| dc.description.abstract | Tuberculosis (TB), caused by Mycobacterium tuberculosis Complex (MTBC), is the joint leading cause of death worldwide due to a single infectious agent, with HIV/AIDS, and remains a major challenge to public health in both low- and high-prevalence countries. It is estimated to have killed 1.5 million people in 2014 (1.1 million co-infected with HIV/AIDS). Latent TB infection is estimated to affect approximately one third of the world’s population. TB has been statutorily notifiable in Ireland since 1947. Cases of TB in Ireland have declined from a high of 230 cases per 100,000 in the early 1950s to 6.9 per 100,000 in 2015. However, TB incidence has plateaued at this level in recent years and prevalence remains high in urban centres. Ireland is a European island nation that has experienced mass emigration and immigration over the last century. Immigration of people from outside the European Union, as well as free movement policies within the European Union, has been associated with the spread of TB, especially multi- and extensively-drug resistant TB (MDR/XDR-TB). MDR-TB is defined as resistance to isoniazid and rifampicin, while XDRTB is defined as resistance to the above plus a fluoroquinolone and aminoglycoside. Drug resistance- associated mutations have been discovered that can predict phenotypic resistance in MTBC. Disruption of transmission chains is a key factor in controlling the spread of TB. Seven global lineages of MTBC have been elucidated. Mycobacterial Interspersed Repetitive Units – Variable Number Tandem Repeat (MIRU-VNTR) genotyping is the established genotyping method. With the advent of ‘sequencing by synthesis’ Whole Genome, or Next Generation, Sequencing (WGS/NGS) has become more accessible, faster, and less costly, to the diagnostic laboratory. It is the ultimate genotyping tool, and its uses could be extended much further than genotyping alone. The primary aim of this study was to examine MTBC genotyping data collected at the Irish Mycobacteria Reference Laboratory (IMRL) from 2010-14 in order to assess the distribution of lineages present and to build on work already published in this regard (n=1,305 strains). It was clear from the results that clusters of MTBC were present in Ireland, and that this was worth further investigation by exploiting the higher resolution achieved by WGS (n=11 informative clusters chosen). It was hypothesised that MIRU-VNTR genotyping had over-estimated TB transmission events, that WGS could resolve these clusters with greater resolution, and that WGS would be a valuable addition to conventional genotyping currently in place in the IMRL. The MIRU-VNTR genotyping results also revealed that MDR/XDR-TB cases had increased in number over the period of the study. A comprehensive survey of MDR/XDR-TB (collected 2001-14, n=42 isolates from 41 patients) was consequently undertaken in order to characterise the strains circulating in Ireland. A number of different WGS analysis platforms were evaluated for their use in drug resistance prediction compared to phenotypic drug susceptibility testing (DST) – an algorithm designed by Walker and Kohl et al, online web-tools PhyResSe and TB Profiler, and ReseqTB data-sharing platform. The IMRL also took part in an international collaborative pilot study that aimed to introduce WGS techniques for mycobacterial identification, MTBC drug resistance prediction, and outbreak detection. This was the most comprehensive molecular characterisation of MTBC strains found in Ireland that has been performed to date. MIRU-VNTR genotyping is an excellent first-line tool for surveillance of the molecular epidemiology of MTBC in Ireland. Higher diversity of lineages was found within strains collected from 2010-14, than in previous Irish studies, due to the presence of six global lineages, including West African lineages 5 and 6. Euro-American lineage 4 remains predominant. MDR/XDR-TB is present in low, but ultimately increasing, numbers of cases, although mono-resistance remains below 5%. Median cluster size was 2. Within Euro-American lineage 4, clusters of 2 cases constituted over 50%, which compared to previous studies. However, 2.7% clusters were greater than 25 cases, which was not seen previously. Cluster analysis results showed that WGS could both rule-in and rule-out outbreaks with greater discrimination than MIRU-VNTR genotyping and could confirm recent transmission events, making it a valuable tool in the fight against TB, especially in cases where MIRU-VNTR genotypes are identical. From this and other studies, it is clear that WGS alone cannot infer all epidemiological information. Epidemiological data, contact tracing and genotyping all remain essential tools for MTBC cluster and outbreak investigation. The molecular characterisation of MDR/XDR-TB strains in Ireland from 2001-14 has proven that these strains are not being readily transmitted within the Irish population, that the drug resistant strains are similar to those found circulating in Europe, and that despite their high diversity, the drug-resistance-associated mutations they harboured were largely similar. WGS genotypic drug resistance prediction matched phenotypic DST in most cases. Statistical sensitivity varied widely, depending on the drug, and analysis platform used, while specificity ranged from 86-100%. When rifampicin and isoniazid alone were analysed, sensitivities ranged from 90-100%, better than rapid molecular tests already available (83-93%). For fluoroquinolones, WGS analysis resulted in sensitivity of 71%. Sensitivity for aminoglycosides and other drugs was lower, although specificity remained high. TB Profiler and PhyReSe resulted in the highest overall sensitivity compared to phenotypic DST. However, a high number of false positives was seen with TB Profiler (n=13). While isoniazid, rifampicin and fluoroquinolone genotypic results were reliable, others require more evidence of phenotypic-genotypic correlation if they are to be used diagnostically. WGS analysis, although currently not at a stage where it could replace phenotypic DST completely, would be an invaluable additional tool within the laboratory for rapid drug resistance prediction of MTBC. The international collaborative pilot study provided proof of principle that WGS could be employed for drug resistance prediction, nearest neighbour relatedness analysis that could flag outbreaks, and mycobacterial identification, improving laboratory turnaround times significantly, and even decreasing costs by 7% annually according to UK site figures. Despite its limitations, WGS represents a ‘game-changing’ technology for MTBC and many other microbiological applications.Tuberculosis (TB), caused by Mycobacterium tuberculosis Complex (MTBC), is the joint leading cause of death worldwide due to a single infectious agent, with HIV/AIDS, and remains a major challenge to public health in both low- and high-prevalence countries. It is estimated to have killed 1.5 million people in 2014 (1.1 million co-infected with HIV/AIDS). Latent TB infection is estimated to affect approximately one third of the world’s population. TB has been statutorily notifiable in Ireland since 1947. Cases of TB in Ireland have declined from a high of 230 cases per 100,000 in the early 1950s to 6.9 per 100,000 in 2015. However, TB incidence has plateaued at this level in recent years and prevalence remains high in urban centres. Ireland is a European island nation that has experienced mass emigration and immigration over the last century. Immigration of people from outside the European Union, as well as free movement policies within the European Union, has been associated with the spread of TB, especially multi- and extensively-drug resistant TB (MDR/XDR-TB). MDR-TB is defined as resistance to isoniazid and rifampicin, while XDRTB is defined as resistance to the above plus a fluoroquinolone and aminoglycoside. Drug resistance- associated mutations have been discovered that can predict phenotypic resistance in MTBC. Disruption of transmission chains is a key factor in controlling the spread of TB. Seven global lineages of MTBC have been elucidated. Mycobacterial Interspersed Repetitive Units – Variable Number Tandem Repeat (MIRU-VNTR) genotyping is the established genotyping method. With the advent of ‘sequencing by synthesis’ Whole Genome, or Next Generation, Sequencing (WGS/NGS) has become more accessible, faster, and less costly, to the diagnostic laboratory. It is the ultimate genotyping tool, and its uses could be extended much further than genotyping alone. The primary aim of this study was to examine MTBC genotyping data collected at the Irish Mycobacteria Reference Laboratory (IMRL) from 2010-14 in order to assess the distribution of lineages present and to build on work already published in this regard (n=1,305 strains). It was clear from the results that clusters of MTBC were present in Ireland, and that this was worth further investigation by exploiting the higher resolution achieved by WGS (n=11 informative clusters chosen). It was hypothesised that MIRU-VNTR genotyping had over-estimated TB transmission events, that WGS could resolve these clusters with greater resolution, and that WGS would be a valuable addition to conventional genotyping currently in place in the IMRL. The MIRU-VNTR genotyping results also revealed that MDR/XDR-TB cases had increased in number over the period of the study. A comprehensive survey of MDR/XDR-TB (collected 2001-14, n=42 isolates from 41 patients) was consequently undertaken in order to characterise the strains circulating in Ireland. A number of different WGS analysis platforms were evaluated for their use in drug resistance prediction compared to phenotypic drug susceptibility testing (DST) – an algorithm designed by Walker and Kohl et al, online web-tools PhyResSe and TB Profiler, and ReseqTB data-sharing platform. The IMRL also took part in an international collaborative pilot study that aimed to introduce WGS techniques for mycobacterial identification, MTBC drug resistance prediction, and outbreak detection. This was the most comprehensive molecular characterisation of MTBC strains found in Ireland that has been performed to date. MIRU-VNTR genotyping is an excellent first-line tool for surveillance of the molecular epidemiology of MTBC in Ireland. Higher diversity of lineages was found within strains collected from 2010-14, than in previous Irish studies, due to the presence of six global lineages, including West African lineages 5 and 6. Euro-American lineage 4 remains predominant. MDR/XDR-TB is present in low, but ultimately increasing, numbers of cases, although mono-resistance remains below 5%. Median cluster size was 2. Within Euro-American lineage 4, clusters of 2 cases constituted over 50%, which compared to previous studies. However, 2.7% clusters were greater than 25 cases, which was not seen previously. Cluster analysis results showed that WGS could both rule-in and rule-out outbreaks with greater discrimination than MIRU-VNTR genotyping and could confirm recent transmission events, making it a valuable tool in the fight against TB, especially in cases where MIRU-VNTR genotypes are identical. From this and other studies, it is clear that WGS alone cannot infer all epidemiological information. Epidemiological data, contact tracing and genotyping all remain essential tools for MTBC cluster and outbreak investigation. The molecular characterisation of MDR/XDR-TB strains in Ireland from 2001-14 has proven that these strains are not being readily transmitted within the Irish population, that the drug resistant strains are similar to those found circulating in Europe, and that despite their high diversity, the drug-resistance-associated mutations they harboured were largely similar. WGS genotypic drug resistance prediction matched phenotypic DST in most cases. Statistical sensitivity varied widely, depending on the drug, and analysis platform used, while specificity ranged from 86-100%. When rifampicin and isoniazid alone were analysed, sensitivities ranged from 90-100%, better than rapid molecular tests already available (83-93%). For fluoroquinolones, WGS analysis resulted in sensitivity of 71%. Sensitivity for aminoglycosides and other drugs was lower, although specificity remained high. TB Profiler and PhyReSe resulted in the highest overall sensitivity compared to phenotypic DST. However, a high number of false positives was seen with TB Profiler (n=13). While isoniazid, rifampicin and fluoroquinolone genotypic results were reliable, others require more evidence of phenotypic-genotypic correlation if they are to be used diagnostically. WGS analysis, although currently not at a stage where it could replace phenotypic DST completely, would be an invaluable additional tool within the laboratory for rapid drug resistance prediction of MTBC. The international collaborative pilot study provided proof of principle that WGS could be employed for drug resistance prediction, nearest neighbour relatedness analysis that could flag outbreaks, and mycobacterial identification, improving laboratory turnaround times significantly, and even decreasing costs by 7% annually according to UK site figures. Despite its limitations, WGS represents a ‘game-changing’ technology for MTBC and many other microbiological applications. | |
