13653156, 2021, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/tmi.13543 by INASP/HINARI - INDONESIA, Wiley Online Library on [29/11/2022].

See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Tropical Medicine and International Health doi:10.1111/tmi.13543

volume 26 no 4 pp 421–427 april 2021

Accuracy of molecular drug susceptibility testing amongst

tuberculosis patients in Karakalpakstan, Uzbekistan
Horacio Gil1, Hasmik Margaryan1, Ismailov Azamat1, Bekturdieva Ziba1, Halmuratov Bayram2,
Pirimqul Nazirov3, Diana Gomez4, Jatinder Singh5, Sayfutdinov Zayniddin6, Nargiza Parpieva7 and Jay Achar8,9
1 Medecins Sans Frontieres (MSF), Nukus, Uzbekistan
2 Republican TB No 1 Hospital Laboratory of Karakalpakstan, Nukus, Uzbekistan
3 Republican TB No 1 Hospital of Karakalpakstan, Nukus, Uzbekistan
4 MSF, Amsterdam, Holland
5 MSF, Tashkent, Uzbekistan
6 Tuberculosis Institute, Tashkent, Uzbekistan
7 National Tuberculosis Reference Laboratory, Tashkent, Uzbekistan
8 MSF, London, UK
9 Karolinska Institutet, Stockholm, Sweden

Abstract objectives In this retrospective study, we evaluated the diagnostic accuracy of molecular tests (MT)
for the detection of DR-TB, compared to the gold standard liquid-based drug susceptibility testing
(DST) in Karakalpakstan.
methods A total of 6670 specimens received in the Republican TB No 1 Hospital Laboratory of
Karakalpakstan between January and July 2017 from new and retreatment patients were analysed.
Samples were tested using Xpert MTB/RIF and line probe assays (LPA) for the detection of mutations
associated with resistance. The sensitivity and specificity of MTs were calculated relative to results
based on DST.
results The accuracy of MT for detection of rifampicin resistance was high, with sensitivity and
specificity over 98%. However, we observed reduced sensitivity of LPA for detection of resistance;
86% for isoniazid (95% CI 82–90%), 86% for fluoroquinolones (95% CI 68-96%), 70% for
capreomycin (95% CI 46–88%) and 23% for kanamycin (95% CI 13–35%).
conclusions We show that MTs are a useful tool for rapid and safe diagnosis of DR-TB; however,
clinicians should be aware of their limitations. Although detection of rifampicin resistance was highly
accurate, our data suggest that resistance mutations circulating in the Republic of Karakalpakstan for
other drugs were not detected by the methods used here. This merits further investigation.

keywords drug-resistant tuberculosis, drug susceptibility testing, Mycobacteria, line probe assay,
GeneXpert, diagnostic accuracy

Sustainable Development Goals (SDGs): 3.3

Rapid, accurate detection of drug resistance is impera-

Introduction
tive for the prompt implementation of effective treatment
Estimates from World Health Organisation (WHO) regimens to interrupt transmission of MDR-TB and
suggest that in 2018, 484 000 people developed TB extensively drug-resistant (XDR-TB) strains, defined as
that was resistant to rifampicin (RR-TB), amongst strains also resistant to any fluoroquinolone, and to any
whom 78% also harboured isoniazid resistance, thereby of the three second-line injectables (amikacin, capre-
fulfilling the criteria for multi-drug-resistant TB (MDR- omycin and kanamycin) [3]. Delays in the identification
TB) [1]. Longer treatment durations, higher treatment of drug resistance result in prolonged transmission
costs, greater toxicity, poorer response and relative through the continued prescription of ineffective or sub-
inaccessibility to diagnostics all mean that RR-TB optimal regimens. Therefore, the priority use of rapid
threatens global progress in achieving WHO’s End TB molecular tests (MTs) for the prediction of drug resis-
Strategy [2]. tance over time-consuming culture-based drug

© 2021 John Wiley & Sons Ltd 421

 13653156, 2021, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/tmi.13543 by INASP/HINARI - INDONESIA, Wiley Online Library on [29/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Tropical Medicine and International Health volume 26 no 4 pp 421–427 april 2021

H. Gil et al. Molecular TB DST performance in Uzbekistan

susceptibility testing (DST) is critical for initial patient Specimens were analysed by acid-fast bacilli smear
management. Better understanding of the molecular and microscopy and Xpert MTB/RIF. Positive samples by
genetic basis of drug resistance is hoped to contribute to either method were cultured and tested with first-line
establishing effective MT, enabling rapid identification of LPA. If rifampicin resistance was detected by either
DR-TB strains [1]. MT accuracy is affected by properties MT, second-line LPA was performed. When rifampicin
of the testing method itself [4, 5]. Additionally, the local resistance was detected, culture isolates were tested by
prevalence of mutations which are undetected or misclas- first-line DST (streptomycin, rifampicin, isoniazid,
sified may affect accuracy of specific MTs and therefore ethambutol and pyrazinamide) and second-line DST
impact their clinical utility [6]. Therefore, ongoing moni- (ofloxacin, kanamycin and capreomycin). All extrapul-
toring and evaluation of the performance of MT within monary samples were cultured and tested by LPA
programmes is key to early detection of decreasing capac- regardless of smear microscopy and Xpert MTB/RIF
ity for detecting drug-resistant strains, as well as for pro- results.
viding molecular epidemiological data regarding
resistance patterns. DR-TB is a public health concern in
Molecular tests
Eastern Europe and Central Asia. Uzbekistan is amongst
the 20 high-burden countries for MDR-TB [1]; its TB Samples were tested using Xpert MTB/RIF (Cepheid,
incidence is 70/100,000 population, with 15% of new Sunnyvale, USA) for the detection of Mycobacterium
cases and 34% of retreatment cases exhibiting rifampicin tuberculosis, and the identification of rifampicin resis-
resistance [1]. Moreover, within the seven countries with tance following manufacturers’ instructions [9]. Positive
XDR-TB cohorts of more than 100 individuals, mortality specimens were subsequently tested using LPA Genotype
was second highest in Uzbekistan (26%). MTBDRplus VER 2.0 (Hain Lifescience, Germany) for
Karakalpakstan, in northwestern Uzbekistan, has a the identification of mutations associated with rifampicin
population of around 1.7 million. Medecins sans Fron- and isoniazid resistance [4]. Samples resistant to rifampi-
tieres (MSF) has been supporting the local TB pro- cin using Xpert MTB/RIF or LPA were further analysed
gramme to strengthen TB surveillance, prevention and with Genotype MTBDRsl VER 1.0 (Hain Lifescience,
care since 2003. MSF’s model of care focuses on rapid Germany) for the detection of mutations associated with
TB diagnosis, and detection of drug resistance, to allow resistance to ethambutol, fluoroquinolones and aminogly-
the provision of effective treatment regimens. Long-term cosides/cyclic peptides. Where LPA results were invalid,
investment in laboratory services has resulted in a regio- testing was repeated on culture isolates if available. For
nal laboratory where phenotypic and genotypic diagnos- cases with no culture isolate, a new sample from the
tic tests are performed. patient was analysed.
In this retrospective study, we describe performance When performing LPA tests, DNA from the specimen
characteristics of Xpert MTB/RIF and LPAs for the detec- or culture was extracted with GenoLysis (HAIN Life-
tion of MDR-TB and XDR-TB, when compared to the science), and amplified and hybridised following manu-
gold standard liquid-based DST in Karakalpakstan. facturers’ instructions [10, 11].

Phenotypic methods
Methods
Isolates were analysed using DST from MGIT 960 liquid
Study samples
culture according to established procedures [12]. The fol-
All diagnostic specimens received between January and lowing antibiotics and concentrations recommended by
July 2017 from new and retreatment patients managed in WHO [13] were used as part of DST: rifampicin (1 µg/
the Karakalpak TB programme were included in this ml), isoniazid (0.1 µg/ml), ethambutol (5 µg/ml), ofloxa-
study. Clinical and laboratory assessment for TB is direc- cin (2 µg/ml), kanamycin (2.5 µg/ml) and capreomycin
ted by local guidelines based on global recommendations (2.5 µg/ml).
[7]. All recommended 2nd line drugs except gatifloxacin
were available and treatment with the short regimen for
Statistical methods
MDR-TB was accessible to eligible patients. Specimens
were tested according to locally approved algorithms for The accuracy, sensitivity, specificity, positive (PPV) and
the diagnosis of TB from pulmonary and extrapulmonary negative (NPV) predictive values for each MT were calcu-
samples [8]. lated along with 95% confidence intervals (CI). Results

422 © 2021 John Wiley & Sons Ltd

 13653156, 2021, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/tmi.13543 by INASP/HINARI - INDONESIA, Wiley Online Library on [29/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Tropical Medicine and International Health volume 26 no 4 pp 421–427 april 2021

H. Gil et al. Molecular TB DST performance in Uzbekistan

from phenotypic MGIT 960 liquid culture were used as a Table 2 Frequency of resistant hybridisation patterns detected
gold- standard reference. Heteroresistant samples, con- with LPA
taining sensitive and resistant M. tuberculosis sub-popula-
Associated No. samples
tions identified by LPA were considered resistant for the Drug Pattern mutation (%)
purposes of this analysis. Concordance between pheno-
typic and MT was described with kappa values. The sta- Rifampicin rpoB MUT3 S531L 254 (93)
tistical analysis was performed using R version 3.6.0 [14]. rpoB WT8 S531Q, 11 (4.0)
absent S531W or
L533P
Ethics rpoB WT3/ D516Y or 4 (1.5)
WT4 absent deletion 515
This research fulfilled the exemption criteria set by the rpoB MUT2A H526Y 2 (0.7)
Medecins Sans Frontiieres Ethics Review Board for a pos- rpoB MUT2B H526D 1 (0.4)
teriori analyses of routinely collected clinical data and rpoB MUT1 D516V 1 (0.4)
thus did not require MSF ethical review. Isoniazid katG MUT1 S315T 257 (100)
inhA MUT1 C-15T 15 (48)
inhA MUT3A T-8C 14 (45)
inhA MUT3B T-8A 1 (3.2)
Results InhA MUT1 C-15T & T- 1 (3.2)
&MUT3A 8C*
Molecular detection of rifampicin resistance
Ethambutol embB MUT1B M306V 106 (74)
Between January and July 2017, 6670 samples were embB MUT1A M306I 33 (23)
received in the laboratory, of which 985 (15%) were pos- embB WT M306I 5 (3.5)
absent
itive using Xpert MTB/RIF. 395 of these (40%) were
Ofloxacin gyrA MUT3C D94G 10 (35)
detected as rifampicin-resistant. Rifampicin-resistant sam- gyrA MUT1 A90V 7 (24)
ples were mostly associated with mutations detected gyrA MUT3B D94N 5 (17)
using probe E (96%), while mutations detected using gyrA MUT3A D94A 4 (14)
probes A (0.3%), B (2.3%) and D (1.3%) were infre- gyrA MUT2 S91P & 1 (3.4)
quently detected. One specimen (0.3%) was detected & MUT3C D94G*
gyrA MUT3B D94N & 1 (3.4)
with both A and B probes.
& MUT3C D94G*
LPA rpoB results were available from 694 samples, of gyrA MUT3A D94A & 1 (3.4)
which 249 (36%) exhibited resistance and 24 samples & MUT3C D94G*
displayed heteroresistance (Table 1). The most frequent Kanamycin & rrs MUT1 A1401G 12 (80)
hybridisation pattern was rpoB MUT3 (93%), associated Capreomycin rrs WT absent C1402T 3 (20)
with the mutation S531L, while all other patterns showed
*Mutations considering mixed infections in the sample.
a frequency lower than 4.0% (Table 2). A high degree of
accuracy (98%; 95% CI 98–99%) was found between
the Xpert MTB/RIF and LPA results. rifampicin using DST. Both MTs showed a high accuracy
(over 99%) against phenotypic DST (Table 3). Sensitivity
and specificity for both MT methods were also high; for
Accuracy of MT for detection of rifampicin resistance
Xpert MTB/RIF, sensitivity was 98% (95% CI 95–99%)
A total of 609 and 584 samples, tested with Xpert MTB/ and specificity 99% (95% CI 97–100%); for LPA, sensi-
RIF and LPA, respectively, had a phenotypic result for tivity was 99% (95% CI 97–100%) and specificity 99%
(95% CI 97–100%; Table 3).

Table 1 Comparison between Xpert MTB/RIF and LPA rpoB

Reduced sensitivity of LPA for detection of isoniazid
methods
resistance
LPA rpoB results
A total of 589 specimens tested with first-line LPA and
Xpert result Heteroresistant Resistant Sensitive Total isoniazid phenotypic DST had valid results. LPA detected
241 resistant specimens, and 26 of these had a heterore-
Resistant 16 247 5 268
sistant pattern, with 88% (236/267) having mutations
Sensitive 8 2 416 426
Total 24 249 421 694
only in katG, 3.8% (10/267) only in inhA, and 7.9%
(21/267) having mutations in both genes. All of the

© 2021 John Wiley & Sons Ltd 423

 13653156, 2021, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/tmi.13543 by INASP/HINARI - INDONESIA, Wiley Online Library on [29/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Tropical Medicine and International Health volume 26 no 4 pp 421–427 april 2021

H. Gil et al. Molecular TB DST performance in Uzbekistan

katG-resistant specimens had the katG MUT1 pattern

99 (98–100)
99 (98–100)
99 (98–99)

(73–85)
(93–99)
(56–72)
(92–99)
Accuracy
(mutation S315T), while the most frequent mutations in
inhA were C-15T and T-8C (Table 2).

79
96
64
95
LPA had high specificity for detecting isoniazid resis-
tance (98%; 95% CI 96–99%), although sensitivity was
0.97 (0.95–0.99)

0.97 (0.96–0.99)
0.84 (0.80–0.88)

(0.29–0.60)
(0.73–0.95)
(0.07–0.41)
(0.61–0.94)
reduced to 86% (95% CI 82–90%). False-negative isoni-
azid resistance was seen in 7.0% (41/589) of samples
Kappa‡

0.45 (Table 3), resulting in a reduction in NPV to 87% (95%

0.84
0.24
0.77
CI 83–91%).
99 (97–100)

99 (98–100)
87 (83–91)

(56–87)
(94–99)
(51–68)
(91–98)
NPV‡

Lack of specificity for detection of ethambutol resistance

74
98
60
95

A total of 178 specimens had valid results for ethambutol

resistance using both LPA and phenotypic DST. Sensitiv-
98 (96–100)

(78–100)
98 (95–99)
98 (95–99)

(73–87)
(68–96)

(66–99)

ity and specificity for detection of ethambutol resistance

using LPA were 93 (95% CI 87–97) and 47% (95% CI
PPV‡

81
86
100
93

33–61), respectively (Table 3). The most frequently

detected mutation was M306V (Table 2). The mutation
†Heteroresistant samples are considered as resistant, for the purposes of carrying out calculations of parameters.
99 (97–100)

99 (97–100)

(95–100)
(95–100)
98 (96–99)

(33–61)
(94–99)

M306I was found in 56% (15/27) of specimens found to

be phenotypically sensitive, but resistant to ethambutol
Spec‡

47
98
100
99

on LPA.
Parameters†

99 (97–100)
98 (95–99)

86 (82–90)

(87–97)
(68–96)
(13–35)
(46–88)

Reduced sensitivity for detection of fluoroquinolone

Sen‡

resistance
93
86
23
70

A total of 201 specimens had valid results for fluoro-

MT

quinolone resistance using LPA and phenotypic DST; 29

Sen

371

351
281

25
168
76
121

of these were detected as resistant using LPA (Table 3).

DST sensitive

The most frequently associated mutation was D94G,

MT
Res

1
1

27
2
0
1

‡For each parameter, percentages and 95% confidence interval are given.

found in 35% (10/29) of resistant specimens (Table 3).

*Number of samples with information for both MT and DST available.
Table 3 Comparison between genotypic and phenotypic DST methods

Although LPA showed a high specificity of 98% (95%

MT
HT

CI 94–99%) for the detection of fluoroquinolone resis-

0

4
5

1
2
0
0

tance (Table 3), sensitivity was reduced to 86% (95%

CI 68–96%).
MT
Sen

2
41

9
4
51
6
DST resistant

MT
Res

229

211
240

116
22
15
14

Lack of sensitivity for detection of resistance to

aminoglycosides/cyclic peptides
MT
HT

A total of 142 specimens had valid results for analysis of

0

15
21

0
3
0
0

resistance to aminoglycosides/cyclic peptides using LPA

Samples*

Number

and DST, 15 of these showed a resistant pattern using

LPA (Table 3). 80% (12/15) of the resistant specimens
609

584
589

178
201
142
142

had the rrs MUT1 pattern (A1401G mutation). Although

specificity using LPA was high for detection of resistance
Xpert MTB/

LPA (embB)
LPA (rpoB)

LPA (gyrA)
LPA (katG
& inhA)

against these drugs, there was a lack of sensitivity. In the

LPA (rrs)
LPA (rrs)

case of capreomycin, sensitivity was 70% (95% CI 46–

Methods compared

RIF
MT

88%); sensitivity was even lower for detection of kana-

mycin resistance (23%; 95% CI 13–35). We observed
Capreomycin
Ethambutol

Kanamycin
Rifampicin

Rifampicin

that 36% (51/142) of specimens were phenotypically

Ofloxacin
Isoniazid

resistant to kanamycin, but observed to be sensitive on

DST

LPA testing; this reduced NPV to 60% (95% CI 51–

424 © 2021 John Wiley & Sons Ltd

 13653156, 2021, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/tmi.13543 by INASP/HINARI - INDONESIA, Wiley Online Library on [29/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Tropical Medicine and International Health volume 26 no 4 pp 421–427 april 2021

H. Gil et al. Molecular TB DST performance in Uzbekistan

68%), and the kappa value to 0.24 (95% CI 0.07–0.41), progression to MDR-TB [25]. These results should
for detection of kanamycin resistance (Table 3). 8.3% (1/ encourage clinicians to continue using culture and DST
12) of specimens with the rrs MUT1(A1401G) mutation to confirm drug resistance patterns, thereby informing
were phenotypically sensitive to capreomycin (Tables 2 selection of proper treatment. This may ultimately
and 3). enable development of future generations of LPA tools
with the ability to detect a larger number of potential
mutations for isoniazid resistance.
Discussion
LPA prediction of ethambutol resistance was not reli-
Determining the drug resistance pattern of the mycobac- able. Poor LPA ethambutol results have also been
teria infecting a TB patient is critical for making deci- observed in other studies [26]. The discrepancies seen
sions on appropriate treatment regimens. In addition, here may be due in part to the mutation M306I, which
confidence in DST results for patients with MDR-TB is induces a low level of resistance [27], potentially explain-
essential when considering treatment with shorter regi- ing the reduced specificity in our DST assay. Indeed, the
mens [15]. The rapid diagnosis made possible by MTs minimum inhibitory concentration between ethambutol-
can be essential for identifying MDR and XDR-TB sensitive and resistant isolates is narrow [28], making
strains. However, since mechanisms of drug resistance in DST complicated to perform. Ethambutol is not a prior-
M. tuberculosis are complex and not always fully defined, ity drug for the selection of treatment in MDR patients,
and resistance-associated mutations vary by geographic and it is not considered currently in the updated LPA ver-
region, the accuracy of commercial MTs should always sion [11].
be evaluated in specific settings. Use of LPA for the detection of second-line drugs
The detection of rifampicin resistance is key to the (MTBDRsl version 1) showed high specificity, but low
classification of MDR-TB. In our study, both GeneXpert sensitivity in our study – dramatically low in the case
MTB-RIF and LPA (GenoType MTBDRplus) showed of kanamycin. This is due likely to mutations not
high accuracy for the detection of rifampicin resistance, detected by LPA. Indeed, the updated LPA (MTBDRsl
with sensitivity and specificity falling within the range version 2) has included the targets gyrB for the fluoro-
seen in other studies [5]. We observed a small number of quinolones and eis for kanamycin, improving the sensi-
discrepancies (<1.5%), which could be explained by the tivity of MT [11]. Preliminary data following
presence of heteroresistant infections [16, 17], silent introduction of an updated version of Genotype
mutations [6], mutations which induce a lower level of MTBDRsl show that the eis mutation is detected in
resistance [18], or paucibacillary samples [19]. Unfortu- 22% of the Karakalpakstan isolates tested (data not
nately, the lack of DNA sequencing capacity in our set- shown), explaining the low sensitivity observed in our
ting prevented detailed molecular evaluation of some of analysis. The reduced sensitivity of LPA for capre-
these discrepancies. omycin shown in our study could be related to muta-
We observed reduced sensitivity using LPA for the tions in tlyA, which are still not included in the LPA
detection of isoniazid resistance. Newly described muta- panel [29]. Similarly to our findings, 7% of isolates
tions in katG, fabG1, as well as unknown molecular with the rrs mutation A1401G are phenotypically sensi-
mechanisms inferred for around 2% of isoniazid-resis- tive to capreomycin, indicating that this mutation is a
tant isolates are not detectable with currently available moderately specific resistance marker for this drug, as
LPA methods [20–23]. Such poorly understood mecha- has been previously described [29].
nisms could explain the 7% of phenotypically isoni- In this study, the lack of access to sequencing to con-
azid-resistant strains which were found to be sensitive firm the presence of mutations not included in LPA pan-
using MTs in our study. Moreover, the isoniazid muta- els was a limitation. National regulations prohibit export
tions not detected by LPA were found to be common of biological specimens and whole-genome sequencing is
within isoniazid mono-resistant (Mono-INHR) isolates not available in the country. We encourage further
[20]. This complication in relation to detection of iso- molecular analysis of the mutations carried by isolates in
niazid mutations could increase the frequency of Karakalpakstan.
Mono-INHR isolates, as has been described in some The molecular information presented here will be of
Asian regions [24]. Indeed, 12% of the isolates from particular value in the rapid clinical detection of MDR
Karakalpakstan are Mono-INHR (data not shown). In and XDR M. tuberculosis strains in the Republic of
addition, inappropriate treatment of Mono-INHR Karakalpakstan. MT used in our setting provides a rapid
patients can lead to poor outcomes and risk of and reliable tool for optimising therapeutic management

© 2021 John Wiley & Sons Ltd 425

 13653156, 2021, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/tmi.13543 by INASP/HINARI - INDONESIA, Wiley Online Library on [29/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Tropical Medicine and International Health volume 26 no 4 pp 421–427 april 2021

H. Gil et al. Molecular TB DST performance in Uzbekistan

of patients. However, MTs should be applied together 12. Mycobacteriology Laboratory Manual. Global Laboratory
with DST, since some resistant strains, which are not Initiative: Geneva, Switzerland, 2014.
detectable using these methods, are circulating in 13. Companion handbook to the WHO guidelines for the pro-
Karakalpakstan. grammatic management of drug-resistant tuberculosis.
World Health Organisation: Geneva, Switzerland, 2014.
14. R: A Language and Environment for Statistical Computing.
Acknowledgements R Foundation for Statistical Computing: Vienna, Austria,
2019.
We thank all the laboratory technicians that participate
15. Van Deun A, Maug AK, Salim MA et al. Short, highly effec-
in the daily analysis of samples for diagnosis and detec- tive, and inexpensive standardized treatment of multidrug-
tion of drug resistance studies at the tuberculosis labora- resistant tuberculosis. Am J Respir Crit Care Med. 2010:
tory in Nukus. Also, we acknowledge the support of all 182(5): 684–692.
the Ministry of Health and MSF staff that work with the 16. Blakemore R, Story E, Helb D et al. Evaluation of the ana-
comprehensive care programme in the Republic of lytical performance of the Xpert MTB/RIF assay. J Clin
Karakalpakstan. Emma Veitch, freelance medical editor, Microbiol. 2010: 48: 2495–2501.
MSF, provided editorial assistance. 17. Zetola NM, Shin SS, Tumedi KA et al. Mixed Mycobac-
terium tuberculosis complex infections and false-negative
results for rifampin resistance by GeneXpert MTB/RIF are
associated with poor clinical outcomes. J Clin Microbiol
References
2014: 52: 2422–2429.
1. Global Tuberculosis Report 2019. World Health Organisa- 18. Rigouts L, Gumusboga M, de Rijk WB et al. Rifampin resis-
tion: Geneva, Switzerland 2019. tance missed in automated liquid culture system for
2. The End TB Strategy. World Health Organisation: Switzer- Mycobacterium tuberculosis isolates with specific rpoB
land, 2014. mutations. J Clin Microbiol 2013: 51: 2641–2645.
3. Technical Report on Critical Concentrations for Drug Sus- 19. Williamson DA, Basu I, Bower J et al. An evaluation of the
ceptibility Testing of Medicines used in the treatment of Xpert MTB/RIF assay and detection of false-positive rifam-
Drug-Resistant Tuberculosis. World Health Organisation: picin resistance in Mycobacterium tuberculosis. Diagn
Geneva, Switzerland, 2018. Microbiol Infect Dis. 2012: 74: 207–209.
4. Crudu V, Stratan E, Romancenco E et al. First evaluation of 20. Torres JN, Paul LV, Rodwell TC et al. Novel katG muta-
an improved assay for molecular genetic detection of tuber- tions causing isoniazid resistance in clinical M. tuberculosis
culosis as well as rifampin and isoniazid resistances. J Clin isolates. Emerg Microbes Infect 2015:4:e42.
Microbiol. 2012: 50: 1264–1269. 21. Wei CJ, Lei B, Musser JM, Tu SC. Isoniazid activation
5. Steingart KR, Schiller I, Horne DJ et al. Xpert(R) MTB/RIF defects in recombinant Mycobacterium tuberculosis cata-
assay for pulmonary tuberculosis and rifampicin resistance lase-peroxidase (KatG) mutants evident in InhA inhibitor
in adults. Cochrane Database Syst Rev. 2014(1): production. Antimicrob Agents Chemother 2003: 47: 670–
CD009593. 675.
6. Koser CU, Feuerriegel S, Summers DK et al. Importance of 22. Siqueira HR, Freitas FA, Oliveira DN et al. Isoniazid-resis-
the genetic diversity within the Mycobacterium tuberculosis tant Mycobacterium tuberculosis strains arising from muta-
complex for the development of novel antibiotics and diag- tions in two different regions of the katG gene. J Bras
nostic tests of drug resistance. Antimicrob Agents Che- Pneumol 2009: 35: 773–779.
mother. 2012: 56: 6080–6087. 23. Ando H, Miyoshi-Akiyama T, Watanabe S, Kirikae T. A
7. WHO Treatment Guidelines for Drug-resistant Tuberculo- silent mutation in mabA confers isoniazid resistance on
sis. World Health Organisation: Geneva, Switzerland, Mycobacterium tuberculosis. Mol Microbiol 2014: 91: 538–
2016. 547.
8. Clinical guidelines on drug sensitive and drug resistant 24. Varahram M, Nasiri MJ, Farnia P et al. A retrospective
tuberculosis. Ministry of Health of Republic of analysis of isoniazid-monoresistant tuberculosis: among Ira-
Karakakalpakstan and Medecins Sans Frontieres: nian pulmonary tuberculosis patients. Open Microbiol J
Karakalpakstan, Uzbekistan; 2017. 2013: 8: 1–5.
9. Helb D, Jones M, Story E et al. Rapid detection of 25. Jacobson KR, Theron D, Victor TC et al. Treatment out-
Mycobacterium tuberculosis and rifampin resistance by use comes of isoniazid-resistant tuberculosis patients, Western
of on-demand, near-patient technology. J Clin Microbiol. Cape Province. South Africa. Clin Infect Dis. 2011: 53:
2010: 48: 229–237. 369–372.
10. GenoType MTBDRsl VER 1.0 instructions for use. HAIN 26. Cheng S, Cui Z, Li Y, Hu Z. Diagnostic accuracy of a
LifeScience: Nehren, Germany, 2012. molecular drug susceptibility testing method for the antitu-
11. GenoType MTBDRplus VER 2.0 instructions for use. HAIN berculosis drug ethambutol: a systematic review and meta-
LifeScience: Nehren, Germany, 2012. analysis. J Clin Microbiol. 2014: 52: 2913–2924.

426 © 2021 John Wiley & Sons Ltd

 13653156, 2021, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/tmi.13543 by INASP/HINARI - INDONESIA, Wiley Online Library on [29/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Tropical Medicine and International Health volume 26 no 4 pp 421–427 april 2021

H. Gil et al. Molecular TB DST performance in Uzbekistan

27. Andres S, Groschel MI, Hillemann D et al. A Diagnostic mycobacterium tuberculosis by agar proportion and radio-
Algorithm To Investigate Pyrazinamide and Ethambutol metric methods. J Clin Microbiol. 2002: 40: 3976–3979.
Resistance in Rifampin-Resistant Mycobacterium tuberculo- 29. Georghiou SB, Magana M, Garfein RS et al. Evaluation of
sis Isolates in a Low-Incidence Setting. Antimicrob Agents genetic mutations associated with Mycobacterium tuberculo-
Chemother. 2019: 63. sis resistance to amikacin, kanamycin and capreomycin: a
28. Madison B, Robinson-Dunn B, George I et al. Multicenter systematic review. PLoS One 2012: 7: e33275.
evaluation of ethambutol susceptibility testing of

Corresponding Author Jay Achar, Department of Global Public Health, Karolinska Institutet, Stockholm 17177, Sweden. E-mail:
jay.achar@doctors.org.uk

© 2021 John Wiley & Sons Ltd 427