See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/328706328 Low sensitivity of the MPT64 identification test to detect lineage 5 of the Mycobacterium tuberculosis complex Article  in  Journal of Medical Microbiology · November 2018 DOI: 10.1099/jmm.0.000846 CITATIONS READS 0 131 16 authors, including: Chakirath NDira Sanoussi Mathieu Odoun Laboratoire de Référence des Mycobactéries Cotonou Benin / Institute Of Tropical… Centre National Hospitalier Universitaire de Pneumo-phtisiologie du Bénin 15 PUBLICATIONS   124 CITATIONS    14 PUBLICATIONS   162 CITATIONS    SEE PROFILE SEE PROFILE Ousman Bodi Dorothy kyerewah Yeboah-Manu LRM Noguchi Memorial Institute for Medical Research 2 PUBLICATIONS   0 CITATIONS    234 PUBLICATIONS   2,212 CITATIONS    SEE PROFILE SEE PROFILE Some of the authors of this publication are also working on these related projects: Molecular epidemiology of tuberculosis in Ghana View project Study C View project All content following this page was uploaded by Dorothy kyerewah Yeboah-Manu on 05 November 2018. The user has requested enhancement of the downloaded file. RESEARCH ARTICLE Sanoussi et al., Journal of Medical Microbiology DOI 10.1099/jmm.0.000846 Low sensitivity of the MPT64 identification test to detect lineage 5 of the Mycobacterium tuberculosis complex C. N Dira Sanoussi,1,2,’ * Bouke C. de Jong,1 Mathieu Odoun,2 Karamatou Arekpa,2,3 Moulikatou Ali Ligali,2 Ousman Bodi,2 Simon Harris,4 Boatema Ofori-Anyinam,5 Dorothy Yeboah-Manu,6 Isaac Darko Otchere,6 Adwoa Asante-Poku,6 Severin Anagonou,2 Sebastien Gagneux,7 Mireia Coscolla,8 Leen Rigouts1,9 and Dissou Affolabi2 Abstract Purpose. Differentiation of the Mycobacterium tuberculosis complex (MTBc) from non-tuberculous mycobacteria (NTM) is important for tuberculosis diagnosis and is a prerequisite for reliable phenotypic drug-resistance testing. We evaluated the performance of the rapid MPT64 antigen identification test for the detection of Mycobacterium africanum lineage 5 (MAF L5). Methodology. Smear-positive tuberculosis patients’ sputa were included prospectively. Culture was performed on Löwenstein–Jensen medium and, when positive, the MPT64 test and the classical para-nitro benzoic acid susceptibility and heat-labile catalase (PNB/catalase) identification tests were performed. The MPT64 test was repeated 14 days after an initially negative first testing. Direct spoligotyping was performed for MTBc lineage determination. Results. In total, 333 isolates were tested for all of the methods. Three hundred and twenty-two (96.7%) were pure MTBc, by agreement between spoligotyping and PNB/catalase, and 11 were NTM or a mixture of MTBc/NTM. The MPT64 test conducted on day zero of culture-positivity correctly identified most of the pure MTBc isolates (93.2%, 300/322), but it failed to detect 24% of the L5 isolates (18/75) versus 2% (4/202) of the L4 ones [OR=15.6 (5.3–45.8), P<0.0001], with improved sensitivity for L5 detection on repeat testing after 14 days. The L5-wide non-synonymous single-nucleotide polymorphism in the mpt64 gene may explain the poor performance of the MPT64 test for L5. Conclusion. The MPT64 test has a lower sensitivity for detecting L5 isolates of the MTBc, and can be considered as a first- screening test that should be confirmed by another identification method when it produces negative results in countries with L5. Given the microbiological bias in both the isolation and identification of MAF lineages, diagnostics with high sensitivity for direct testing on clinical material are preferable. INTRODUCTION (Mycobacterium tuberculosis complex) from non-tubercu- lous mycobacteria (NTM) [2]. The advent of the molecular Tuberculosis (TB) remains a public health problem, espe- Cepheid GeneXpert MTB/RIF test partially resolved this cially in low-resource countries, where 95% of global tuber- problem, as it can confirm the presence of MTBc but not culosis is detected [1]. Microscopic detection of acid-fast NTM, and is also more sensitive than AFB microscopy for bacilli (AFB) has been the main TB diagnostic tool for more TB diagnosis [2]. Nevertheless, compared to culture, GeneX- than a century, yet it cannot distinguish MTBc pert MTB/RIF still has a lower sensitivity for the diagnosis of Received 25 May 2018; Accepted 7 September 2018 1 Author affiliations: Mycobacteriology Unit, Institute of Tropical Medicine, Antwerp, Belgium; 2Laboratoire de Reference des Mycobacteries, Cotonou, Benin; 3Genie de Biologie Humaine, Ecole Polytechnique d’Abomey-Calavi, Universite d’Abomey-Calavi, Benin; 4Wellcome Trust Sanger Institute, Hinxton CB10 1SA, UK; 5Vaccine and Immunity Theme, Medical Research Council (MRC) Unit, Serrekunda, The Gambia; 6Noguchi Memorial Institute for Medical Research, Legon, Accra, Ghana; 7Swiss Tropical and Public Health Institute, Basel, Switzerland; 8University of Valencia, 46010 Valencia, Spain; 9Department of Biomedical Sciences, University of Antwerp, Antwerp, Belgium. *Correspondence: C. N’Dira Sanoussi, ndirasanoussi@gmail.com Keywords: Rapid MPT64 identification test; Mycobacterium tuberculosis complex lineages; Mycobacterium africanum West African 1 (Lineage 5); non- tuberculous mycobacteria; mpt64 gene (Rv1980c); non-synonymous SNP. Abbreviations: AFB, acid-fast bacilli; CI, confidence interval; DST, drug susceptibility testing; D0, date when primary colonies were large enough that they could be scraped from the culture medium slant; D14, 14 days after D0; HCl, hydrochloric acid; IUATLD, International Union Against Tuberculosis and Lung diseases; L1, lineage 1; L2, lineage 2; L3, lineage 3; L4, lineage 4; L5, lineage 5; L6, lineage 6; LJ, Löwenstein–Jensen; LR, negative likelihood ratio; LR+, positive likelihood ratio; MAF, Mycobacterium africanum; MGIT, mycobacterial growth indicator tube; MTBc, Mycobacterium tuber- culosis complex; NaOH, sodium hydroxide; nSNP, non-synonymous single-nucleotide polymorphism; NTM, non-tuberculous mycobacteria; OR, odds ratio; PNB, para-nitro-benzoic acid; Sen, sensitivity; Spe, specificity; TB, tuberculosis. Two supplementary tables are available with the online version of this article. 000846 ã 2018 The Authors Downloaded from www.microbiologyresearch.org by This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. IP: 197.255.119.29 1 On: Mon, 05 Nov 2018 13:59:02 Sanoussi et al., Journal of Medical Microbiology TB in smear-negative pulmonary and extra-pulmonary TB clinic, all retreatment TB patients and a sample of new specimens [3, 4], as well as for TB in children [4]. Culture TB patients (the next four diagnosed after a retreatment TB thus remains the most sensitive laboratory test for the diag- patient) were included. Their sputa were collected before nosis of TB. The isolation of strains in culture is also a prereq- the initiation of TB treatment and then shipped to the uisite for conventional phenotypic drug susceptibility testing National Reference Laboratory for Mycobacteria (Labora- (DST) techniques. When a culture is positive, the rapid dif- toire de Reference des Mycobacteries) in Cotonou, Benin, ferentiation of MTBc from NTM is necessary for prompt TB where further laboratory analyses were performed. treatment initiation to ensure a better outcome. Moreover, Specimen preparation: sputum decontamination the differentiation of MTBc from NTM is crucial for a valid and culture of mycobacteria interpretation of resistance patterns in phenotypic DST assays, as NTM can be intrinsically resistant to anti-TB drugs The sputa were decontaminated using the Petroff method and can bemistaken formultidrug-resistantMTBc. [5] (2% NaOH final for 15min and neutralization using 1N HCl containing phenol red), followed by centrifugation Many techniques have been described for the differentiation at 3000 g and 4 C for 20min. The pellet was resuspended in of MTBc from NTM, such as growth in the presence of 1.5ml of phosphate-buffered saline (PBS). Two standard para-nitro benzoic acid (PNB), the heat-labile catalase test, Löwenstein–Jensen (LJ) slants and one LJ slant without cord-formation ability, hybridization with specific molecu- glycerol but supplemented with 0.5% sodium pyruvate were lar probes and high-performance liquid chromatography inoculated per patient’s sputum and then incubated at 37 C [5, 6]. Due to their speed and simplicity, the most popular and read weekly for 13weeks (90 days) before a negative tests are the rapid and simple immuno-chromatographic culture was reported. An aliquot of the sediment was stored methods, which yield results in 15 min [2]. The MPT64 at 20 C for direct spoligotype analysis. antigen test was reported to be highly sensitive and specific for the identification of MTBc in a systematic review [7]. MPT64 antigen rapid identification test Few studies have included strains isolated from solid The SD Bioline TB Ag MPT 64 Rapid assay (Standard Diag- medium [7, 8] or measured the performance of the test for nostics, Republic of Korea) was used for the identification of different MTBc lineages [9]. isolates from culture-positive specimens following the man- The human-adapted MTBc members comprise M. tubercu- ufacturer’s instructions [15]. The test was performed at day losis sensu stricto and Mycobacterium africanum (MAF). 0 (D0) and, if negative, was repeated 14 days later (D14). D0 The latter is subdivided into MAF West African 1 and 2, was defined as the date when primary colonies were large also called lineages 5 (L5) and 6 (L6), respectively, and is enough that they could be scraped from the slant. The test mostly restricted to the West African region, where it causes was repeated at D14 on colonies remaining on the primary  up to 40% of TB [10]. Recently, a study in The Gambia, slant after it had been reincubated at 37 C. A few colonies where L6 is prevalent, found that the MPT64 antigen test were suspended in 200 µl of the extraction buffer provided has low sensitivity for the detection of L6 isolates grown in in the kit or in the condensation fluid (if any) of the slant automated liquid mycobacterial growth indicator tubes [15]. Then, 100 µl of this suspension was deposited in the (MGITs) [9], while the sensitivity of the test for the detec- sample well of the MPT64 device [15]. After 15min incuba- tion of L5 isolates remains unknown. Nevertheless, a recent tion, the MPT64 was reported as negative if a colour signal study found a substitution (I43N) in the mpt64 gene of L5 line only appeared in the control band and not in the test isolates in Ghana [11], while Gagneux et al. [12] suggested band, and as positive if a colour signal line appeared in the that L5 has a non-synonymous single-nucleotide polymor- test band as well[15]. For MPT64-positive isolates, the phism (nSNP; not specified) in the mpt64 gene that could intensity of the positive colour signal line was also recorded impact negatively on the performance of the MPT64 anti- as ‘strong’ (strong intensity) or ‘faint’ (weaker intensity). gen test for L5 detection, affecting countries where this line- MPT64 devices were double-read by another person who age is endemic. was blinded for the results of the first reading, mainly because of faint positivity signals. The MPT64 antigen test We evaluated the performance of the MPT64 antigen rapid was performed without prior knowledge of the spoligotype identification test for L5 detection on isolates from solid cul- analysis results (spoligotype analysis for all specimens was ture medium in Benin, where L5 causes up to one-third of performed in batches after the availability of the MPT64 human TB [13, 14], and compared the available L5 and L6 results). The MPT64 test, spoligotype analysis and conven- genomes for nSNPs in the mpt64 gene. tional identification were performed blindly by different persons. METHODS Conventional identification with PNB/catalase along Patient selection and specimens with DST Presumptive TB patients were systematically screened with- Identification with PNB on LJ medium was performed AFB microscopy nationwide through the 24 largest TB clin- along with first-line DST. Mycobacterial suspensions were ics in Benin from April to December 2016. TB patients with inoculated on LJ medium with a final concentration of smear-positive sputa were prospectively recruited. In each 500 µgml1 PNB and on LJ control slants, as described by Downloaded from www.microbiologyresearch.org by IP: 197.2255.119.29 On: Mon, 05 Nov 2018 13:59:02 Sanoussi et al., Journal of Medical Microbiology Fig. 1. Specimen flow chart. *, using PNB/catalase and GenoType CM. †, MPT64 could not be repeated at D14 as very few colonies remained on the LJ slants, which were used for DST and PNB/catalase. ‡, eugonic isolate: isolate with easily scrapable colonies that grow easily in subculture. §, dysgonic isolate: isolate with very small dry flat or convex [27], hardly scrapable colonies, not improved by subculture. Dysgonic colonies can be hardly visible. the International Union Against Tuberculosis and Lung Determination of NTM species and prevalence in Diseases (IUATLD) [5]. The first-line DST included, as the study population usual, rifampicin, isoniazid, streptomycin and ethambutol, The Genotype Mycobacterium CM version 1 (Hain Life- respectively, at the critical concentrations of 40 µg ml1, 0.2 science) [16], molecular species identification test was used µg ml1, 4 µg ml1 and 2 µg ml1 using the proportion to identify presumed NTM isolates or MTBc/NTM mixtures method on LJ medium [5, 16]. As recommended by the based on PNB, catalase and spoligotyping results. DNA was IUATLD [5], when an isolate was PNB-susceptible and extracted from the isolates by heat inactivation (100 C for resistant to at least one of the first-line drugs, it was tested 20min)[17, 19]. PCR followed by hybridization of the PCR  for the production of 68 C-labile catalase to rule out NTM. products on the Genotype Hain CM strip were performed DNA extraction and spoligotyping as described by the manufacturer [22]. The H37Rv reference strain was included as a positive control and distilled water For each specimen, 200 µl of decontaminated sputum from was used as a negative control. After hybridization, the  the aliquot that had been stored at 20 C was heat-inacti- strips were fixed on an interpretation sheet and interpreted  vated at 100 C for 5min [17], followed by DNA extraction following the species profiles provided by the manufacturer. in 300 µl elution buffer using the Promega Maxwell16 Tis- sue DNA Purification kit (AS1030)[18] with the Promega Single-nucleotide polymorphism (SNP) analysis of Maxwell 16 machine model AS2000 version 4.9, as mpt64 gene in L5 described previously [19]. A mycobacterial sediment known To better understand the MPT64 test performance obtained to be PCR-positive for MTBc was included as a positive for the L5 strains in our study, we investigated whether extraction control and distilled water was used as a negative mutations are present in their mpt64 gene (Rv1980c) using extraction control. Spoligotyping (PCR followed by hybrid- Snippy 3.1 [23] with H37Rv (NC_000962.3) as the reference ization) was then performed as previously described [19, MTBc genome. The available whole-genome sequences 20]. The M. tuberculosis H37Rv and Mycobacterium bovis (FastQ files, File S1) from 25 strains belonging to L5 previ- BCG reference strains were included as positive PCR con- ously isolated in different countries were used (21 from trols and distilled water was used as a negative control. The Benin, 2 from Ivory Coast and 2 from Guinea). As a control MTBc lineage was assigned using the online tool, the TB- group, we also checked for the presence of mutations in the Lineage database [21]. mpt64 gene of the 18 L6 strains (2 from Benin, 4 from Downloaded from www.microbiologyresearch.org by IP: 197.2355.119.29 On: Mon, 05 Nov 2018 13:59:02 Sanoussi et al., Journal of Medical Microbiology Table 1. Overall performance of the MPT64 antigen test at day 0 (D0) and at day 14 (D14) MPT64 Spoligotyping PNB/catalase Combined PNB/catalase + spoligotyping Sensitivity (95% CI) Specificity (95% CI) LR+ (95% CI) LR (95% CI) MPT64 at D0 MTBc No profile (NTM or test Total MTBc NTM Total Pure NTM Mixture Total Sen: 93.2 (89.8– failure) MTBc spol MTBc + pnb 95.7) NTM Spe: 80 (28.4– + (MTBc) 366 1 367 300 3 303 300 1* 2† 303 99.5) (NTM) 26 4 30 22 8 30 22 4 4‡ 30 LR+: 4.7 (0.8– Total 392 5 397 322 11 333 322 5 6 333 26.9) LR: 0.09 (0.05– 0.16) MPT64 at MTBc NTM Mixture Total Sen: 96.2 (93.5– D0/D14 spol MTBc + pnb 98) NTM Spe: 80 (28.4– + (MTBc) 307 1* 2† 310 99.5) (NTM) 12 4 4‡ 20 LR+: 4.8 (0.8– Total 319§ 5 6 330§ 27.8) LR: 0.05 (0.02– 0.1) *Identified as MTBc by MPT64 but NTM by PNB/catalase and no spoligo bands or MTBc found using Genotype Mycobacterium CM, which identified this isolate as Mycobacterium scrofulaceum (or parafunicum or parasrofulaceum). †MTBc/NTM mixtures confirmed using Genotype Mycobacterium CM on the isolates. ‡Identified as NTM by MPT64 and PNB/catalase (agreement). Spoligotyping of specimens detected the presence of MTBc (two L6, one L4 and one M. bovis), yet no mixture (MTBc/NTM) found using Genotype CM on the isolates. §Three isolates that were MPT64-negative at D0 did not have MPT64 repeated at D14 and so are not included in the table. Burkina-Faso, 3 from Guinea, 5 from Ivory Coast and 4 Overall performance of MPT64 test from Senegal) for which genomes were available. For MPT64 at D0 genome sequencing, multiplexed Illumina libraries were In total, 397/434 (91.5%) positive cultures had MPT64 prepared following manufacturer’s guidelines using custom results available (Fig. 1, Table 1). Using either spoligotyping multiplex tags. Pooled samples were sequenced on an Illu- [19, 20] or PNB/catalase [5] as the reference standard for mina HiSeq2000 using the V4 kit to produce paired-end MTBc identification, 93.4% (366/392; 95%CI: 90.4–95.6) reads that were 100 bp in length. We aimed to attain an and 93.2% (300/322; 95%CI: 89.8–95.7) of MTBc isolates average depth of coverage of ~100-fold for each sample. were correctly classified as MTBc by the MPT64 test, respectively (Table 1). The PNB/catalase results showed that Statistical analyses six isolates (2 MPT64-positive and 4 MPT64-negative) iden- tified as MTBc using spoligotyping were identified pheno- We used Stata12.0 (StataCorp, USA) for statistical analyses. typically as NTM (PNB/catalase), suggesting the presence of The McNemar exact Chi-squared test was used to compare MTBc and NTM in the same specimen. This could be due paired data, and Fisher’s exact test was used for to a mixed infection of MTBc/NTM (Table 1). Excluding independent data. these isolates from the analysis, 93.2% (300/322; 95%CI: The odds ratios (OR), sensitivity, specificity, positive likeli- 89.8–95.7) of MTBc isolates were correctly classified as hood ratio (LR+) and negative likelihood ratio (LR) were MTBc by the MPT64 test on D0, with a specificity of 80% all calculated, along with the 95% confidence interval (CI). (4/5; 95%CI:28.4–99.5 (Table 1). The difference was considered significant when the two- MPT64 at D14 sided P-value was below 0.05. The MPT64 test was repeated at D14 for 19 of the 22 iso- lates with a negative MPT64 test at D0 (Fig. 1). Seven of the RESULTS 19 isolates became MPT64-positive, increasing the sensitiv- ity of the MPT64 test to 96.2% (307/319; 95%CI: 93.5–98), Specimens testing flow considering PNB/catalase and spoligotyping agreement as The specimen flow chart is presented in Fig. 1. the reference standard (Table 1). Downloaded from www.microbiologyresearch.org by IP: 197.2455.119.29 On: Mon, 05 Nov 2018 13:59:02 Sanoussi et al., Journal of Medical Microbiology Table 2. Performance of the MPT64 test across MTBc lineages Only pure MTBc isolates (detected using PNB/catalase and spoligotyping) are included in this comparison PNB/catalase (All MTBc) MTBc MPT64 at D0 OR with 95% CI, P* lineages Positive Negative Total (MTBc) L1 15 (100%) 0 15 – L2 14 (100%) 0 14 – L3 3 (100%) 0 3 – L4 198 (98%) 4 202 . OR=0.06 (0.02 to 0.19) L5 57 (76%) 18 75 . P<0.0001 L6 8 (100%) 0 8 – M. bovis 5 (100%) 0 5 – Total 300 22 322 Overall P<0.001 MPT64 at D0-14 L1 15 (100%) 0 15 – L2 14 (100%) 0 14 – L3 3 (100%) 0 3 – L4 198 (98.5%) 3† 201 . OR=0.11 (0.03 to 0.38) L5 64 (87.7%) 9‡ 73 . P=0.0005 L6 8 (100%) 0 8 – M. bovis 5 (100%) 0 5 – Total 307 12 319 Overall P=0.018 *P-values were calculated using Fisher’s exact test. †MPT64 was not performed at D14 for one specimen that was MPT64-negative at D0 and so it was excluded. ‡MPT64 was not performed at D14 for two specimens that were MPT64-negative at D0 and so they were excluded; MPT64 became positive for seven specimens. Species of NTM identified Performance of MPT64 test across MTBc lineages Of the 11 isolates phenotypically identified as not being For the following comparisons, NTMs and mixtures were pure MTBc, 6 were considered to be mixtures (MTBc/ excluded; only pure MTBc (PNB/catalase and spoligotyp- NTM) based on the double banding profile on the Genotype ing-confirmed) isolates (322/333, Table 1) were included. Mycobacterium CM strip and/or a spoligotyping profile MPT64 at D0 in MTBc isolates from the sputum specimen combined with growth of the At D0, MPT64 positivity varied significantly across lineages isolate on the PNB tube. The five pure NTM isolates (P<0.001, Table 2). MPT64 was positive for all isolates from included one M. kansasii, one M. abscessus (or M. immuno- lineages 1, 2, 3, 6 and M. bovis, whereas almost one-quarter genum), one M. malmoense (or M. haemophilum, M. pal- (24%) of the MAF L5 isolates and 2% of the L4 isolates ustre or M. nebraskense), one M. scrofulaceum (or M. were MPT64-negative. MPT64-positive isolates were signifi- paraffunicum or M. parascrofulaceum) and one M. species. cantly under-represented among L5 versus L4 isolates, the Among the MTBc/NTM mixtures, two isolates were posi- most prevalent lineage (OR=0.06, 95%CI: 0.02–0.19, tive in the MPT64 assay (Table 1). They were confirmed as P<0.0001; Table 2), corresponding to 15.6-fold odds mixtures by Genotype CM and the spoligotype banding (95%CI: 5.3–45.8) of MPT64 false-negativity in L5 isolates profiles, revealing L4 + M. fortuitum and L6 + M. species. (P<0.0001). For the four remaining presumed mixtures, spoligotyping of MPT64 at D14 in MTBc isolates the sputum suggested the co-existence of MTBc, whereas no The 19 isolates that were MPT64-negative at D0 and repeat MTBc was detected in the isolates with only the NTM probe tested at D14 included 16 L5 and 3 L4 isolates. MPT64 of the Genotype CM reacting and growth on PNB. The became positive for an additional seven (43.8%) L5 isolates, combined results suggested the following mixtures: M. bovis significantly increasing the positivity of MPT64 from 76% + M. intracellulare, L4 + M. fortuitum and 2 L6 + M. at D0 to 87.7% at D14 (RR=1.1, 95%CI: 1.0–1.2, P=0.016, intracellulare. The overall proportion of NTM identified exact McNemar). None of the three L4 isolates that were was 3.3% (11/333). M. intracellulare followed by M. fortui- MPT64-negative at D0 became positive. Despite the positiv- tum were the most observed. ity at D14 of some previously MPT64-negative isolates, Downloaded from www.microbiologyresearch.org by IP: 197.2555.119.29 On: Mon, 05 Nov 2018 13:59:02 Sanoussi et al., Journal of Medical Microbiology Table 3. Variation of the intensity of the positivity signal line of the MPT64 strip across MTBc lineages Intensity of MPT64 cartridge positivity signal band (MPT64 positive specimens) MPT64 positive at D0 OR (95% CI), P* Strong signal Faint signal Total MTBc Lineages L1 14 1 (6.7%) 15 - L2 14 0 14 - L3 3 0 3 - L4 195 3 (1.5%) 198 .OR=21.2 (6.2–71.5) L5 43 14 (24.6%) 57 .P<0.0001 L6 7 1 (12.5%) 8 - M. bovis 5 0 5 - Total 281 19 300 Modern MTBc (L2+L3+L4) 212 3 (1.4%) 215 .OR=17.7 (5.3–58.4) Ancestral MTBc (L1 +L5+L6) 64 16 (20%) 80 .P<0.0001 Other than MAF (L1+L2+L3+L4) 226 4 (1.7%) 230 .OR=16.9 (5.6–50.6) MAF (L5+L6) 50 15 (23.1%) 55 .P<0.0001 MPT64 positive at D0-14 L1 14 1 (6.7%) 15 - L2 14 0 14 - L3 3 0 3 - L4 195 3 (1.5%) 198 .OR=21.7 (6.4–72.2) L5 48 16 (25%) 64 .P<0.0001 L6 7 1 (12.5%) 8 - M. bovis 5 0 5 - Total 286 21 307 Modern MTBc (L2+L3+L4) 212 3 (1.4%) 215 .OR=18.4 (5.6–60.3) Ancestral MTBc (L1 +L5+L6) 69 18 (20.7%) 87 .P<0.0001 Other than MAF (L1+L2+L3+L4) 226 4 (1.7%) 230 .OR=17.5 (5.9–51.4) MAF (L5+L6) 55 17 (23.6%) 72 .P<0.0001 *P-values were calculated using Fisher’s exact test. MPT64 positivity still varied across lineages (Table 2). The Variation of the time from inoculation (start of difference in MPT64 false-negativity between L5 versus L4 incubation) to the realization of the MPT64 test (Dt). isolates was still strongly significant, with 12.3% (9/73) of The median Dt at D0 was shortest for L1, L3 and L4 L5 isolates versus 1.5% (3/201) from L4 remaining negative (3 weeks) followed by L2 and M. bovis (4 weeks), L5 at D14 (false-negativity OR=9.3, 95%CI: 2.6–32.6, (6 weeks) and L6 (8 weeks). Including repeated testing, the P=0.0005). median Dt increased to 8weeks for L5, while it remained similar for the other lineages (Table S2). Variation of the intensity of the positivity signal (colour) line of the MPT64 cartridge SNP analysis of mpt64 gene in L5 At D0, a total of 19 (6.3%) of the 300 MPT64-positive iso- All 25 of the L5 isolate genomes shared the same nSNP lates had a faint signal. Most of them were from L5 (24.6%, 128T>A (I43N) in the mpt64 gene (File S1). One of the L5 14/57), followed by L6 (12.5 %, 1/8), versus only 1.5% (3/ genomes had in addition, a synonymous SNP 519G>A 198) from L4 (Table 3). The odds of obtaining a faint signal (V197V). Among the L6 isolate genomes, there was no in L5 isolates was 21.2-fold (95%CI: 6.2–71.5) the odds of mutation in the mpt64 gene for 17 of the 18 isolates ana- that for L4 isolates (P<0.0001), which did not decrease after lysed. One genome harboured a synonymous SNP 81C>T the D14 repeat (OR=21.7, 95%CI: 6.5–72.2, P<0.0001) (T27T) in the gene. (Table 3). The proportion of isolates with a faint signal was Lineage distribution among population of isolates also higher in the MAF group (L5+L6) versus the non-MAF included versus excluded in the MPT64 comparison group (P<0.0001) or the ancestral lineages group (L1+L5 The lineage distribution in the isolates (pure MTBc) +L6) compared to the modern lineages group (P<0.0001, included in the MPT64 vs combined spoligotyping and Table 3). PNB/catalase comparison differed significantly to the Downloaded from www.microbiologyresearch.org by IP: 197.2655.119.29 On: Mon, 05 Nov 2018 13:59:02 Sanoussi et al., Journal of Medical Microbiology Downloaded from www.microbiologyresearch.org by IP: 197.2755.119.29 On: Mon, 05 Nov 2018 13:59:02 Table 4. MTBc isolate culture outcome across lineages: lineage distribution among the population of isolates that was included versus that among the population that was excluded (dysgonic/ partially contaminated) in the MPT64/PNB comparison. P-values were calculated using Fischer’s exact test All specimens (culture All culture- Included Included vs excluded for quality (dysgonic, partial contamination) Included (eugonic) vs excluded (dysgonic only) positive, negative and positive isolates Excluded isolates OR %Difference contaminated) % (n) specimens (pure excluded/ excluded - P Dysgonic ORdysgonic among %Difference dysgonic P (dysgonic/partially included, 95%CI included, 95%CI among% (n) MTBc) % excluded/ eugonic among excluded - eugonic MTBc contaminated) % excluded (included) , 95%CI(n) (included) , 95%CI lineages (n) % (n) Total n=513* n=428† n=322 n=79‡ n=72 L1 3.9 (20) 4.2 (18) 4.7 (15) 3.8 (3) 0.8 (0.2 to 2.7) 0.9 (5.7 to 3.9) 1 2.8 (2) 0.6 (0 to 2.4) 1.9 (6.3 to 2.6) 0.749 L2 4.3 (22) 4.7 (20) 4.3 (14) 3.8 (3) 0.9 (0.3 to 2.9) 0.6 (5.3 to 4.2) 1 2.8 (2) 0.6 (0 to 2.5) 1.6 (6 to 2.8) 0.747 L3 0.8 (4) 0.7 (3) 0.9 (3) 0 (0) 0 (0 to 5.3) 0.9 (2 to 0.1) 1 0 (0) 0 (0 to 5.8) 0.9 (2 to 0.1) 1 L4 52.4 (269) 56.1 (240) 62.7 (202) 29.1 (23) 0.2 (0.1 to 0.4) 33.6 (44.9 to 22.3) <0.0001 25 (18) 0.2 (0.1 to 0.4) 37.7 (49 to 26.4) <0.0001 L5 27.9 (143) 26.4 (113) 23.3 (75) 43.0 (34) 2.5 (1.5 to 4.2) 19.7 (7.9 to 31.6) 0.0007 47.2 (34) 3 (1.7 to 5) 23.9 (11.5 to 36.4) 0.0001 L6 8.4 (43) 6.1 (26) 2.5 (8) 17.7 (14) 8.5 (3.5 to 20.5) 15.2 (6.6 to 23.8) <0.0001 19.4 (14) 9.5 (3.9 to 23.1) 17 (7.7 to 26.3) <0.0001 M. bovis 2.3 (12) 1.9 (8) 1.6 (5) 2.5 (2) 1.7 (0 to 7.5) 1 (2.7 to 4.7) 0.628 2.8 (2) 1.8 (0 to 8.3) 1.2 (2.8 to 5.3) 0.616 Other than 63.7 (327) 67.5 (289) 74.2 (239) 39.2 (31) 0.2 (0.1 to 0.4) 35 (46.8 to 23.2) <0.0001 33.3 (24) 0.2 (0.1 to 0.3) 40.9 (52.8 to 29) <0.0001 MAF MAF 36.3 (186) 32.5 (139) 25.8 (83) 60.8 (48) 4.5 (2.7 to 7.5) 35 (23.2 to 46.8) 66.7 (48) 5.8 (3.3 to 9.9) 40.9 (29 to 52.8) *There were 11 specimens that did not yield a spoligotype profile among the 524 specimens. †There were 6 specimens that did not yield a spoligotype profile among the 434 culture-positive specimens. ‡There was 1 specimen with failed spoligotyping among the 80 dysgonic/partially contaminated specimens (101 excluded in total – 21 with PNB available and MPT64 missing). Sanoussi et al., Journal of Medical Microbiology distribution in isolates excluded for the unavailability of Oettinger et al. created five monoclonal antibodies MPT64 and/or PNB/catalase results (poor quality for subse- (C24b1-3, L24b4-5) that reacted with four epitopes of the quent tests) (Table 4). MAF L5 and L6 were significantly MPT64 antigen [26]. The MPT64 antigen’s epitope for the over-represented in that excluded group relative to the C24b3 antibody consists of two structural domains found included group (OR=4.5, 95% CI: 2.7–7.5, P<0.0001). The in the sequences Ala1-Leu43 and Ala108-Ser152 [26]. The excluded group of isolates (n=80) included 72 dysgonic and Ile43Asn mutation (I43N) found in all L5 isolates genomes 8 partially contaminated isolates. When comparing the coincides with the final amino- acid of the first part of included group solely to the dysgonic isolates among the this epitope. This could lead to partial binding and lower excluded group, we found that the dysgonic nature of the adherence or prevent the binding of the C24b3 antibody isolates was strongly associated with the lineage, with L5 to its partially modified epitope in the MPT64 antigen, and L6 being over-represented among dysgonic isolates resulting in a faint positivity signal band or negative (OR=5.7, 95%CI: 3.3–9.9, P<0.0001), especially L6 MPT64 test if that antibody was used in the development (OR=9.5, 95%CI: 3.9–23.1, P<0.0001) (Table 4). of the MPT64 test. The mouse antibodies (at least three) [15] used in the development of the MPT64 test are not DISCUSSION specified, but are likely included among the five detected Our evaluation of LJ-based primary cultures from smear- by Oettinger et al. [26], as this paper is cited in the SD positive TB patients in Benin in a nationwide prospective Bioline MPT64 sheet [15]. This could explain why, despite study suggests that the SD Bioline MPT64 antigen test offers the non-synonymous mutation in the mpt64 gene of L5, relatively low performance for the rapid identification of the there was a significant proportion of isolates with faint L5 of MTBc. Its lower sensitivity for the confirmation of L5 positivity signal bands and isolates with strong positivity as MTBc leads to the misclassification of L5 as NTM. signal bands, and a significant proportion of MPT64-nega- Repeating the test 14 days after a first negative result tive isolates (even after a D14 repeat). Nevertheless, chang- improved the sensitivity for L5 detection significantly, but ing the mouse antibody from which the corresponding not completely. The few L4 isolates that tested negative at MPT64 epitope is mutated in L5 isolates may improve the D0 (1.5%) remained negative at D14 testing, which could sensitivity for L5. It was reported that the MPT64 test has mean that those L4 isolates and the L5 isolates that were lower sensitivity for L6 strains [9], although no missense MPT64-negative at D14 need an incubation time beyond mpt64 mutations were identified in L6 genomes. So 14 days after D0, or could point to mutations in the mpt64 another mechanism may account for the lower sensitivity gene, as found in L4 (including a 63 bp deletion [24]) in in MAF L5 like for example in MAF L6 strains, in which other studies [24, 25]. The MPT64 test positivity at D0 for the expression of the mpt64 gene was lower than in L5 isolates cultured on LJ (solid) medium in our study M. tuberculosis sensu stricto [9]. More extended gene (76%) was similar to that for L6 isolates cultured in MGIT expression and regulation studies should be conducted in (liquid) medium (78.4 %) at the same D0 testing time-point order to confirm these possible causes for the lower per- in the study by Ofori-Anyinam et al., while there was a simi- formance of MPT64 tests for the L5 and L6 MAF lineages. lar increase in positivity to that found for L5 at D14 in our study (87.7%) for L6 at D15 (90.2 %)[9]. In our study, how- Importantly, in West Africa where MAF is common, ever, all L6 isolates (n=8) were positive in the MPT64 test, MPT64-negative tests should be confirmed by another iden- even at D0. This observed difference – albeit for a small tification method (such as Genotype Hain CM, IS6110 PCR, number – may be explained by the lower performance of spoligotyping, Cepheid GeneXpert, PNB/catalase) before MPT64 for L6 cultured in MGIT than for isolates cultured being classified as NTM. If available, molecular analyses on LJ, as mycobacterial growth is more rapid in MGIT (liq- [Genotype Mycobacterium CM, IS6110-PCR, spoligotyping, uid) than on LJ (solid) medium, allowing the production of GeneXpert MTB/RIF (on a diluted bacterial suspension)] a higher quantity of MPT64 protein on LJ compared to in should be prioritized for this confirmation, as these are MGIT. Indeed, a shorter incubation time posed an indepen- more rapid and sensitive (they can identify MTBc in mix- dent risk for a false-negative MPT64 test in the MGIT-based tures of MTBc/NTM isolates) than PNB/catalase. If only study [9]. PNB/catalase is available, further MPT64 testing can be done at at least day 14 after the first testing (while PNB/cat- SNP analysis of the mpt64 gene in L5 genomes confirmed alase is underway) as the result (if positive) can be obtained that L5 isolates harbour an nSNP in this gene [11] (also more rapidly than that for PNB/catalase. confirmed in all the 367 L5 genomes available in another genome collection from various countries; M. Coscolla, per- In our study, MAF isolates were more likely to be dysgonic, sonal communication), leading to a modification of the as previously reported [27, 28], particularly for L6. Gehre amino acid chain of the MPT64 produced by L5 (I43N), et al. found that L6 has non-synonymous mutations in probably impacting on the protein structure. Jiang et al. genes related to growth in culture (aceE, recA, Rv2112 and [24] found that nSNPs in the mpt64 gene rarely changed the Rv0862) that may explain its attenuated growth in culture structure and function of the protein, in contrast to a 63 bp [28]. A previous study also indicated that L5 is less likely to deletion (amino acids 66–86) that is mostly observed in L4, grow in culture compared to M. tuberculosis sensu stricto but also in some L1 isolates [24–26]. [19]. Our findings suggest that, in addition to the lower Downloaded from www.microbiologyresearch.org by IP: 197.2855.119.29 On: Mon, 05 Nov 2018 13:59:02 Sanoussi et al., Journal of Medical Microbiology ability of MAF strains to yield a positive culture, the MAF References isolates that are successfully grown are more likely to be 1. Raviglione MC, Snider DE, Kochi A. Global epidemiology of tuber- culosis. Morbidity and mortality of a worldwide epidemic. JAMA dysgonic, jeopardizing phenotypic post-culture tests. These 1995;273:220–226. findings highlight the need for simple phenotypic or geno- 2. WHO. TB Diagnostics and Laboratory Services: Information note typic diagnostic tests that can be applied directly to patient (laboratory info note GF R 11 ver 2). www.who.int/tb/dots/lab.pdf. sputa/uncultured specimens to detect all species of the 3. Sharma SK, Kohli M, Yadav RN, Chaubey J, Bhasin D et al. Evalu- MTBc, distinguish MTBc strains from NTM, and detect ating the Diagnostic Accuracy of Xpert MTB/RIF Assay in Pulmo- possible mixed MTBc/NTM infections and drug resistance nary Tuberculosis. PLoS One 2015;10:e0141011. against multiple antibiotics. 4. Zeka AN, Tasbakan S, Cavusoglu C. Evaluation of the GeneXpert MTB/RIF assay for rapid diagnosis of tuberculosis and detection One strength of our study is that it was conducted in a set- of rifampin resistance in pulmonary and extrapulmonary speci- mens. J Clin Microbiol 2011;49:4138–4141. ting where L5 is common, allowing the prospective assess- ment of the performance of the MPT64 test for identifying 5. Rieder H LVA, Kam KM, Kim SJ, Chonde TM, Tre bucq A et al. Pri- orities for tuberculosis bacteriology services in low-income coun- this lineage as an MTBc member. Testing on primary isola- tries. International Union Against Tuberculosis and Lung Disease tion reflects how the MPT64 test is used in routine practice, (The Union), 2nd ed; 2007. typically on cultures from smear-negative or GeneXpert 6. Leao S CMA, Mejia MG I, Palomino JC, Robledo RJ, Telles M A d S MTB/Rif-negative specimens, or extra-pulmonary speci- et al. Practical Hanbook for Phenotypic and Genotypic Identification of Mycobacteria; 2004. mens, and for rapid screening identification before pheno- typic DST. Our testing was essentially blinded to the strain 7. Yin X, Zheng L, Lin L, Hu Y, Zheng F et al. Commercial MPT64- based tests for rapid identification of Mycobacterium tuberculosis lineage, which was determined later and by different techni- complex: a meta-analysis. J Infect 2013;67:369–377. cians. Possible limitations of our study include the fact that 8. Hopprich R, Shephard L, Taing B, Kralj S, Smith A et al. Evalua- we only used one of the four commercially available MPT64 tion of (SD) MPT64 antigen rapid test, for fast and accurate identi- assays (Capillia, SD Bioline, MGIT TBc and TB Check), fication of Mycobacterium tuberculosis complex. Pathology 2012; 44:642–643. although a study on L6 did not find a difference between the MPT64 tests provided by different manufacturers [9]. 9. Ofori-Anyinam B, Kanuteh F, Agbla SC, Adetifa I, Okoi C et al. Impact of the Mycobaterium africanumWest Africa 2 Lineage on TB Diagnos- We expect that our findings will therefore be generalizable tics in West Africa: decreased sensitivity of rapid identification tests in to other settings, especially in West and Central Africa, the Gambia. PLoS Negl Trop Dis 2016;10:e0004801. where L5 is prevalent. 10. de Jong BC, Antonio M, Gagneux S. Mycobacterium africanum- review of an important cause of human tuberculosis in West We therefore strongly recommend that the differences in Africa. PLoS Negl Trop Dis 2010;4:e744. MTBc lineage characteristics should be considered in diag- 11. Otchere ID, Harris SR, Busso SL, Asante-Poku A, Osei-Wusu S nostics development, so tests will perform equally well in et al. The first population structure and comparative genomics ancestral lineages such as L5 and L6. Furthermore, their analysis of Mycobacterium africanum strains from Ghana reveals higher diversity of Lineage 5. Int J Mycobacteriol 2016;5:S80–S81. performance should be evaluated in different settings, 12. Gagneux S, Small PM. Global phylogeography of Mycobacterium especially where such ancestral lineages are common. tuberculosis and implications for tuberculosis product develop- ment. Lancet Infect Dis 2007;7:328–337. 13. Gehre F, Antonio M, Faïhun F, Odoun M, Uwizeye C et al. The first Funding information phylogeographic population structure and analysis of transmis- This work was supported by funds from the Directorate General for sion dynamics of M. africanum West African 1-combining molecu- Development (DGD), Belgium (FA4) to C. N. S., B. C. d. J. and D. A., and lar data from Benin, Nigeria and Sierra Leone. PLoS One 2013;8: the European Research Council-INTERRUPTB starting grant no. e77000. 311725 to B. C. d. J., L. R. and B. O. The funders had no role in study 14. Affolabi D, Anyo G, Faïhun F, Sanoussi N, Shamputa IC et al. First design, data collection and interpretation or the decision to submit the molecular epidemiological study of tuberculosis in Benin. Int J work for publication. Tuberc Lung Dis 2009;13:317–322. 15. Standard Diagnostics Inc. SD Bioline TB Ag MPT64 Rapid. 2018 Acknowledgements www.biosystemsantioquia.com.co/images/docs/reactivos/enfer- The MAF genomes analysed in this study were sequenced by the Well- medades-infecciosas/otras/08fk50-sd-tb-ag-mpt64-d.pdf. come Trust Sanger Institute. We thank Julian Parkhill (Wellcome Trust 16. European Centre for Disease Prevention and Control. Handbook Sanger Institute, United Kingdom) for contributing to the genome on TB Laboratory Diagnostic Methods for the European Union, 2016. sequencing and Conor Meehan (Institute of Tropical Medicine, Antwerp, Stockholm: ECDC; 2016. Belgium) for advising on genome analysis and providing feedback on the manuscript. 17. Bemer-Melchior P, Drugeon HB. Inactivation of Mycobacterium tuber- culosis for DNA typing analysis. J Clin Microbiol 1999;37:2350–2351. Conflicts of interest 18. Promega. Maxwell 16 DNA Purification Kits Technical Manual. The authors declare that there are no conflicts of interest 2018 https://worldwide.promega.com/resources/protocols/tech- nical-manuals/0/maxwell-16-dna-purification-kits-protocol/. Ethical statement 19. Sanoussi CN, Affolabi D, Rigouts L, Anagonou S, de Jong B. Geno- This study is part of the BeniDiT study that has been approved by the typic characterization directly applied to sputum improves the Ethics Committee of Benin, and those of the Institute of Tropical Medi- detection of Mycobacterium africanum West African 1, under-rep- cine and the University of Antwerp, Belgium. Its registration number resented in positive cultures. PLoS Negl Trop Dis 2017;11: on ClinicalTrials.gov is NCT02744469. All included patients provided e0005900. written informed consent. Moreover, all sputa were anonymized before 20. Kamerbeek J, Schouls L, Kolk A, van Agterveld M, van Soolingen laboratory analyses. D et al. Simultaneous detection and strain differentiation of Downloaded from www.microbiologyresearch.org by IP: 197.2955.119.29 On: Mon, 05 Nov 2018 13:59:02 Sanoussi et al., Journal of Medical Microbiology Mycobacterium tuberculosis for diagnosis and epidemiology. J Clin 25. Hirano K, Aono A, Takahashi M, Abe C. Mutations including Microbiol 1997;35:907–914. IS6110 insertion in the gene encoding the MPB64 protein of Capi- 21. Shabbeer A, Cowan LS, Ozcaglar C, Rastogi N, Vandenberg SL lia TB-negative Mycobacterium tuberculosis isolates. J Clin et al. TB-Lineage: an online tool for classification and analysis of Microbiol 2004;42:390–392. strains of Mycobacterium tuberculosis complex. Infect Genet Evol 26. Oettinger T, Andersen AB. Cloning and B-cell-epitope mapping of 2012;12:789–797. MPT64 from Mycobacterium tuberculosis H37Rv. Infect Immun 22. Hain Lifesciences. Genotype Mycobacterium CM. www.hain- 1994;62:2058–2064. lifescience.de/en/products/microbiology/mycobacteria/ntm/ 27. Pattyn SR, Portaels F, Spanoghe L, Magos J. Further studies on genotype-mycobacterium-cm.html. [accessed 09 May 2018]. African strains of Mycobacterium tuberculosis: comparison with M. 23. Seemaan T. Snippy: rapid haploid variant calling and core SNP bovis and M. microti. Ann Soc Belges Med Trop Parasitol Mycol phylogeny. 2018. https://github.com/tseemann/snippy. 1970;50:211–227. 24. Jiang Y, Liu H, Wang H, Dou X, Zhao X et al. Polymorphism of anti- 28. Gehre F, Otu J, Deriemer K, de Sessions PF, Hibberd ML et al. gen MPT64 in Mycobacterium tuberculosis strains. J Clin Microbiol Deciphering the growth behaviour of Mycobacterium africanum. 2013;51:1558–1562. PLoS Negl Trop Dis 2013;7:e2220. Five reasons to publish your next article with a Microbiology Society journal 1. The Microbiology Society is a not-for-profit organization. 2. We offer fast and rigorous peer review – average time to first decision is 4–6 weeks. 3. Our journals have a global readership with subscriptions held in research institutions around the world. 4. 80% of our authors rate our submission process as ‘excellent’ or ‘very good’. 5. Your article will be published on an interactive journal platform with advanced metrics. Find out more and submit your article at microbiologyresearch.org. Downloaded from www.microbiologyresearch.org by IP: 197.21505.119.29 On: Mon, 05 Nov 2018 13:59:02 View publication stats