Phase 2b placebo-controlled trial of M72/AS01E candidate vaccine to prevent active tuberculosis in adults

Background: A tuberculosis vaccine to interrupt transmission is urgently needed. We assessed the safety and efficacy of the candidate tuberculosis vaccine, M72/AS01E, against progression to bacteriologically-confirmed active pulmonary tuberculosis disease in adults with latent Mycobacterium tuberculosis (Mtb) infection. Methods: In a randomized, double-blind, placebo-controlled, phase 2b trial conducted in Kenya, South Africa and Zambia, human immunodeficiency virus (HIV)-negative adults aged 18-50 years with latent Mtb infection (positive by interferon-gamma release assay) were randomized (1:1) to receive two doses of either M72/AS01E or placebo intramuscularly on days 0 and 30. Clinical suspicion of tuberculosis was confirmed from sputum using a polymerase chain reaction test and/or mycobacterial culture. Results: This paper reports the primary analysis, conducted after a mean follow-up of 2.3 years. 1786 participants received M72/AS01E and 1787 received placebo. In the vaccine group, 10 cases met the primary case definition (bacteriologically-confirmed active pulmonary tuberculosis confirmed prior to treatment, not associated with HIV infection) versus 22 cases in the placebo group (0.3 vs. 0.6 cases per 100 person-years, respectively): vaccine efficacy 54.0% (90% confidence interval 13.9-75.4; 95%CI 2.9-78.2; p=0.04). Solicited and unsolicited adverse events within 7 days post-injection were more frequent among M72/AS01E recipients (91.2%) than placebo recipients (68.9%), the difference attributed mainly to injection site reactions and flu-like symptoms. Serious adverse events, potential immune-mediated diseases and deaths occurred with similar low frequencies between groups. Conclusions: M72/AS01E was associated with a clinically acceptable safety profile and provided 54.0% protection for Mtb-infected adults against active pulmonary tuberculosis disease.


Trial registration number: NCT01755598
Introduction (2981, max 2700 words) One-quarter of the global population is estimated to be infected with Mycobacterium tuberculosis (Mtb), and tuberculosis (TB) is the leading infectious cause of death worldwide. 1,2 There were an estimated 10.4 million new TB cases and 1.7 million TB deaths in 2016. An effective TB vaccine for Mtb-infected persons would have a marked impact on TB control, including drug-resistant TB, through interruption of transmission. 3,4 Modelling suggests the most effective contribution to TB control would be a vaccine preventing pulmonary TB in adolescents and young adults. 4 The only licensed TB vaccine, BCG (bacille Calmette-Guérin), does not offer significant protection against pulmonary TB in Mtb-infected adults. 5 The M72/AS01 E (GSK) candidate vaccine contains the M72 recombinant fusion protein derived from two immunogenic Mtb antigens (Mtb32A and Mtb39A), combined with the Adjuvant System AS01, which is also a component of the malaria (Mosquirix, GSK) and recombinant zoster (Shingrix, GSK) vaccines. The Mtb39A and Mtb32A components of the recombinant antigen elicited specific lymphoproliferation and/or interferon-gamma (IFN-ƴ) production in individuals with latent and active TB. [6][7][8] In phase 2 studies, M72/AS01 exhibited a clinically acceptable safety profile, and induced humoral and cell-mediated immune (CMI) responses in healthy and human immunodeficiency virus (HIV)-infected individuals, Mtbinfected adults and adolescents, and in BCG-vaccinated infants (Table S1). [9][10][11][12][13][14][15][16] Despite the caveats associated with the available animal models, 17 non-clinical evaluations (antigen-selection approach and in vivo preclinical data), clinical safety and immunogenicity evidence, based on the ability of the candidate vaccine to induce Th-1 type responses, supported a proof-of-concept human trial. [6][7][8][9][18][19][20][21][22] We conducted a proof-of-concept phase 2b trial to evaluate M72/AS01 E in preventing bacteriologically-confirmed pulmonary TB in HIV-negative adults with Mtb infection, defined by a positive interferon-gamma release assay (IGRA) (www.clinicaltrials.gov. NCT01755598). This population was selected based on its higher incidence of pulmonary TB compared to IGRA-negative individuals, which allowed a smaller sample size for proofof-concept. 23

Study design
The study is a multi-center, double-blind, randomized (1:1), placebo-controlled trial conducted in three TB endemic African countries (Kenya, South Africa and Zambia). The randomization was not stratified but was performed using a minimization algorithm accounting for gender and center (see Supplement for details). Eleven study sites were selected based on local TB prevalence and ability to perform the study according to Good Clinical Practice guidelines (GCP). The QuantiFERON TB Gold in-Tube assay (QFT, Qiagen) was used at the manufacturer's recommended cut-off to identify latent Mtb infection.
The study population is being followed up for 3 years after vaccination with M72/AS01 E or placebo. A pre-specified primary analysis was performed when all participants had completed at least 2 years of follow-up. Immunogenicity and reactogenicity were assessed in a subgroup of 300 participants. The final analysis after 3 years of follow-up and secondary study objectives including CMI responses will be reported in a subsequent publication.
The study was undertaken in accordance with GCP and the Declaration of Helsinki. The protocol was approved by ethics committees and regulatory authorities in each participating country. All participants provided informed consent. The study protocol is available at NEJM.org. Unblinded safety data are reviewed by an independent data monitoring committee (IDMC). Anonymized individual participant data and study documents can be requested for further research from www.clinicalstudydatarequest.com.

Population
Adults 18-50 years of age were eligible if they were healthy or had stable chronic medical conditions, were HIV-negative, had no TB symptoms, were QFT-positive, and had a sputum sample negative for Mtb at baseline using polymerase chain reaction (PCR) (GeneXpert MTB/RIF, Cepheid).

Vaccination
Two doses of M72/AS01 E or placebo were administered intramuscularly into the deltoid one month apart.

Efficacy endpoints
The primary study objective was to evaluate M72/AS01 E efficacy to prevent active pulmonary TB according to the first case definition (primary endpoint; see Table 1 for case definitions). Secondary study objectives were vaccine efficacy (VE) according to additional case definitions, immunogenicity, safety and reactogenicity of the vaccine.

Evaluation of safety and reactogenicity
Serious adverse events (SAEs), potential immune-mediated diseases (pIMDs) and pregnancies were recorded until 6 months after the second vaccination. SAEs deemed related to study product were recorded until study end. Unsolicited AEs were recorded for 30 days after each dose. Local and systemic symptoms were solicited from the subgroup using diary cards for 7 days after each injection. Laboratory testing for clinical chemistry and hematology was performed in the subgroup on Days 0, 7, 30 and 37.

Evaluation of immunogenicity
Blood samples were collected from the subgroup before dose 1, one month post-dose 2, and annually until year 3. Anti-M72 IgG antibodies were measured using enzyme-linked immunosorbent assay (ELISA) as previously described (cut-off 2.8 ELISA units/ml). 13 TB surveillance TB surveillance used both active (visits, phone calls and text messages), and passive (selfreporting) methods. Participants with clinical suspicion of pulmonary TB provided three sputum samples collected over one week for PCR and liquid culture by Mycobacterial Growth Indicator Tube. Samples were preferably to be taken before initiation of TB treatment, but samples collected up to 4 weeks after treatment initiation were accepted (Table 1, case definitions 3 and 4). Diagnostic and treatment decisions were made by treating non-study physicians. HIV retesting and screening for diabetes (HbA1c) were performed in all participants with confirmed TB disease.

Statistical analysis
Using a log rank test with 80% power assuming a true VE of 70% (hazard ratio of 30%) and a 2-sided 10% significance level, 21 cases were required for a fixed sample design assuming proportional hazard rates. To obtain 21 cases, assuming a mean yearly attack rate of 0.55% in the control group, 2 years of follow-up for each subject and an attrition rate of 15% over the 2-year period, 3,506 participants needed to be enrolled. As per protocol, the primary analysis could occur at 21 cases or at completion of 24-month follow-up. VE was analyzed in the according-to-protocol efficacy cohort, using a Cox proportional hazard regression models (VE=1-hazard ratio) with 90% confidence intervals (CIs) and Wald p-value. Descriptive post-hoc 95% CIs are also provided. The primary endpoint was met if the lower limit of the 2-sided 90% CI for VE against bacteriologically-confirmed pulmonary TB (first case definition) was >0%. If the primary endpoint was met, the first secondary endpoint (VE for the second case definition) was to be analyzed using the same success criterion. A pre-planned exploratory analysis compared the effect of 6 pre-specified covariates (giving 14 subgroups) on VE (interpretation should be performed cautiously as the risk of having at least one false significant result ranges between 51%-77%).
The total vaccinated cohort (all subjects who received at least one vaccination) was used to assess safety. Analysis of immunogenicity was performed on the according-to-protocol immunogenicity cohort for the subgroup.
Statistical analyses were performed with SAS version 9.2 or above on SAS Drug Development. Only the external statisticians and IDMC have been unblinded at the level of individual subject data.
The Supplementary Appendix provides the eligibility criteria and screening procedures, vaccine and placebo composition, safety monitoring and surveillance activities.

Results
Out of 3,575 randomized participants, 3,573 received at least one dose of M72/AS01 E or placebo between August 2014 and November 2015. The mean age of participants was 28.9 (standard deviation [SD] 8.3 years); 43% were women. The study groups were balanced in terms of pre-specified demographic characteristics (Table S2).

Vaccine efficacy
There were 3,283 participants included in the according-to-protocol efficacy analysis ( Figure   1). Ten cases of active pulmonary TB in the vaccine group and 22 cases in the placebo group met the primary case definition after mean follow-up of 2.3 years (SD 0.4) ( Table 2).
A planned sensitivity analysis of the first case definition was restricted to participants positive for Mtb by at least two diagnostic tests (culture and/or PCR) performed on the sputa collected (Table S3). This analysis included 5 cases in the M72/AS01 E group and 17 cases in the placebo group; VE was 70.3% (90% CI 31.3-87.1%; p=0.02) ( Table 2). Piece-wise analysis of cases (first case definition) occurring before versus after the median follow-up time (1.12 years) showed VE of 39.0% (90% CI -42.5-73.9) in the first period and 66.5% (90% CI 13.3-87.0) in the second period.
Pre-specified subgroup analyses using case definition 1 showed VE in men of 75.2% (p=0.03) and in women of 27.4% (p=0.52), and VE in participants aged ≤25 years of 84.4% (p=0.01) and VE for those aged >25-50 years of 10.2% (p=0.82) ( Table 3). A post-hoc hierarchical test was performed to assess the interaction between group and gender (p=0.31) and between group and age (p=0.07) in the complete model containing all main effects as well as the two interaction terms (Table S4).

Reactogenicity and safety
The percentage of participants who experienced at least one SAE within 6 months postvaccination was similar between the groups (1.6% in the M72/AS01 E group and 1.8% in the placebo group) ( Table 4). One SAE in each group was considered causally related to vaccination (pyrexia and hypertensive encephalopathy, currently blinded to group). pIMDs were reported by two participants in the M72/AS01 E group and 5 in the placebo group.
There were 24 deaths (14 trauma-related) during the study, 7 in the M72/AS01 E group (5 trauma-related) and 17 in the placebo group (9 trauma-related) ( Table 4). No death was assessed as related to study vaccination. One participant died of pneumonia for whom there was also a suspicion of intestinal TB, but this latter diagnosis was not confirmed. Vaccination did not significantly affect hematology and biochemistry parameters ( Figure S1). A post-hoc analysis showed 33 pregnancies, of which 28 resulted in delivery of a healthy infant. There were three ectopic pregnancies, one spontaneous abortion, and one pregnant woman was lost-to-follow-up. No birth defects were noted. Regular IDMC review of unblinded safety data resulted in recommendations to continue the study unchanged.
More participants reported unsolicited AE in the M72/AS01 E group (67.4%) than in the placebo group (45.4%). The excess was driven by injection-site reactions and influenza-like symptoms (Table S5). Swelling reactions larger than 100mm diameter were reported by 53 participants (3.0%) in the M72/AS01 E group and by one participant in the placebo group.
The median duration of these large swelling reactions was 4 days.
In the subgroup, local and systemic solicited symptoms were reported more frequently by M72/AS01 E recipients than placebo recipients (Table S6). Among local solicited symptoms, pain was the most frequently reported (81.8% of M72/AS01 E recipients and 34.4% of placebo recipients, with 24.3% and 3.3%, respectively, reporting grade 3 pain). Redness and swelling were uncommon in both groups. Fatigue, headache, malaise, or myalgia was each reported by 58.1%-68.9% of M72/AS01 E recipients and 26.5%-47.0% of placebo recipients.
The immunogenicity results indicate 100% of participants in the M72/AS01E group seroconverted at Month 2 and 99% were still seropositive at Month 12 ( Figure S2). Figure   S3 elaborates on the clinical relevance of this study that could be shared with patients by healthcare professionals.

Discussion
There is no TB vaccine recommended for use in Mtb-infected adults, who represent a large reservoir of future cases of active TB. Here, we demonstrate that protection against TB disease may be achieved by vaccination of Mtb-infected adults with an adjuvanted subunit vaccine containing two Mtb proteins. The finding of efficacy for the primary endpoint was supported by the more stringent sensitivity analysis, and by the analysis of the second case definition. Less stringent case definitions 3-5 showed similar, but non-significant, trends for efficacy. This is the first efficacy trial with M72/AS01 E , and the results confirm the clinically acceptable safety and reactogenicity profile observed previously. Antibody responses were in the same range as previously observed in M72/AS01 E -vaccinated adults living in TB endemic regions. 9,10 Since the trial included only Mtb-infected individuals, it is not possible to determine the extent to which Mtb infection influences VE. In previous TB efficacy trials, the viral-vectored candidate vaccine MVA85A showed no additional protection beyond that provided by BCG in Mtb-uninfected infants; 24 multiple doses of inactivated M. vaccae (obuense) administered to HIV-infected adults reduced the hazard of definite TB, which was a secondary endpoint, by 39%, with no effect modification by baseline Mtb infection status. 25 A comprehensive global TB vaccination strategy should be targeted at both Mtb-uninfected and -infected adolescents and adults. 4 Our study in Mtb-infected adults complements the findings of a recent trial that demonstrated 45% efficacy of BCG revaccination for protection of Mtb non-infected adolescents against sustained QFT seroconversion (In press). 26 These results suggest a potential role for M72/AS01 E among vaccination strategies against TB.
Recent research suggests that progression from latent Mtb infection to active TB is not a single definitive event, but rather a transition through a spectrum of inflammatory and infected states reflecting the activity of individual granulomas. 27 Clinically, this spectrum results in heterogeneous disease states within and between individuals. In this study, participants with clinical suspicion of TB underwent diagnostic investigation. Approximately one-third of confirmed pulmonary TB cases were only confirmed by a single test out of the six performed (either culture or PCR). 'Single positive' cases were evenly distributed between the vaccine and placebo arms and became positive by culture (7 cases) after an unusually long period or by PCR (3 cases) after an unusually high number of amplification cycles. We hypothesize that active surveillance of study participants detected pulmonary TB with low bacterial load, consistent with early stages of disease or reinfection. Interestingly, three (out of 10) 'single positive' participants (blinded as to group) did not receive TB treatment and remain well, suggesting successful immune control and lack of disease progression. The sensitivity analysis suggested higher VE in participants with at least two positive tests, consistent with higher bacterial load. Piece-wise and time-to-event analyses did not demonstrate significant VE during year 1. We hypothesize this may be because at least some individuals who developed active TB during this time already had incipient TB at baseline, against which the vaccine could not be expected to have impact, or that the study did not have power to demonstrate a difference in the first year, or that this was a chance finding. Whilst we made reasonable efforts to exclude participants with active TB at screening (single PCR test on one sputum specimen), a limitation of the study was that we could not exclude that early active cases were missed, given the frequently low bacillary load and sporadic nature of bacillary shedding in early stages of TB disease. 28 Unexpectedly, we observed a higher point estimate for VE in men than women (attack rate in the placebo group 0.6 per 100 person-years for men and women), and in those aged ≤25 years versus 26-50 years (attack rate 0.8 versus 0.4 per 100 person-years, respectively). A post-hoc demographic analysis showed an imbalance in gender in the group aged ≤25 years (66% men and 34% women), while the older age group was well-balanced, suggesting that the apparent difference observed by gender was confounded by the effect of age and is artefactual. Additionally, in a post-hoc interaction test, VE did not seem to be different by gender (p=0.3), while efficacy tended to be heterogeneous across age groups (p=0.07 in a hierarchical model containing both interactions). Interpretation of all post-hoc and exploratory subgroup analyses should be performed cautiously because the study was not powered to detect differences between subgroups and multiplicity was not accounted for.
Age could potentially affect VE through a differential vaccine effect according to time since primary Mtb infection or BCG priming. 29 We hypothesize that those further from primary infection are more likely to have infection under immune system control, with little additional benefit conveyed by vaccination. Increasing age is associated with increased probability of more remote infection based on several studies [30][31][32][33][34] and screening data from the current study, in which 55.1%-66.6% of the screened individuals were already Mtb-infected. 31 Alternatively, the circumstances that lead to reactivation may be less amenable to immunologic control by booster vaccination further from primary BCG vaccination or initial Mtb infection, and therefore the benefits of vaccination may be more limited. Given the age of the study population, immune senescence is unlikely to impact VE.
PCR had sensitivity of 80% compared to culture (Table S7), consistent with more events meeting the first case definition than the second. Future VE trials should therefore utilize automated liquid culture in addition to PCR to maximize case detection. TB treatment of adult drug-sensitive pulmonary TB leads to negative sputum culture and PCR in 8 weeks in some participants; 35 therefore case definitions 3 and 4 likely underestimate TB incidence.
Strengths of the study were the inclusion of a large, well-defined cohort, exclusion of active TB disease at baseline, statistical power to address the primary endpoint, and the use of alternative case definitions for the efficacy endpoint that reflect applicability in the real world.
Finally, 99% of participants consented to biobanking of pre-and post-vaccination blood samples. These samples offer the opportunity to discover potential immune correlates of vaccine-mediated protection against TB, which if confirmed, will be critical to reduce the size of future efficacy trials (www.clinicaltrials.gov NCT02097095).
In conclusion, M72/AS01 E had an acceptable safety and reactogenicity profile, and significantly reduced the incidence of pulmonary TB in healthy Mtb-infected HIV-negative adults. These promising results provide a unique opportunity to better understand the mechanisms by which this vaccine confers protection against TB and could lead to future improvements in global tuberculosis control.

Trademark statement
Mosquirix and Shingrix are trademarks of the GSK group of companies.