Identifying biomarkers for HIV prevention with Ying Huang

Vaccine and Infectious Disease Division

Identifying biomarkers for HIV prevention with Ying Huang

Ying Huang

Dr. Ying Huang did her graduate studies at UW and a postdoc at Fred Hutch. After that she worked as an assistant professor of Biostatistics at Columbia in NYC between 2009 and 2010 and returned to Fred Hutch in 2010.

Biomarkers, objects that indicate something more complex and hard to measure, can be used to diagnose and/or predict diseases. In randomized HIV vaccine clinical trials, a certain immune response (the biomarker) is measured post randomization and the question is whether the vaccine’s effect on those immune responses can predict the vaccine’s effect in preventing HIV infection. Those immune responses, also known as surrogate markers, can be used to guide the development of efficacious vaccines. For more than a decade, the HVTN and SCHARP have been dedicated to ending the HIV pandemic with a successful vaccine.

“We use statistical methods to identify and develop biomarkers that will be useful in prevention and control of diseases such as infectious diseases and cancer,” said Dr. Ying Huang, associate member of VIDD, who hopes to identify biomarkers that correlate with vaccine efficacy against HIV acquisition.

While the recent RV144 clinical trial resulted in partial efficacy, it provided a wealth of information for follow-up studies to uncover potential mechanisms of protection. The presence of two types of IgG antibodies and absence of one IgA antibody were associated with reduced acquisition of HIV in vaccinated subjects compared to those who received placebo. These biomarkers are a huge step in designing even better vaccines to improve efficacy. However, because the human immune system and HIV pathogenesis are so complex, designing an efficacious vaccine is anything but simple.

“We now know of many biomarkers for numerous immune responses: how do we best select them?” Huang continued, “How do we combine all this high dimensional information to develop rules that can allow us to predict risk of HIV?”

Down Selection Project

Before a large scale vaccine clinical trial, the many attributes that make up the vaccine regimen must be fully optimized. Examples of these components include the antigen(s) for which the immune system responds to, the delivery backbone to use, the numbers boosters to give, and the population to test. Assuming you can choose from 1-4 different antigens, 1-2 backbones, and up to 3 booster combinations, the total number of regimens (or ‘arms’ of a study) quickly becomes exponentially unfeasible. Sometimes even slight changes in a vaccine molecular design or booster schedule can drastically alter the outcome/efficacy. Because scientists cannot test every possible combination in a human vaccine trial, there must be a system for determining which combinations to move forward into the next stage. 

Down Selection Project

Figure legend. Integrated methods for addressing down selection criteria. In down selection, we are looking for regimens that stimulate superior immune responses and also differ from one another; i.e., the immune responses they generate are unique. We integrate three different statistical methods including hypothesis testing, clustering, and ranking to select regimens that satisfy these criteria.

Figure provided by Y. Huang.

“We definitely want vaccine regimens that generate superior immune responses. And because we only have so much money,” Huang said, “we don’t want regimens that are redundant. They should elicit unique immunogenicity patterns.”

The HVTN will launch a phase II trial in the near future that will compare 1-3 candidate regimens designed to stimulate robust immune responses to HIV risk reduction. They are currently selecting which regimens to use based on approximately 15 phase I immunogenicity studies and will have to narrow that number down based on how robust, sensitive and specific the immune profile biomarkers are for each regimen. Huang is developing statistical methods for determining which vaccines to push forward. These statistical techniques involve inputting the vaccine regimen parameters into a computer program and running simulations that in essence identifies how useful that particular regimen will be in predicting the vaccine’s ability to reduce HIV infection.

“In the real world and practically speaking, resources such as money and the number of trial participants are limited. The availability of high throughput, complex statistical programs offers us a feasible route to  find the best vaccine,” said Huang.

In addition to biomarker statistical methods, Huang currently studies causal interference, personalized treatment selection and genome wide association research. She is also an associate member of the Public Health Sciences Division and an affiliate associate professor of Biostatistics at the UW. 

Ying Huang Faculty Profile


How to reach VIDD News: