The best cooks know that when you steam vegetables, you shouldn't throw out the broth.
That time-worn kitchen advice may be equally helpful to scientists who develop vaccines, especially if the vaccines target highly variable bugs like HIV, the virus that causes AIDS.
The ultimate goal of a vaccine trial is, of course, to demonstrate high levels of vaccine protection against infection. But those individuals who slip through the cracks and get infected despite vaccination play a valuable role in the development of new vaccines.
Thanks to an analytical tool known as sieve analysis, developed in part by Dr. Peter Gilbert of the Hutch's Statistical Center for HIV/AIDS Research & Prevention (SCHARP), researchers can extract important clues from vaccine trials, even if the vaccines aren't entirely successful. Because sieve analysis identifies "holes" in vaccines that compromise their effectiveness, scientists can design new versions with increased breadth of protection.
Big challenge of diversity
Gilbert said the diversity of HIV, his research specialty, is one of the biggest challenges to vaccine design. He developed sieve analysis to enable researchers to identify the distinguishing properties of viruses that escape the immunization barrier.
"My research focuses on analysis of the characteristics of viruses that still manage to infect people in vaccine trials," he said. "The idea is to compare the characteristics of viruses that infected vaccine recipients with those that infected individuals who received a placebo."
In the case of HIV, strains may differ from one another because they provoke unique immune responses or use different host receptors as portals to sneak inside cells.
The immunogens (components that provoke an immune system response) in the vaccine that fail to induce protection against genetically divergent HIV strains are like holes in a sieve, said Gilbert, who joined the Public Health Sciences Division last year.
"We expect that some viruses will make it through by eluding the immune response," he said. "We want to figure out where the holes are in the vaccine and come up with a picture of the kind of virus the vaccine works against."
Scientists can take that information back to the laboratory to reformulate the vaccine and plug the holes.
"Hopefully," Gilbert said, "the sieve will get tighter."
Effective vaccines trigger protective immunity by challenging the body's infection-fighting system with weakened versions of the pathogens or snippets of their coats or other components. Such simulated "infection," when successful, readies the immune system to ward off future attacks by the actual pathogen. But when microbes are highly variable, immunity may be limited to a single infectious strain.
Vaccines against smallpox and polio were successful because the viral genomes contain regions that are similar in all strains, enabling vaccines to be designed that act against these so-called conserved regions, Gilbert said. In contrast, conserved regions of the HIV genome don't represent parts of the virus that interact with the immune system. But if an immune response could be generated against viral components in conserved regions, then vaccine protection against a broad array of strains might be feasible.
"The early HIV vaccines have been against variable regions because these are the parts of the virus that provoke an immune response," he said. "Many of the conserved regions are well protected, buried within the virus."
Gilbert began work on sieve analysis while he was a University of Washington graduate student working with SCHARP director Dr. Steven Self, who is also a UW professor of biostatistics. As described in a paper they published together last year, sieve analysis can be applied to vaccine trials for variable pathogens besides HIV, including those that cause cholera, malaria and hepatitis.
One of their key points was that data from animal studies and small-scale, early human studies are insufficient for identifying the attributes of HIV that are critical for designing broadly protective vaccines. They argue that only data from efficacy trials, which are larger comparative studies, allow this direct assessment and should serve as motivation for such trials to be conducted.
In his new position at SCHARP, Gilbert will take part in a multitude of HIV prevention studies, including some that are non-vaccine based.
One ongoing prevention study of mother-to-child HIV transmission involves 1,200 HIV-positive women in Botswana, where about 40 percent of the population is infected with HIV, he said. Women are being randomized to either breast-feed their infants accompanied by feedings of AZT syrup or to feed infants with formula.
Info to refine future trials
Another project is to analyze a variety of parameters in individuals who become infected with HIV while enrolled in Phase I or II vaccine trials (such studies are small trials designed to examine safety and dosage of a new vaccine or drug and to make preliminary comparisons of vaccine candidates). This information will help refine future vaccine trials.
"Trial participants who get infected roll over into another study, where we look at white blood-cell counts, viral genetics and other virological, immunological and clinical parameters," he said.
Gilbert will also take part in design, operations and interpretation, which will include sieve analysis, of large-scale vaccine efficacy trials that may begin in the next few years.
Prior to joining the Hutch, Gilbert was a postdoctoral fellow and then an assistant professor in the department of biostatistics at Harvard University, where he worked closely with the Harvard AIDS Institute. During that time, he traveled frequently to Thailand, where he taught workshops to train statisticians in clinical trials and management.
He also collaborates with scientists in Botswana to conduct vaccine trials there. Last December, Gilbert and Self traveled to Botswana for the opening of a new AIDS research lab.
Gilbert was especially pleased to accept the position at SCHARP to be near his close collaborator Self and part of a stimulating research environment. "I'm impressed with the infrastructure here," he said. "Computer and data managers, operations people and statisticians are together in one place. There's a real excitement here about the research."