Immunotherapy trial simulators: Using mathematical, computational and statistical techniques to simulate trial design for new products in the treatment, vaccination and immunoprophylaxis of disease.

Proposal
1137

Title of Proposed Research

Lead Researcher

Roy Anderson, PhD

Affiliation

Imperial College LondonLondonUK

Funding Source

The research project is fully supported by Crucell Vaccine Institute (CVI), Janssen Centre of Excellence in Leiden, the Netherlands.

Potential Conflicts of Interest

Lead Researcher: Independent Non-Executive Director for GSK

Data Sharing Agreement Date

16 April 2015

Lay Summary

Background of the Project

Influenza A infection has a huge impact on human health; the WHO estimates that annual epidemics result in around 3-5 million cases of severe illness, and 250-500,000 deaths. The Influenza virus mutates frequently, meaning drugs and vaccines have a limited lifespan and development of new treatment and vaccines is necessary. Mathematical models can aid our understanding of disease progression and can therefore help in the design of drugs, vaccines and immunotherapies. In addition, simulating clinical trials can help predict outcomes of novel therapy and make the trials more productive for researchers and less intrusive for patients. Sophisticated computational models of the immune response to Influenza can predict various things such as; when to administer treatment, what samples need to be collected, what needs to be measured and how many people must be tested to prove an effect.
Aim and Objectives

The aim of our research project is to design a clinical trial simulator, to examine new vaccines and immunotherapy approaches against Influenza A infection. We will develop deterministic and stochastic mathematical models of the time course of Influenza within the human host and the associated immune response. The aim is to keep these models as simple and tractable as possible and only to include the essential features that characterise the immune response to influenza. To reliably identify these features we will need access to high-quality human data. Once the essential features have been identified we will introduce potential immunomodulatory treatments into the model to predict their effects.

Integral to the development of these model simulators will be the construction of databases for Influenza A. The goal is to collect the best cohort studies from around the world to be able to include reliable data on both the progression and incidence of Influenza A. The compilation of these databases themselves is a valuable exercise, rarely attempted in clinical studies.

Study Data Provided

[{ "PostingID": 281, "Title": "GSK-NAI30008", "Description": "A Double-Blind, Randomized, Placebo-Controlled, Parallel Group, Multicenter Study To Investigate The Efficacy And Safety Of Zanamivir (GG167) 10mg Administered Twice A Day For Five Days In The Treatment Of Influenza In Patients 12 Years Or Over With Asthma Or Chronic Obstructive Pulmonary Disease (COPD)

Medicine: zanamivir, Condition: Influenza, Human, Phase: 3, Clinical Study ID: NAI30008, Sponsor: GSK" },{ "PostingID": 282, "Title": "GSK-NAI30009", "Description": "A double-blind, randomized, placebo-controlled, parallel-group, multicenter study to investigate the efficacy and safety of zanamivir (GG167) 10 mg administered by inhalation twice daily for five days in the treatment of symptomatic influenza A and B viral infections in children ages 5-12.

Medicine: zanamivir, Condition: Influenza, Human, Phase: 3, Clinical Study ID: NAI30009, Sponsor: GSK" },{ "PostingID": 284, "Title": "GSK-NAI30012", "Description": "A Double-Blind, Randomised, Placebo-Controlled, Parallel-Group, Multicentre Study to Investigate the Efficacy and Safety of Inhaled Zanamivir 10mg Administered Twice Daily for Five Days in the Treatment of Symptomatic Influenza A and B Viral Infections in Subjects Aged >= 65 Years.

Medicine: zanamivir, Condition: Influenza, Human, Phase: 3, Clinical Study ID: NAI30012, Sponsor: GSK" },{ "PostingID": 285, "Title": "GSK-NAI30015", "Description": "A double-blind, randomised, placebo-controlled, parallel-group, multicentre study to investigate the efficacy and safety of inhaled zanamivir 10mg administered twice daily for five days in the treatment of symptomatic influenza A and B viral infections in armed services personnel.

Medicine: zanamivir, Condition: Influenza, Human, Phase: 4, Clinical Study ID: NAI30015, Sponsor: GSK" },{ "PostingID": 535, "Title": "GSK-NAIA2005", "Description": "A Double-Blind, Randomized Placebo-Controlled Multicenter Study to Investigate the Efficacy and Safety of GG167 in the Treatment of Influenza A and B Viral Infection

Medicine: zanamivir, Condition: Influenza, Human, Phase: 2, Clinical Study ID: NAIA2005, Sponsor: GSK" },{ "PostingID": 537, "Title": "GSK-NAIA2008", "Description": "A Double-Blind, Randomized, Placebo-Controlled, Multicenter, Parallel and Group Study to Investigate the Efficacy and Safety of GG167 Administered Twice or Four Times a Day for the Treatment of Influenza A and B Viral Infections

Medicine: zanamivir, Condition: Influenza, Human, Phase: 2, Clinical Study ID: NAIA2008, Sponsor: GSK" },{ "PostingID": 543, "Title": "GSK-NAIB2005", "Description": "A Double-Blind, Randomised, Placebo-Controlled, Parallel-Group, Multicentre Study to Investigate the Efficacy and Safety of Inhaled and Intranasal GG167 in the Treatment of Influenza A and B Viral Infections

Medicine: zanamivir, Condition: Influenza, Human, Phase: 2, Clinical Study ID: NAIB2005, Sponsor: GSK" },{ "PostingID": 546, "Title": "GSK-NAIB2008", "Description": "A Double-Blind, Randomized, Placebo-Controlled, Multicenter, Parallel-Group Study to Investigate the Efficacy and Safety of GG167 Administered Twice or Four Times a Day for the Treatment of Influenza A and B Viral Infections.

Medicine: zanamivir, Condition: Influenza, Human, Phase: 2, Clinical Study ID: NAIB2008, Sponsor: GSK" },{ "PostingID": 547, "Title": "GSK-NAIB2009", "Description": "A Double-Blind, Randomized, Placebo-Controlled, Multicenter, Parallel-Group Study to Investigate the Efficacy and Safety of GG167 in the Prevention and/or Progression of Influenza A and B Viral Infections

Medicine: zanamivir, Condition: Influenza, Human, Phase: 2, Clinical Study ID: NAIB2009, Sponsor: GSK" },{"PostingID":3165,"Title":"GSK-NAIA1001","Description":"A Study to Investigate the Effect of Intranasal GR121167X on Infection Rates in Healthy Male Volunteers when Experimentally Inoculated with Influenza A/Texas/91 (H1N1) virus, Sponsor: GSK"},{"PostingID":3166,"Title":"GSK-NAIA1002","Description":"A Study to Investigate the Effect of Intranasal GG167 Initiated at Various Intervals Post Inoculation on Infection in Healthy Volunteers when Experimentally Inoculated with Influenza A/Texas/91 (H1N1) Virus, Sponsor: GSK"},{"PostingID":3167,"Title":"GSK-NAIA1003","Description":"A Study to Investigate the Effect of Intranasal GG167 at Various Dosing Frequencies on Infection in Healthy Volunteers when Experimentally Inoculated with Influenza A/Texas/91 (H1N1) Virus, Sponsor: GSK"},{"PostingID":3168,"Title":"GSK-NAIA1004","Description":"A Study to Investigate the Effect of Intranasal GG167 as Nasal Drops and Nasal Spray on Infection in Healthy Volunteers Experimentally Inoculated with Influenza A/Texas/91 (H1N1) Virus, Sponsor: GSK"},{"PostingID":3169,"Title":"GSK-NAIA1005","Description":"A Study to Investigate the Effect of Intranasal GG167 on Infection in Healthy Volunteers Experimentally Inoculated with Influenza B/YAMAGATA/16/88 Virus, Sponsor: GSK"}]

Statistical Analysis Plan

Fitting approaches - Estimation of the model parameters
We will use Bayesian inference and the Markov Chain Monte Carlo (MCMC) method to estimate the parameters for the within-host influenza models that will be developed. Following this approach, the parameters are assumed to be random variables drawn from an assumed probability distribution. MCMC methods simulate random samples from a parameter space and assign a statistical probability to each sample. By exploring a wider range of the possible parameter space, MCMC methods are more likely to find good parameter estimates and at the same time more accurately quantify the uncertainty of the estimates than conventional fitting methods (e.g. non-linear least squares fitting).

Sub-group analysis
We will fit our models to datasets of individual patients. This will allow us to gain an overview of the variability among different patients infected with influenza A and thus help us to interpret uncertainty of parameter estimates. Moreover, we will be able to analyse variability by following sub-categories as:
- Virus strain;
- Characteristics of the patient, such as age group, gender, where he lives;
- Presence or absence of co-morbidities;
- Vaccination history;
- Antiviral use.
Standard statistical methods will be used to evaluate the results of MCMC analysis (convergence diagnostics, autocorrelation diagnostics).

Validation Process
Once the parameters will be determined, we will compare the simulated data produced by our model with real experimental data to validate the model.

Correlation
Correlation relationship between viral load (peak and shedding parameters) and age of the patient, vaccination history to previous influenza strains, antiviral use, influenza strain and clinical symptoms over time will be examined.

Software to be used
The R(1) package will be used for the analysis while the MCMC simulations will be performed using the pyMC(2) software package.

(1) R Core Team (2014). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL http://www.R-project.org/.
(2) Patil, A., D. Huard and C.J. Fonnesbeck. 2010. PyMC: Bayesian Stochastic Modelling in Python. Journal of Statistical Software, 35(4), pp. 1-81.
The statistical analysis plan will be added after the research is published.

Publication Citation

Cite this article: Hadjichrysanthou C, Caue¨t E,
Lawrence E, Vegvari C, de Wolf F, Anderson
RM. 2016 Understanding the within-host
dynamics of influenza A virus: from theory
to clinical implications. J. R. Soc. Interface 13:
20160289.
http://dx.doi.org/10.1098/rsif.2016.0289