My Profile

Search abstracts

Lewis Sheiner

Ljubljana, Slovenia

Stockholm, Sweden

Montreux, Switzerland

Budapest, Hungary

Lisboa, Portugal

Hersonissos, Crete, Greece

Alicante, Spain

Glasgow, Scotland

Venice, Italy

Athens, Greece

Berlin, Germany

St. Petersburg, Russia

Marseille, France

København, Denmark

Brugge/Bruges, Belgium

Pamplona, Spain

Uppsala, Sweden

Verona, Italy

Paris, France

Basel, Switzerland

Salamanca, Spain

Saintes, France

Wuppertal, Germany

Glasgow, Scotland

Sandwich, UK

Frankfurt, Germany

Greenford, UK

Paris, France

Basel, Switzerland

Printable version

PAGE. Abstracts of the Annual Meeting of the Population Approach Group in Europe.
ISSN 1871-6032

PAGE 28 (2019) Abstr 9193 []

Oral: Drug/Disease modelling

B-06 Marjorie Imperial Stratified medicine approaches for drug susceptible tuberculosis patients

Marjorie Z. Imperial (1), Payam Nahid (1), Patrick P. J. Phillips (1), Geraint R. Davies (2), Katherine Fielding (3), Debra Hanna (4,5), David Hermann (5), Robert S. Wallis (6), John L. Johnson (7,8), Christian Lienhardt (9,10) and Rada M. Savic (1)

(1) University of California, San Francisco, San Francisco, CA, USA. (2) University of Liverpool, Liverpool, UK. (3) London School of Hygiene and Tropical Medicine, London, UK. (4) Critical Path Institute, Tucson, AZ, USA. (5) Bill and Melinda Gates Foundation, Seattle, WA, USA. (6) Aurum Institute and ACT4TB/HIV, Johannesburg, South Africa. (7) Case Western Reserve University, Cleveland, OH, USA. (8) University Hospitals Cleveland Medical Center, Cleveland, OH, USA. (9) Global Tuberculosis Programme, World Health Organization, Geneva, Switzerland. (10) Unité Mixte Internationale TransVIHMI (UMI 233 IRD–U1175 INSERM–Université de Montpellier), Institut de Recherche pour le Développement (IRD), Montpellier, France.


Tuberculosis (TB) kills more people than any other infectious disease. All current approaches to TB treatment are based on a one-size-fits-all approach, which leads to under-treatment of patients with severe forms of disease and entails unnecessarily long treatment with potential toxicities for many patients in whom the disease is less severe.[1] Currently, all TB drug development efforts are aimed at shortening treatment duration using the same one-size-fits-all paradigm.[2]  However, all recent Phase 3 clinical trials (OFLOTUB[3], REMoxTB[4], and RIFAQUIN[5]) failed to show non-inferiority between 4-month fluoroquinolone-containing regimens and 6-month standard of care. Nonetheless, the 4-month treatments achieved 80% cure rates, confirming that a significant proportion of global TB burden is eligible for short duration, only if major characteristics of these patients are identified. To that end, we pooled individual patient data from these three trials to 1.) identify populations eligible for short course therapy, and conversely, hard-to-treat populations requiring longer courses, 2.) assess the value of sputum culture conversion biomarker, common Phase 2B endpoint,  as a predictor of outcomes, 3.) evaluate the impact of adherence and dosing strategies on outcomes, and 4.) develop data-driven clinical tools that can be used to provide recommendations for treatment interventions in stratified groups.


Standardized individual patient data were obtained from a public repository.[6] Each trial evaluated later-generation fluoroquinolones as substitutions for ethambutol or isoniazid with the objective of shortening treatment duration from the standard six months to four months. A fourth trial, DMID 01-009[7], in patients with non-cavitary disease treated with a 4-month standard regimen was used for external validation.

The primary efficacy endpoint was time to an unfavorable outcome up to 24 months after start of treatment. We performed multivariate Cox proportional hazard analysis to identify risk factors of unfavorable outcomes. Baseline predictors that were common across trials were included in the analysis: age, race, body mass index, sex, HIV status, presence of cavitation and smear status (measure of bacterial burden). Month 2 culture status and patient adherence were considered as on-treatment predictors.

Non-inferiority analyses were performed in patient subgroups according to identified risk factors in the Cox analysis. The difference in proportion of unfavorable outcomes was calculated using inverse probability Kaplan-Meier (KM) estimates at 24 months after start of treatment. Non-inferiority was assessed using the upper bound of the two-sided 90% confidence interval (CI), determined by bootstrapping (n=500), and a non-inferiority margin of 6 percentage points.

A parametric survival model was developed next to obtain a clinical trial simulation tool and patient stratification algorithm. A competing risks model was developed for two types of unfavorable outcomes: a.) TB related events (i.e. relapse) and b.) non-TB related events (i.e. dropout). Gompertz, Weibull, and surge hazard models were explored. Baseline characteristics, treatment exposure (treatment duration, cumulative number of treatment days, and regimen composition), and on treatment culture positivity were evaluated as predictors of outcomes using a stepwise model selection approach with forward inclusion (p<0.05) and backward deletion (p>0.01) steps.

The final parametric model was used to evaluate in silico novel clinical trial designs and novel strategies to TB treatment based on stratified medicine principles – in which individualized treatment duration is based on patient phenotypes.

All analyses were performed in R 3.4 and NONMEM 7.4. Clinical simulation and clinical management tools were developed using the ‘shiny’ package in R 3.4.


Of 3405 patients, 1404 were randomized to a 6-month control regimen and 2001 to 4-month experimental regimens. In patients assigned to experimental regimens, baseline smear 3+ relative to smear negative or 1+ and HIV seropositive were the two major clinical risk factors for unfavorable outcomes with an adjusted hazard ratio (HR) of 1.6 (95% CI,1.2-2.3) and 1.5 (95% CI 1.1-2.0), respectively. HIV seropositive was also a major clinical risk factor in patients assigned to the control regimen (HR 3.1; 95% CI 2.0-4.6). Non-adherence was the most significant risk factor of unfavorable outcome irrespective of regimen with HR of 5.7 (95% CI, 3.3-9.9) and 5.9 (95% CI, 3.3-10.5) for patients who miss 10% or more doses relative to fully adherent patients following a 4- and 6-month regimen, respectively.[8]

In an easy-to-treat phenotype, the proportion of unfavorable outcomes for patients with a baseline smear <2+ grade or non-cavitary disease, representing 47% of the population, was non-inferior in 4- vs 6-month groups (difference in KM estimate, 3.4; 90% CI, 1.5 to 5.4), indicating that these patients can receive short course therapy. Patients with smear 3+ and cavitary disease, consisting of 34% of the study population, were inferior with the 4-month regimens (difference of 8.8; 90% CI, 6.4-11.3). The easy-to-treat population was externally validated in an independent dataset from the DMID 01-009 trial.[8]

The parametric model confirmed results of the Cox analysis and showed that high baseline smear, HIV seropositive, increased number of missed treatment days, and month 2 culture positivity increases the risk of TB related events. Additionally, older patients are at increased risk of non-TB related events.

Three stratified virtual populations were investigated in clinical trial simulations: a.) easy-to-treat defined as smear <2+ or non-cavitary disease, b.) moderate-to-treat defined as smear 2+ and cavitary disease, and c.) hard-to-treat defined as smear 3+ and cavitary disease. Clinical trial simulations indicated that stratified medicine approaches to TB care, where treatment duration is selected with precision based on patient risk, can result in high cure rates and enable implementation of superiority trial designs in TB drug development.


In this pooled analysis of three recent Phase 3 treatment shortening trials, we have identified easy- and hard-to-treat phenotypes in drug susceptible TB patients. The rifampin-containing regimens tested in these trials are unforgiving with minimal non-adherence resulting in significantly increased risk for unfavorable outcomes. Based on these results, we have developed a risk stratification algorithm and clinical trial simulation tool that was used to investigate optimal treatment interventions for stratified populations. Our results have led to two novel Phase 3 trials currently being designed and developed to evaluate principles of stratified medicine for treatment of drug susceptible and multi-drug resistant TB, which is a paradigm-shifting approach to tackling the TB epidemic.

[1] P. Nahid et al., “Official American Thoracic Society/Centers for Disease Control and Prevention/Infectious Diseases Society of America Clinical Practice Guidelines: Treatment of Drug-Susceptible Tuberculosis,” Clin. Infect. Dis., vol. 63, no. 7, pp. e147–e195, 2016.
[2] S. Goldberg, “TBTC Study 31: Rifapentine-containing Tuberculosis Treatment Shortening Regimens (S31/A5349).” [Online]. Available:
[3] C. S. Merle et al., “A Four-Month Gatifloxacin-Containing Regimen for Treating Tuberculosis,” N. Engl. J. Med, vol. 371, no. 23, pp. 1588–1598, 2014.
[4] S. H. Gillespie et al., “Four-Month Moxifloxacin-Based Regimens for Drug-Sensitive Tuberculosis,” N. Engl. J. Med., vol. 371, no. 17, pp. 1577–1587, 2014.
[5] A. Jindani et al., “High-Dose Rifapentine with Moxifloxacin for Pulmonary Tuberculosis,” N. Engl. J. Med., vol. 371, no. 17, pp. 1599–1608, 2014.
[6] “Platform for Aggregation of Clinical TB Studies, TB-PACTS,” Critical Path Institute. .
[7] J. L. Johnson et al., “Shortening Treatment in Adults with Noncavitary Tuberculosis and 2-Month Culture Conversion,” Am J Respir Crit Care Med, vol. 180, pp. 558–563, 2009.
[8] M. Z. Imperial et al., “A patient-level pooled analysis of treatment-shortening regimens for drug-susceptible pulmonary tuberculosis,” Nat. Med., vol. 24, no. 11, pp. 1708–1715, Nov. 2018.