Scientists overcome missing data to demonstrate ART effectiveness in HIV-infected infants

Now, scientists from The Wistar Institute, in collaboration with the Center for Interdisciplinary Research on AIDS at Yale University, have demonstrated that the issue of missing data can be successfully overcome using appropriate statistical methods, and as a result, they were able to show how early initiation of ART in infants preserves an expansion of naïve T-cells and allows the infant's immune system to be properly reconstructed.

The results of the study were published in the journal PLOS ONE.

Children born with HIV who do not receive treatment with ART experience progressive immunodeficiency that can lead to death, and if they do survive, their immune cell levels can be permanently offset. These children have low levels of naïve CD4+ T-cells and other immune effectors. Studying the levels of immune cells in these infected infants could help scientists develop better treatment strategies.

Scientists struggle with the frequent issue of missing data when it comes to studying these infants. Blood samples taken from these infants are routinely used for monitoring their health and response to treatment. However, it is often difficult to collect larger volumes of blood for immunological studies. Additionally, in resource-constrained settings, many infants miss visits because their caregivers may be unable to get them to a central location where samples can be routinely taken.

"Despite the best efforts of pediatricians and pediatric nurses, insufficient samples and missed visits have been the norm for pediatric studies in developing countries," said Luis J. Montaner, D.V.M., D.Phil., Director of the HIV-1 Immunopathogenesis Laboratory and Herbert Kean, M.D., Family Professor at The Wistar Institute. "Loss of data lead to loss of statistical power, so it's extremely important to develop methods that allow us to analyze data sets where data are randomly missing."

To solve this issue, Dr. Montaner teamed with Dr. Russell Barbour from the Center for Interdisciplinary Research on AIDS at Yale University and an international team of scientists to determine which methods could successfully overcome data missingness in a dataset collected in the course of a study on HIV-infected infants. While numerous methods have been suggested to address this issue, many have intrinsic problems like not being able to anticipate change over time and artificially reducing standard error. The team settled on two approaches that they believed would effectively address missing data. The first, called the Multiple Imputation (MI) method, uses the observed values as well as different imputations -- values that are substituted for missing data -- to fill in these gaps while not changing anything else about the data. The second approach was based on a Bayesian model, which was used to create 5,000 "alternative" instances of the dataset based on the observed data.

A total of 66 HIV-infected or seronegative infants born of HIV-infected mothers were recruited at the Chris Hani Baragwanath Hospital in Soweto, South Africa. These infants were enrolled in the Children with HIV Early Antiretroviral Therapy (CHER) trial and randomized to receive ART either as soon as they were enrolled or when their CD4+ T-cell count dropped below 20 percent. Blood samples were taken once every six months.

The use of these statistical methods allowed Dr. Montaner and his collaborators to demonstrate that early ART treatment results in higher CD4+ T-cell frequency, lower cellular activation, and had higher proportions of naïve T-cells. These results suggest that in infants with perinatal HIV infection early treatment may improve the chances of developing a functional immune system.

"Our study offers a field-based proof of concept that certain type of data missingness can be tolerated without affecting the integrity of a study," Montaner said. "We hope this will encourage other scientists to target hard-to-reach populations, particularly in resource-constrained settings."

This work was supported by the National Institutes of Health grant AI062512. Additional support was provided by The Philadelphia Foundation (Robert I. Jacobs Fund), the Stengel-Miller family, and AIDS funds from the Commonwealth of Pennsylvania and the Commonwealth Universal Research Enhancement (CURE) Program.

Other co-authors of this study from The Wistar Institute include Livio Azzoni and Emmanouil Papasavvas. Other co-authors of this study include Deborah Glencross and Wendy Stevens from the University of the Witwatersrand and National Health Laboratory Service in Johannesburg, South Africa; Mark Cotton from the Department of Paediatrics and Child Health at Stellenbosch University in Cape Town, South Africa; and Avy Violari from the Perinatal HIV Research Unit at the University of the Witwatersrand.