Challenges to Developing a Rotavirus Vaccine

Rotaviruses are the most important cause of gastroenteritis throughout the world. Symptoms include high fever, vomiting, and diarrhea progressing, in severe cases, to dehydration, shock, and death (15).

In developed countries, where standards of hygiene in the home and sanitation in the country are high, rotavirus is the most common cause of infant diarrhea (6). Virtually all children are infected by 2 to 3 years of age. In the United States, before the availability of rotavirus vaccines, rotavirus accounted for ∼2.7 million cases of gastroenteritis, 500,000 physician visits, 55,000 to 70,000 hospital admissions, and 20–60 deaths per year (18,23). The high rate of hospital admissions for dehydration is caused by the virus's propensity to cause diarrhea, fever, and, most importantly, frequent, severe, and persistent vomiting, making it difficult to rehydrate the child orally (35).

In developing countries, rotavirus infections are one of the most common causes of death in infants and young children. Before the availability of rotavirus vaccines, the World Health Organization (WHO) estimated that between 480,000 and 640,000 infants and young children died every year—about 2,000 children every day—from dehydration caused by rotavirus (1,28,37,40).

Although oral rehydration therapy in the developing world has helped to decrease the number of severe and fatal infections (1), the continuing mortality of rotavirus underlined the need for a safe and effective vaccine.

In 1978, Drs. Stanley Plotkin and Fred Clark, with funding from Pasteur-Mérieux-Connaught (PMC; Sanofi Pasteur today), established a rotavirus research team at the Children's Hospital of Philadelphia (CHOP). I joined that team in 1981. Dr. Plotkin was able to convince PMC to fund the project because rotavirus was an important disease worldwide, a rotavirus vaccine would likely be licensed for universal use in U.S. infants (making higher profits likely), and Dr. Plotkin had enormous credibility with vaccine makers, having successfully researched and developed a rubella vaccine and directed clinical trials of a human-cell-based rabies vaccine.

In 1983, several years after the rotavirus team at CHOP was formed, Bishop et al. provided the first evidence that a vaccine to prevent rotavirus was possible (5). Bishop found that children infected with rotavirus during the first month of life were protected against moderate-to-severe disease caused by reinfection; these children were not, however, protected against mild disease. Conversely, children not infected during the first month of life were susceptible to severe rotavirus gastroenteritis. Bishop's findings were later replicated in children infected during the first few years of life (3,16,38,41). The Bishop study set a limit on what could be expected from a rotavirus vaccine; a realistic goal would be to keep children out of the hospital and out of the morgue but not to prevent reinfection or even mild disease caused by reinfection.

Bishop's findings were consistent with previous observations that effector functions at mucosal surfaces are short lived; for example, rotavirus-specific secretory immunoglobulin A (sIgA) is not typically detected at the intestinal mucosal surface 1 year after symptomatic infection (14,25). Although mucosal, virus-specific sIgA at the time of exposure can protect against all rotavirus disease, protection against moderate-to-severe disease but not mild disease is most likely mediated by virus-specific sIgA produced by memory rotavirus-specific B cells in the intestinal lamina propria (26). Because it takes several days for memory B cells to be activated and to differentiate to antibody-producing plasma cells, modification and not complete protection against disease results.

Before the first rotavirus vaccine could be developed, several important questions about rotavirus pathogenesis and immunogenesis had to be addressed. These questions were answered using both large (calves, sheep, horses, and pigs) and small (mice) animal models. Researchers found the following:

• Species-specific strains of rotavirus infect the young of many mammalian and some avian species (32).

• Species barriers were high. For example, bovine rotaviruses caused diarrhea in calves but not in babies and human rotaviruses caused diarrhea in babies but not in calves (32).

• Rotaviruses replicated primarily in mature villous epithelial cells of the small intestine; although rotaviruses have been detected in the bloodstream, neither viremia nor replication at sites distant to the small intestine was an important part of viral pathogenesis (31). Therefore, protection against rotavirus disease depended on inducing an immune response active at the intestinal mucosal surface. Because sIgA resists degradation by acids and proteases, it is the critical immunoglobulin present at the intestinal mucosal surface. Monomeric IgG in serum, because it lacks secretory piece, is not transported through the polymeric immunoglobulin receptor—which is located at the base of villous epithelial cells—to the intestinal mucosal surface. Therefore, the goal of a rotavirus vaccine was to induce high titers of intestinal virus-specific sIgA.

• The most effective way to induce rotavirus-specific sIgA at the intestinal mucosal surface and high frequencies of virus-specific memory B and T cells in the intestinal lamina propria was by oral inoculation; parenteral inoculation was less effective (30).

• Although human and animal rotaviruses include a variety of distinguishable serotypes, studies in animals and children showed that heterotypic immunity existed (21).

For these reasons, our first attempt at a rotavirus vaccine centered on oral inoculation with a bovine rotavirus strain that was distinct from known human serotypes.

This source of our first rotavirus vaccine was a calf with diarrhea in Chester County, Pennsylvania in 1981. This virus was serially passaged 12 times in African green monkey kidney cells at The Wistar Institute. Because this was the third of several strains isolated, it was called Wistar Calf 3 (WC3) (13). In an initial double-blinded, placebo-controlled, efficacy trial performed in suburban Philadelphia, three doses of WC3 were given by mouth at 2, 4, and 6 months of age resulting in a 76% reduction in all rotavirus infections and 100% protection against moderate-to-severe disease (12). These findings were not, however, reproduced in subsequent efficacy trials conducted in Cincinnati and Bangui, Central African Republic, where protection against disease was insignificant (4,17). Heterotypic protection, which was induced by antibodies directed against outer capsid proteins, appeared to be inconsistent. Therefore, the WC3 vaccine was abandoned as a vaccine candidate.

It was back to the drawing board. The goal at this point was to determine which rotaviral genes coded for proteins that evoked virus-neutralizing antibodies and which rotaviral genes determined viral virulence. With this information in hand, the plan was to create a series of reassortant rotavirus strains that retained the attenuated virulence characteristics of WC3 (which had been shown to be safe in clinical trials) but did not include rotaviral genes associated with viral virulence. Again, animal models were used to determine the genetic basis of rotavirus virulence and rotavirus neutralization phenotype.

Animal model studies revealed the following:

• At least four rotavirus genes (VP3, VP4, VP7, and NSP4) determined viral virulence. Inclusion of all four genes was necessary to confer pathogenicity (20).

• Both rotaviral surface proteins (VP4 and VP7) independently evoked rotavirus-specific neutralizing antibodies (19,29). Rotaviruses, therefore, were similar to the influenza viruses, where two viral surface proteins determine neutralization phenotype. Rotavirus surface protein VP4 was responsible for viral cell binding. Because cleavage of VP4 by intestinal proteases was required for cell entry, VP4 serotypes were termed P types. Because rotavirus surface protein VP7 is a glycoprotein, VP7 serotypes were termed G types. Global surveillance has led to the characterization of at least 37 P genotypes and 27 G serotypes in humans and, because rotavirus has a segmented genome, gene reassortment could theoretically lead to almost 200 different P and G combinations. However, although >60 P-G combinations have been found in humans, five strains (P[8], G1; P[4], G2; P[8], G3; P[8], G4; and P[8], G9) are associated with 80% to 90% of the childhood rotavirus disease burden globally (32).

With this information in hand, five bovine–human reassortant rotavirus strains that included the attenuating genes of WC3 and either the P or G genes from prevalent human rotavirus serotypes were created. Unfortunately, at this stage, PMC abandoned the project and Dr. Plotkin left CHOP. Fred Clark and I then presented what we had done to scientists at three vaccine makers: Merck, Glaxo-Smith-Kline, and Wyeth. Merck was willing to evaluate our bovine–human reassortant rotaviruses as a vaccine. A formal working relationship was established in 1992.

Unfortunately, an event that occurred in 1998 dramatically changed the landscape of how future rotavirus vaccine trials would be conducted.

On August 31, 1998, the first animal–human reassortant rotavirus vaccine for use in infants was licensed in the United States. The vaccine (Rotashield), which was developed by researchers at the National Institutes of Health (NIH) in collaboration with Wyeth Laboratories, was given by mouth to infants at 2, 4, and 6 months of age and contained one simian rotavirus and three simian–human reassortant rotaviruses (7). The simian rotavirus strain, rhesus rotavirus (RRV), was similar to human serotype G3. The other three viruses were reassortants containing 10 genes from RRV and 1 gene (gene segment 8 or 9) from human rotaviruses of serotypes G1, G2, or G4. The choice of simian and simian–human rotaviruses for use as a vaccine for infants was based on inclusion of the attenuated virulence characteristics of the nonhuman strain RRV, and inclusion of human rotavirus genes that encoded surface protein vp7. Rotashield replicated less efficiently in the human intestine than in natural human rotaviruses, evoked virus-specific neutralizing antibodies against human G types 1–4, and was eventually shown to afford excellent protection against challenge in prospective, placebo-controlled studies (2,22,33,34,36). In August 1998, Rotashield was licensed for use in U.S. infants. Although Rotashield had been shown to be safe in a 10,000-child, prospective, placebo-controlled, prelicensure study, a rare adverse event appeared postlicensure (although several cases of this adverse event had also been seen both in the vaccine and placebo group prelicensure).

About 1 year after licensure, after Rotashield had been administered to about 1 million infants in the United States, 15 cases of intussusception after its use were reported to the Vaccine Adverse Events Reporting System (8). Intussusception occurs when a segment of the small intestine—typically around the ileocecal junction—telescopes into itself causing a blockage. When this occurs, severe intestinal bleeding or entrance of intestinal bacteria into the bloodstream can result in sepsis. In July of 1999, the Centers for Disease Control and Prevention (CDC) temporarily suspended the use of Rotashield, pending the results of a case–control analysis. Subsequent studies found that Rotashield was a rare cause of intussusception in children—about 1 case of intussusception for every 10,000 children vaccinated (24,27).

Because Rotashield was found to cause intussusception, both the CDC and the American Academy of Pediatrics withdrew their recommendation in October 1999 (9).

Although Rotashield was withdrawn from use, several investigators argued, reasonably, that even in U.S. children natural infection was still far more likely to cause hospitalization and death than immunization. Nonetheless, the manufacturer withdrew Rotashield for U.S. use.

More tragic was the loss of Rotashield for the developing world. Because the United States was no longer using Rotashield, health officials in the developing world argued that if the vaccine was unsafe for U.S. children, then it was also unsafe for their children, even though the benefit-to-risk ratio was dramatically different. As a result, Rotashield, a vaccine that had the capacity to save as many as 2,000 lives a day, went unused. Seven years passed before the next rotavirus vaccine became available.

The cause of intussusception after administration of Rotashield remains unknown.

The problem with the first rotavirus vaccine did not end attempts to make a safer vaccine, but it did increase the size and expense of prelicensure trials. Now the task was not simply to rule out relatively uncommon side effects prelicensure, but rather to rule out rare events prelicensure. Merck struggled with the cost of this development project, which would soon exceed $1 billion. The argument that most persuaded Merck at this point, when it became clear that a Phase III trial would be large, was that Glaxo-Smith-Kline, a competitor, was also moving forward with its own rotavirus vaccine (a live attenuated human strain). Another limitation and expense to making a rotavirus vaccine was that no clear immunological correlate of protection had been found. Neither virus-specific binding nor neutralizing antibodies in the serum or at the intestinal mucosal surface was predictive of protection. Based on animal model studies, the correlate most likely predictive of protection was the frequency of virus-specific B and T cells in the intestinal lamina propria (26). However, these data could have only been provided by performing intestinal biopsies on otherwise healthy children—an impossibility.

Although it took about 10 years to do the basic science research required to construct the bovine–human rotavirus strains that became a vaccine, it took 16 years to do the research of development, which included small Phase I and Phase II studies for safety and efficacy to prove that each of the strains needed to be in the final formulation (proof-of-concept studies), to determine the correct dose of each strain (dose-ranging studies), to prove that the shelf life of the vaccine was compatible with distribution in the developing world (real-time stability studies), and to design a convenient plastic squirt vial that allowed for easy inoculation of a fully liquid product (32). One of the biggest hurdles was deciding how to mass-produce the vaccine. GlaxoSmithKline's (GSK's) microcarrier technology was probably the most efficient way to mass-produce live attenuated rotavirus vaccines. Merck did not have this technology, eventually settling on Nunc cell factories. This, too, was one of the many go-no-go decisions that stalled the vaccine.

The Merck bovine–human rotavirus vaccine, RotaTeq, is a live, oral vaccine that contains five bovine–human reassortant strains. The bovine strain WC3 is the backbone virus for the vaccine. The four human strains reassorted into the WC3 backbone include human serotypes G1, G2, G3, or G4 (derived from outer capsid protein VP7) as well as P1A (derived from outer capsid protein VP4) (11).

The efficacy of RotaTeq against all rotavirus gastroenteritis was evaluated in a Phase III trial of >70,000 infants (39). This trial, which was a placebo-controlled, 11 country, 4-year, prospective study, cost about $350 million. The original study, which was designed in collaboration with the Food and Drug Administration, was a 40,000-child, placebo-controlled trial. But intussusception is a relatively rare event. Background rates are low. Therefore, the trial continued, adding increments of 10,000 children beyond the original 40,000 until statistical differences between the incidence of intussusception in the vaccine and control group were reached. During this time, as 10s of millions of dollars were added to the expense of the trial, Merck struggled to decide whether it wanted to continue to pursue this vaccine.

RotaTeq was effective. Protection against rotavirus disease of any severity was 74%, against severe rotavirus disease 98%, against rotavirus-associated hospitalizations 96%, against emergency visits 94%, against office visits 87%, and against hospitalizations of any etiology 59%; the last finding proving the importance of rotavirus as a cause of severe gastroenteritis in children. One year after vaccination, the efficacy of RotaTeq was 63% against all rotavirus diseases and 88% against severe disease (39).

The large Phase III trial also showed that RotaTeq neither caused nor prevented intussusception. Within 14 days of inoculation, 1 case occurred in the vaccine group and 1 in the placebo group; within 42 days of inoculation, 6 cases in the vaccine group and 5 in the placebo group; and within 1 year of inoculation, 12 cases in the vaccine group and 15 in the placebo group (39).

In February 2006, the Advisory Committe on Immunization Practices (ACIP) recommended routine immunization of U.S. infants with three doses of RotaTeq to be administered by mouth at 2, 4, and 6 months of age (7). Since licensure, rotavirus vaccines have caused an approximate 90% decrease in the incidence of clinically significant rotavirus disease in the United States (10). Furthermore, the introduction of rotavirus vaccines into the United States has not increased the overall incidence of intussusception. Globally, as of May 2016, 81 countries, including 38 low-income countries, have implemented rotavirus vaccines as part of their national immunization programs.