Safety and immunogenicity of a parenteral trivalent P2-VP8 subunit rotavirus vaccine: a multisite, randomised, double-blind, placebo-controlled trial

Background: A monovalent, parenteral, subunit rotavirus vaccine was well tolerated and immunogenic in adults in the USA and in toddlers and infants in South Africa, but elicited poor responses against heterotypic rotavirus strains. We aimed to evaluate safety and immunogenicity of a trivalent vaccine formulation (P2-VP8-P[4],[6],[8]). Methods: A double-blind, randomised, placebo-controlled, dose-escalation, phase 1/2 study was done at three South African research sites. Healthy adults (aged 18–45 years), toddlers (aged 2–3 years), and infants (aged 6–8 weeks, ≥37 weeks’ gestation, and without previous receipt of rotavirus vaccination), all without HIV infection, were eligible for enrolment. In the dose-escalation phase, adults and toddlers were randomly assigned in blocks (block size of five) to receive 30 μg or 90 μg of vaccine, or placebo, and infants were randomly assigned in blocks (block size of four) to receive 15 μg, 30 μg, or 90 μg of vaccine, or placebo. In the expanded phase, infants were randomly assigned in a 1:1:1:1 ratio to receive 15 μg, 30 μg, or 90 μg of vaccine, or placebo, in block sizes of four. Participants, parents of participants, and clinical, data, and laboratory staff were masked to treatment assignment. Adults received an intramuscular injection of vaccine or placebo in the deltoid muscle on the day of randomisation (day 0), day 28, and day 56; toddlers received a single injection of vaccine or placebo in the anterolateral thigh on day 0. Infants in both phases received an injection of vaccine or placebo in the anterolateral thigh on days 0, 28, and 56, at approximately 6, 10, and 14 weeks of age. Primary safety endpoints were local and systemic reactions (grade 2 or worse) within 7 days and adverse events and serious adverse events within 28 days after each injection in all participants who received at least one injection. Primary immunogenicity endpoints were analysed in infants in either phase who received all planned injections, had blood samples analysed at the relevant timepoints, and presented no major protocol violations considered to have an effect on the immunogenicity results of the study, and included serum anti-P2-VP8 IgA, IgG, and neutralising antibody geometric mean titres and responses measured 4 weeks after the final injection in vaccine compared with placebo groups. This trial is registered with ClinicalTrials.gov, NCT02646891. Findings: Between Feb 15, 2016, and Dec 22, 2017, 30 adults (12 each in the 30 μg and 90 μg groups and six in the placebo group), 30 toddlers (12 each in the 30 μg and 90 μg groups and six in the placebo group), and 557 infants (139 in the 15 μg group, 140 in the 30 μg group, 139 in the 90 μg group, and 139 in the placebo group) were randomly assigned, received at least one dose, and were assessed for safety. There were no significant differences in local or systemic adverse events, or unsolicited adverse events, between vaccine and placebo groups. There were no serious adverse events within 28 days of injection in adults, whereas one serious adverse event occurred in a toddler (febrile convulsion in the 30 μg group) and 23 serious adverse events (four in placebo, ten in 15 μg, four in 30 μg, and five in 90 μg groups) occurred among 20 infants, most commonly respiratory tract infections. One death occurred in an infant within 28 days of injection due to pneumococcal meningitis. In 528 infants (130 in placebo, 132 in 15 μg, 132 in 30 μg, and 134 in 90 μg groups), adjusted anti-P2-VP8 IgG seroresponses (≥4-fold increase from baseline) to P[4], P[6], and P[8] antigens were significantly higher in the 15 μg, 30 μg, and 90 μg groups (99–100%) than in the placebo group (10–29%; p<0·0001). Although significantly higher than in placebo recipients (9–10%), anti-P2-VP8 IgA seroresponses (≥4-fold increase from baseline) to each individual antigen were modest (20–34%) across the 15 μg, 30 μg, and 90 μg groups. Adjusted neutralising antibody seroresponses in infants (≥2·7-fold increase from baseline) to DS-1 (P[4]), 1076 (P[6]), and Wa (P[8]) were higher in vaccine recipients than in placebo recipients: p<0·0001 for all comparisons. Interpretation: The trivalent P2-VP8 vaccine was well tolerated, with promising anti-P2-VP8 IgG and neutralising antibody responses across the three vaccine P types. Our findings support advancing the vaccine to efficacy testing. Funding: Bill & Melinda Gates Foundation. This phase 1/2, double-blind, randomised, placebo-controlled, descending-age, dose-escalation trial consisted of a dose-escalation phase and an expanded phase. In the dose-escalation phase, 30 μg then 90 μg doses of the vaccine were assessed for safety, tolerability, and immunogenicity in adults, followed by assessment of the same doses in toddlers and 15 μg, 30 μg, and 90 μg doses in infants (appendix p 1). The expanded phase recruited additional infants and evaluated all three doses of the vaccine. In the dose-escalation phase, progression from one dose to the next and from one age group to the next required review by a safety review committee of safety data up to 7 days following the first injection at each dose or in each age group (appendix pp 2–4). The expanded-cohort phase was done after completion of enrolment into the dose-escalation phase and safety assessment of each dose (appendix pp 7, 14). The dose-escalation phase was done at the Respiratory and Meningeal Pathogens Research Unit (RMPRU; Johannesburg, South Africa), and the expanded-cohort phase was done at the RMPRU, the Wits Reproductive Health and HIV Institute Shandukani Research Centre (Johannesburg, South Africa), and the Family Clinical Research Unit (Cape Town, South Africa). The protocol (appendix p 20) was approved by the Human Research Ethics Committee of the University of the Witwatersrand (Johannesburg, South Africa), Stellenbosch University Health Research Ethics Committee (Cape Town, South Africa), the Western Institutional Review Board (Puyallup, WA, USA), and the South African Health Products Regulatory Authority (Pretoria, South Africa), and was done under a US Food and Drug Administration investigational new drug application. Eligibility was assessed through medical history, physical examination, and screening laboratory tests. Healthy adults (aged 18–45 years), toddlers (aged 2–3 years), and infants (aged 6–8 weeks, ≥37 weeks' gestation, and without previous receipt of rotavirus vaccination), all without HIV infection, were eligible for enrolment. Exclusion criteria included acute illness, pregnant or breastfeeding women, presence of malnutrition or any systemic disorder that would compromise the participant's health or result in non-conformance to the protocol, congenital disorders, known or suspected impaired immunological function based on medical history and physical examination, immunoglobulin therapy or chronic immunosuppressant medications, a clinically significant screening laboratory value, HIV infection, and concurrent participation in another clinical trial. Full inclusion and exclusion criteria are listed in the appendix, pp 1–2. Investigators used their clinical judgment in considering a participant's overall fitness for inclusion in the trial. All adult participants were literate and provided written informed consent. Children were enrolled if their parents were literate and provided written informed consent. At RMPRU, adults were invited for screening by advertisements in the Soweto community, and parents of toddlers identified from hospital birth registers and infants identified in postnatal wards were invited to bring their children to RMPRU for screening. At the Family Clinical Research Unit, pregnant women were informed of the study at antenatal clinics and followed up at the obstetric unit and labour ward. At Shandukani Research Centre, pregnant women at antenatal clinics and mothers of infants attending the day 3 postnatal visit or 6-week vaccination visit were approached. In the dose-escalation phase, adults and toddlers were randomly assigned to receive 30 μg or 90 μg of vaccine, or placebo; infants were randomly assigned to receive 15 μg, 30 μg, or 90 μg of vaccine, or placebo (appendix p 1). Permuted block randomisation was used throughout. Adults and toddlers were randomly assigned to receive the vaccine or placebo in groups of 15, using three blocks of five participants (four vaccine, one placebo) per group. Infants in the dose-escalation phase were randomly assigned in groups of 16: four blocks of four infants (three vaccine, one placebo) per group. In the expanded phase, infants were randomly assigned in a 1:1:1:1 ratio to the 15 μg, 30 μg, or 90 μg dose groups or placebo in block sizes of four. The randomisation sequence was computer-generated and maintained by the Statistical and Data Management Group at The Emmes Corporation (Rockville, MD, USA). A masked study investigator enrolled participants, who were then randomly assigned electronically. An unmasked pharmacist prepared and dispensed the injection, which was masked using an opaque sticker and administered by the masked study investigator. Participants, parents of participants, and clinical, data, and laboratory staff were masked to treatment assignment. Any deviation from the protocol was reported. In the dose-escalation phase, adults received an intramuscular injection of vaccine or placebo in the deltoid on the day of randomisation (day 0), day 28, and day 56; toddlers received a single injection of vaccine or placebo in the anterolateral thigh on day 0. Infants in both the dose-escalation phase and the expanded phase received an injection of vaccine or placebo in the anterolateral thigh on day 0, day 28, and day 56, which roughly corresponded to ages 6, 10, and 14 weeks. The trivalent P2-VP8 vaccine was manufactured and supplied by the Walter Reed Army Institute of Research Pilot Bioproduction Facility (Silver Spring, MD, USA), as described previously.19 Vaccine, formulated as a sterile suspension containing 360 μg of protein—120 μg of each VP8 antigen derived from P[4] (DS-1), P[6] (1076), and P[8] (Wa) rotavirus strains—per mL adsorbed to aluminium hydroxide (Alhydrogel, Brenntag Biosector, Frederikssund, Denmark; 1·125 mg of aluminium per mL), was diluted with aluminium hydroxide diluent (1·125 mg/mL) within 6 h of administration to yield dose concentrations of 15 μg, 30 μg, and 90 μg per 0·5 mL containing 0·56 mg aluminium hydroxide. Sterile saline was used as placebo. In infants, two additional vaccines were given in the opposite thigh to the P2-VP8 vaccine or placebo: Hexaxim (Sanofi Pasteur, France), a diptheria, tetanus, pertussis, poliovirus, hepatitis B virus, and Haemophilus influenzae type B vaccine, was given at 6, 10, and 14 weeks of age; and Prevnar 13 (Pfizer, USA), a 13-valent pneumococcal conjugate vaccine, was given at 6 and 14 weeks of age. All infants in both phases received three oral doses of Rotarix, one each at 4, 8, and 12 weeks after the third study injection. Participants were observed for 30 min after the administration of each injection. Local symptoms (injection site pain or tenderness, redness, swelling, and itching) and systemic symptoms (fever, headache, vomiting, nausea, fatigue, chills, and myalgia in adults, and fever, vomiting, poor appetite, irritability, and decreased activity in toddlers and infants) were recorded daily for 7 days following each injection. Clinic visits were done 7 days after each injection in adults, and 3 and 7 days after each injection in toddlers and infants. Unsolicited adverse events were recorded from randomisation until the final study visit, 6 months after the last injection. Adverse events were graded by investigators from mild (grade 1) to life threatening (grade 4) using a grading scale developed on the basis of the Division of AIDS Table for Grading the Severity of Adult and Pediatric Adverse Events, version 2.0, November 2014, of the US National Institutes of Health, with modifications to reflect local population norms (appendix pp 2–4). Safety data were reviewed by the safety review committee and an independent data safety and monitoring board periodically throughout the study. Haemoglobin, white blood cell count, platelet count, total bilirubin, creatinine, and alanine transaminase were measured at baseline in all participants in both phases and 7 days after the first injection in the adult, toddler, and infant dose-escalation cohorts. Serum albumin was assessed at baseline only. A serum pregnancy test at screening and a urine pregnancy test before each injection was done for adult female participants. Serum was collected at baseline and 4 weeks after the final injection in all participants, as well as 4 weeks after the first and second injections in adults and 4 weeks after the second injection in infants. Anti-P2-VP8 IgG and IgA against P[4], P[6], and P[8] antigens were quantitated using standard ELISA assay techniques.19 Neutralising antibodies to DS-1 (G2P[4]), 1076 (G2P[6]), and Wa (G1P[8]) rotavirus strains were measured as previously described.23 Details of the serological testing are provided in appendix p 5. Serological testing was done at Division of Infectious Diseases, Cincinnati Children's Hospital Medical Center, Cincinnati, OH, USA. Stool was collected from infants 5, 7, and 9 days after the first dose of Rotarix in the subset of infants enrolled at the RMPRU and tested for the presence of rotavirus using the commercially available ProsPecT Rotavirus Microplate Assay (Oxoid, Ely, UK), according to the manufacturer's instructions. ELISA-positive specimens were confirmed and genotyped by PCR amplification of the VP7 and VP4 genes, as previously described.20 Stool testing was done at the National Institute for Communicable Diseases (Johannesburg, South Africa). The primary safety endpoints in all three age groups were the number of serious adverse events and adverse events up to 28 days after the last injection and the number of local and systemic reactions (grade 2 or worse) during the 7 days after vaccination in vaccine recipients compared with placebo recipients. The primary immunogenicity endpoints were the proportion of infants with anti-P2-VP8 IgG and IgA seroresponses (at least a 4-fold increase in antibody titres between baseline and 4 weeks after the third study injection) for each of the three vaccine antigens (P[4], P[6], and P[8]); the proportion of infants with neutralising antibody responses (at least a 2·7-fold increase in antibody titres between baseline and 4 weeks after the third study injection) to each of the three rotavirus strains (DS-1, 1076, and Wa) from which the vaccine antigens were derived; the proportion of infants with neutralising antibody responses to at least two of the three strains from which the vaccine antigens were derived; and the change in geometric mean titres (GMTs) of anti-P2-VP8 IgG, IgA, and neutralising antibodies from baseline to 4 weeks after the third injection in infants. A 4-fold increase in neutralising antibody responses to each of the three rotavirus strains between baseline and 4 weeks after the third study injection was evaluated in infants as an exploratory endpoint. Secondary safety endpoints were the number of serious adverse events and adverse events up to 6 months after the last vaccination. Secondary immunogenicity endpoints were anti-P2-VP8 IgG, IgA, and neutralising antibody responses and GMTs between baseline and 4 weeks after the second injection in infants or after the final injection in adults and toddlers (appendix p 6). Assessments of immunogenicity in adults at timepoints other than 4 weeks after the final injection were exploratory. Also as an exploratory endpoint, we assessed the proportion of infants enrolled at the RMPRU site who tested rotavirus-positive on an ELISA stool test at any time (5, 7, or 9 days) after administration of the first dose of Rotarix. For the adult and toddler cohorts, 12 vaccine recipients per dose provided a greater than 90% chance of observing an adverse event that had a frequency of 17·5%, and 24 vaccine recipients for the two doses combined provided a greater than 90% chance of observing an adverse event that had a frequency of 9·2%. In the infant cohorts, the 150 vaccine recipients planned per dose provided a greater than 90% chance of observing an adverse event that had a frequency of 1·6%, and the 450 vaccine recipients planned for the three dose groups combined provided a greater than 90% chance of observing an adverse event that had a frequency of 0·5%. On the basis of the results in South African infants who received monovalent P2-VP8 vaccine or placebo, the strain-specific seroresponse rates were expected to be 80% or more for at least one of the three P2-VP8 vaccine doses and less than 20% for the placebo group.20 For the infant cohort, 135 evaluable vaccine recipients per dose (assuming 10% loss of study participants due to drop-out) provided at least 74% power (Fisher's exact test) to detect a difference of 15 percentage points and at least 95% power to detect a difference of 20 percentage points in seroresponse rates between any two dose groups (ie, assuming true rates of 65% vs 80% and 60% vs 80%), and at least 99% power to detect a difference of 30 percentage points or more between a vaccine group and the combined placebo groups. Safety was analysed by treatment received, and the safety population included all participants in the dose-escalation and expansion cohorts who were randomly assigned and received at least one dose of vaccine or placebo. Immunogenicity was assessed in the per-protocol population, which included all randomly assigned participants who received all planned injections, had blood samples analysed at the relevant timepoints, and presented no major protocol violations considered to have an effect on the immunogenicity results of the study. Categorical results (serum IgA, IgG, and neutralising antibody seroresponses) are presented as frequency, proportion (%), and exact two-sided binomial (Clopper-Pearson) 95% CI. Continuous outcomes (serum IgA, IgG, and neutralising antibody responses) are presented as GMT and two-sided 95% CI obtained from the t-distribution on log-transformed titres. To compare safety outcomes between treatment groups, we used Fisher's exact two-tailed test, or χ2 test if the expected number of events was sufficient (all expected cell frequencies equal to five or more). The binomial immunogenicity response variables were compared between each dose group and placebo with logistic regression, and p values from prespecified pairwise comparisons are reported (if there was a significant difference overall between groups, then pairwise comparison was done). Continuous immunogenicity variables (GMTs) were modelled by ANOVA. To adjust for decay in maternal antibody occurring concurrently with IgG and neutralising antibody immune responses to the vaccine in infants, we did an analysis of the adjusted seroresponse rates using the exponential decay function based on the estimated half-life (for each immunogenicity assay separately) of maternal antibodies in infants in the placebo group who had detectable baseline titres that were higher than at the post-injection visit. Adjusted seroresponse was defined as at least a 4-fold increase in titre between baseline and 4 weeks after the third injection (adjusted titre) in infants with an unadjusted post-injection titre greater than the limit of detection. Shedding of Rotarix virus was assessed for each of the three specified post-vaccination days (5, 7, or 9 days) and for any of the three days. Proportions of infants with shedding were compared between the placebo group and each dose group and between the placebo group and all dose groups combined using Fisher's exact test. Data were analysed with SAS software (version 9.3), and statistical significance was defined as a two-tailed p value of less than 0·05. This trial is registered with ClinicalTrials.gov, {"type":"clinical-trial","attrs":{"text":"NCT02646891","term_id":"NCT02646891"}}NCT02646891. The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.