Within-host infectious disease models accommodating cellular coinfection, with an application to influenza

Katia Koelle, Alex P Farrell, Christopher B Brooke, Ruian Ke

Virus Evolution

July 8, 2019


Within-host models are useful tools for understanding the processes regulating viral load dynamics. While existing models have considered a wide range of within-host processes, at their core these models have shown remarkable structural similarity. Specifically, the structure of these models generally consider target cells to be either uninfected or infected, with the possibility of accommodating further resolution (e.g. cells that are in an eclipse phase). Recent findings, however, indicate that cellular coinfection is the norm rather than the exception for many viral infectious diseases, and that cells with high multiplicity of infection are present over at least some duration of an infection. The reality of these cellular coinfection dynamics is not accommodated in current within-host models although it may be critical for understanding within-host dynamics. This is particularly the case if multiplicity of infection impacts infected cell phenotypes such as their death rate and their viral production rates. Here, we present a new class of within-host disease models that allow for cellular coinfection in a scalable manner by retaining the low-dimensionality that is a desirable feature of many current within-host models. The models we propose adopt the general structure of epidemiological ‘macroparasite’ models that allow hosts to be variably infected by parasites such as nematodes and host phenotypes to flexibly depend on parasite burden. Specifically, our within-host models consider target cells as ‘hosts’ and viral particles as ‘macroparasites’, and allow viral output and infected cell lifespans, among other phenotypes, to depend on a cell’s multiplicity of infection. We show with an application to influenza that these models can be statistically fit to viral load and other within-host data, and demonstrate using model selection approaches that they have the ability to outperform traditional within-host viral dynamic models. Important in vivo quantities such as the mean multiplicity of cellular infection and time-evolving reassortant frequencies can also be quantified in a straightforward manner once these macroparasite models have been parameterized. The within-host model structure we develop here provides a mathematical way forward to address questions related to the roles of cellular coinfection, collective viral interactions, and viral complementation in within-host viral dynamics and evolution.

Design of vaccine efficacy trials during public health emergencies

Natalie E. Dean, Pierre-Stéphane Gsell, Ron Brookmeyer, Victor De Gruttola, Christl A. Donnelly, M. Elizabeth Halloran, Momodou Jasseh, Martha Nason, Ximena Riveros, Conall H. Watson, Ana Maria Henao-Restrepo, Ira M. Longini

Science Translational Medicine

July 3, 2019


Public health emergencies, such as an Ebola disease outbreak, provide a complex and challenging environment for the evaluation of candidate vaccines. Here, we outline the need for flexible and responsive vaccine trial designs to be used in public health emergencies, and we summarize recommendations for their use in this setting.

A general framework for modelling the impact of co-infections on pathogen evolution

Mary Bushman and Rustom Antia

Royal Society Interface

June 26, 2019


Theoretical models suggest that mixed-strain infections, or co-infections, are an important driver of pathogen evolution. However, the within-host dynamics of co-infections vary enormously, which complicates efforts to develop a general understanding of how co-infections affect evolution. Here, we develop a general framework which condenses the within-host dynamics of co-infections into a few key outcomes, the most important of which is the overall R0 of the co-infection. Similar to how fitness is determined by two different alleles in a heterozygote, the R0 of a co-infection is a product of the R0 values of the co-infecting strains, shaped by the interaction of those strains at the within-host level. Extending the analogy, we propose that the overall R0 reflects the dominance of the co-infecting strains, and that the ability of a mutant strain to invade a population is a function of its dominance in co-infections. To illustrate the utility of these concepts, we use a within-host model to show how dominance arises from the within-host dynamics of a co-infection, and then use an epidemiological model to demonstrate that dominance is a robust predictor of the ability of a mutant strain to save a maladapted wild-type strain from extinction (evolutionary emergence).

Successes and failures of the live-attenuated influenza vaccine, can we do better?

Laura Matrajt, M. Elizabeth Halloran, Rustom Antia

Clinical Infectious Diseases

May 6, 2019


Live-attenuated vaccines are usually highly effective against many acute viral infections. However, the effective- ness of the live attenuated influenza vaccine (LAIV) can vary widely, ranging from 0% effectiveness in some studies done in the United States to 50% in studies done in Europe. The reasons for these discrepancies remain largely unclear. In this paper we use mathematical models to explore how the efficacy of LAIV is affected by the degree of mismatch with the currently circulating influenza strain and interference with pre-existing immunity. The model incorporates two key antigenic distances - the distance between pre-existing immunity and the currently circulating strain as well as the LAIV strain. Our models show that a LAIV that is matched with the currently circulating strain is likely to have only modest efficacy. Our results suggest that the efficacy of the vaccine would be increased (optimized) if, rather than being matched to the circulating strain, it is antigenically slightly further from pre-existing immunity compared with the circulating strain. The models also suggest two regimes in which LAIV that is matched to circulating strains may provide effective protection. The first is in children before they have built immunity from circulating strains. The second is in response to novel strains (such as antigenic shifts) which are at substantial antigenic distance from previously circulating strains. Our models provide an explanation for the variation in vaccine effectiveness, both between children and adults as well as between studies of vaccine effectiveness observed during the 2014-15 influenza season in different countries.

A Spatio-Temporal Modeling Framework for Surveillance Data of Multiple Infectious Pathogens With Small Laboratory Validation Sets

Xueying Tang, Yang Yang, Hong-Jie Yu, Qiao-Hong Liao, Nikolay Bliznyuk

Journal of the American Statistical Association

April 30, 2019


Many surveillance systems of infectious diseases are syndrome-based, capturing patients by clinical manifestation. Only a fraction of patients, mostly severe cases, undergo laboratory validation to identify the underlying pathogen. Motivated by the need to understand transmission dynamics and associate risk factors of enteroviruses causing the hand, foot, and mouth disease (HFMD) in China, we developed a Bayesian spatio-temporal modeling framework for surveillance data of infectious diseases with small validation sets. A novel approach was proposed to sample unobserved pathogen-specific patient counts over space and time and was compared to an existing sampling approach. The practical utility of this framework in identifying key parameters was assessed in simulations for a range of realistic sizes of the validation set. Several designs of sampling patients for laboratory validation were compared with and without aggregation of sparse validation data. The methodology was applied to the 2009 HFMD epidemic in southern China to evaluate transmissibility and the effects of climatic conditions for the leading pathogens of the disease, enterovirus 71, and Coxsackie A16. Supplementary materials for this article are available online.

Estimating effective population size changes from preferentially sampled genetic sequences

Michael D. Karcher, Marc A. Suchard, Gytis Dudas, Vladimir N. Minin


March 28, 2019


Coalescent theory combined with statistical modeling allows us to estimate effective population size fluctuations from molecular sequences of individuals sampled from a population of interest. When sequences are sampled serially through time and the distribution of the sampling times depends on the effective population size, explicit statistical modeling of sampling times improves population size estimation. Previous work assumed that the genealogy relating sampled sequences is known and modeled sampling times as an inhomogeneous Poisson process with log-intensity equal to a linear function of the log-transformed effective population size. We improve this approach in two ways. First, we extend the method to allow for joint Bayesian estimation of the genealogy, effective population size trajectory, and other model parameters. Next, we improve the sampling time model by incorporating additional sources of information in the form of time-varying covariates. We validate our new modeling framework using a simulation study and apply our new methodology to analyses of population dynamics of seasonal influenza and to the recent Ebola virus outbreak in West Africa.

Effects of infection history on dengue virus infection and pathogenicity

Tim K. Tsang, Samson L. Ghebremariam, Lionel Gresh, Aubree Gordon, M. Elizabeth Halloran, Leah C. Katzelnick, Diana Patricia Rojas, Guillermina Kuan, Angel Balmaseda, Jonathan Sugimoto, Eva Harris,
Ira M. Longini Jr., Yang Yang

Nature Communications

March 18, 2019


The understanding of immunological interactions among the four dengue virus (DENV) serotypes and their epidemiological implications is often hampered by the lack of individual-level infection history. Using a statistical framework that infers full infection history, we analyze a prospective pediatric cohort in Nicaragua to characterize how infection history modulates the risks of DENV infection and subsequent clinical disease. After controlling for age, one prior infection is associated with 54% lower, while two or more are associated with 91% higher, risk of a new infection, compared to DENV-naive children. Children >8 years old have 55% and 120% higher risks of infection and subsequent disease, respectively, than their younger peers. Among children with ≥1 prior infection, intermediate antibody titers increase, whereas high titers lower, the risk of subsequent infection, compared with undetectable titers. Such complex dependency needs to be considered in the design of dengue vaccines and vaccination strategies.

Fitting stochastic epidemic models to gene genealogies using linear noise approximation

Mingwei Tang, Gytis Dudas, Trevor Bedford, Vladimir N. Minin


February 24, 2019


Phylodynamics is a set of population genetics tools that aim at reconstructing demographic history of a population based on molecular sequences of individuals sampled from the population of interest. One important task in phylodynamics is to estimate changes in (effective) population size. When applied to infectious disease sequences such estimation of population size trajectories can provide information about changes in the number of infections. To model changes in the number of infected individuals, current phylodynamic methods use non-parametric approaches, parametric approaches, and stochastic modeling in conjunction with likelihood-free Bayesian methods. The first class of methods yields results that are hard-to-interpret epidemiologically. The second class of methods provides estimates of important epidemiological parameters, such as infection and removal/recovery rates, but ignores variation in the dynamics of infectious disease spread. The third class of methods is the most advantageous statistically, but relies on computationally intensive particle filtering techniques that limits its applications. We propose a Bayesian model that combines phylodynamic inference and stochastic epidemic models, and achieves computational tractability by using a linear noise approximation (LNA) --- a technique that allows us to approximate probability densities of stochastic epidemic model trajectories. LNA opens the door for using modern Markov chain Monte Carlo tools to approximate the joint posterior distribution of the disease transmission parameters and of high dimensional vectors describing unobserved changes in the stochastic epidemic model compartment sizes (e.g., numbers of infectious and susceptible individuals). We apply our estimation technique to Ebola genealogies estimated using viral genetic data from the 2014 epidemic in Sierra Leone and Liberia.

Impact of rotavirus vaccine introduction in children less than 2 years of age presenting for medical care with diarrhea in rural Matlab, Bangladesh

Lauren M Schwartz, K Zaman, Md Yunus, Ahasan-ul H Basunia, Abu Syed Golam Faruque, Tahmeed Ahmed, Mustafizur Rahman, Jonathan D Sugimoto, M Elizabeth Halloran, Ali Rowhani-Rahbar, Kathleen M Neuzil, John C Victor

Clinical Infectious Diseases

February 12, 2019



Following the conclusion of a Rotarix vaccine (HRV) cluster-randomized controlled trial (CRT) in Matlab, Bangladesh, HRV was included in Matlab’s routine immunization program. We describe the population-level impact of programmatic rotavirus vaccination in Bangladesh in children <2 years of age


Interrupted time series were used to estimate the impact of HRVintroduction. Diarrheal surveillance collected between 2000 and 2014 within the two service delivery areas (icddr,b service area [ISA] and government service area [GSA]) of the Matlab Health and Demographic Surveillance System administered by icddr,b was used. Age-group specific incidence rates were calculated for both rotavirus-positive (RV+) and rotavirus-negative (RV-) diarrhea of any severity presenting to the hospital. Two models were used to assess impact within each service area: Model 1 used the pre-vaccine time period in all villages (HRV- and control-only) and Model 2 combined the pre-vaccine time period and the CRT time period using outcomes from control-only villages.


Both models demonstrated a downward trend in RV+ diarrheal incidence in the ISA villages during 3.5 years of routine HRV use, though only Model 2 was statistically significant. Significant impact of HRV on RV+ diarrhea incidence in GSA villages was not observed in either model. Differences in population-level impact between the two delivery areas may be due to varied rotavirus vaccine coverage and presentation rate to the hospital.


This study provides initial evidence of the population-level impact of rotavirus vaccines in children <2 years of age in Matlab, Bangladesh. Further studies of rotavirus vaccine impact after nationwide introduction in Bangladesh are needed.