Mathematical modeling can be a powerful tool for taking a systems-based approach to understanding virus-host interactions and simulating clinical trials that test therapeutic efficacy. Drs. Elizabeth Duke, E. Fabian Cardozo-Ojeda, and Joshua Schiffer from the Schiffer Lab aimed to model natural Cytomegalovirus (CMV) infection in bone marrow transplant recipients and published their findings in a recent article in Viruses.
CMV is a slow replicating herpesvirus that is of particular concern for immunocompromised individuals, including bone marrow transplant recipients, by causing gastroenteritis, pneumonia, and retinitis. Ganciclovir is an effective, first line antiviral therapy for CMV infection, although it is not without a level of toxicity. One side-effect of ganciclovir is bone marrow suppression, which is a concern especially for bone marrow transplant recipients. For this reason, understanding the viral kinetics of CMV in bone marrow transplant recipients is of great importance.
To investigate natural CMV infection kinetics in immunocompromised patients, the researchers took a modeling approach based on virus abundance measurements from blood samples of bone marrow transplant recipients. The blood samples provided a unique opportunity to determine viral kinetics in individuals not treated with antiviral therapies, thus representing a natural infection in bone marrow transplant recipients. Blood samples from at least three different time points were available for 33 patients over a 100-day timeline. The amount of virus detected from the patient samples was highly variable across individuals but CMV growth kinetics in each individual fit into four main groupings: rapid growth, initial rapid growth that later slows, growth and partial clearance, and growth and complete clearance. CMV DNA abundance declined by half every 2.5 days in patients who cleared the virus, supporting slow viral clearance kinetics in untreated patients as compared to data published for ganciclovir treated patients with a 1.5-day rate of decline. This initial calculation from the virus abundance measurements in patient blood samples is informative on its own by defining the natural rate of CMV clearance for this patient population, which is needed to evaluate drug efficacies.
To understand the infection kinetics of CMV, 38 models were generated and evaluated using model selection theory that tests each mathematical model for agreement with actual data and chooses the model that represents the data with the greatest accuracy. Specifically, the Akaike Information Criterion was used to determine the likelihood that the model will agree with the actual data; the relative standard error was calculated to evaluate which parameters in the mathematical model are able to be “identified”, meaning there is enough data to inform the value of the parameters independently from one another. From the models tested, two sufficiently aligned with the actual virus abundance data measured over the duration of infection in the patient population. Due to the high variability of CMV abundance in patient blood samples, a model reflecting a dynamic immune response against infection was most suitable in aligning with the patient data. Models that did not fit the data well included those that assumed that CMV had limited target cell availability for infection. This outcome is logical as CMV can infect most cell types and therefore cell availability would not be a limiting factor in this system. For the best model selected using model selection theory, the model predicted a CMV half-life decay rate of 2.6 days, which closely mirrors the actual decay rate. The researchers were also able to deduce from these models an estimated doubling time of 1.8 days for CMV infection in bone marrow transplant recipients. While all models come with limitations in data interpretation, the researchers propose that the baseline model can be used to estimate the efficacy and dosing of CMV-targeting therapies for bone marrow transplant recipients.
Lead author Dr. Duke stated that they will use this mathematical model to “estimate the efficacy of ganciclovir at different doses and simulate dosing earlier or later.” Additionally, this model of CMV growth kinetics may allow for pre-clinical evaluation of the next line of anti-CMV therapies specific for bone marrow transplant recipients. With a few antiviral drugs available with lower toxicity than ganciclovir, Dr. Duke commented that they “are interested in whether combination therapy might benefit transplant recipients and could simulate combination therapy trials with the [mathematical] model.”
The research spotlighted in this article was funded by Merck Sharp & Dohme Corp., a subsidiary of Merck & Co., Inc., and Fred Hutchinson Cancer Research Center Vaccine and Infectious Disease Division (Biorepository).
UW/Fred Hutch Cancer Consortium member Michael Boeckh contributed to this work.
Duke ER, Boshier FAT, Boeckh M, Schiffer JT, Cardozo-Ojeda EF. 2021. Mathematical Modeling of Within-Host, Untreated, Cytomegalovirus Infection Dynamics after Allogeneic Transplantation. Viruses. 13(11):2292. doi: 10.3390/v13112292.