- Split View
-
Views
-
Cite
Cite
Philip C Hill, Frank Cobelens, Leonardo Martinez, Marcel A Behr, Gavin Churchyard, Tom Evans, Andrew J Fiore-Gartland, Alberto L Garcia-Basteiro, Willem Hanekom, Molebogeng X Rangaka, Johan Vekemans, Richard G White, An Aspiration to Radically Shorten Phase 3 Tuberculosis Vaccine Trials, The Journal of Infectious Diseases, Volume 228, Issue 9, 1 November 2023, Pages 1150–1153, https://doi.org/10.1093/infdis/jiad356
- Share Icon Share
Abstract
A new tuberculosis vaccine is a high priority. However, the classical development pathway is a major deterrent. Most tuberculosis cases arise within 2 years after Mycobacterium tuberculosis exposure, suggesting a 3-year trial period should be possible if sample size is large to maximize the number of early exposures. Increased sample size could be facilitated by working alongside optimized routine services for case ascertainment, with strategies for enhanced case detection and safety monitoring. Shortening enrolment could be achieved by simplifying screening criteria and procedures and strengthening site capacity. Together, these measures could enable radically shortened phase 3 tuberculosis vaccine trials.
(See the Editorial Commentary by McShane on pages 1147–9.)
Tuberculosis is responsible for over 1.5 million deaths annually [1]. The bacille Calmette-Guérin (BCG) vaccine is inadequate to facilitate tuberculosis elimination, but there are only 3 new tuberculosis vaccines (VPM1002, M72/AS01E, and MTBVAC) currently entering phase 3 prevention of disease (POD) trials. Recently, stakeholders emphasized the need to explore innovative tuberculosis vaccine designs and build trial capacity in high-incidence countries [2]. In addition, careful selection of trial end points is important to optimize trial efficiency [3].
Lessons from coronavirus disease 2019 (COVID-19) include conducting human trials and phases in parallel, mobilization of large-scale funding, harmonization of efforts, and use of efficient designs [4]. For Mycobacterium tuberculosis, the annual risk of infection is estimated to average only 2% in tuberculosis endemic settings [5], rarely reaching up to 10% [6]. Furthermore, fewer than 10% of those with new M. tuberculosis infection progress to tuberculosis disease in 5 years [7]. Phase 2b tuberculosis vaccine trials recruit around 3000 participants with up to 3 years of follow-up [8, 9]. Phase 3 trials then randomize tens of thousands of individuals with around 5 years of follow-up.
However, noting that the majority of those who progress to tuberculosis disease do so within 2 years after M. tuberculosis exposure [10], a robust phase 3 adolescent/adult POD tuberculosis vaccine trial may only need 2 to 3 years of follow-up if a sufficient number of participants are exposed to M. tuberculosis early after receiving vaccination/placebo. Here we consider options to achieve a shortened trial in relation to study size, participant characteristics, and enrolment procedures (Table 1).
Proposed Approaches . | Advantages . | Limitations . |
---|---|---|
Traditional design | ||
Enroll high-risk populations | Higher number of end point accruals per population over the study period | Nonrepresentative of ultimate target population |
May have weak immune responses leading to an underestimate of vaccine efficacy | ||
Due to known high-risk, preventive therapy may be mandatory | ||
Alternative design | ||
Enhanced routine passive case detection | Cost per tuberculosis case detected is likely to be much lower | Cases detected by routine health services may be more advanced clinically |
May more accurately measure real-world effect of vaccine | Risk of reduced case capture and underdiagnosis of incident tuberculosis cases | |
Nonreliance on TST/IGRA testing at screening | More representative of target population | Study may be suboptimally balanced in relation to the proportions with and without Mycobacterium tuberculosis infection |
TST/IGRA testing can be performed late, in batches efficacy estimates generated according to baseline infection status | Potential source of bias if vaccine efficacy is different according to baseline infection status | |
Limit preenrollment testing to symptoms and GeneXpert | Sensitivity to detect clinical tuberculosis will remain high despite reduction in testing | May affect ability to estimate vaccine efficacy against subclinical tuberculosis |
Increase site capacity | Required number of participants can be enrolled within a short period | Increased cost and logistical challenges |
Proposed Approaches . | Advantages . | Limitations . |
---|---|---|
Traditional design | ||
Enroll high-risk populations | Higher number of end point accruals per population over the study period | Nonrepresentative of ultimate target population |
May have weak immune responses leading to an underestimate of vaccine efficacy | ||
Due to known high-risk, preventive therapy may be mandatory | ||
Alternative design | ||
Enhanced routine passive case detection | Cost per tuberculosis case detected is likely to be much lower | Cases detected by routine health services may be more advanced clinically |
May more accurately measure real-world effect of vaccine | Risk of reduced case capture and underdiagnosis of incident tuberculosis cases | |
Nonreliance on TST/IGRA testing at screening | More representative of target population | Study may be suboptimally balanced in relation to the proportions with and without Mycobacterium tuberculosis infection |
TST/IGRA testing can be performed late, in batches efficacy estimates generated according to baseline infection status | Potential source of bias if vaccine efficacy is different according to baseline infection status | |
Limit preenrollment testing to symptoms and GeneXpert | Sensitivity to detect clinical tuberculosis will remain high despite reduction in testing | May affect ability to estimate vaccine efficacy against subclinical tuberculosis |
Increase site capacity | Required number of participants can be enrolled within a short period | Increased cost and logistical challenges |
Abbreviations: IGRA, interferon γ release assay; TST, tuberculin skin test.
Proposed Approaches . | Advantages . | Limitations . |
---|---|---|
Traditional design | ||
Enroll high-risk populations | Higher number of end point accruals per population over the study period | Nonrepresentative of ultimate target population |
May have weak immune responses leading to an underestimate of vaccine efficacy | ||
Due to known high-risk, preventive therapy may be mandatory | ||
Alternative design | ||
Enhanced routine passive case detection | Cost per tuberculosis case detected is likely to be much lower | Cases detected by routine health services may be more advanced clinically |
May more accurately measure real-world effect of vaccine | Risk of reduced case capture and underdiagnosis of incident tuberculosis cases | |
Nonreliance on TST/IGRA testing at screening | More representative of target population | Study may be suboptimally balanced in relation to the proportions with and without Mycobacterium tuberculosis infection |
TST/IGRA testing can be performed late, in batches efficacy estimates generated according to baseline infection status | Potential source of bias if vaccine efficacy is different according to baseline infection status | |
Limit preenrollment testing to symptoms and GeneXpert | Sensitivity to detect clinical tuberculosis will remain high despite reduction in testing | May affect ability to estimate vaccine efficacy against subclinical tuberculosis |
Increase site capacity | Required number of participants can be enrolled within a short period | Increased cost and logistical challenges |
Proposed Approaches . | Advantages . | Limitations . |
---|---|---|
Traditional design | ||
Enroll high-risk populations | Higher number of end point accruals per population over the study period | Nonrepresentative of ultimate target population |
May have weak immune responses leading to an underestimate of vaccine efficacy | ||
Due to known high-risk, preventive therapy may be mandatory | ||
Alternative design | ||
Enhanced routine passive case detection | Cost per tuberculosis case detected is likely to be much lower | Cases detected by routine health services may be more advanced clinically |
May more accurately measure real-world effect of vaccine | Risk of reduced case capture and underdiagnosis of incident tuberculosis cases | |
Nonreliance on TST/IGRA testing at screening | More representative of target population | Study may be suboptimally balanced in relation to the proportions with and without Mycobacterium tuberculosis infection |
TST/IGRA testing can be performed late, in batches efficacy estimates generated according to baseline infection status | Potential source of bias if vaccine efficacy is different according to baseline infection status | |
Limit preenrollment testing to symptoms and GeneXpert | Sensitivity to detect clinical tuberculosis will remain high despite reduction in testing | May affect ability to estimate vaccine efficacy against subclinical tuberculosis |
Increase site capacity | Required number of participants can be enrolled within a short period | Increased cost and logistical challenges |
Abbreviations: IGRA, interferon γ release assay; TST, tuberculin skin test.
STUDY SIZE AND PARTICIPANTS
One (more standard) approach to accrue the required number of tuberculosis cases without having to increase the study size is to focus enrolment on high-risk subpopulations, such as people with directly impaired immunity (eg, people with human immunodeficiency virus [HIV] have an up to 37-fold increased risk of developing tuberculosis [11] and people with diabetes a 2 times increase [12], and a positive interferon γ release assay [IGRA] 2 times increased risk vs a negative test [13]), or contacts of tuberculosis cases in the first 2 years after exposure.
An advantage of this approach is that the higher incidence of disease will accrue end points more quickly.
The limitations include that these populations may be of small size and not representative of those to which the vaccine will eventually be given. In addition, those with immunocompromise have relatively weak immune responses, potentially leading to an underestimate of vaccine efficacy. Furthermore, tuberculosis case contacts can be considered for post-, but not preexposure vaccine trials. Finally, tuberculosis preventive treatment is indicated in people with HIV and young children (plus older children and adults in some settings) who are household contacts of a tuberculosis case.
Another approach, which could enable a larger study size, is to work alongside and optimize routine health services for case detection. This is usually reserved for postlicensure phase 4 evaluations, but was part of the case-finding strategy of the Chingleput BCG trial in India [14]. Such an approach should aim to maximize case capture as well as support increased capacity to detect cases, while still requiring and ensuring robust case confirmation ascertainment according to the protocol.
An advantage of this approach is that the cost per study participant may be reduced, enabling enrolment of a larger number of individuals within a short timeframe. Improved routine passive case detection systems would strengthen the health system with better standardization of indications and diagnostic processes and quality assurance capacity building. The use of routine services would support generalizability of trial findings to real life conditions.
A limitation of this approach is that use of routine services for case capture may lead to a reduced incidence of end points accrued and identify cases that would be in general more severe. However, possible late or underdiagnosis is likely to be nondifferential, and vaccine efficacy estimates would not be affected, as high diagnostic specificity would be maintained. Extra sample collection and processing may be challenging with this design, limiting study of, for example, immunological correlates of protection.
There could be 4 additional enhancements to such a trial design.
For licensure, a minimum number of participants, consecutively enrolled from the beginning, could have active follow-up for safety and tuberculosis disease through regular clinic attendance. A planned early interim safety analysis could be done. The remaining participants could be monitored using an intensified adverse drug event reporting system, supported by providing participants access to telephone/web-based safety monitoring. These systems could also remind them to present to clinics for diagnosis when unwell.
Routine case detection could be supplemented with active case finding among contacts of incident tuberculosis cases found during follow-up. If a high proportion of a defined population participates in the trial, then a high proportion of cases diagnosed during follow-up will have contacts who had been enrolled and received vaccine or placebo. These could be actively investigated as a high-risk subpopulation. Case contact recruitment can also support assessment of vaccine efficacy against M. tuberculosis transmission.
An end-of-trial tuberculosis prevalence survey in those enrolled could be conducted based on both chest X-ray and symptom-driven GeneXpert sputum testing [15]. A prevalence survey was used in the BCG trial in Chingleput, which followed 90 000 individuals [14].
Follow-up of all participants could continue until the last person randomized completes 2 years of follow-up, as has been done in other trials [16], and provides a median participant follow-up period of approximately 2.5 years.
Advantages of these enhancements include that active case-finding of household contacts of identified tuberculosis cases engages a high-risk group and a prevalence survey addresses the potential problems associated with cases being missed by passive detection. A secondary end point of asymptomatic, subclinical, X-ray–positive tuberculosis disease could also be introduced.
Limitations include that active case finding in household contacts may add only a small percentage increase to diagnosed case accrual. Also, fewer incident index tuberculosis cases in the intervention arm potentially will lead to fewer households undergoing active case finding than in the placebo arm, which may require an adjusted statistical analysis. Further, a prevalence survey adds significant costs and one assumes that pathology in missed cases persists until the end of follow-up.
REDUCING ENROLMENT TIME
Reducing enrolment time to increase the numbers able to be recruited might be achieved in a number of ways.
Trial enrolment may be quicker if results from preenrolment testing for M. tuberculosis infection are not required to determine eligibility.
The advantages of this approach include quicker recruitment. The study population may also be more consistent with the population in which the vaccine will eventually be rolled out. Local populations will most likely be well characterized pretrial, through observational studies of M. tuberculosis infection and disease.
The limitations of this approach include the lack of testing for M. tuberculosis infection. However, batch IGRA testing could still be carried out, enabling stratification of efficacy and safety results by infection status. Furthermore, the benefit may be negated by the turnaround times of other tests required at screening.
Preenrolment testing for tuberculosis disease could be limited to symptom screen and GeneXpert sputum testing; this approach could also be used during the trial for case capture, in participants first screened for tuberculosis symptoms. Testing for tuberculosis disease would then identify clinical tuberculosis disease in need of treatment. Sputum culture could be done in those who are GeneXpert positive, to study the influence of strain on efficacy [17].
The advantages of this approach include that chest X-ray and sputum culture could be avoided at enrolment, with little reduction to sensitivity to detect clinical tuberculosis disease [15].
A limitation of this approach is that if the vaccine does not protect against subclinical disease, observed vaccine efficacy would be reduced. Incident tuberculosis cases occurring in the first few months of trial follow-up could be excluded to address this [9].
Trial enrolment may be quicker if the number of trial sites is increased.
An advantage of this approach includes that larger numbers of trial participants can be enrolled and randomized within a short period.
The limitations include that increasing the number of trial sites might increase the administrative complexity and cost.
CONCLUSION
A phase 3 adolescent/adult POD trial of 3 years’ duration requires both a short enrolment time and a very large study population to maximize the number of early M. tuberculosis exposures postvaccination. Such a design would have the added benefit of increased generalizability (Figure 1). Important considerations such as costs, combining phases 2b and 3, and formal sample size calculations should be explored. Operational research could explore feasibility, speed, and efficiency gains. Trial simulation modeling could help prioritize design components. Formal observation beyond the trial period could help understand longer-term effectiveness and safety, as done for COVID-19 and hepatitis B vaccines [18, 19]. Consensus-building consultations may be needed with regulators, local authorities, and country tuberculosis programs. Pretrial capacity strengthening and epidemiological studies would be important. Domestic and international funders, and policy and regulatory bodies should come together to achieve this aspiration, and accelerate the time to potential availability of a new tuberculosis vaccine.
Notes
Financial support. No financial support was received for this work.
Potential conflicts of interest. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed