We describe five TB studies conducted serially at our site and the lessons learnt in each.
1. Adolescent Cohort Study (ACS)
The ACS began in 2007, with the objective of recruiting adolescents to establish the optimal way to access them in the study area and determine TB prevalence and incidence in preparation for TB vaccine trials. A mobile field site (MFS) was used to enroll adolescents. (Supplementary Figure 5-Mobile Field Site). The MFS was a collection of tents, a mobile radiography unit and mobile generator pitched in a grassy open field in a school, church compound or other open public arena. Potential MFS sites were identified by study staff, and the site assembled every week. The tents housed study staff, parents and adolescents who moved serially from consenting, TB symptom screening, TST administration or reading, to chest radiography and then to dispatch. Data entry was on Personal Digital Assistants and laptops, using a local area network powered by the mobile generator. Upon exhausting all potentials at that catchment area, the site would demobilize and set up in a contiguous site and the process would start over again. Follow-up visits were conducted in similar fashion. The challenges of operating in an open field were particularly apparent during the rainy seasons. The site would get flooded or barely accessible, disrupting study processes. In that case it would be disassembled and relocated. A total of 5004 adolescents were recruited on target over 12 months and 83% retained (4). Most of this team rolled over into the infant study.
2. Infant Cohort Study (ICS)
Pediatric TB estimates from program data were not age-disaggregated, masking the disproportionate affliction of young infants with TB. In 2008, the thinking of the TB vaccine community leaned heavily towards infants as a target population. Therefore, the ICS was conceived to provide estimates of TB incidence using comprehensive diagnostic methods (5), and to demonstrate capacity to deliver BCG within 96 hours of birth. This was to anticipate the administration of BCG replacement candidate vaccines.
At the time, it was reported that 80% of mothers delivered at home, with the help of a traditional birth attendant (TBA). We engaged TBAs to assist in notifying births to the study staff. The recruitment coordinator would then dispatch a nurse on a motorbike to perform enrolment procedures. Thereafter, a nurse would deliver BCG to enrolled infants. The median age of enrollees was 10 days, therefore in many cases it was not possible to deliver BCG within 96 hours of birth due to late notifications and problems locating mothers who were highly mobile in the study area. This was accentuated when due to insufficient notifications, the study area was extended to Boro, a nearby sub-county, which was not under the HDSS.
In addition, the terrain was frequently challenging. Nurses were inexperienced motorbike riders and predominantly female. There were several falls that injured their morale more than anything. Nevertheless, the team rallied, and 2900 infants were enrolled on target within 12 months.
Fortunately, the number of home births in the study area continued to decline. By 2011, at least 40% occurred at health facilities (6). Future trials of BCG replacement vaccine trials can therefore enroll infants at health facilities.
The ICS operated out of two tents in a grassy field of the Siaya Hospital, seeing scheduled and unscheduled visitors. The hospital allocated a store in the pediatric ward for creation of a negative pressure room for specialized sputum collection procedures. We renovated a defunct, dilapidated amenity wing designed for private patients, and converted into a case verification ward for overnight admissions for TB investigation procedures.
The existing hodgepodge of facilities did not meet trial requirements, as there was no GCP compliant pharmacy, archival facilities, or laboratories. Therefore, in 2009, construction of a state-of-the-art research annex (Supplementary Figure 4-Siaya Clinical Research Annex) with all the requisite infrastructure at the hospital grounds began, funded by several partners. It was completed in 2012 and auspiciously launched.
The study also created a previously non-existent capacity to diagnose pediatric TB using induced sputa and gastric aspirates as well as structured chest radiograph review.
There were thousands of ancillary care visits that contributed the largest number of presumptive TB cases, given the non-specific presentation of TB in infants.
3. Aeras-402 (Nochak)
The first TB vaccine trial at the site followed closely on the heels of the ICS. It was a phase IIb trial of AERAS 402 in healthy, HIV uninfected, BCG vaccinated infants (7). The site was the first to be initiated into the trial in 2010. The study team was now proficient in recruitment of infants and requisite community links had been forged. There was widespread enthusiasm for study participation due to intense community mobilization by site staff. Locals had branded the study ‘NoChaK’ (Nonro mar Chanjo mar Kahera), translated ‘a new TB vaccine’ in Luo, the local language. (Supplementary Figure 2-Community Engagement)
Disappointment ensued when there was an inordinate number of screen failures due to abnormal biochemistries, particularly elevated total bilirubin in infants who were in every other sense healthy. To determine eligibility, the National Institutes of Health (NIH) reference ranges were used, which, in hindsight, were clearly derived from a distinct population. It would have been unethical to continue screening, and therefore enrolment was paused to investigate the issue. To exclude invalid laboratory results, the local laboratory that was also being used for the RTS,S malaria vaccine trial, was scrutinized to determine the reproducibility and repeatability of affected parameters, as well as External Quality Assurance (EQA) program results. The laboratory cleared. We found no antecedent literature describing what we were observing. Following wide consultations, we developed a new set of reference ranges. This allowed the study to proceed with enrolment. The area was hyperendemic for malaria and that was considered the prime reason for the elevations in healthy infants. Years later, two publications emerged on the subject, but for an adult population proximal to the study area, which mirrored our observations (8, 9).
There were significant problems with collection and processing of Peripheral Blood Mononuclear Cells (PBMCs). The cells needed to be separated and frozen within 4 hours of collection. The study site was about 80 km away (about 1.5-hour drive) from the immunology laboratory, with predictable inaccessibility in wet weather. The process of separation took at least two hours. In addition, eight ml of blood were collected in several tall vials. Like many African communities, there were strong sensitivities to blood sampling, despite our best efforts to explain that the relative blood volumes sampled would not adversely impact the babies given their age and weight. Furthermore, as infants were significantly dark skinned and healthy, if more than one attempt was needed to collect these samples, parents would get distressed and the time taken also derailed the tight separation timelines. In response, tight orchestration of the visits with PBMC collection was done. An experienced phlebotomist, efficient in cutting down the collection time was added to the team. Technicians were alerted once the sample left and were on stand-by ready to immediately begin to process the samples.
Unfortunately, the laboratory team had no prior immunology experience. They failed proficiency testing, some on more than one occasion and this skill was only gradually acquired. In addition, on one occasion the dry ice shipper containing EQA samples arrived at the destination laboratory dry, compromising cell viability. The laboratory teams were retrained, and this capacity was then available for the future trial among adults.
The stability of AERAS 402 vaccine at 2-8 degrees was short-lived. It was stored at -80 0C in a research pharmacy in Kisumu (Supplementary Figure 3-Logistics). Twenty-four hours prior to dosing, it would be transferred from the freezer to the refrigerator and could keep for only six weeks. There were frequent power outages at the clinic, therefore it was not possible to safely keep vaccine on site. Therefore, an alternative thaw protocol was used. Vials would be removed from the freezer, thawed at room temperature for one to two hours, and then shipped to site under rigorous temperature monitoring. After review of potential participants, the study doctor would call up the vaccine from the central pharmacy, and vaccinations would happen upon arrival two hours later. As there were mandatory post-vaccination observation procedures, many mothers were dropped off at their homes late in the night, which created friction in families. Mothers agreed to assemble for early pick-ups on days of randomization, and thus there was only one vaccine run to Kisumu.
4. TB018 (M72)
The second vaccine trial in this site was a phase IIb trial of M72/AS01E against placebo (10). The site’s successful experience with the MFS made it a compelling recruitment strategy for this trial. A total of 538 individuals were randomized over a period of eight months. The laboratory was now experienced in handling immunology samples and there were few hitches.
5. VPM1002
In 2018, towards the end of the M72 study, our site was selected to participate in a phase III trial of VPM1002 against BCG among infants. The cumulative site experience in immunology, infant recruitment, TB case detection, and adverse event surveillance continues to be applied for this trial. Recruitment is expected to end by the second quarter of 2022, with a minimum follow-up time of 12 months.