Predicting Post-Treatment Control Before It Happens

Jeffrey Laurence, M.D., and Rowena Johnston, Ph.D.

Dr. Mohamed Abdel-Mohsen: One [model] predicted the probability of HIV reappearing following ART cessation with almost 98% accuracy.

Research question
There is a pressing need for identification of blood tests that could help predict the likelihood of HIV remission after stopping antiretroviral therapy (ART) in the setting of HIV cure trials. Ideally, such markers would not only improve the safety of a treatment interruption, they might also predict which individuals could control virus without ART and provide insight into how.

In a pilot study recently highlighted in a previous update, an amfAR research team had found several sugars and sugar-proteins in blood that appeared to predict the probability of rebound, as well as the time it took for such viral relapse to occur. But those studies involved small numbers of individuals and did not take into account a multitude of potential confounding variables, from a person’s age and sex to the timing and duration of their ART. The team, led by Dr. Mohamed Abdel-Mohsen of the Wistar Institute, extended the findings in additional individuals, and added analyses of metabolites and their interactions with immune and other functions.


Two groups of participants in HIV cure studies with treatment interruptions had blood samples tested before and at various points after ART interruption by this same research team. One was a cohort of 24 from Philadelphia; the other, 74 individuals from six NIH-supported clinical trial sites. Importantly, they included 27 participants known to control HIV after ART was withdrawn. Given the assumption that a single blood test would likely not have a high predictive value in terms of HIV rebound, machine-learning tools were also utilized to combine the data from almost two hundred blood markers.

Two best-fit combinations, or models, were uncovered. One predicted the probability of HIV reappearing following ART cessation with almost 98% accuracy, and the other estimated the time it took for such relapse with almost 75% accuracy. As an added benefit, these blood-based biomarkers, all involved with metabolic systems in the host, offered clues as to the mechanisms by which those metabolic systems may contribute to changes in the likelihood of, or delay to, viral rebound.


The authors concluded that “Our signatures have the potential to fill a major gap in the HIV cure field … providing means for selecting only the most promising therapies and most likely individuals to achieve viral remission to be tested by analytic treatment interruptions.”

amfAR’s role

amfAR was a funder of this research.  

Original article

Dr. Laurence is amfAR’s senior scientific consultant and Dr. Johnston is vice president and director of research.

Share This: