Transcriptional adaptation of Mycobacterium tuberculosis that survives prolonged multi-drug treatment in mice.

To address the ongoing global tuberculosis crisis, there is a need for shorter, more effective treatments. A major reason why tuberculosis requires prolonged treatment is that, following a short initial phase of rapid killing, the residual Mycobacterium tuberculosis withstands drug killing. Because existing methods lack sensitivity to quantify low-abundance mycobacterial RNA in drug-treated animals, cellular adaptations of drug-exposed bacterial phenotypes in vivo remain poorly understood. Here, we used a novel RNA-seq method called SEARCH-TB to elucidate the Mycobacterium tuberculosis transcriptome in mice treated for up to 28 days with standard doses of isoniazid, rifampin, pyrazinamide, and ethambutol. We compared murine results with in vitro SEARCH-TB results during exposure to the same regimen. Treatment suppressed genes associated with growth, transcription, translation, synthesis of rRNA proteins, and immunogenic secretory peptides. Bacteria that survived prolonged treatment appeared to transition from ATP-maximizing respiration toward lower-efficiency pathways and showed modification and recycling of cell wall components, large-scale regulatory reprogramming, and reconfiguration of efflux pump expression. Although the pre-treatment in vivo and in vitro transcriptomes differed profoundly, genes differentially expressed following treatment in vivo and in vitro were similar, with differences likely attributable to immunity and drug pharmacokinetics in mice. These results reveal cellular adaptations of Mycobacterium tuberculosis that withstand prolonged drug exposure in vivo, demonstrating proof of concept that SEARCH-TB is a highly granular pharmacodynamic readout. The surprising finding that differential expression is concordant in vivo and in vitro suggests that insights from transcriptional analyses in vitro may translate to the mouse. IMPORTANCE A major reason that curing tuberculosis requires prolonged treatment is that drug exposure changes bacterial phenotypes. The physiologic adaptations of Mycobacterium tuberculosis that survive drug exposure in vivo have been obscure due to low sensitivity of existing methods in drug-treated animals. Using the novel SEARCH-TB RNA-seq platform, we elucidated Mycobacterium tuberculosis phenotypes in mice treated for with the global standard 4-drug regimen and compared them with the effect of the same regimen in vitro. This first view of the transcriptome of the minority Mycobacterium tuberculosis population that withstands treatment in vivo reveals adaptation of a broad range of cellular processes, including a shift in metabolism and cell wall modification. Surprisingly, the change in gene expression induced by treatment in vivo and in vitro was largely similar. This apparent "portability" from in vitro to the mouse provides important new context for in vitro transcriptional analyses that may support early preclinical drug evaluation.

Lung microenvironments harbor Mycobacterium tuberculosis phenotypes with distinct treatment responses.

Tuberculosis lung lesions are complex and harbor heterogeneous microenvironments that influence antibiotic effectiveness. Major strides have been made recently in understanding drug pharmacokinetics in pulmonary lesions, but the bacterial phenotypes that arise under these conditions and their contribution to drug tolerance are poorly understood. A pharmacodynamic marker called the RS ratio® quantifies ongoing rRNA synthesis based on the abundance of newly synthesized precursor rRNA relative to mature structural rRNA. Application of the RS ratio in the C3HeB/FeJ mouse model demonstrated that Mycobacterium tuberculosis populations residing in different tissue microenvironments are phenotypically distinct and respond differently to drug treatment with rifampin, isoniazid, or bedaquiline. This work provides a foundational basis required to address how anatomic and pathologic microenvironmental niches may contribute to long treatment duration and drug tolerance during the treatment of human tuberculosis.

Transcriptional adaptation of drug-tolerant Mycobacterium tuberculosis in mice.

Transcriptome evaluation of Mycobacterium tuberculosis in the lungs of laboratory animals during long-term treatment has been limited by extremely low abundance of bacterial mRNA relative to eukaryotic RNA. Here we report a targeted amplification RNA sequencing method called SEARCH-TB. After confirming that SEARCH-TB recapitulates conventional RNA-seq in vitro, we applied SEARCH-TB to Mycobacterium tuberculosis-infected BALB/c mice treated for up to 28 days with the global standard isoniazid, rifampin, pyrazinamide, and ethambutol regimen. We compared results in mice with 8-day exposure to the same regimen in vitro. After treatment of mice for 28 days, SEARCH-TB suggested broad suppression of genes associated with bacterial growth, transcription, translation, synthesis of rRNA proteins and immunogenic secretory peptides. Adaptation of drug-stressed Mycobacterium tuberculosis appeared to include a metabolic transition from ATP-maximizing respiration towards lower-efficiency pathways, modification and recycling of cell wall components, large-scale regulatory reprogramming, and reconfiguration of efflux pumps expression. Despite markedly different expression at pre-treatment baseline, murine and in vitro samples had broadly similar transcriptional change during treatment. The differences observed likely indicate the importance of immunity and pharmacokinetics in the mouse. By elucidating the long-term effect of tuberculosis treatment on bacterial cellular processes in vivo, SEARCH-TB represents a highly granular pharmacodynamic monitoring tool with potential to enhance evaluation of new regimens and thereby accelerate progress towards a new generation of more effective tuberculosis treatment.

Standardized RS Ratio Metrics To Assess Tuberculosis Antimicrobial Efficacy and Potency.

The sigmoid Emax model was used to describe the rRNA synthesis ratio (RS ratio) response of Mycobacterium tuberculosis to antimicrobial concentration. RS-Emax measures the maximal ability of a drug to inhibit the RS ratio and can be used to rank-order drugs based on their RS ratio effect. RS-EC90 is the concentration needed to achieve 90% of the RS-Emax, which may guide dose selection to achieve a maximal RS ratio effect in vivo.

Combination of Mycobacterium tuberculosis RS Ratio and CFU Improves the Ability of Murine Efficacy Experiments to Distinguish between Drug Treatments.

Murine tuberculosis drug efficacy studies have historically monitored bacterial burden based on CFU of Mycobacterium tuberculosis in lung homogenate. In an alternative approach, a recently described molecular pharmacodynamic marker called the RS ratio quantifies drug effect on a fundamental cellular process, ongoing rRNA synthesis. Here, we evaluated the ability of different pharmacodynamic markers to distinguish between treatments in three BALB/c mouse experiments at two institutions. We confirmed that different pharmacodynamic markers measure distinct biological responses. We found that a combination of pharmacodynamic markers distinguishes between treatments better than any single marker. The combination of the RS ratio with CFU showed the greatest ability to recapitulate the rank order of regimen treatment-shortening activity, providing proof of concept that simultaneous assessment of pharmacodynamic markers measuring different properties will enhance insight gained from animal models and accelerate development of new combination regimens. These results suggest potential for a new era in which antimicrobial therapies are evaluated not only on culture-based measures of bacterial burden but also on molecular assays that indicate how drugs impact the physiological state of the pathogen.

MOVER approximated CV: A tool for quantifying precision in ratiometric droplet digital PCR assays.

Droplet digital PCR is a particularly valuable tool for ratiometric assays because it provides simultaneous absolute quantification of two target sequences in a single assay. This manuscript addresses a challenge in establishing a new ratiometric droplet digital PCR assay for use in sputum, the rRNA synthesis ratio. In principle, the methods established to evaluate precision and determine the limit of quantification for a single measurand cannot be applied to a ratiometric assay. The precision of a ratio depends on precision in both the numerator and denominator. Here, we evaluated the MOVER approximated coefficient of variation as indicator of assay precision that does not require technical replicates. We estimated the MOVER approximated coefficient of variation in dilution series and routine assays and evaluated its agreement with the traditional coefficient of variation. We found that the MOVER approximated coefficient of variation was able to recapitulate the traditional coefficient of variation without the requirement for replicate assays. We also demonstrated that the MOVER approximated coefficient of variation threshold can be used to define the limit of quantification of the rRNA synthesis Ratio. In conclusion, the MOVER approximated coefficient of variation may be useful not only for the rRNA synthesis ratio but for other assays that measure ratios via droplet digital PCR.

Reproducibility of the Ribosomal RNA Synthesis Ratio in Sputum and Association with Markers of Mycobacterium tuberculosis Burden.

There is a critical need for improved pharmacodynamic markers for use in human tuberculosis (TB) drug trials. Pharmacodynamic monitoring in TB has conventionally used culture or molecular methods to enumerate the burden of Mycobacterium tuberculosis organisms in sputum. A recently proposed assay called the rRNA synthesis (RS) ratio measures a fundamentally novel property, how drugs impact ongoing bacterial rRNA synthesis. Here, we evaluated RS ratio as a potential pharmacodynamic monitoring tool by testing pretreatment sputa from 38 Ugandan adults with drug-susceptible pulmonary TB. We quantified the RS ratio in paired pretreatment sputa and evaluated the relationship between the RS ratio and microbiologic and molecular markers of M. tuberculosis burden. We found that the RS ratio was highly repeatable and reproducible in sputum samples. The RS ratio was independent of M. tuberculosis burden, confirming that it measures a distinct new property. In contrast, markers of M. tuberculosis burden were strongly associated with each other. These results indicate that the RS ratio is repeatable and reproducible and provides a distinct type of information from markers of M. tuberculosis burden. IMPORTANCE This study takes a major next step toward practical application of a novel pharmacodynamic marker that we believe will have transformative implications for tuberculosis. This article follows our recent report in Nature Communications that an assay called the rRNA synthesis (RS) ratio indicates the treatment-shortening of drugs and regimens. Distinct from traditional measures of bacterial burden, the RS ratio measures a fundamentally novel property, how drugs impact ongoing bacterial rRNA synthesis.

Mycobacterium tuberculosis precursor rRNA as a measure of treatment-shortening activity of drugs and regimens.

There is urgent need for new drug regimens that more rapidly cure tuberculosis (TB). Existing TB drugs and regimens vary in treatment-shortening activity, but the molecular basis of these differences is unclear, and no existing assay directly quantifies the ability of a drug or regimen to shorten treatment. Here, we show that drugs historically classified as sterilizing and non-sterilizing have distinct impacts on a fundamental aspect of Mycobacterium tuberculosis physiology: ribosomal RNA (rRNA) synthesis. In culture, in mice, and in human studies, measurement of precursor rRNA reveals that sterilizing drugs and highly effective drug regimens profoundly suppress M. tuberculosis rRNA synthesis, whereas non-sterilizing drugs and weaker regimens do not. The rRNA synthesis ratio provides a readout of drug effect that is orthogonal to traditional measures of bacterial burden. We propose that this metric of drug activity may accelerate the development of shorter TB regimens.

C-Reactive Protein Testing for Active Tuberculosis among Inpatients without HIV in Uganda: a Diagnostic Accuracy Study.

The objective of this prospective cross-sectional study, conducted at a national referral hospital in Kampala, Uganda, was to determine diagnostic performance of serum C-reactive protein (CRP) as a triage test for tuberculosis (TB) among HIV-seronegative inpatients. We calculated the sensitivity, specificity, positive and negative likelihood ratios, and positive and negative predictive values to determine the diagnostic performance of a CRP enzyme-linked immunosorbent assay (ELISA) (Eurolyser) in comparison to that of a reference standard of Mycobacterium tuberculosis culture on two sputum samples. We constructed receiver operating curves and reported performance in reference to the manufacturer's cutoff and also to a threshold chosen to achieve sensitivity of >90%, in accordance with the WHO's target-product profile for a triage test. Among 119 HIV-seronegative inpatients, 46 (39%) had culture-positive pulmonary TB. In reference to M. tuberculosis culture, CRP had a sensitivity of 78% (95% confidence interval [CI], 64 to 89%) and a specificity of 52% (95% CI, 40 to 64%) at the manufacturer's threshold of 10 mg/liter. At a threshold of 1.5 mg/liter, the sensitivity was 91% (95% CI, 79 to 98%) but the specificity was only 21% (95% CI, 12 to 32%). Performance did not differ when stratified by illness severity at either threshold. In conclusion, among HIV-seronegative inpatients, CRP testing performed substantially below targets for a TB triage test. Additional studies among HIV-seronegative individuals in clinics and community settings are needed to assess the utility of CRP for TB screening.

Antibiotic Lethality and Membrane Bioenergetics.

A growing body of research suggests bacterial metabolism and membrane bioenergetics affect the lethality of a broad spectrum of antibiotics. Electrochemical gradients spanning energy-transducing membranes are the foundation of the chemiosmotic hypothesis and are essential for life; accordingly, their dysfunction appears to be a critical factor in bacterial death. Proton flux across energy-transducing membranes is central for cellular homeostasis as vectorial proton translocation generates a proton motive force used for ATP synthesis, pH homeostasis, and maintenance of solute gradients. Our recent investigations indicate that maintenance of pH homeostasis is a critical factor in antibiotic killing and suggest an imbalance in proton flux initiates disruptions in chemiosmotic gradients that lead to cell death. The complex and interconnected relationships between electron transport systems, central carbon metabolism, oxidative stress generation, pH homeostasis, and electrochemical gradients provide challenging obstacles to deciphering the roles for each of these processes in antibiotic lethality. In this chapter, we will present evidence for the pH homeostasis hypothesis of antibiotic lethality that bactericidal activity flows from disruption of cellular energetics and loss of chemiosmotic homeostasis. A holistic understanding of the interconnection of energetic processes and antibiotic activity may direct future research toward the development of more effective therapeutic interventions.

Does discovery of differentially culturable M tuberculosis really demand a new treatment paradigm? Longitudinal analysis of DNA clearance from sputum.

BACKGROUND: According to the traditional tuberculosis (TB) treatment paradigm, the initial doses of treatment rapidly kill most Mycobacterium tuberculosis (Mtb) bacilli in sputum, yet many more months of daily treatment are required to eliminate a small, residual subpopulation of drug-tolerant bacilli. This paradigm has recently been challenged following the discovery that up to 90% of Mtb bacilli in sputum are culturable only with growth-factor supplementation. These "differentially culturable" bacilli are hypothesized to be more drug-tolerant than routinely culturable bacilli. This hypothesis implies an alternative paradigm in which TB treatment does not rapidly reduce the total Mtb population but only the small, routinely culturable subpopulation. To evaluate these competing paradigms, we developed a culture-independent method for quantifying the viable fraction of Mtb bacilli in sputum during treatment. METHODS: We used GeneXpert MTB/RIF to quantify Mtb DNA in sputa collected longitudinally from Ugandan adults taking standard 4-drug treatment for drug-susceptible pulmonary TB. We modeled GeneXpert cycle thresholds over time using nonlinear mixed-effects regression. We adjusted these models for clearance of DNA from killed-but-not-yet-degraded bacilli, assuming clearance half-lives ranging from 0 to 1.25 days. We used a convolution integral to quantify DNA from viable bacilli only, and converted cycle thresholds to Mtb genomic equivalents. We replicated our results in a South African cohort. RESULTS: We enrolled 41 TB patients in Uganda. Assuming a DNA-clearance half-life of 0 days, genomic equivalents of viable sputum bacilli decreased by 0.22 log/day until 8.8 days, then by 0.07 log/day afterwards. Assuming a DNA-clearance half-life of 1.25 days, genomic equivalents of viable bacilli decreased by 0.36 log/day until 5.0 days, then by 0.06 log/day afterwards. By day 7, viable Mtb had decreased by 97.2-98.8%. We found similar results for 19 TB patients in South Africa. DISCUSSION: Using a culture-independent method, we found that TB treatment rapidly eliminates most viable Mtb in sputum. These findings are incompatible with the hypothesis that differentially culturable bacilli are drug-tolerant. CONCLUSIONS: A culture-independent method for measuring viable Mtb in sputum during treatment corroborates the traditional TB treatment paradigm in which a rapid bactericidal phase precedes slow, elimination of a small, residual bacillary subpopulation.

Effect of anti-retroviral therapy on oxidative stress in hospitalized HIV-infected adults with and without TB.

BACKGROUND: HIV infection and opportunistic infections cause oxidative stress (OS), which is associated with tissue damage. Anti-retroviral therapy (ART) is used to treat HIV and decrease the risk of opportunistic infections, but it is unclear whether ART reduces OS. Association of ART with OS was investigated. METHODS: We stratified a convenience sample of frozen serum or plasma from HIV-infected, ART-naïve (n=21); HIV-infected, ART-treated (n=14); HIV and PTB co-infected, ART-naïve (n=21); HIV and PTB co-infected, ART-treated (n=25) patients. Controls (n=21) were HIV-negative adults without TB symptoms. Concentration of OS markers namely: transaminases (ALT and AST), gamma glutamyl transpeptidase (GGT), albumin, total protein, malondialdehyde (MDA), vitamin C, and total anti-oxidant status (TAS) were determined. RESULTS: AST (p<0.001), GGT (p<0.001), total protein (p=0.001) and MDA (p<0.001) were higher in HIV patients compared to controls. Vitamin C (P<0.0001) and albumin (p<0.01) were lower in HIV-patients relative to controls. ART was only associated with higher albumin (p=0.001), higher GGT (p=0.02) and lower vitamin C (p=0.009). HIV and PTB co-infection was only significantly associated with higher GGT (p=0.01) and AST (p=0.03). CONCLUSION: We identified severe OS among HIV-patients. ART was associated with both increased and reduced markers of OS hence suggesting that ART may not attenuate OS.

Complete Genome Sequence of Mycobacterium avium subsp. hominissuis Strain H87 Isolated from an Indoor Water Sample.

Mycobacterium avium subsp. hominissuis is an environmentally acquired bacterium known to cause pulmonary and soft tissue infections, lymphadenitis, and disseminated disease in humans. We report here the complete genome sequence of strain H87, isolated from an indoor water sample, as a single circular chromosome of 5,626,623 bp with a G+C content of 68.8%.

Significant under expression of the DosR regulon in M. tuberculosis complex lineage 6 in sputum.

Mycobacterium africanum lineage (L) 6 is an important pathogen in West Africa, causing up to 40% of pulmonary tuberculosis (TB). The biology underlying the clinical differences between M. africanum and M. tuberculosis sensu stricto remains poorly understood. We performed ex vivo expression of 2179 genes of the most geographically dispersed cause of human TB, M. tuberculosis L4 and the geographically restricted, M. africanum L6 directly from sputa of 11 HIV-negative TB patients from The Gambia who had not started treatment. The DosR regulon was the most significantly decreased category in L6 relative to L4. Further, we identified nonsynonymous mutations in major DosR regulon genes of 44 L6 genomes of TB patients from The Gambia and Ghana. Using Lebek's test, we assessed differences in oxygen requirements for growth. L4 grew only at the aerobic surface while L6 grew throughout the medium. In the host, the DosR regulon is critical for M. tuberculosis in adaptation to oxygen limitation. However, M. africanum L6 appears to have adapted to growth under hypoxic conditions or to different biological niches. The observed under expression of DosR in L6 fits with the genomic changes in DosR genes, microaerobic growth and the association with extrapulmonary disease.

Adaptation of Mycobacterium tuberculosis to Impaired Host Immunity in HIV-Infected Patients.

BACKGROUND: It is unknown whether immunosuppression influences the physiologic state of Mycobacterium tuberculosis in vivo. We evaluated the impact of host immunity by comparing M. tuberculosis and human gene transcription in sputum between human immunodeficiency virus (HIV)-infected and uninfected patients with tuberculosis. METHODS: We collected sputum specimens before treatment from Gambians and Ugandans with pulmonary tuberculosis, revealed by positive results of acid-fast bacillus smears. We quantified expression of 2179 M. tuberculosis genes and 234 human immune genes via quantitative reverse transcription-polymerase chain reaction. We summarized genes from key functional categories with significantly increased or decreased expression. RESULTS: A total of 24 of 65 patients with tuberculosis were HIV infected. M. tuberculosis DosR regulon genes were less highly expressed among HIV-infected patients with tuberculosis than among HIV-uninfected patients with tuberculosis (Gambia, P < .0001; Uganda, P = .037). In profiling of human genes from the same sputa, HIV-infected patients had 3.4-fold lower expression of IFNG (P = .005), 4.9-fold higher expression of ARG1 (P = .0006), and 3.4-fold higher expression of IL10 (P = .0002) than in HIV-uninfected patients with tuberculosis. CONCLUSIONS: M. tuberculosis in HIV-infected patients had lower expression of the DosR regulon, a critical metabolic and immunomodulatory switch induced by NO, carbon monoxide, and hypoxia. Our human data suggest that decreased DosR expression may result from alternative pathway activation of macrophages, with consequent decreased NO expression and/or by poor granuloma formation with consequent decreased hypoxic stress.

Sputum is a surrogate for bronchoalveolar lavage for monitoring Mycobacterium tuberculosis transcriptional profiles in TB patients.

Pathogen-targeted transcriptional profiling in human sputum may elucidate the physiologic state of Mycobacterium tuberculosis (M. tuberculosis) during infection and treatment. However, whether M. tuberculosis transcription in sputum recapitulates transcription in the lung is uncertain. We therefore compared M. tuberculosis transcription in human sputum and bronchoalveolar lavage (BAL) samples from 11 HIV-negative South African patients with pulmonary tuberculosis. We additionally compared these clinical samples with in vitro log phase aerobic growth and hypoxic non-replicating persistence (NRP-2). Of 2179 M. tuberculosis transcripts assayed in sputum and BAL via multiplex RT-PCR, 194 (8.9%) had a p-value <0.05, but none were significant after correction for multiple testing. Categorical enrichment analysis indicated that expression of the hypoxia-responsive DosR regulon was higher in BAL than in sputum. M. tuberculosis transcription in BAL and sputum was distinct from both aerobic growth and NRP-2, with a range of 396-1020 transcripts significantly differentially expressed after multiple testing correction. Collectively, our results indicate that M. tuberculosis transcription in sputum approximates M. tuberculosis transcription in the lung. Minor differences between M. tuberculosis transcription in BAL and sputum suggested lower oxygen concentrations or higher nitric oxide concentrations in BAL. M. tuberculosis-targeted transcriptional profiling of sputa may be a powerful tool for understanding M. tuberculosis pathogenesis and monitoring treatment responses in vivo.

Empiric TB Treatment of Severely Ill Patients With HIV and Presumed Pulmonary TB Improves Survival.

RATIONALE: In 2007, World Health Organization (WHO) issued emergency recommendations on empiric treatment of sputum acid-fast bacillus smear-negative patients with possible tuberculosis (TB) in HIV-prevalent areas, and called for operational research to evaluate their effectiveness. We sought to determine if early, empiric TB treatment of possible TB patients with abnormal chest radiography or severe illness as suggested by the 2007 WHO guidelines, is associated with improved survival. METHODS: We prospectively enrolled consecutive HIV-seropositive inpatients at Mulago Hospital in Kampala, Uganda, from 2007 to 2011 with cough for ≥2 weeks. We retrospectively examined the effect of empiric TB treatment before discharge on 8-week survival among those with and without a WHO-defined "danger sign," including fever >39°C, tachycardia >120 beats per minute, or tachypnea >30 breaths per minute. We modeled the interaction between empiric TB treatment and danger signs, and their combined effect on 8-week survival, and adjusted for relevant covariates. RESULTS: Among 631 sputum smear-negative patients, 322 (51%) had danger signs. Cumulative 8-week survival of patients with danger signs was significantly higher with empiric TB treatment (80%) than without treatment (64%, P < 0.001). After adjusting for duration of cough and concurrent hypoxemia, patients with danger signs who received empiric TB treatment had a 44% reduction in 8-week mortality (risk ratio 0.56, 95% confidence interval: 0.34-0.91, P = 0.020). CONCLUSIONS: Empiric TB treatment of HIV-seropositive, smear-negative, presumed pulmonary TB patients with 1 or more danger signs is associated with improved 8-week survival. Enhanced implementation of the 2007 WHO empiric treatment recommendations should be encouraged whenever and wherever rapid and highly sensitive diagnostic tests for TB are unavailable.

Treatment of Non-Tuberculous Mycobacterial Lung Disease.

Treatment of non-tuberculous mycobacterial lung disease (NTM-LD) is challenging for several reasons including the relative resistance of NTM to currently available drugs and the difficulty in tolerating prolonged treatment with multiple drugs. Yet-to-be-done, large, multicenter, prospective randomized studies to establish the best regimens will also be arduous because multiple NTM species are known to cause human lung disease, differences in virulence and response to treatment between different species and strains within a species will make randomization more difficult, the need to distinguish relapse from a new infection, and the difficulty in adhering to the prescribed treatment due to intolerance, toxicity, and/or drug-drug interactions, often necessitating modification of therapeutic regimens. Furthermore, the out-of-state resident status of many patients seen at the relatively few centers that care for large number of NTM-LD patients pose logistical issues in monitoring response to treatment. Thus, current treatment regimens for NTM-LD is largely based on small case series, retrospective analyses, and guidelines based on expert opinions. It has been nearly 10 years since the publication of a consensus guideline for the treatment of NTM-LD. This review is a summary of the available evidence on the treatment of the major NTM-LD until more definitive studies and guidelines become available.

Blood Transcriptional Biomarkers for Active Tuberculosis among Patients in the United States: a Case-Control Study with Systematic Cross-Classifier Evaluation.

UNLABELLED: Blood transcriptional signatures are promising for tuberculosis (TB) diagnosis but have not been evaluated among U.S. PATIENTS: To be used clinically, transcriptional classifiers need reproducible accuracy in diverse populations that vary in genetic composition, disease spectrum and severity, and comorbidities. In a prospective case-control study, we identified novel transcriptional classifiers for active TB among U.S. patients and systematically compared their accuracy to classifiers from published studies. Blood samples from HIV-uninfected U.S. adults with active TB, pneumonia, or latent TB infection underwent whole-transcriptome microarray. We used support vector machines to classify disease state based on transcriptional patterns. We externally validated our classifiers using data from sub-Saharan African cohorts and evaluated previously published transcriptional classifiers in our population. Our classifier distinguishing active TB from pneumonia had an area under the concentration-time curve (AUC) of 96.5% (95.4% to 97.6%) among U.S. patients, but the AUC was lower (90.6% [89.6% to 91.7%]) in HIV-uninfected Sub-Saharan Africans. Previously published comparable classifiers had AUC values of 90.0% (87.7% to 92.3%) and 82.9% (80.8% to 85.1%) when tested in U.S. PATIENTS: Our classifier distinguishing active TB from latent TB had AUC values of 95.9% (95.2% to 96.6%) among U.S. patients and 95.3% (94.7% to 96.0%) among Sub-Saharan Africans. Previously published comparable classifiers had AUC values of 98.0% (97.4% to 98.7%) and 94.8% (92.9% to 96.8%) when tested in U.S. PATIENTS: Blood transcriptional classifiers accurately detected active TB among U.S. adults. The accuracy of classifiers for active TB versus that of other diseases decreased when tested in new populations with different disease controls, suggesting additional studies are required to enhance generalizability. Classifiers that distinguish active TB from latent TB are accurate and generalizable across populations and can be explored as screening assays.