Background
The emergence of drug-resistant tuberculosis, especially multidrug-resistant tuberculosis (MDR-TB), threatens TB control efforts in most countries including Cameroon. MDR-TB is caused by strains of
Mycobacterium tuberculosis (Mtb) with in-vitro resistance to two of the most potent anti-tuberculosis drugs, Rifampicin (RIF) and Isoniazid (INH) [
1,
2]. Recent World Health Organization (WHO) estimates suggest that each year about 460/20882 (2.2%) new pulmonary TB (PTB) patients in Cameroon develop MDR-TB, and another 170/1575 (11%) previously treated PTB patients develop MDR-TB. Unfortunately, only a small fraction (~40%) of these patients have access to drug susceptibility testing (DST) for MDR-TB [
3]. In response to the need to expand access to diagnostics for patients at risk of MDR-TB, the government of Cameroon signed an agreement with the EXPAND-TB program (Expand Access to New Diagnostics for TB) project in 2010. The EXPAND-TB program is a UNITAID funded program with the overall objective to increase access to diagnostics for patients at risk of MDR-TB in 27 low-income and high TB burden countries, including Cameroon. In Cameroon, this program plans to strengthen existing MDR-TB diagnostic laboratories and establish new laboratories with the capacity to culture and conduct culture-based DST of
Mtb. Also, the program introduced the Genotype MTBDRplus assay for the rapid diagnosis of MDR-TB [
4].
The Genotype MTBDRplus assay (Hain Lifescience GmbH, Nehren, Germany) is a molecular-based test that can rapidly diagnose MDR-TB. This assay detects MDR-TB by identifying mutations in the
rpoB gene associated with resistance to RIF and in the
KatG gene and promoter region of the
inhA gene associated with INH resistance [
5,
6]. Two independent systematic reviews concluded that the Genotype MTBDRplus assay was highly accurate in diagnosing MDR-TB when compared to the culture-based proportion method [
7,
8]. These two reviews reported pooled sensitivities of 88.7% [
7] and 91% [
8] and specificities of 99.2% [
7] and 99% [
8], respectively, for detection of MDR-TB. Although there is evidence of the accuracy of the MTBDRplus assay, the prevalence of MDR-TB varies widely worldwide. This variation in prevalence of MDR-TB might have a significant impact on the predictive value of the MTBDRplus assay [
9]. Thus, it is important to evaluate the assay before its use in routine diagnosis.
In this study, we sought to determine the accuracy of the MTBDRplus assay to diagnose MDR-TB cases in Cameroon. Furthermore, the government of Cameroon has recently adopted, an alternative molecular-based test the GeneXpert MTB/RIF assay (Cepheid, Sunnyvale, CA, USA) for the initial screening of TB patients at risk of MDR-TB [
10‐
12]. The GeneXpert MTB/RIF assay simultaneously detects TB and resistance to RIF directly from clinical specimen within 2 h. At present, the government has introduced the GeneXpert MTB/RIF assay in 3 of the 10 regions of the country but plans to roll-out the assay to all parts of the Country [
12]. As such, we discuss on the additional use and role of the Genotype MTBDRplus assay in the diagnosis of MDR-TB amidst the availability of the GeneXpert MTB/RIF.
Methods
Study design and setting
Participants were enrolled at six health facilities in three towns in Cameroon, between April 2015 and February 2016. The health facilities were the Jamot Hospital in Yaounde (3.8480 N, 11.5020 E), the Bamenda Regional Hospital in Bamenda (5.9630 N, 10.159°E), and 4 facilities in Douala (4.0510 N, 9.7680 E) including Laquintinie Hospital, the St Albert the Great Hospital, the dispensary Catholic Barcelone and the New Bell District Hospital. These health facilities are the major TB diagnosis and treatment centers in their respective locations.
Study participants
Participants aged ≥15 years old visiting any of the six health facilities and diagnosed by the laboratory of the health facility to be sputum smear positive for acid-fast bacilli (AFB) or diagnosed by the consulting clinician at the health facility to have clinical symptoms and signs suggestive of PTB were recruited into the study. This recruitment approach, allowed us to enroll patients from whom sputum samples will likely yield positive Mtb culture.
Although there are many naïve PTB patients (n = 20,882), very few are MDR-TB (2.2%; n = 460/20882). To include a high proportion of patients likely to have MDR-TB, the initial focus was to enrol previously treated PTB patients in whom the proportion of MDR-TB is higher (11%; n = 170/1575) and then expand enrollment to include naïve PTB patients. Written informed assent and consent was obtained from all study participants, and a structured questionnaire was used to collect demographic and medical information, including prior PTB history, age, gender, and HIV status. An expectorated sputum specimen was collected from each participant and transported to the study laboratory for mycobacterial culture, DST and MTBDRplus assay.
Specimen processing and microbiology testing
Mycobacterial culture and drug susceptibility testing were performed at the Mycobacteriology unit of Centre Pasteur du Cameroun in Yaounde. Within 48 h of receipt of sputum samples in the laboratory, smears were made, stained using the auramine staining technique and examined by fluorescence microscopy. The smears were read and graded according to WHO guidelines and that of the International Union against Tuberculosis and Lung Diseases (IUALTD) [
13,
14]. The sputum samples were then processed for culturing by decontaminating in Sodium hydroxide-Sodium citrate-N-acetyl L-cysteine (NaOH-NaC-NALC) solution and cultured in MGIT tubes using the BACTEC MGIT 960 system (Becton Dickinson, Franklin Lakes, NJ). MGIT tubes that gave a positive fluorescent signal with the BACTEC MGIT 960 equipment were checked for acid-fast-bacilli using Ziehl-Neelsen staining and confirmed for
Mtb complex (MTBC) using the TB Ag MPT64 test (SD Bioline, Standard Diagnostics, Suwon, Korea). Cultures growing MTBC were assessed for contamination with other bacteria or fungi by growth on blood agar medium for 24 h at 37 °C. If no contaminants (i.e.
, Bacteria or fungi) were detected on the blood agar, drug susceptibility testing (DST) was performed using the MGIT 960 Streptomycin-Isoniazid-Rifampicin-Ethambutol (SIRE) kit (Becton Dickinson Diagnostic Systems). If contaminants were detected on the blood agar, the MTBC cultures were decontaminated and the culturing process repeated. The critical concentrations of Streptomycin, Rifampicin, Isoniazid and Ethambutol used in the MGIT 960 SIRE kit were respectively 1.0 μg/ml, 1.0 μg/ml, 0.1 μg/ml and 5.0 μg/ml.
Immediately the DST result of a patient’s sample was available, an aliquot of the culture was made and stored at +4 °C. Once the total number of cultures with DST results reached 226, the stored aliquots were screened using the genotype MTBDRplus assay.
Genotype MTBDRplus assay
Samples were screened using the genotype MTBDRplus assay according to manufacturer’s instructions [
5] and interpreted without knowledge of susceptibility results determined by the MGIT SIRE 960 system. Testing consisted of three steps: DNA extraction using the Genolyse kit (Hain Lifescience, Nehren, Germany), multiplex PCR amplification using biotinylated primers and reverse hybridization. The three steps were carried out in three separated rooms.
Each MTBDRplus strip had 27 reaction zones or probes that hybridize DNA (amplicons). Six of the probes were positive controls, while 21 probes were used to detect resistance to RIF and INH. For the detection of RIF resistance, the strip contained 8 probes that hybridize DNA from codons 506 to 533 of the rpoB gene and 4 mutation probes (rpoBMUT1 (D516V), rpoBMUT2A (H526Y), rpoBMUT2B (H526D) and rpoBMUT3 (S531 L)). Similarly, for INH resistance, the strip contains probes that hybridized DNA at codon 315 of the KatG gene and positions −1 to −22 on the inhA promoter region. The mutation probes associated with INH resistance were katGMUT1 (S315 T1), katGMUT2 (S315 T2), inhAMUT1 (C15T), inhAMUT2 (A16G), inhAMUT3A (T8C), and inhAMUT3B (T8A). When the DNA amplicons hybridized to the probes on the MTBDRplus strip following hybridization, a dark band was produced that was easily interpreted as positive.
The MTBDRplus strips were interpreted in a two-stage process. First, the presence of the 6 control bands was confirmed, demonstrating the assay worked and that MTBC was present. Secondly, susceptibilities to RIF and INH were assessed. A sample was considered to be resistant to the drug if at least one of the wild-type bands was absent or a band indicating a common mutation in the drug resistance-related gene was present. Likewise, a sample was considered sensitive to the drug if all the wild-type bands of the gene were present and no common mutation was detected.
Statistical methods
Participants were classified as new, previously treated PTB patients or patients with unknown PTB history. Participant characteristics were compared between new and previously treated TB patients using two-tailed Fisher’s exact test or Chi-square test for categorical variables and non-parametric Mann-Whitney test for continuous numerical variables. No adjustment for multiple comparisons was made because only a few planned comparisons were made and the data evaluated are actual observations [
15].
To determine the accuracy of the MTBDRplus assay to diagnose MDR-TB, MTBDRplus results were compared to the gold standard MGIT 960 SIRE system. We calculated the sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) with 95% confidence intervals (CI) of the MTBDRplus assay for the detection of MDR-TB and resistance to RIF and INH. Sensitivity was defined as the proportion of isolates correctly determined as resistant by the MTBDRplus assay compared with MGIT 960 SIRE system. Specificity was defined as the proportion of isolates correctly determined as susceptible by the MTBDRplus assay compared with MGIT 960 SIRE system. PPV was defined as the proportion of resistant isolates determined by the MGIT 960 SIRE system among isolates determined as resistant by the MTBDRplus assay. NPV was defined as the proportion of susceptible isolates determined by the MGIT 960 SIRE system among isolates determined as susceptible by the MTBDRplus assay. The degree of agreement between MGIT 960 SIRE system and the MTBDRplus assay was also assessed using Cohen’s kappa (
ĸ) coefficient.
ĸ values of >0.75 defined as showing very good agreement,
ĸ values of <0.4 defined as showing poor agreement and
ĸ values of 0.4–0.75 defined as showing fair to good agreement [
16].
All analyses and comparisons were done with GraphPad prism software, version 6 (GraphPad Software, California, USA) and all results having a p < 0.05 were considered statistically significant.
Discussion
This study reports the accuracy of the Genotype MTBDRplus assay for diagnosing MDR-TB in new and previously treated PTB patients in Cameroon. Our analysis showed that based on 49 MDR-TB cases, the MTBDRplus assay had a sensitivity and specificity of 94% and 100%, respectively. Such high sensitivity and specificity values make the Genotype MTBDRplus very suitable for use to diagnose MTB-DR in Cameroon. The specificity and sensitivity values reported in our study are similar to those reported in the two systematic reviews of the MTBDRplus [
7,
8]. Our participant recruitment strategy led to both a high number of positive
Mtb cultures and MDR-TB patients (22%; 49/225). This proportion of MDR-TB patients is higher than the population estimate of 2.2% (460/20882) among newly diagnosed TB patients and 11% (170/1575) among previously treated TB patients. However, a sensitivity analysis indicates that the high proportion of MDR-TB patients in our study does not affect the positive predictive value of the Genotype MTBDRplus in diagnosing MDR-TB due to the high specificity value (100%) obtained. Based on our data of 94% sensitivity and an estimated 630 new and previously treated cases of MDR-TB annually in Cameroon, the MTBDRplus assay would fail annually to diagnose 32 cases. These misdiagnosed cases would compromise the goal to identify every patient with MDR-TB in Cameroon. However, among the 3 MDR-TB cases not diagnosed as resistant to RIF and INH by the MTBDRplus assay, 2 were diagnosed as resistant to RIF and 1 as resistant to INH. As such, even if the MTBDRplus assay were to misdiagnose a few MDR-TB cases, the presence of resistance to either INH or RIF among these cases would be detected. Overall, our data suggests that the Genotype MTBDRplus assay can be used to diagnose MDR-TB in Cameroon.
The majority 44/48 (92%) of RIF-resistant isolates detected by the MTBDRplus assay had mutations at codon 531 41/48 (85%) and 526 (6%). These mutations are known to be the most prevalent RIF-resistance associated mutations [
17‐
19], but their frequencies vary worldwide. The frequency of codon 531 mutation in our study was 85% which is higher than generally reported [
20]. A similar high frequency of codon 531 mutation among RIF-resistant isolates in Cameroon was also reported [
18]. As such, the codon 531 mutation may actually be the most prevalent RIF-resistant associated mutation among RIF-resistant isolates in Cameroon. The relevance of a predominant codon 531 mutations is unclear but could reflect on-going transmission of isolates carrying this mutation. RIF-resistant associated mutation at positions 516 was not detected in this study, but have been reported in other studies in Cameroon [
18,
21].
In this study, four RIF-resistant isolates failed to hybridize with one or two of the wild type (WT) probes and did not hybridize with any of the probes representing known mutations. These results suggest that there is either a technical problem or a new previously unreported mutation. The WT probes with no hybridization were mostly WT2, WT3, WT4 and WT8. First, Seifert and colleagues (2016) suggest that this type of result is likely due to the failure of the mutant to hybridize with the mutation probe and is not due to the presence of a rare or new mutation. They concluded that in such situations, improved optimization of the MTBDRplus will demonstrate hybridization to the mutation probes [
22]. However, the absence of hybridization could also indicate that there is a mutation at positions 511 (WT2), 516 (WT3), 526 (WT4), and/or 533 (WT8). Mutations at these positions have been reported, but not all mutations have been associated with RIF-resistance [
23‐
26]. In the current study, the four isolates that failed to hybridize to one or two WT probes were RIF-resistant in the MGIT DST, making it likely that unknown mutations associated with RIF resistance could be present. Unfortunately, DNA sequencing could not be done in our study to confirm or identify the mutations; however, since they were resistant by MGIT DST, further studies are warranted.
Besides identifying resistance to RIF or INH, the MTBDRplus assay provides information that is necessary for patient treatment and understanding the evolution of drug resistance. Mycobacterial isolates with the
KatG codon 315 mutation have reduced ability to activate INH, but the isolate still has catalase and peroxidase activities. As such, isolates with this mutation can persist and be transmitted without any negative selection [
27]. Administration of a high-dose of INH (900 mg) per day to patients harboring isolates with the
inhA promoter mutations, might lead to better treatment outcome [
2,
28‐
30]. Isolates with
inhA promoter are also resistant to the anti-TB drug Ethionamide (ETH). As such, use of ETH will not be beneficial to the patient [
31]. A recent study suggests that routine evaluation of the frequency of the
inhA promoter mutation can help predict progression to more severe forms of drug resistance. The authors of the study observed an increase in the frequency of
inhA promoter mutations as isolates progressed to more severe forms of drug resistance, from MDR-TB to pre-extremely drug resistant TB (pre-XDR-TB) and XDR-TB [
30,
32].
At present, the WHO recommends that the MTBDRplus assay be used directly on specimen without culturing if the specimen is smear positive and on Mtb isolates obtained after culture [
33]. However, among smear-positive specimens, a recent study showed that the MTBDRplus assay will perform best if specimens were graded ≥AFB2+ [
34]. In our study, 71% (192/270) of samples were ≥AFB2+. Gauthier and colleagues (2014) have proposed a diagnostic algorithm where the MTBDRplus assay should be used directly on ≥AFB2+ specimens [
35], which is appealing because it helps accelerate diagnosis of drug resistance. In reality, a significant proportion of patients will be <AFB2+ and will not be eligible for direct testing according to the algorithm proposed by Gauthier and colleagues (2014). In our study, we had 29% (78/270) of such patients. As such, use of the MTBDRplus assay for direct testing might not be adequate. Furthermore, most laboratories are now equipped with the GeneXpert MTB/RIF assay (Cepheid, Sunnyvale, CA, USA) [
11], that performs better than the MTBDRplus for direct testing of clinical specimens [
36]. However, the MTBDRplus assay offers great benefit when used on cultured isolates. First, results are available within 48 h in contrast to the 7 days to weeks with culture-based DST. Secondly, the MTBDRplus assay is not affected by the presence of contamination with bacteria, fungi or non-tuberculosis mycobacteria as are most cultured-based DST systems [
37,
38]. In our study, 29% (70/239) of the cultures were contaminated and had to be decontaminated and re-cultured before DST could be done. Among the 70 repeats, we were successful in obtaining pure cultures in only 57 cultures; whereas the other 13 remained contaminated and culture-based DST could not be performed. The presence of contaminants increases laboratory processing and reporting times that could be avoided if the MTBDRplus assay is used.
Acknowledgements
We thank the study participants for their time, interest and willingness to participate in the study. We also thank the staff at each clinical site for their exceptional efforts to enroll participants, conduct interviews and collect samples. We also acknowledge the support of the staff of Centre Pasteur du Cameroon; in particular the entire staff of the Mycobacteriology Unit of CPC, as well as Yannick Kamdem, Felix Fotso, Dr. Valerie Donkeng and Dr. Abby Collier (Faculty of Pharmaceutical Sciences, University of British Columbia, Canada).
We are also grateful to the organizers of the Wellcome Trust workshop ‘Molecular Approaches to Clinical Microbiology in Africa’ for providing training to NNA on the Genotype MTBDRplus assay.