Comparison of rapid tests for detection of rifampicin-resistant Mycobacterium tuberculosis in Kampala, Uganda.

Abstract

Drug resistant tuberculosis (TB) is a growing concern worldwide. Rapid detection of resistance expedites appropriate intervention to control the disease. Several technologies have recently been reported to detect rifampicin resistant Mycobacterium tuberculosis directly in sputum samples. These include phenotypic culture based methods, tests for gene mutations and tests based on bacteriophage replication. The aim of the present study was to assess the feasibility of implementing technology for rapid detection of rifampicin resistance in a high disease burden setting in Africa.
Sputum specimens from re-treatment TB patients presenting to the Mulago Hospital National TB Treatment Centre in Kampala, Uganda, were examined by conventional methods and simultaneously used in one of the four direct susceptibility tests, namely direct BACTEC 460, Etest, "in-house" phage test, and INNO- Rif.TB. The reference method was the BACTEC 460 indirect culture drug susceptibility testing. Test performance, cost and turn around times were assessed.
In comparison with indirect BACTEC 460, the respective sensitivities and specificities for detecting rifampicin resistance were 100% and 100% for direct BACTEC and the Etest, 94% and 95% for the phage test, and 87% and 87% for the Inno-LiPA assay. Turn around times ranged from an average of 3 days for the INNO-LiPA and phage tests, 8 days for the direct BACTEC 460 and 20 days for the Etest. All methods were faster than the indirect BACTEC 460 which had a mean turn around time of 24 days. The cost per test, including labour ranged from $18.60 to $41.92 (USD).
All four rapid technologies were shown capable of detecting rifampicin resistance directly from sputum. The LiPA proved rapid, but was the most expensive. It was noted, however, that the LiPA test allows sterilization of samples prior to testing thereby reducing the risk of accidental laboratory transmission. In contrast the Etest was low cost, but slow and would be of limited assistance when treating patients. The phage test was the least reproducible test studied with failure rate of 27%. The test preferred by the laboratory personnel, direct BACTEC 460, requires further study to determine its accuracy in real-time treatment decisions in Uganda.

Full text

Page 1

BioMed Central
Open Access
Page 1 of 9
(page number not for citation purposes)
BMC Infectious Diseases
Research article
Comparison of rapid tests for detection of rifampicin-resistant
Mycobacterium tuberculosis in Kampala, Uganda
Sam Ogwang1,6, Benon B Asiimwe2, Hamidou Traore3, Francis Mumbowa1,8,
Alphonse Okwera4, Kathleen D Eisenach5, Susan Kayes1,6, Edward C Jones-
López7,8, Ruth McNerney3, William Worodria8, Irene Ayakaka8,
Roy D Mugerwa4,6, Peter G Smith3, Jerrold Ellner7,8 and Moses L Joloba*2,6
Address: 1Joint Clinical Research Centre, Kampala, Uganda, 2Department of Medical Microbiology, Makerere University College of Health
Sciences, Kampala, Uganda, 3Department of Infectious and Tropical Diseases, London School of Hygiene & Tropical Medicine, London, UK,
4Department of Medicine, Makerere University Medical School, Kampala, Uganda, 5University of Arkansas for Medical Sciences, Little Rock, AR,
USA, 6Uganda-Case Western Reserve University Research Collaboration, Kampala, Uganda, 7Department of Medicine, New Jersey Medical School
– University of Medicine and Dentistry of New Jersey (UMDNJ), Newark, NJ, USA and 8Makerere University-UMDNJ Research Collaboration,
Kampala, Uganda
Email: Sam Ogwang - ogwangsam@yahoo.com; Benon B Asiimwe - basiimwe@med.mak.ac.ug; Hamidou Traore - hmdt@hotmail.com;
Francis Mumbowa - francmumbowa@yahoo.co.uk; Alphonse Okwera - a_okwera@mucwru.or.ug;
Kathleen D Eisenach - EisenachKathleenD@uams.edu; Susan Kayes - susan.kayes@gmail.com; Edward C Jones-López - jonesec@umdnj.edu;
Ruth McNerney - Ruth.Mcnerney@lshtm.ac.uk; William Worodria - worodria@yahoo.com; Irene Ayakaka - ayakaka@gmail.com;
Roy D Mugerwa - profrdm@imul.com; Peter G Smith - Peter.Smith@lshtm.ac.uk; Jerrold Ellner - jerroldellner@yahoo.com;
Moses L Joloba* - moses.joloba@case.edu
* Corresponding author
Abstract
Background: Drug resistant tuberculosis (TB) is a growing concern worldwide. Rapid detection
of resistance expedites appropriate intervention to control the disease. Several technologies have
recently been reported to detect rifampicin resistant Mycobacterium tuberculosis directly in sputum
samples. These include phenotypic culture based methods, tests for gene mutations and tests based
on bacteriophage replication. The aim of the present study was to assess the feasibility of
implementing technology for rapid detection of rifampicin resistance in a high disease burden
setting in Africa.
Methods: Sputum specimens from re-treatment TB patients presenting to the Mulago Hospital
National TB Treatment Centre in Kampala, Uganda, were examined by conventional methods and
simultaneously used in one of the four direct susceptibility tests, namely direct BACTEC 460, Etest,
"in-house" phage test, and INNO- Rif.TB. The reference method was the BACTEC 460 indirect
culture drug susceptibility testing. Test performance, cost and turn around times were assessed.
Results: In comparison with indirect BACTEC 460, the respective sensitivities and specificities for
detecting rifampicin resistance were 100% and 100% for direct BACTEC and the Etest, 94% and
95% for the phage test, and 87% and 87% for the Inno-LiPA assay. Turn around times ranged from
an average of 3 days for the INNO-LiPA and phage tests, 8 days for the direct BACTEC 460 and
20 days for the Etest. All methods were faster than the indirect BACTEC 460 which had a mean
turn around time of 24 days. The cost per test, including labour ranged from $18.60 to $41.92
(USD).
Published: 26 August 2009
BMC Infectious Diseases 2009, 9:139 doi:10.1186/1471-2334-9-139
Received: 6 February 2009
Accepted: 26 August 2009
This article is available from: http://www.biomedcentral.com/1471-2334/9/139
© 2009 Ogwang et al; licensee BioMed Central Ltd.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0),
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Page 2

BMC Infectious Diseases 2009, 9:139http://www.biomedcentral.com/1471-2334/9/139
Page 2 of 9
(page number not for citation purposes)
Conclusion: All four rapid technologies were shown capable of detecting rifampicin resistance
directly from sputum. The LiPA proved rapid, but was the most expensive. It was noted, however,
that the LiPA test allows sterilization of samples prior to testing thereby reducing the risk of
accidental laboratory transmission. In contrast the Etest was low cost, but slow and would be of
limited assistance when treating patients. The phage test was the least reproducible test studied
with failure rate of 27%. The test preferred by the laboratory personnel, direct BACTEC 460,
requires further study to determine its accuracy in real-time treatment decisions in Uganda.
Background
Developing countries account for 95% of active tubercu-
losis (TB) cases and deaths due to TB worldwide [1,2]. In
developing countries, many national TB control programs
have low case detection rates and once a case is detected,
cure may also be difficult because of poor case holding,
high default rates and insufficient control of drug pre-
scription [3]. In such a setting, an added threat lies with
the potential for development and spread of multi drug-
resistant tuberculosis (MDR-TB), defined as resistance to
at least isoniazid and rifampicin, as well as extensively
drug-resistant tuberculosis (XDR -TB) caused by MDR
strains resistant to any fluoroquinolone and at least one of
three injectable second-line drugs (i.e., amikacin, kan-
amycin, or capreomycin) [4,5]. Although the DOTS strat-
egy is effective in the management of drug susceptible TB,
outbreaks of MDR-TB and XDR-TB strains that are untreat-
able by the DOTS program threaten to complicate TB con-
trol [6-8]. In places where drug susceptibility testing is not
routinely performed patients with MDR-TB are likely to
become re-treatment cases and will be managed using a
standardized regimen, such as 2 months of streptomycin
[S], isoniazid [H], rifampicin [R], pyrazinamide [Z] and
ethambutol [E] followed by 1 month of H, R, Z, E and 5
months of H, R, E. As the majority of re-treatment cases
have susceptible TB and will respond to the simpler stand-
ard regimen (2HRZE/6HE) this practice is expensive for
the program and also exposes the majority of re-treatment
cases to unnecessary drug toxicity. For those with MDR-
TB, the re-treatment regimen has a risk of adding a single
drug to a failing regimen, which leads to worsening resist-
ance profile and outcomes [9]. Previous studies have indi-
cated that delayed initiation of appropriate treatment for
MDR-TB is associated with increased mortality and spread
of the MDR strains [10-13]. Typically in developing coun-
tries, if susceptibility testing is done, results are not avail-
able for 3–4 months. To improve patient survival, reduce
unnecessary costs and toxicity during re-treatment, and
minimize transmission of MDR-TB strains, a rapid
method for detection of MDR-TB is a high priority for
developing countries.
Uganda has a high burden of tuberculosis with an esti-
mated prevalence of 561 cases per 100,000 of the popula-
tion [2]. The prevalence of MDR-TB in new cases of TB has
been reported to be low at less than 2% [2,14]; however,
we have recently reported that 12.7% of re-treatment cases
attending the National Tuberculosis and Leprosy Program
(NTLP) clinic in Kampala had MDR-TB [9]. Although
drug susceptibility testing is not routinely undertaken in
Uganda it would be advantageous to know the drug sus-
ceptibility status of re-treatment patients to facilitate
appropriate patient management. As part of a larger
project to investigate strategies for controlling MDR-TB in
re-treatment cases in Kampala we undertook an assess-
ment of new technologies for rapid detection of MDR-TB.
In this population resistance to rifampicin may be consid-
ered a good predictor of MDR-TB where less than 0.05%
of re-treatment patients were found to have mono resist-
ance to the drug [9]. There are several competing technol-
ogies that have been proposed for rapid detection of
resistance directly from the sputum of tuberculosis
patients. We have investigated four different technologies
to assess their suitability for implementation in a TB refer-
ence diagnostic laboratory in Africa for the rapid assess-
ment of resistance to rifampicin in smear positive re-
treatment patients. The purpose of the study was to com-
pare the feasibility of the technologies prior to selecting
one for implementation in a larger study on the impact of
new interventions to control MDR-TB in Kampala. The
four technologies investigated were liquid culture
(BACTEC 460, Becton Dickinson), an in-house bacteri-
ophage based test, a Line Probe Assay (INNO-LiPA Rif.TB,
Innogenetics) and drug impregnated strips on solid cul-
ture (E-Test, AB Biodisk). Performance characteristics and
costs were compared using the indirect BACTEC 460 drug
susceptibility test on culture as the reference standard.
Methods
Study setting
The culture-based methods were carried out in the TB lab-
oratory of the Joint Clinical Research Centre (JCRC) in
Kampala, while molecular assays were performed in the
Department of Medical Microbiology at Makerere Univer-
sity. The JCRC TB laboratory supports clinical studies as
well as routine patient care. On average, each month the
JCRC TB laboratory receives 600 sputum specimens, for
either diagnosis or follow-up. The Department of Medical
Microbiology performs both applied and basic TB
research and works closely with the JCRC laboratory.

Page 3

BMC Infectious Diseases 2009, 9:139http://www.biomedcentral.com/1471-2334/9/139
Page 3 of 9
(page number not for citation purposes)
Study population and Ethics
The study population comprised adult (≥ 18 years) TB
cases who presented to the Uganda NTLP clinic at Mulago
Hospital in Kampala, Uganda during the period March
2004 to December 2005. Patients with smear positive dis-
ease who had been previously treated for TB and received
at least two months of therapy were invited to join the
study. All of the patients were admitted to the hospital
ward to initiate therapy, which included daily injections
of streptomycin. Subjects were classified according to rea-
son for re-treatment, i.e., treatment failure, relapse, or
defaulted treatment based on a previous treatment card or
chart review. Sputum specimens were collected for culture
and sensitivity testing, sampling was consecutive and not
randomized. This being a feasibility study results from the
rapid methods under evaluation were not used for patient
management and only drug susceptibility results from the
indirect testing were reported to the clinic.
The study protocol was approved by the AIDS research
sub-committee of the Uganda National Council of Sci-
ence and Technology together with the Institutional
Review Boards at the University of Medicine and Dentistry
of New Jersey and London School of Hygiene & Tropical
Medicine and Hygiene.
Study design
Each of the rapid tests was evaluated using sputum speci-
mens collected from re-treatment patients with smear
positive tuberculosis. In addition to testing by one of the
rapid methods specimens were assessed by smear micros-
copy and cultured in the BACTEC 460. Indirect drug sus-
ceptibility testing was performed on isolates obtained.
Technicians performing the rapid tests were kept blind to
the results of the indirect drug susceptibility tests until the
final comparative analysis. For the purposes of this feasi-
bility study, the rapid tests were evaluated consecutively
rather than concurrently, with the test under scrutiny
being run alongside the existing routine activities of the
laboratory. A maximum of five specimens were tested
each week. As this study was to gauge the feasibility of
integrating the various technologies into a routine labora-
tory, each test was assessed for a period of at least three
months.
Laboratory procedures
Early morning spot sputum specimens were collected in
sterile disposable 50 ml polypropylene centrifuge tubes
and taken to the laboratory. Specimens delivered to the
lab after 2 pm, were refrigerated at 2 – 8°C and processed
the following morning. Specimens (2.5 to 10 ml) were
decontaminated using a final concentration of 1% NaOH
and concentrated at 3000 × g for 15 minutes. The sedi-
ment was reconstituted to 2.5 ml with phosphate buffer
pH 6.8, and used to prepare smears and cultures on Mid-
dlebrook 7H10 agar and in BACTEC 12B. Smear micros-
copy was performed using auramine and graded
according to ATS/CDC guidelines [15]. Only smear-posi-
tive specimens were selected for subsequent testing by one
of the rapid methods.
Growth from 12B vials was used for indirect drug suscep-
tibility testing in the BACTEC 460, according to the man-
ufactures' instruction [16]. Quality control for the indirect
BACTEC drug susceptibility test included weekly monitor-
ing with H37Rv, a pan-susceptible strain, according to the
guidelines of the Clinical and Laboratory Standards Insti-
tute (CLSI; formerly the National Committee for Clinical
laboratory Standards). In addition, new lots of BACTEC
12 B medium and new drug lots were controlled with drug
resistant strains. External quality assessment included par-
ticipation in the quality assurance program for drug sus-
ceptibility testing administered through the WHO/
IUATLD Supranational Laboratory Network. Confirma-
tion of M. tuberculosis complex was determined by either
the BACTEC NAP test (Becton Dickinson) or PCR for IS
6110 [17]. Following sampling for microscopy and cul-
ture, part of the remaining inoculum was used for the
rapid direct test under investigation. The direct tests were
set up at the same time as the routine cultures, with the
exception of the LiPA where samples were batched.
To set up the direct BACTEC susceptibility test, 0.1 mL of
stock rifampicin solution (80 μg/ml) was added to a 12B
vial, the same final concentration (2.0 μg/ml) as used with
the indirect method. This vial was then inoculated with
0.1 mL of 1:10 dilution of the digested and decontami-
nated sputum sediment. A second BACTEC 12B vial, with-
out rifampicin, was prepared in the same manner to serve
as a growth control. Vials were pre-incubated before test-
ing on the BACTEC 460 instrument, depending on the
degree of smear positivity. 1+ and 2+ smears were held 1
day, and 3+ and 4+ smears were held 2 days before testing
commenced [16]. Vials were read daily until a control
Growth Index (GI) of at least 20 was reached. Susceptibil-
ity to rifampicin was determined by comparing the
change in GI values between the rifampicin-containing
vial and the control vial. When the GI in the drug vial
decreased in relation to the GI of the control vial, the
organism was considered "susceptible". Conversely, when
the GI of the drug vial increased in relation to the GI of the
control vial, the organism was considered "resistant".
Indirect testing confirmed the direct susceptibility results.
For the INNO-LiPA Rif.TB test, the sputum sediment was
heated to 80°C for 60 minutes to lyse the MTB. To
improve lysis, samples were sonicated for 10 minutes.
DNA in the lysate was amplified in a PCR targeting the
rpoB gene where mutations associated with rifampicin
resistance are located. Subsequently, the amplicon was

Page 4

BMC Infectious Diseases 2009, 9:139http://www.biomedcentral.com/1471-2334/9/139
Page 4 of 9
(page number not for citation purposes)
hybridized to immobilized-membrane bound probes
covering overlapping wild-type sequences [18,19]. The
interpretation of the results was done in accordance with
the manufacturer's instructions (Innogenetics, Inc, Ghent,
Belgium).
The direct phage assay was set up according to the method
described by McNerney, et al. [20,21]. Briefly, 1 ml of the
inoculum (direct sediment or 1:10 dilution) was added to
1 ml of 7H9 broth containing rifampicin (final concentra-
tion 4 μg/ml). A second drug-free control tube was also
similarly set up. The tubes were pre-incubated at 37°C for
24 hours. Then 50 μl of D29 phages was added to each
tube for a final concentration of 8 × 106/ml. The tubes
were incubated for 90 minutes at 37°C to allow the
phages to infect the mycobacteria cells. 0.25 ml of 100
mM ferrous ammonium sulphate was added to neutralize
extracellular phages. The mixture was added to 1 – 2 ml of
stationary phase grown Mycobacterium smegmatis mc2 155,
in a tube to which 9 ml of molten LB agar (1.5%) was
added. The mixture was poured to a 90 mm Petri dish. The
plates were sealed and incubated at 37°C for 16 hours.
The plates were examined for plaques and the number of
plaques recorded. The presence of plaques in the drug-free
sample and no plaques in the rifampicin-containing sam-
ple indicated that the MTB in the sputum was susceptible
to rifampicin. However, if plaques were seen in both sam-
ples, the MTB was considered rifampicin resistant.
For direct Etest, the manufacturer's directions (AB Biodisk,
Solna, Sweden) were followed with the exception of using
the sputum sediment in place of a mycobacteria suspen-
sion [22,23]. In order to reduce contamination, two 7H10
plates were made selective by addition of Polymixin B
(200 IU/ml), Amphotericin B (10 μg/ml), Carbenicillin
(50 μg/ml) and Trimethoprim (20 μg/ml); and each plate
was then inoculated with 0.3 ml of sediment, swabbing
evenly over the entire plate. Plates were pre-incubated for
48 hours at 37°C in a CO2 incubator, after which a
rifampicin Etest strip was added to one plate. This proce-
dural modification was necessary to obtain good growth
on the plates with the Etest strip. The second plate without
an Etest strip was used as a growth control. Plates were
sealed and read weekly, until an elliptical zone of inhibi-
tion was observed. A critical concentration of 1.0 μg/ml,
similar to that used for agar dilution in Middlebrook, was
used as a break point for determining rifampicin resist-
ance.
Statistical analysis
The statistical significance of the differences in the sensi-
tivities and the specificities of the tests evaluated were
assessed using Fisher's Exact test for comparing binomial
proportions.
Cost per test
Cost analyses, in US dollars, were performed for the four
direct rapid methods and the indirect BACTEC 460 drug
susceptibility test. Overall expenses were determined by
calculating the direct costs associated with each test,
including reagents, consumables and labor. The cost of
preventive maintenance of equipment and general operat-
ing supplies, such as gloves, gowns, masks, disinfectants
were included. Since sputum specimens needed to be
decontaminated, buffered and concentrated before being
tested, routine specimen processing costs were included in
the overall cost assessment. Since smear-positive samples
were required for testing, the cost for microscopy was also
included. Labor costs were based on remuneration of lab-
oratory personnel for the time in hours spent undertaking
the task. General overheads and items of a site specific
nature such as manager and director salaries, clerical
wages, capital equipment outlay, and facility operational
expenses were excluded. Such costs vary widely among
Ugandan laboratories as well as throughout the rest of
Africa.
Turn around times
Turn around time (TAT) was estimated as the time, in
days, from the receipt of specimens to the time results
were available in the laboratory. For the LiPA test, estima-
tion of the turn around times commenced from when a
batch of samples, stored at -20°C, was retrieved for anal-
ysis.
Results
A total of 203 smear-positive specimens yielding MTB iso-
lates were tested. Eighty percent of the specimens had 3+
or 4+ smears (17, 23, 50 and 113 for 1+, 2+, 3+ and 4+
graded smears, respectively). The distribution of speci-
mens tested by the direct methods was 39 BACTEC, 44
Etest, 77 phage and 43 LiPA. Out of 203 specimens exam-
ined, 17 primary BACTEC cultures had no growth, thus
indirect drug susceptibility results were not available for
comparison with the direct test results. Additionally, 27
direct tests were not interpretable due to various reasons:
i) growth failures occurred in one direct BACTEC, one
Etest, and seven phage; ii) control failures occurred in one
direct BACTEC and eight phage; iii) two contaminated
cultures in each of the direct BACTEC, Etest, and phage;
and iv) three PCR failures in the LiPA. With the direct
tests, failures were observed throughout the smear positiv-
ity range, and were not associated with only specimens
containing low numbers of MTB.
The ability of each direct test and the indirect BACTEC
drug susceptibility test to yield interpretable results is
shown in Table 1. Direct testing by BACTEC, Etest, phage
or LiPA yielded a result in 87, 89, 73, and 86% of samples

Page 5

BMC Infectious Diseases 2009, 9:139http://www.biomedcentral.com/1471-2334/9/139
Page 5 of 9
(page number not for citation purposes)
tested, respectively. The LiPA was the only direct test to
provide a higher number of test results than the indirect
test (37 vs. 33). The phage method performed least well
with a success rate of 73% compared to 96% for the indi-
rect method.
Performance of direct tests
Comparable results between the direct methods and the
indirect BACTEC were available for 159 (78%) of the 203
specimens comprising 34, 44, 56 and 30 samples tested
by direct BACTEC, Etest, phage and LiPA respectively.
Overall, of the 159 isolates tested in the indirect BACTEC,
115 (72%) were rifampicin susceptible and 44 (28%)
resistant. Performance characteristics of the four methods
are shown in Table 2. Using the indirect BACTEC as the
gold standard, the sensitivity and specificity for detecting
rifampicin resistance in direct BACTEC was 100% and
100%, for direct E-test 100% and 100%, direct phage 94%
and 95% and for direct LiPA 87% and 87%. Overall con-
cordance with the indirect BACTEC for determining
rifampicin susceptibility and resistance ranged from
100% (direct BACTEC and Etest) to 93% (LiPA). How-
ever, it should be noted that none of the differences in
sensitivity and specificity observed were statistically sig-
nificant.
Test costs
Total costs for each rapid method and the indirect
BACTEC susceptibility test are shown in Table 3. The dis-
tribution of costs attributed to each of the rapid tests by
reagents/consumables and labor are indicated. The indi-
rect BACTEC and LiPA had similar total costs and were the
most expensive. Total costs for the Etest and phage assay
were similar and the least expensive. In-house preparation
of media for these tests kept test-specific reagent costs low.
The direct BACTEC cost was in mid-range. Since testing a
sample by LiPA required two extra control strips, testing in
batches produced substantial costs savings.
Turn around times (TAT)
The LiPA and phage results were available in 2–4 days,
being the most rapid of the four tests. The average time to
a positive direct BACTEC was eight days. The Etest had the
longest TAT of 14–37 days (mean, 20), which was still
faster than the conventional indirect BACTEC with time to
positive ranging from 8–60 days (mean, 24). The TAT gen-
erally coincided with degree of smear positivity. Direct
BACTEC results were available as early as four days with
all 4+ specimens and one 3+ specimen. The remaining 3+
and half of the 2+ specimens had direct BACTEC results
within seven days. The other half of 2+ specimens took
longer, up to 20 days. The direct BACTEC result for the
one analyzable 1+ specimen was available in 6 days.
Discussion
The current conventional methods for determining drug
susceptibility are based on testing isolates rather than
direct testing of organisms in patient specimens. To pro-
vide susceptibility data in a clinical and public health rel-
evant time frame, results should be available within a few
days of specimen collection. Molecular and phage tests,
both commercial and those developed in-house may pro-
vide results in 1–4 days. Such examples are the INNO-
LiPA tested in this study, HAIN Rif.TB (HAIN Lifescience
GmbH, Nehren, Germany) [24,25], molecular beacon
assay [26], FASTPlaqueTB-RIF (Biotec Laboratories Ltd.,
Ipswich, UK) [27], and in-house phage test [28]. Methods
that provide results in around 8–12 days rely on culture
and use some means of detecting growth before it is mac-
roscopically evident. These include BACTEC 460 radio-
metric TB system [16], BACTEC MGIT 960[29], Etest
[22,23], nitrate reductase assay (NRA) [30], microscopic
observation drug susceptibility (MODS) assay [31], and
other microplate assays using oxidation-reduction indica-
tors [32] for early detection of growth. All of these are
promising methods for rapid detection of rifampicin and
isoniazid resistant MTB and could potentially be set up to
Table 1: Comparison of various direct and the indirect BACTEC 460 susceptibility methods according to AFB smear grade.
Smear
Grade
Direct BACTEC
(n = 39)
Direct
Available
Direct E-test
(n = 44)
Direct
Available
Direct phage
(n = 77)
Direct
Available
Direct LiPA
(n = 43)
Direct
Available
No.
Tested
Indirect
Available
No.
Tested
Indirect
Available
No.
Tested
Indirect
Available
No.
Tested
Indirect
Available
1+
31
(33%)
6
(100%)
5
(71%)
22
(96%)
1
(33%)
6
(100%)
7
(100%)
23
(100%)
201
(50%)
4
(100%)
16
(45%)
21
(100%)
20
(0%)
4
(44%)
14
(74%)
38
(81%)
1
(50%)
6
(67%)
18
(95%)
47
(100%)
107
(70%)
3
(75%)
7
(100%)
20
(91%)
2
(20%)
3
(75%)
7
(100%)
21
(95%)
2+
644
(100%)
15
(88%)
20
(95%)
94
3+
7 17 197
4+
23 214722
TOTAL 39 34
(87%)
37
(95%)
4439
(89%)
42
(96%)
7756
(73%)
74
(96%)
4337
(86%)
33
(77%)

Page 6

BMC Infectious Diseases 2009, 9:139http://www.biomedcentral.com/1471-2334/9/139
Page 6 of 9
(page number not for citation purposes)
test second-line drugs, which would facilitate rapid detec-
tion of XDR-TB. Most of these methods have previously
been used with AFB smear-positive specimens; but few
evaluations have been done in high disease burden set-
tings such as Uganda. Since our interest was in examining
the feasibility of introducing one these technologies into
a a TB reference diagnostic laboratory we trialed each of
the tests independently on a limited number of samples
each week in order to reflect the caseload of re-treatment
patients attending the clinic.
The LiPA met the qualifications of providing results very
quickly and results could have been available even sooner
if they had not been batch tested. This method yielded
more interpretable results than the indirect culture based
method when applied to the same samples (Table 1). Our
results are not surprising in that this test looks for muta-
tions in the genome and does not require viable bacilli.
Other authors have reported the LiPA used directly with
sputum specimens detects MTB in specimens missed by
culture [33]. The lower proportion of evaluable specimens
using the BACTEC culture in this comparison may have
been due to non-viable MTB in patients on drug therapy
or killing of bacilli during the NaOH decontamination
process. Three of the PCR reactions for the LiPA failed,
possibly due to inhibition. It is probable that the yield of
positive LiPA results would be increased if a more rigorous
DNA extraction method were used instead of crude
lysates. However, this would increase the hands on time
required and overall cost of the test. When compared to
the indirect direct susceptibility results, there were four
discrepant LiPA results (Table 2). Two of these (resistant
by indirect BACTEC, susceptible by LiPA) may be due
Table 2: Performance of rapid test methods for detecting
rifampicin resistance compared to indirect BACTEC.
Indirect BACTEC
Resistant Susceptible Total
Direct
BACTEC
Resistant404
Susceptible0 3030
Total4 3034
Direct
Etest
Resistant909
Susceptible0 3030
Total9 3039
Direct
Phage
Resistant 152 17
Susceptible138 39
Total 16 4056
Direct
LiPA
Resistant 13215
Susceptible213 15
Total1515 34
Table 3: Cost comparison (US$) of indirect BACTEC and four rapid methods
Indirect BACTECDirect BACTECDirect Etest Direct Phage Direct LiPA
Reagents and Consumables:
Sputum Processing
Fluorescent Microscopy
Test-specific
Overhead*
Total
1.01
1.33
24.44
3.23
30.01
1.01
1.33
16.64
3.23
22.21
1.01
1.33
6.99
3.23
12.56
1.01
1.33
5.63
3.23
11.20
1.01
1.33
30.60
3.55
36.49
Labor:
Sputum Processing
Fluorescent Microscopy
Test-specific
Overhead
Total
0.56
1.04
8.23
1.82
11.65
0.56
1.04
2.99
1.82
6.41
0.56
1.04
2.62
1.82
6.04
0.56
1.04
4.78
1.82
8.20
0.56
1.04
1.83
2.00
5.43
Total Cost per Specimen41.6628.6218.6019.4041.92
* Overhead included preventive maintenance of equipment and general operating supplies, such as gloves, gowns, masks and disinfectants.

Page 7

BMC Infectious Diseases 2009, 9:139http://www.biomedcentral.com/1471-2334/9/139
Page 7 of 9
(page number not for citation purposes)
mutations outside the "hotspot" region in the rpoB gene
which are not covered by the LiPA test. Two discrepancies
in which the LiPA was resistant and the indirect test was
susceptible are unexplainable, but could be due to a mix-
ture of resistant and susceptible strains in one sample. All
four LiPA discrepancies where confirmed with at least one
other indirect BACTEC test. Our results suggest that this
test could be used to triage patients for appropriate TB
treatment regimen and control measures. Commercially
available line probe assays were recently endorsed by
WHO for this purpose, subject to the availability of appro-
priate training and supervision. In addition to speed, the
test has a second advantage over culture based methods in
that samples can be treated to render them non- infec-
tious. The major disadvantage of the LiPA compared to
the other technologies we investigated is the high cost and
requirement for separate clean room facilities to avoid
contamination of the amplification reaction.
The in-house phage test was also ultra-rapid. Unfortu-
nately, many of the tests did not yield results (Table 1).
Some of the reasons for this were microbiological con-
tamination and failure of Mycobacterium smegmatis indica-
tor plates or unexpected control results. Three discrepant
drug susceptibility results were observed; one R indirect
BACTEC but S phage; and two S indirect BACTEC but R
phage. The one discrepant isolate which was sensitive by
the phage test but resistant by indirect BACTEC was resist-
ant by repeat indirect BACTEC and LiPA susceptibility
tests. However, the isolate remained sensitive by repeat
phage test suggesting phage failure to detect resistance.
The two discrepant isolates that were resistant by phage
but susceptible by indirect BACTEC were susceptible by
repeat indirect phage as well as BACTEC. The major
advantage of the direct phage is its low cost. However, in
our experience it was technically difficult to adapt, per-
form and interpret when compared to the other tests.
The direct BACTEC was found to be fairly rapid, techni-
cally easy and fitted well in the work flow of the labora-
tory. For this reason it was the preferred test of the
laboratory staff. Although most of the specimens tested
were from the multi-bacillary patients, the sensitivity and
specificity of direct BACTEC of 100% is encouraging. In
principle, the test could be easily extended to other liquid
culture systems such as the Mycobacteria Growth Indica-
tor Tube (MGIT) BACTEC 960 (Becton Dickinson, Tow-
son, MD),) and BacT/Alert (bioMérieux, Hazelwood,
MO) and recent work by El-Sayed et al [34] supports this.
It has the added advantage that the technology may also
be used for diagnosis and, if required, testing susceptibil-
ity to other anti-tuberculosis drugs. The laboratory is cur-
rently undertaking further studies with liquid culture
including the direct detection of resistance to isoniazid
and rifampicin using MGIT 960.
The direct Etest also had 100% agreement with the indi-
rect BACTEC. It was simple to set up, learn, and perform
and is inexpensive. We determined that the rifampicin
susceptible isolates had MICs in the range of 0.002 to 1.0
μg/ml and all resistant isolates had a MIC >256 μg/ml, an
added benefit of the Etest. However, TATs were much
longer than for the other tests suggesting the Etest would
be of limited value for triaging patients.
We believe this is the first report comparing head-to-head
these four technologies for detecting drug resistance from
sputum in a TB endemic country with automated liquid
culture. It should be noted that our evaluation was based
on samples with a high percentage of heavy smear-posi-
tive specimens, more specimens in the 1+ to 2+ range
need to be tested to determine the sensitivity of these
methods with low numbers of organisms. Similarly, the
lack of statistical significance of the test performance data
reflects the small sample size. A larger study would be
required to fully determine the sensitivity, specificity and
predictive values of the tests in this setting.
An important consideration when testing M. tuberculosis
in the laboratory is safety as great care must be taken to
avoid exposure to infected aerosols. Indirect culture based
methods require handling of large numbers of bacilli and
stringent safety precautions are necessary, including nega-
tive pressure rooms and microbiological safety cabinets.
Direct testing of sputum involves much lower numbers of
bacilli reducing the risk to laboratory personnel. How-
ever, inhalation of a very small number of bacilli can
result in infection and tuberculosis disease and the direct
BACTEC, Etest and phage all require protective facilities to
microbiological safety category level III. The risk of aero-
sol production and accidental exposure increases each
time viable bacteria are handled. The BACTEC test
requires less manipulation of open samples than either
the Etest or the Phage assay as so might be considered the
least risky procedure. As previously mentioned the LiPA
may be considered the safest technology due to the oppor-
tunity to sterilize samples prior to testing, when they may
be handled in a standard laboratory without fear of infec-
tion.
The majority of diagnostic laboratories in TB endemic
countries are poorly resourced and do not have the capac-
ity to undertake culture of M. tuberculosis, and drug suscep-
tibility testing in such settings is currently restricted to a
few central referral laboratories. Until recently these refer-
ence laboratories were restricted to culture using home-
made Lowenstein Jensen slopes, a slow and cumbersome
procedure. However, a recent change in international pol-
icy and the launch of the Global Laboratory Strengthening
Initiative by the STOP-TB partnership has changed the
outlook for TB laboratories. A program of refurbishment

Page 8

BMC Infectious Diseases 2009, 9:139http://www.biomedcentral.com/1471-2334/9/139
Page 8 of 9
(page number not for citation purposes)
funded by international donors has led to the introduc-
tion of automated liquid culture systems. This has coin-
cided with an expansion of activities to treat and control
MDR-TB. The tests we have investigated would be appro-
priate for implementation in the reference laboratories of
developing countries where treatment for MDR-TB is
available. Ideally new tests should be robust, fast and effi-
cient. Our finding that the direct BACTEC method was the
most convenient in this a TB reference laboratory suggests
that further evaluation of automatic liquid culture meth-
ods is warranted. The LiPA test was rapid and has advan-
tages of safety. However, the necessity to avoid
contamination of the PCR requires dedicated clean room
facilities. Investment will also be needed to retrain techni-
cal staff in molecular skills and this option may be less
attractive for some laboratories.
There are concerns that resistance to rifampicin may not
be a reliable marker for MDR-TB in some settings, espe-
cially where rifampicin mono-resistant isolates are com-
mon. Review of indirect BACTEC drug susceptibility data
from the laboratory during 2000–2006 showed that 164
(90%) of the 181 rifampicin resistant isolates were also
isoniazid resistant. However, there may be clinical situa-
tions in which rapid detection of isoniazid resistance, in
the absence of rifampicin resistance, may be useful to con-
struct an effective treatment regimen. Thus, inclusion of
isoniazid in rapid tests may be desirable. Furthermore,
with concern about possible emergence of XDR-TB in
Africa, adding testing for resistance to second-line drugs
(one fluoroquinolone and two aminoglycosides) to the
rapid test is logical. The rapid culture-based methods can
easily be adapted to test for isoniazid resistance and addi-
tional drugs. Several of the molecular techniques, Geno-
type MTBDR (Hain Lifescience) and molecular beacon
assay, already test for the katG and inhA gene mutations
that are associated with isoniazid resistance. It is likely
that the molecular assays can be easily modified to
include gene targets for second-line drugs. However, the
sensitivity of such tests has yet to be determined and pre-
liminary reports suggest they may be less accurate than
tests for rifampicin.
Conclusion
A high correlation was observed between the indirect
rifampicin susceptibility results using the BACTEC 460
and those from the four rapid methods when applied to
smear-positive sputum specimens. Although the number
of specimens tested was small, the results demonstrate
that personnel in this lab setting can be trained to success-
fully use a number of different test platforms. The test pre-
ferred by laboratory personnel, the direct BACTEC, was
neither the fastest nor the cheapest but was preferred on
grounds of convenience. The line probe assay was consid-
ered the safest test and was rapid but was expensive and
required separate laboratory facilities to avoid PCR con-
tamination. We recommend that further investigation be
undertaken to determine the efficacy and cost effective-
ness of direct liquid culture and the commercial line
probe assay for detection of MDR-TB in an sub-Sahara
African setting. We suggest that the delays in obtaining
results from the direct Etest and the lack of robustness and
inconvenience of the in-house phage based test preclude
their application for detection of drug resistant tuberculo-
sis in a clinical setting.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
SO performed the phage assays and participated in draft-
ing the manuscript.
HT assisted with the phage assays and training for LiPA.
BBA performed the LiPA assays and participated in draft-
ing the manuscript.
FM performed the Bactec and E-tests.
SK participated in general supervision of laboratory work,
collection of cost data and drafting the manuscript.
KDE, EJ, RM, PGS, RDM, JE and MLJ participated in the
conception and design of the study, general supervision of
the research, and critical revision of the manuscript.
AO, IA and WW: participated as study physicians.
All authors read and approved the final version of the
manuscript.
Acknowledgements
This study received financial support from the UNICEF/UNDP/World
Bank/WHO Special Programme for Research and Training in Tropical Dis-
eases (TDR), the Wellcome Trust – Burroughs Wellcome Fund Infectious
Diseases Initiative 063410/ABC/00/Z, the contract NOI-A195383 (Tuber-
culosis Research and Prevention Control Unit) National Institutes of
Allergy and Infectious Diseases/National Institutes of Health and the
Department for International Development (UK) TARGETS Research
Consortium on Communicable Diseases, Risk and Poverty and was carried
out with institutional support of the Joint Clinical Research Center (JCRC)
and Makerere University, Kampala, Uganda. Etest strips were kindly pro-
vided by AB Biodisk, Solna, Sweden. D29 phage were supplied by Ruth
McNerney of the London School or Hygiene & Tropical Medicine. We are
grateful for the contributions of the ward staff at the Old Mulago Hospital
and the technical support provided by the Molecular Biology Laboratory at
Makerere University Medical School and the Mycobacteriology Lab at
JCRC.
References
1.Corbett EL, Watt CJ, Walker N, Maher D, Williams BG, Raviglione
MC, Dye C: The growing burden of tuberculosis: global trends

Page 9

BMC Infectious Diseases 2009, 9:139http://www.biomedcentral.com/1471-2334/9/139
Page 9 of 9
(page number not for citation purposes)
and interactions with the HIV epidemic. Arch Intern Med 2003,
163(9):1009-1021.
World Health Organisation: Global tuberculosis control: surveil-
lance, planning, financing. Geneva: WHO; 2009.
Guwatudde D, Zalwango S, Kamya MR, Debanne SM, Diaz MI, Okw-
era A, Mugerwa RD, King C, Whalen CC: Burden of tuberculosis
in Kampala, Uganda. Bull World Health Organ 2003,
81(11):799-805.
CDC: Notice to Readers: Revised Definition of Extensively
Drug-Resistant Tuberculosis. JAMA 2006, 296(23):2792.
Zager EM, McNerney R: Multidrug-resistant tuberculosis. BMC
Infect Dis 2008, 8:10.
Case definition for extensively drug-resistant tuberculosis.
Wkly Epidemiol Rec 2006, 81(42):408.
Extensively drug-resistant tuberculosis – United States,
1993–2006. MMWR Morb Mortal Wkly Rep 2007, 56(11):250-253.
Sakoane R: XDR-TB in South Africa: back to TB sanatoria per-
haps? PLoS Med 2007, 4(4):e160.
Temple B, Ayakaka I, Ogwang S, Nabanjja H, Kayes S, Nakubulwa S,
Worodria W, Levin J, Joloba M, Okwera A, et al.: Rate and ampli-
fication of drug resistance among previously-treated
patients with tuberculosis in Kampala, Uganda. Clin Infect Dis
2008, 47(9):1126-1134.
Center for Disease Control and Prevention: Epidemiologic notes
and reports nosocomial transmission of multidrug-resistant
tuberculosis among HIV-infected persons-Florida and New
York, 1988–1991. Morb Mortal Wkly Rep 1991, 40:585-591.
Center for Disease Control and Prevention: Transmission of
multi-drug-resistant tuberculosis among immunocompro-
mised persons in a correctional system-New York, 1991.
Morb Mortal Wkly Rep 1992, 41:507-509.
Sharp V, Sepkowitz KA, Lockhart B: Survival with MDR-TB and
AIDS depends on proper selection of initial therapy. The
Eighth International Symposium on Infections in the Immunocompromised
Host: June 19–22 1994; Davos, Switzerland 1994.
Telzak E, Turett G, Torian L, Fazal B, Weisfuse I: Multidrug-resist-
ant tuberculosis (MDRTB) in the South Bronx, New York
City: clinical characteristics and improved outcome. The
Eighth International Symposium on Infections in the Immunocompromised
Host: June 19–22 1994; Davos, Switzerland 1994.
Joloba ML, Whalen CC, Cave DM, Eisenach KD, Johnson JL, Okwera
A, Morrissey A, Bajaksouzian S, Feagin J, Mugerwa R, et al.: Determi-
nation of drug susceptibility and DNA fingerprint patterns of
clinical isolates of Mycobacterium tuberculosis from Kam-
pala, Uganda. East Afr Med J 2000, 77(2):111-115.
American, Thoracic, Society: Diagnostic Standards and Classifi-
cation of Tuberculosis in Adults and Children. Am J Respir Crit
Care Med 2000, 161:1376-1395.
Siddiqi SH: BACTEC 460 TB System. Product and Procedure
Manual. Sparks, MD: Becton Dickinson Diagnostic Instrument Sys-
tems; 1995.
Muhumuza J, Asiimwe BB, Kayes S, Mugyenyi R, Whalen C, Mugerwa
RD, Boom H, Eisenach KD, Joloba ML: Introduction of an in-
house PCR for routine identification of M. tuberculosis in a
low-income country.
Int
10(11):1262-1267.
Tortoli E, Mariottini A, Mazzarelli G: Evaluation of INNO-LiPA
MYCOBACTERIA v2: improved reverse hybridization mul-
tiple DNA probe assay for mycobacterial identification. J Clin
Microbiol 2003, 41(9):4418-4420.
Tortoli E, Nanetti A, Piersimoni C, Cichero P, Farina C, Mucignat G,
Scarparo C, Bartolini L, Valentini R, Nista D, et al.: Performance
assessment of new multiplex probe assay for identification of
mycobacteria. J Clin Microbiol 2001, 39(3):1079-1084.
McNerney R: Phage replication technology for diagnosis and
susceptibility testing. In Mycobacterium tuberculosis. Volume
54. NY: Humana Press, Totowa; 2001.
McNerney R, Kambashi BS, Kinkese J, Tembwe R, Godfrey-Faussett
P: Development of a bacteriophage phage replication assay
for diagnosis of pulmonary tuberculosis. J Clin Microbiol 2004,
42(5):2115-2120.
Wanger A, Mills K: Etest for susceptibility testing of Mycobac-
terium tuberculosis and Mycobacterium avium-intracellu-
lare. Diagn Microbiol Infect Dis 1994, 19(3):179-181.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
J Tuberc Lung Dis 2006,
18.
19.
20.
21.
22.
23. Wanger A, Mills K: Testing of Mycobacterium tuberculosis sus-
ceptibility to ethambutol, isoniazid, rifampin, and strepto-
mycin by using Etest. J Clin Microbiol 1996, 34(7):1672-1676.
Hillemann D, Rusch-Gerdes S, Richter E: Evaluation of the Geno-
Type MTBDRplus Assay for Rifampin and Isoniazid Suscepti-
bility Testing of Mycobacterium tuberculosis Strains and
Clinical Specimens. J Clin Microbiol 2007, 45(8):2635-2640.
Miotto P, Piana F, Penati V, Canducci F, Migliori GB, Cirillo DM: Use
of genotype MTBDR assay for molecular detection of
rifampin and isoniazid resistance in Mycobacterium tubercu-
losis clinical strains isolated in Italy. J Clin Microbiol 2006,
44(7):2485-2491.
Lin SY, Probert W, Lo M, Desmond E: Rapid detection of isoni-
azid and rifampin resistance mutations in Mycobacterium
tuberculosis complex from cultures or smear-positive sputa
by use of molecular beacons. J Clin Microbiol 2004,
42(9):4204-4208.
Albert H, Trollip A, Seaman T, Mole RJ: Simple, phage-based
(FASTPplaque) technology to determine rifampicin resist-
ance of Mycobacterium tuberculosis directly from sputum.
Int J Tuberc Lung Dis 2004, 8(9):1114-1119.
Traore H, Ogwang S, Mallard K, Joloba ML, Mumbowa F, Narayan K,
Kayes S, Jones-Lopez EC, Smith PG, Ellner JJ, et al.: Low-cost rapid
detection of rifampicin resistant tuberculosis using bacteri-
ophage in Kampala, Uganda. Ann Clin Microbiol Antimicrob 2007,
6:1.
Palaci M, Ueki SY, Sato DN, Da Silva Telles MA, Curcio M, Silva EA:
Evaluation of mycobacteria growth indicator tube for recov-
ery and drug susceptibility testing of Mycobacterium tuber-
culosis isolates from respiratory specimens. J Clin Microbiol
1996, 34(3):762-764.
Lemus D, Montoro E, Echemendia M, Martin A, Portaels F, Palomino
JC: Nitrate reductase assay for detection of drug resistance
in Mycobacterium tuberculosis: simple and inexpensive
method for low-resource laboratories. J Med Microbiol 2006,
55(Pt 7):861-863.
Moore DA, Evans CA, Gilman RH, Caviedes L, Coronel J, Vivar A,
Sanchez E, Pinedo Y, Saravia JC, Salazar C, et al.: Microscopic-
observation drug-susceptibility assay for the diagnosis of TB.
N Engl J Med 2006, 355(15):1539-1550.
Martin A, Camacho M, Portaels F, Palomino JC: Resazurin micro-
titer assay plate testing of Mycobacterium tuberculosis sus-
ceptibilities to second-line drugs: rapid, simple, and
inexpensive method. Antimicrob Agents Chemother 2003,
47(11):3616-3619.
Traore H, van Deun A, Shamputa IC, Rigouts L, Portaels F: Direct
detection of Mycobacterium tuberculosis complex DNA and
rifampin resistance in clinical specimens from tuberculosis
patients by line probe assay. J Clin Microbiol 2006,
44(12):4384-4388.
El-Sayed Zaki M, Goda T: Rapid phenotypic assay of antimyco-
bacterial susceptibility pattern by direct mycobacteria
growth indicator tube and phage amplified biological assay
compared to BACTEC 460 TB. Tuberculosis (Edinb) 2007,
87(2):102-108.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
Pre-publication history
The pre-publication history for this paper can be accessed
here:
http://www.biomedcentral.com/1471-2334/9/139/pre
pub