Using Modeling and Simulation to Optimize the Timing of Maternal Influenza Vaccination

Using Modeling and Simulation to Optimize the Timing of Maternal Influenza Vaccination

Model-Informed Drug Development (MIDD) has become an important quantitative tool in drug development to characterize drug disposition and effects across a broad range of therapeutic areas including infectious diseases. MIDD approaches for infectious disease therapeutics have been used in the development of Palivizumab, a monoclonal antibody used for the prevention of Respiratory Syncytial Virus (RSV) in infants at high risk of RSV infection, Rilpivirine and Raltegravir (HIV), Simeprivir (Hepatitis C Virus), Letemovir (Cytomegalovirus), Pretomanid (Tuberculosis), Botulism Toxin Heptavalent (bioterrorism), Tecovirimat (smallpox), and Oseltamivir (influenza virus). Notably, an integrated pharmacology strategy was instrumental in achieving a successful and harmonized approval in Europe and the U.S. for Oseltamivir dosing in babies two weeks of age and older in Europe and the US.1 Further, viral kinetic models have been used to describe the changes in viral load with time in an infected patient. These models can provide important information about the cell infection rate, viral production rate, and viral clearance rate

MIDD for Vaccine Development

Although MIDD is used extensively in anti-infectious drug development, its use to support vaccine development is not widespread. Certara, in collaboration with the Bill and Melinda Gates Foundation and leveraging data from the Maternal Immunization Working Group in the Centers for Disease Control (CDC), developed an M&S model that would better assess the timing and efficacy of maternal vaccinations. This model could then be further employed to predict infant antibody (Ab) levels at birth. Currently, vaccination timing for infants six months and older is scheduled around routine checkups. Until 6 months of age, infants rely on transferred maternal antibodies, typically IgGs, from the placenta. This transfer depends on maternal antibody levels and fetal gestational age. The U.S. maternal vaccination policy set by the Advisory Committee on Immunization Practices (ACIP), a branch of the CDC, recommends that pregnant women receive a variety of vaccinations. Vaccination for influenza can be administered at any time during pregnancy. However, both maternal antibody levels and gestational age are time-dependent which suggests that “any time during pregnancy” may not be optimal.

Causal Chain of Events and Key Questions

To develop the vaccine-timing model, we established a causal chain of events and key questions that would facilitate the analysis to support vaccines similar to the critical thinking used for traditional MIDD.

Causal Chain of Events for Maternal Influenza Vaccination Timing Model

Once we understand what the infant Ab titer looks like, we want to map the protection. Waning Ab levels convey protection to infants. At some point, the anti-viral efficacy begins to decline – similar to the same type of analysis performed on a drug concentration pharmacodynamic effort. The vaccine provides less protection below a certain antibody titer threshold. Our ultimate goal is to understand this chain of events.

Leveraging Data from the Maternal Immunization Working Group

The Maternal Immunization Working Group is a large multinational group that conducted prospective studies to assess the efficacy of maternal vaccination using data from three different regions:  South Africa, Mali and Nepal. To train our model, we used the datasets from the South Africa and Mali trials, and the Nepal data was used for external validation.

The three geographies reported different timings of the vaccinations. That heterogeneity in vaccine timing allowed us to study its impact on differences in transferred IgGs to the infants. In addition, a population-approach analysis is well-suited to benefit from discrepant study designs.

How do we leverage the different literature and relatively sparse data including immunization and sampling schedules to develop the quantitative model that determines the optimal vaccination time? The clear differences in the protocol specification have windows of vaccination. Can we leverage the sparse data in thousands of mother-infant pairs to determine the optimal time to vaccinate the mother? We turned to our causal chain of events framework to work through the complex system and designed a structural vaccination timing model.

Influenza Vaccination Timing Model

The development of this M&S timing model offers an alternate approach to dedicated timing studies to determine the best time to vaccinate pregnant women. In a future blog, I will delve into our findings on how this model fit the existing maternal antibody data, determined the time to the production of maximal maternal antibodies, and ultimately predicted infant antibody levels at birth.

To learn more on how MIDD can be used in infectious disease drug development, read this article by my colleague Craig Rayner who describes how MIDD translational medicine strategies were used for RSV and Influenza drug development.

  1. S. Food and Drug Administration. FDA News Release: FDA expands Tamiflu’s use to treat children younger than 1 year. 21 December 2012. http://www.fda.gov/NewsEvents/Newsroom/PressAnnouncements/ucm333205.htm
Michael Dodds

About the Author

Mike Dodds has been working in the pharmacometrics field since 2005. He joined Certara (formerly Quantitative Solutions) in 2015 in the role of Director Consulting Services and is based in Seattle, Washington. Previously, he worked for ZymoGenetics in the Department of Pharmacokinetics and Pharmacodynamics, providing predictive modeling support for emerging drug candidates in the areas of autoimmunity, coagulation and oncology. Prior to that, he worked for Amgen in the Department of Pharmacokinetics and Drug Metabolism and then Clinical Pharmacology Modeling and Simulation as a pharmacometrics subject matter expert, providing modeling and simulation expertise to guide development program decisions around target selection, candidate selection, preclinical-to-clinical translation, Phase 1, 2 and 3 dose and regimen selection, and biosimilar development. Mike’s research focuses on the application of pharmacometrics: mathematical models of biology, (patho)-physiology, pharmacology and disease that quantify beneficial and undesirable interactions between drugs and patients to predict outcome [Barrett 2008]. Importantly, accurate and precise quantitative prediction of patient outcome allows for informed drug-development decisions. Mike has authored 20+ peer-reviewed articles, has presented at ACoP, ASCPT, PAGE, and holds two patents. He received his BS in Chemical Engineering and BS in Biochemistry from North Carolina State University, MS in Chemical Engineering from Montana State University, and PhD in Bioengineering from the University of Washington. In his free time, Mike enjoys making wine, playing tabletop games, creating large-scale, interactive, flame effect sculpture and exploring the beautiful Pacific Northwest.