Investigating the non-specific effects of BCG vaccination on the innate immune system in Ugandan neonates: Study protocol for a randomised controlled trial

Background: The potential for Bacillus Calmette-Guérin (BCG) vaccination to protect infants against non-mycobacterial disease has been suggested by a randomised controlled trial conducted in low birth-weight infants in West Africa. Trials to confirm these findings in healthy term infants, and in a non-West African setting, have not yet been carried out. In addition, a biological mechanism to explain such heterologous effects of BCG in the neonatal period has not been confirmed. This trial aims to address these issues by evaluating whether BCG non-specifically enhances the innate immune system in term Ugandan neonates, leading to increased protection from a variety of infectious diseases. Methods: This trial will be an investigator-blinded, randomised controlled trial of 560 Ugandan neonates, comparing those receiving BCG at birth with those receiving BCG at 6 weeks of age. This design allows comparison of outcomes between BCG-vaccinated and -naïve infants until 6 weeks of age, and between early and delayed BCG-vaccinated infants from 6 weeks of age onwards. The primary outcomes of the study will be a panel of innate immune parameters. Secondary outcomes will include clinical illness measures. Discussion: Investigation of the possible broadly protective effects of neonatal BCG immunisation, and the optimal vaccination timing to produce these effects, could have profound implications for public healthcare policy. Evidence of protection against heterologous pathogens would underscore the importance of prioritising BCG administration in a timely manner for all infants, provide advocacy against the termination of BCG’s use and support novel anti-tuberculous vaccine strategies that would safeguard such beneficial effects.

Infants will be recruited on the day of birth from the maternity ward of Entebbe Grade B hospital, a government hospital located in Wakiso District in central Uganda. The region is populated mainly by semi-urban, rural and fishing communities. Neonatal mortality rates in Uganda remain high at 28/1,000 live births, with a large proportion attributable to invasive infectious diseases. The inclusion criteria for this study are: No specific weight or gestational age limit has been set for this study. Clinical responses to early BCG are suggested to have the greatest effect in infants of the lowest birth weight [18]; thus, it is important to include these infants in data collection. No increased rate of detrimental side-effects or reduction of immunological efficacy has been shown with BCG immunisation of premature infants [48]. Written informed consent will be obtained from the mothers of all infants prior to their enrolment in the study. Neonates will be excluded from the study if: All infants will receive 0.05 ml of BCG-Statens Serum Institute (SSI, Copenhagen, Denmark) (Danish Strain 1331) intra-dermally into the right deltoid. This will be given either at birth (Early intervention arm) or at 6 weeks of age (Delayed intervention arm). Intervention and blood sampling time-point allocation will be determined by block randomisation, stratified by sex. This will be carried out by an independent statistician, prior to the trial commencement, using Stata (StataCorp, College Station, TX, USA) to generate the allocation sequence. Allocations will be concealed within sequentially numbered, sealed opaque envelopes, prepared by two research assistants who are independent of the trial. Upon delivery of an eligible infant, assignment of allocation will be carried out by midwives who will select the next sequential envelope according to the infant’s gender. This study will be single blind. Mothers will not be blinded to intervention allocation due to lack of feasibility (BCG produces a visible reaction) and to reduce confusion if a child who is lost to follow-up presents to a community immunisation clinic. Staff involved in administering BCG immunisation to the participants, either at birth or at 6 weeks of age, will not be involved in clinical follow-up or assessment of outcomes. Investigators performing clinical assessment of children will be blinded to intervention allocation by means of a plaster placed over the area corresponding to BCG vaccination site. This will be placed by a nurse not involved in clinical assessment, prior to physician assessment. If a child is presenting due to concerns about the immunisation site it will be left uncovered and the un-blinding documented. Illness events arising from concerns or complications directly related to the BCG immunisation will not be included in the analysis of illness events, but will be presented separately. Immunological investigations will be conducted on blood samples identified only by study number. The intervention allocation code will only be broken once laboratory analysis is complete and the data have been cleaned and locked. Figure 1 shows the SPIRIT (Standard Protocol Items: Recommendations for Interventional Trials) diagram for the trial procedures. On presentation to labour ward, mothers in active labour will be screened for their eligibility and informed consent will be taken. Following delivery the infant will be assessed for eligibility and placental cord blood collected. Infants who are eligible for the study will be randomised as described above, to receive BCG vaccination either immediately or at 6 weeks of age. All infants will be followed-up until 10 completed weeks of age. During this time 2 × 2 ml venous blood samples and 2 stool samples will be collected and all routine immunisations will be given (Oral Polio Vaccination (OPV) at birth and primary immunisations at 6 and 10 weeks of age). Clinical follow-up of the infants will be carried out by weekly telephone interviews to check the well-being of participants, and physician review and anthropometry at each routine clinic visit for blood samples/routine immunisations (on average four visits per participant). Unwell participants presenting to the study clinic or Entebbe Grade B hospital will also be reviewed and managed by the study team, free of charge. Study follow-up is complete once the child has completed 10 weeks of age. Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) diagram of study procedures. Sensitisation of parents to the study will occur during antenatal classes via posters, group discussions and during individual midwife-led consultations. Mothers will then be approached for consent by trained midwives when presenting in active labour to Entebbe Grade B hospital. The study will be explained in detail verbally and the information sheet provided (or read to illiterate mothers). Information sheets will be available in English and Luganda. Consent will also be taken to allow for storage of excess samples and use of data in future research studies. Although consent during labour is not optimal, it is necessary to enable cord blood collection. However, consent will be verbally reconfirmed with mothers following delivery prior to any intervention. This method of consent and recruitment has been piloted in the same hospital and shown to be an appropriate and successful method. Demographic details, anthropometric measurements and socio-economic indices will be collected at enrolment including gender, gestational age, birth weight, occipito-frontal circumference and length, maternal age and parity, parental ethnicity, parental educational level attained, accommodation type and assets. Global Positioning System (GPS) co-ordinates of the participant’s home address will also be collected to aid follow-up. During routine clinic visits anthropometric and vital sign measurements will be collected. All mothers will be interviewed about illness episodes in the participant since they were last seen in clinic and any current concerns. Physical examination findings will be documented. A standardised illness episode case report form will be completed whenever a child presents unwell to the research clinic or paediatric ward at Entebbe Grade B hospital. This will include anthropometric and vital sign measurement, symptoms and signs, investigation results, final diagnosis and outcome. All participants will be interviewed by telephone on a weekly basis by a fieldworker using a standardised case report form to ensure the health of the infant. Any infants for whom there are concerns will be reviewed in clinic. This intensive follow-up will enhance identification of clinical illness episodes, which are secondary outcomes for the study. More importantly, however, it will allow early identification and management of any cases of perinatal TB, particularly in the delayed intervention group. Any suspected or confirmed cases of TB occurring during the study will be reported to the ethics committees and Data Safety Monitoring Board (DSMB), who will decide whether the study needs to be stopped early for safety. Direct electronic data entry will occur for all case report forms. This will be verified and optimized by co-documentation with paper case report forms at the beginning of the study. Data will be maintained in encrypted, password protected forms, to maintain confidentiality. All participants in the study will have 10 ml placental cord blood collected at birth; divided into 5 ml of heparinised and 5 ml of ethylenediaminetetraacetic acid (EDTA)-anticoagulated blood. They will then have 2-ml venous blood samples collected at 2 time-points between birth and their exit from the study at 10 completed weeks of age. Each sub-study has up to four possible time-points where blood samples are collected, but each infant will only be bled at two of these time-points (randomly allocated) to avoid undue stress for the baby and the mother. The time-points have been selected to enable interrogation of the changes to the innate immune system induced by BCG both acutely following vaccination and in the longer term. The timing of the blood samples in the iron sub-study differs slightly from those in the cytokine and epigenetic sub-studies (see Table 2). These differences are necessitated by the systemic nature of iron metabolism. As hepcidin is produced mainly in the liver this precludes analysis of iron metabolism following in-vitro non-specific stimulation. Thus, the iron sub-study will use routine primary immunisations as in-vivo non-specific stimuli and measure the resulting changes to iron parameters. Blood sample time-points (T) according to immunological sub-study Stool samples will be collected at the 6-week and 10-week time-points and stored to allow for future analysis, funding permitting. Whenever an unwell participant presents to the study team investigations and treatments will be conducted as directed by the attending clinician. Investigations will include cultures for accurate diagnosis of febrile illness. An extra 2-ml blood sample will be taken from any participant under-going phlebotomy provided that this will not compromise the child’s health or well-being. This will allow a sub-study to be conducted to compare primary immunological outcomes in unwell children according to BCG status. Overnight whole blood stimulation with the non-specific stimulants S. aureus, S. pneumoniae, E. coli, C. albicans and polyinosinic:polycytidylic acid/C-phosphate-G (Poly I:C/CpG) will be carried out using fresh sodium-heparinised blood. Measurement of the pro-inflammatory cytokines IL-1β, IL-6, TNF-α and IFN-γ by ELISA (BD-OptEIA, Becton, Dickinson and Company, Oxford, UK) will then be conducted on the harvested supernatant following storage at −80°C. These stimulants have been chosen because they are the most common pathogens isolated from septic neonates in Uganda [49] and because they represent a range of pathogen types and toll-like receptor pathways. The levels of trimethylation of H3K4 at the promoter region of pro-inflammatory cytokines will be assessed using chromatin immunoprecipitation followed by qPCR. Peripheral blood mononuclear cell (PBMC) isolation for this work will occur by density-centrifugation over histopaque (Sigma-Aldrich, Dorset, UK). Measures of iron status will be conducted on the plasma fraction of lithium-heparinised blood following storage at −80°C. Serum iron, Unbound Iron Binding Capacity (UIBC), Total Iron Binding Capacity (TIBC), Transferrin Saturation (TSAT) and ferritin will be measured using the automated Cobas Integra (Roche Diagnostics, Switzerland). The hormone hepcidin will be quantified using ELISA (Bachem-25, Bachem, Switzerland). Red cell parameters will be measured from fresh EDTA whole blood using a Coulter AC.T 5 Diff CP haematology analyser (Beckman Coulter, Inc, CA, USA). Primary outcomes in each sub-study will be compared between the 2 intervention groups both acutely following BCG (up to 1 week after birth/6 weeks of age) and at time-points distant from vaccination (6 and 10 weeks of age). The above clinical outcomes for the three sub-studies will be analysed together to increase power. The iron sub-study will also have the following secondary outcomes: In a secondary analysis, longitudinal within-infant changes in primary outcomes will also be analysed for each sub-study. Each sub-study is powered for its own primary outcomes. The overall sample size is the summation of the participants required for each sub-study. Due to paucity of published data in this area, an approach based on standard deviation (SD) difference in average population cytokine levels has been used. Forty-eight subjects per intervention group (BCG immunisation at birth or at 6 weeks of age) will be needed at each time point to show a 0.66 SD difference in average population cytokine levels with 90% power and 5% significance. Sixty infants per intervention group per time point will be recruited to allow for attrition. As each recruited infant will be bled at 2 time-points, 240 infants will be recruited in total to allow for the 4 time-points. The only previous study in this area (which was performed in adults) required 20 subjects per intervention arm [47]. We will recruit 40 subjects to each intervention arm to allow for attrition and also due to the requirement for a full 2-ml blood sample for epigenetic analysis, which is unlikely to be obtained for all subjects. Due to funding constraints, epigenetic analysis will be restricted to the first two sampling time-points, and each infant will be bled at both time-points, eighty subjects will be recruited in total. Sample size determination was performed using TSAT as it is the only primary outcome parameter currently of clinical relevance. Average neonatal TSAT in low-income settings is 55% [50]. Fifty infants in each group at each time point will be needed to show a 30% reduction in transferrin saturation (reduction to average TSAT levels in low income settings) with 90% power and 5% significance. Sixty subjects will be recruited to each intervention group at each time point to allow for attrition. As each recruited infant will be bled at 2 time-points, 240 infants will be recruited in total. Combined analysis of clinical end-points from all three sub-studies will be conducted as secondary analysis. Based on data from a previous study in Entebbe [51] we expect 80% power to detect a ≥ 40% reduction in physician-diagnosed invasive infections with 5% significance. The effect of BCG is unlikely to be this pronounced, but this preliminary data combined with the primary immunological outcomes, should provide sufficient evidence to determine whether expanding the cohort would be valuable. This is a randomised controlled trial with datasets generated from clinical questionnaires and laboratory assays. A combination of direct electronic capture and paper forms will be used, linked by a unique participant identifier. Microsoft Access (Redmond, WA, USA) will be utilised to produce the study database. Data will be exported from Microsoft Access to Stata (StataCorp, College Station, TX, USA) for statistical analysis. A detailed data dictionary with range checks will be used to reduce data entry errors. Quality control checks will be run by the data clerk, on a weekly basis, who will highlight any queries to the principal investigator. Data will only be uploaded onto the master database once any queries highlighted by quality control checks have been resolved. Group characteristics will be compared using Pearson’s Chi-squared test for categorical variables and the t-test for continuous variables. Cross-sectional comparisons between intervention groups at each time-point will be carried-out using the t-test for differences between means. Non-normally distributed outcome data will be log-transformed before analysis; Mann–Whitney two-tailed test will be used for persistently skewed data. If potential confounders remain unbalanced between the groups despite randomisation: for instance season of birth, these will be adjusted for using multiple linear regression analysis. Paired/longitudinal analysis of within infant changes in parameters over time will be conducted using the paired student t-test or Wilcoxon matched-pairs test. Incidence rate of invasive infectious disease in the first 10 weeks of life will be compared by Poisson regression with a random effects model to allow for within-child clustering. Statistical significance will be assessed at the 2-sided 0.05 level but interpretation of results will not be solely reliant on P-values. This clinical trial will be conducted according to Good Clinical Practice standards. An internal study monitor will oversee the day-to-day running of the trial locally, with external oversight and monitoring co-ordinated by the London School of Hygiene and Tropical Medicine. This may include internal audit by the Clinical Trials Quality Assurance Manager and external audits by a third party. A Trial Steering Committee (TSC) and an independent DSMB have been set up for this study. The DSMB will look at a number of clinical outcome measures, documented in ‘real time’ during the study, to assess whether the study needs to be stopped early for safety. Safety reporting for this trial will follow standard Uganda Virus Research Institute and London School of Hygiene and Tropical Medicine procedures. This includes notification of Serious Adverse Events (SAEs) to the local ethics committee within 24 hours, notification of Suspected Unexpected Serious Adverse Reactions (SUSARs) to the sponsor within 7 days if life-threatening or 15 days if non-life-threatening. The manufacturer of the BCG vaccine, Staten Serum Institute, will also be notified of any SAE/SUSAR. As this trial will alter the timing of BCG from the current Ugandan guidelines (BCG at birth) in half of the study infants, a thorough risk-benefit analysis of a 6-week delay in vaccination has been conducted. In summary, we feel that the risks of delay are minimal for the following reasons: There is also evidence that delay in BCG vaccination from birth to 6 weeks may be beneficial for participants because: All infants in the study, whether in the early or delayed BCG group will benefit from regular physician reviews and free access to medical review and treatment if participants become unwell. They will also benefit from receiving all other primary vaccinations at the correct time as part of the study. The most recent survey of vaccination rates in Uganda showed that 56% of infants have not received their first set of primary immunisations (diphtheria/tetanus/pertussis/hepatitis B/Haemophilus influenzae (HiB) and oral polio vaccine) by 12 weeks of age, with 26% still not having received it by 1 year of age. This produces a substantial risk for those children of contracting serious, preventable illnesses, which participation in the study will negate. Thus, we believe the general benefits of taking part in the study will outweigh the extremely small risks from a 6-week delay in BCG vaccination. The full risk-benefit analysis for this study can be found in Additional file 1. This trial has been approved by ethics boards at the Uganda Virus Research Institute on AIDS (Ref: GC/127/13/11/432), the Uganda National Council for Science and Technology (Ref: HS 1524), The Office of the President of Uganda and the London School of Hygiene and Tropical Medicine (Ref: 6545). This study will be conducted according to the principles of the Declaration of Helsinki. The primary immunisation schedule imposes a number of constraints on the design of this study, as blood samples need to be timed to limit the potentially confounding influence of non-BCG vaccinations on innate immune responses. This is particularly relevant for comparison of the longer-term non-specific effects of BCG between the Early and Delayed intervention arms at 10 weeks, where BCG will have been given more recently in the Delayed intervention arm. As we are investigating the acute response to non-tuberculous stimulants, we believe that this should not be a problem, as any bystander effect of BCG vaccination itself is likely to be lost by 4 weeks of age. However, we are actively seeking funding for a longer-term follow-up time-point that should help to clarify this issue as well as to provide information about the duration of any non-specific effects of BCG vaccination on the innate immune system. Although it is important to understand the biological mechanism underlying any non-specific effects of BCG vaccination, ultimately the impact on all-cause clinical illness episodes and mortality will be the outcome measures that are likely to have impacts on public healthcare policy. This study has limited power to detect differences in such outcomes, due to its small sample size. However, if suggested by the immunological and preliminary clinical data in this study, additional funding will be sought to expand the cohort to allow full interrogation of clinical outcomes.

• prentice, sarah
• webb, emily l.
• dockrell, hazel m.
• kaleebu, pontiano
• elliott, alison m.
• cose, stephen