Review

Ground glass hepatocytes provide targets for therapy or prevention of hepatitis B virus-related hepatocellular carcinoma

  • Received: 12 November 2017 Accepted: 08 March 2018 Published: 14 March 2018
  • Ground glass hepatocyte (GGH) represents a histologic hallmark of chronic hepatitis B virus (HBV) infection and is characterized by the accumulation of pre-S mutant surface antigens in the endoplasmic reticulum (ER). In the past decade, GGHs have been recognized as pre-neoplastic lesions of hepatocellular carcinoma (HCC). The accumulation of pre-S mutant protein in ER may induce a misfolded protein response or ER stress signals with activation of VEGF/Akt/mTOR and COX-2/NF-κB signals, leading to oxidative DNA damage, aneuploidy, and genomic instability. Molecular studies revealed clonal HBV DNA integration in type II GGHs which continue to express and secrete surface antigens, representing the sustained surface antigens in the serum after NA antiviral treatment. The persistence of GGHs in the liver after anti-viral therapy not only constitute the challenge to eliminate HBV infection but also carry the high risk to develop HCC. DNA chip and ELISA kit are designing to detect the pre-S mutants in serum. Novel or second generation anti-HBV drugs are under phase II development and include the combination of anti-virals, immunomodulators, agents for host DNA damage, and siRNA to target at the transcription of HBsAg gene in cccDNA or integrated HBV DNA. In the past years, we explored the possibility to provide drugs or natural agents targeting at ER stress signals in GGHs to prevent HCC development. In a transgenic mice model of pre-S mutant and HBx, a combination of silymarin and resveratrol targeting at mTOR and NF-κB signals could reduce a 80% of HCC development. In a pilot clinical trial, liposomal curcumin (Meriva®) combined with anti-virals and immumodulators P1101 have been designed and attempted to eliminate the serum surface antigen and hence the recurrence of HCC after surgical resection. Therefore, the detection of pre-S mutants in serum or GGHs in the liver should provide rational target design for therapy or prevention of HCC in these high risk patients of chronic HBV infection.

    Citation: Hong-Yi Chang, Hung-Wen Tsai, Chiao-Fang Teng, Lily Hui-Ching Wang, Wenya Huang, Ih-Jen Su. Ground glass hepatocytes provide targets for therapy or prevention of hepatitis B virus-related hepatocellular carcinoma[J]. AIMS Medical Science, 2018, 5(2): 90-101. doi: 10.3934/medsci.2018.2.90

    Related Papers:

    [1] Shuang Hong Ma, Hai Feng Huo, Hong Xiang, Shuang Lin Jing . Global dynamics of a delayed alcoholism model with the effect of health education. Mathematical Biosciences and Engineering, 2021, 18(1): 904-932. doi: 10.3934/mbe.2021048
    [2] Pengyan Liu, Hong-Xu Li . Global behavior of a multi-group SEIR epidemic model with age structure and spatial diffusion. Mathematical Biosciences and Engineering, 2020, 17(6): 7248-7273. doi: 10.3934/mbe.2020372
    [3] Xiaomei Feng, Zhidong Teng, Fengqin Zhang . Global dynamics of a general class of multi-group epidemic models with latency and relapse. Mathematical Biosciences and Engineering, 2015, 12(1): 99-115. doi: 10.3934/mbe.2015.12.99
    [4] Qianqian Cui . Global stability of multi-group SIR epidemic model with group mixing and human movement. Mathematical Biosciences and Engineering, 2019, 16(4): 1798-1814. doi: 10.3934/mbe.2019087
    [5] Simphiwe M. Simelane, Phumlani G. Dlamini, Fadekemi J. Osaye, George Obaido, Blessing Ogbukiri, Kehinde Aruleba, Cadavious M. Jones, Chidozie W. Chukwu, Oluwaseun F. Egbelowo . Modeling the impact of public health education on tungiasis dynamics with saturated treatment: Insight through the Caputo fractional derivative. Mathematical Biosciences and Engineering, 2023, 20(5): 7696-7720. doi: 10.3934/mbe.2023332
    [6] Salih Djillali, Soufiane Bentout, Tarik Mohammed Touaoula, Abdessamad Tridane . Global dynamics of alcoholism epidemic model with distributed delays. Mathematical Biosciences and Engineering, 2021, 18(6): 8245-8256. doi: 10.3934/mbe.2021409
    [7] Simphiwe M. Simelane, Justin B. Munyakazi, Phumlani G. Dlamini, Oluwaseun F. Egbelowo . Projections of human papillomavirus vaccination and its impact on cervical cancer using the Caputo fractional derivative. Mathematical Biosciences and Engineering, 2023, 20(7): 11605-11626. doi: 10.3934/mbe.2023515
    [8] Jinhu Xu, Yicang Zhou . Global stability of a multi-group model with vaccination age, distributed delay and random perturbation. Mathematical Biosciences and Engineering, 2015, 12(5): 1083-1106. doi: 10.3934/mbe.2015.12.1083
    [9] Zhisheng Shuai, P. van den Driessche . Impact of heterogeneity on the dynamics of an SEIR epidemic model. Mathematical Biosciences and Engineering, 2012, 9(2): 393-411. doi: 10.3934/mbe.2012.9.393
    [10] Amira Bouhali, Walid Ben Aribi, Slimane Ben Miled, Amira Kebir . Impact of immunity loss on the optimal vaccination strategy for an age-structured epidemiological model. Mathematical Biosciences and Engineering, 2024, 21(6): 6372-6392. doi: 10.3934/mbe.2024278
  • Ground glass hepatocyte (GGH) represents a histologic hallmark of chronic hepatitis B virus (HBV) infection and is characterized by the accumulation of pre-S mutant surface antigens in the endoplasmic reticulum (ER). In the past decade, GGHs have been recognized as pre-neoplastic lesions of hepatocellular carcinoma (HCC). The accumulation of pre-S mutant protein in ER may induce a misfolded protein response or ER stress signals with activation of VEGF/Akt/mTOR and COX-2/NF-κB signals, leading to oxidative DNA damage, aneuploidy, and genomic instability. Molecular studies revealed clonal HBV DNA integration in type II GGHs which continue to express and secrete surface antigens, representing the sustained surface antigens in the serum after NA antiviral treatment. The persistence of GGHs in the liver after anti-viral therapy not only constitute the challenge to eliminate HBV infection but also carry the high risk to develop HCC. DNA chip and ELISA kit are designing to detect the pre-S mutants in serum. Novel or second generation anti-HBV drugs are under phase II development and include the combination of anti-virals, immunomodulators, agents for host DNA damage, and siRNA to target at the transcription of HBsAg gene in cccDNA or integrated HBV DNA. In the past years, we explored the possibility to provide drugs or natural agents targeting at ER stress signals in GGHs to prevent HCC development. In a transgenic mice model of pre-S mutant and HBx, a combination of silymarin and resveratrol targeting at mTOR and NF-κB signals could reduce a 80% of HCC development. In a pilot clinical trial, liposomal curcumin (Meriva®) combined with anti-virals and immumodulators P1101 have been designed and attempted to eliminate the serum surface antigen and hence the recurrence of HCC after surgical resection. Therefore, the detection of pre-S mutants in serum or GGHs in the liver should provide rational target design for therapy or prevention of HCC in these high risk patients of chronic HBV infection.


    A small intake of alcohol may be beneficial to health, but alcoholism, also known as alcohol dependence or alcohol abuse, is among the main healthy risky behavior due to the high relevance of negative health and social effect. Alcohol consumption has been identified as a major contributor to the global burden of chronic disease, injury and economic cost [1,2,3]. The World Health Organization reports the harmful use of alcohol causes approximately 3.3 million deaths every year (or 5.9% of all the global deaths), and 5.1% of the global burden of disease is attributable to alcohol consumption [4]. Furthermore, alcoholism is reported to be among the major concerns to public health in many countries, like UK [5,6]. Recently, the study of alcoholism model has become an important aspect of social epidemic. A great number of literatures can be found where healthy risky behavior, including drinking, smoking, drug use and obesity, are viewed as a treatable contagious disease, see [7,8,9,10] and the reference contained therein. In particular, several alcohol models described by ordinary differential equations or delay differential equations have been investigated extensively (see [11,12,13,14,15,16,17,18,19,20,21,22]). In [11], the authors modelled alcoholism as a contagious disease and studied how "infected" drinking buddies spread problem drinking. Manthey et al. [12] studied campus drinking and suggested that the basic reproductive numbers are not sufficient to predict whether drinking behavior will persist on campus and that the pattern of recruiting new members play a significant role in the reduction of campus alcohol problems. The impact of environmental factors and peer influences on the distribution of heavy drinking was studied in [13,16]. In addition, the two-stage models: one stage where people who admit to having a alcohol problem and other stage where people who do not admit to having a alcohol problem have been developed in [17,18]. Bhunu [19] studied the co-interaction of alcoholism and smoking in a community. Walters et al. [20] also discussed alcohol problems, and their results showed that an increase in the recovery rate decreased the proportion of binge drinkers in the population. Xiang et al. [21] considered the effect of constant immigration on drinking behavior. Wang et al. [22] presented a deterministic mathematical model for the spread of alcoholism with two control strategies to gain insights into this increasingly concerned about health and social phenomenon. The optimal control strategies are derived by proposing an objective functional and using Pontryagin's Maximum Principle. Zhu et al. [23] formulated a alcohol model with the impact of tax and investigated their dynamical behaviors. Other related drinking epidemic or population models, we refer to see [24,25,26,27,28,29,30,31,32,33].

    Public health education (e.g. Radio, Newspapers, Billboards, TV, and Internet, etc.) has been used to control the alcohol problems, which can not only influence the individuals' behavior but also increase the governmental health care involvement to control the spread of heavy drinking. These behavioral responses can change the transmission patterns and decline to drink. In recent years, many mathematical models [34,35,36,37] have been used for studying the impact of awareness programs by media on drinking problem. In [35], the authors studied drinking dynamics and focused on awareness programs and treatment in the modelling process. They extended the model in [34] via including a treatment class and established some sufficient conditions for the stability of the alcohol-free and alcohol-present equilibria. Xiang et al. [36] also studied a drinking model with public health education campaigns. Their results showed that awareness programs is an effective measure in reducing alcohol problems. These studies suggested that public health education and media had huge impact in controlling the spread of alcoholism. Recently, Ma et al. [37] proposed an alcohol consumption model with awareness programs and time delay described by including media function $ S\frac{M}{k_{0}+M} $, where $ M $ was the cumulative density of awareness programs driven by media. The results showed that the time delay in alcohol consumption habit which developed in susceptible population might result in a Hopf bifurcation.

    Actually, it has been found that heterogeneity (e.g., age, sex, space and so on) exists in many aspects of social epidemic transmission processes. Since multi-group models play important roles in considering the heterogeneity of host population, the study of the multi-group models can contribute to clarify the transmission pattern of infectious diseases in more realistic situations. There are lots of studies of the global stability of multi-group epidemic models, in which a general approach is used (see [38,39,40,41,42,43,44,45,46]). This graph-theoretic approach is sufficiently general to be applicable to a variety of coupled systems. On the other hand, since the pioneering work of Hoppensteadt established an age-dependent epidemic model in 1974 [47], the effects of the age factor on the multi-group epidemic models have already been studied by many researchers (see [48,49,50,51]). Dynamical behaviors of these multi-group epidemic models with age structure have been studied, respectively. However, to our knowledge, there are few studies on the alcoholism model in heterogeneous populations. In particular, we should notice that the impact of the age of alcoholism on other people is different. Thus, the study of the multi-group alcoholism model with public health education and age of alcoholism can contribute to the control of alcohol problems in more realistic situations.

    In this paper, motivated by the above works, we formulate a novel and more reasonable multi-group alcoholism model with public health education and alcoholism age to describe alcoholism spread in a heterogeneous host population. Inspired by the method developed in [38,39,40], we construct Lyapunov functionals and obtain the global stability of the alcohol-free equilibrium and alcohol-present equilibrium. Our results demonstrate that, for age structured multi-group alcoholism model, this graph-theoretic approach can be successfully applied by choosing an appropriate weighted matrix as well. At the same time, our main results indicate that public health education is beneficial for alcoholism control, and alcoholism age structure does not alter the dynamical behaviors.

    This paper is organized as follows. In next section, we formulate a more reasonable multi-group alcoholism model with public health education and alcoholism age. Some preliminary setting for the multi-group and age-dependent alcoholism model are presented in Section 3. In Section 4, we state our main results of this paper. We prove the global asymptotic stability of the alcohol-free equilibrium $ P_{0} $ for $ R_{0}\leq1 $ by using the theory of non-negative matrices and the classical method of Lyapunov functional. By applications of the graph-theoretic approach to the method of Lyapunov functionals, we prove the existence, uniqueness and global asymptotic stability of the alcohol-present equilibrium $ P^{*} $ for $ R_{0}>1 $. In Section 5, the effects of the public health education and alcoholism age are given. A brief discussion is also given in last section.

    Let $ n\in\mathbb{N} $ be the number of groups, the heterogeneous host population is divided into $ n $ homogeneous groups. For $ i $-th ($ 1\leq i\leq n $) group, it is further classified as five compartments: the uneducated susceptible individuals $ S_{i}(t) $ who do not drink or drink only moderately and do not accepted the public health education, but may one day develop light drinkers, the educated susceptible individuals $ E_{i}(t) $ who do not drink or consume alcohol in moderation and accepted the public health education, but may one day also develop light drinkers, the light drinkers $ L_{i}(t) $ who often consume alcohol but they don't influence other people, the alcoholics $ A_{i}(t) $ who have drinking problems or addictions (i.e. heavy drinkers) and they will influence susceptible individuals, and the recovered drinkers $ R_{i}(t) $ who are the recovered drinkers and permanently quit drinking. In $ i $-th ($ 1\leq i\leq n $) group, we assume that the susceptible individuals, the light drinkers and the recovered drinkers are homogeneous at time $ t $. The alcoholics individuals is structured by the age of alcoholism $ \theta $, and $ a_{i}(t, \theta) $ be the alcoholism age density of the individuals in the alcoholism individuals with alcoholism age $ \theta $ at time $ t $. Suppose that $ a_{i}(t, \theta) = 0 $ for all sufficiently large $ \theta>\theta^{+} $, where $ \theta^{+} $ is the maximum ages of alcoholism which is finite. Then $ A_{i}(t) = \int_{0}^{\infty}a_{i}(t, \theta)d\theta $ is the number of heavy drinkers at time $ t $. Furthermore, we make the following assumptions:

    (H1) In group $ i $, at any moment in time, new recruits enter the uneducated susceptible population at a rate $ \Lambda_{i}>0 $.

    (H2) In group $ i $, via public health education, the uneducated susceptible individuals in $ S_{i} $ class turn into the educated susceptible individuals $ E_{i} $ class at a constant rate $ \xi_{i}>0 $.

    (H3) In group $ i $, the coefficient of alcoholism transmission for susceptible individuals ($ S_{i} $ or $ E_{i} $) turn into light drinkers ($ L_{i} $) is $ \beta_{ij}\geq 0 $ for some reasons (e.g. through peer pressure), in which a susceptible individual contact with heavy drinkers $ A_{j} $ come from the $ j $-th group.

    (H4) Some educated individuals who never have a drink due to the effect of the public health educational campaigns, and the proportion $ \alpha_{i}>0 $ is that fraction of $ E_{i} $ who move into the $ R_{i} $ class.

    (H5) The heavy drinkers in $ A_{i} $ class can recover (due to counselling, health reasons, treatment, prohibition, or tax hiked on alcohol beverages, etc.) and will permanently quit drinking. The proportion $ \mu_{i}>0 $ is that fraction of $ A_{i} $ who move into the $ R_{i} $ class.

    (H6) The $ n $-square contact matrix $ B^{(n)} = (\beta_{ij})_{n\times n} $ is irreducible [41], where $ \beta_{ij} \geq 0 $.

    Remark 1: Assumption (H6) implies that every pair of groups is joined by an infectious path so that the presence of a heavy drinking individual in the first group can cause "infection" in the second group.

    The other parameters description of the model are presented in Table 1.

    Table 1.  The other parameters description of the alcoholism model.
    Variables Description
    $ p_{ij}(\theta) $ the infectivity between $ A_{j}(t) $ and $ S_{i}(t) $ at the age of alcoholism $ \theta $
    $ q_{ij}(\theta) $ the infectivity between $ A_{j}(t) $ and $ E_{i}(t) $ at the age of alcoholism $ \theta $
    $ \gamma_{i} $ the rate of the light drinkers enter into the alcoholism compartment in $ i $-th group
    $ d_{i}^{S} $ the natural death rate of uneducated susceptible individuals in $ i $-th group
    $ d_{i}^{E} $ the natural death rate of educated susceptible individuals in $ i $-th group
    $ d_{i}^{L} $ the natural death rate of light drinking individuals in $ i $-th group
    $ d_{i}^{A} $ the natural death rate of heavy drinking individuals in $ i $-th group
    $ d_{i}^{R} $ the natural death rate of recovered individuals in $ i $-th group
    $ m_{i}(\theta) $ the death rate of alcoholics due to excessive drinking in $ i $-th group

     | Show Table
    DownLoad: CSV

    Here, all the parameters are positive. Furthermore, the following assumptions are made.

    (H7) We assume that $ p_{ij}(\theta), q_{ij}(\theta), m_{i}(\theta)\in L_{+}^{1}(0, +\infty) $ are nonnegative and essential bounded, and they are all non-decreasing functions of the age of alcoholism $ \theta $. If $ \theta\geq \theta_{0} $, then $ p_{ij}(\theta)\equiv p_{max} $, $ q_{ij}(\theta)\equiv q_{max} $, and $ m_{i}(\theta)\equiv m_{max} $, where $ 0<\theta_{0}\leq \theta^{+} $.

    Because the effect of public health education, here the infectivity $ q_{ij}(\theta) $ is reduced relatively for educated individuals, i.e., $ q_{ij}(\theta) \leq p_{ij}(\theta) $ (the smaller, the better). For simplicity, we set $ P_{ij} = \int_{0}^{\infty}p_{ij}(\theta)a_{j}(t, \theta)d\theta $, $ Q_{ij} = \int_{0}^{\infty}q_{ij}(\theta)a_{j}(t, \theta)d\theta $, and $ [\mu_{i}A_{i}] $ stands for $ \int_{0}^{\infty}\mu_{i}a_{i}(t, \theta)d\theta. $ Under assumptions (H1)-(H6), the alcoholism transmission diagram is depicted in Figure 1.

    Figure 1.  The transfer diagram for our multi-group age-structured alcoholism model.

    The transfer diagram leads to the following multi-group age structured SELAR alcoholism model with public health education:

    $ {dSi(t)dt=Λinj=1βijSi(t)0pij(θ)aj(t,θ)dθ(dSi+ξi)Si(t),dEi(t)dt=ξiSi(t)nj=1βijEi(t)0qij(θ)aj(t,θ)dθ(dEi+αi)Ei(t),dLi(t)dt=nj=1βijSi(t)0pij(θ)aj(t,θ)dθ+nj=1βijEi(t)0qij(θ)aj(t,θ)dθ(dLi+γi)Li(t),ai(t,θ)t+ai(t,θ)θ=[dAi+mi(θ)+μi]ai(t,θ),dRi(t)dt=αiEi(t)+μi0ai(t,θ)dθdRiRi(t),i=1,2,...,n.
    $
    (2.1)

    The initial conditions and boundary conditions of model (2.1) are respectively given by:

    $ Si(0)=ϕi10,Ei(0)=ϕi20,Li(0)=ϕi30,Ri(0)=ϕi40,ai(0,θ)=a0i(θ)L1+(0,+),ai(t,0)=γiLi(t),i=1,2,...,n,
    $
    (2.2)

    where $ L_{+}^{1}(0, +\infty) $ is the space of functions that are nonnegative and Lebesgue integrable.

    Remark 2: Neglecting the heterogeneity and alcoholism age, Xiang et al. [36] constructed a SEARQ drinking model with public health educational campaigns, its threshold stability was achieved.

    In this section, some primary results are presented for establishing our main conclusions.

    Note that the variables $ R_{i}(t) $ does not appear in the first four equations of (2.1), it suffices to study the dynamical behaviors of the following reduced sub-system (3.1):

    $ {dSi(t)dt=Λinj=1βijSi(t)0pij(θ)aj(t,θ)dθ(dSi+ξi)Si(t),dEi(t)dt=ξiSi(t)nj=1βijEi(t)0qij(θ)aj(t,θ)dθ(dEi+αi)Ei(t),dLi(t)dt=nj=1βijSi(t)0pij(θ)aj(t,θ)dθ+nj=1βijEi(t)0qij(θ)aj(t,θ)dθ(dLi+γi)Li(t),ai(t,θ)t+ai(t,θ)θ=[dAi+mi(θ)+μi]ai(t,θ).
    $
    (3.1)

    Let

    $ Γ0i(θ)=e0[dAi+mi(s)+μi]ds,
    $
    (3.2)

    then $ \Gamma_{0i}(\theta) $ is the probability of an alcoholism individual in the $ i $-th group surviving to alcoholism-age $ \theta $. Using the approach introduced by Webb [53], integrating along the characteristic lines and incorporating the boundary conditions, we can obtain that

    $ ai(t,θ)={γiLi(tθ)Γ0i(θ),t>θ0,a0i(θt)Γ0i(θ)Γ0i(θt),θt0.
    $
    (3.3)

    Substituting (3.3) into the equations for $ S_{i}'(t), E_{i}'(t) $ and $ L_{i}'(t) $ of system (3.1), we have

    $ {dSi(t)dt=Λinj=1βijSi(t)t0pij(θ)γjLj(tθ)Γ0j(θ)dθ(dSi+ξi)Si(t)F1(t),dEi(t)dt=ξiSi(t)nj=1βijEi(t)t0qij(θ)γjLj(tθ)Γ0j(θ)dθ(dEi+αi)Ei(t)F2(t),dLi(t)dt=nj=1βijSi(t)t0pij(θ)γjLj(tθ)Γ0j(θ)dθ+nj=1βijEi(t)t0qij(θ)γjLj(tθ)Γ0j(θ)dθ(dLi+γi)Li(t)+F1(t)+F2(t),
    $
    (3.4)

    where

    $ F_{1}(t) = \sum\limits_{j = 1}^{n}\beta_{ij} S_{i}(t)\int_{t}^{\infty}p_{ij}(\theta)a_{j}^{0}(\theta-t)\frac{\Gamma_{0j}(\theta)}{\Gamma_{0j}(\theta-t)}d\theta, $
    $ F_{2}(t) = \sum\limits_{j = 1}^{n}\beta_{ij} E_{i}(t)\int_{t}^{\infty}q_{ij}(\theta)a_{j}^{0}(\theta-t)\frac{\Gamma_{0j}(\theta)}{\Gamma_{0j}(\theta-t)}d\theta. $

    It is easy to check that $ \lim_{t\longrightarrow \infty} F_{1}(t) = 0, \lim_{t\longrightarrow \infty} F_{2}(t) = 0 $. Furthermore, set

    $ h_{i}(\theta) = p_{ij}(\theta)\gamma_{i}\Gamma_{0i}(\theta), g_{i}(\theta) = q_{ij}(\theta)\gamma_{i}\Gamma_{0i}(\theta), i = 1, 2, \cdots, n. $

    Obviously, $ h_{i}(\theta)\geq g_{i}(\theta), i = 1, 2, \cdots, n $. Then, by application of the results in [54], any equilibrium of system (3.4) (if it exists), must be a solution of the following limiting system associated with (3.4):

    $ {dSi(t)dt=Λinj=1βijSi(t)0hj(θ)Lj(tθ)dθ(dSi+ξi)Si(t),dEi(t)dt=ξiSi(t)nj=1βijEi(t)0gj(θ)Lj(tθ)dθ(dEi+αi)Ei(t),dLi(t)dt=nj=1βijSi(t)0hj(θ)Lj(tθ)dθ+nj=1βijEi(t)0gj(θ)Lj(tθ)dθ(dLi+γi)Li(t).
    $
    (3.5)

    The behavior of system (3.1) is equivalent to the system (3.5). Once the solution of system (3.5) is determined, we can obtain $ a_{i}(t, \theta) $ from (3.3). So that the stability of the equilibrium of system (2.1) is the same as that of system (3.5). In this paper, we focus on the system (3.5).

    For system (3.5), we need the appropriate fading memory space of continuous functions as follows (see Atkinson et al. [52]):

    $ C_{\lambda_{i}} = \left\{\phi \in C((-\infty, 0], \mathbb{R}_{+}):sup_{s\leq0}\mid\phi(s)\mid e^{\lambda_{i}s} \lt +\infty \right\}, $

    which is a Banach space endowed with the norm $ \parallel\phi\parallel_{\lambda_{i}}: = sup_{s\leq0}\mid\phi(s)\mid e^{\lambda_{i}s}<+\infty $, where $ \mathbb{R}_{+} = [0, +\infty] $, $ \lambda_{i} $ is a positive constant. Let $ \phi_{it}(s)\in C_{\lambda_{i}} $ be such that $ \phi_{it}(s) = \phi_{i}(t+s), s\in(-\infty, 0]. $ Thus, we consider the system (3.5) in the phase space

    $ X = \prod\limits_{i = 1}^{n}(C_{\lambda_{i}}\times C_{\lambda_{i}}\times C_{\lambda_{i}}). $

    For system (3.5), the existence and uniqueness of the solution with the initial and boundary conditions (2.2) can be checked by the standard approaches in [53]. Furthermore, we clams that any solution of system (3.5) with nonnegative initial conditions remains nonnegative.

    We define a continuous solution semi-flow $ \Phi: \mathbb{R}_{+} \times X\longrightarrow X $ associated with (3.5):

    $ \Phi(t, x_{0}) = \Phi_{t}(x_{0}) = \big(S_{1}(t), E_{1}(t), L_{1}(t), \cdots, S_{n}(t), E_{n}(t), L_{n}(t)\big), x_{0} \in X, t \geq0. $

    Let

    $ Ω={(S1(),E1(),L1(),,Sn(),En(),Ln())X0Si(0)+Ei(0)+Li(0)Λiηi,Si(s),Ei(s),Li(s)0,s(,0],i=1,2,,n},
    $

    where $ \eta_{i} = \min\{d_{i}^{S}, d_{i}^{E}, d_{i}^{L}, d_{i}^{A}, d_{i}^{R}\} $. And denote the interior of $ \Omega $ as:

    $ Ωo={(S1(),E1(),L1(),,Sn(),En(),Ln())X0<Si(0)+Ei(0)+Li(0)<Λiηi,Si(s),Ei(s),Li(s)>0,s(,0],i=1,2,,n}.
    $

    Lemma 3.1. For system (3.5), $ \Omega $ is positively invariant for $ \Phi $, i.e., $ \Phi(t, x_{0})\in \Omega, \forall x_{0} \in \Omega, t \geq0 $.

    Proof. First, we prove that all solutions of system (3.5) with the initial and boundary conditions (2.2) remains nonnegative. By continuity of the solutions of system (3.5) and $ S_{i}(0) = \phi_{1}^{i}\geq0 $, we claim $ S_{i}(t)\geq0 $ for all $ t\geq0, i = 1, 2, \cdots, n. $ In fact, we assume that there exists a $ i_{1}\in\{1, 2, , \cdots, n\} $ such that $ S_{i_{1}}(t) $ lost its positivity for the first time at $ t_{1}>0 $, i.e., $ S_{i_{1}}(t_{1}) = 0 $. However, from the first equation of (3.5), we can see $ \frac{ {\rm d} S_{i_{1}}(t_{1})}{ {\rm d} t} = \Lambda_{i_{1}} > 0 $ which is a contradiction to the fact that $ S_{i_{1}}(t)>S_{i_{1}}(t_{1}) = 0 $ for any $ 0\leq t<t_{1} $.

    Similarly, we shall show that $ E_{i}(t)\geq0 $ for all $ t \geq0, i = 1, 2, \cdots, n. $ Otherwise, we assume that there exists a $ i_{2}\in\{1, 2, , \cdots, n\} $ and the first time $ t_{2}>0 $ such that $ E_{i_{2}}(t_{2}) = 0 $. However, from the second equation of (3.5) one obtains $ \frac{{\rm d}E_{i_{2}}(t_{2})}{{\rm d}t} = \xi_{i_{2}}S_{i_{2}}(t_{2}) > 0 $ which follows from the nonnegativity of $ S_{i}(t) $ for all $ t\geq0, i = 1, 2, \cdots, n. $ Thus, $ E_{i}(t)\geq0 $ for all $ t\geq0, i = 1, 2, \cdots, n. $

    Furthermore, we conclude that there is a $ i_{3}\in\{1, 2, , \cdots, n\} $ and the first time $ t_{3}>0 $ such that $ L_{i_{3}}(t_{3}) = 0 $. Note that $ 0<t_{3}-\theta<t_{3} $, thus, $ L_{i_{3}}(t_{3}-\theta)>0 $. Using the similar arguments and the third equation of (3.5), we gets $ \frac{{\rm d}L_{i_{3}}(t_{3})}{{\rm d}t} > 0. $ This is a contradiction which implies the nonnegativity of $ L_{i}(t) $ for all $ t\geq0, i = 1, 2, \cdots, n. $

    Second, we prove that all solutions of system (3.5) with the initial and boundary conditions (2.2) remains bounded. The total population size is $ N(t) = \sum_{i = 1}^{n}N_{i}(t) $. In the $ i $-th group, $ N_{i}(t) = S_{i}(t)+E_{i}(t)+L_{i}(t)+\int_{0}^{\infty}a_{i}(t, \theta)d\theta+R_{i}(t). $ Let $ \eta_{i} = \min\{d_{i}^{S}, d_{i}^{E}, d_{i}^{L}, d_{i}^{A}, d_{i}^{R}\} $, then

    $ \frac{{\rm d}N_{i}(t)}{{\rm d}t})\leq\Lambda_{i}-\eta_{i}N_{i}(t)-\int_{0}^{\infty}m_{i}(\theta)a_{i}(t, \theta)d\theta\leq\Lambda_{i}-\eta_{i}N_{i}(t), $

    this suggesting

    $ \limsup\limits_{t\longrightarrow \infty}N_{i}(t)\leq\frac{\Lambda_{i}}{\eta_{i}}. $

    Then, we have

    $ \limsup\limits_{t\longrightarrow \infty}(S_{i}(t)+E_{i}(t)+L_{i}(t))\leq\frac{\Lambda_{i}}{\eta_{i}}. $

    Therefore, we can obtain that $ \Omega $ is positively invariant for $ \Phi $, i.e., $ \Phi(t, x_{0})\in \Omega, \forall x_{0} \in \Omega, t \geq0 $. This proof is completed.

    Our results in this paper will be stated for system (3.5) in $ \Omega $, and can be translated straightforwardly to system (3.1). Moreover, all positive semi-orbits in $ \Omega $ have compact closure in $ X $ (see [52]), and thus have non-empty $ \omega $-limit sets. We have the following result.

    Lemma 3.2. All positive semi-orbits in $ \Omega $ have non-empty $ \omega $-limit sets.

    Lemma 3.3. System (3.5) is point dissipative, that is, there exists constants $ M_{i}>0 $ such that for each solution of (3.5) there is a $ T_{i}>0 $ such that $ \parallel S_{it}\parallel_{\lambda_{i}}\leq M_{i}, \parallel E_{it}\parallel_{\lambda_{i}}\leq M_{i} $ and $ \parallel L_{it}\parallel_{\lambda_{i}}\leq M_{i} $ for all $ t\geq T_{i}, i = 1, 2, \cdots, n $.

    Proof. Let $ W_{i} = \max_{t\in[0, T_{i}]}S_{i}(t)>0 $, then, for any $ t\geq T_{i} $ and any $ \varepsilon_{i}>0 $, we get

    $ \parallel S_{it}\parallel_{\lambda_{i}} = \sup\limits_{s\leq 0}S_{it}(s)e^{\lambda_{i}s} = \sup\limits_{u\leq t}S_{it}(u)e^{\lambda_{i}u}e^{-\lambda_{i}t}\leq\max\left\{e^{-\lambda_{i}t}\parallel S_{i0}\parallel_{\lambda_{i}}, W_{i}e^{\lambda_{i}T_{i}}e^{-\lambda_{i}t}, \frac{\Lambda_{i}}{\eta_{i}}+\varepsilon_{i}\right\}, $

    where the last estimation was obtained by three separations to $ u\leq 0, 0\leq u\leq T_{i} $ and $ T_{i}\leq u\leq t. $ Hence, we can choose $ M^{S}_{i}>0 $ such that $ \parallel S_{it}\parallel_{\lambda_{i}}\leq M^{S}_{i} $. Similary, we can also prove that $ \parallel E_{it}\parallel_{\lambda_{i}}\leq M^{E}_{i} $ and $ \parallel L_{it}\parallel_{\lambda_{i}}\leq M^{L}_{i} $. Set $ M_{i} = \max\{M^{S}_{i}, M^{E}_{i}, M^{L}_{i}\}>0 $, this proof is completed for all $ t\geq T_{i}, i = 1, 2, \cdots, n $.

    System (3.5) always has a alcohol-free equilibrium $ P_{0} = (S_{1}^{0}, E_{1}^{0}, 0, ..., S_{n}^{0}, E_{n}^{0}, 0)\in\mathbb{R}_{+}^{3n} $, where

    $ S_{i}^{0} = \frac{\Lambda_{i}}{d_{i}^{S}+\xi_{i}}, E_{i}^{0} = \frac{\xi_{i}S_{i}^{0}}{d_{i}^{E}+\alpha_{i}} = \frac{\xi_{i}\Lambda_{i}}{(d_{i}^{S}+\xi_{i})(d_{i}^{E}+\alpha_{i})}, i = 1, 2, ..., n. $

    Let

    $ B_{i} = \int_{0}^{\infty}h_{i}(\theta)d\theta \gt 0, C_{i} = \int_{0}^{\infty}g_{i}(\theta)d\theta \gt 0. $

    The alcohol-present equilibrium of system (3.5) is given by

    $ P^{*} = (S_{1}^{*}, E_{1}^{*}, L_{1}^{*}, ..., S_{n}^{*}, E_{n}^{*}, L_{n}^{*})\in\Omega^{o}, $

    then it is determined by the following system of equations

    $ {Λi=(dSi+ξi)Si+nj=1βijBjSiLj,ξiSi=(dEi+αi)Ei+nj=1βijCjEiLj,(dLi+γi)Li=nj=1βijBjSiLj+nj=1βijCjEiLj.
    $
    (3.6)

    The basic reproduction number $ R_{0} $ (see [38,55,56]) is defined as the spectral radius of a matrix $ M_{0} $, that is,

    $ R_{0} = \rho(M_{0}),~~~~~ M_{0} = (βijBjS0i+βijCjE0idLi+γi)
    _{n\times n}. $

    It acts as a threshold as is shown in the following result. We shall establish that the dynamical behaviors of system (3.5) are completely determined by values of $ R_{0} $.

    In this section we state and prove our main results concerning the global dynamics of system (3.5).

    Theorem 4.1. Assume that the matrix $ B^{(n)} = (\beta_{ij})_{n\times n} $ is irreducible.

    (1) If $ R_{0}\leq1 $, then the unique alcohol-free equilibrium $ P_{0} $ of system (3.5) is globally asymptotically stable in $ \Omega $;

    (2) If $ R_{0}>1 $, then $ P_{0} $ is unstable and system (3.5) is uniformly persistent, i.e., there exists a positive constant $ c> 0 $ such that $ \limsup_{t\longrightarrow \infty}S_{i}(t)\geq c, \limsup_{t\longrightarrow \infty}E_{i}(t)\geq c, \limsup_{t\longrightarrow \infty} L_{i}(t)\geq c, i = 1, 2, \ldots, n $. Furthermore, system (3.5) has at least one alcohol-present equilibrium $ P^{*} $ in $ \Omega^{o} $.

    Proof. (1) Let $ S = (S_{1}, S_{2}, ..., S_{n})^{T}, E = (E_{1}, E_{2}, ..., E_{n})^{T}, $ we define a matrix-valued function

    $ M(S, E) = :(βij(SiBj+CjEi)dLi+γi)
    _{n\times n}. $

    It is easy to see that $ M(S^{0}, E^{0}) = M_{0} $, where $ S^{0} = (S_{1}^{0}, S_{2}^{0}, ..., S_{n}^{0})^{T}, $ $ E^{0} = (E_{1}^{0}, E_{2}^{0}, ..., E_{n}^{0})^{T} $. We also have

    $ 0\leq M(S, E)\leq M(S^{0}, E^{0}) = M_{0}, 0\leq S_{i} \leq S_{i}^{0}, 1\leq i \leq n. $

    First, we will prove the uniqueness of alcohol-free equilibrium in $ \Omega $. If $ R_{0}\leq1 $, $ (S, E)\neq (S^{0}, E^{0}) $, then we obtain

    $ 0\leq M(S, E) \lt M_{0}. $

    Since $ B^{(n)} = (\beta_{ij})_{n\times n} $ is irreducible, we know nonnegative matrix $ M(S, E) $ and $ M_{0} $ are also irreducible. Using the Perron-Frobenius theorem (see Theorem 2.1.4 or Corollary 2.1.5 in [58]), we get

    $ \rho(M(S, E)) \lt \rho(M_{0})\leq1, (S, E)\neq (S^{0}, E^{0}). $

    Therefore, $ \rho(M(S, E)) < 1 $ holds when $ R_{0} = \rho(M_{0})\leq 1 $ and $ (S, E)\neq (S^{0}, E^{0}) $. This implies that the vector equation $ M(S, E)L = L $ has only the trivial solution $ L = 0 $, where $ L = (L_{1}, L_{2}, ..., L_{n})^{T} $. Thus $ P_{0} $ is the unique equilibrium of system (3.5) in $ \Omega $ if $ R_{0}\leq1 $.

    Next, we claim that the alcohol-free equilibrium $ P_{0} $ is globally asymptotically stable in $ \Omega $. Since $ M_{0} $ is irreducible, there exists a positive left eigenvector $ \omega $ of $ M_{0} $ corresponding to $ \rho(M_{0}) $, i.e.,

    $ (\omega_{1}, \omega_{2}, \cdots, \omega_{n})M_{0} = (\omega_{1}, \omega_{2}, \cdots, \omega_{n})\rho(M_{0}), $

    where $ \omega = (\omega_{1}, \omega_{2}, ..., \omega_{n}) $ and $ \omega_{i} > 0, i = 1, 2, ..., n $. We introduce the Volterra-type function $ \varphi(x) = x-1-\ln x, x>0, $ which is positive definite and attains its global minimum $ \varphi(1) = 0 $ at $ x = 1 $. Note that $ \omega $ is a strictly positive left eigenvector. We define a Lyapunov functional $ L_{afe}(t):(S_{i}, E_{i}, L_{i})\rightarrow\mathbb{R} $ as

    $ L_{afe}(t) = H_{1}(t)+H_{2}(t)+H_{3}(t), $

    which is nonnegative and continuously differentiable. Here

    $ H1(t)=ni=1ci{S0iφ(SiS0i)+E0iφ(EiE0i)+Li},H2(t)=ni=1ci{nj=1βijS0i0mj(θ)Lj(tθ)dθ},H3(t)=ni=1ci{nj=1βijE0i0nj(θ)Lj(tθ)dθ},
    $

    where $ c_{i} = \frac{\omega_{i}}{d_{i}^{L}+\gamma_{i}} $, $ m_{j}(\theta) = \int_{\theta}^{\infty}h_{j}(s)ds $ and $ n_{j}(\theta) = \int_{\theta}^{\infty}g_{j}(s)ds $. Obviously, $ m_{j}(0) = \int_{0}^{\infty}h_{j}(s)ds = B_{j}, \frac{d m_{j}(\theta)}{d\theta} = -h_{j}(\theta) $, and $ n_{j}(0) = \int_{0}^{\infty}g_{j}(s)ds = C_{j}, \frac{d n_{j}(\theta)}{d\theta} = -g_{j}(\theta) $.

    It follows from Lemmas 3.1 and 3.3 that $ L_{afe}(t) $ is nonnegative and bounded for big enough positive $ t>0 $. It is clear that $ L_{afe}(t) $ with the equality holds if and only if $ S_{i} = S_{i}^{0}, E_{i} = E_{i}^{0}, L_{i} = 0 $. First, calculating the time derivative of $ H_{1}(t) $ along with the solutions of system (3.5), using the equation $ \Lambda_{i} = (d_{i}^{S}+\xi_{i})S_{i}^{0} $, we have

    $ H1(t)=ni=1ci{(1S0iSi)dSi(t)dt+(1E0iEi)dEi(t)dt+dLi(t)dt}=ni=1ci{(dSi+ξi)(S0iSi)(dSi+ξi)S0iSi(S0iSi)+nj=1βijS0i0hj(θ)Lj(tθ)dθ+ξiSi(t)(dEi+αi)EiξiSiE0iEi+(dEi+αi)E0i+nj=1βijE0i0gj(θ)Lj(tθ)dθ(dLi+γi)Li}.
    $
    (4.1)

    Second, calculating the time derivative of $ H_{2}(t) $ along with the solutions of system (3.5). Let $ t-\theta = s $, then $ -\infty<s<t $, we obtain

    $ H2(t)={ni=1nj=1ciβijS0itmj(ts)Lj(s)ds}=ni=1nj=1ciβijS0imj(0)Lj(t)+ni=1nj=1ciβijS0itmj(ts)dtEj(s)ds=ni=1nj=1ciβijS0iBjLj(t)ni=1nj=1ciβijS0i0hj(θ)Lj(tθ)dθ.
    $
    (4.2)

    Similar to $ H^{'}_{2}(t) $, we get derivative of $ H_{3}(t) $ along with the solutions of system (3.5). Let $ t-\theta = s $, then $ -\infty<s<t $, we obtain

    $ H3(t)=ni=1nj=1ciβijE0iCjLj(t)ni=1nj=1ciβijE0i0gj(θ)Lj(tθ)dθ.
    $
    (4.3)

    Using (4.1)-(4.3), we have

    $ dLafe(t)dt=ni=1ci{(dSi+ξi)(S0iSi)(dSi+ξi)S0iSi(S0iSi)+ξiSi(t)(dEi+αi)EiξiSiE0iEi+(dEi+αi)E0i(dLi+γi)Li+ni=jβijLj(S0iBj+E0iCj)}=ni=1ci{(dSi+ξi)Si(S0iSi)2+ξiS0i(2E0iEiEiE0i)}+ni=1(nj=1ωiβij(S0iBj+E0iCj)LjdLi+γiωiLi)ni=1(nj=1ωiβij(S0iBj+E0iCj)LjdLi+γiωiLi)=(ω1,ω2,,ωn)(M0LL)=(ρ(M01)(ω1,ω2,,ωn)L=(R01)(ω1,ω2,,ωn)L0,ifR01.
    $
    (4.4)

    If $ R_{0} = \rho(M_{0}) < 1 $, then (4.4) implies $ L = 0 $, and $ \frac{d L_{afe}(t)}{dt} = 0 $ if and only if $ L = 0 $.

    If $ R_{0} = \rho(M_{0}) = 1 $, from (4.4), we see $ \frac{d L_{afe}(t)}{dt} = 0 $ implies $ (\omega_{1}, \omega_{2}, \ldots, \omega_{n})\cdot(M_{0}L-L) = 0, $ that is

    $ (ω1,ω2,,ωn)(M0In)L=0,
    $
    (4.5)

    where $ I_{n} $ denote the $ n $-dimensional identity matrix. Hence, if $ (S, E)\neq(S^{0}, E^{0}) $, then we have

    $ (\omega_{1}, \ldots, \omega_{n})M_{0} = (\omega_{1}, \ldots , \omega_{n})\rho(M_{0}) = (\omega_{1}, \ldots , \omega_{n}), $

    so the solution of (4.5) is only the trivial value $ L = 0 $. Summarizing the above statements, we see that $ \frac{d L_{afe}(t)}{dt} = 0 $ implies that either $ L = 0 $ or $ R_{0} = 1 $ and $ (S, E) = (S^{0}, E^{0}) $. It can be verified that the largest compact invariant subset of $ \left\{(S_{1}, E_{1}, L_{1}, ... , S_{n}, E_{n}, L_{n})\in\Omega: \frac{d L_{afe}(t)}{dt} = 0\right\} $ is the singleton $ \left\{P_{0}\right\} $. Therefore, by Lemma 3.2 and the Lyapunov-LaSalle invariance principle (see Theorem 3.4.7 in [59]), we can see that $ P_{0} $ is globally attractive in $ \Omega $ if $ R_{0}\leq1 $. The local asymptotical stability of the disease-free equilibrium $ P_{0} $ comes from the relationship between the eigenvalues of the linearized matrix and $ R_{0} $, which can be proved using the same proof as one for Corollary 5.3.1 in [60] (The detailed process is omitted here). Then, together with the local asymptotical stability of $ P_{0} $, we know that the alcohol-free equilibrium $ P_{0} $ is globally asymptotically stable in $ \Omega $ when $ R_{0}\leq1 $.

    (2) Suppose that $ R_{0} = \rho(M_{0})>1 $, and $ L\neq0 $. By continuity and (4.4), we have

    $ (\omega_{1}, \omega_{2}, \ldots, \omega_{n})(M_{0}-I_{n})L = (\rho(M_{0})-1)(\omega_{1}, \omega_{2}, \ldots, \omega_{n})L \gt 0, $

    which implies that $ \frac{d L_{afe}(t)}{dt}>0 $ in a sufficiently small neighborhood of the alcohol-free equilibrium $ P_{0} $ in $ \Omega^{0} $. This implies that the alcohol-free equilibrium $ P_{0} $ is unstable when $ R_{0}>1 $.

    Next, with a uniform persistence result in [61] and a similar argument as in the proof of Proposition 3.3 of [62], we can show that the instability of $ P_{0} $ implies that system (3.5) is uniformly persistent when $ R_{0}>1 $, i.e., there exists a positive constant $ c> 0 $ such that

    $ \limsup\limits_{t\longrightarrow \infty}S_{i}(t)\geq c, \limsup\limits_{t\longrightarrow \infty}E_{i}(t)\geq c, \limsup\limits_{t\longrightarrow \infty} L_{i}(t)\geq c, i = 1, 2, ..., n, $

    provided $ \left(S_{1}(0), E_{1}(0), L_{1}(0), ..., S_{n}(0), E_{n}(0), L_{n}(0)\right)\in \Omega^{0} $.

    Furthermore, using the uniform persistence of system (3.5), together with the uniform boundedness of solutions, we shall establish the existence of the alcohol-present equilibrium $ P^{*} $ (see Theorem 2.8.6 in [63] or Theorem D.3 in [64]). This completes the proof of theorem 4.1.

    In following section, we devote to the alcohol-present equilibrium of system (3.5) is globally asymptotically stable. In order to prove our results, we will construct the proper Lyapunov functional and apply subtle grouping technique in estimating the derivatives of Lyapunov functional guided by graph theory, which was recently developed by Guo et al. in [38,39] and Li et al. [40]. Here, we quote some results from graph theory which will be used in the proof of our main results. We refer the reader to [38,39,40] and the references cited therein for more details of these concepts and results.

    Given a non-negative matrix $ A = (a_{ij})_{n\times n} $, the $ directed ~ graph $ $ G(A) $ associated with $ A $ has vertices {1, 2, ..., n} with a directed arc $ (i, j) $ from $ i $ to $ j $ if and only if $ a_{ij}\neq0 $. It is $ strongly ~ connected $ if any two distinct vertices are joined by an oriented path. Matrix $ A $ is irreducible if and only if $ G(A) $ is strongly connected.

    A tree is a connected graph with no cycles. A subtree $ T $ of a graph $ G $ is said to be $ spanning $ if $ T $ contains all the vertices of $ G $. A $ directed tree $ is a tree in which each edge has been replaced by an arc directed one way or the other. A directed tree is said to be $ rooted $ at a vertex, called the root, if every arc is oriented in the direction towards to the root. An $ oriented cycle $ in a directed graph is a simple closed oriented path. A $ unicyclic graph $ is a directed graph consisting of a collection of disjoint rooted directed trees whose root are on an oriented cycle.

    For a given nonnegative matrix $ (\bar{\beta}_{ij})_{n\times n} $, $ \bar{\beta}_{ij}\geq0, 1\leq i, j\leq n $. Let

    $ ˉB=(j1ˉβ1jˉβ21ˉβn1ˉβ12j2ˉβ2jˉβn2ˉβ1nˉβ2njnˉβnj)
    $
    (4.6)

    be the Laplacian matrix of the directed graph $ (\bar{\beta}_{ij})_{n\times n} $, and $ C_{ij} $ denote the cofactor of the $ (i, j) $ entry of $ \bar{B} $. For the linear system

    $ ˉBv=0,
    $
    (4.7)

    the following results hold (see Lemma 2.1 in [38] and Theorem 2.2 in [40]).

    Lemma 4.1 ([38]) Assume that $ (\bar{\beta}_{ij})_{n\times n} $ is irreducible and $ n\geq2 $. Then following results hold:

    (1) The solution space of system (4.7) has dimension 1;

    (2) A basis of the solution space is given by

    $ (v_{1}, v_{2}, ..., v_{n}) = (C_{11}, C_{22}, ..., C_{nn}), $

    where $ C_{ii} $ denotes the cofactor of the $ k $-th diagonal entry of $ \bar{B} $, $ 1\leq i\leq n $;

    (3) For all $ 1\leq i\leq n, $

    $ C_{ii} = \sum\limits_{T\in\mathbb{T}_{i}}w(T) = \sum\limits_{T\in\mathbb{T}_{i}}\prod\limits_{(r, m)\in E(T)}\bar{\beta}_{rm}, $

    where $ \mathbb{T}_{i} $ is the set of all directed spanning subtrees of $ G(\bar{B}) $ that are rooted at the $ i-th $ vertex, $ w(T) $ is the weight of a directed tree $ T $, and $ E(T) $ denotes the set of directed arcs in a directed tree $ T $;

    (4) For all $ 1\leq i\leq n $,

    $ C_{ii} \gt 0. $

    Lemma 4.2. ([40]) Assume that $ v_{i} $ is the same meaning in Lemma 4.1, and $ n\geq2 $, then

    $ \sum\limits_{i, j = 1}^{n}v_{i}\bar{\beta}_{ij}F_{ij}(x_{i}, y_{j}) = \sum\limits_{T\in\mathbb{T}}w(T)\sum\limits_{(r, m)\in E(CT)}F_{rm}(x_{r}, y_{m}), $

    where $ F_{ij}(x_{i}, y_{j}), 1\leq i, j\leq n $ is an arbitrary function, $ \mathbb{T} $ is the set of all spanning unicyclic graphs of $ G(\bar{B}) $, $ w(T) $ is the weight of a directed tree $ T $, $ CT $ denotes the oriented cycle in a unicyclic graph $ T $, and $ E(CT) $ denotes the set of directed arcs in $ CT $.

    Let $ P^{*} = (S_{1}^{*}, E_{1}^{*}, L_{1}^{*}, ..., S_{n}^{*}, E_{n}^{*}, L_{n}^{*}) $ be the alcohol-present equilibrium of system (3.5), then $ S_{i}^{*}, E_{i}^{*}, L_{i}^{*} ( i = 1, 2, ..., n) $ are determined by the following equations

    $ {Λi=(dSi+ξi)Si+nj=1βijBjSiLj,ξiSi=(dEi+αi)Ei+nj=1βijCjEiLj,(dLi+γi)Li=nj=1βijBjSiLj+nj=1βijCjEiLj.
    $
    (4.8)

    In the following section, we prove that the alcohol-present equilibrium $ P^{*} $ is globally asymptotically stable when $ R_{0}>1 $. In particular, this proof implies that the endemic equilibrium is unique in the region $ \Omega^{0} $ when it exists. Therefore, we have the following main result on the uniqueness and global stability of the positive equilibrium $ P^{*} $ when $ R_{0}>1 $.

    Theorem 4.2. Assume that $ B^{(n)} = (\beta_{ij})_{n\times n} $ is irreducible. If $ R_{0}>1 $, then the alcohol-present equilibrium $ P^{*} $ of system (3.5) is globally asymptotically stable in $ \Omega^{0} $ and thus is the unique positive equilibrium.

    Proof. Case Ⅰ: $ n = 1 $

    The multi-group system (3.5) reduces to a single-group system as follows

    $ {dS(t)dt=ΛβS(t)0h(θ)L(tθ)dθ(dS+ξ)S(t),dE(t)dt=ξS(t)βE(t)0g(θ)L(tθ)dθ(dE+α)E(t),dL(t)dt=βS(t)0h(θ)L(tθ)dθ+βE(t)0g(θ)L(tθ)dθ(dL+γ)L(t),
    $
    (4.9)

    where $ h(\theta) = \gamma p(\theta)\Gamma_{0}(\theta), \int_{0}^{\infty}h(\theta)d\theta = B $ and $ g(\theta) = \gamma q(\theta)\Gamma_{0}(\theta), \int_{0}^{\infty}g(\theta)d\theta = C $. The alcohol-present equilibrium $ P^{*} $ of system (4.9) satisfies the following equations

    $ {Λ=(dS+ξ)S+βSLB,ξS=(dE+α)E+βELC,(dL+γ)L=βSLB+βELC.
    $
    (4.10)

    Let $ (S(t), E(t), L(t)) $ be any solution of system (4.9) with non-negative initial data, we construct a Lyapunov functional $ V_{ape}(t):(S, E, L)\rightarrow\mathbb{R} $ as follows

    $ V_{ape}(t) = V_{1}(t)+V_{2}(t)+V_{3}(t), $

    which is nonnegative and continuously differentiable. Here we define

    $ V1(t)=Sφ(SS)+S+Eφ(EE)+Lφ(LL),V2(t)=βS0m(θ)Lφ(L(tθ)L)dθ,V3(t)=βE0n(θ)Lφ(L(tθ)L)dθ,
    $

    where $ \varphi(x) = x-1-\ln x $, $ m(\theta) = \int_{\theta}^{\infty}h(s)ds, m(0) = B, \frac{d m(\theta)}{d\theta} = -h(\theta), $ and $ n(\theta) = \int_{\theta}^{\infty}g(s)ds, n(0) = C, \frac{d n(\theta)}{d\theta} = -g(\theta). $ It is clear that $ V_{ape}(t) $ is bounded for all $ t\geq0 $, and $ V_{ape}(t)\geq 0 $ with the equality holds if and only if $ S(t) = S^{*}, E(t) = E^{*}, L(t) = L(t-\theta) = L^{*} $.

    Using the equations in (4.9) and $ \Lambda = (d^{S}+\xi)S^{*}+\beta S^{*}L^{*}B $, and differentiating $ V_{1}(t) $ along system (4.8), we have

    $ V1(t)=(1SS)dS(t)dt+dS(t)dt+(1EE)dE(t)dt+(1LL)dL(t)dt=Λ(1SS)+(dS+ξ)(2SS)+βSLB+βS0h(θ)L(tθ)dθβS0h(θ)L(tθ)dθ(dE+α)EξSEEdSS+(dE+α)E+βE0g(θ)L(tθ)dθβE0g(θ)L(tθ)dθ(dL+γ)LLLβS0h(θ)L(tθ)dθLLβE0g(θ)L(tθ)dθ+(dL+γ)L=dS+ξS(SS)2βBSLSS+βS0h(θ)L(tθ)dθβS0h(θ)L(tθ)dθξSEEdSS(dE+α)(EE)+βE0g(θ)L(tθ)dθβE0g(θ)L(tθ)dθ(dL+γ)(LL)LLβS0h(θ)L(tθ)dθLLβE0g(θ)L(tθ)dθ.
    $
    (4.11)

    Further, let $ s = t-\theta $ and base on fact that $ \frac{d m(\theta)}{d\theta} = -h(\theta) $, we obtain

    $ V2(t)=βSddt[tm(ts)Lφ(L(t)L)ds]=βS[m(0)Lφ(L(t)L)+tdm(ts)dtLφ(L(t)L)ds]=βS[0h(θ)Lφ(L(t)L)dθ+0dm(θ)dθLφ(L(tθ)L)dθ]=βSL0h(θ)[L(t)LL(tθ)L+lnL(tθ)L(t)]dθ=βS0h(θ)L(t)dθβS0h(θ)L(tθ)dθ+βSL0h(θ)lnL(tθ)L(t)dθ.
    $
    (4.12)

    Similarly, base on fact that $ \frac{d n(\theta)}{d\theta} = -g(\theta) $, we have

    $ V3(t)=βEL0g(θ)[L(t)LL(tθ)L+lnL(tθ)L(t)]dθ=βE0g(θ)L(t)dθβE0g(θ)L(tθ)dθ+βEL0g(θ)lnL(tθ)L(t)dθ.
    $
    (4.13)

    Combining (4.10)-(4.13), we get

    $ dVape(t)dt=dS+ξS(SS)2βBSLSS+βS0h(θ)L(tθ)dθβS0h(θ)L(tθ)dθξSEEdSS(dE+α)(EE)+βE0g(θ)L(tθ)dθβE0g(θ)L(tθ)dθ(dL+γ)(LL)LLβS0h(θ)L(tθ)dθLLβE0g(θ)L(tθ)dθ+βSL0h(θ)[LLL(tθ)L+lnL(tθ)L]dθ+βEL0g(θ)[LLL(tθ)L+lnL(tθ)L]dθ=dS+ξS(SS)2βS0h(θ)L(tθ)dθξSEEdSS(dE+α)EβE0g(θ)L(tθ)dθ(dL+γ)LLLβS0h(θ)L(tθ)dθLLβE0g(θ)L(tθ)dθβSL0h(θ)[φ(SS)+φ(LL)+φ(SL(tθ)SL)]dθβEL0g(θ)[φ(LL)+φ(L(tθ)L)]dθ.
    $
    (4.14)

    From (4.14), we conclude that $ \frac{d V_{ape}(t)}{dt}\leq 0 $ and with the equality holds if and only if $ S(t) = S^{*}, E(t) = E^{*}, L(t) = L(t-\theta) = L^{*} $ for all $ t\geq0 $. Thus, the largest invariant set $ \{\frac{d V_{ape}(t)}{dt} = 0\} = \{P^{*}\} $. Therefore, using the LaSalle's Invariance Principle, we get the alcohol-present equilibrium $ P^{*} $ is globally attractive in $ \Omega^{0} $ if $ R_{0}>1 $ for $ n = 1 $.

    Case Ⅱ: $ n\geq2 $

    In the following, we are going to consider the case $ n\geq2 $. Let

    $ \bar{\beta}_{ij} = \beta_{ij}L_{j}^{*}(S_{i}^{*}B_{j}+E_{i}^{*}C_{j}), 1\leq i, j\leq n, n\geq2, $

    and matrix $ \bar{B} $ as given in (4.6). Since $ B^{(n)} = (\beta_{ij})_{n\times n} $ is irreducible, we know the matrix $ \bar{B} $ is also irreducible. Let $ v = \{v_{1}, \ldots, v_{n}\}, v_{ i}> 0 $ be a basis for the solution space of linear system (4.7), i.e., $ \bar{B}v = 0 $ as described in Lemma 4.1.

    Let $ (S_{i}(t), E_{i}(t), L_{i}(t)) ( 1\leq i\leq n) $ be any solution of system (3.5) with non-negative initial data. For such $ v = \{v_{1}, \ldots, v_{n}\} $, we define a Volterra-type Lyapunov functional $ U_{ape}(t):(S_{i}, E_{i}, L_{i})\rightarrow\mathbb{R} $ as follows

    $ U_{ape}(t) = U_{1}(t)+U_{2}(t)+U_{3}(t), $

    which is nonnegative and continuously differentiable, where

    $ U1(t)=ni=1vi{Siφ(SiSi)+Eiφ(EiEi)+Liφ(LiLi)},U2(t)=ni=1vi{nj=1βijSi0mj(θ)Ljφ(Lj(tθ)Lj)dθ},U3(t)=ni=1vi{nj=1βijEi0nj(θ)Ljφ(Lj(tθ)Lj)dθ},
    $

    where $ m_{j}(\theta) = \int_{\theta}^{\infty}h_{j}(s)ds, m_{j}(0) = B_{j}, \frac{d m_{j}(\theta)}{d\theta} = -h_{j}(\theta), $ and $ n_{j}(\theta) = \int_{\theta}^{\infty}g_{j}(s)ds, n_{j}(0) = C_{j}, \frac{d n_{j}(\theta)}{d\theta} = -g_{j}(\theta). $

    By the definition of the fading memory space, Lemmas 3.2 and 3.3, we known that $ U_{ape}(t) $ is well-defined, that is, $ U_{ape}(t) $ is bounded for all $ t\geq0 $. It is clear that $ U_{ape}(t)\geq0 $ with the equality holds if and only if $ S_{i}(t) = S_{i}^{*}, E_{i}(t) = E_{i}^{*}, L_{i}(t-\theta) = L_{i}(t) = L_{i}^{*}. $ Differentiating $ U_{1}(t), U_{2}(t), U_{3}(t) $ along the solutions of system (3.5), and using the equilibrium equations (4.8), we get

    $ U1(t)=ni=1vi{(1SiSi)dSi(t)dt+(1EiEi)dEi(t)dt+(1LiLi)dLi(t)dt}=ni=1vi{dSi+ξiSi(SiSi)2nj=1βijBjSiLj(SiSi1)+nj=1βijSi0hj(θ)Lj(tθ)dθξiSi(EiEi1)(dEi+αi)(EiEi)nj=1βijEi0gj(θ)Lj(tθ)dθ+nj=1βijEi0gj(θ)Lj(tθ)dθ(dLii+γi)(LiLi)nj=1LjLjβijSi0hj(θ)Lj(tθ)dθnj=1LjLjβijEi0gj(θ)Lj(tθ)dθ}.
    $
    (4.15)

    Further, let $ s = t-\theta $ and base on fact that $ \frac{d m_{j}(\theta)}{d\theta} = -h_{j}(\theta) $, we obtain

    $ U2(t)=ni=1nj=1viβijSiLj0hj(θ)(Lj(t)LjLj(tθ)Lj+lnLj(tθ)Lj(t))dθ=ni=1nj=1viβijSi{0hj(θ)Lj(t)dθ0hj(θ)Lj(tθ)dθ+Lj0hj(θ)lnLj(tθ)Lj(t)dθ}.
    $
    (4.16)

    Similarly, base on fact that $ \frac{d n_{j}(\theta)}{d\theta} = -g_{j}(\theta) $, we have

    $ U3(t)=ni=1nj=1viβijEiLj0gj(θ)(Lj(t)LjLj(tθ)Lj+lnLj(tθ)Lj(t))dθ=ni=1nj=1viβijEi{0gj(θ)Lj(t)dθ0gj(θ)Lj(tθ)dθ+Lj0gj(θ)lnLj(tθ)Lj(t)dθ}.
    $
    (4.17)

    Combining (4.15)-(4.17), we get

    $ dUape(t)dt=3k=1Uk(t)=ni=1vi{dSi+ξiSi(SiSi)2nj=1βijBjSiLj(SiSi1)+nj=1βijSi0hj(θ)Lj(tθ)dθξiSi(EiEi1)(dEi+αi)(EiEi)nj=1βijEi0gj(θ)Lj(tθ)dθ+nj=1βijEi0gj(θ)Lj(tθ)dθ(dLii+γi)(LiLi)nj=1LjLjβijSi0hj(θ)Lj(tθ)dθnj=1LjLjβijEi0gj(θ)Lj(tθ)dθ}+ni=1nj=1viβijSi{0hj(θ)Lj(t)dθ0hj(θ)Lj(tθ)dθ+Lj0hj(θ)lnLj(tθ)Lj(t)dθ}+ni=1nj=1viβijEi{0gj(θ)Lj(t)dθ0gj(θ)Lj(tθ)dθ+Lj0gj(θ)lnLj(tθ)Lj(t)dθ}.
    $
    (4.18)

    Using $ \varphi(x) = x-1-\ln x $ and $ \bar{\beta}_{ij} = \beta_{ij}L_{j}^{*}(S_{i}^{*}B_{j}+E_{i}^{*}C_{j}) $, we rearrange the terms in (4.18) as follows

    $ dUape(t)dt=ni=1vi{dSi+ξiSi(SiSi)2nj=1βijBjSiLjSiSiξiSiEiEi(dEi+αi)Ei(dLii+γi)Linj=1βijSiLj0hj(θ)[φ(SiSi)+φ(SiLj(tθ)SiLj)]dθnj=1βijEiLj0gj(θ)[φ(LjLj)+φ(LiLj(tθ)LiLj)]dθ}+ni=1nj=1viˉβij(LjLjLiLi)ni=1nj=1viˉβijlnLiLjLiLj.
    $
    (4.19)

    Now, we are going to show that $ \frac{d U_{ape}(t)}{dt}\leq 0 $. From the properties of function $ \varphi(x) $, it is easy to see that we only need to consider the last two items in (4.19). In what follows, we first prove that the penultimate item

    $ ni=1nj=1viˉβij(LjLjLiLi)0.
    $
    (4.20)

    In fact, from (4.7) $ \bar{B}v = 0 $, we get

    $ v_{i}\sum\limits_{j = 1}^{n}\bar{\beta}_{ij} = \sum\limits_{j = 1}^{n}\bar{\beta}_{ji}v_{j}, i = 1, \ldots, n. $

    Note that $ \bar{\beta}_{ij} = \beta_{ij}L_{j}^{*}(S_{i}^{*}B_{j}+E_{i}^{*}C_{j}) $, we obtain

    $ v_{i}\sum\limits_{j = 1}^{n}\beta_{ij}L_{j}^{*}(S_{i}^{*}B_{j}+E_{i}^{*}C_{j}) = \sum\limits_{j = 1}^{n}\beta_{ji}L_{i}^{*}(S_{j}^{*}B_{i}+E_{j}^{*}C_{i})v_{j}, i = 1, \ldots, n. $

    Using this equation, we can obtain

    $ ni=1nj=1viˉβijLjLj=ni=1nj=1viβij(SiBj+EiCj)Lj=ni=1Linj=1vjβji(SjBi+EjCi)=ni=1LiLinj=1vjβjiLi(SjBi+EjCi)=ni=1nj=1vjβjiLi(SjBi+EjCi)LiLi=ni=1nj=1viβijLj(SiBj+EiCj)LiLi=ni=1nj=1viˉβijLiLi.
    $
    (4.21)

    Therefore, we have

    $ \sum\limits_{i = 1}^{n}\sum\limits_{j = 1}^{n}v_{i}\bar{\beta}_{ij}\left(\frac{L_{j}}{L_{j}^{*}}-\frac{L_{i}}{L_{i}^{*}}\right)\equiv 0 $

    holds for all $ L_{1}, L_{2}, \cdots, L_{n}>0 $.

    Next, we will show that the last item is also equal to zero. Let

    $ W_{n} = \sum\limits_{i = 1}^{n}\sum\limits_{j = 1}^{n}v_{i}\bar{\beta}_{ij}\ln\frac{L_{i}^{*}L_{j}}{L_{i}L_{j}^{*}}. $

    Then, we will find that

    $ Wn0
    $
    (4.22)

    holds for all $ L_{1}, L_{2}, \cdots, L_{n}>0 $.

    In fact, by the Kirchhoff's Matrix-Tree Theorem in Lemmas 4.1 and 4.2 (see [38]-[40]), we known that $ v_{i} = C_{ii} $ is a sum of weights of all directed spanning subtrees $ T $ of $ G $ that are rooted at vertex $ i $. So, each term $ v_{i}\bar{\beta}_{ij} $ is the weight $ w(Q) $ of a unicyclic subetaaph $ Q $ of $ G $, obtained from such a tree $ T $ by adding a directed arc $ (i, j) $ from the root $ i $ to vertex $ j $. Thus, the meaning of double sum in (4.22) can be considered as a sum over all unicyclic subetaaphs $ Q $ containing vertices $ \{1, 2, \ldots, n\} $, that is

    $ W_{n} = \sum\limits_{Q}W_{n, Q} = \sum\limits_{Q}w(Q)\cdot\sum\limits_{(i, j)\in E(CQ)}\ln\frac{L_{i}^{*}L_{j}}{L_{i}L_{j}^{*}} = \sum\limits_{Q}w(Q)\cdot\ln \left(\prod\limits_{(i, j)\in E(CQ)}\frac{L_{i}^{*}L_{j}}{L_{i}L_{j}^{*}}\right). $

    For each unicyclic subetaaph $ Q $, we can see that

    $ \prod\limits_{(i, j)\in E(CQ)}\frac{L_{i}^{*}L_{j}}{L_{i}L_{j}^{*}} = 1, $

    which implies that

    $ \ln \left(\prod\limits_{(i, j)\in E(CQ)}\frac{L_{i}^{*}L_{j}}{L_{i}L_{j}^{*}}\right) = 0. $

    For example, we set $ n = 2 $, the unique cycle $ CQ $ has two vertices with the cycle $ 1\rightarrow 2 \rightarrow 1 $, and $ E(CQ) = \{(1, 2), (2, 1)\} $. Then we can get $ v_{1} = \bar{\beta}_{21} $, $ v_{2} = \bar{\beta}_{12} $, and

    $ \prod\limits_{(i, j)\in E(CQ)}\frac{L_{i}^{*}L_{j}}{L_{i}L_{j}^{*}} = \frac{L_{1}^{*}L_{2}}{L_{1}L_{2}^{*}}\cdot\frac{L_{2}^{*}L_{1}}{L_{2}L_{1}^{*}} = 1. $

    So, we have $ W_{n}\equiv 0 $ holds for all $ L_{1}, L_{2}, \cdots, L_{n}>0 $.

    Therefore, combining with (4.19)-(4.20), (4.22) and using properties of $ \varphi(x) $, we have $ \frac{d U_{ape}(t)}{dt}\leq 0 $ for all $ (S_{1}, E_{1}, A_{1}, \ldots, S_{n}, E_{n}, A_{n})\in\Omega^{0} $, and with the equality holds if and only if $ S_{i}(t) = S_{i}^{*}, E_{i}(t) = E_{i}^{*}, L_{i}(t) = L_{i}(t-\theta) = L_{i}^{*} $ for all $ t\geq0 $, $ \theta\in [0, \theta^{+}] $, $ i = 1, \ldots, n $. We conclude that the largest invariant set $ \{\frac{d U_{ape}(t)}{dt} = 0\} = \{P^{*}\} $. The $ \omega $-limit set of $ \Omega^{0} $ consist of just the alcohol-present equilibrium $ P^{*} $. Therefore, using the LaSalle's Invariance Principle, we see that the alcohol-present equilibrium $ P^{*} $ is globally attractive in $ \Omega^{0} $ if $ R_{0}>1 $ for $ n\geq 2 $.

    As for the local asymptotical stability of the alcohol-present equilibrium $ P^{*} $, which can be proved by the way of Corollary 5.3.1 in [60]. Therefore, using an argument similar to that in the proof of Theorem 4.1 (1), the alcohol-present equilibrium $ P^{*} $ is globally asymptotically stable in $ \Omega^{0} $ if $ R_{0}>1 $, which consequently implies that the alcohol-present equilibrium $ P^{*} $ is unique. The proof of Theorem 4.2 is completed.

    In this section, we will discuss the effects of the public health education and alcoholism age on the alcohol control. It follows from Theorems 4.1 and 4.2 that the global dynamics of system (3.5) are completely determined by the basic reproduction number

    $ R_{0} = \rho(M_{0}), ~~~~~~M_{0} = (βij(BjS0i+CjE0i)dLi+γi)
    _{n\times n}, $

    where $ S_{i}^{0} = \frac{\Lambda_{i}}{d_{i}^{S}+\xi_{i}} $, $ E_{i}^{0} = \frac{\xi_{i}\Lambda_{i}}{(d_{i}^{S}+\xi_{i})(d_{i}^{E}+\alpha_{i})} $, and $ B_{j} = \int_{0}^{\infty}h_{j}(\theta)d\theta $, $ C_{j} = \int_{0}^{\infty}g_{j}(\theta)d\theta $.

    For convenience, we assume that the natural death rate of susceptible population (including uneducated and educated susceptible population) is the same, that is, $ d_{i}^{S} = d_{i}^{E} $. To investigate the effect of the public health education, we consider the special case with $ \xi_{i} = 0 $, which is the rate of the susceptible population who accepted the public health education entering into the educated class. Then for $ \xi_{i} = 0 $, we have $ \hat{E_{i}^{0}} = 0 $ and $ \hat{S_{i}^{0}} = \frac{\Lambda_{i}}{d_{i}^{S}} $. Note that $ \frac{\Lambda_{i}}{d_{i}^{S}}-\frac{\Lambda_{i}}{d_{i}^{S}+\xi_{i}} = \frac{\xi_{i}\Lambda_{i}}{d_{i}^{S}(d_{i}^{S}+\xi_{i})}>\frac{\xi_{i}\Lambda_{i}}{(d_{i}^{S}+\xi_{i})(d_{i}^{E}+\alpha_{i})} $ and $ B_{j}\geq C_{j} $, we can obtain that

    $ \hat{R_{0}} = R_{0}|_{\xi_{i} = 0} \gt R_{0}. $

    This implies that public health education leads to the basic reproduction number decline. On the other hand, the reproduction number $ R_{0} $ is an increasing function of transmission coefficient $ \beta_{ij} $. By increasing public health educational campaigns at all social levels, the value of $ \xi_{i} $ will be increase. Hence, the awareness about drinking will alert the susceptible individuals so that they isolate themselves and decline to drink or drink moderately. This leads to a decrease in the value of $ \beta_{ij} $. In this case, the value of the reproduction number $ R_{0} $ will be decrease. Considering both the cost and the practical purposes, efforts to increase public health education are more effective in controlling the spread of alcohol problems than efforts to increase the number of individuals who have access to treatment. Therefore, public health education is one of the effective measures to control the spread of alcohol problems and it is beneficial for alcoholism control in the whole society.

    However, $ R_{0} $ is a decreasing function of $ \gamma_{i} $ which is the rate of the light drinkers enter into the alcoholism compartment. From the assumption (H7), alcoholism age $ \theta $ will cause the value of the reproduction number $ R_{0} $ increase, which is bad for alcoholism control. It suggests that the longer the light drinkers stay in their compartment, the better alcohol problems will be controlled.

    The goal of this paper is to analysis threshold dynamics of a multi-group alcoholism epidemic model with public health education and alcoholism age in heterogeneous populations. Our results expands the previous related works which have been obtained in single-group models without alcoholism age. Mathematical analysis shown that the global asymptotic behavior of multi-group alcoholism model is completely determined by the size of the basic reproduction number $ R_{0} $. By using the theory of non-negative matrices and the classical method of Lyapunov functional, Theorem 4.1 implies that the alcohol problems dies out in the sense that alcoholism fractions go to zero from all the groups if $ R_{0}\leq1 $. By applications of the graph-theoretic approach to the method of Lyapunov functionals, we proved the existence, uniqueness and global asymptotic stability of the alcohol-present equilibrium $ P^{*} $ for $ R_{0}>1 $, see Theorem 4.2. Our results implies that the alcoholism will persist in all the groups of the population and will eventually settle at a constant level in each group.

    Our main results indicate that the dynamics of alcoholism model (2.1) is similar to that for the models without considering multi-group and alcoholism age. That is, heterogeneity of populations does not alter the dynamical behaviors as shown in [36]. However, our model is more realistic than the corresponding models already established. Because by decomposing the heterogeneous population into several subetaoups, the effects of both the intra-group and inter-group are considered. Moreover, inclusion of alcoholism population with alcoholism age leads to the basic reproduction number $ R_{0} $ is increase. So the heterogeneity of populations plays an important role on $ R_{0} $, it affects the global dynamics of the model. Strengthening public health education and controlling the age of drinking have a positive role in alcoholism control. In addition, there is an innovation in the method of study in this paper. Our findings in this paper may be valuable for the health workers who are performing alcoholism control. These results are provided with intention to inform and assist policy-makers in targeting education and treatment resources for maximum effectiveness.

    Of course, other factors, such as time lag for alcoholism or nonlinear transmission rate, taking into account the model which will make the model more realistic. In addition, relapse is very common when it comes to drinking. Research on these issues remains to be completed in the future.

    This research is supported by the NNSF of China (11461041 and 11661050), the NSF of Gansu Province (148RJZA024 and 061706), and the Development Program for HongLiu Outstanding Young Teachers in Lanzhou University of Technology. We are grateful to the anonymous referees and the editors for their valuable comments and suggestions which improved the quality of the paper.

    The authors declare that there is no conflict of interests regarding the publication of this paper.

    [1] Hadziyannis S, Gerber MA, Vissoulis C, et al. (1973) Cytoplasmic hepatitis B antigen in "ground-glass" hepatocytes of carriers. Arch Pathol 96: 327–330.
    [2] Pópper H (1975) The ground glass hepatocyte as a diagnostic hint. Hum Pathol 6: 517–520. doi: 10.1016/S0046-8177(75)80069-4
    [3] Shikata T (1973) Australia antigen in liver tissue-an immunofluorescent and immunoelectron microscopic study. Jpn J Exp Med 43: 231–245.
    [4] Shikata T, Uzawa T, Yoshiwara N, et al. (1974) Staining methods of Australia antigen in paraffin section-detection of cytoplasmic inclusion bodies. Jpn J Exp Med 44: 25–36.
    [5] Gudat F, Bianchi L, Sonnabend W, et al. (1975) Pattern of core and surface expression in liver tissue reflects state of specific immune response in hepatitis B. Lab Invest 32: 1–9.
    [6] Hsu H, Lai M, Su I, et al. (1988) Correlation of hepatocyte HBsAg expression with virus replication and liver pathology. Hepatology 8: 749–754. doi: 10.1002/hep.1840080408
    [7] Wang HC, Wu HC, Chen CF, et al. (2003) Different types of ground glass hepatocytes in chronic hepatitis B virus infection contain specific pre-S mutants that may induce endoplasmic reticulum stress. Am J Pathol 163: 2441–2449. doi: 10.1016/S0002-9440(10)63599-7
    [8] Su IJ, Kuo TT, Liaw YF (1985) Hepatocyte hepatitis B surface antigen. Diagnostic evaluation of patients with clinically acute hepatitis B surface antigen-positive hepatitis. Arch Pathol Lab Med 109: 400–402.
    [9] Hsu HC, Lin YH, Chang MH, et al. (1988) Pathology of chronic hepatitis B virus infection in children: with special reference to the intrahepatic expression of hepatitis B virus antigens. Hepatology 8: 378–382. doi: 10.1002/hep.1840080232
    [10] Fan YF, Lu CC, Chang YC, et al. (2000) Identification of a pre-S2 mutant in hepatocytes expressing a novel marginal pattern of surface antigen in advanced diseases of chronic hepatitis B virus infection. J Gastroenterol Hepatol 15: 519–528. doi: 10.1046/j.1440-1746.2000.02187.x
    [11] Su IJ, Lai MY, Hsu HC, et al. (1986) Diverse virological, histopathological and prognostic implications of seroconversion from hepatitis B e antigen to anti-HBe in chronic hepatitis B virus infection. J Hepatol 3: 182–189. doi: 10.1016/S0168-8278(86)80024-1
    [12] Tai PC, Banik D, Lin GI, et al. (1997) Novel and frequent mutations of hepatitis B virus coincide with a major histocompatibility complex class I-restricted T-cell epitope of the surface antigen. J Virol 71: 4852–4856.
    [13] Hsu HC, Wu TT, Wu MZ, et al. (1988) Evolution of expression of hepatitis B surface and core antigens (HBsAg, HBcAg) in resected primary and recurrent hepatocellular carcinoma in HBsAg carriers in Taiwan. Correlation with local host immune response. Cancer 62: 915–921.
    [14] Mimms L (1995) Hepatitis B virus escape mutants: "Pushing the envelope" of chronic hepatitis B virus infection. Hepatology 21: 884–887.
    [15] Hsieh YH, Su IJ, Wang HC, et al. (2007) Hepatitis B virus pre-S2 mutant surface antigen induces degradation of cyclin-dependent kinase inhibitor p27Kip1 through c-Jun activation domain-binding protein 1. Mol Cancer Res 5: 1063–1072. doi: 10.1158/1541-7786.MCR-07-0098
    [16] Liu S, Zhang H, Gu C, et al. (2009) Associations between hepatitis B virus mutations and the risk of hepatocellular carcinoma: A meta-analysis. J Natl Cancer Inst 101: 1066–1082. doi: 10.1093/jnci/djp180
    [17] Santantonio T, Jung MC, Schneider R, et al. (1992) Hepatitis B virus genomes that cannot synthesize pre-S2 proteins occur frequently and as dominant virus populations in chronic carriers in Italy. Virology 188: 948–952. doi: 10.1016/0042-6822(92)90559-8
    [18] Fan YF, Lu CC, Chen WC, et al. (2001) Prevalence and significance of hepatitis B virus (HBV) pre-S mutants in serum and liver at different replicative stages of chronic HBV infection. Hepatology 33: 277–286. doi: 10.1053/jhep.2001.21163
    [19] Huy TT, Ushijima H, Win KM, et al. (2003) High prevalence of hepatitis B virus pre-S mutant in countries where it is endemic and its relationship with genotype and chronicity. J Clin Microbiol 41: 5449–5455. doi: 10.1128/JCM.41.12.5449-5455.2003
    [20] Fernholz D, Galle PR, Stemler M, et al. (1993) Infectious hepatitis B virus variant defective in pre-S2 protein expression in a chronic carrier. Virology 194: 137–148. doi: 10.1006/viro.1993.1243
    [21] Le SJ, Chouteau P, Cannie I, et al. (1998) Role of the pre-S2 domain of the large envelope protein in hepatitis B virus assembly and infectivity. J Virol 72: 5573–5578.
    [22] Su IJ, Wang HC, Wu HC, et al. (2008) Ground glass hepatocytes contain pre-S mutants and represent preneoplastic lesions in chronic hepatitis B virus infection. J Gastroenterol Hepatol 23: 1169–1174. doi: 10.1111/j.1440-1746.2008.05348.x
    [23] Wang M, Kaufman RJ (2014) The impact of the endoplasmic reticulum protein-folding environment on cancer development. Nat Rev Cancer 14: 581–597. doi: 10.1038/nrc3800
    [24] Hsieh YH, Su IH, Chang WW, et al. (2004) Pre-S mutant surface antigens in chronic hepatitis B virus infection induce oxidative stress and DNA damage. Carcinogenesis 25: 2023–2032. doi: 10.1093/carcin/bgh207
    [25] Hung JH, Su IJ, Lei HY, et al. (2004) Endoplasmic reticulum stress stimulates the expression of cyclooxygenase-2 through activation of NF-kappaB and pp38 mitogen-activated protein kinase. J Biol Chem 279: 46384–46392. doi: 10.1074/jbc.M403568200
    [26] Yen TT, Yang A, Chiu WT, et al. (2016) Hepatitis B virus PreS2-mutant large surface antigen activates store-operated calcium entry and promotes chromosome instability. Oncotarget 7: 23346–23360.
    [27] Smyth JT, Hwang SY, Tomita T, et al. (2010) Activation and regulation of store-operated calcium entry. J Cell Mol Med 14: 2337–2349. doi: 10.1111/j.1582-4934.2010.01168.x
    [28] Wang LH, Huang W, Lai MD, et al. (2012) Aberrant cyclin A expression and centrosome overduplication induced by hepatitis B virus pre-S2 mutants and its implication in hepatocarcinogenesis. Carcinogenesis 33: 466–472. doi: 10.1093/carcin/bgr296
    [29] Wang HC, Chang WT, Chang WW, et al. (2005) Hepatitis B virus pre-S2 mutant upregulates cyclin A expression and induces nodular proliferation of hepatocytes. Hepatology 41: 761–770. doi: 10.1002/hep.20615
    [30] Sunami Y, Ringelhan M, Kokai E, et al. (2016) Canonical NF-kappaB signaling in hepatocytes acts as a tumor-suppressor in hepatitis B virus surface antigen-driven hepatocellular carcinoma by controlling the unfolded protein response. Hepatology 63: 1592–1607. doi: 10.1002/hep.28435
    [31] Hsieh YH, Chang YY, Su IJ, et al. (2015) Hepatitis B virus pre-S2 mutant large surface protein inhibits DNA double-strand break repair and leads to genome instability in hepatocarcinogenesis. J Pathol 236: 337–347. doi: 10.1002/path.4531
    [32] Hsieh YH, Hsu JL, Su IJ, et al. (2011) Genomic instability caused by hepatitis B virus: Into the hepatoma inferno. Front Biosci 16: 2586–2597. doi: 10.2741/3874
    [33] Akbar SK, Onji M (1998) Hepatitis B virus (HBV)-transgenic mice as an investigative tool to study immunopathology during HBV infection. Int J Exp Pathol 79: 279–291.
    [34] Huang SN, Chisari FV (1995) Strong, sustained hepatocellular proliferation precedes hepatocarcinogenesis in hepatitis B surface antigen transgenic mice. Hepatology 21: 620–626.
    [35] Hagen TM, Huang S, Curnutte J, et al. (1994) Extensive oxidative DNA damage in hepatocytes of transgenic mice with chronic active hepatitis destined to develop hepatocellular carcinoma. Proc Natl Acad Sci U S A 91: 12808–12812. doi: 10.1073/pnas.91.26.12808
    [36] Wang HC, Huang W, Lai MD, et al. (2006) Hepatitis B virus pre-S mutants, endoplasmic reticulum stress and hepatocarcinogenesis. Cancer Sci 97: 683–688. doi: 10.1111/j.1349-7006.2006.00235.x
    [37] Zheng Y, Chen WL, Louie SG, et al. (2007) Hepatitis B virus promotes hepatocarcinogenesis in transgenic mice. Hepatology 45: 16–21. doi: 10.1002/hep.21445
    [38] Teng YC, Neo JC, Wu JC, et al. (2017) Expression of a hepatitis B virus pre-S2 deletion mutant in the liver results in hepatomegaly and hepatocellular carcinoma in mice. J Pathol 241: 463–474. doi: 10.1002/path.4850
    [39] Shen FC, Su IJ, Wu HC, et al. (2009) A pre-S gene chip to detect pre-S deletions in hepatitis B virus large surface antigen as a predictive marker for hepatoma risk in chronic hepatitis B virus carriers. J Biomed Sci 16: 84. doi: 10.1186/1423-0127-16-84
    [40] Tsai HW, Lin YJ, Wu HC, et al. (2016) Resistance of ground glass hepatocytes to oral antivirals in chronic hepatitis B patients and implication for the development of hepatocellular carcinoma. Oncotarget 7: 27724–27734.
    [41] Chang TT, Liaw YF, Wu SS, et al. (2010) Long-term entecavir therapy results in the reversal of fibrosis/cirrhosis and continued histological improvement in patients with chronic hepatitis B. Hepatology 52: 886–893. doi: 10.1002/hep.23785
    [42] Singal AK, Salameh H, Kuo YF, et al. (2013) Meta-analysis: The impact of oral anti-viral agents on the incidence of hepatocellular carcinoma in chronic hepatitis B. Aliment Pharmacol Ther 38: 98–106. doi: 10.1111/apt.12344
    [43] Wu CY, Chen YJ, Ho HJ, et al. (2012) Association between nucleoside analogues and risk of hepatitis B virus-related hepatocellular carcinoma recurrence following liver resection. JAMA 308: 1906–1914. doi: 10.1001/2012.jama.11975
    [44] Tsai HW, Lin YJ, Lin PW, et al. (2011) A clustered ground-glass hepatocyte pattern represents a new prognostic marker for the recurrence of hepatocellular carcinoma after surgery. Cancer 117: 2951–2960. doi: 10.1002/cncr.25837
    [45] Manesis EK, Hadziyannis ES, Angelopoulou OP, et al. (2007) Prediction of treatment-related HBsAg loss in HBeAG-negative chronic hepatitis B: A clue from serum HBsAg levels. Antiviral Ther 12: 73–82.
    [46] Wong DK, Cheung AM, O'Rourke K, et al. (1993) Effect of alpha-interferon treatment in patients with hepatitis B e antigen-positive chronic hepatitis B. A meta-analysis. Ann Intern Med 119: 312–323. doi: 10.7326/0003-4819-119-4-199308150-00011
    [47] Teng CF, Wu HC, Hsieh WC, et al. (2015) Activation of ATP citrate lyase by mTOR signal induces disturbed lipid metabolism in hepatitis B virus pre-S2 mutant tumorigenesis. J Virol 89: 605–614. doi: 10.1128/JVI.02363-14
    [48] Loguercio C, Andreone P, Brisc C, et al. (2012) Silybin combined with phosphatidylcholine and vitamin E in patients with nonalcoholic fatty liver disease: A randomized controlled trial. Free Radical Biol Med 52: 1658–1665. doi: 10.1016/j.freeradbiomed.2012.02.008
    [49] Polyak SJ, Morishima C, Lohmann V, et al. (2010) Identification of hepatoprotective flavonolignans from silymarin. Proc Natl Acad Sci U S A 107: 5995–5999. doi: 10.1073/pnas.0914009107
    [50] Jang M, Cai L, Udeani GO, et al. (1997) Cancer chemopreventive activity of resveratrol, a natural product derived from grapes. Science 275: 218–220. doi: 10.1126/science.275.5297.218
    [51] Gusman J, Malonne H, Atassi G (2001) A reappraisal of the potential chemopreventive and chemotherapeutic properties of resveratrol. Carcinogenesis 22: 1111–1117. doi: 10.1093/carcin/22.8.1111
    [52] Fulda S (2010) Resveratrol and derivatives for the prevention and treatment of cancer. Drug Discovery Today 15: 757–765. doi: 10.1016/j.drudis.2010.07.005
    [53] Guarente L (2006) Sirtuins as potential targets for metabolic syndrome. Nature 444: 868–874. doi: 10.1038/nature05486
    [54] Lekli I, Ray D, Das DK (2010) Longevity nutrients resveratrol, wines and grapes. Genes Nutr 5: 55–60. doi: 10.1007/s12263-009-0145-2
    [55] Teng CF, Hsieh WC, Wu HC, et al. (2015) Hepatitis B Virus Pre-S2 Mutant Induces Aerobic Glycolysis through Mammalian Target of Rapamycin Signal Cascade. PLoS One 10: e0122373. doi: 10.1371/journal.pone.0122373
  • This article has been cited by:

    1. Zizhen Zhang, Junchen Zou, Soumen Kundu, Bifurcation and optimal control analysis of a delayed drinking model, 2020, 2020, 1687-1847, 10.1186/s13662-020-02987-1
    2. Zizhen Zhang, Junchen Zou, Ranjit Kumar Upadhyay, Stability and Hopf bifurcation of a delayed giving up smoking model with harmonic mean type incidence rate and relapse, 2020, 19, 22113797, 103619, 10.1016/j.rinp.2020.103619
    3. Divine Wanduku, On the almost sure convergence of a stochastic process in a class of nonlinear multi-population behavioral models for HIV/AIDS with delayed ART treatment, 2020, 0736-2994, 1, 10.1080/07362994.2020.1848593
    4. Divine Wanduku, Finite- and multi-dimensional state representations and some fundamental asymptotic properties of a family of nonlinear multi-population models for HIV/AIDS with ART treatment and distributed delays, 2021, 0, 1937-1179, 0, 10.3934/dcdss.2021005
    5. Banghua Yang, Xuelin Gu, Shouwei Gao, Ding Xu, Classification accuracy and functional difference prediction in different brain regions of drug abuser prefrontal lobe basing on machine-learning, 2021, 18, 1551-0018, 5692, 10.3934/mbe.2021288
  • 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(9340) PDF downloads(1079) Cited by(5)

Figures and Tables

Figures(4)

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog