There are few biological agents that have a greater impact on childhood morbidity and mortality than rotaviruses. They are the primary cause of severe diarrhoea in young children and are responsible for an estimated 500 000 deaths worldwide in this population every year.1–3 Because nearly every child in the world experiences at least one rotavirus infection by the time they reach 3 years of age, the range of rotavirus disease is universal. Although children in developed nations typically do not die as a result of rotavirus infections, most cases of severe diarrhoea necessitating hospitalizations in these countries were previously caused by rotaviruses. This has changed in the past few years with the introduction and wide usage of new rotavirus vaccines in many nations. For example, the Centers for Disease Control has reported that, in the USA, as a result of the introduction of new rotavirus vaccines, the annual number hospital admissions due to rotavirus in children younger than 4 years plummeted by 80% by 2010.4 The use of these vaccines also resulted in herd immunity, in that rotavirus hospitalizations have decreased by 70% among children between 5 years and 14 years of age who were too old to have received the rotavirus vaccines when they were introduced between 2006 and 2008.3,5 Although most infants born in Latin America are being vaccinated against rotavirus, there has been a slow uptake of these vaccines into national immunization programmes in much of the developing world. This landscape is slowly changing, inspired by consistent evidence from multiple epidemiological studies that rotaviruses are the primary cause of fatal severe diarrhoea in young children everywhere in the world. It is now estimated that, within the next few years, 240 000 deaths due to rotavirus could be prevented each year through the use of rotavirus vaccines.
The first rotaviruses to be described were murine strains that were classified under the general description as the agents responsible for epizootic diarrhoea of infant mice.6,7 These agents were described as 70-nm particles that had a wheel-like appearance and were later designated ‘rota’ viruses from the Latin word for wheel. The correlation between these viruses and severe diarrhoea in young children was first reported in 1973 by Bishop et al.,8,9 who used electron microscopy to examine biopsy specimens of duodenal mucosa from children with acute gastroenteritis. Shortly after these studies, other investigators confirmed the association between the presence of rotavirus in faeces and acute gastroenteritis.
Rotavirus is a double-stranded RNA virus with 11 gene segments. There are six structural proteins – VP1–VP4, VP6 and VP7 – and six non-structural proteins, NSP1–NSP6.10,11 Each segment encodes one rotavirus protein, except segment 11, which encodes both NSP5 and NSP6.12 The two outer capsid proteins, VP7 and VP4, contain the only neutralization epitopes for this virus. Early studies using sera from hyperimmunized animals in cross-neutralization assays described a number of different serotypes from infected humans and animals based solely on the VP7 protein. Viruses are assigned G serotypes that refer to the glycosylated structure of this VP7 protein. VP7 serotypes are determined by cross-neutralization and by using panels of monoclonal antibodies;13–15 however, VP4 serotype determination is more difficult.13–18 Two numeric systems were established to classify the VP4 protein or P-type referring to the protease sensitivity of this protein. The P serotype is based on neutralization assays using antisera against recombinant expressed VP4 proteins or viruses with specific VP4 genes.19,20 A second classification system, one that is now more widely used to characterize rotavirus isolates, is based on sequence analyses and rotaviruses are described as different genotypes.18–22 The P serotype is indicated by an open number and the genotype is denoted with a bracketed number. For example, the most common P-type worldwide belongs to serotype P1A and genotype 8 and is therefore designated P1A. Currently, rotaviruses are most often classified only by genotyping for the P protein, i.e. P. Rotaviruses, similar to other viruses with segmented genomes, have the ability to form reassortants. This ability to reassort genome segments has contributed to the diversity of G- and P-types found throughout the world.
Infection with rotavirus often results in more severe form of illness than other pathogens and, therefore, before the use of vaccines, rotavirus accounted for a higher percentage of gastroenteritis episodes requiring medical intervention.20–24 During peak rotavirus seasons, approximately 70% of all gastrointestinal hospitalizations in the USA were due to rotavirus-associated gastroenteritis25 and rotavirus illness resulted in 20–70 deaths annually.26–28 It is estimated that one child in seven required a doctor’s visit at a clinic or emergency room and about 1 in 75 were hospitalized because of a rotavirus illness.27,29 Based on these estimates, rotaviruses were responsible for 5–10% of all gastroenteritis episodes in children under 5 years of age, leading to over 50 000 hospitalizations. Similar estimates of the disease burden due to rotavirus have been reported in studies conducted in European countries.30–32
Globally, rotavirus disease has an even more dramatic and significant impact on infant health. Although rates of rotavirus illness among children are similar throughout the world, the resulting mortality differs substantially. As already noted, it is estimated that rotavirus illness is responsible for approximately 500 000 deaths annually, representing 5% of all deaths in children younger than 5 years of age worldwide.33 More than 90% of these deaths occur in Africa and Asia, and more than 100 000 deaths occur in India and sub-Saharan Africa and 35 000 in China.33–35
History of rotavirus vaccines
The history of human rotavirus vaccines began within a few years after the discovery of human rotaviruses in 1973. The initial vaccines were developed from bovine rotavirus strains because it was believed that they would be naturally attenuated for humans. Although the protection against rotavirus disease elicited by these vaccines during trials in infants was insufficient to warrant their further development, attempts to find an animal rotavirus strain that would elicit satisfactory protection when administered to infants continued. After multiple trials with a simian rotavirus called rhesus rotavirus (RRV), it was determined that consistent protection with this animal strain required that it be modified to contain human rotavirus genes. Thus, a single gene from three different serotypes of human was incorporated into RRV by reassortment, creating three new RRV strains belonging to human rotavirus serotypes. These three strains, together with RRV itself, became the tetravalent RRV vaccine later marketed as Rotashield.
In less than 1 year after its introduction into the childhood immunization programme in the USA in 1998, evidence was obtained by the Centers for Disease Control and Prevention that Rotashield produced a small excess of intussusception (IS) cases in vaccinees. With this evidence, Rotashield was removed from the marketplace and its use was never extended beyond the borders of the USA.
During this same period, genes from human rotavirus were incorporated into a bovine strain called WC3, creating five new rotavirus strains containing serotype-specific, human rotavirus genes. In spite of the disaster surrounding Rotashield, this pentavalent rotavirus was eventually licensed and incorporated into the childhood immunization programme in the USA under the name RotaTeq. This is now one of the two major rotavirus vaccines being administered to infants in the world today.
In July 2004, a new rotavirus vaccine was licensed in Mexico under the name RotarixTM (GlaxoSmithKline, Brentford, UK). This vaccine was developed from a rotavirus vaccine composed of a single rotavirus strain obtained from a child in Cincinnati, OH, USA. Distribution of the vaccine within Mexico was launched in January 2005, which was the first time a rotavirus vaccine had been internationally distributed and more than 5 years after Rotashield was withdrawn from the market in the USA. Today, Rotarix is licensed in more than 110 countries and has been recommended for universal usage by the World Health Organization (WHO). Many developed countries, including the USA, have incorporated its use into their childhood immunization programmes along with RotaTeq. In Latin America, most children receive Rotarix as infants, and countries in Africa are gradually following suit. If this pattern continues, the use of Rotarix is expected to substantially reduce global mortality caused by rotavirus within the next few years.
Rationale for development of Rotarix
Human infections are caused by rotaviruses belonging to a variety of serotypes as defined by antigenic determinants present in the outer capsid VP4 (P) and VP7 (G) proteins.10 Although reinfection with rotavirus occurs routinely, a single rotavirus infection has been reported to prevent most subsequent rotavirus illnesses, particularly those classified as severe.36 Thus, protection provided by a single rotavirus infection appears to cross serotype lines. Based on this as well as other observations, it was considered possible to develop a single-strain, live rotavirus vaccine that would be effective against multiple rotavirus serotypes if its administration were to mimic the outcome found after natural rotavirus infection. Proof of this concept was eventually provided from numerous trials with Rotarix in which evidence of protection against human rotaviruses of all serotypes circulating in the study population has been consistently observed.
Development of the 89-12 rotavirus vaccine, the precursor to Rotarix
In 1988, a trial was conducted in Cincinnati to evaluate a new rotavirus vaccine, the WC3 strain of bovine rotavirus. This vaccine had been reported to provide significant efficacy against subsequent rotavirus illnesses when evaluated in a small trial in Philadelphia, PA, USA.37 After administration of one dose of this vaccine or placebo (to 103 subjects each) to Cincinnati infants ranging in age from 2 to 12 months, no significant protection was detected in vaccines against subsequent mild or severe illnesses caused by a circulating G1P rotavirus strain.38 This disappointing result was in direct juxtaposition to that observed when subjects who became naturally infected with rotavirus during this study were followed for a second year.39 Of the 57 subjects who developed an immune response to rotavirus during the first year of the study, including 20 who did not develop symptoms of gastroenteritis and whose infections were detected through seroconversions only, none became symptomatically infected with rotavirus during year 2 and only two were even infected. In contrast, 11 out of 85 subjects who did not develop immune responses in year 1 had a symptomatic rotavirus infection in year 2 and an additional 20 subjects experienced asymptomatic infections. Only rotaviruses belonging to the G1 serotype were detected during both years of the study. Of particular interest was the observation that almost all rotavirus strains circulating in the first year had identical electropherotypes.38 Furthermore, these G1 strains stimulated neutralizing antibody responses not only against G1 rotaviruses, but also against G3 and G4 strains, representative of other major serotypes that share the same P VP4 genotype.40 This was not unexpected because approximately 75% of the neutralizing antibodies generated against the infecting rotaviruses were found to be directed against the VP4 or P protein.41 Coupled with the protection data, these results suggested that, if the original circulating strain could be attenuated, it should be a highly effective vaccine.
Attenuation of the virus was performed by the classical method, i.e. multiple passages in cell culture to select viral mutants that grow better in vitro but less well in their natural host. The unpassaged virus was obtained from the stool of a placebo recipient (subject #2) in the 1988–89 WC3 vaccine trial and designated as strain 89-12. A vaccine lot derived after 33 passages of this strain in African green monkey kidney cells was evaluated first in adults, then in children previously infected with rotavirus, and finally in infants. These studies were sponsored under a licence with a small pharmaceutical company called Virus Research Institute in Cambridge, MA, USA. The vaccine was well tolerated and 19 out of 20 previously uninfected subjects developed serum rotavirus IgA responses.42
Based on these promising results, the 89-12 vaccine was evaluated in a multicentre, placebo-controlled (1:1), two-dose efficacy trial of 215 subjects.43 Although the vaccinees experienced a greater number of low-grade fevers than the placebo recipients, no other symptoms were associated with the vaccine. Nearly all the subjects (94.4%) developed serum rotavirus IgA or neutralizing antibody responses after two doses. This two-dose regimen was, therefore, maintained in essentially all trials conducted during the development of this vaccine. Protection against all rotavirus disease in this first efficacy trial was 89%, and this increased to 100% for subjects requiring medical intervention, a pattern that changed little when the subjects were followed for a second year.44 As at least 80% of the rotaviruses obtained from infected subjects were typed as G1, the same G-type as the vaccine, protection against rotaviruses of other serotypes was not determinable in this trial. Regardless, these initial results showed outstanding promise for this vaccine that was subsequently sublicensed to SmithKline Beecham (SKB), which became GlaxoSmithKline (GSK) in 2000, for further development as a vaccine candidate.
The results of this highly successful but small efficacy trial were published in July 1999 and appeared in print the same week that vaccinees administered Rotashield were found to develop excess cases of IS. Although the 89-12 vaccine had already been sublicensed to SKB, this major adverse reaction to a rotavirus vaccine caused all rotavirus vaccine developers to pause. Fortunately, development of the 89-12 vaccine into Rotarix and the pentavalent bovine reassortant vaccine into RotaTeq proceeded in spite of the cloud of apprehension hanging over both.
Isolation of RIX4414 from the 89-12 vaccine
After 33 passages of the 89-12 virus, this vaccine preparation was composed of multiple genetic variants of the original virus. To obtain a single virus isolate from the 89-12 vaccine preparation, passage 33 was end-point diluted three consecutive times in Vero cells (passages 34–36). Partial sequence analyses of the genome segments of 10 selected isolates revealed a low degree of genetic variation and the isolate that best matched the consensus sequence of virus in the passage 33 vaccine was chosen for further development. A final lyophilized vaccine preparation was developed after an additional seven passages of this isolate (RIX4414) in Vero cells.
Phase I and II evaluations of RIX4414
After initial safety and immunogenicity studies in adults and 1- to 3-year-old toddlers with serological evidence of previous rotavirus infections, the first trial with RIX4414 was conducted with previously uninfected infants in Finland.45 Infants aged 6–12 weeks received two oral doses of the vaccine ranging from 104.1 to 105.8 ffu in a four-part, placebo-controlled, dose escalation study. The vaccine appeared to be safe and immunogenic. Serum rotavirus IgA responses were dose dependent and detected in 73–96% of the subjects. Similar results were obtained in a placebo-controlled safety/immunogenicity trial of this vaccine with 529 infants aged 5–15 weeks conducted in the USA and Canada.46 Of note, the fever associated with administration of 89-12 in the earliest trials was not detected in these trials of RIX4414. This may have been due to either the clonal selection or the additional passages of the vaccine virus.
The first efficacy trial with RIX4414 was conducted in Finland, in which 405 infants aged 6–12 weeks were administered two doses of 104.7 ffu of RIX4414 separated by a 2-month interval and followed up for 2 years.47 Efficacy over the two seasons was 72% and 90% against all and severe rotavirus illness, respectively. Again, however, almost all illnesses were due to G1 rotaviruses. Thus, cross-G-type protection by this vaccine could not be evaluated. This was followed by a much larger trial in Singapore, in which 2464 infants aged 11–17 weeks were administered two doses of 104.7, 105.2 or 106.1 ffu of RIX4414 or placebo.48 Serum rotavirus IgA responses were detected in between 76% and 91% of the subjects, but efficacy could not be evaluated because of the low number of rotavirus illnesses.
The largest of the initial efficacy trials was conducted in three less-developed nations in Latin America, i.e. Brazil, Mexico and Venezuela. Infants aged 6–12 weeks were administered two doses of RIX4414, ranging between 104.7 and 105.8 ffu (1618 subjects), or placebo (537 subjects) and followed up for rotavirus illnesses until 1 year of age.49 Immunogenicity results were discouraging, in that only 61–65% of the vaccinees developed detectable serum rotavirus IgA responses. However, protection elicited by the highest dose of vaccine was greater than the measured immune responses (70% and 86% against all and severe rotavirus diseases, respectively). As local, intestinal immune responses are perceived to be more relevant to protection after rotavirus infection than systemic responses, it is not surprising that serum rotavirus IgA levels were not perfect correlates of protection after RIX4414 vaccination. Overall, however, a clear correlation was found between serum rotavirus IgA responses and clinical protection against rotavirus gastroenteritis in the studies conducted in both Finland and Latin America.50 In fact, this is the only rotavirus vaccine for which a consistent correlate of protection has been found.
The Latin American study was the first RIX4414 trial in which a significant number of the illnesses were caused by non-G1 strains, particularly those belonging to G9, thus allowing some measure of heterotypic immunity. Protection against severe disease in those receiving the highest dose was similar against the non-G1 and the G1 strains (77% vs. 88%, respectively).49
For a rotavirus vaccine to be most useful it should be compatible with administration of other childhood vaccines. Thus, it was important to note that in both this and other RIX4414 trials, for which routine immunizations with other recommended vaccines were administered concomitantly, no interference of immune responses to any of these vaccine antigens was found. These highly encouraging results provided justification for subsequent large phase III safety/efficacy trials.
Phase III trials with RIX4414
Withdrawal of the first licensed rotavirus vaccine (Rotashield) in 1999, owing to its association with increased rates of IS, formed a cloud over the future of live rotavirus vaccines. It was clear that a risk of IS would have to be essentially eliminated for any new candidate to be licensed. Because this is a rare disease, sample size estimates revealed this would require phase III studies much larger than those routinely used to evaluate vaccines. Therefore, RIX4414 was evaluated in a placebo-controlled trial of 63 225 infants in 11 Latin American countries and Finland.51 Subjects received two doses of either 106.5 cell culture infective doses of vaccine or placebo (1:1 ratio), beginning at between 6 and 13 weeks of age and separated by 1–2 months. The safety objectives were to access the risk of IS within 31 days after administration of each dose of vaccine and the occurrence of any serious adverse event, including IS, during the entire 1-year study period. Vaccine efficacy against rotavirus gastroenteritis of sufficient magnitude to require overnight hospitalization or rehydration therapy was also evaluated in a subset of 20 169 infants.
The risk of definite IS in vaccinees within 31 days after doses 1 and 2 was essentially equal in vaccine and placebo recipients (six compared with seven cases, respectively). Interestingly, of the 12 cases of IS reported after the 31-day window, only three were in the vaccine group, a nearly significant decrease (P = 0.08) suggestive of protection against IS due to RIX4414 vaccination, perhaps by preventing rotavirus illness. Most importantly, no risk of IS was associated with administration of this vaccine.
The efficacy of the vaccine in this large trial was equivalent to that noted in the previous Rotarix study in Latin America,49 i.e. 85% against severe rotavirus disease of any serotype (Table 1). This protection was essentially equivalent against G1 strains and non-G1 strains that shared the P genotype with the G1 RIX4414 vaccine, i.e. G3, G4 and G9 strains. Protection against totally heterotypic G2P strains was 41%, but there were only 16 illnesses due to these viruses and thus wide confidence intervals (CIs) to this assessment. When the efficacy evaluation for this vaccine was extended another year, overall protection against severe rotavirus gastroenteritis remained high (81%) and the levels of protection against the different rotavirus serotypes changed little in the 15 183 subjects who were followed.52 Efficacy against very severe forms of the disease (a score of ≥ 19 on the Vesikari scale) during the entire 2-year period was nearly 100%. Of interest, in this study, the incidence of severe gastroenteritis of any cause was reduced by approximately 40% in vaccinees (see Table 1), a somewhat greater reduction than anticipated based on the expected percentage owing to rotavirus.52
Perhaps the most relevant study to predict efficacy in developed nations is a 2-year efficacy trial of RIX4414 with 3994 participants conducted in six developed European countries.53 Two doses of vaccine (106.5 cell culture infective doses each) were co-administered with routine vaccinations according to the normal schedules. Two-year protection against severe rotavirus of 90.4% was somewhat improved relative to results found in Latin America (95.8% in the first year and 85.6% in the second) (Table 2). This difference in efficacy comparing less-developed with developed settings, and especially with developing country settings, has been consistently noted regardless of the vaccine being studied (see Evaluation of RIX4414 in Third World nations). Rotavirus illnesses of any severity were also monitored for 2 years in this European study in which protection seemed to decline in year 2 compared with year 1 (71.9% vs. 87.1%), suggestive of some waning in immunity in developed countries where re-exposure to rotavirus with an accompanying boost in immunity is expected to be lower than in less-developed and developing nations. Overall protection against rotaviruses belonging to G serotypes 1–4 and 9 (including G2P strains) was all highly significant over the 2-year period, ranging from 58.3% to 89.8% against any rotavirus disease (data not shown) and from 85.5% to 96.4% against severe illnesses (see Table 2). This was the most definitive proof of heterotypic immunity for a rotavirus vaccine at that time. Remarkably, admissions for gastroenteritis of any cause were reduced by 72% overall, 75% in year 1 and 65% in year 2, indicating that either the relative contribution of rotavirus as a cause of more severe gastroenteritis is underestimated in developed nations or vaccination with RIX4414 stimulated protection against other causes of gastrointestinal illness.
Some insight into the cause for the seemingly excess protection against all severe gastroenteritis in this European study53 was provided by a report containing additional analyses of the Mexican subset of subjects enrolled in the earlier phase II trial conducted in three Latin American countries.54 Although the number of subjects in the subset was relatively small (n = 405), it was intriguing to note that even though the three dose levels of vaccine administered (104.7, 105.2 and 105.8) provided comparable protection against rotavirus illnesses, protection against severe diarrhoea episodes of all causes was 18%, 48%, and 70%, respectively, making this outcome highly dose dependent (P = 0.005). These combined results suggest that prevention of rotavirus disease or infection by RIX4414 vaccination can decrease the severity of subsequent infection by other gastroenteritis-producing agents, perhaps by preventing damage to the gastrointestinal tract by subclinical rotavirus infections.
A phase III efficacy trial with RIX4414 was also conducted in developed settings in Asia (Singapore, Hong Kong and Taiwan). In this large (n = 10 708 total subjects), 2-year, placebo-controlled (1:1 randomization) trial, administration of two doses of vaccine beginning at 6–17 weeks of age provided 96.1% protection against severe rotavirus gastroenteritis (100% against G1 strains and 93.6% against pooled non-G1 strains), thus reinforcing the conclusions obtained from the European trial.
As RotarixTM is a monovalent G1[P8] vaccine and contains neither G2, G9 nor [P4], there has been interest in examining the efficacy of RotarixTM against fully heterotypic strains such as G2[P4] and G9[P4]. In an integrated analysis, efficacy over the combined study period against G2[P4] was 58.3% against disease of any severity and 85.5% against severe disease.53,55 More recently, a study in Mexico showed efficacy of 94% against rotavirus hospitalizations owing to a fully heterotypic G9P rotavirus strain.56
Evaluation of RIX4414 in Third World nations
Additional phase III trials with RIX4414 have been, and continue to be, conducted in developing nations. In a study of African infants conducted in South Africa (n = 3166 infants) and Malawi (n = 1773 infants), two or three doses of vaccine were compared with placebo for prevention of severe rotavirus gastroenteritis.57 A total of 4939 infants were enrolled, with 1647 receiving two doses and 1651 receiving three doses of vaccine, while 1641 received placebo. The combined efficacy in the two countries against severe disease was 61.2% (95% CI 44.0–73.2%). Vaccine efficacy was lower in Malawi (49.4%) than in South Africa (76.9%).57 It is important to understand that because of the higher mortality rates, more lives would be saved in Malawi even than in South Africa despite the lower efficacy. Comparable efficacies were demonstrated against a diverse group of circulating strains in this African study.58 Efficacy against all-cause severe gastroenteritis in this study was 30.32%.
The post-licensure effectiveness of Rotarix has been verified in several studies.59,60 Effectiveness studies, conducted in developed or less-developed nations, have recently included nations classified as developing nations, which account for 95% of deaths due to rotavirus occur each year. In El Salvador, a low- to middle-income country, effectiveness was 76% against rotavirus hospitalization.61 In Brazil, effectiveness was 77% against rotavirus hospitalizations or emergency department treatments.62 Effectiveness was markedly reduced in those over 12 months of age. In a study conducted in Bolivia, protection against hospitalization due to rotavirus was between 69% and 77% during the first year of life thanks to the Rotarix vaccination.63 Similar effectiveness was reported to have been sustained through the first 2 years of life. Protection was again significant against diverse serotypes, partially or fully heterotypic to the G1P vaccine strain. Of interest, in a G9 rotavirus outbreak in Australia, effectiveness was 85%.64
Perhaps most importantly, vaccination has been associated with reduced mortality from diarrhoea-associated disease. In a study in Mexico,65 diarrhoea-related mortality fell from an annual median of 18.1 deaths per 100 000 children under 5 years of age before the introduction of Rotarix vaccination to 11.8 per 100 000 children in 2008 (about 2 years after the introduction of the vaccine). Among infants who were 11 months of age or younger, diarrhoea-related mortality fell from 61.5 deaths per 100 000 children at baseline to 36.0 per 100 000 children in 2008. These findings were recently extended to show a sustained reduction in diarrhoea-related mortality for children under the age of 5 years for 3 full years.66 Compared with baseline, diarrhoea-related mortality fell by 56% during rotavirus season after vaccination.
In the USA, vaccine effectiveness was 91% (95% CI 80–95%) among children aged ≥ 8 months. The vaccine effectiveness against G2P disease was high at 94% (95% CI 78–98%), as was that against G1P disease at 89% (95% CI 70–96%). Effectiveness was sustained for 23 months among children aged 12 years [vaccine effectiveness (VE) 91%; 95% CI 75–96%].67
The implications of these studies have led the WHO to recommend that all countries introduce rotavirus vaccines into their national immunization programmes but, to date, only 45 – including eight in Africa – have done so. However, 22 African nations had applied for support from Gavi to introduce rotavirus vaccines. Therefore, although uptake of rotavirus vaccines in countries where they are needed most has been slow, it has been steady, thus providing optimism to the suggestion that rotavirus vaccines may save 240 000 lives annually within the next few years.