Emergence of Fluoroquinolone Resistance in Mycobacterium tuberculosis during Continuously Dosed Moxifloxacin Monotherapy in a Mouse Model

Presently, fluoroquinolones are under study for first-line treatment of tuberculosis (9, 11). However, fluoroquinolone resistance among Mycobacterium tuberculosis strains is emerging, with important implications for treatment (1, 2, 3). To determine whether selection of fluoroquinolone-resistant mutants would occur in the murine experimental model of tuberculosis as in humans, we treated infected mice with moxifloxacin (MXF) mixed into the diet. Such a procedure permits the large differences in MXF exposure between the two species to be overcome because the half-life of MXF after oral administration is about 1.5 h in mice whereas that of MXF after oral administration in humans is 9 to 12 h (8, 9, 12). In addition, we sequenced the quinolone resistance-determining regions (QRDRs) of gyrA and gyrB in MXF-resistant isolates.

A culture of M. tuberculosis H37Rv prepared in Middlebrook 7H9-oleic acid-albumin-dextrose-catalase (OADC) was allowed to grow to the end of log phase. The MIC of MXF was determined by the agar dilution method by plating appropriate dilutions of 10⁵ organisms/ml broth culture suspension on 7H10-OADC agar plates without and with MXF (0.06 to 8 μg/ml). The MIC of MXF was 0.25 μg/ml.

After the concentration by centrifugation of an M. tuberculosis H37Rv broth culture, 1 ml (4.35 × 10⁸ CFU) was plated on 150 × 15 cm² 7H10-OADC agar plates without and with MXF (0.5 to 8 μg/ml). Four weeks later, an innumerable amount of CFU (>1 × 10⁻⁶) and 296 (6.8 × 10⁻⁷), 126 (2.9 × 10⁻⁷), 18 (4.1 × 10⁻⁸), and 0 (<10⁻⁸) CFU were recovered, respectively (prevalences of MXF-resistant CFU are shown in parentheses). Five single colonies were picked from 1 to 4 μg/ml MXF-containing plates, placed in 5 ml Middlebrook 7H9-OADC broth, and allowed to grow to stationary phase for sequencing of the QRDRs of gyrA and gyrB, the genes responsible for DNA gyrase-mediated fluoroquinolone resistance in M. tuberculosis. Genomic DNA was purified and amplified by PCR. Oligonucleotide primers for the QRDRs of gyrA and gyrB were used to amplify a 320-bp region of gyrA and a 375-bp region of gyrB (7, 14). The oligonucleotides for gyrA were 5′-CAGCTACATCGACTATGCGA-3′ and 5′-GGGCTTCGGTGTACCTCAT-3′, and those for gyrB were 5′-CCACCGACATCGGTGGATT-3′ and 5′-CTGCCACTTGAGTTTGTACA-3′. PCR products were purified and sequenced.

Colonies that grew on MXF-containing plates uniformly revealed single gyrA mutations, although two also exhibited single novel gyrB mutations (Table 1). For the colonies that grew at lower MXF concentrations (1 to 2 μg/ml), the missense mutation appeared in codon 94; for those that grew at 4 μg/ml, a single mutation could also be found in codon 88.

Seventy-two 6-week-old female BALB/c mice were infected by the aerosol route with an M. tuberculosis H37Rv broth culture, concentrated by centrifugation, that implanted 1.1 × 10⁵ ± 2.1 × 10⁴ (95% confidence interval) CFU, i.e., one log₁₀ above the expected load. Two weeks later, prior to the initiation of treatment for 56 days, mice were moribund and the lung CFU counts of the six sacrificed mice were 5.5 × 10⁷, 2.9 × 10⁸, 6.9 × 10⁸, 1.1 × 10⁹, 3.0 × 10⁹, and 9.6 × 10⁹. Mice were then randomized to one of seven treatment groups, with 18 mice in the control group and 9 mice each included in groups receiving diets with the following six concentrations of MXF: 0.03%, 0.06%, 0.12%, 0.25%, 0.5%, and 1%. Due to its bitter taste, MXF was mixed with powdered sugar in a 1:10 ratio and added to the powdered food. Based on preliminary data obtained in our laboratory correlating MXF in the diet with serum levels in mice (10), it was expected that these concentrations would continuously provide MXF levels of 0.25, 0.5, 1.0, 2.0, 4.0, and 8.0 μg/ml, respectively, in serum.

Both remaining food and surviving mice were weighed every other day to monitor mean ingestion of the diet and increase in body weight. All mice had decreased diet intakes and weight loss (Table 2). All untreated control mice and mice in the lower-MXF-concentration groups (0.03 and 0.06%) died within the first 12 days of treatment, most within the first week. Only 10 mice survived after day 12 in the groups receiving MXF concentrations of 0.12% to 1.0%. Surviving mice ate 3 to 5 g diet/day. At day 56, serum from mice treated with 0.12% (one mouse), 0.25% (four mice), 0.5% (three mice), and 1.0% (two mice) MXF in the diet had concentrations of 2.4, 4.1, 5.3, and 17.9 μg/ml, respectively, in serum. These were higher than expected by approximately double. The lungs of the single surviving mouse in the 0.12% concentration group had 6.7 × 10⁴ CFU. The lung CFU counts of the four mice treated with a 0.25% concentration were 4.8 × 10³, 6.9 × 10³, 4.3 × 10⁴, and 5.0 × 10⁴ (mean, 2.6 × 10⁴). The CFU counts in the lungs of the three mice in the 0.5% concentration group were 2.5 × 10², 5.0 × 10², and 1.9 × 10³ (mean, 8.8 × 10²), and those in the two mice in the 1% concentration group were 1.0 × 10² and 2.0 × 10² (mean, 1.5 × 10²). These data suggest an increased reduction of the CFU/lung values in relation to the increase of MXF concentration in the diet (Table 3).

Up to five single colonies were picked from each MXF-containing plate, placed in 5 ml Middlebrook 7H9-OADC broth, and allowed to grow to stationary phase for PCR amplification and sequencing of the QRDRs of gyrA and gyrB. All of the tested isolates from mice fed on MXF-containing diets yielded MXF-resistant mutants with mutations in QRDRs of gyrA. Sequencing revealed single mutations in codons 90, 91, and 94 of gyrA (Table 4). No double mutations were detected.

Although the collective mouse and food weights were monitored closely throughout the experiment, there was only a one-time collective measurement of MXF in the serum per group taken on completion of the experiment. Thus, we have no direct measurement of the serum levels of MXF achieved during the entire course of the experiment. While mice eventually manifested very high MXF concentrations in blood, poor diet consumption early in the course of therapy likely resulted in lower MXF concentrations at the crucial therapeutic stage when the organism load and the risk of selecting resistant mutants were highest. Therefore, the present experiment indirectly demonstrates the importance of achieving the maximal antibacterial effect early in the course of therapy to reduce the risk of resistance. Finally, one might consider that the poor diet consumption of the mice was analogous to poor adherence in humans, so that the present study also inadvertently demonstrates the adverse impact of nonadherence on the promotion of fluoroquinolone resistance. This is cause for concern in the current context of increasing numbers of fluoroquinolone prescriptions worldwide (1, 3, 4, 5, 6, 13).

• American Society for Microbiology