Evaluation of Etest for Susceptibility Testing of Mycobacterium tuberculosis

Abstract

The Etest method for susceptibility testing of Mycobacterium tuberculosis was compared to the agar proportion method using four first-line agents and two fluoroquinolones. Catergorical agreement between the methods was 100% for rifampin, ethambutol, streptomycin, and ofloxacin and 98% for isoniazid. Results were obtained in 6 to 10 days by Etest. The Etest method is suitable for testing the agents evaluated against M. tuberculosis.

There has been a resurgence of tuberculosis in both developed and developing countries since the early 1980s (2, 5, 8, 18). However, 95% of tuberculosis morbidity and mortality occurs in developing nations (5, 8). Together with the highly prevalent human immunodeficiency virus, tuberculosis has become a serious public health problem in such countries (4, 5, 8, 14). The emergence and spread of multidrug-resistant tuberculosis, associated with dismal clinical outcome, have been demonstrated (1, 3, 6, 11, 17). With inadequate tuberculosis control programs characterized by a limited control of drug prescriptions, episodes of poor drug supplies, low compliance rates, and errors in drug prescriptions, the selection and dissemination of drug-resistant strains remain potential global threats (5, 14; Uganda National Tuberculosis and Leprosy Programme, 1992 annual report). A cost-effective drug susceptibility method is required to guide tuberculosis treatment and evaluate the performance of control programs.

The current recommended methods for Mycobacterium tuberculosis susceptibility testing, the agar proportion and BACTEC broth methods (10), use relative growth at a single critical drug concentration, arbitrarily set at 1%, for determination of drug resistance (9). These methods are either very cumbersome or too expensive for many countries. The Etest was developed for susceptibility testing of M. tuberculosis by Wanger and Mills (15, 16), and recent preliminary data have shown that its results correlate well with those of the agar proportion and BACTEC radiometric methods (12, 13, 15, 16). Although expensive, the Etest method will be valuable for developing countries, as it is relatively easy to perform, provides susceptibility results in about a week, and does not require expensive instruments and media.

To further evaluate the Etest method for susceptibility testing of M. tuberculosis, we compared the Etest with the agar proportion method using four first-line antituberculous agents and two fluoroquinolones and both susceptible and resistant strains.

A total of 59 isolates, maintained as stock cultures in glycerol at −70°C at University Hospitals of Cleveland, Cleveland, Ohio, were used in this study. We included the quality-control strains M. tuberculosis H37RV, which is susceptible to all first-line drugs; ATCC 35838, which is resistant to rifampin; ATCC 35837, which is resistant to ethambutol; ATCC 35822, which is resistant to isoniazid; ATCC 35820, which is resistant to streptomycin; and BCG 104, which is resistant to all first-line drugs and quinolones. The strains were subcultured on Lowenstein-Jensen slants (Remel, Lenexa, Kans.), and their species was reconfirmed on the basis of results of standard biochemical tests (9) and NAP susceptibility (S. H. Siddiqi, BACTEC TB System, Product and Procedure Manual, Becton Dickinson Diagnostic Instrument Systems, Sparks, Md., 1995).

All strains were tested for isoniazid, rifampin, streptomycin, ethambutol, ofloxacin, and ciprofloxacin susceptibility using an indirect-proportion method on antibiotic-incorporated 7H10 medium at the critical concentrations shown in Table 1 (9). All the strains were also tested by the Etest method as described by Wanger and Mills (15, 16). Briefly, 3- to 4-week-old colonies of M. tuberculosis from Lowenstein-Jensen agar slants were suspended in 3 ml of Middlebrook 7H9 broth in tubes containing glass beads. The tubes were then vigorously vortexed for 3 to 5 min, large particles were allowed to settle for about 30 min, and the supernatant was adjusted to a turbidity equivalent to a no. 3.0 McFarland standard for use as the inoculum. Using cotton applicators (Hardwood Products Company, Guilford, Maine), the entire surface of 90-mm-diameter plates of freshly prepared Middlebrook 7H11 agar supplemented with OADC (Difco Laboratories, Detroit, Mich.) was swabbed with the inoculum for confluent growth. The plates were incubated at 37°C in 5 to 10% CO₂ for 24 h. Etest strips (AB BIODISK, Solna, Sweden), containing antibiotic gradients of 0.016 to 256 μg/ml for isoniazid, streptomycin, and ethambutol and 0.002 to 32 μg/ml for rifampin, ofloxacin, and ciprofloxacin, were then placed on each plate and reincubated as described above for 21 days. When the growth became visible and ellipses of inhibition were seen, the MICs were read off at the point where the ellipses intersected the Etest strips according to the Etest technical guidelines (Susceptibility Testing of Mycobacteria. Etest technical guide no. 6, AB BIODISK, N.A., Inc., Piscataway, N.J., 1998). Etest plates were also read at 21 days.

TABLE 1.

Critical concentrations used for the agar proportion method on Middlebrook 7H10 agar

Agent	Critical concn (μg/ml)
Isoniazid	0.2
Rifampin	1.0
Streptomycin	2.0
Ethambutol	5.0
Ofloxacin	2.0
Ciprofloxacin	2.0

All quality-control strains produced expected results by both methods. Comparative results of drug susceptibility testing of test isolates are shown in Table 2. Results were able to be read at 6 to 10 days by Etest and 21 days by agar incorporation. Prolonged incubation to 21 days did not alter the Etest results. Using the critical concentrations recommended for the agar proportion method as the breakpoint MICs for the Etest method (9) (Table 1), there was 100% categorical agreement between the Etest and agar proportion methods for rifampin, ethambutol, streptomycin, and ofloxacin and 98% for isoniazid. The one discrepant strain was resistant to isoniazid by agar proportion (> 1% growth at 0.2 μg/ml) but susceptible by Etest, with an MIC of 0.1 μg/ml; however, the MIC for this isolate was considerably higher than those for the other isoniazid-susceptible isolates, which were all < 0.016 μg/ml.

TABLE 2.

Comparison of susceptible and resistant M. tuberculosis strains by the Etest and agar proportion methods

Agent	No. of isolates with indicated result by:				% Agreement
	Agar proportion method		Etest method
	Sa	Rb	S	R
Isoniazid	48	11	49	10	98
Rifampin	50	9	50	9	100
Ethambutol	50	9	50	9	100
Streptomycin	48	11	48	11	100
Ofloxacin	49	10	49	10	100
Ciprofloxacin	49	10	NAc	NA	NA

^aS, susceptible. 

^bR, resistant. 

^cNA, not applicable. 

Ciprofloxacin results by the agar incorporation method were the same as for ofloxacin. However, the Etest susceptibility results for ciprofloxacin could not be read in many instances, as the drug gradient appeared to fail after a few days and irregular growth up to the Etest strip occurred.

The MICs at which 50% of the isolates were inhibited (MIC₅₀s) and MIC₉₀s for susceptible and resistant strains are shown in Table 3. Bimodal MIC distributions were found with all agents tested, with most MICs being at the upper or lower limits of the Etest ranges. The susceptible isolates had markedly low MICs, with MIC₅₀s and MIC₉₀s being nearly equivalent.

TABLE 3.

MIC₅₀s and MIC₉₀s for susceptible and resistant strains by the Etest method

Agent	Susceptible strains			Resistant strains
	No. (%)	MIC50 (μg/ml)	MIC90 (μg/ml)	No. (%)	MIC50 (μg/ml)	MIC90 (μg/ml)
Isoniazid	47 (80.0)	<0.016	<0.016	12 (20.0)	8	>256
Streptomycin	48 (81.4)	<0.016	<0.016	11 (18.6)	>256	>256
Rifampin	50 (84.7)	<0.002	<0.002	9 (15.3)	>32	>32
Ethambutol	50 (84.7)	<0.016	<0.047	9 (15.3)	>256	>256
Ofloxacin	49 (83.0)	<0.002	<0.002	10 (17.0)	8	>32

Due to the worldwide increase in the incidence of tuberculosis, complicated by the AIDS pandemic, the development and spread of multidrug-resistant strains, and the existence of a dynamic world population, a rapid, accurate, and cost-effective susceptibility testing method for M. tuberculosis is needed. Although the overall time required for isolation and susceptibility testing is shortest with the BACTEC system, this method uses radioactive material. The nonradiometric systems in which detection of growth in liquid media is based on measurement of either consumption of oxygen or release of CO₂, such as the BACTEC 9000MB, BACTEC MGIT 960, MB/Bact, and ESP, are being evaluated for isolation and susceptibility testing of M. tuberculosis. These fully automated systems, as well as the BACTEC-460, are currently too expensive for less developed countries, where the prevalence of tuberculosis is high. Etest is currently not used to test the susceptibility of M. tuberculosis to pyrazinamide, as this agent is active only when it is deaminated into pyrazinoic acid by pyrazinamidase produced by M. tuberculosis. Pyrazinoic acid is active at an acidic pH (5.5), while M. tuberculosis grows poorly at this pH. However, efforts are being made to develop new agar media for susceptibility testing of M. tuberculosis to pyrazinamide (7).

This study demonstrated excellent agreement between the Etest and the conventional agar proportion method, with results being available in 6 to 10 days by Etest as opposed to 21 days by the agar proportion method. A similar study by Wanger and Mills found good correlations between Etest and agar proportion of 100, 94, 94, 93, 90, and 82% for rifampin, streptomycin, ciprofloxacin, isoniazid, ethambutol, and ofloxacin, respectively, and excellent agreement between BACTEC and Etest of 100% for isoniazid and rifampin, 98% for streptomycin, and 97% for ethambutol (15, 16).

The Etest method therefore yields susceptibility results that are essentially equivalent to those with BACTEC and in the same time span. In comparison to the other methods, Etest may be more cost-effective for developing countries. In addition, by providing a wide range of precise MICs, the Etest method provides quantitative results, which can be used to define susceptibility breakpoints for the older antituberculous drugs in current use as well as to establish breakpoints for newer agents (16).