Clustering of tuberculosis cases based on variable-number tandem-repeat typing in relation to the population structure of Mycobacterium tuberculosis in the Netherlands

Molecular diversity of Mycobacterium tuberculosis complex in Sikkim, India and prediction of dominant spoligotypes using artificial intelligence

In India, tuberculosis is an enormous public health problem. This study provides the first description of molecular diversity of the Mycobacterium tuberculosis complex (MTBC) from Sikkim, India. A total of 399 Acid Fast Bacilli sputum positive samples were cultured on Lőwenstein-Jensen media and genetic characterisation was done by spoligotyping and 24-loci MIRU-VNTR typing. Spoligotyping revealed the occurrence of 58 different spoligotypes. Beijing spoligotype was the most dominant type constituting 62.41% of the total isolates and was associated with Multiple Drug Resistance. Minimum Spanning tree analysis of 249 Beijing strains based on 24-loci MIRU-VNTR analysis identified 12 clonal complexes (Single Locus Variants). The principal component analysis was used to visualise possible grouping of MTBC isolates from Sikkim belonging to major spoligotypes using 24-MIRU VNTR profiles. Artificial intelligence-based machine learning (ML) methods such as Random Forests (RF), Support Vector Machines (SVM) and Artificial Neural Networks (ANN) were used to predict dominant spoligotypes of MTBC using MIRU-VNTR data. K-fold cross-validation and validation using unseen testing data set revealed high accuracy of ANN, RF, and SVM for predicting Beijing, CAS1_Delhi, and T1 Spoligotypes (93-99%). However, prediction using the external new validation data set revealed that the RF model was more accurate than SVM and ANN.

Universal genotyping reveals province-level differences in the molecular epidemiology of tuberculosis

OBJECTIVES

Compare the molecular epidemiology of tuberculosis (TB) between two large Canadian provinces-Ontario and British Columbia (BC)-to identify genotypic clusters within and across both provinces, allowing for an improved understanding of genotype data and providing context to more accurately identify clusters representing local transmission.

DESIGN

We compared 24-locus Mycobacterial Interspersed Repetitive Units-Variable Number of Tandem Repeats (MIRU-VNTR) genotyping for 3,314 Ontario and 1,602 BC clinical Mycobacterium tuberculosis isolates collected from 2008 through 2014. Laboratory data for each isolate was linked to case-level records to obtain clinical and demographic data.

RESULTS

The demographic characteristics of persons with TB varied between provinces, most notably in the proportion of persons born outside Canada, which was reflected in the large number of unique genotypes (n = 3,461). The proportion of clustered isolates was significantly higher in BC. Substantial clustering amongst non-Lineage 4 TB strains was observed within and across the provinces. Only two large clusters (≥10 cases/cluster) representing within province transmission had interprovincial genotype matches.

CONCLUSION

We recommend expanding analysis of shared genotypes to include neighbouring jurisdictions, and implementing whole genome sequencing to improve identification of TB transmission, recognize outbreaks, and monitor changing trends in TB epidemiology.

Assessment of the use and need for an integrated molecular surveillance of tuberculosis: an online survey in Germany

Background

The implementation of an integrated molecular surveillance (IMS) of tuberculosis (TB) is of high priority for TB control. IMS is defined as the systematic inclusion of molecular typing results in the national TB surveillance system. Although not standardized, an IMS of TB is already implemented in several low TB incidence countries. Germany is in the process of implementing a nationwide IMS of TB. This requires close collaboration between national and local health authorities. We conducted an online survey to understand the current use of molecular typing results for TB surveillance among the local public health offices (PHO)s in Germany, and to collect their perception and expectations towards the implementation of a nationwide IMS of TB.

Methods

The online survey was developed using the software Voxco and included 31 questions. The survey was sent to all the 377 local PHOs in Germany in April 2017. Responses were collected until June 2017.

Results

A total of 174/377 (46.2%) local PHOs participated in our survey, and 88/377 (23.3%) used molecular typing results in their routine TB surveillance work. The PHOs used molecular typing results especially as support for epidemiological contact tracing (62/88, 70.4%). We found statistically significant differences between answers of PHOs that did not use molecular typing results (n = 86) vs. PHOs that did use molecular typing results (n = 88): the latter perceived the use of molecular typing results as more beneficial for their work compared to the former (65.9% vs. 34.9%, p < 0.05). Moreover, the PHOs using molecular typing results expect for the future more support and coordination from regional and national public health institutes, especially regarding the identification and analysis of molecular clusters.

Conclusions

Our study is a step forward in the broader goal of implementing an IMS of TB in Germany. The local PHOs currently using the molecular typing results highlighted their positive attitude towards the implementation of an IMS, but also their needs of more support. Similar assessments might serve as an example for other countries which are on the way to implement a nationwide IMS of TB.

Electronic supplementary material

The online version of this article (10.1186/s12889-019-6631-6) contains supplementary material, which is available to authorized users.

A Quantitative Evaluation of MIRU-VNTR Typing Against Whole-Genome Sequencing for Identifying Mycobacterium tuberculosis Transmission: A Prospective Observational Cohort Study

Background

Mycobacterial Interspersed Repetitive Unit-Variable Number Tandem Repeat (MIRU-VNTR) typing is widely used in high-income countries to determine Mycobacterium tuberculosis relatedness. Whole-genome sequencing (WGS) is known to deliver greater specificity, but no quantitative prospective comparison has yet been undertaken.

Methods

We studied isolates from the English Midlands, sampled consecutively between 1 January 2012 and 31 December 2015. In addition to routinely performed MIRU-VNTR typing, DNA was extracted from liquid cultures and sequenced using Illumina technology. Demographic and epidemiological data for the relevant patients were extracted from the Enhanced Tuberculosis Surveillance system run by Public Health England. Closely related samples, defined using a threshold of five single nucleotide variants (SNVs), were compared to samples with identical MIRU-VNTR profiles, to samples from individuals with shared epidemiological risk factors, and to those with both characteristics.

Findings

1999 patients were identified for whom at least one M. tuberculosis isolate had been MIRU-VNTR typed and sequenced. Comparing epidemiological risk factors with close genetic relatedness, only co-residence had a positive predictive value of over 5%. Excluding co-resident individuals, 18.6% of patients with identical MIRU-VNTR profiles were within 5 SNVs. Where patients also shared social risk factors and ethnic group, this rose to 48%. Only 8% of MIRU-VNTR linked pairs in lineage 1 were within 5 SNV, compared to 31% in lineage 4.

Interpretation

In the setting studied, this molecular epidemiological study shows MIRU-VNTR typing and epidemiological risk factors are poorly predictive of close genomic relatedness, assessed by SNV. MIRU-VNTR performance varies markedly by lineage.

Funding

Public Health England, Health Innovation Challenge Fund, NIHR Health Protection Research Unit Oxford, NIHR Oxford Biomedical Research Centre.

•

M. tuberculosis relatedness was measured by MIRU-VNTR typing and single nucleotide variants (SNV).

•

In consecutive isolates in England, only 20% of samples with identical MIRU-VNTR profiles were within 5 SNVs.

•

MIRU-VNTR is more predictive of close relatedness in lineage 4 (Euro-American) lineages than in other lineages.

Epidemiological links between tuberculosis cases identified twice as efficiently by whole genome sequencing than conventional molecular typing: A population-based study

BACKGROUND

Patients with Mycobacterium tuberculosis isolates sharing identical DNA fingerprint patterns can be epidemiologically linked. However, municipal health services in the Netherlands are able to confirm an epidemiological link in only around 23% of the patients with isolates clustered by the conventional variable number of tandem repeat (VNTR) genotyping. This research aims to investigate whether whole genome sequencing (WGS) is a more reliable predictor of epidemiological links between tuberculosis patients than VNTR genotyping.

METHODS

VNTR genotyping and WGS were performed in parallel on all Mycobacterium tuberculosis complex isolates received at the Netherlands National Institute for Public Health and the Environment in 2016. Isolates were clustered by VNTR when they shared identical 24-loci VNTR patterns; isolates were assigned to a WGS cluster when the pair-wise genetic distance was ≤ 12 single nucleotide polymorphisms (SNPs). Cluster investigation was performed by municipal health services on all isolates clustered by VNTR in 2016. The proportion of epidemiological links identified among patients clustered by either method was calculated.

RESULTS

In total, 535 isolates were genotyped, of which 25% (134/535) were clustered by VNTR and 14% (76/535) by WGS; the concordance between both typing methods was 86%. The proportion of epidemiological links among WGS clustered cases (57%) was twice as common than among VNTR clustered cases (31%).

CONCLUSION

When WGS was applied, the number of clustered isolates was halved, while all epidemiologically linked cases remained clustered. WGS is therefore a more reliable tool to predict epidemiological links between tuberculosis cases than VNTR genotyping and will allow more efficient transmission tracing, as epidemiological investigations based on false clustering can be avoided.

A Predominant Variable-Number Tandem-Repeat Cluster of Mycobacterium tuberculosis Isolates among Asylum Seekers in the Netherlands and Denmark, Deciphered by Whole-Genome Sequencing

In many countries, Mycobacterium tuberculosis isolates are routinely subjected to variable-number tandem-repeat (VNTR) typing to investigate M. tuberculosis transmission. Unexpectedly, cross-border clusters were identified among African refugees in the Netherlands and Denmark, although transmission in those countries was unlikely. Whole-genome sequencing (WGS) was applied to analyze transmission in depth and to assess the precision of VNTR typing. WGS was applied to 40 M. tuberculosis isolates from refugees in the Netherlands and Denmark (most of whom were from the Horn of Africa) that shared the exact same VNTR profile. Cluster investigations were undertaken to identify in-country epidemiological links. Combining WGS results for the isolates (all members of the central Asian strain [CAS]/Delhi genotype), from both European countries, an average genetic distance of 80 single-nucleotide polymorphisms (SNPs) (maximum, 153 SNPs) was observed. The few pairs of isolates with confirmed epidemiological links, except for one pair, had a maximum distance of 12 SNPs. WGS divided this refugee cluster into several subclusters of patients from the same country of origin. Although the M. tuberculosis cases, mainly originating from African countries, shared the exact same VNTR profile, most were clearly distinguished by WGS. The average genetic distance in this specific VNTR cluster was 2 times greater than that in other VNTR clusters. Thus, identical VNTR profiles did not represent recent direct M. tuberculosis transmission for this group of patients. It appears that either these strains from Africa are extremely conserved genetically or there is ongoing transmission of this genotype among refugees on their long migration routes from Africa to Europe.

PyroTyping, a novel pyrosequencing-based assay for Mycobacterium tuberculosis genotyping

We developed a novel method, PyroTyping, for discrimination of Mycobacterium tuberculosis isolates combining pyrosequencing and IS6110 polymorphism. A total of 100 isolates were analysed with IS6110-restriction fragment length polymorphism (RFLP), spoligotyping, mycobacterial interspersed repetitive units - variable number tandem repeats (MIRU-VNTR), and PyroTyping. PyroTyping results regarding clustering or discrimination of the isolates were highly concordant with the other typing methods performed. PyroTyping is more rapid than RFLP and presents the same discriminatory power, thus, it may be useful for taking timely decisions for tuberculosis control.

Acquired and Transmitted Multidrug Resistant Tuberculosis: The Role of Social Determinants

Although risk factors for multi-drug resistant tuberculosis are known, few studies have differentiated between acquired and transmitted resistance. It is important to identify factors associated with these different mechanisms to optimize control measures. We conducted a prospective cohort study of index TB patients and their household contacts in Lima, Peru to identify risk factors associated with acquired and transmitted resistance, respectively. Patients with higher socioeconomic status (SES) had a 3-fold increased risk of transmitted resistance compared to those with lower SES when acquired resistance served as the baseline. Quality of housing mediated most of the impact of SES.

Tuberculosis microepidemics among dispersed migrants, Birmingham, UK, 2004-2013

To determine if local transmission was responsible for rising tuberculosis incidence in a recently dispersed migrant community in Birmingham, UK, during 2004–2013, we conducted enhanced epidemiologic investigation of molecular clusters. This technique identified exact locations of social mixing and chains of apparent recent transmission, which can be helpful for directing resources.

Genotypic characterization and historical perspective of Mycobacterium tuberculosis among older and younger Finns, 2008-2011

The Mycobacterium tuberculosis genotypes obtained from elderly Finns were assessed and compared with those obtained from younger Finns to comprehend the epidemiology of tuberculosis (TB) in Finland. From 2008 to 2011, a total of 1021 M. tuberculosis isolates were characterized by spoligotyping and 15-locus mycobacterial interspersed repetitive units-variable number tandem repeat typing. In total, 733 Finnish-born cases were included in the study, of which 466 (64%) were born before 1945 (older Finns). Of these, 63 (14%) shared an M. tuberculosis genotype with foreign-born or younger Finnish cases (born after 1945), and 59 (13%) shared a genotype with older Finnish cases. Eighty-five per cent had a unique genotypic profile while 70% belonged to T or Haarlem families, suggesting that ongoing transmission is infrequent among young and elderly Finns. Simultaneous reactivation of TB among older Finns was the most likely cause for clustering. As most isolates belonged to Haarlem or T, Finland was most likely affected by a similar TB epidemic at the beginning of the twentieth century as that seen in Sweden and Norway. Younger Finns were significantly more likely to be clustered (56% versus 27%, p<0.001), have pulmonary TB (87% versus 71%, p<0.001) and to be sputum smear positive (57% versus 48%, p<0.05) indicating that the risk of TB transmission from younger Finns is likely to be larger than from older Finns. The M. tuberculosis isolates from elderly Finns were associated with dominant lineages of the early twentieth century and differed from the heterogeneous lineages found among younger TB patients. Additionally, younger TB patients were more likely to transmit TB than elderly Finns.

Genomics and Machine Learning for Taxonomy Consensus: The Mycobacterium tuberculosis Complex Paradigm

Infra-species taxonomy is a prerequisite to compare features such as virulence in different pathogen lineages. Mycobacterium tuberculosis complex taxonomy has rapidly evolved in the last 20 years through intensive clinical isolation, advances in sequencing and in the description of fast-evolving loci (CRISPR and MIRU-VNTR). On-line tools to describe new isolates have been set up based on known diversity either on CRISPRs (also known as spoligotypes) or on MIRU-VNTR profiles. The underlying taxonomies are largely concordant but use different names and offer different depths. The objectives of this study were 1) to explicit the consensus that exists between the alternative taxonomies, and 2) to provide an on-line tool to ease classification of new isolates. Genotyping (24-VNTR, 43-spacers spoligotypes, IS6110-RFLP) was undertaken for 3,454 clinical isolates from the Netherlands (2004-2008). The resulting database was enlarged with African isolates to include most human tuberculosis diversity. Assignations were obtained using TB-Lineage, MIRU-VNTRPlus, SITVITWEB and an algorithm from Borile et al. By identifying the recurrent concordances between the alternative taxonomies, we proposed a consensus including 22 sublineages. Original and consensus assignations of the all isolates from the database were subsequently implemented into an ensemble learning approach based on Machine Learning tool Weka to derive a classification scheme. All assignations were reproduced with very good sensibilities and specificities. When applied to independent datasets, it was able to suggest new sublineages such as pseudo-Beijing. This Lineage Prediction tool, efficient on 15-MIRU, 24-VNTR and spoligotype data is available on the web interface “TBminer.” Another section of this website helps summarizing key molecular epidemiological data, easing tuberculosis surveillance. Altogether, we successfully used Machine Learning on a large dataset to set up and make available the first consensual taxonomy for human Mycobacterium tuberculosis complex. Additional developments using SNPs will help stabilizing it.

Comparison between RFLP and MIRU-VNTR genotyping of Mycobacterium tuberculosis strains isolated in Stockholm 2009 to 2011

Our aim was to analyze the difference between methods for genotyping of Mycobacterium tuberculosis complex isolates. We collected genotyping results from Restriction Fragment Length Polymorphism (RFLP) and Mycobacterial Interspersed Repetitive Units - Variable Numbers of Tandem Repeat (MIRU-VNTR) in a geographically limited area (Stockholm) during a period of three years. The number and proportion of isolates belonging to clusters was reduced by 45 and 35% respectively when combining the two methods compared with using RFLP or MIRU-VNTR only. The mean size of the clusters was smaller when combining methods and smaller with RFLP compared to MIRU-VNTR. In clusters with confirmed epidemiological links RFLP coincided slightly better than MIRU-VNTR but where there was a difference, the variation in MIRU-VNTR pattern was only in a single locus. In isolates with few IS6110 bands in RFLP, MIRU-VNTR differentiated the isolates more, dividing the RFLP clusters. Since MIRU-VNTR is faster and less labour-intensive it is the method of choice for routine genotyping. In most cases it will be sufficient for epidemiological purposes but true clustering might still be considered if there are epidemiological links and the MIRU-VNTR results differ in only one of its 24 loci.

Genotyping of clinical Mycobacterium tuberculosis isolates based on IS6110 and MIRU-VNTR polymorphisms

In this study, 155 clinical Mycobacterium tuberculosis isolates were subject to genotyping with fast ligation-mediated PCR (FLiP). This typing method is a modified mixed-linker PCR, a rapid approach based on the PCR amplification of HhaI restriction fragments of genomic DNA containing the 3' end of IS6110 and resolving the amplicons by polyacrylamide gel electrophoresis. The results were compared with previous data of the more commonly used methods, 15-locus mycobacterial interspersed repetitive unit-variable number tandem repeat (MIRU-VNTR) typing and, to verify combined FLiP/MIRU-VNTR clusters, the reference IS6110 restriction fragment length polymorphism (RFLP). FLiP banding patterns were highly reproducible and polymorphic. This method differentiated 119 types among the study set compared to 108 distinct MIRU-VNTR profiles. The discriminatory power of FLiP was slightly higher than that of MIRU-VNTR analysis (Hunter-Gaston Discriminatory Index = 0.991 and 0.990, resp.). Detailed comparison of the clusters defined by each of the methods revealed, however, a more apparent difference in the discriminatory abilities that favored FLiP. Clustering of strains by using combined results of these two PCR-based methods correlated well with IS6110 RFLP-defined clusters, further confirming high discriminatory potential of FLiP typing. These results indicate that FLiP could be an attractive and valuable secondary typing technique for verification of MIRU-VNTR clusters of M. tuberculosis strains.

Proposal of a consensus set of hypervariable mycobacterial interspersed repetitive-unit-variable-number tandem-repeat loci for subtyping of Mycobacterium tuberculosis Beijing isolates

Mycobacterium tuberculosis Beijing strains represent targets of special importance for molecular surveillance of tuberculosis (TB), especially because they are associated with spread of multidrug resistance in some world regions. Standard 24-locus mycobacterial interspersed repetitive-unit-variable-number tandem-repeat (MIRU-VNTR) typing lacks resolution power for accurately discriminating closely related clones that often compose Beijing strain populations. Therefore, we evaluated a set of 7 additional, hypervariable MIRU-VNTR loci for better resolution and tracing of such strains, using a collection of 535 Beijing isolates from six world regions where these strains are known to be prevalent. The typeability and interlaboratory reproducibility of these hypervariable loci were lower than those of the 24 standard loci. Three loci (2163a, 3155, and 3336) were excluded because of their redundant variability and/or more frequent noninterpretable results compared to the 4 other markers. The use of the remaining 4-locus set (1982, 3232, 3820, and 4120) increased the number of types by 52% (from 223 to 340) and reduced the clustering rate from 58.3 to 36.6%, when combined with the use of the standard 24-locus set. Known major clonal complexes/24-locus-based clusters were all subdivided, although the degree of subdivision varied depending on the complex. Only five single-locus variations were detected among the hypervariable loci of an additional panel of 92 isolates, representing 15 years of clonal spread of a single Beijing strain in a geographically restricted setting. On this calibrated basis, we propose this 4-locus set as a consensus for subtyping Beijing clonal complexes and clusters, after standard typing.