- Email: [email protected]
A proposed framework for evaluating and comparing efficacy estimates in clinical trials of new rotavirus vaccines
Vaccine 32S (2014) A179–A184
Contents lists available at ScienceDirect
Vaccine journal homepage: www.elsevier.com/locate/vaccine
Review
A proposed framework for evaluating and comparing efficacy estimates in clinical trials of new rotavirus vaccines Kathleen M. Neuzil a,∗ , K. Zaman b , John C. Victor a a b
PATH, Seattle, WA, USA International Centre for Diarrhoeal Disease Research, Bangladesh (icddr,b), Dhaka, Bangladesh
a b s t r a c t Oral rotavirus vaccines have yielded different point estimates of efficacy when tested in different populations. While population and environmental factors may account for these differences, study design characteristics should also be considered. We review the study design elements of rotavirus vaccine trials that may affect point estimates of efficacy, and propose a framework for evaluating new rotavirus vaccines. © 2014 Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).
The World Health Organization (WHO) has recommended oral rotavirus vaccines for all infants worldwide [1]. As of May 20, 2014, 60 countries worldwide and 26 GAVI-eligible countries had introduced rotavirus vaccine (RV) into their national immunization programs [2]. (Fig. 1) Major barriers to more rapid introduction of rotavirus vaccines in low-resource settings have been related to vaccine cold chain constraints in some countries and limited product-of-choice availability for others. Thus, the availability of additional, affordable rotavirus vaccines is a high priority to enhance rotavirus disease control efforts. Clinical trials under real-world conditions in low-resource countries established the public health benefit of RotaTeq® (Merck & Co.) and Rotarix® (GlaxoSmithKline), and informed the WHO recommendation for their use [1,3–5]. Much has been written about the lower point estimates of efficacy in these trials compared with trials performed in higher resource settings. Among the reasons given for the lower efficacy are higher maternal antibody in low-resource settings, environmental enteropathy, differences in the gut microbiome among children in different resource settings, nutritional status, breastfeeding practices and interference by oral poliovirus vaccines [6–9]. In addition to these factors, we propose that the contribution of study design differences should be considered when comparing point estimates of efficacy across trials. In addition, the biologic factors and study design factors may be interrelated; for example, the higher antibody in low resource settings may be due to both an increased exposure to rotavirus and to
∗ Corresponding author at: PATH, 2201 Westlake Avenue, Suite 200, Seattle, WA 98121, USA. Tel.: +1 206 285 3500; fax: +1 206 825 6619. E-mail address: [email protected] (K.M. Neuzil).
the younger age at administration of routine childhood vaccines, including rotavirus vaccines. Currently, there is no accepted correlate of protection for rotavirus vaccines, and thus development of new rotavirus vaccines for the foreseeable future will likely require evaluations of clinical efficacy. Recently, a new rotavirus vaccine, ROTAVAC® , based on the 116E rotavirus strain and manufactured by Bharat Biotech International Limited of India, demonstrated efficacy in a pivotal clinical trial in India [10,11]. Additional rotavirus vaccines are in various stages of preclinical and clinical development. The parameters for the success of such trials from a regulatory perspective will likely differ from the parameters for policy or vaccine introduction decisions, and thus the various study designs used to evaluate efficacy in these trials are likely to differ. To properly frame the results of clinical trials conducted with new vaccines, we reviewed the available literature on efficacy trials of rotavirus vaccines in low-resource settings in Africa and Asia. While acknowledging the importance of safety in regulatory and policy decisions, we limited this review to efficacy outcomes, and to the currently approved and recommended vaccines (Rotarix® , RotaTeq® ). Both Rotarix® and RotaTeq® were already approved by international regulatory authorities when tested in Africa and Asia, and thus those trials were conducted primarily to inform policy. Under the assumption that aspects of study design and population characteristics will influence the point estimates of efficacy obtained, we propose that comparisons of point estimates of efficacy from different trials may be challenging, and should be done with a clear understanding of trial design and the variables that could influence such comparisons. Table 1 provides a number of factors that are known or hypothesized to influence rotavirus vaccine immunogenicity and/or efficacy, with references and examples
http://dx.doi.org/10.1016/j.vaccine.2014.04.074 0264-410X/© 2014 Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).
A180
K.M. Neuzil et al. / Vaccine 32S (2014) A179–A184
Fig. 1. Rotavirus vaccine introductions in GAVI-eligible countries as of May 1, 2014.
from clinical trials. We then used these study design characteristics as a framework for evaluating the efficacy data from the new oral rotavirus vaccine, ROTAVAC® as an example of how to interpret appropriately new efficacy results (Table 2). Concomitant administration of oral poliovirus vaccines (OPV) with oral rotavirus vaccines reduces the immunogenicity of rotavirus vaccines, as measured by serum IgA antibody responses and rotavirus vaccine shedding, when compared with administration of the two vaccines separated in time by 1–2 weeks (Table 1) [12–14]. This lower immunogenicity would be expected to result in no effect, or a reduction in efficacy, against clinical outcomes. In moderate to high resource settings, rotavirus vaccines were administered with inactivated poliovirus vaccines (IPV), or separated from OPV administration by at least 2 weeks. In trials performed to date in low resource settings, most of the children received OPV concomitantly with RVs as shown in Table 2. The exception was the trial of RotaTeq® in Africa, where only 35% of children received OPV with RV. In this example, however, the administration was not separated by two weeks, but may have been separated by only a day or two. The effect of OPV in that situation is not known, but might be expected to be even greater than concomitant administration given the replication kinetics of OPVs. Overall, the global plans to move from trivalent to bivalent OPVs, and eventually to inactivated poliovirus vaccines (IPV) would be expected to have favorable effects on the immunogenicity of oral RVs in low-resource settings. A major issue emerging from rotavirus vaccine trials in high mortality/low resource settings compared with low mortality/high
resource settings has been the observation of possible waning of efficacy in the second year of life. Thus, in developing world trials that include follow-up time beyond the first year of life (or over multiple years) the relative person-time accumulated estimate reported during the first versus second year of life is critical to interpreting the summary point estimate of efficacy. For example, the RotaTeq® trial in Africa ended on a specific date, and so the primary outcome included follow-up to a median of 21 months of age [5]. Thus, the overall efficacy reported in this trial reflects cases occurring at various ages. Relatively more cases during the first year of life when vaccine protection appears to be highest would lead to higher overall cumulative efficacy. Additionally, sites had different follow-up time and contributed cases differently to the first versus second years of life. In the RotaTeq® study in Africa, for example, the site in Mali, with lower point estimates of efficacy during both years, contributed relatively more cases in the second year of life as compared with the first year. So comparisons of efficacy beyond the first year of life are particularly problematic without a full understanding of the mix of cases by year and by site [15,16]. Another important element to consider when comparing results from different trials is the outcome measure. Most trials have focused on severe gastroenteritis as measured by the Vesikari scoring system, as the primary outcome measure. Even in circumstances where the outcome is relatively uniform, how the scoring system is utilized may differ between sites [17]. In addition, secondary outcome measures (e.g. efficacy according to severity of disease, all-cause gastroenteritis) may offer additional
K.M. Neuzil et al. / Vaccine 32S (2014) A179–A184 Table 1 Effect of study design and population characteristics on estimates of point efficacy in rotavirus vaccine trials.
A181
A182
K.M. Neuzil et al. / Vaccine 32S (2014) A179–A184
Table 1 (Continued ).
a b c
Oral polio vaccine. Gastroenteritis. Oral rehydration solution.
information on the public health value of a vaccine, but also require interpretation of point estimates in the context of the definitions employed. For example, in rural Kenya, multiple measures of severe gastroenteritis were used for children in the trial as a substudy of the larger multicenter RotaTeq® efficacy trial in Africa [18]. The primary outcome measure for the multicenter trial was severe gastroenteritis as measured in healthcare facilities using the 20-point modified Vesikari scoring system. A second measure of acute gastroenteritis was employed during weekly home visits, using WHO’s Integrated Management of Childhood Illness (IMCI) criteria for dehydration. RotaTeq® was 83.4% (95% CI 25.5, 98.2) efficacious in the first year of life against severe rotavirus gastroenteritis (Vesikari score ≥ 11) and 34.4% (95% CI 5.3, 54.6) efficacious in preventing acute gastroenteritis associated with severe dehydration in the home setting. These differences highlight the need to critically evaluate the degree to which outcome measures and the tools to measure them are tuned to the population being studied. The preceding example also illustrates why comparisons of point estimates of efficacy alone provide an incomplete assessment of vaccine performance. Absolute disease rates and case reductions are more relevant measures of the public health impact of vaccines. This concept was endorsed by an international panel of experts convened in 2007, prior to the release of data from the rotavirus vaccine trials in developing countries [19]. Using the example above in Kenya, rotavirus vaccines prevented an estimated 3.3 cases per 100 child-years of severe rotavirus gastroenteritis, as defined by the Vesikari score and measured at health facilities, and 19 cases per 100 child-years of acute gastroenteritis with severe dehydration as defined by IMCI criteria and measured in the home [18]. This illustrates the importance of outcome definition, both from the perspective of comparisons, but also from the perspective of public health value. In rural Africa, prevention of home cases of dehydration may be a more relevant measure of prevention as children have limited access to care, and thus the use of different outcome definitions can provide a more complete assessment of disease reduction afforded by vaccines [18]. It will be important to report incidence rates in placebo and vaccine groups for a number of outcomes in trials of new rotavirus vaccines, again with an understanding of how differences in case ascertainment and case definition could affect those incidence rates. Age is an important influencer of vaccine immunogenicity, with immune responses to vaccine generally improving with age. It is difficult to determine whether the lower immune responses
reflect an immature immune system, or interference by high concentrations of maternal antibody that wane over time. What is known is that the high levels of serum neutralizing antibody against the human rotavirus serotypes in the RotaTeq® vaccine measured before vaccination, and thus presumed to be maternal antibody, has only been observed in the low resource settings of Africa and Asia [9]. Additional factors that could impact point estimates of efficacy are shown in Table 1. Some are difficult to quantify, and a full description of each of these parameters may not always be provided in published manuscripts. For example, while pivotal studies for licensure generally have strict inclusion criteria, the Rotarix® and RotaTeq® trials in Africa and Asia had more lenient inclusion criteria. While one would expect healthier children to have better immune responses than children with underlying disease, few specific data are available. The limited studies performed in HIVinfected children suggest a satisfactory immune response [3,19]. Another example is the routine use of interventions, such as oral rehydration solution (ORS) that could affect the outcome of interest – severe rotavirus gastroenteritis – and potentially mask the full effects of the vaccine on severe disease [21]. Likewise, the timing of vaccination and the method of analysis in relation to rotavirus circulation may affect efficacy estimates, although the direction of the effect may be difficult to predict. For example, in the efficacy trial in the South Africa site, all vaccinations were completed prior to the start of the rotavirus season. Thus, children exposed to rotavirus had received vaccine relatively recently, which may favor vaccine efficacy estimates if there is any waning of immunity over time. In the same trial, at the Malawi site, vaccinations occurred throughout the year, including time periods when rotavirus circulated. These differences are reflected in the percentage of children in the placebo group with detectable rotavirus IgA antibody at 18 weeks of age at the two sites – 40.5% in Malawi as compared to 11.6% in South Africa. Another example is the RotaTeq® trial that included a cohort in Mali, where vaccinations were given before and during rotavirus season. As the per protocol definition required cases to occur at least 2 weeks following the last dose of vaccine, fewer cases were available for the per protocol evaluation. The intention to treat analysis is arguably the more relevant from the public health perspective, as rotavirus vaccines are given with other childhood vaccines on a year-round schedule. The use of the PP definition has led many to conclude that the vaccine was not efficacious in Mali [22]. While both the ITT and PP point estimates are imprecise due
Table 2 Summary of efficacy trials of currently approved rotavirus vaccines in low-resource settings in Africa and Asia. Africa Median age (weeks)
Randomization
Outcome measures
Number of doses
Sites
Other
Efficacy (follow-up to 1 year of age) (95% CI)
Efficacy (second year of life) (95% CI)
RotaTeq®
35
7, 12, 16
Individual
3
6, 11, 16
Individual
65% birth dose OPV 3 dose arm also included
19.6% (−15.7–44.4)
99
Ghana, Kenya, Mali [5] Malawi, South Africa [3]
64.2% (40.2–79.4)
Rotarix®
Severe rotavirus GEb (Vesikari ≥ 11) Severe rotavirus GE (Vesikari ≥ 11)
58.7% (35.7–74.0)
Results available by site [14,15]
Concomitant OPV use (%)
Median age (weeks)
Randomization
Outcome measures
Number of doses
Sites
Other
Efficacy (follow-up to 1 year of age) (95% CI)
Efficacy (follow-up to 2 years of age) (95% CI)
95
9, 14, 18
Individual
3
45.5% (1.2–70.7)
8, 13, 17
Individual
25% birth dose OPV
45.7% (−1.2–71.8)
39.3% (−18.3–69.7)
10, 14, 19
Individual
Severe rotavirus GE (Vesikari ≥ 11)
3
Bangladesh, Vietnam [4] Bangladesh site only (Matlab: icddr,b area) [4] Vietnam only [4]
51.0% (12.8–71.8)
99
Severe rotavirus GE (Vesikari ≥ 11) Severe rotavirus GE (Vesikari ≥ 11)
No birth dose of OPV
72.3% (−45.2–97.2)
64.6% (−47.7–93.9)
Cluster
Rotavirus diarrhea presenting to treatment facility (case capture identical as to RotaTeq® trial) Severe rotavirus GE (Vesikari≥11)
2
Matlab, Bangladesh [21]
42.3%c (24–56)
n/a
3
Multiple sites in India [10,11]
56.3% (36.7–69.9)
48.9% (17.4–68.4)
2
Asia
RotaTeq®
90% Rotarix®
ROTAVAC® a b c d
∼95
99d
8, 13
7, 12, 16
Individual
3
K.M. Neuzil et al. / Vaccine 32S (2014) A179–A184
Concomitant OPVa use (%)
Oral polio vaccine. Gastroenteritis. Total effectiveness (direct + indirect). Personal communication.
A183
A184
K.M. Neuzil et al. / Vaccine 32S (2014) A179–A184
to the small number of cases that occurred in the first year of life, the ITT point estimate of 42.7% (95% CI −124.7 to 87.7) is more in line with the point estimates of efficacy from the sites in Ghana and Kenya that were part of that multicenter trial [5]. As we do not yet have a complete understanding of the protective mechanism of rotavirus vaccines in low-resource settings, additional factors that are not yet understood or easily measured could also affect trial results. In Table 2, realizing that all factors may not be fully delineated or reported, the studies of rotavirus vaccines in low-resource settings, including the recent results from the ROTAVAC® efficacy trial conducted in India [10,11], are categorized by important design characteristics. For the major variables of age, use of OPV, outcome definition, and type of randomization, the ROTAVAC® efficacy trial design is similar to the design of the individually randomized RotaTeq® and Rotarix® studies. Acknowledging uncertainties that may still remain in the implementation of the study outcome measurement, differing rotavirus seasonality and epidemiology between sites, and other unknown differences between populations, the point estimates of efficacy from the recently reported ROTAVAC® trial compare quite favorably to those from Rotarix® and RotaTeq® trials in low-resource settings. While the RotaTeq® trial in Asia was designed and conducted as a multicenter trial in Bangladesh and Vietnam, we also present the estimates for the two sites separately, in order to provide what we hypothesize to be the most relevant comparisons to the ROTAVAC® trial in India. In the RotaTeq® trial, the point estimates for efficacy against severe rotavirus gastroenteritis in the first year of life were 51.0% (95% CI 12.8–73.3) for the entire cohort, 45.7% (95% CI −1.2 to 71.9) for the Bangladesh cohort and 72.3% (−45.2 to 97.2) for the Vietnam cohort. The ROTAVAC® point estimate of efficacy for the same outcome in the first year of life was 56.4% (95% CI 36.7–69.9). The apparent maintenance of efficacy in the second year of life in the ROTAVAC® trial is encouraging, and similar to what was seen in the RotaTeq® trial in Asia, recognizing that point estimates of efficacy in the second year of life are less precise, given the smaller number of outcomes. This is indeed an exciting time for rotavirus vaccines. Ultimately, multiple safe and efficacious choices should allow for optimal price and supply conditions, resulting in maximal numbers of children vaccinated. Head-to-head comparisons of different vaccines would be the best way to control for study design and population differences, and may be more common in the future given the global roll-out of rotavirus vaccines. In the meantime, this proposed framework should be useful in comparing efficacy estimates of new rotavirus vaccines conducted with placebo controls in various settings. We have proposed important design elements to be considered in those comparisons, including age at receipt of vaccine; co-administration of other vaccines, most notably OPV; definition and method of ascertainment of outcome measure; inclusion and exclusion criteria; and the pattern of rotavirus circulation. Ultimately, vaccine choices by individual countries are unlikely to be based on efficacy alone, and will include considerations of rotavirus disease burden, vaccine safety, cost and feasibility. Conflict of interest None reported.
References [1] World Health Organization. Meeting of the immunization strategic advisory group of experts, April 2009—conclusions and recommendations. Wkly Epidemiol Rec 2009;84(23):220–36. [2] PATH. Country introductions of rotavirus vaccines. Maps and list, Figure 1; 2014. Available at: http://sites.path.org/rotavirusvaccine/rotavirusadvocacy-and-communications-toolkit/country-introduction-maps-and-list/ [accessed 30.01.14]. [3] Madhi SA, Cunliffe NA, Steele D, Witte D, Kirsten M, Louw C, et al. Effect of human rotavirus vaccine on severe diarrhea in African infants. New Engl J Med 2010;362(4):289–98. [4] Zaman K, Dang DA, Victor JC, Shin JC, Yunus M, Dallas M, et al. Efficacy of pentavalent rotavirus vaccine against severe rotavirus gastroenteritis in infants in developing countries in Asia: a randomised, double-blind, placebo-controlled trial. Lancet 2010;376(9741):615–23. [5] Armah GE, Sow SO, Breiman RF, Dallas M, Tapia M, Feikin D, et al. Efficacy of pentavalent rotavirus vaccine against severe rotavirus gastroenteritis in infants in developing countries in sub-Saharan Africa: a randomised, double-blind, placebo-controlled trial. Lancet 2010;376(9741):606–14. [6] Neuzil KM, Parashar UD, Steele AD. Rotavirus vaccines for children in developing countries: understanding the science, maximizing the impact, and sustaining the effort. Vaccine 2012;30(Suppl. 1):A1–2. [7] Glass RI, Parashar U, Patel M, Gentsch J, Jiang B. Rotavirus vaccines: successes and challenges. J Infect 2014;68(Suppl. 1):S9–18. [8] Patel M, Shane AL, Parashar UD, Jiang B, Gentsch JR, Glass RI. Oral rotavirus vaccines: how well will they work where they are needed most? J Infect Dis 2009;200(Suppl. 1):S39–48. [9] Armah GE, Breiman RF, Tapia MD, Dallas M, Neuzil K, Binka F, et al. Immunogenicity of the pentavalent rotavirus vaccine in African infants. Vaccine 2012;30(Suppl. 1):A86–93. [10] Bhandari N, Rongsen-Chandola T, Bavdekar A, John J, Antony K, Taneja S, et al. Efficacy of a monovalent human-bovine (116E) rotavirus vaccine in Indian infants: a randomised double blind placebo controlled trial. Lancet 2014:62630–6. S0140-6736(13). [11] Bhandari N, Rongsen-Chandola T, Bavdekar A, John J, Antony K, Taneja S, et al. Efficacy of a monovalent human-bovine (116E) rotavirus vaccine in Indian children in the second year of life. Vaccine 2014 (in press). [12] Zaman K, Sack DA, Yunus M, Arifeen S, Popper G, Azim T, et al. Successful co-administration of a human rotavirus and oral poliovirus vaccines in Bangladeshi infants in a 2-dose schedule at 12 and 16 weeks of age. Vaccine 2009;27(9):1333–40. [13] Patel M, Steele AD, Parashar UD. Influence of oral polio vaccines on performance of the monovalent and pentavalent rotavirus vaccines. Vaccine 2012;30(Suppl. 1):A30–5. [14] Ciarlet M, Sani-Grosso R, Yuan G, Lui G, Heaton P, Gottesdiener K, et al. Concomitant use of the oral pentavalent human-bovine reassortant rotavirus vaccine and oral poliovirus vaccine. Pediatr Infect Dis J 2008;27(10):874–80. [15] Madhi SA, Kirsten M, Louw C, Bos P, Aspinall S, Bouchenooghe A, et al. Efficacy and immunogenicity of two or three dose rotavirus-vaccine regimen in South African children over two consecutive rotavirus-seasons: a randomized, double-blind, placebo-controlled trial. Vaccine 2012;30(Suppl. 1):A44–51. [16] Cunliffe NA, Witte D, Ngwira B, Todd S, Bostock N, Turner A, et al. Efficacy of human rotavirus vaccine against severe gastroenteritis in Malawian children in the first two years of life. Vaccine 2012;30(Suppl 1:):A36–43. [17] Lewis KD, Dallas MJ, Victor JC, Ciarlet M, Mast T, Ji M, et al. Comparison of two clinical severity scoring systems in two multi-center, developing country rotavirus vaccine trials in Africa and Asia. Vaccine 2012;30(Suppl. 1):A159–66. [18] Feikin DR, Laserson KF, Ojwando J, Nyambane G, Ssempijja V, Audi A, et al. Efficacy of pentavalent rotavirus vaccine in a high HIV prevalence population in Kenya. Vaccine 2012;30(Suppl. 1):A52–60. [19] Steele AD, Patel M, Parashar UD, Victor JC, Aguado T, Neuzil KM. Rotavirus vaccines for infants in developing countries in Africa and Asia: considerations from a world health organization-sponsored consultation. J Infect Dis 2009;200(Suppl. 1):S63–9. [20] Breiman RF, Zaman K, Armah G, Sow S, Anh D, Victor J, et al. Analyses of health outcomes from the 5 sites participating in the Africa and Asia clinical efficacy trials of the oral pentavalent rotavirus vaccine. Vaccine 2012;30(Suppl. 1):A24–9. [21] Zaman K. Introduction of a live, oral human rotavirus vaccine (Rotarix) in Matlab, Bangladesh. In: International Rotavirus Symposium. 2012. [22] Nelson A, Glass R. Rotavirus: realizing the potential of a promising vaccine. Lancet 2010;376(9741):568–70. [23] Sow S, Tapia M, Haidara F, Ciarlet M, Diallo F, Kodio M, et al. Efficacy of the oral pentavalent rotavirus vaccine in Mali. Vaccine 2011;30(1):A71–8.
Contents lists available at ScienceDirect
Vaccine journal homepage: www.elsevier.com/locate/vaccine
Review
A proposed framework for evaluating and comparing efficacy estimates in clinical trials of new rotavirus vaccines Kathleen M. Neuzil a,∗ , K. Zaman b , John C. Victor a a b
PATH, Seattle, WA, USA International Centre for Diarrhoeal Disease Research, Bangladesh (icddr,b), Dhaka, Bangladesh
a b s t r a c t Oral rotavirus vaccines have yielded different point estimates of efficacy when tested in different populations. While population and environmental factors may account for these differences, study design characteristics should also be considered. We review the study design elements of rotavirus vaccine trials that may affect point estimates of efficacy, and propose a framework for evaluating new rotavirus vaccines. © 2014 Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).
The World Health Organization (WHO) has recommended oral rotavirus vaccines for all infants worldwide [1]. As of May 20, 2014, 60 countries worldwide and 26 GAVI-eligible countries had introduced rotavirus vaccine (RV) into their national immunization programs [2]. (Fig. 1) Major barriers to more rapid introduction of rotavirus vaccines in low-resource settings have been related to vaccine cold chain constraints in some countries and limited product-of-choice availability for others. Thus, the availability of additional, affordable rotavirus vaccines is a high priority to enhance rotavirus disease control efforts. Clinical trials under real-world conditions in low-resource countries established the public health benefit of RotaTeq® (Merck & Co.) and Rotarix® (GlaxoSmithKline), and informed the WHO recommendation for their use [1,3–5]. Much has been written about the lower point estimates of efficacy in these trials compared with trials performed in higher resource settings. Among the reasons given for the lower efficacy are higher maternal antibody in low-resource settings, environmental enteropathy, differences in the gut microbiome among children in different resource settings, nutritional status, breastfeeding practices and interference by oral poliovirus vaccines [6–9]. In addition to these factors, we propose that the contribution of study design differences should be considered when comparing point estimates of efficacy across trials. In addition, the biologic factors and study design factors may be interrelated; for example, the higher antibody in low resource settings may be due to both an increased exposure to rotavirus and to
∗ Corresponding author at: PATH, 2201 Westlake Avenue, Suite 200, Seattle, WA 98121, USA. Tel.: +1 206 285 3500; fax: +1 206 825 6619. E-mail address: [email protected] (K.M. Neuzil).
the younger age at administration of routine childhood vaccines, including rotavirus vaccines. Currently, there is no accepted correlate of protection for rotavirus vaccines, and thus development of new rotavirus vaccines for the foreseeable future will likely require evaluations of clinical efficacy. Recently, a new rotavirus vaccine, ROTAVAC® , based on the 116E rotavirus strain and manufactured by Bharat Biotech International Limited of India, demonstrated efficacy in a pivotal clinical trial in India [10,11]. Additional rotavirus vaccines are in various stages of preclinical and clinical development. The parameters for the success of such trials from a regulatory perspective will likely differ from the parameters for policy or vaccine introduction decisions, and thus the various study designs used to evaluate efficacy in these trials are likely to differ. To properly frame the results of clinical trials conducted with new vaccines, we reviewed the available literature on efficacy trials of rotavirus vaccines in low-resource settings in Africa and Asia. While acknowledging the importance of safety in regulatory and policy decisions, we limited this review to efficacy outcomes, and to the currently approved and recommended vaccines (Rotarix® , RotaTeq® ). Both Rotarix® and RotaTeq® were already approved by international regulatory authorities when tested in Africa and Asia, and thus those trials were conducted primarily to inform policy. Under the assumption that aspects of study design and population characteristics will influence the point estimates of efficacy obtained, we propose that comparisons of point estimates of efficacy from different trials may be challenging, and should be done with a clear understanding of trial design and the variables that could influence such comparisons. Table 1 provides a number of factors that are known or hypothesized to influence rotavirus vaccine immunogenicity and/or efficacy, with references and examples
http://dx.doi.org/10.1016/j.vaccine.2014.04.074 0264-410X/© 2014 Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).
A180
K.M. Neuzil et al. / Vaccine 32S (2014) A179–A184
Fig. 1. Rotavirus vaccine introductions in GAVI-eligible countries as of May 1, 2014.
from clinical trials. We then used these study design characteristics as a framework for evaluating the efficacy data from the new oral rotavirus vaccine, ROTAVAC® as an example of how to interpret appropriately new efficacy results (Table 2). Concomitant administration of oral poliovirus vaccines (OPV) with oral rotavirus vaccines reduces the immunogenicity of rotavirus vaccines, as measured by serum IgA antibody responses and rotavirus vaccine shedding, when compared with administration of the two vaccines separated in time by 1–2 weeks (Table 1) [12–14]. This lower immunogenicity would be expected to result in no effect, or a reduction in efficacy, against clinical outcomes. In moderate to high resource settings, rotavirus vaccines were administered with inactivated poliovirus vaccines (IPV), or separated from OPV administration by at least 2 weeks. In trials performed to date in low resource settings, most of the children received OPV concomitantly with RVs as shown in Table 2. The exception was the trial of RotaTeq® in Africa, where only 35% of children received OPV with RV. In this example, however, the administration was not separated by two weeks, but may have been separated by only a day or two. The effect of OPV in that situation is not known, but might be expected to be even greater than concomitant administration given the replication kinetics of OPVs. Overall, the global plans to move from trivalent to bivalent OPVs, and eventually to inactivated poliovirus vaccines (IPV) would be expected to have favorable effects on the immunogenicity of oral RVs in low-resource settings. A major issue emerging from rotavirus vaccine trials in high mortality/low resource settings compared with low mortality/high
resource settings has been the observation of possible waning of efficacy in the second year of life. Thus, in developing world trials that include follow-up time beyond the first year of life (or over multiple years) the relative person-time accumulated estimate reported during the first versus second year of life is critical to interpreting the summary point estimate of efficacy. For example, the RotaTeq® trial in Africa ended on a specific date, and so the primary outcome included follow-up to a median of 21 months of age [5]. Thus, the overall efficacy reported in this trial reflects cases occurring at various ages. Relatively more cases during the first year of life when vaccine protection appears to be highest would lead to higher overall cumulative efficacy. Additionally, sites had different follow-up time and contributed cases differently to the first versus second years of life. In the RotaTeq® study in Africa, for example, the site in Mali, with lower point estimates of efficacy during both years, contributed relatively more cases in the second year of life as compared with the first year. So comparisons of efficacy beyond the first year of life are particularly problematic without a full understanding of the mix of cases by year and by site [15,16]. Another important element to consider when comparing results from different trials is the outcome measure. Most trials have focused on severe gastroenteritis as measured by the Vesikari scoring system, as the primary outcome measure. Even in circumstances where the outcome is relatively uniform, how the scoring system is utilized may differ between sites [17]. In addition, secondary outcome measures (e.g. efficacy according to severity of disease, all-cause gastroenteritis) may offer additional
K.M. Neuzil et al. / Vaccine 32S (2014) A179–A184 Table 1 Effect of study design and population characteristics on estimates of point efficacy in rotavirus vaccine trials.
A181
A182
K.M. Neuzil et al. / Vaccine 32S (2014) A179–A184
Table 1 (Continued ).
a b c
Oral polio vaccine. Gastroenteritis. Oral rehydration solution.
information on the public health value of a vaccine, but also require interpretation of point estimates in the context of the definitions employed. For example, in rural Kenya, multiple measures of severe gastroenteritis were used for children in the trial as a substudy of the larger multicenter RotaTeq® efficacy trial in Africa [18]. The primary outcome measure for the multicenter trial was severe gastroenteritis as measured in healthcare facilities using the 20-point modified Vesikari scoring system. A second measure of acute gastroenteritis was employed during weekly home visits, using WHO’s Integrated Management of Childhood Illness (IMCI) criteria for dehydration. RotaTeq® was 83.4% (95% CI 25.5, 98.2) efficacious in the first year of life against severe rotavirus gastroenteritis (Vesikari score ≥ 11) and 34.4% (95% CI 5.3, 54.6) efficacious in preventing acute gastroenteritis associated with severe dehydration in the home setting. These differences highlight the need to critically evaluate the degree to which outcome measures and the tools to measure them are tuned to the population being studied. The preceding example also illustrates why comparisons of point estimates of efficacy alone provide an incomplete assessment of vaccine performance. Absolute disease rates and case reductions are more relevant measures of the public health impact of vaccines. This concept was endorsed by an international panel of experts convened in 2007, prior to the release of data from the rotavirus vaccine trials in developing countries [19]. Using the example above in Kenya, rotavirus vaccines prevented an estimated 3.3 cases per 100 child-years of severe rotavirus gastroenteritis, as defined by the Vesikari score and measured at health facilities, and 19 cases per 100 child-years of acute gastroenteritis with severe dehydration as defined by IMCI criteria and measured in the home [18]. This illustrates the importance of outcome definition, both from the perspective of comparisons, but also from the perspective of public health value. In rural Africa, prevention of home cases of dehydration may be a more relevant measure of prevention as children have limited access to care, and thus the use of different outcome definitions can provide a more complete assessment of disease reduction afforded by vaccines [18]. It will be important to report incidence rates in placebo and vaccine groups for a number of outcomes in trials of new rotavirus vaccines, again with an understanding of how differences in case ascertainment and case definition could affect those incidence rates. Age is an important influencer of vaccine immunogenicity, with immune responses to vaccine generally improving with age. It is difficult to determine whether the lower immune responses
reflect an immature immune system, or interference by high concentrations of maternal antibody that wane over time. What is known is that the high levels of serum neutralizing antibody against the human rotavirus serotypes in the RotaTeq® vaccine measured before vaccination, and thus presumed to be maternal antibody, has only been observed in the low resource settings of Africa and Asia [9]. Additional factors that could impact point estimates of efficacy are shown in Table 1. Some are difficult to quantify, and a full description of each of these parameters may not always be provided in published manuscripts. For example, while pivotal studies for licensure generally have strict inclusion criteria, the Rotarix® and RotaTeq® trials in Africa and Asia had more lenient inclusion criteria. While one would expect healthier children to have better immune responses than children with underlying disease, few specific data are available. The limited studies performed in HIVinfected children suggest a satisfactory immune response [3,19]. Another example is the routine use of interventions, such as oral rehydration solution (ORS) that could affect the outcome of interest – severe rotavirus gastroenteritis – and potentially mask the full effects of the vaccine on severe disease [21]. Likewise, the timing of vaccination and the method of analysis in relation to rotavirus circulation may affect efficacy estimates, although the direction of the effect may be difficult to predict. For example, in the efficacy trial in the South Africa site, all vaccinations were completed prior to the start of the rotavirus season. Thus, children exposed to rotavirus had received vaccine relatively recently, which may favor vaccine efficacy estimates if there is any waning of immunity over time. In the same trial, at the Malawi site, vaccinations occurred throughout the year, including time periods when rotavirus circulated. These differences are reflected in the percentage of children in the placebo group with detectable rotavirus IgA antibody at 18 weeks of age at the two sites – 40.5% in Malawi as compared to 11.6% in South Africa. Another example is the RotaTeq® trial that included a cohort in Mali, where vaccinations were given before and during rotavirus season. As the per protocol definition required cases to occur at least 2 weeks following the last dose of vaccine, fewer cases were available for the per protocol evaluation. The intention to treat analysis is arguably the more relevant from the public health perspective, as rotavirus vaccines are given with other childhood vaccines on a year-round schedule. The use of the PP definition has led many to conclude that the vaccine was not efficacious in Mali [22]. While both the ITT and PP point estimates are imprecise due
Table 2 Summary of efficacy trials of currently approved rotavirus vaccines in low-resource settings in Africa and Asia. Africa Median age (weeks)
Randomization
Outcome measures
Number of doses
Sites
Other
Efficacy (follow-up to 1 year of age) (95% CI)
Efficacy (second year of life) (95% CI)
RotaTeq®
35
7, 12, 16
Individual
3
6, 11, 16
Individual
65% birth dose OPV 3 dose arm also included
19.6% (−15.7–44.4)
99
Ghana, Kenya, Mali [5] Malawi, South Africa [3]
64.2% (40.2–79.4)
Rotarix®
Severe rotavirus GEb (Vesikari ≥ 11) Severe rotavirus GE (Vesikari ≥ 11)
58.7% (35.7–74.0)
Results available by site [14,15]
Concomitant OPV use (%)
Median age (weeks)
Randomization
Outcome measures
Number of doses
Sites
Other
Efficacy (follow-up to 1 year of age) (95% CI)
Efficacy (follow-up to 2 years of age) (95% CI)
95
9, 14, 18
Individual
3
45.5% (1.2–70.7)
8, 13, 17
Individual
25% birth dose OPV
45.7% (−1.2–71.8)
39.3% (−18.3–69.7)
10, 14, 19
Individual
Severe rotavirus GE (Vesikari ≥ 11)
3
Bangladesh, Vietnam [4] Bangladesh site only (Matlab: icddr,b area) [4] Vietnam only [4]
51.0% (12.8–71.8)
99
Severe rotavirus GE (Vesikari ≥ 11) Severe rotavirus GE (Vesikari ≥ 11)
No birth dose of OPV
72.3% (−45.2–97.2)
64.6% (−47.7–93.9)
Cluster
Rotavirus diarrhea presenting to treatment facility (case capture identical as to RotaTeq® trial) Severe rotavirus GE (Vesikari≥11)
2
Matlab, Bangladesh [21]
42.3%c (24–56)
n/a
3
Multiple sites in India [10,11]
56.3% (36.7–69.9)
48.9% (17.4–68.4)
2
Asia
RotaTeq®
90% Rotarix®
ROTAVAC® a b c d
∼95
99d
8, 13
7, 12, 16
Individual
3
K.M. Neuzil et al. / Vaccine 32S (2014) A179–A184
Concomitant OPVa use (%)
Oral polio vaccine. Gastroenteritis. Total effectiveness (direct + indirect). Personal communication.
A183
A184
K.M. Neuzil et al. / Vaccine 32S (2014) A179–A184
to the small number of cases that occurred in the first year of life, the ITT point estimate of 42.7% (95% CI −124.7 to 87.7) is more in line with the point estimates of efficacy from the sites in Ghana and Kenya that were part of that multicenter trial [5]. As we do not yet have a complete understanding of the protective mechanism of rotavirus vaccines in low-resource settings, additional factors that are not yet understood or easily measured could also affect trial results. In Table 2, realizing that all factors may not be fully delineated or reported, the studies of rotavirus vaccines in low-resource settings, including the recent results from the ROTAVAC® efficacy trial conducted in India [10,11], are categorized by important design characteristics. For the major variables of age, use of OPV, outcome definition, and type of randomization, the ROTAVAC® efficacy trial design is similar to the design of the individually randomized RotaTeq® and Rotarix® studies. Acknowledging uncertainties that may still remain in the implementation of the study outcome measurement, differing rotavirus seasonality and epidemiology between sites, and other unknown differences between populations, the point estimates of efficacy from the recently reported ROTAVAC® trial compare quite favorably to those from Rotarix® and RotaTeq® trials in low-resource settings. While the RotaTeq® trial in Asia was designed and conducted as a multicenter trial in Bangladesh and Vietnam, we also present the estimates for the two sites separately, in order to provide what we hypothesize to be the most relevant comparisons to the ROTAVAC® trial in India. In the RotaTeq® trial, the point estimates for efficacy against severe rotavirus gastroenteritis in the first year of life were 51.0% (95% CI 12.8–73.3) for the entire cohort, 45.7% (95% CI −1.2 to 71.9) for the Bangladesh cohort and 72.3% (−45.2 to 97.2) for the Vietnam cohort. The ROTAVAC® point estimate of efficacy for the same outcome in the first year of life was 56.4% (95% CI 36.7–69.9). The apparent maintenance of efficacy in the second year of life in the ROTAVAC® trial is encouraging, and similar to what was seen in the RotaTeq® trial in Asia, recognizing that point estimates of efficacy in the second year of life are less precise, given the smaller number of outcomes. This is indeed an exciting time for rotavirus vaccines. Ultimately, multiple safe and efficacious choices should allow for optimal price and supply conditions, resulting in maximal numbers of children vaccinated. Head-to-head comparisons of different vaccines would be the best way to control for study design and population differences, and may be more common in the future given the global roll-out of rotavirus vaccines. In the meantime, this proposed framework should be useful in comparing efficacy estimates of new rotavirus vaccines conducted with placebo controls in various settings. We have proposed important design elements to be considered in those comparisons, including age at receipt of vaccine; co-administration of other vaccines, most notably OPV; definition and method of ascertainment of outcome measure; inclusion and exclusion criteria; and the pattern of rotavirus circulation. Ultimately, vaccine choices by individual countries are unlikely to be based on efficacy alone, and will include considerations of rotavirus disease burden, vaccine safety, cost and feasibility. Conflict of interest None reported.
References [1] World Health Organization. Meeting of the immunization strategic advisory group of experts, April 2009—conclusions and recommendations. Wkly Epidemiol Rec 2009;84(23):220–36. [2] PATH. Country introductions of rotavirus vaccines. Maps and list, Figure 1; 2014. Available at: http://sites.path.org/rotavirusvaccine/rotavirusadvocacy-and-communications-toolkit/country-introduction-maps-and-list/ [accessed 30.01.14]. [3] Madhi SA, Cunliffe NA, Steele D, Witte D, Kirsten M, Louw C, et al. Effect of human rotavirus vaccine on severe diarrhea in African infants. New Engl J Med 2010;362(4):289–98. [4] Zaman K, Dang DA, Victor JC, Shin JC, Yunus M, Dallas M, et al. Efficacy of pentavalent rotavirus vaccine against severe rotavirus gastroenteritis in infants in developing countries in Asia: a randomised, double-blind, placebo-controlled trial. Lancet 2010;376(9741):615–23. [5] Armah GE, Sow SO, Breiman RF, Dallas M, Tapia M, Feikin D, et al. Efficacy of pentavalent rotavirus vaccine against severe rotavirus gastroenteritis in infants in developing countries in sub-Saharan Africa: a randomised, double-blind, placebo-controlled trial. Lancet 2010;376(9741):606–14. [6] Neuzil KM, Parashar UD, Steele AD. Rotavirus vaccines for children in developing countries: understanding the science, maximizing the impact, and sustaining the effort. Vaccine 2012;30(Suppl. 1):A1–2. [7] Glass RI, Parashar U, Patel M, Gentsch J, Jiang B. Rotavirus vaccines: successes and challenges. J Infect 2014;68(Suppl. 1):S9–18. [8] Patel M, Shane AL, Parashar UD, Jiang B, Gentsch JR, Glass RI. Oral rotavirus vaccines: how well will they work where they are needed most? J Infect Dis 2009;200(Suppl. 1):S39–48. [9] Armah GE, Breiman RF, Tapia MD, Dallas M, Neuzil K, Binka F, et al. Immunogenicity of the pentavalent rotavirus vaccine in African infants. Vaccine 2012;30(Suppl. 1):A86–93. [10] Bhandari N, Rongsen-Chandola T, Bavdekar A, John J, Antony K, Taneja S, et al. Efficacy of a monovalent human-bovine (116E) rotavirus vaccine in Indian infants: a randomised double blind placebo controlled trial. Lancet 2014:62630–6. S0140-6736(13). [11] Bhandari N, Rongsen-Chandola T, Bavdekar A, John J, Antony K, Taneja S, et al. Efficacy of a monovalent human-bovine (116E) rotavirus vaccine in Indian children in the second year of life. Vaccine 2014 (in press). [12] Zaman K, Sack DA, Yunus M, Arifeen S, Popper G, Azim T, et al. Successful co-administration of a human rotavirus and oral poliovirus vaccines in Bangladeshi infants in a 2-dose schedule at 12 and 16 weeks of age. Vaccine 2009;27(9):1333–40. [13] Patel M, Steele AD, Parashar UD. Influence of oral polio vaccines on performance of the monovalent and pentavalent rotavirus vaccines. Vaccine 2012;30(Suppl. 1):A30–5. [14] Ciarlet M, Sani-Grosso R, Yuan G, Lui G, Heaton P, Gottesdiener K, et al. Concomitant use of the oral pentavalent human-bovine reassortant rotavirus vaccine and oral poliovirus vaccine. Pediatr Infect Dis J 2008;27(10):874–80. [15] Madhi SA, Kirsten M, Louw C, Bos P, Aspinall S, Bouchenooghe A, et al. Efficacy and immunogenicity of two or three dose rotavirus-vaccine regimen in South African children over two consecutive rotavirus-seasons: a randomized, double-blind, placebo-controlled trial. Vaccine 2012;30(Suppl. 1):A44–51. [16] Cunliffe NA, Witte D, Ngwira B, Todd S, Bostock N, Turner A, et al. Efficacy of human rotavirus vaccine against severe gastroenteritis in Malawian children in the first two years of life. Vaccine 2012;30(Suppl 1:):A36–43. [17] Lewis KD, Dallas MJ, Victor JC, Ciarlet M, Mast T, Ji M, et al. Comparison of two clinical severity scoring systems in two multi-center, developing country rotavirus vaccine trials in Africa and Asia. Vaccine 2012;30(Suppl. 1):A159–66. [18] Feikin DR, Laserson KF, Ojwando J, Nyambane G, Ssempijja V, Audi A, et al. Efficacy of pentavalent rotavirus vaccine in a high HIV prevalence population in Kenya. Vaccine 2012;30(Suppl. 1):A52–60. [19] Steele AD, Patel M, Parashar UD, Victor JC, Aguado T, Neuzil KM. Rotavirus vaccines for infants in developing countries in Africa and Asia: considerations from a world health organization-sponsored consultation. J Infect Dis 2009;200(Suppl. 1):S63–9. [20] Breiman RF, Zaman K, Armah G, Sow S, Anh D, Victor J, et al. Analyses of health outcomes from the 5 sites participating in the Africa and Asia clinical efficacy trials of the oral pentavalent rotavirus vaccine. Vaccine 2012;30(Suppl. 1):A24–9. [21] Zaman K. Introduction of a live, oral human rotavirus vaccine (Rotarix) in Matlab, Bangladesh. In: International Rotavirus Symposium. 2012. [22] Nelson A, Glass R. Rotavirus: realizing the potential of a promising vaccine. Lancet 2010;376(9741):568–70. [23] Sow S, Tapia M, Haidara F, Ciarlet M, Diallo F, Kodio M, et al. Efficacy of the oral pentavalent rotavirus vaccine in Mali. Vaccine 2011;30(1):A71–8.