Direct Drug Susceptibility Testing of Mycobacterium tuberculosis for Rapid Detection of Multidrug Resistance Using the Bactec MGIT 960 System: a Multicenter Study

Abstract

Conventional indirect drug susceptibility testing of Mycobacterium tuberculosis with liquid medium is well established and offers time-saving and reliable results. This multicenter study was carried out to evaluate if drug susceptibility testing (DST) can be successfully carried out directly from processed smear-positive specimens (direct DST) and if this approach could offer substantial time savings. Sputum specimens were digested, decontaminated, and concentrated by the laboratory routine procedure and were inoculated in Bactec MGIT 960 as well as Lowenstein-Jensen (LJ) medium for primary isolation. All the processed specimens which were acid-fast bacterium (AFB) smear positive were used for setting up direct DST for isoniazid (INH) and rifampin (RIF). After the antimicrobial mixture of polymyxin B, amphotericin B, nalidixic acid, trimethoprim, and azlocillin (PANTA) was added, the tubes were entered in the MGIT 960 instrument using the 21-day protocol (Bactec 960 pyrazinamide [PZA] protocol). Results obtained by direct DST were compared with those obtained by indirect DST to establish accuracy and time savings by this approach. Of a total of 360 AFB smear-positive sputum specimens set up for direct DST at four sites in three different countries, 307 (85%) specimens yielded reportable results. Average reporting time for direct DST was 11 days (range, 10 to 12 days). The average time savings by direct DST compared to indirect DST, which included time to isolate a culture and perform DST, was 8 days (range, 6 to 9 days). When results of direct DST were compared with those of indirect DST, there was 95.1% concordance with INH and 96.1% with rifampin. These findings indicate that direct DST with the Bactec MGIT 960 system offers further time savings and is a quick method to reliably detect multidrug resistance (MDR) cases.

INTRODUCTION

According to the WHO, drug resistance in tuberculosis (TB) is a global problem (30). Resistance against isoniazid (INH) and rifampin (RIF), defined as multidrug resistance (MDR), is increasing in many countries (3, 4, 31). If these cases are not treated properly, they can develop resistance to other drugs as well, such as fluoroquinolones and injectable aminoglycosides, defined as extensive drug resistant (XDR), and may in turn infect others with a drug-resistant strain (5, 10, 22). For a better management of drug-resistant cases, early detection of resistance is extremely important so that effective treatment can be prescribed. Rapid drug susceptibility testing plays an important role in the detection and control of MDR/XDR TB (23, 31).

Drug susceptibility testing (DST) of Mycobacterium tuberculosis is generally carried out after a culture is isolated from a clinical specimen. This takes a long time, first to isolate a culture and then to perform drug susceptibility testing (indirect DST). If DST could be set up at the same time as when a processed specimen is inoculated in solid and or liquid medium (direct DST), it could save significant time for the detection of drug resistance.

Direct DST in the conventional solid medium has been well established (9, 14, 15). The only disadvantage is that it takes a long time to obtain results on solid medium, as the growth rate on such media is lower. With the introduction of Bactec 12B liquid medium (Becton Dickinson Diagnostic Systems, Sparks, MD), the time to report results was significantly reduced (13, 19, 24, 25). In 1993, the CDC recommended to use liquid medium based on its better performance and earlier results (27). Direct DST with the Bactec 460 liquid system has been tried successfully (16). However, the use of the Bactec 460 radiometric method is phasing out due to the concerns of the radioactive waste disposal. This system is being replaced by the nonradiometric Bactec MGIT 960 system (Becton Dickinson Diagnostic Systems, Sparks, MD). Indirect DST is well established in this liquid system (1, 2, 6–8, 12, 20). In 2001, a small study using the manual BBL MGIT system was reported on direct DST for INH and RIF with excellent results (11). In 2007, WHO published a policy statement on the recommendation for the use of liquid medium for low- and middle-income countries (32).

This multicenter study was designed to establish direct DST feasibility in four different clinical laboratories with different patient populations and test parameters. The primary objective of this study was to establish a standard protocol for the direct DST using the Bactec MGIT 960 automated system. Results were compared with the indirect DST results to establish accuracy as well as time savings with this approach.

MATERIALS AND METHODS

Study sites.

This study was carried out at four different sites in three countries: (i) site 1, P.D. Hinduja National Hospital and Medical Research Center, a tertiary care hospital, Mumbai, India; (ii) site 2, LRS Institute of Tuberculosis and Lung Diseases, a dedicated TB hospital and national reference laboratory (NRL), New Delhi, India; (iii) site 3, Indus Hospital, a general charity hospital focused on TB, especially MDR management, Karachi, Pakistan; (iv) site 4, National Reference Center for Mycobacteria, national center (with access to clinical samples from areas of high MDR endemicity in other countries), Borstel, Germany.

Specimens.

Only sputum specimens from patients strongly suspected of tuberculosis as well as those from old chronic cases, especially those expected to have MDR TB, were included in this study. Specimens which were found to be smear positive for acid-fast bacteria (AFB) irrespective of the degree of smear positivity were included in the study. Sputum specimens were transported to the laboratory with minimum delay and were refrigerated if the processing was not done immediately.

Specimen processing.

All specimens were processed following the standard NALC-NaOH method for digestion, decontamination, and concentration (14, 26). The concentrated sediment was resuspended in about 2 to 3 ml phosphate buffer (pH 6.8) and mixed thoroughly. A smear was prepared for acid-fast staining, and culture media were inoculated according to the laboratory standard procedure for primary isolation. It was ensured that a little more than 1 ml of suspension was left for the direct DST.

AFB smears.

All the smears were stained with the Ziehl-Neelsen and/or fluorochrome methods. Smears were graded following WHO guidelines (28) and based on the number of AFB found during examination. The smears were graded as scanty (1 to 9 AFB/100 fields), 1+ (10 to 99/100 fields), 2+ (1 to 10 AFB/field), or 3+ (more than 10 AFB/field). Two laboratories did not have the scanty category in their grading system (these smears were included in the 1+ category).

Inoculation of culture media for primary isolation.

All four sites used the Bactec MGIT 960 system for liquid, and for the solid medium they used one Lowenstein-Jensen (LJ) slant. These media were inoculated following the established individual laboratory standard operating procedure (SOP). For MGIT medium, standard recommended procedures were followed (manufacturer's recommendations and the MGIT manual by FIND [26]). After preparing smears and inoculations for culture, the remainder of specimen was refrigerated immediately at 2°C to 8°C and was used for setting up direct DST as soon as the smear results were available.

Direct DST procedure.

There were three major differences in the direct DST procedure compared to the standard indirect DST procedure for MGIT 960: (i) direct DST was a 4- to 21-day protocol, while indirect DST was a 4- to 13-day protocol; (ii) the control was diluted 1:10 in direct DST, while in indirect DST it was diluted 1:100; (iii) in the direct DST, an antimicrobial mixture of polymyxin B, amphotericin B, nalidixic acid, trimethoprim, and azlocillin (PANTA) (Becton Dickinson Diagnostic Systems, Sparks, MD) was added to the control as well as in the drug-containing MGIT tubes to suppress contamination.

All the media and other reagents were the same as those used in the routine indirect DST: MGIT medium (7-ml bar-coded MGIT tubes), SIRE supplement for DST, and the lyophilized Bactec MGIT INH and RIF drugs (Becton Dickinson Diagnostic Systems, Sparks, MD).

Prior to setting up the direct DST, lyophilized PANTA was reconstituted using 15 ml of SIRE supplement (not growth supplement) (Becton Dickinson Diagnostic Systems, Sparks, MD) and mixed well until completely dissolved. Drug vials (lyophilized drugs, same as those used in MGIT indirect DST) of INH and RIF were reconstituted with 4 ml of sterile deionized (DI) water and mixed well. Sets of four MGIT tubes were prepared per specimen for performing the direct DST. Two tubes were labeled “growth control” (GC), one for INH and the other for RIF. The third tube was labeled “INH” and the fourth was labeled “RIF.”

Once dissolved, 0.8 ml of PANTA-SIRE supplement mixture was added into each of the four labeled MGIT tubes. The next step was the addition of drugs. In the INH-labeled tube, 0.1 ml of reconstituted lyophilized INH drug was added (0.1 μg/ml final concentration). Similarly 0.1 ml of reconstituted RIF was added to the RIF-labeled tube (1.0 μg/ml). These concentrations were the same as those in the indirect DST procedure (package insert [26]). After mixing the medium, 0.5 ml of the well-mixed reconstituted sediment was inoculated in each of the two drug-containing tubes. For the control, the resuspended sediment was diluted 1:10 by adding 0.2 ml of the well-mixed sediment into 1.8 ml of sterile saline or water. After being thoroughly mixed, 0.5 ml was inoculated into each of the two GC tubes. The tubes were mixed again by inverting several times.

For direct DST, an extended protocol which is used routinely for indirect PZA DST setup was followed since direct DST requires a 21-day protocol to complete the test (18, 26). Two two-tube DST set carriers were used. Growth control and INH tubes were placed in one set carrier (GC and INH), and growth control and RIF tubes were placed in the other set carrier (GC and RIF). These set carriers were entered in the instrument as the PZA test. The first tube in the set carrier was always the control tube.

At one site, another approach of handling the 21-day protocol was followed. In this second option for direct DST, the procedures were the same except that only one GC was used. The GC and INH tubes were placed in one two-tube carrier and entered as the PZA DST. The RIF tube was entered as a regular growth tube (42-day protocol) and was placed close to the INH set carrier.

Safety precautions.

All four sites have well-equipped BSL III laboratory facilities. Standard safety precautions were followed for specimen processing, inoculation, and DST (29).

Interpretation of direct drug susceptibility testing.

When the GC reached the growth unit (GU) value of 400 or more, the instrument indicated that the test was complete, the susceptibility set was removed after scanning, and an inventory report was printed. Susceptibility results for both INH and RIF (in the first option) were interpreted by the instrument as “S” or “R.” At the time the GU value of the GC was 400 or more and if the GU value of the drug tube was less than 100, the test result was reported as “susceptible,” while if the GU value of the drug tube was 100 or more the result was interpreted as “resistant.” The GU values of both the DST sets were retrieved and recorded. In case the GU value of the control did not reach 400 within 21 days, the instrument indicated an X200 error, indicating insufficient growth. On the other hand, if the GU reached 400 earlier than day 4, the instrument gave an X400 error, indicating contamination or overinoculation.

In the second option, the GC and INH tubes were taken out once the instrument indicated that the test was complete and results were retrieved. The GU values of the INH set were retrieved. Since the RIF tube was incubated separately in the system as a regular growth tube, the instrument would not interpret the results. At this point, the GU value of the RIF tube was also retrieved by printing an instrument inventory report. The results were interpreted manually following the criteria given above.

For calculation of time to obtain direct DST results, the time it took for the instrument to complete the DST test was recorded. Since the instrument gives time in hours, any time equal to half a day or more was taken as a full day.

Reference method: indirect drug susceptibility testing.

The primary isolation tubes of the specimens that were included in the study were followed. Once an inoculated specimen was culture positive in MGIT and confirmed to have pure culture of M. tuberculosis, the indirect DST was set up following the manufacturer's recommended procedures for the MGIT 960 system. Results of indirect DST and time to complete the test were retrieved from the instrument and recorded.

For calculation of time required for reportable indirect DST results for a specimen, time to get a positive culture plus time to achieve indirect DST results was added to get the total time. Since indirect DST is set up after 1 to 5 days after the instrument gives a positive signal, additionally an average of 2 days was added to the total time. Occasionally, the DST was not set up within 5 days of culture positivity in MGIT and a subculturing was required. This time has not been documented. The time required for indirect DST was compared with that of direct DST of the same specimen. The difference was considered the time savings.

At site 4, routine indirect DST was also carried out on LJ medium, but results have not been included in this analysis.

Identification of the isolated mycobacteria.

The MGIT 960 DST method is recommended for M. tuberculosis, and therefore the isolated mycobacteria from the culture-positive specimens were identified using the routine method used in the laboratories to identify M. tuberculosis. Only those cultures confirmed as M. tuberculosis were included in this study. Direct DST results on those specimens which were contaminated or where a mycobacterium was other than M. tuberculosis were excluded.

Discrepancy testing.

The specimens that showed discrepant results between the direct and indirect methods were retested by repeating the indirect method. If the second testing results were the same as the first indirect results, the direct susceptibility results were considered discrepant. If the repeat testing results disagreed with the previous indirect results and were in agreement with the direct results, then the results were considered the correct one.

QC testing.

M. tuberculosis H37Rv strain (ATCC 27294) was used for quality control (QC) testing in DST. This strain was introduced at each time when a batch of DST was set up or as every 6th isolate in a run. If any resistance in the QC strain was observed, all the other results in that batch were considered invalid.

RESULTS

Of the total 360 specimens processed, 307 (85%) DST results were reportable (Table 1). The majority of those which were not reportable were those where the control did not reach the required threshold (X200 error), while some were contaminated (X400 error). Site 1 processed 126 sputum specimens and reported 113 results (90%), site 2 processed 122 specimens and reported 103 results (84%), site 3 processed 74 specimens and reported 58 results (84%), while site 4 had 33 reportable results out of 38 specimens processed (87%). Specimens which were negative for culture, were contaminated, or had nontuberculous mycobacteria (NTM) were excluded to calculate reportable results. There were a total of 15 specimens which had contamination, 8 were identified as NTM, 19 had the X200 error, and 5 had the X400 error.

Table 1.

Overall summary of testing

Sitea	No. of tests set up	No. of reportable results
1	126	113
2	122	103
3	74	58
4	38	33
Total	360	307 (85%)

^aSites: 1, PDH, India; 2, LRS, India; 3, Indus, Pakistan; 4, TB reference lab, Germany.

The time to report results of positive cultures was analyzed according to the degree of AFB smear positivity (Table 2). The majority of specimens were 2 to 3+ smear positive at all the sites. The time to detection of culture positive was not significantly different between different smear grades. Overall, cultures were positive at an average of 8 days, with a range of 8 to 10 days.

Table 2.

Time to detect MGIT-positive cultures

Smear grading by site	No. of specimens turned positive after days:				Total no. of specimens turned positive	Avg time to detect (no. of days)
	3–7	8–14	15–21	>21
Site 1
Scanty	4	6	1	0	11	9
1+	14	7	3	0	24	9
2+	14	14	1	0	29	8
3+	45	3	1	0	49	6
Total	77	30	6	0	113	8
Site 2
Scanty	0	0	0	0	0	0
1+	10	18	4	0	32	10
2+	17	11	2	0	30	8
3+	35	6	0	0	41	6
Total	62	35	6	0	103	8
Site 3
Scanty	0	0	0	0	0	0
1+	8	15	3	0	26	10
2+	4	1	1	0	6	8
3+	21	3	1	1	26	6
Total	33	19	5	1	58	8
Site 4
Scanty	1	4	1	0	6	12
1+	1	6	0	1	8	12
2+	2	3	1	0	6	11
3+	10	3	0	0	13	7
Total	14	16	2	1	33	10

The time to complete direct DST from processed specimens was calculated according to the smear-positive categories (Table 3). There was no significant difference between times to complete direct DST results in different smear-positive categories. The majority of results were ready within 8 to 14 days, with an overall average of 10 days at site 1, 11 days at sites 2 and 3, and 12 days at site 4.

Table 3.

Time to report direct drug susceptibility results from processed specimens

Smear (no. of specimens)	No. of tests after days:			Avg time to detect (no. of days)
	3–7	8–14	15–21
Scanty (11)	1	7	3	12
1+ (24)	0	20	4	12
2+ (29)	2	21	6	11
3+ (49)	18	31	0	8
Avg time to detect of total				10
Site 2
Scanty	0	0	0	0
1+ (32)	3	21	8	13
2+ (30)	3	23	4	11
3+ (41)	17	21	3	9
Avg time to detect of total				11
Site 3
Scanty	0	0	0	0
1+ (26)	3	13	10	13
2+ (6)	2	4	0	9
3+ (26)	8	16	2	10
Avg time to detect of total				11
Site 4
Scanty (6)	0	4	2	15
1+ (8)	0	3	5	14
2+ (6)	0	6	0	11
3+ (13)	1	11	1	10
Avg time to detect of total				12

In Table 4, the average time to complete indirect DST from isolated cultures is reported without taking into consideration the time to culture positivity. The average time to report indirect DST from isolated culture ranged from 6 days (site 4) to 10 days (site 2).

Table 4.

Time to report results by indirect susceptibility testing from isolated cultures^a

Site	No. of tests with reportable results after days:		Avg time to report (no. of days)
	4–7	8–13
1	27	86	8
2	16	87	10
3	18	40	9
4	30	3	6

^aTime to report indirect DST is the total time required from the day of setting up DST to the day results were ready. It does not include the time to get a positive culture from processed specimen, 1 to 5 additional days for setting up DST from the day the instrument gives a positive signal, plus in some cases additional time required for subculturing if needed.

The total time required from the time the specimen was processed to the time when indirect DST results were available is included in Table 6. It ranged from 18 to 20 days.

Table 6.

Time savings by direct susceptibility testing from processed specimens

Site	Avg time to report (days)		Time savings (no. of days)
	Indirecta	Direct
1	18	10	8
2	20	11	9
3	19	11	8
4	18	12	6
All sites	19	11	8

^aTime to report indirect DST is the total time from inoculation of a processed specimen into MGIT and the time when indirect DST results were available. This includes additional time, averaging 2 days for setting up DST from the day an MGIT culture was instrument positive. In some cases, additional time was required for subculturing if needed, which has not been documented.

Discrepant results were analyzed on all direct DST tests on which confirmed indirect DST results were available (Table 5). Of the 113 specimens reported by site 1, there were 5 specimens that showed discrepant results between direct and indirect methods for INH (4.4%): three results were reported as false resistant and two results as false susceptible. For RIF, there were three discrepant results (2.7%), two being false susceptible and one being false resistant to RIF. At site 2, out of 103 specimens, three showed discrepant results for INH (2.9%), all false susceptible. Also for RIF, there were a total of three discrepant results (2.9%): two false resistant and one false susceptible. Site 3 had six discrepant results out of 58 tests for INH (10.3%), four being false resistant and two false susceptible. For RIF, there were five discrepant results (8.6%), two being false resistant and three false susceptible. At site 4, out of 33 tests there was only one (3.0%) false resistant to INH and one (3.0%) false susceptible to RIF. Thus, there were overall 4.9% discrepant results for INH and 3.9% for RIF among all the sites.

Table 5.

Discrepant results between direct and indirect DST methods

Site (total no. of tests)	No. (%) of specimens
	INH			RIF
	False S	False R	Total	False S	False R	Total
1 (113)	2 (1.8)	3 (2.7)		2 (1.8)	1 (0.9)
2 (103)	3 (2.9)	0		1 (1)	2 (1.9)
3 (58)	2 (3.4)	4 (7)		3 (5.2)	2 (3.4)
4 (33)	0	1 (3)		1 (3)	0
Total (307)	7 (2.3)	8 (2.6)	15 (4.9)	7 (2.5)	5 (1.6)	12 (3.9)

Findings of time savings by direct DST have been given in Table 6. Overall time saving was 8 days, ranging from 6 to 9 days among all the sites.

For quality control, H37Rv was set up along with direct and indirect DST batches. There was not a single incidence at any site where H37 Rv failed to give expected results.

DISCUSSION

Results obtained with the conventional indirect susceptibility testing methods, especially with solid media, become available too late to influence a timely decision on patient management. Hence, more rapid TB susceptibility tests directly applied on clinical specimens are needed. Some of the noncommercially available direct tests include the nitrate reductase assay (NRA) and microscopic observation drug susceptibility (MODS) assay. These tests have been developed as “in house” assays with the aim to overcome the high costs of the commercially available techniques. In NRA, the addition of the NRA reagent requires tubes to be regularly opened, which poses significant risk of aerosol generation. Reading of the MODS plates has to be performed on a daily basis and is laborious and time-consuming (34). Commercially available molecular assays, such as Genotype MTBDR Plus (Hain Lifescience, Nehren, Germany), can be applied directly to smear-positive specimens and have less turnaround time, thus saving several weeks. However, none of the established molecular tests target all possible genes involved in resistance, and thus a variable proportion of resistant strains may not be detected (17, 33). Liquid culture has been established as a gold standard and is most rapid for phenotypic DST. This was a research study designed to establish time savings with the direct DST approach compared to the routine indirect approach in liquid medium, and thus molecular testing was not included in the study.

Liquid culture offers a more sensitive and rapid method for isolation of M. tuberculosis and performing susceptibility testing against a variety of first-line and second-line antituberculosis drugs (13, 21). However, most of the DST studies have been carried out by the routine indirect method using isolated cultures. This is the first large-scale multicenter study to evaluate direct susceptibility testing of M. tuberculosis from clinical specimens using the MGIT 960 automated liquid culture system.

Since MDR tuberculosis is one of the main concerns in a TB control program and DST results play an important role in the control of TB, we focused on direct DST on only two drugs, INH and RIF. This does not mean that this approach is applicable to only these two antimicrobials. It is anticipated that this study could provide a guideline for a rapid broth-based direct DST for other first-line and second-line anti-TB drugs, which will help especially if a patient is suspected to have resistance to those drugs.

One of the main concerns was how to carry out direct DST. It was established earlier that for the direct DST using liquid medium, the length of the test protocol should be extended from 14 days to 21 days, since M. tuberculosis present in the clinical specimen does not grow as rapidly as it does from an isolated culture (16). In the proposed protocol, the direct DST is set up from an AFB smear-positive specimen with any degree of smear positivity by any staining procedure; thus, in some cases, the bacterial count present in the inoculum is low and may take longer than 14 days to reach the required growth level or GU. The MGIT 960 indirect DST procedure is designed with a 4- to 13-day protocol, except for PZA DST, which is a 4- to 21-day test. However, the PZA test is designed for only a two-tube system. We tried several different workflow approaches to achieve results with the 21-day protocol. The first approach was to set up each drug with its own control and then enter in the instrument in a two-tube set carrier as a PZA test. This is a simple, straightforward, and preferred procedure, and the instrument interprets results automatically. However, some investigators thought that this approach was costly, because a control is needed for each drug, meaning requirement of more MGIT tubes. To cut down the cost, we came up with another approach, where the first drug, in this case growth control and INH, is entered into the instrument with the PZA protocol. The other drug tube, namely, RIF, is entered in the instrument as a growth and not a DST tube. The tube is placed in the same drawer close to the INH DST tubes. Once the INH set is ready (GU of control reaches 400 or more), the instrument flags it as complete and interprets results as “S” or “R.” At that time, the GU values of the RIF tube are retrieved by asking the inventory GU values of the incubated tubes in the instrument, and then the GU value of the RIF tube is recorded. This is done by printing the inventory report without scanning the tube out of the instrument prior to printing, as the GU values may be lost. In case it is difficult to locate a single tube in a drawer, there is another possibility of putting this RIF tube in a two-tube DST set carrier, placing first a blank uninoculated MGIT medium tube as a growth control and then an RIF tube. This set carrier is entered in the instrument as the PZA DST and is placed close to the INH set. Since the control is not inoculated, this set carrier should be taken out at the same time as the INH set carrier. The GU values are retrieved and recorded before scanning out this second carrier set; otherwise, the GU values will be lost. Interpretation of RIF DST is done manually based on the formula given in the earlier section. The uninoculated growth control tube may be used again and again. The above-described procedure may be followed for any and as many drugs as needed.

This study was carried out in well-established laboratories. The culture positivity rate of the smear-positive specimen was very high (above 95%), with acceptable contamination rates (4 to 8%) and very low prevalence of NTM. The overall success rate of DST from smear-positive specimens was 85%. That means that only about 10 to 15% of total DST setups were uninterpretable, due to several reasons, such as contamination or presence of NTM (X400 errors) or no growth or not enough growth in the control (X200 errors). A few, though AFB smear positive (even 3+), either failed to grow in the primary isolation tube and were culture negative or the growth was not enough to interpret direct DST results (X200 error). The information on the success rate of direct DST is important to evaluate the cost-effectiveness of the direct DST approach.

The most important aspect of our findings is the time savings by direct DST. The time to report Bactec MGIT indirect DST from positive cultures varied from 6 to 10 days, which is concordant with many earlier reports (1, 2, 7, 20). This time to report varies depending on many factors, including variability in the standard procedure followed in a laboratory, patient population, and prevalence of drug resistance. It is known that drug-resistant isolates tend to take a longer time to grow than the drug-susceptible ones. The total time for report of indirect DST was calculated as the time to isolate a culture, the time required to set up DST, and the time to get results of the susceptibility test from positive culture. On the other hand, the time to report direct DST was the time to achieve DST results after inoculation of a processed specimen. The time savings by direct DST was overall 8 days and did not vary too much from site to site. This time savings is significantly important, as every day counts in an MDR case. The results indicate that direct DST further reduces the time to report susceptibility results significantly. Following direct DST, on average, sites 1 and 4 reported direct DST results after 2 days of culture-positive results, and sites 2 and 3 reported after three days of culture-positive results. These sites handle a large number of MDR cases, and thus there was a good representation of drug-resistant cases among the total cases that we studied.

Another objective of this study was to evaluate the accuracy of results obtained by the direct method. In this calculation, indirect DST was considered the gold standard, and in case of discordant results indirect DST was repeated. Overall, among the four sites, only 4.9% of results were discordant in case of INH and 3.9% in case of RIF, with 2.3% false susceptible (very major error) for INH and 2.5% for rifampin. False resistance (major error) was 2.6% for INH and 1.6% for rifampin. One site had higher discordance than the other three sites. There was no clear pattern of discordant results, as both false resistant and false susceptible were observed with INH and RIF.

In summary, this multicenter study established a standard protocol for performing direct susceptibility with the Bactec MGIT automated system. DST results may be reported 2 to 3 days after the culture positivity results are available. This significant time savings could offer a great help in prescribing effective treatment, especially in MDR cases. If cost is a concern and the prevalence of monoresistance to RIF is not common, only RIF may be tested and could be considered a surrogate marker for MDR. Direct DST is a reliable test, as results obtained by direct DST had 95 to 96% concordance with those obtained by the indirect method.