(b,t)=y03(b)eλt,  y4(t)=y04eλt, (31)

    where y01,y02(a),y03(b),y04 are to be determined. Substituting (31) into (30) yields

    λy01=d0y01βTy04, (32)
    {λy02(a)+dy02(a)da=θ1(a)y02(a),y02(0)=fd(01)y01+fβTy04, (33)
    {λy03(b)+dy03(b)db=θ2(b)y03(b),y03(0)=(1f)d(01)y01+(1f)βTy04+y02(0)0ξ(a)eλaa0θ1(s)dsda, (34)

    and

    λy04=0p(b)y03(b)dbcy04. (35)

    We integrate the first equation of (33), (34) from 0 to a,

    y02(a)= y02(0)eλaa0θ1(s)ds,

    and

    y03(b)= y03(0)eλbb0θ2(s)ds= [(1f)d(01)y01+(1f)βTy04]eλbb0θ2(s)ds +y02(0)0ξ(a)eλaa0θ1(s)dsdaeλbb0θ2(s)ds.

    and from (35), we have

    y04= 0p(b)y03(b)dbλ+c= 1fλ+c(d(01)y01+βTy04)0p(b)eλbb0θ2(s)dsdb+y02(0)λ+c0ξ(a)eλaa0θ1(s)dsda0p(b)eλbb0θ2(s)dsdb. (36)

    Combining (32), (33) into (36), yields the characteristic equation at P that

    G(λ)=(λ+d)W1(λ)λd0=0, (37)

    where

    W1(λ)= (1f)βTλ+c0p(b)eλbb0θ2(s)dsdb+fβTλ+c0ξ(a)eλaa0θ1(s)dsda0p(b)eλbb0θ2(s)dsdb.

    It is sufficient to show that (37) has no roots with non-negative real parts. Suppose that it has a root λ=μ+νi with μ0. Then we have

    (μ+νi+d)W1(μ+νi)μνid0=0.

    Separating the real part of the above equality gives

    Re W1(μ+νi)=(μ+d0)(μ+d)+ν2(μ+d)2+ν2>1. (38)

    Noticing that W1(0)=T0T0=1 and W1 is a decreasing function, we have

    Re W1(μ+νi)|W1(μ)|=W1(μ)W1(0)=1,

    which yields a contradiction. This completes the proof.


    6. Global stability of equilibria

    This section is devoted to investigate the global stability of the equilibria by using Lyapunov functionals under the threshold value. In what follows, we introduce an important function g on (0,) defined by g(x)=x1lnx for x(0,). This function is continuous and concave up with g(1)=0. By Theorem 5.1, it is suffice to show that equilibria of (1) are globally attractive in Y.

    Theorem 6.1. The infection-free equilibrium P0 of (1) is globally attractive if 01.

    Proof. Considering the candidate Lyapunov function as follows,

    LIFE(t)=L1(t)+L2(t)+L3(t)+L4(t),

    where L1(t)=T0g(T(t)T0), L2(t)=0ϕ(a)e(a,t)da, L3(t)=0ψ(b)i(b,t)db, and L4(t)=βT0cV(t). Here the nonnegative kernel functions ϕ(a) and ψ(b) will be determined later. Firstly, we calculate the derivative of Li, i=1,2,3,4, respectively,

    dL1(t)dt=dT0(T0T+TT02)βTV+βT0V.

    By integration by parts, we calculate the derivative of L2,

    dL2(t)dt= 0ϕ(a)e(a,t)tda=0ϕ(a)[θ1(a)e(a,t)+e(a,t)a]da= ϕ(a)e(a,t)|0+0ϕ(a)e(a,t)da0ϕ(a)θ1(a)e(a,t)da= ϕ(0)e(0,t)+0(ϕ(a)ϕ(a)θ1(a))e(a,t)da.

    An argument similar to the one used in calculating the derivative of L2, we get

    dL3(t)dt=ψ(0)i(0,t)+0(ψ(b)ψ(b)θ2(b))i(b,t)db.

    We calculate the derivative of L4,

    dL4(t)dt=βT0c0p(b)i(b,t)dbβT0V.

    Secondly, we have

    dLIFE(t)dt=dT0(T0T+TT02)βTV+ϕ(0)fβTV+ψ(0)(1f)βTV+0(ϕ(a)ϕ(a)θ1(a)+ψ(0)ξ(a))e(a,t)da+0(ψ(b)ψ(b)θ2(b)+βT0cp(b))i(b,t)db.

    Choosing

    {ψ(b)=bβT0cp(u)eubθ2(ω)dωdu,ϕ(a)=aψ(0)ξ(u)euaθ1(ω)dωdu.

    Then it is easy to see that

    {ψ(0)=βT0Jc, ϕ(0)=βT0JKc,ψ(b)ψ(b)θ2(b)+βT0cp(b)=0,ϕ(a)ϕ(a)θ1(a)+ψ(0)ξ(a)=0.

    Consequently, LIFE satisfies

    dLIFE(t)dt=dT0(T0T+TT02)+(01)βTV.

    Notice that dLIFE(t)dt=0 implies that T=T0. It can be verified that the largest invariant set where dLIFE(t)dt=0 is the singleton {P0}. Therefore, by the invariance principle, P0 is globally attractive when 01.

    To establish the global stability of the infection equilibrium, we introduce the following Lemma.

    Lemma 6.2. Suppose that 0>1. Then, for any solution (T(t),e(a,t),i(b,t),V(t)) of (1), the following equalities hold,

    (1f)βTV[1e(0,t)i(0)e(0)i(0,t)]+0ξ(a)e(a)[1e(a,t)i(0)e(a)i(0,t)]da=0, (39)

    Proof. We give the proof for (39). In fact,

     (1f)βTV+0ξ(a)e(a)da (1f)βTVe(0,t)i(0)e(0)i(0,t)0ξ(a)e(a)e(a,t)i(0)e(a)i(0,t)da
    = i(0)((1f)βTV+0ξ(a)e(a,t)da)i(0)i(0,t)= 0

    This immediately gives (39).

    Theorem 6.3. If 0>1, then the infection equilibrium P=(T,e(a),i(a),V) of (1) is globally attractive.

    Proof. Let

    G[x,y]=xyylnxy, for x,y>0.

    It is easy to see that G is non-negative on (0,)×(0,) with the minimum value 0 only when x=y. Furthermore, it is easy to verify that xGx[x,y]+yGy[x,y]=G[x,y].

    Considering the following candidate Lyapunov function,

    LEE(t)=H1(t)+H2(t)+H3(t)+H4(t),

    where

     H1(t)=G[T,T], H2(t)=0ϕ1(a)G[e(a,t),e(a)]da, H3(t)=0ψ1(b)G[i(b,t),i(b)]db, H4(t)=βTcG[V,V].

    We define ϕ1(a) and ψ1(b) as

    ψ1(b)=bβTcp(u)eubθ2(ω)dωdu,

    and

    ϕ1(a)=aψ1(0)ξ(u)euaθ1(ω)dωdu,

    it follows that ψ1(0)=βTJc, ϕ1(0)=βTKJc and

    ψ1(b)ψ1(b)θ2(b)= βTcp(b).
    ϕ1(a)ϕ1(a)θ1(a)= ψ1(0)ξ(a).

    Firstly, we calculate the derivative of Hi, i=1,2,3,4, respectively,

    dH1(t)dt=dT(TT+TT2)+1f(1TT)(e(0)e(0,t))

    By using (4),

    H2(t)= t0ϕ1(a)G[e(0,ta)Ω(a),e(a)]da +tϕ1(a)G[e0(at)eaatθ1(ω)dω,e(a)]da= t0ϕ1(tr)G[e(0,r)Ω(tr),e(tr)]dr +0ϕ1(t+r)G[e0(r)et+rrθ1(ω)dω,e(t+r)]dr= B1(t)+B2(t).

    The derivative of B1 and B2 take the following form,

     dB1(t)dt=ϕ1(0)G[e(0,t),e(0)]+t0ϕ1(tr)G[e(0,r)etr0θ1(ω)dω,e(tr)]dr t0ϕ1(tr)θ1(tr)[e(0,r)etr0θ1(ω)dωGx[e(0,r)etr0θ1(ω)dω,e(tr)] +e(tr)Gy[e(0,r)etr0θ1(ω)dω,e(tr)]]dr,

    and

     dB2(t)dt=0ϕ1(t+r)G[e0(r)et+rrθ1(ω)dω,e(t+r)]dr 0ϕ1(t+r)θ1(t+r)[e0(r)et+rrθ1(ω)dωGx[e0(r)et+rrθ1(ω)dω,e(t+r)] +e(t+r)Gy[e0(r)et+rrθ1(ω)dω,e(t+r)]]dr.

    We obtain the derivative of H2(t),

    dH2(t)dt=ϕ1(0)G[e(0,t),e(0)]+0[ϕ1(a)ϕ1(a)θ1(a)]G[e(a,t),e(a)]da=ϕ1(0)G[e(0,t),e(0)]0ψ1(0)ξ(a)G[e(a,t),e(a)]da.

    A similar argument as in the derivative of H2, we calculate the derivative of H3,

    dH3(t)dt= ψ1(0)G[i(0,t),i(0)]+0[ψ(b)ψ(b)θ2(b)]G[i(b,t),i(b)]db= ψ1(0)G[i(0,t),i(0)]0βTcp(b)G[i(b,t),i(b)]db.

    We calculate the derivative of H4,

    dH4(t)dt=βTc0p(b)i(b,t)dbβTV+βTVβTVcV0p(b)i(b,t)db.

    If follows from ψ1(0)=βTJc and ϕ1(0)=βTKJc that

    dLEEdt=dT(TT+TT2)+1f(1TT)(e(0)e(0,t))+ϕ1(0)G[e(0,t),e(0)]0ψ1(0)ξ(a)G[e(a,t),e(a)]da+ψ1(0)G[i(0,t),i(0)]0βTcp(b)G[i(b,t),i(b)]db+0βTcp(b)i(b,t)db+βTVβTVVV0βTcp(b)i(b,t)db. (40)

    Recall that

    (1f)(βTVβTV)+0ξ(a)(e(a)e(a,t))da=i(0)i(0,t),

    and

    fβTKJc+(1f)βTJc=(fβKJc+(1f)βJc)T00=1.

    Thus (40) becomes

    dLEE(t)dt=dT(TT+TT2)+1f(1TT)(e(0)e(0,t))+1fG[e(0,t),e(0)]0βTcp(b)G[i(b,t),i(b)]db+βTJc[(1f)βTVlne(0,t)i(0)e(0)i(0,t)+0ξ(a)e(a)lne(a,t)i(0)e(a)i(0,t)da]+0βTcp(b)i(b,t)db+βTVβTVVV0βTcp(b)i(b,t)db.

    It follows that,

    dLEE(t)dt=dT(TT+TT2)1fe(0)(TTlne(0,t)e(0))0βTcp(b)G[i(b,t),i(b)]db+βTJc[(1f)βTVlne(0,t)i(0)e(0)i(0,t)+0ξ(a)e(a)lne(a,t)i(0)e(a)i(0,t)da]+0βTcp(b)i(b,t)db+βTVVV0βTcp(b)i(b,t)db. (41)

    Recall that e(0)=fβTV and 0p(b)i(b)db=cV in (11). Collecting the terms of (41) yields

     dLEE(t)dt=dT(TT+TT2) +βTJc[(1f)βTVlne(0,t)i(0)e(0)i(0,t)+0ξ(a)e(a)lne(a,t)i(0)e(a)i(0,t)da] +0βTcp(b)i(b)(2+lni(b,t)i(b)TTlne(0,t)e(0)Vi(b,t)Vi(b))db.

    Further, we have

    dLEE(t)dt=dT(TT+TT2)+βTJc(1f)βTV(1e(0,t)i(0)e(0)i(0,t)+lne(0,t)i(0)e(0)i(0,t))+βTJc0ξ(a)e(a)(1e(a,t)i(0)e(a)i(0,t)+lne(a,t)i(0)e(a)i(0,t))da+0βTp(b)ci(b)(2TT+lnTTVi(b,t)Vi(b)+lnVi(b,t)Vi(b))db (42)
     βTJc{(1f)βTV[1e(0,t)i(0)e(0)i(0,t)] +0ξ(a)e(a)[1e(a,t)i(0)e(a)i(0,t)]da}.

    Recall that Lemma 6.2 holds. Putting (39) into (42), we have

    dLEE(t)dt= dT(TT+TT2) βTJc[0ξ(a)e(a)g(e(a,t)i(0)e(a)i(0,t))da +(1f)0βTcp(b)i(b)g(e(0,t)i(0)e(0)i(0,t))db] 0βTcp(b)i(b)[g(TT)+g(Vi(b,t)Vi(b))]db 0

    and dLEE(t)dt=0 implies that T=T and

    i(b,t)i(b)=i(0,t)i(0)=VV=e(0,t)e(0)=e(a,t)e(a),  for all ab0.

    It is not difficult to check that the largest invariant subset {dLEE(t)dt=0} is the singleton {P}. By the invariance principle, P is globally attractive and hence the proof is complete.


    7. Discussion

    This paper is devoted to the global dynamics of an HIV infection model subject to latency age and infection age, where the model formulation, basic reproduction number computation, and rigorous mathematical analysis, such as relative compactness and persistence of the solution semi-flows, and existence of a global attractor are involved. We have shown that the existence of a compact attractor of all compact sets of nonnegative initial data and used the Lyapunov functional to show that this attractor is the singleton set containing the equilibrium. Given that the model is so complex, the proof does require some rigorous calculation. The dynamics (at least the long-term dynamics) of the model do not appear to have been altered by adding the e(a,t) component. We hope the model studied here have a contribution to improve the broader contexts of investigating viral infection subject to age structure.


    Acknowledgments

    The authors would like to thank the anonymous referees and editor for very helpful suggestions and comments which led to improvements of our original manuscript. J. Wang is supported by National Natural Science Foundation of China (No. 11226255 and No. 11201128), the Science and Technology Innovation Team in Higher Education Institutions of Heilongjiang Province (No. 2014TD005). X. Dong is supported by Graduate Students Innovation Research Program of Heilongjiang University (No. YJSCX2017-177HLJU).


    [1] [ A. Alshorman,C. Samarasinghe,W. Lu,L. Rong, An HIV model with age-structured latently infected cells, J. Biol. Dyna., 11 (2017): 192-215.
    [2] [ C. J. Browne, A multi-strain virus model with infected cell age structure: Application to HIV, Nonlinear Anal. RWA, 22 (2015): 354-372.
    [3] [ C. Castillo-Chavez,Z. Feng, To treat or not to treat: The case of tuberculosis, J. Math. Biol., 35 (1997): 629-656.
    [4] [ G. Huang,X. Liu,Y. Takeuchi, Lyapunov functions and global stability for age-structured HIV infection model, SIAM J. Appl. Math., 72 (2012): 25-38.
    [5] [ J. K. Hale, Asymptotic Behavior of Dissipative Systems, AMS, Providence, 1988.
    [6] [ M. Iannelli, Mathematical Theory of Age-Structured Population Dynamics, Appl. Math. Monogr. CNR, vol. 7, Giadini Editori e Stampator, Pisa, 1994.
    [7] [ H. Kim and A. S. Perelson, Viral and latent reservoir persistence in HIV-1-infected patients on therapy, PLoS Comput. Biol., 10 (2006), e135.
    [8] [ M. Y. Li,H. Shu, Impact of intracellular delays and target-cell dynamics on in vivo viral infections, SIAM J. Appl. Math., 70 (2010): 2434-2448.
    [9] [ X. Lai,X. Zou, Modeling the HIV-1 virus dynamics with both virus-to-cell and cell-to-cell transmission, SIAM J. Appl. Math., 74 (2014): 898-917.
    [10] [ M. Y. Li,H. Shu, Global dynamics of an in-host viral model with intracellular delay, Bull. Math. Bio., 72 (2010): 1492-1505.
    [11] [ V. Muller,J. F. Vigueras-Gomez,S. Bonhoeffer, Decelerating decay of latently infected cells during prolonged therapy for human immunodeficiency virus type 1 infection, J. Virol., 76 (2002): 8963-8965.
    [12] [ M. Martcheva,C. Castillo-Chavez, Diseases with chronic stage in a population with varying size, Math. Biosci., 182 (2003): 1-25.
    [13] [ C. C. McCluskey, Global stability for an SEI epidemiological model with continuous age-structure in the exposed and infectious classes, Math. Biosci. Eng., 9 (2012): 819-841.
    [14] [ P. Magal, Compact attractors for time periodic age-structured population models, Electron. J. Differ. Equ., 65 (2001): 1-35.
    [15] [ P. W. Nelson,M. A. Gilchrist,D. Coombs,J. M. Hyman,A. S. Perelson, An age-structured model of HIV infection that allow for variations in the production rate of viral particles and the death rate of productively infected cells, Math. Biosci. Eng., 1 (2004): 267-288.
    [16] [ Y. Nakata, Global dynamics of a viral infection model with a latent period and beddington-DeAngelis response, Nonlinear Anal. TMA, 74 (2011): 2929-2940.
    [17] [ Y. Nakata, Global dynamics of a cell mediated immunity in viral infection models with distributed delays, J. Math. Anal. Appl., 375 (2011): 14-27.
    [18] [ A. S. Perelson,P. Essunger,Y. Cao,M. Vesanen,A. Hurley,K. Saksela,M. Markowitz,D. D. Ho, Decay characteristics of HIV-1-infected compartments during combination therapy, Nature, 387 (1997): 188-191.
    [19] [ L. Rong and A. S. Perelson, Modeling latently infected cell activation: Viral and latent reservoir persistence, and viral blips in HIV-infected patients on potent therapy, PLoS Comput. Biol., 5 (2009), e1000533, 18pp.
    [20] [ L. Rong,A. S. Perelson, Asymmetric division of activated latently infected cells may explain the decay kinetics of the HIV-1 latent reservoir and intermittent viral blips, Math. Biosci., 217 (2009): 77-87.
    [21] [ L. Rong,A. S. Perelson, Modeling HIV persistence, the latent reservoir, and viral blips, J. Theoret. Biol., 260 (2009): 308-331.
    [22] [ M. C. Strain,H. F. Gunthard,D. V. Havlir,C. C. Ignacio,D. M. Smith,A. J. Leigh-Brown,T. R. Macaranas,R. Y. Lam,O. A. Daly,M. Fischer,M. Opravil,H. Levine,L. Bacheler,C. A. Spina,D. D. Richman,J. K. Wong, Heterogeneous clearance rates of long-lived lymphocytes infected with HIV: Intrinsic stability predicts lifelong persistence, Proc. Natl. Acad. Sci. USA,, 191 (2005): 1410-1418.
    [23] [ M. C. Strain,S. J. Little,E. S. Daar,D. V. Havlir,H. F. Günthard,R. Y. Lam,O. A. Daly,J. Nguyen,C. C. Ignacio,C. A. Spina,D. D. Richman,J. K. Wong, Effect of treatment, during primary infection, on establishment and clearance of cellular reservoirs ofHIV-1, J. Infect. Dis., 191 (2005): 1410-1418.
    [24] [ H. Shu,L. Wang,J. Watmough, Global stability of a nonlinear viral infection model with infinitely distributed intracellular delays and CTL immune responses, SIAM J. Appl. Math., 73 (2013): 1280-1302.
    [25] [ H. L. Smith and H. R. Thieme, Dynamical Systems and Population Persistence, Amer. Math. Soc., Providence, 2011.
    [26] [ H. L. Smith, Mathematics in Population Biology, Princeton University Press, 2003.
    [27] [ H. R. Thieme, Semiflows generated by Lipschitz perturbations of non-densely defined operators, Dif. Int. Eqs., 3 (1990): 1035-1066.
    [28] [ H. R. Thieme, Global stability of the endemic equilibrium in infinite dimension: Lyapunov functions and positive operators, J. Differential Equations, 250 (2011): 3772-3801.
    [29] [ H. R. Thieme,C. Castillo-Chavez, How may infection-age-dependent infectivity affect the dynamics of HIV/AIDS?, SIAM J. Appl. Math., 53 (1993): 1447-1479.
    [30] [ V. Volterra, Leçns Sur la Thérie Mathématique de la Lutte Pour la vie, Éditions Jacques Gabay, Sceaux, 1990.
    [31] [ J. Wang,S. Liu, The stability analysis of a general viral infection model with distributed delays and multi-staged infected progression, Commun. Nonlinear Sci. Numer. Simulat., 20 (2015): 263-272.
    [32] [ J. Wang,R. Zhang,T. Kuniya, Global dynamics for a class of age-infection HIV models with nonlinear infection rate, J. Math. Anal. Appl., 432 (2015): 289-313.
    [33] [ J. Wang,R. Zhang,T. Kuniya, Mathematical analysis for an age-structured HIV infection model with saturation infection rate, Electron. J. Differential Equations,, 33 (2015): 1-19.
    [34] [ J. Wang,G. Huang,Y. Takeuchi, Global asymptotic stability for HIV-1 dynamics with two distributed delays, Math. Med. Biol., 29 (2012): 283-300.
    [35] [ J. Wang,L. Guan, Global stability for a HIV-1 infection model with cell-mediated immune response and intracellular delay, Discrete Cont. Dyn. Sys. B, 17 (2012): 297-302.
    [36] [ J. Wang,J. Pang,T. Kuniya,Y. Enatsu, Global threshold dynamics in a five-dimensional virus model with cell-mediated, humoral immune responses and distributed delays, Appl. Math. Comput., 241 (2014): 298-316.
    [37] [ J. Wang,J. Lang,X. Zou, Analysis of a structured HIV infection model with both virus-to-cell infection and cell-to-cell transmission, Nonlinear Analysis: RWA, 34 (2017): 75-96.
    [38] [ J. Wang,R. Zhang,T. Kuniya, The dynamics of an SVIR epidemiological model with infection age, IMA J. Appl. Math., 81 (2016): 321-343.
    [39] [ X. Wang,S. Liu, A class of delayed viral models with saturation infection rate and immune response, Math. Meth. Appl. Sci., 36 (2013): 125-142.
    [40] [ X. Wang,S. Tang,X. Song,L. Rong, Mathematical analysis of an HIV latent infection model including both virus-to-cell infection and cell-to-cell transmission, J. Biol. Dyna., 11 (2017): 455-483.
    [41] [ D. Wodarz, -Hepatitis c virus dynamics and pathology: The role of CTL and antibody responses, J. Gen. Virol., 84 (2003): 1743-1750.
    [42] [ G. F. Webb, Theory of Nonlinear Age-Dependent Population Dynamics, Marcel Dekker, New York and Basel, 1985.
    [43] [ J. A. Walker, Dynamical Systems and Evolution Equations, Plenum Press, New York and London, 1980.
    [44] [ Y. Yan,W. Wang, Global stability of a five-dimensional model with immune responses and delay, Discrete Cont. Dyn. Sys. B, 17 (2012): 401-416.
  • This article has been cited by:

    1. Cameron J. Browne, Xuejun Pan, Hongying Shu, Xiang-Sheng Wang, Resonance of Periodic Combination Antiviral Therapy and Intracellular Delays in Virus Model, 2020, 82, 0092-8240, 10.1007/s11538-020-00704-3
    2. Jahangir Alam, Ghulam Murtaza, Efstratios Tzirtzilakis, Mohammad Ferdows, Biomagnetic Fluid Flow and Heat Transfer Study of Blood with Gold Nanoparticles over a Stretching Sheet in the Presence of Magnetic Dipole, 2021, 6, 2311-5521, 113, 10.3390/fluids6030113
    3. Zizi Wang, Qian Zhang, Yong Luo, A general non-local delay model on oncolytic virus therapy, 2022, 102, 0307904X, 423, 10.1016/j.apm.2021.09.045
    4. Zhongzhong Xie, Xiuxiang Liu, Global dynamics in an age-structured HIV model with humoral immunity, 2021, 14, 1793-5245, 2150047, 10.1142/S1793524521500479
    5. Junfeng Yin, Xianqin Zhang, M. Israr Ur Rehman, Aamir Hamid, Thermal radiation aspect of bioconvection flow of magnetized Sisko nanofluid along a stretching cylinder with swimming microorganisms, 2022, 30, 2214157X, 101771, 10.1016/j.csite.2022.101771
    6. Chunyang Li, Xiu Dong, Jinliang Wang, Stability analysis of an age-structured viral infection model with latency, 2022, 2022, 1072-6691, 16, 10.58997/ejde.2022.16
    7. Abderrazak Nabti, Salih Djilali, Malek Belghit, Dynamics of a Double Age-Structured SEIRI Epidemic Model, 2025, 196, 0167-8019, 10.1007/s10440-025-00723-z
  • Reader Comments
  • © 2018 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(4862) PDF downloads(813) Cited by(7)

Article outline

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog