VPM1002 for tuberculosis prevention in India: a 1,296-compartment dynamic model with disaggregated pulmonary and extrapulmonary efficacy, Bayesian evidence synthesis, and dual-perspective health economics
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Background
Tuberculosis (TB) kills 1·3 million people annually. Global efforts focus on ending pulmonary TB (PTB); however, extrapulmonary TB (EPTB) is rising and poses significant health and economic burden. The PreVenTB Phase III trial evaluated VPM1002 in household contacts aged ≥6 years across India, and did not meet its primary composite endpoint for all TB. Disaggregated prespecified secondary endpoints revealed a substantially stronger EPTB signal (vaccine efficacy 42·3%, 95% CI −9·1 to 69·4, p=0·09) in the modified intention-to-treat (mITT) population. No existing TB model translates these disaggregated hazard ratios into population-level effectiveness across heterogeneous demographics, or against a dynamic baseline accounting for ongoing National Tuberculosis Elimination Programme (NTEP)-driven incidence decline.
Methods
We developed a 1,296-compartment deterministic dynamic compartmental model (4 age × 2 HIV × 3 BMI × 3 socioeconomic strata × 18 states). Separate PTB and EPTB vaccine-efficacy posteriors were derived by Bayesian evidence synthesis of the PreVenTB trial’s per-protocol and mITT analyses (power-prior-adjusted conjugate normal-normal update; α=0·082) and propagated through Monte Carlo simulation (n=1,000 iterations per scenario). A dynamic no-vaccine baseline was constructed by fitting time-varying case-detection-rate CDR(t) and treatment-success-rate TSR(t) logistic curves to WHO/NTEP data (2015 to 2024; incidence validation R 2 =0·896, RMSE 4·90 per 100,000), projecting PTB and EPTB incidence to 2050. Economic analysis used societal (value-of-statistical-life-inclusive BCR) and health-system (cost-effectiveness acceptability curves) perspectives, discounted at 3% annually.
Findings
Posterior vaccine effectiveness was substantially higher against EPTB than PTB (EPTB 40·0% [95% CI −5·6 to 69·1%] vs PTB 12·8% [−19·0 to 37·9%]; all-TB 16·2% [−11·7 to 38·6%]). EPTB accounted for 74% of deaths averted (381 of 513) and 73% of DALYs averted (5,347 of 7,308) in the 10-year/dynamic scenario. EPTB cases averted exceeded PTB and concurrent disease combined in every scenario. Mean cases averted ranged from 2,047 (3-year protection, dynamic baseline) to 3,394 (10-year, static) per 1,000,000 vaccinated; mean disability-adjusted life years (DALYs) averted ranged from 4,792 to 8,125. The benefit-cost ratio (BCR; societal perspective) ranged from 3·4 (3-year protection, dynamic baseline) to 8·2 (10-year protection, static baseline), exceeding break-even in every scenario. Median gross incremental cost-effectiveness ratio (ICER) ranged from US$647 (10-year static) to US$1,568 (3-year dynamic) per DALY averted, below India’s 3× gross domestic product (GDP)-per-capita threshold (approximately US$8,084) in every scenario.
Interpretation
This 1,296-compartment model provides the first dynamically-baselined, dual-perspective health-economic evaluation of VPM1002 to separately track pulmonary and extrapulmonary outcomes. EPTB protection is VPM1002’s proportionally larger and more statistically reliable efficacy signal and drives a majority of averted cases, mortality, and DALYs. Policy assessments anchored to composite pulmonary endpoints systematically underestimate this vaccine’s population value.
Funding
In part funded by Serum Life Science Europe GmbH.
Research in context
Evidence before this study
We searched PubMed and preprint servers (medRxiv, bioRxiv) up to July 17, 2026, using various combinations of the terms “tuberculosis vaccine model”, “LMICs”, “TB vaccine cost-effectiveness”, “VPM1002”, “pulmonary and extrapulmonary tuberculosis (E/PTB) model”, and “TB dynamic transmission model India”. Searches returned 4 to 313 results depending on the term combination used; references were screened by title and abstract for relevance to TB vaccine modelling, cost-effectiveness, or EPTB. Existing TB vaccine modelling studies for India and other high-burden low- and middle-income countries (LMICs) have relied on aggregate vaccine efficacy estimates combining PTB and EPTB into a single endpoint. Landmark analyses by Clark, Yerramsetti, Harris, and Weerasuriya and colleagues used efficacy assumptions derived primarily from M72/AS01□ or BCG-revaccination trials, in adults or paediatric models, none of which reported disaggregated PTB or EPTB outcomes. Models have assumed homogeneous mixing and evaluated vaccine impact against static programmatic baselines. No published model has tracked EPTB as a separate disease compartment, stratified impact simultaneously by age, HIV status, BMI, and socioeconomic status (SES), or projected impact against a dynamic NTEP trajectory calibrated to observed incidence data.
Added value of this study
This 1,296-compartment model is the first to propagate disaggregated PTB and EPTB VPM1002 efficacy posteriors, derived by Bayesian evidence synthesis of the PreVenTB Phase III trial, through a stratified population model. We construct a dynamic no-vaccine baseline calibrated to WHO/NTEP data from 2015 to 2024, enabling marginal vaccine impact to be expressed against a realistically declining programmatic trajectory. Dual-perspective economic outputs (societal benefit-cost ratio and health-system incremental cost-effectiveness ratio) are reported alongside cost-effectiveness acceptability curves. The model shows that EPTB accounts for the majority of averted cases, deaths, and DALYs across all scenarios, a finding that is indiscernible in models using composite efficacy endpoints.
Implications of all the available evidence
VPM1002 is cost-effective at conventional willingness-to-pay thresholds in every modelled scenario. The EPTB-specific efficacy signal, statistically more reliable than the composite endpoint and consistent with VPM1002’s mechanism of action, drives a disproportionate share of averted mortality and DALY burden. Policy assessments relying solely on composite pulmonary endpoints will underestimate this vaccine’s population value. The framework is directly relevant to National Technical Advisory Group on Immunisation (NTAGI) deliberations and Health Technology Assessment in India (HTAIn) appraisals, and is extensible to other high-burden LMICs and TB vaccine candidates.
