The Hidden Epidemic of Isoniazid-Resistant Tuberculosis in South Africa.

BACKGROUND: Isoniazid-resistant tuberculosis (Hr-TB) is often overlooked in diagnostic algorithms due to reliance on first-line molecular assays testing only for rifampicin resistance. OBJECTIVES: To determine the prevalence, outcomes and molecular mechanisms associated with Hr-TB in the Eastern Cape, South Africa. METHODS: Between April 2016 and October 2017, sputum samples were collected from patients with rifampin-susceptible TB at baseline and at weeks 7 and 23 of drug-susceptible TB treatment. We performed isoniazid phenotypic and genotypic drug susceptibility testing, FluorotypeMTBDR, Sanger sequencing, targeted next-generation sequencing (tNGS), and whole genome sequencing. RESULTS: We analysed baseline isolates from 766 patients with rifampin-susceptible TB. Of 89 patients (11.7%) found to have Hr-TB, 39 (44%) had canonical katG or inhA promoter mutations; 35 (39%) had non-canonical katG mutations (including 5 with underlying large deletions); 4 (5%) had mutations in other candidate genes associated with isoniazid resistance. For 11 (12.4%), no cause of resistance was found. CONCLUSIONS: Among patients with rifampin-susceptible TB diagnosed using first-line molecular TB assays, there is a high prevalence of Hr-TB. Phenotypic DST remains the gold standard. To improve performance of genetic-based phenotyping tests, all isoniazid resistance associated regions should be included, and such tests should have the ability to identify underlying mutations.

Bedaquiline for treatment of non-tuberculous mycobacteria (NTM): a systematic review and meta-analysis.

BACKGROUND: Non-tuberculous mycobacteria (NTM) infections are increasing in incidence and associated mortality. NTM are naturally resistant to a variety of antibiotics, complicating treatment. We conducted a literature assessment on the efficacy of bedaquiline in treating NTM species in vitro and in vivo (animal models and humans); meta-analyses were performed where possible. METHOD: Four databases were searched using specific terms. Publications were included according to predefined criteria. Bedaquiline's impact on NTM in vitro, MICs and epidemiological cut-off (ECOFF) values were evaluated. A meta-analysis of bedaquiline efficacy against NTM infections in animal models was performed. Culture conversion, cure and/or relapse-free cure were used to evaluate the efficacy of bedaquiline in treating NTM infection in humans. RESULTS: Fifty studies met the inclusion criteria: 33 assessed bedaquiline's impact on NTM in vitro, 9 in animal models and 8 in humans. Three studies assessed bedaquiline's efficacy both in vitro and in vivo. Due to data paucity, an ECOFF value of 0.5 mg/mL was estimated for Mycobacterium abscessus only. Meta-analysis of animal studies showed a 1.86× reduction in bacterial load in bedaquiline-treated versus no treatment within 30 days. In humans, bedaquiline-including regimens were effective in treating NTM extrapulmonary infection but not pulmonary infection. CONCLUSIONS: Bedaquiline demonstrated strong antibacterial activity against various NTM species and is a promising drug to treat NTM infections. However, data on the genomic mutations associated with bedaquiline resistance were scarce, preventing statistical analyses for most mutations and NTM species. Further studies are urgently needed to better inform treatment strategies.

Site-directed mutagenesis of Mycobacterium tuberculosis and functional validation to investigate potential bedaquiline resistance-causing mutations.

Molecular detection of bedaquiline resistant tuberculosis is challenging as only a small proportion of mutations in candidate bedaquiline resistance genes have been statistically associated with phenotypic resistance. We introduced two mutations, atpE Ile66Val and Rv0678 Thr33Ala, in the Mycobacterium tuberculosis H37Rv reference strain using homologous recombineering or recombination to investigate the phenotypic effect of these mutations. The genotype of the resulting strains was confirmed by Sanger- and whole genome sequencing, and bedaquiline susceptibility was assessed by minimal inhibitory concentration (MIC) assays. The impact of the mutations on protein stability and interactions was predicted using mutation Cutoff Scanning Matrix (mCSM) tools. The atpE Ile66Val mutation did not elevate the MIC above the critical concentration (MIC 0.25-0.5 µg/ml), while the MIC of the Rv0678 Thr33Ala mutant strains (> 1.0 µg/ml) classifies the strain as resistant, confirming clinical findings. In silico analyses confirmed that the atpE Ile66Val mutation minimally disrupts the bedaquiline-ATP synthase interaction, while the Rv0678 Thr33Ala mutation substantially affects the DNA binding affinity of the MmpR transcriptional repressor. Based on a combination of wet-lab and computational methods, our results suggest that the Rv0678 Thr33Ala mutation confers resistance to BDQ, while the atpE Ile66Val mutation does not, but definite proof can only be provided by complementation studies given the presence of secondary mutations.

COVID-19 detection in cough, breath and speech using deep transfer learning and bottleneck features.

We present an experimental investigation into the effectiveness of transfer learning and bottleneck feature extraction in detecting COVID-19 from audio recordings of cough, breath and speech. This type of screening is non-contact, does not require specialist medical expertise or laboratory facilities and can be deployed on inexpensive consumer hardware such as a smartphone. We use datasets that contain cough, sneeze, speech and other noises, but do not contain COVID-19 labels, to pre-train three deep neural networks: a CNN, an LSTM and a Resnet50. These pre-trained networks are subsequently either fine-tuned using smaller datasets of coughing with COVID-19 labels in the process of transfer learning, or are used as bottleneck feature extractors. Results show that a Resnet50 classifier trained by this transfer learning process delivers optimal or near-optimal performance across all datasets achieving areas under the receiver operating characteristic (ROC AUC) of 0.98, 0.94 and 0.92 respectively for all three sound classes: coughs, breaths and speech. This indicates that coughs carry the strongest COVID-19 signature, followed by breath and speech. Our results also show that applying transfer learning and extracting bottleneck features using the larger datasets without COVID-19 labels led not only to improved performance, but also to a marked reduction in the standard deviation of the classifier AUCs measured over the outer folds during nested cross-validation, indicating better generalisation. We conclude that deep transfer learning and bottleneck feature extraction can improve COVID-19 cough, breath and speech audio classification, yielding automatic COVID-19 detection with a better and more consistent overall performance.

Automatic cough classification for tuberculosis screening in a real-world environment.

Objective.The automatic discrimination between the coughing sounds produced by patients with tuberculosis (TB) and those produced by patients with other lung ailments.Approach.We present experiments based on a dataset of 1358 forced cough recordings obtained in a developing-world clinic from 16 patients with confirmed active pulmonary TB and 35 patients suffering from respiratory conditions suggestive of TB but confirmed to be TB negative. Using nested cross-validation, we have trained and evaluated five machine learning classifiers: logistic regression (LR), support vector machines, k-nearest neighbour, multilayer perceptrons and convolutional neural networks.Main Results.Although classification is possible in all cases, the best performance is achieved using LR. In combination with feature selection by sequential forward selection, our best LR system achieves an area under the ROC curve (AUC) of 0.94 using 23 features selected from a set of 78 high-resolution mel-frequency cepstral coefficients. This system achieves a sensitivity of 93% at a specificity of 95% and thus exceeds the 90% sensitivity at 70% specificity specification considered by the World Health Organisation (WHO) as a minimal requirement for a community-based TB triage test.Significance.The automatic classification of cough audio sounds, when applied to symptomatic patients requiring investigation for TB, can meet the WHO triage specifications for the identification of patients who should undergo expensive molecular downstream testing. This makes it a promising and viable means of low cost, easily deployable frontline screening for TB, which can benefit especially developing countries with a heavy TB burden.

COVID-19 cough classification using machine learning and global smartphone recordings.

We present a machine learning based COVID-19 cough classifier which can discriminate COVID-19 positive coughs from both COVID-19 negative and healthy coughs recorded on a smartphone. This type of screening is non-contact, easy to apply, and can reduce the workload in testing centres as well as limit transmission by recommending early self-isolation to those who have a cough suggestive of COVID-19. The datasets used in this study include subjects from all six continents and contain both forced and natural coughs, indicating that the approach is widely applicable. The publicly available Coswara dataset contains 92 COVID-19 positive and 1079 healthy subjects, while the second smaller dataset was collected mostly in South Africa and contains 18 COVID-19 positive and 26 COVID-19 negative subjects who have undergone a SARS-CoV laboratory test. Both datasets indicate that COVID-19 positive coughs are 15%-20% shorter than non-COVID coughs. Dataset skew was addressed by applying the synthetic minority oversampling technique (SMOTE). A leave-p-out cross-validation scheme was used to train and evaluate seven machine learning classifiers: logistic regression (LR), k-nearest neighbour (KNN), support vector machine (SVM), multilayer perceptron (MLP), convolutional neural network (CNN), long short-term memory (LSTM) and a residual-based neural network architecture (Resnet50). Our results show that although all classifiers were able to identify COVID-19 coughs, the best performance was exhibited by the Resnet50 classifier, which was best able to discriminate between the COVID-19 positive and the healthy coughs with an area under the ROC curve (AUC) of 0.98. An LSTM classifier was best able to discriminate between the COVID-19 positive and COVID-19 negative coughs, with an AUC of 0.94 after selecting the best 13 features from a sequential forward selection (SFS). Since this type of cough audio classification is cost-effective and easy to deploy, it is potentially a useful and viable means of non-contact COVID-19 screening.

MDR M. tuberculosis outbreak clone in Eswatini missed by Xpert has elevated bedaquiline resistance dated to the pre-treatment era.

BACKGROUND: Multidrug-resistant (MDR) Mycobacterium tuberculosis complex strains not detected by commercial molecular drug susceptibility testing (mDST) assays due to the RpoB I491F resistance mutation are threatening the control of MDR tuberculosis (MDR-TB) in Eswatini. METHODS: We investigate the evolution and spread of MDR strains in Eswatini with a focus on bedaquiline (BDQ) and clofazimine (CFZ) resistance using whole-genome sequencing in two collections ((1) national drug resistance survey, 2009-2010; (2) MDR strains from the Nhlangano region, 2014-2017). RESULTS: MDR strains in collection 1 had a high cluster rate (95%, 117/123 MDR strains) with 55% grouped into the two largest clusters (gCL3, n = 28; gCL10, n = 40). All gCL10 isolates, which likely emerged around 1993 (95% highest posterior density 1987-1998), carried the mutation RpoB I491F that is missed by commercial mDST assays. In addition, 21 (53%) gCL10 isolates shared a Rv0678 M146T mutation that correlated with elevated minimum inhibitory concentrations (MICs) to BDQ and CFZ compared to wild type isolates. gCL10 isolates with the Rv0678 M146T mutation were also detected in collection 2. CONCLUSION: The high clustering rate suggests that transmission has been driving the MDR-TB epidemic in Eswatini for three decades. The presence of MDR strains in Eswatini that are not detected by commercial mDST assays and have elevated MICs to BDQ and CFZ potentially jeopardizes the successful implementation of new MDR-TB treatment guidelines. Measures to limit the spread of these outbreak isolates need to be implemented urgently.

A landscape of genomic alterations at the root of a near-untreatable tuberculosis epidemic.

BACKGROUND: Atypical Beijing genotype Mycobacterium tuberculosis strains are widespread in South Africa and have acquired resistance to up to 13 drugs on multiple occasions. It is puzzling that these strains have retained fitness and transmissibility despite the potential fitness cost associated with drug resistance mutations. METHODS: We conducted Illumina sequencing of 211 Beijing genotype M. tuberculosis isolates to facilitate the detection of genomic features that may promote acquisition of drug resistance and restore fitness in highly resistant atypical Beijing forms. Phylogenetic and comparative genomic analysis was done to determine changes that are unique to the resistant strains that also transmit well. Minimum inhibitory concentration (MIC) determination for streptomycin and bedaquiline was done for a limited number of isolates to demonstrate a difference in MIC between isolates with and without certain variants. RESULTS: Phylogenetic analysis confirmed that two clades of atypical Beijing strains have independently developed resistance to virtually all the potent drugs included in standard (pre-bedaquiline) drug-resistant TB treatment regimens. We show that undetected drug resistance in a progenitor strain was likely instrumental in this resistance acquisition. In this cohort, ethionamide (ethA A381P) resistance would be missed in first-line drug-susceptible isolates, and streptomycin (gidB L79S) resistance may be missed due to an MIC close to the critical concentration. Subsequent inadequate treatment historically led to amplification of resistance and facilitated spread of the strains. Bedaquiline resistance was found in a small number of isolates, despite lack of exposure to the drug. The highly resistant clades also carry inhA promoter mutations, which arose after ethA and katG mutations. In these isolates, inhA promoter mutations do not alter drug resistance, suggesting a possible alternative role. CONCLUSION: The presence of the ethA mutation in otherwise susceptible isolates from ethionamide-naïve patients demonstrates that known exposure is not an adequate indicator of drug susceptibility. Similarly, it is demonstrated that bedaquiline resistance can occur without exposure to the drug. Inappropriate treatment regimens, due to missed resistance, leads to amplification of resistance, and transmission. We put these results into the context of current WHO treatment regimens, underscoring the risks of treatment without knowledge of the full drug resistance profile.

Geospatial distribution of Mycobacterium tuberculosis genotypes in Africa.

OBJECTIVE: To investigate the distribution of Mycobacterium tuberculosis genotypes across Africa. METHODS: The SITVIT2 global repository and PUBMED were searched for spoligotype and published genotype data respectively, of M. tuberculosis from Africa. M. tuberculosis lineages in Africa were described and compared across regions and with those from 7 European and 6 South-Asian countries. Further analysis of the major lineages and sub-lineages using Principal Component analysis (PCA) and hierarchical cluster analysis were done to describe clustering by geographical regions. Evolutionary relationships were assessed using phylogenetic tree analysis. RESULTS: A total of 14727 isolates from 35 African countries were included in the analysis and of these 13607 were assigned to one of 10 major lineages, whilst 1120 were unknown. There were differences in geographical distribution of major lineages and their sub-lineages with regional clustering. Southern African countries were grouped based on high prevalence of LAM11-ZWE strains; strains which have an origin in Portugal. The grouping of North African countries was due to the high percentage of LAM9 strains, which have an origin in the Eastern Mediterranean region. East African countries were grouped based on Central Asian (CAS) and East-African Indian (EAI) strain lineage possibly reflecting historic sea trade with Asia, while West African Countries were grouped based on Cameroon lineage of unknown origin. A high percentage of the Haarlem lineage isolates were observed in the Central African Republic, Guinea, Gambia and Tunisia, however, a mixed distribution prevented close clustering. CONCLUSIONS: This study highlighted that the TB epidemic in Africa is driven by regional epidemics characterized by genetically distinct lineages of M. tuberculosis. M. tuberculosis in these regions may have been introduced from either Europe or Asia and has spread through pastoralism, mining and war. The vast array of genotypes and their associated phenotypes should be considered when designing future vaccines, diagnostics and anti-TB drugs.

Isoniazid Resistance and Dosage as Treatment for Patients with Tuberculosis.

BACKGROUND: The first-line TB antibiotic isoniazid (INH) serves as a central component of combined first-line anti-tuberculosis drug therapy. However, resistance to INH has hindered the functioning of this drug. Resistance is caused by several known and unknown mutations in genes/regions in Mycobacterium tuberculosis (M. tuberculosis), followed by selection of these mutants in the presence of the drug. INH resistance can be categorised as either "high-level" (minimum inhibitory concentration (MIC) of > 1µg/mL to INH) or "low-level" (MIC between 0.1-1.0 µg/L) resistance and is dependent on the specific mutation acquired. The level of resistance is relevant, as INH resistance is often considered to be the first step in development of Multi-Drug Resistant (MDR) and extremely resistant (XDR) TB. Isoniazid is a pro-drug in which first pass metabolism happens via N-acetyltransferase and is fast, intermediate or slow, depending on the genetics of the host. Thus, low-level INH resistance, particularly in the presence of fast metabolism, could allow additional mutations, development of high-level resistance and progression to multi-drug resistance. METHODS: A structured search of bibliographic databases for peer-reviewed research literature was performed. Set parameters and specific inclusion criteria were used to filter the literature, based on our specified review questions. The quality and relevance of included papers was deduced using standard tools. The relevant content of cited papers was described, and an inferential qualitative content analysis methodology was utilised to analyse the inferences and findings of included studies using a conceptual framework. RESULTS: Seventy-eight papers were included in the review, of which a sub-set (36) of the papers describe how different genetic mutations result in low or high-level resistance to INH. These papers were also used to set up a diagram detailing how each mutation affects INH functionality in order to visualise the interactome of INH and M.tuberculosis A further twenty-eight out of the seventy-eight papers detail the methods for testing for INH resistance, current treatment regimens and factors that influence treatment outcome in order to better understand the role of INH within the current anti-tuberculosis treatment therapy and how its use can be optimised. CONCLUSION: The findings of this review suggest that low-level INH resistance, in the presence of fast-acetylation, is an underrated component of the global TB epidemic worldwide, and may be a significant problem in terms of treatment outcome and progression to antibiotic resistance. Thus, more research must be done to test whether personalised diagnostics and targeted high dose treatment with INH will reduce the incidence of isoniazid mono-resistant and multi-drug resistant (MDR) tuberculosis.

Programmatically selected multidrug-resistant strains drive the emergence of extensively drug-resistant tuberculosis in South Africa.

BACKGROUND: South Africa shows one of the highest global burdens of multidrug-resistant (MDR) and extensively drug-resistant (XDR) tuberculosis (TB). Since 2002, MDR-TB in South Africa has been treated by a standardized combination therapy, which until 2010 included ofloxacin, kanamycin, ethionamide, ethambutol and pyrazinamide. Since 2010, ethambutol has been replaced by cycloserine or terizidone. The effect of standardized treatment on the acquisition of XDR-TB is not currently known. METHODS: We genetically characterized a random sample of 4,667 patient isolates of drug-sensitive, MDR and XDR-TB cases collected from three South African provinces, namely, the Western Cape, Eastern Cape and KwaZulu-Natal. Drug resistance patterns of a subset of isolates were analyzed for the presence of commonly observed resistance mutations. RESULTS: Our analyses revealed a strong association between distinct strain genotypes and the emergence of XDR-TB in three neighbouring provinces of South Africa. Strains predominant in XDR-TB increased in proportion by more than 20-fold from drug-sensitive to XDR-TB and accounted for up to 95% of the XDR-TB cases. A high degree of clustering for drug resistance mutation patterns was detected. For example, the largest cluster of XDR-TB associated strains in the Eastern Cape, affecting more than 40% of all MDR patients in this province, harboured identical mutations concurrently conferring resistance to isoniazid, rifampicin, pyrazinamide, ethambutol, streptomycin, ethionamide, kanamycin, amikacin and capreomycin. CONCLUSIONS: XDR-TB associated genotypes in South Africa probably were programmatically selected as a result of the standard treatment regimen being ineffective in preventing their transmission. Our findings call for an immediate adaptation of standard treatment regimens for M/XDR-TB in South Africa.

Emergence and spread of extensively and totally drug-resistant tuberculosis, South Africa.

Factors driving the increase in drug-resistant tuberculosis (TB) in the Eastern Cape Province, South Africa, are not understood. A convenience sample of 309 drug-susceptible and 342 multidrug-resistant (MDR) TB isolates, collected July 2008-July 2009, were characterized by spoligotyping, DNA fingerprinting, insertion site mapping, and targeted DNA sequencing. Analysis of molecular-based data showed diverse genetic backgrounds among drug-sensitive and MDR TB sensu stricto isolates in contrast to restricted genetic backgrounds among pre-extensively drug-resistant (pre-XDR) TB and XDR TB isolates. Second-line drug resistance was significantly associated with the atypical Beijing genotype. DNA fingerprinting and sequencing demonstrated that the pre-XDR and XDR atypical Beijing isolates evolved from a common progenitor; 85% and 92%, respectively, were clustered, indicating transmission. Ninety-three percent of atypical XDR Beijing isolates had mutations that confer resistance to 10 anti-TB drugs, and some isolates also were resistant to para-aminosalicylic acid. These findings suggest the emergence of totally drug-resistant TB.

The rationale for using rifabutin in the treatment of MDR and XDR tuberculosis outbreaks.

Genetically related Mycobacterium tuberculosis strains with alterations at codon 516 in the rpoB gene were observed amongst a substantial number of patients with drug resistant tuberculosis in the Eastern Cape Province (ECP) of South Africa. Mutations at codon 516 are usually associated with lower level rifampicin (RIF) resistance, while susceptibility to rifabutin (RFB) remains intact. This study was conducted to assess the rationale for using RFB as a substitution for RIF in the treatment of MDR and XDR tuberculosis outbreaks. Minimum inhibitory concentrations (MICs) of 34 drug resistant clinical isolates of M tuberculosis were determined by MGIT 960 and correlated with rpoB mutations. RFB MICs ranged from 0.125 to 0.25 µg/ml in the 34 test isolates thereby confirming phenotypic susceptibility as per critical concentration (CC) of 0.5 µg/ml. The corresponding RIF MICs ranged between 5 and 15 µg/ml, which is well above the CC of 1.0 µg/ml. Molecular-based drug susceptibility testing provides important pharmacogenetic insight by demonstrating a direct correlation between defined rpoB mutation and the level of RFB susceptibility. We suggest that isolates with marginally reduced susceptibility as compared to the epidemiological cut-off for wild-type strains (0.064 µg/ml), but lower than the current CC (≤0.5 µg/ml), are categorised as intermediate. Two breakpoints (0.064 µg/ml and 0.5 µg/ml) are recommended to distinguish between susceptible, intermediate and RFB resistant strains. This concept may assist clinicians and policy makers to make objective therapeutic decisions, especially in situations where therapeutic options are limited. The use of RFB in the ECP may improve therapeutic success and consequently minimise the risk of ongoing transmission of drug resistant M. tuberculosis strains.

BDNF Val66Met and DRD2 Taq1A polymorphisms interact to influence PTSD symptom severity: a preliminary investigation in a South African population.

BACKGROUND: We evaluated the role that selected variants in serotonin transporter (5-HTT), dopamine receptor 2 (DRD2) and brain-derived neurotrophic factor (BDNF) genes play in PTSD symptom severity in an at-risk population. We also investigated the interaction between the genetic variants to determine whether these variables and the interactions between the variables influenced the severity of PTSD symptoms. METHODS: PTSD symptoms were quantitatively assessed using the Davidson Trauma Scale (DTS) in 150 participants from an at-risk South African population. All participants were genotyped for the 5-HTTLPR, DRD2 Taq1A and BDNF Val66Met polymorphisms. Gene-gene interactions were investigated using various linear models. All analyses were adjusted for age, gender, major depressive disorder diagnosis, level of resilience, level of social support and alcohol dependence. RESULTS: A significant interaction effect between DRD2 Taq1A and BDNF Val66Met variants on DTS score was observed. On the background of the BDNF Val66Val genotype, DTS score increased significantly with the addition of a DRD2 Taq1A A1 allele. However, on the BDNF Met66 allele background, the addition of an A1 allele was found to reduce total DTS score. CONCLUSIONS: This study provides preliminary evidence for an epistatic interaction between BDNF Val66Met and DRD2 Taq1A polymorphisms on the severity of PTSD symptoms, where both too little and too much dopamine can result in increased PTSD symptom severity.

Epistasis between antibiotic resistance mutations drives the evolution of extensively drug-resistant tuberculosis.

BACKGROUND AND OBJECTIVES: Multidrug resistant (MDR) bacteria are a growing threat to global health. Studies focusing on single antibiotics have shown that drug resistance is often associated with a fitness cost in the absence of drug. However, little is known about the fitness cost associated with resistance to multiple antibiotics. METHODOLOGY: We used Mycobacterium smegmatis as a model for human tuberculosis (TB) and an in vitro competitive fitness assay to explore the combined fitness effects and interaction between mutations conferring resistance to rifampicin (RIF) and ofloxacin (OFX); two of the most important first- and second-line anti-TB drugs, respectively. RESULTS: We found that 4 out of 17 M. smegmatis mutants (24%) resistant to RIF and OFX showed a statistically significantly higher or lower competitive fitness than expected when assuming a multiplicative model of fitness effects of each individual mutation. Moreover, 6 out of the 17 double drug-resistant mutants (35%) had a significantly higher fitness than at least one of the corresponding single drug-resistant mutants. The particular combinations of resistance mutations associated with no fitness deficit in M. smegmatis were the most frequent among 151 clinical isolates of MDR and extensively drug-resistant (XDR) Mycobacterium tuberculosis from South Africa. CONCLUSIONS AND IMPLICATIONS: Our results suggest that epistasis between drug resistance mutations in mycobacteria can lead to MDR strains with no fitness deficit, and that these strains are positively selected in settings with a high burden of drug-resistant TB. Taken together, our findings support a role for epistasis in the evolution and epidemiology of MDR- and XDR-TB.