Genomic Sequencing from Sputum for Tuberculosis Disease Diagnosis, Lineage Determination, and Drug Susceptibility Prediction

Diagnostic accuracy of nanopore sequencing for detecting Mycobacterium tuberculosis and drug-resistant strains: a systematic review and meta-analysis

Tuberculosis (TB), caused by Mycobacterium tuberculosis (MTB) infection, remains a significant public health threat. The timeliness, portability, and capacity of nanopore sequencing for diagnostics can aid in early detection and drug susceptibility testing (DST), which is crucial for effective TB control. This study synthesized current evidence on the diagnostic accuracy of the nanopore sequencing technology in detecting MTB and its DST profile. A comprehensive literature search in PubMed, Scopus, MEDLINE, Cochrane, EMBASE, Web of Science, AIM, IMEMR, IMSEAR, LILACS, WPRO, HERDIN Plus, MedRxiv, and BioRxiv was performed. Quality was assessed using the Quality Assessment of Diagnostic Accuracy Studies-2 tool. Pooled sensitivity, specificity, predictive values (PV), diagnostic odds ratio (DOR), and area under the curve (AUC) were calculated. Thirty-two studies were included; 13 addressed MTB detection only, 15 focused on DST only, and 4 examined both MTB detection and DST. No study used Flongle or PromethION. Seven studies were eligible for meta-analysis on MTB detection and five for DST; studies for MTB detection used GridION only while those for DST profile used MinION only. Our results indicate that GridION device has high sensitivity [88.61%; 95% CI (83.81-92.12%)] and specificity [93.18%; 95% CI (85.32-96.98%)], high positive predictive value [94.71%; 95% CI (89.99-97.27%)], moderately high negative predictive value [84.33%; 95% CI (72.02-91.84%)], and excellent DOR [107.23; 95% CI (35.15-327.15)] and AUC (0.932) in detecting MTB. Based on DOR and AUC, the MinION excelled in detecting pyrazinamide and rifampicin resistance; however, it underperformed in detecting isoniazid and ethambutol resistance. Additional studies will be needed to provide more precise estimates for MinION's sensitivity in detecting drug-resistance, as well as DOR in detecting resistance to pyrazinamide, streptomycin, and ofloxacin. Studies on detecting resistance to bedaquiline, pretomanid, and linezolid are lacking. Subgroup analyses suggest that overall accuracy of MTB detection tends to be higher with prospective study design and use of standards other than CSTB (Chinese national standard for diagnosing TB). Sensitivity analyses reveal that retrospective study design, use of GridION, and use of Illumina whole-genome sequencing (WGS) decrease overall accuracy in detecting any drug-resistant MTB. Findings from both types of analyses, however, should be interpreted with caution because of the low number of studies and uneven distribution of studies in each subgroup.

Targeted sputum sequencing for rapid and broad drug resistance of Mycobacterium tuberculosis

PURPOSE

Rapid detection of drug resistance in Mycobacterium tuberculosis (Mtb) from clinical samples facilitates the timely provision of optimal treatment regimens for tuberculosis (TB) patients.

METHODS

In November, 2023, the WHO released its second catalogue of resistance-conferring mutations in Mtb. Utilizing this information, we developed a single 17-plex PCR assay covering 16 key resistance genes and modified thermo-protection buffer to amplify 30 kbp DNA directly from sputum samples for nanopore sequencing. We implemented our protocol using rapid barcoding for sequencing with both a Flongle and a MinION flow cell.

RESULTS

The single multiplex PCR assay was successfully validated on clinical sputum samples using the thermo-protection buffer. The protocol was applied to both Flongle and MinION flow cells, analyzing 12 and 40 samples, respectively. Data analysis suggested that optimal performance could be achieved by processing 6 and 12 samples with similar microscope staining scores on these two platforms. This approach facilitated rapid antimicrobial resistance (AMR) predictions directly from sputum on the day of collection or the following day, with a cost of less than $35 per sample. Compared to AMR predictions based on whole-genome sequencing (WGS) using Mykrobe and TBProfiler, our amplicon-based analysis tool, ARapidTb, demonstrated superior resistance detection capabilities. When analyzing publicly available nanopore WGS datasets for 442 isolates, ARapidTb achieved agreement rates of 95.8% and 98.0%, outperforming Mykrobe (89.4% and 98.3%) and TBProfiler (75.6% and 89.8%).

CONCLUSIONS

Our study significantly reduces the time required for drug resistance detection, enabling quicker initiation of appropriate treatments and potentially improving patient outcomes and TB management.

Benchmarking reveals superiority of deep learning variant callers on bacterial nanopore sequence data

Variant calling is fundamental in bacterial genomics, underpinning the identification of disease transmission clusters, the construction of phylogenetic trees, and antimicrobial resistance detection. This study presents a comprehensive benchmarking of variant calling accuracy in bacterial genomes using Oxford Nanopore Technologies (ONT) sequencing data. We evaluated three ONT basecalling models and both simplex (single-strand) and duplex (dual-strand) read types across 14 diverse bacterial species. Our findings reveal that deep learning-based variant callers, particularly Clair3 and DeepVariant, significantly outperform traditional methods and even exceed the accuracy of Illumina sequencing, especially when applied to ONT's super-high accuracy model. ONT's superior performance is attributed to its ability to overcome Illumina's errors, which often arise from difficulties in aligning reads in repetitive and variant-dense genomic regions. Moreover, the use of high-performing variant callers with ONT's super-high accuracy data mitigates ONT's traditional errors in homopolymers. We also investigated the impact of read depth on variant calling, demonstrating that 10× depth of ONT super-accuracy data can achieve precision and recall comparable to, or better than, full-depth Illumina sequencing. These results underscore the potential of ONT sequencing, combined with advanced variant calling algorithms, to replace traditional short-read sequencing methods in bacterial genomics, particularly in resource-limited settings.

Bridging the gap between molecular and genomic epidemiology in tuberculosis: inferring MIRU-VNTR patterns from genomic data

The transition from MIRU-VNTR-based epidemiology studies in tuberculosis (TB) to genomic epidemiology has transformed how we track transmission. However, short-read sequencing is poor at analyzing repetitive regions such as the MIRU-VNTR loci. This causes a gap between the new genomic data and the large amount of information stored in historical databases. Long-read sequencing could bridge this knowledge gap by allowing analysis of repetitive regions. However, the feasibility of extracting MIRU-VNTRs from long reads and linking them to historical data has not been evaluated. In our study, an in silico arm, consisting of inference of MIRU patterns from long-read sequences (using MIRUReader program), was compared with an experimental arm, involving standard amplification and fragment sizing. We analyzed overall performance on 39 isolates from South Africa and confirmed reproducibility in a sample enriched with 62 clustered cases from Spain. Finally, we ran 25 consecutive incident cases, demonstrating the feasibility of correctly assigning new clustered/orphan cases by linking data inferred from genomic analysis to MIRU-VNTR databases. Of the 3,024 loci analyzed, only 11 discrepancies (0.36%) were found between the two arms: three attributed to experimental error and eight to misassigned alleles from long-read sequencing. A second round of analysis of these discrepancies resulted in agreement between the experimental and in silico arms in all but one locus. Adjusting the MIRUReader program code allowed us to flag potential in silico misassignments due to suboptimal coverage or unfixed double alleles. Our study indicates that long-read sequencing could help address potential chronological and geographical gaps arising from the transition from molecular to genomic epidemiology of tuberculosis.

IMPORTANCE

The transition from molecular epidemiology in tuberculosis (TB), based on the analysis of repetitive regions (VNTR-based genotyping), to genomic epidemiology transforms in the precision with which we track transmission. However, short-read sequencing, the most common method for performing genomic analysis, is poor at analyzing repetitive regions. This means that we face a gap between the new genomic data and the large amount of information stored in historical databases, which is also an obstacle to cross-national surveillance involving settings where only molecular data are available. Long-read sequencing could help bridge this knowledge gap by allowing analysis of repetitive regions. Our study demonstrates that MIRU-VNTR patterns can be successfully inferred from long-read sequences, allowing the correct assignment of new cases as clustered/orphan by linking new data extracted from genomic analysis to historical MIRU-VNTR databases. Our data may provide a starting point for bridging the knowledge gap between the molecular and genomic eras in tuberculosis epidemiology.

Genetic diversity within diagnostic sputum samples is mirrored in the culture of Mycobacterium tuberculosis across different settings

Culturing and genomic sequencing of Mycobacterium tuberculosis (MTB) from tuberculosis (TB) cases is the basis for many research and clinical applications. The alternative, culture-free sequencing from diagnostic samples, is promising but poses challenges to obtain and analyse the MTB genome. Paradoxically, culture is assumed to impose a diversity bottleneck, which, if true, would entail unexplored consequences. To unravel this paradox we generate high-quality genomes of sputum-culture pairs from two different settings after developing a workflow for sequencing from sputum and a tailored bioinformatics analysis. Careful downstream comparisons reveal sources of sputum-culture incongruences due to false positive/negative variation associated with factors like low input MTB DNA or variable genomic depths. After accounting for these factors, contrary to the bottleneck dogma, we identify a 97% variant agreement within sputum-culture pairs, with a high correlation also in the variants' frequency (0.98). The combined analysis from five different settings and more than 100 available samples shows that our results can be extrapolated to different TB epidemic scenarios, demonstrating that for the cases tested culture accurately mirrors clinical samples.

Improving the diagnosis of tuberculosis: old and new laboratory tools

INTRODUCTION

Despite recent advances in diagnostic technologies and new drugs becoming available, tuberculosis (TB) remains a major global health burden. If detected early, screened for drug resistance, and fully treated, TB could be easily controlled.

AREAS COVERED

Here the authors discuss M. tuberculosis culture methods which are considered the definitive confirmation of M. tuberculosis infection, and limited advances made to build on these core elements of TB laboratory diagnosis. Literature searches showed that molecular techniques provide enhanced speed of turnaround, sensitivity, and richness of data. Sequencing of the whole genome, is becoming well established for identification and inference of drug resistance. PubMed® literature searches were conducted (November 2022-March 2024).

EXPERT OPINION

This section highlights future advances in diagnosis and infection control. Prevention of prolonged hospital admissions and rapid TAT are of the most benefit to the overall patient experience. Host transcriptional blood markers have been used in treatment monitoring studies and, with appropriate evaluation, could be rolled out in a diagnostic setting. Additionally, the MBLA is being incorporated into latest clinical trial designs. Whole genome sequencing has enhanced epidemiological evidence. Artificial intelligence, along with machine learning, have the ability to revolutionize TB diagnosis and susceptibility testing within the next decade.

Predictive phage therapy for Escherichia coli urinary tract infections: Cocktail selection for therapy based on machine learning models

This study supports the development of predictive bacteriophage (phage) therapy: the concept of phage cocktail selection to treat a bacterial infection based on machine learning (ML) models. For this purpose, ML models were trained on thousands of measured interactions between a panel of phage and sequenced bacterial isolates. The concept was applied to Escherichia coli associated with urinary tract infections. This is an important common infection in humans and companion animals from which multidrug-resistant (MDR) bloodstream infections can originate. The global threat of MDR infection has reinvigorated international efforts into alternatives to antibiotics including phage therapy. E. coli exhibit extensive genome-level variation due to horizontal gene transfer via phage and plasmids. Associated with this, phage selection for E. coli is difficult as individual isolates can exhibit considerable variation in phage susceptibility due to differences in factors important to phage infection including phage receptor profiles and resistance mechanisms. The activity of 31 phage was measured on 314 isolates with growth curves in artificial urine. Random Forest models were built for each phage from bacterial genome features, and the more generalist phage, acting on over 20% of the bacterial population, exhibited F1 scores of >0.6 and could be used to predict phage cocktails effective against previously untested strains. The study demonstrates the potential of predictive ML models which integrate bacterial genomics with phage activity datasets allowing their use on data derived from direct sequencing of clinical samples to inform rapid and effective phage therapy.

Evaluation of Nanopore Sequencing for Diagnosing Pulmonary Tuberculosis Using Negative Smear Clinical Specimens

Purpose

This study aimed to evaluate the efficacy of nanopore sequencing for diagnosing pulmonary tuberculosis (PTB) using smear-negative clinical specimens.

Methods

We conducted a retrospective study based on a review of patient medical records to assess the accuracy of nanopore sequencing as a diagnostic tool for smear-negative PTB. Compared with clinical diagnosis, we determined the sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and area under the curve (AUC) of nanopore sequencing.

Results

A total of 647 patients were evaluated. Nanopore sequencing demonstrated an overall sensitivity of 91.7%, specificity of 85.3%, PPV of 95.1%, NPV of 76.4%, and AUC of 0.88. Notably, the overall diagnostic accuracy of nanopore sequencing was significantly higher than that of Mycobacterium tuberculosis (MTB) culture technique.

Conclusion

Nanopore sequencing exhibited satisfactory overall diagnostic accuracy for smear-negative PTB, regardless of MTB culture status. Therefore, if conditions permit, nanopore sequencing is recommended as a diagnostic method for smear-negative PTB.

Pangenome databases improve host removal and mycobacteria classification from clinical metagenomic data

BACKGROUND

Culture-free real-time sequencing of clinical metagenomic samples promises both rapid pathogen detection and antimicrobial resistance profiling. However, this approach introduces the risk of patient DNA leakage. To mitigate this risk, we need near-comprehensive removal of human DNA sequences at the point of sequencing, typically involving the use of resource-constrained devices. Existing benchmarks have largely focused on the use of standardized databases and largely ignored the computational requirements of depletion pipelines as well as the impact of human genome diversity.

RESULTS

We benchmarked host removal pipelines on simulated and artificial real Illumina and Nanopore metagenomic samples. We found that construction of a custom kraken database containing diverse human genomes results in the best balance of accuracy and computational resource usage. In addition, we benchmarked pipelines using kraken and minimap2 for taxonomic classification of Mycobacterium reads using standard and custom databases. With a database representative of the Mycobacterium genus, both tools obtained improved specificity and sensitivity, compared to the standard databases for classification of Mycobacterium tuberculosis. Computational efficiency of these custom databases was superior to most standard approaches, allowing them to be executed on a laptop device.

CONCLUSIONS

Customized pangenome databases provide the best balance of accuracy and computational efficiency when compared to standard databases for the task of human read removal and M. tuberculosis read classification from metagenomic samples. Such databases allow for execution on a laptop, without sacrificing accuracy, an especially important consideration in low-resource settings. We make all customized databases and pipelines freely available.

Molecular epidemiology of SARS-CoV-2 in Mongolia, first experience with nanopore sequencing in lower- and middle-income countries setting

BACKGROUND

Coronavirus disease (COVID-19) has had a significant impact globally, and extensive genomic research has been conducted on severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) lineage patterns and its variants. Mongolia's effective response resulted in low prevalence until vaccinations became available. However, due to the lack of systematically collected data and absence of whole genome sequencing capabilities, we conducted a two-stepped, nationally representative molecular epidemiologic study of SARS-CoV-2 in Mongolia for 2020 and 2021.

METHODS

We used retrospective analysis of stored biological samples from November 2020 to October 2021 and a variant-specific real-time reverse transcription polymerase chain reaction (RT-PCR) test to detect SARS-CoV-2 variants, followed by whole genome sequencing by Nanopore technology. Samples were retrieved from different sites and stored at -70°C deep freezer, and tests were performed on samples with cycle threshold <30.

RESULTS

Out of 4879 nucleic acid tests, 799 whole genome sequencing had been carried out. Among the stored samples of earlier local transmission, we found the 20B (B.1.1.46) variant predominated in the earlier local transmission period. A slower introduction and circulation of alpha and delta variants were observed compared to global dynamics in 2020 and 2021. Beta or Gamma variants were not detected between November 2020 and September 2021 in Mongolia.

CONCLUSIONS

SARS-CoV-2 variants of concerns including alpha and delta were delayed in circulation potentially due to public health stringencies in Mongolia. We are sharing our initial experience with whole genome sequencing of SARS-CoV-2 from Mongolia, where sequencing data is sparse.