Performance of the GenoType MTBDRplus Assay Directly on Sputum Specimens from Brazilian Patients with Tuberculosis Treatment Failure or Relapse

Abstract

Rapid identification of drug resistance in clinical isolates of Mycobacterium tuberculosis is important in determining treatment for tuberculosis. The aim of this work was evaluate the performance of the GenoType MDRTBplus assay directly on sputum of patients who had treatment failure or relapse in a routine outpatient setting in southern Brazil.

TEXT

Isoniazid (INH) and rifampin (RIF) are the main first-line drugs used for tuberculosis (TB) treatment. The rapid identification of drug resistance, particularly of multidrug-resistant Mycobacterium tuberculosis (MDR-TB) clinical isolates (defined as strains resistant to at least INH and RIF) is an important challenge to developing rapid and adequate antituberculosis chemotherapy and preventing the spread of resistant strains. The GenoType MTBDRplus assay has been validated in several countries (1, 2, 3, 4) but not yet in Brazil. In the present study, we compared the performance of GenoType MDRTBplus to that of DNA sequencing and conventional drug susceptibility testing (DST) for detecting drug-resistant strains of M. tuberculosis directly in sputum specimens from patients who had treatment failure or relapse in a routine setting between August 2009 and June 2011. We enrolled patients >18 years old who were treated at a reference hospital for tuberculosis, in Porto Alegre, the capital of Rio Grande do Sul State in southern Brazil. The incidence of the disease in this city is 112 cases per 100,000 people (5). Culture by Petroff's method and identification were performed according to WHO recommendations (6). After culture, DST was performed using the proportion method on Lowenstein-Jensen medium with the critical concentrations 0.2 μg/ml INH, 40 μg/ml RIF, 4 μg/ml streptomycin (SM), and 2 μg/ml ethambutol (EMB) (7), with a critical proportion of 1% indicating that an isolate is resistant. Samples were analyzed by DNA sequencing of the key regions involved in the development of resistance (rpoB, katG, and inhA) (8, 9). After pretreatment using Petroff's method, the DNA extraction (directly from sputum), PCR amplification, strip hybridization, and interpretation of profiles of MTBDRplus were performed according to the manufacturer's instructions. Statistical analysis was performed using SPSS software (SPSS Inc., Chicago, IL). We determined the κ scores to assess concordance, and they were interpreted as follows: <0.45 = poor; 0.45 to 0.70 = fair; and >0.70 = excellent. Statistical significance was defined as a P value of <0.05.

During the period analyzed, we collected 112 sputum specimens; 68 grew on solid media, and all cultures belonged to the M. tuberculosis complex (niacin positive and paranitrobenzoic acid [PNB] negative). Of the 68 isolates analyzed by DST, 33 (48.5%) were MDR (RIF/INH, n = 29; RIF/INH/SM, n = 4), 18 (26.4%) were monoresistant (RIF, n = 1; INH, n = 16; SM, n = 1), 5 (7.3%) were polyresistant (INH/SM), and 12 (17.6%) were sensitive to all four drugs. All 68 positive cultures were sequenced, and 32 (47%) had mutations in the rpoB gene. Of these 32 isolates, 19 (59.4%) had the mutation S531L, the most common mutation, consistent with the results of other studies (1, 10), 6 (18.8%) had a 12-nucleotide (CCAGAACAACCC) insertion between codons 516 and 517 (a rare insertion, but in this study the second most frequent), 5 (15.6%) had a mutation in codon 526 (H526D, n = 1; H526Y, n = 4), and 2 (6.2%) had a double mutation (D516H and H526D). Of 32 samples (47%) with mutations in the katG gene, 26 (81.2%) had S315T, 5 (15.6%) had S315N, and one (3.1%) had S315R. The last two mutations are rare and were described in other studies (8, 10, 11, 12). Three samples had a mutation in inhA (−15C→T). Based on sequencing of the 68 samples tested, 23 (33.8%) were classified as MDR (Tables 1 and 2). Table 2 shows the results of DST, MTBDRplus, and sequencing of 62 clinical strains; on one occasion, the MTBDRplus results for 4 (6%) samples could not be interpreted. Another two samples were excluded since they were smear negative (Table 1); according to manufacturer's instructions MTBDRplus allows resistance identification of smear-positive samples.

Table 1.

Strains that have no interpretable patterns in GenoType TBMDRplus assay (n = 6)

Sputum AFB grade (n)a	DST result		Sequence			GenoType MTBDRplusb
	RMP	INH	RpoB	KatG	InhA
+ (1)	R	R	H526Y (CAC-TAC)	S315R (AGC-AGA)	WT	Bands were too faint to be interpreted
+ (2), Neg (1)	R	R	ins in codon 516 CCAGAACAACCC	S315T (AGC-ACC)	WT	Bands were too faint to be interpreted
Neg (1)	R	R	S531L (TCG-TTG)	S315T (AGC-ACC)	WT	Lacked the TUB band
+++ (1)	R	R	S531L (TCG-TTG)	WT	WT	Lacked the TUB band

^aNeg, 0 acid-fast bacilli (AFB)/100 fields; +, 10 to 99 AFB/100 fields; +++, >10 AFB/20 fields.

^bThe TUB band indicates that the sample belongs to the M. tuberculosis complex.

Table 2.

Patterns of results of GenoType MTBDRplus in comparison to DST and sequencing (n = 62)

Agreement	No. of samples	DST resulta		Sequencing result			GenoType MTBDRplus resulta,b
		RMP	INH	RpoB	KatG	InhA	RpoB	KatG	InhA
Total agreement	13c	S	S	WT	WT	WT	WT	WT	WT
	11	R	R	S531L (TCG-TTG)	S315T (AGC-ACC)	WT	Δwt8 + Mut3	Mut1e	WT
	8	S	R	WT	S315T (AGC-ACC)	WT	WT	Mutd	WT
	3	R	R	H526Y (CAC-TAC)	S315N (AGC-AAC)	WT	Δwt7 + Mut2Ae	ΔkatGwt	WT
	1	R	S	S531L (TCG-TTG)	WT	WT	Δwt8 + Mut3	WT	WT
	1	S	R	WT	S315N (AGC-AAC)	−15C-T	WT	ΔkatGwt	Mut1e
	1	R	R	Insertion in codon 516 (CCAGAACAACCC)	S315T (AGC-ACC)	WT	Δwt3	Mut1e	WT
	1	R	R	S531L (TCG-TTG)	WT	−15C-T	Δwt8 + Mut3e	WT	Mut1e
	1	S	R	WT	WT	−15C-T	WT	WT	Mut1e
Some disagreement	1	R*	R	WT	S315N (AGC-AAC)	WT	WT	ΔkatGwt	WT
	2	R	R	Insertion in codon 516 (CCAGAACAACCC)	S315T (AGC-ACC)	WT	WT*	Mut1e	WT
	9	S	R*	WT	WT	WT	WT	WT	WT
	1	S	R*	WT	WT	WT	Δwt8 + Mut3	WT	WT
	1	S*	R*	S531L (TCG-TTG)	WT	WT	Δwt8 + Mut3e	WT	WT
	1	R	R*	H526D (CAC-GAC)	WT	WT	Δwt7 + Mut2Be	WT	WT
	3	R	R*	S531L (TCG-TTG)	WT	WT	Δwt8 + Mut3e	WT	WT
	2	R	R*	D516H (GAC-CAC) + H526D (CAC-GAC)	WT	WT	Δwt3, Δwt4, Δwt7, and Mut2Be	WT	WT
	2	R*	R*	WT	WT	WT	WT	WT	WT

^a*, discordant result.

^bΔ, absence of hybridization signal with the wild-type probes.

^cOne sample was monoresistant to streptomycin.

^dTwo strains showed a wild-type and mutation probe-positive hybridization pattern together.

^ePositive hybridization result with the specified mutant probe.

The diagnostic performance of MTBDRplus compared to DST and sequencing is summarized in Table 3. The results of sensitivity and specificity of the MTBDRplus assay in our study (Table 3) were somewhat lower than those in other recent studies, in which the sensitivity of the MTBDRplus assay for detecting RIF and INH resistance ranged from 92% to 99% and from 73% to 92%, respectively (13, 14, 15). We observed in our study that the MTBDRplus assay is more sensitive for detecting RIF resistance than INH resistance (82.0% versus 60.0% using the conventional DST) (Table 3), and the same was observed in some other studies (10, 16, 17). This could be explained by the fact that, for RIF resistance, the mutations in the rpoB gene occur in a well-defined region, while for INH resistance, it is well known that about 10% to 25% of INH-resistant strains are thought to have mutations outside katG and inhA loci (18). Efflux systems may also play a role in the development of INH resistance (19, 20). Five strains classified as RIF resistant by DST were classified as RIF susceptible using the MTBDRplus assay. The sequencing revealed that three of these strains were rpoB wild type, and the other two contained an insertion in codons 516 to 517. This insertion has the same sequence as codons 517 to 520 and may have caused hybridization of the corresponding wild-type probe (wt3). For the three wild-type samples, it is possible that the mutations were located in regions other than the rpoB hot spot. In two other samples, the opposite occurred, as they were classified as RIF resistant by the MTBDRplus assay, rpoB Δwt8 + Mut3 (S531L), but were classified as RIF susceptible by conventional DST. Sequencing revealed the S531L mutation in one strain but no mutations in the other. This phenomenon may be due to incorrect DST assessment or the presence of strains exhibiting heteroresistance (4, 21). Sequencing revealed that 19 strains classified as INH susceptible by the MTBDRplus assay but INH resistant by DST all had wild-type katG and inhA genes. This may have occurred because of the presence of mutations in other genes associated with INH resistance or efflux systems (18, 19, 22).

Table 3.

Diagnostic performance of GenoType MTBDRplus in comparison to DST and sequencing^a

Test	Sensitivity	Specificity	PPV	NPV	Kappa
DST
RMP resistance	23/28 (82)	32/34 (94)	23/25 (95.8)	32/37 (86.4)	0.77
INH resistance	29/48 (60)	14/14 (100)	29/29 (100)	14/33 (42)	0.41
MDR-TBb	16/27 (59.2)	35/35 (100)	16/16 (100)	35/46 (76.0)	0.62
Sequencing
RMP resistance	24/26 (92.3)	35/36 (97.2)	24/25 (96)	35/37 (94.6)	0.90
INH resistance	29/29 (100)	33/33 (100)	33/33 (100)	29/29 (100)	1.00
MDR-TBb	16/18 (88.9)	44/44 (100)	16/16 (100)	44/46 (95.6)	0.92

^aPPV, positive predictive value; NPV, negative predictive value. Values are number of infected patients/total number of patients (%).

^bResistant to both INH and RMP.

As the MTBDRplus assay shows 100% specificity and positive predictive value for detecting MDR-TB and good sensitivity for detecting RIF resistance directly from sputum, we conclude that this assay will allow prompt identification and treatment of MDR-TB before the results of conventional DST are available. Because there is still some discordance between conventional and molecular DST methods, we recommend that the MTBDRplus assay should be used to provide initial guidance on therapy. Due to the small number of samples, further studies are needed to evaluate the impact of introducing the MTBDRplus assay on treatment outcomes and the feasibility and cost associated with its widespread implementation.