Rifaximin disc diffusion test for in vitro susceptibility testing of Clostridium difficile

Rifaximin is a rifampicin derivative, poorly absorbed by the gastro-intestinal tract. We studied the in vitro susceptibility to rifamixin of 1082 Clostridium difficile isolates; among these,184 isolates from a strain collection were tested by an in-house rifaximin disc (40 µg) diffusion test, by an in-house rifaximin broth microdilution test, by rifampicin Etest and by rpoB gene sequencing. In the absence of respective CLSI or EUCAST MIC breakpoints for rifaximin and rifampicin against C. difficile we chose MIC ≥32 µg ml−1 as criterion for reduced in vitro susceptibility. To further validate the disc diffusion test 898 consecutive clinical isolates were analysed using the disc diffusion test, the Etest and rpoB gene sequence analysis for all resistant strains. Rifaximin broth microdilution tests of the 184 reference strains yielded rifaximin MICs ranging from 0.001 (n = 1) to ≥1024 µg ml−1 (n = 61); 62 isolates showed a reduced susceptibility (MIC ≥32 µg ml−1). All of these 62 strains showed rpoB gene mutations producing amino acid substitutions; the rifampicin- and rifaximin-susceptible strains showed either a wild-type sequence or silent amino acid substitutions (19 strains). For 11 arbitrarily chosen isolates with rifaximin MICs of >1024 µg ml−1, rifaximin end-point MICs were determined by broth dilution: 4096 µg ml−1 (n = 2), 8192 µg ml−1 (n = 6), 16 384 µg ml−1 (n = 2) and 32 678 µg ml−1 (n = 1). Rifampicin Etests on the 184 C. difficile reference strains yielded MICs ranging from ≤0.002 (n = 117) to ≥32 µg ml−1 (n = 59). Using a 38 mm inhibition zone as breakpoint for reduced susceptibility the use of rifaximin disc diffusion yielded 59 results correlating with those obtained by use of rifaximin broth microdilution in 98.4 % of the 184 strains tested. Rifampicin Etests performed on the 898 clinical isolates revealed that 67 isolates had MICs of ≥32 µg ml−1. There were no discordant results observed among these isolates with reduced susceptibility using an MIC of ≥32 µg ml−1 as breakpoint for reduced rifampicin susceptibility and a <38 mm inhibition zone as breakpoint for reduced rifaximin susceptibility. The prevalence of reduced susceptibility was 7.5 % for all isolates tested. However, for PCR ribotype 027 the prevalence of reduced susceptibility was 26 %. Susceptibility testing in the microbiology laboratory therefore could have an impact on the care and outcome of patients with infection. Our results show that rifaximin – despite its water-insolubility – may be a suitable candidate for disc diffusion testing.


INTRODUCTION
Clostridium difficile is a spore-forming Gram-positive anaerobic bacterium and a major cause of nosocomial and community-acquired diarrhoea (Wiström et al., 2001). Metronidazole and oral vancomycin are the main antibiotics used to treat C. difficile infection (CDI). In 2005, in Austria, oral rifaximin was licensed for 'treatment of gastrointestinal diseases caused or partially caused by rifaximin-susceptible bacteria, e.g. gastrointestinal infections, pseudomembranous colitis due to C. difficile, hepatic encephalopathy, small bowel bacterial overgrowth and diverticulitis' (Willerroider, 2009). However, to our knowledge, no guidelines have yet been published to test in vitro susceptibility of C. difficile to rifaximin. Two study groups previously used ¢32 mg l 21 (O'Connor et al., 2008;Jiang et al., 2010) as breakpoint for resistance using agar dilution testing, a method not widely employed in routine testing of clinical isolates. Jiang et al. (2000) reported high rifaximin concentrations (4000 to 8000 mg g 21 ) in stools 3 days after a single oral administration. Johnson et al. (2007) reported on the possible prevention of recurrence of CDI by administering rifaximin immediately after completion of the last course of vancomycin therapy of CDI.
In Austria, rifaximin is widely used to treat CDI, and microbiological laboratories often receive requests for in vitro susceptibility testing of clinical C. difficile isolates against this substance. The aim of this study was to evaluate whether the disc diffusion method can be applied to test in vitro susceptibility of C. difficile to rifaximin.

METHODS
Micro-organisms. One hundred and eighty-four C. difficile isolates were obtained from the reference strain collection of the Austrian national C. difficile reference centre, including ATCC strain 9689, and 50 strains previously provided by Leeds University Hospital (UK), 3 strains from the Public Health Laboratory Maribor (Slovenia) and 24 strains from Leiden University Hospital (Netherlands). In addition, 898 non-duplicate clinical isolates were tested, cultured in 21 Austrian medical laboratories in 2009.
Antimicrobial agents and susceptibility testing. The 184 isolates from the strain collection were tested by an in-house rifaximin disc (40 mg) diffusion test (Oxoid; custom-made product), by an in-house rifaximin broth microdilution test and by rifampicin Epsilon-test (Etest) (bioMérieux). The 898 clinical isolates from 2009 were tested only by disc diffusion test and rifampicin Etest (bioMérieux). Rifaximin discs were custom-made (Oxoid) by Gebro Pharma. The disc diffusion tests were performed on Brucella blood agar plates supplemented with 5 mg haemin l 21 and 1 mg vitamin K l 21 (Oxoid).
The rifaximin broth microdilution test was performed as follows. Rifaximin was purchased as a powder from Alfa Wassermann. Sterile stock solutions were prepared according to the instructions of CLSI for testing of anaerobes (Clinical and Laboratory Standards Institute, 2007). In short, rifaximin was dissolved in methanol and then diluted in 0.9 % saline solution. Serial twofold dilutions in ATB S-medium containing menadione (vitamin K 3 ) at 0.5 mg l 21 and haemin at 15 mg l 21 (bioMérieux) were prepared for rifaximin concentrations covering a range from 0.000125 to 1024 mg ml 21 ; 96-well microdilution plates were filled with ATB S-medium (bioMérieux) containing the respective antibiotic concentrations. The antibiotic stocks were freshly prepared on the day of testing. For inoculum preparation, test organisms were cultured for 48 h anaerobically on Columbia blood agar plates (bioMérieux) at 37 uC; bacteria were suspended in 0.9 % saline solution to yield McFarland 0.5 and diluted 1 : 10 into the medium, so that the final test concentration of bacteria was approximately 1610 6 c.f.u. ml 21 . Minimal inhibitory concentration (MIC) was defined as the lowest concentration at which no growth was observed after incubation for 48 h at 37 uC in an anaerobic atmosphere using anaerobic jars and GasPak (BD). Growth controls were performed by inoculation of antibiotic-free medium with an aliquot of the primary inoculum at a concentration of 10 5 c.f.u. per well; purity testing was performed by transferring an aliquot of 10 ml onto two Columbia blood agar plates with incubation under aerobic and anaerobic conditions, respectively.
Detection of single-nucleotide polymorphisms (SNPs) within the rpoB gene. DNA was extracted from cultures using the MagNA Pure Compact (Roche Diagnostics) according to the producer's manual to a final volume of 50 ml. Primers RifFOR (59-CAAGATA-TGGAAGCTATAAC-39) and RifREVlang (59-GTGATTCTATAAAT-CCAAATTC-39) were used in PCRs containing 25 ml HotStar Taq Master Mix (Qiagen), 5 ml (5 pmol ml 21 ) of each primer, 13 ml water and 2 ml DNA. Amplification was performed in a PCR thermocycler (15 min 96 uC, 30 cycles of 1 min 94 uC, 1 min 52 uC and 1 min 72 uC, and finally 10 min 72 uC). PCR products were cleaned up with Escherichia coli exonuclease I, and shrimp alkaline phosphatase (Fermentas) according to the manufacturer's instructions.
Sequencing PCR containing 2 ml Big-Dye-Mix (Applied Biosystems), 1 ml Sequencing Buffer (Applied Biosystems) 4 ml water, 1 ml RifFOR or RifREVlang primer (10 pmol 21 ) and 2 ml DNA was performed in a commercial PCR thermocycler (1 min 96 uC, 30 cycles of 20 s 96 uC, 20 s 50 uC and 4 min 60 uC). The amplified products were cleaned up with Centri Sep 96-well plates or Centri Sep 8 well strips (Applied Biosystems) for dye terminator clean-up according to the manufacturer's manual. Samples were analysed in an ABI 3130 genetic analyser (Applied Biosystems) with 36 cm capillary loaded with a POP7 gel (Applied Biosystems).
Sequences were analysed for the presence of SNPs within the rpoB gene using Kodon (Applied Maths) version 3.5 by aligning the samples to the rpoB gene sequence of a reference wild-type C. difficile strain (CD630; NC_009089) downloaded from the NCBI database.
Using an inhibition zone of ,38 mm as breakpoint for reduced susceptibility, the use of rifaximin disc diffusion yielded results correlating with those received by the use of rifaximin broth microdilution in 180 ( Table 3).
Correlation of rifampicin MICs obtained by broth dilution with rifaximin disc diameters was tested by Kendall's tau-b correlation coefficient: 20.42 (P,0.001 for the hypothesis that both parameters were independent).

Results for clinical isolates
Rifaximin disc (40 mg) diffusion testing performed on 898 clinical C. difficile isolates yielded inhibition zone diameters ranging from 6 mm (n567) to 78 mm (n52) (Fig.  2). Using an inhibition zone ,38 mm as breakpoint for reduced susceptibility a total of 68 strains with reduced susceptibility were identified.
Rifampicin Etests performed on the 898 isolates revealed that 67 isolates had MICs of ¢32 mg ml 21 for rifampicin and that 819 had MICs of ¡0.002 mg ml 21 ; 12 isolates (1.3 %) exhibited rifampicin MICs between these extremes. By rifaximin disc diffusion test, the 67 strains with reduced rifampicin susceptibility exhibited no inhibition zone. No discordant results were observed among these 67 isolates with reduced susceptibility using an MIC of ¢32 mg ml 21 as breakpoint for reduced rifampicin susceptibility and a ,38 mm inhibition zone as breakpoint for reduced rifaximin susceptibility. Isolate 2663/PCR ribotype 053 had a rifampicin Etest MIC of 0.064 mg ml 21 and a rifaximin inhibition zone of 36 mm; molecular analyses revealed a D492N mutation in the rpoB gene (Table 3).

DISCUSSION
All the rifamycins are semisynthetic derivatives of rifamycin B, a fermentation product of Amycolatopsis mediterranei, formerly named Streptomyces mediterranei. Rifamycin B exerts poor antimicrobial activity, but is easily produced and readily converted chemically into rifamycin S, from which most active derivatives are prepared (Parenti & Lancini, 2003). Rifaximin is a semi-synthetic derivative of rifamycin S formulated for oral administration. In the 1980s it was only available in Italy, but today it is marketed worldwide mainly for the treatment of gastrointestinal infections and the treatment of chronic hepatic encephalopathy (Corazza et al., 1988;Festi et al., 1992). Jiang et al. (2000) reported high rifaximin concentrations (4000 to 8000 mg g 21 ) in stools 3 days after single oral administration.
Although adequate antimicrobial drug concentrations should be achieved by the high dosage of the nonabsorbable rifaximin, the constraints imposed by possible drug resistance (O'Connor et al., 2008) often impel clinicians to request in vitro susceptibility testing of C. difficile isolates from patients not responding to therapy. Our results on 11 arbitrarily chosen isolates from the group with reduced susceptibility to rifaximin (MIC .1024 mg ml 21 ) showed MICs ranging from 4096 to 32 678 mg ml 21 , so that a breakpoint for resistance between .32 mg ml 21 and 1024 mg ml 21 can be considered for the future.
We evaluated a rifaximin disc diffusion test and found it substantially equivalent to an in-house rifaximin broth microdilution test. The level of performance considered acceptable for US Food and Drug Administration (FDA) clearance in premarket notification of commercial antimicrobial susceptibility systems is (among others) .89.9 % categorical agreement (same susceptible, intermediate or resistant classification), ¡1.5 % very major errors (false susceptibility based on the number of resistant organisms) and ¡3 % major errors (false resistance based on the number of susceptible isolates) (Richter & Ferraro, 2007). Accepting an MIC ¢32 mg ml 21 as criterion for reduced rifaximin susceptibility, a rifaximin inhibition zone of ,38 mm would represent a valid resistance breakpoint in the 40 mg disc diffusion test described here.
We also found good correlation between the rifaximin and rifampicin susceptibility testing results. The number of discordant results (n58) concerning supposed resistance based on the rifampicin Etest and rifaximin disc diffusion test on C. difficile isolates could be seen as an argument to introduce the criterion 'intermediate' for rifampicin susceptibility testing. If we take the EUCAST breakpoints for staphylococci (S¡0.06/R.0.5 mg ml 21 ) for the rifampicin Etest, the correlation would increase to 99.45 % (five discordant results) (EUCAST, 2010). Our suggestions for preliminary breakpoints between susceptible, indeterminate and resistant are summarized in Table 4 for the Etest and the broth microdilution method to be validated in future studies. In our opinion no reliable intermediate breakpoint result can be given for the disc test; however, all isolates in question showed rpoB mutations yielding an amino acid change.
Rifamycin resistance is commonly the result of a mutation that alters the b-subunit of RNA polymerase, reducing its binding affinity for rifamycins (Struelens, 2003). This study found rpoB mutations in all of the 62 strains showing reduced susceptibility with the rifaximin broth microdilution test. Interestingly, only H502N and D492V mutations showed discordant results between the rifampicin Etest and the rifaximin broth microdilution test in the 184 strains analysed, indicating a potential connection between mutation in the rpoB region and the efficacy of the antibiotic used. Future studies would enable examination of this potential connection.
Our molecular investigations confirmed mutations resulting in amino acid substitutions in RpoB for all eight isolates with discordant results with reduced susceptibility to rifaximin and rifampicin. Interestingly, two strains showing seven or eight silent mutations also showed reduced susceptibility with the disc diffusion test, but not with any of the other methods tested. Since no other mutation could be found in the rpoB gene a mutation in another gene influencing the activity of rifaximin has to be   Jiang et al. (2010) suggests that the use of acetone as solvent for rifaximin and rifampicin by Jiang's group is the reason for the differences.
To answer this question future studies should be undertaken. However, in our opinion, in vitro susceptibility testing of rifampicin can be used to predict resistance to rifaximin when done according to our method.
Rifaximin exhibits high activity against C. difficile in vitro, with a very low MIC 50 and a very high MIC 90 (0.032 mg ml 21 and 1024 mg ml 21 , respectively). By using our suggested breakpoints (Table 4) as criteria for in vitro resistance, the prevalence of resistance was 7.5 % for all clinical isolates tested. Among the so-called hypervirulent ribotype RT 027, the prevalence of resistance was as high as 26 %. These data suggest that the development of resistance is in some way dependent on the PCR ribotype investigated; this is in concordance with general findings that some strains can become resistant more easily than others, e.g. C. difficile PCR ribotype 027, Mycobacterium tuberculosis spoligotype Beijing.
Susceptibility testing in the microbiology laboratory therefore could have an impact on the care and outcome of patients with infection (Doern et al., 1994). The rifaximin agar dilution and broth microdilution test systems facilitate reading of MICs and are often considered as gold standards for susceptibility testing. However, susceptibility testing by the disc diffusion method has the advantages of simplicity, low cost and a high degree of flexibility in the selection of agents tested (Richter & Ferraro, 2007). Our results show that rifaximin -despite its water-insolubility -may be a suitable candidate for disc diffusion testing. Whether 40 mg per disc is the ideal concentration must be examined in further studies. Whether it was prudent to officially license a substance 'for treatment of all enteric infections caused by susceptible organisms' also remains to be answered.