Editor's note: This is the third in a seven-part series investigating the prospects for development of vaccines to head off the threat of an influenza pandemic posed by the H5N1 avian influenza virus. The series puts promising advances in vaccine technology in perspective by illuminating the formidable barriers to producing large amounts of an effective and widely usable vaccine in a short time frame. Part 2 discussed the huge gap between current global vaccine production capacity and the likely demand for vaccine in the event of a pandemic.
Oct 29, 2007 (CIDRAP News) – Many of the difficulties facing achievement of a pandemic influenza vaccine could not have been anticipated before the pandemic threat arose: They are intrinsic to the H5N1 virus itself.
Some are obvious. Because all flu viruses mutate rapidly, antigenic drift and division into clades, or subgroups, has already rendered some early vaccine candidates less potent against circulating strains (see Bibliography: Smith 2006, Riley 2007). Because this flu virus is highly pathogenic, it must be handled initially in one of a relatively small number of high-biosecurity laboratories (see Bibliography: WHO 2004) and also requires the use of reverse genetics to create a seed strain that will reproduce in eggs. Reverse genetics and the stringent pre-release testing that follows add a minimum of 6 weeks to the vaccine production process (see Bibliography: Wood 2007: Reference viruses). And absent regulatory changes, the resulting seed strain could be considered a genetically modified organism under European Union rules or a select agent under US law, further restricting the laboratories, personnel, and manufacturers who could work with it from then on (see Bibliography: Stephenson 2006: Development and evaluation).
Reverse genetics may be responsible for a phenomenon noted by several researchers: H5N1 does not reproduce well in vitro, a potentially major obstacle for vaccine production because it would greatly lengthen the manufacturing timeline. "Most manufacturers report that yields of antigen from reverse genetics–derived H5N1 viruses are 30-40% of the average of seasonal influenza viruses, reducing the quantity of antigen available for vaccine formulation," Iain Stephenson of University Hospital-Leicester and colleagues wrote last year in Lancet Infectious Diseases (see Bibliography: Stephenson 2006: Development of vaccines).
Vaccines show poor immunogenicity
Furthermore, when candidate vaccines have been produced from them, H5N1 and similar viruses have not done a good job of provoking an immune response. This was noted as early as 1998, when an early attempt to create a vaccine in the wake of the 1997 Hong Kong outbreak used an antigenically similar H5N3 strain—only to find it was poorly immunogenic even at the highest concentration of two 30-microgram (mcg) doses (see Bibliography: Nicholson 2001). There were similarly poor results in 2006 and 2007 trials that used an H5N1 strain isolated from Vietnam in 2004 and modified by reverse genetics (see Bibliography: Bresson 2006, Leroux-Roels 2007): Even at the highest dosage given (two 30-mcg doses), neither trial induced levels of immune protection that would be acceptable to the Food and Drug Administration (FDA) or the European Union's Committee for Medicinal Products for Human Use (CHMP).
But the best example of H5N1's poor immunogenicity is the 2006 trial that led to the first FDA licensing of an H5N1 vaccine. The trial, which used the same modified 2004 Vietnam strain as the others, achieved acceptable levels of protection only at the highest amounts given, two doses of 90 mcg each, or 12 times the 15-mcg dose that induces immunity in a seasonal vaccine (see Bibliography: Treanor 2006: Safety and immunogenicity of an inactivated subvirion influenza A [H5N1] vaccine). Even at that dosage, only 54% or 58% of the subjects (by two different measures) exhibited antibody titers that matched FDA and CHMP regulations, compared with the 70% to 90% usually achieved with seasonal vaccine (see Bibliography: CDC 2005).
The trial's investigators acknowledged that the extraordinarily high dose necessary to protect was an unsatisfying result, saying, "The need for a vaccine with a total dose of 180 mcg would pose a considerable barrier to rapid production of a supply that would be adequate to meet the world's requirements should a pandemic occur" (see Bibliography: Treanor 2006: Safety and immunogenicity). Experts who reviewed the data during FDA deliberations on licensure agreed, signaling that they hoped for improved results as research advances. "There are numerous vaccines under development that are potentially better, if you will, than this vaccine. This is an interim vaccine," Norman Baylor, PhD, director of the FDA's Office of Vaccines Research and Review, said during the FDA hearing (see Bibliography: FDA 2007: Committee meeting transcript).
The amount of antigen needed in that vaccine concerns researchers several times over. It is so high that the vaccine would stress the manufacturing system if put into broad production, although regulators at the licensure hearings said the vaccine is intended only for federal stockpiling and not for commercial sale. In addition, they fear the high dose—which at 180 mcg total is four times the total seasonal trivalent dose—could provoke an unusual rate of adverse reactions (see Bibliography: FDA 2007: Committee meeting transcript).
(Regulators may be quietly bracing for adverse publicity as well. Testimony at the same FDA committee meeting indicated that, because of limitations on fill/finish capacity, the 90-mcg vaccine being manufactured for the US stockpile will be packaged in multi-dose vials which, to guard against contamination, will contain the still-controversial preservative thimerosal.)
Immunity hard to measure
The 2006 trial's unsatisfying results highlighted a chronic concern in flu-vaccine research: the lack of reliable correlates of immunity for pandemic vaccines. Given the lack of a vaccine, moderate antiviral resistance, and a case-fatality rate of more than 60%, humans cannot ethically be experimentally exposed to the H5N1 virus. Yet none of the animal models—mice, ferrets, or the recently proposed guinea pigs (see Bibliography: Lowen 2006)—is a perfect substitute. Hence the research community relies instead on measures of human antibody response that are neither uniform across laboratories nor universally agreed to by regulatory bodies.
"I think this is our biggest scientific issue—that we are not sure what the appropriate surrogate for protection is, given the fact we have no ability to challenge [expose humans] and are not likely to," said flu epidemiologist Dr. Arnold Monto of the University of Michigan (see Bibliography: Monto 2007).
In the United States, the FDA-accepted surrogate for immunity in flu-vaccine trials is a hemagglutination-inhibition (HI) antibody assay that returns a post-vaccination titer of more than 1:40 for 70% of those vaccinated (see Bibliography: FDA 2007: Guidance for industry: clinical data needed to support the licensure of seasonal inactivated influenza vaccines). But that measure is known to be imperfect even for seasonal flu: Patients with higher titers have contracted flu, while those with lower post-vaccination titers have apparently been protected (see Bibliography: Poland 2006). Additionally, the HI test appears in the lab to be less sensitive to H5 antibodies than it is to the seasonal strains H1 and H3; a second test, virus microneutralization, appears more sensitive but also has no agreed-upon clinical correlates (see Bibliography: Stephenson 2004). Researchers have resorted to using a 1:40 result as the best available measure of immunologic response, while conceding that it may not indicate actual degrees of protection.
"It's important to understand that this choice of a 1:40 endpoint is not validated in any way as an actual assessment of protection against H5 in humans," Dr. John Treanor of the University of Rochester, principal investigator for the trial that produced the licensed H5 vaccine, said at the FDA licensure hearings. "It might be just as valid to choose a 1:20 or a 1:80 or a 1:10 endpoint. But it's really more a convenient way in order to discriminate responses between groups" (see Bibliography: FDA 2007: Committee meeting transcript).
Unfortunately, HI titer results vary wildly. In two studies, European labs performing the same test for seasonal flu strains returned results that varied from 16- to 128-fold, and for H5N1, labs have achieved different results depending on the source and age of the animal cells used in the assay (see Bibliography: Wood 2007: International standards). The WHO is pursuing international agreement on a standard for H5N1 tests, similar to standards achieved earlier for measles, polio, rubella, and other infectious diseases.
But the problem with finding correlates of immunity goes further. The HI test may not return reliable results for vaccines that provoke types of immunity other than antibody response. That makes it an unreliable measure for the effectiveness of two promising alternative classes of vaccines: live-attenuated vaccines, which have the potential to generate immunity against multiple strains of flu, and inactivated adjuvanted vaccines (those containing a chemical immune-system stimulant), which could help solve the supply bottleneck by allowing much smaller amounts of antigen to be dispensed in each vaccine dose (see Bibliography: Subbarao 2007, Wood 2007: Author interview).
Part 1: Flu research: a legacy of neglect
Part 2: Vaccine production capacity falls far short
Part 3: H5N1 poses major immunologic challenges
Part 4: The promise and problems of adjuvants
Part 5: What role for prepandemic vaccination?
Part 6: Looking to novel vaccine technologies
Part 7: Time for a vaccine 'Manhattan Project'?