G Model ANTAGE-4285; No. of Pages 5
ARTICLE IN PRESS International Journal of Antimicrobial Agents xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
International Journal of Antimicrobial Agents journal homepage: http://www.elsevier.com/locate/ijantimicag
Short Communication
Species-level assessment of the molecular basis of fluoroquinolone resistance among viridans group streptococci causing bacteraemia in cancer patients ˜ a , Luis Velazquez a , Miguel Saldana ˜ a, Pranoti Sahasrabhojane a , Jessica Galloway-Pena a b a,c,∗ Nicola Horstmann , Jeffrey Tarrand , Samuel A. Shelburne a
Department of Infectious Diseases, MD Anderson Cancer Center, Houston, TX 77030, USA Department of Laboratory Medicine, MD Anderson Cancer Center, Houston, TX 77030, USA c Department of Genomic Medicine, MD Anderson Cancer Center, Houston, TX 77030, USA b
a r t i c l e
i n f o
Article history: Received 11 November 2013 Accepted 30 January 2014 Keywords: Streptococcus spp. Fluoroquinolone Molecular epidemiology Bacteraemia Neutropenia
a b s t r a c t Viridans group streptococci (VGS) are a major cause of bacteraemia in neutropenic cancer patients, particularly those receiving fluoroquinolone prophylaxis. In this study, we sought to understand the molecular basis for fluoroquinolone resistance in VGS causing bacteraemia in cancer patients by assigning 115 VGS bloodstream isolates to specific species using multilocus sequence analysis (MLSA), by sequencing the quinolone resistance-determining regions (QRDRs) of gyrA, gyrB, parC and parE, and by testing strain susceptibility to various fluoroquinolones. Non-susceptibility to one or more fluoroquinolones was observed for 78% of isolates, however only 68.7% of patients were receiving fluoroquinolone prophylaxis. All but one of the determinative QRDR polymorphisms occurred in GyrA or ParC, yet the pattern of determinative QRDR polymorphisms was significantly associated with the fluoroquinolone prophylaxis received. By combining MLSA and QRDR data, multiple patients infected with genetically indistinguishable fluoroquinolone-resistant Streptococcus mitis or Streptococcus oralis strains were discovered. Together these data delineate the molecular mechanisms of fluoroquinolone resistance in VGS isolates causing bacteraemia and suggest possible transmission of fluoroquinolone-resistant S. mitis and S. oralis isolates among cancer patients. © 2014 Elsevier B.V. and the International Society of Chemotherapy. All rights reserved.
1. Introduction Infection is a leading cause of morbidity and mortality in neutropenic cancer patients, who are often given prophylactic antimicrobials in an attempt to decrease infection rates. Fluoroquinolones are the major class of antibiotics used for infection prophylaxis, particularly levofloxacin and ciprofloxacin [1]. However, widespread use of fluoroquinolone prophylaxis has led to increasing emergence of antimicrobial-resistant organisms [1]. Viridans group streptococci (VGS) are a large category of genetically
∗ Corresponding author at: Department of Infectious Diseases and Department of Genomic Medicine, MD Anderson Cancer Center, Unit 1460, 1515 Holcombe Boulevard, Houston, TX 77030, USA. Tel.: +1 713 792 3629; fax: +1 713 792 5381. E-mail addresses: [email protected], [email protected] (S.A. Shelburne).
heterogeneous streptococci that often cause serious infections in cancer patients, in part because of their propensity to develop fluoroquinolone resistance [1]. Although VGS encompass at least 15 distinct species capable of causing disease in humans, commonly used phenotypic and genotypic techniques often fail to accurately classify VGS strains to species level [2]. Multilocus sequence analysis (MLSA) was recently developed and is now considered a new ‘gold standard’ for VGS species identification as it has been the most successful approach to date at resolving species clusters for this taxonomically challenging group of organisms [2]. Previous studies of fluoroquinolone resistance in VGS have employed methodologies known to be problematic for proper VGS species identification [3,4]. Therefore, data that include accurate species identification when considering the specific genetic mechanisms underlying fluoroquinolone resistance and its epidemiology in particular VGS species are needed. In order to address this, we
http://dx.doi.org/10.1016/j.ijantimicag.2014.01.031 0924-8579/© 2014 Elsevier B.V. and the International Society of Chemotherapy. All rights reserved.
Please cite this article in press as: Sahasrabhojane P, et al. Species-level assessment of the molecular basis of fluoroquinolone resistance among viridans group streptococci causing bacteraemia in cancer patients. Int J Antimicrob Agents (2014), http://dx.doi.org/10.1016/j.ijantimicag.2014.01.031
G Model
ARTICLE IN PRESS
ANTAGE-4285; No. of Pages 5
P. Sahasrabhojane et al. / International Journal of Antimicrobial Agents xxx (2014) xxx–xxx
2
% of isolates non-susceptible to indicated fluoroquinolone
Ciprofloxacin
Levofloxacin
Moxifloxacin
100 80 60 40 20
S. S. m i in S tis fa ( nt . o ra N= Sa is ng /au lis 67) (N st ui r Sa nis ali =22 s ) liv gr ( ar ou N = iu p 4 s ( gr N = ) ou 16 ) To p ( ta N = l( 4) N = 11 S. 5) S. m i in S. tis fa n or (N Sa tis = a ng /au lis 67) (N st ui r Sa nis ali =22 liv gr s (N ) ar ou = iu p 4 s ( gr N = ) ou 16 ) To p ( ta N = l( 4) N = 11 S. 5) S. m i in S. tis fa n or ( N Sa tis = a ng /au lis 67) (N st ui r Sa nis ali =22 liv gr s (N ) ar ou = iu p 4 s ( gr N = ) ou 16 ) To p ( ta N = l( 4) N = 11 5)
0
VGS Species/Groups Fig. 1. Fluoroquinolone susceptibilities among the different viridans group streptococci (VGS) species causing bacteraemia in cancer patients. The percentage of strains that are non-susceptible to ciprofloxacin, levofloxacin and moxifloxacin is designated by species, with some VGS isolates placed into groups for ease of depiction. The number of isolates per species/group is indicated at the bottom of the graph. The two Anginosus group strains, which were both fully susceptible to all tested fluoroquinolones, are not included for space purposes.
took advantage of a large sample of VGS bacteraemia isolates and patient-specific clinical data and combined these resources with the improved speciation technique of MLSA. 2. Materials and methods 2.1. Bacterial strains, culture conditions, minimum inhibitory concentrations (MICs) and species determination VGS bloodstream isolates were collected from unique patients at MD Anderson Cancer Center hospital (Houston, TX) from 15 April 2011 to 31 December 2012. Clinical and microbiological data, including receipt of fluoroquinolones, were abstracted from electronic medical records using a standardised data collection form. Initial identification of VGS strains was performed using standard microbiological techniques. Strains were grown in Todd–Hewitt broth with 0.2% yeast extract (Becton, Dickinson, and Co., Sparks, MD) at 37 ◦ C with 5% CO2 . The MIC of various fluoroquinolones was determined by Etest (bioMérieux, Marcy l’Étoile, France). MICs were interpreted according to Clinical and Laboratory Standards Institute (CLSI) criteria as follows: for ciprofloxacin and levofloxacin, sensitive ≤2 g/mL, intermediate 4 g/mL and resistant ≥8 g/mL; and for moxifloxacin, sensitive ≤1 g/mL, intermediate 2 g/mL and resistant ≥4 g/mL. VGS species determination and phylogenetic analyses were performed using MLSA as previously described [2]. 2.2. Sequencing of gyrA, gyrB, parC and parE quinolone resistance-determining regions (QRDRs) Amplification and sequencing of gyrA, gyrB, parC and parE loci in each of the VGS strains was done using previously published primers and protocols [3]. Delineation of the QRDRs corresponded to the following amino acid residues from Escherichia coli: 46–172 for GyrA; 371–512 for GyrB; 35–157 for ParC; and 398–483 for ParE [5]. QRDRs were compared with the consensus sequence for
Streptococcus pneumoniae (National Center for Biotechnology Information) given that a consensus sequence, or susceptible allele, has not been established for each VGS species. Polymorphisms in the VGS QRDRs were considered determinative if the polymorphisms had been previously described as contributing to fluoroquinolone resistance, or if the polymorphism was present in non-susceptible VGS strains with no other polymorphisms known to contribute to fluoroquinolone resistance [1,3,6–8]. Polymorphisms not meeting these criteria were considered non-determinative. 2.3. Statistical analysis The relationship between QRDR polymorphisms and type of fluoroquinolone prophylaxis was determined using 2 analysis. The statistical significance of the differences between the rates of fluoroquinolone non-susceptibility among distinct VGS species/groups was analysed by Fisher’s exact test. All tests of significance were two-sided, and statistical significance was defined at P ≤ 0.05. Statistical analysis was performed using IBM SPSS Statistics for Windows v.19.0 (IBM Corp., Armonk, NY). 3. Results and discussion 3.1. Clinical characteristics of infected patients and fluoroquinolone susceptibility of viridans group streptococci isolates During the study period, 115 unique patients with VGS bacteraemia were identified; 93 (80.9%) were neutropenic and 79 (68.7%) were receiving fluoroquinolone prophylaxis at the time of infection onset. Of the 79 patients receiving fluoroquinolone prophylaxis, 57 (72%) were receiving levofloxacin, 17 (22%) ciprofloxacin and 5 (6%) moxifloxacin (Supplementary Table S1). Rates of VGS non-susceptibility to each of the tested fluoroquinolones were as follows: ciprofloxacin, 78% [MIC50 (MIC required to inhibit 50% of the isolates) ≥32 g/mL; MIC90 (MIC
Please cite this article in press as: Sahasrabhojane P, et al. Species-level assessment of the molecular basis of fluoroquinolone resistance among viridans group streptococci causing bacteraemia in cancer patients. Int J Antimicrob Agents (2014), http://dx.doi.org/10.1016/j.ijantimicag.2014.01.031
G Model ANTAGE-4285; No. of Pages 5
ARTICLE IN PRESS P. Sahasrabhojane et al. / International Journal of Antimicrobial Agents xxx (2014) xxx–xxx
3.3. The majority of fluoroquinolone-resistant viridans group streptococci strains have determinative polymorphisms in GyrA and ParC Since there are limited data available regarding the genetic mechanisms of fluoroquinolone resistance among invasive VGS [1], we next sequenced the QRDRs of gyrA, gyrB, parC and parE in all 115 VGS bacteraemia isolates. First we attempted to define the susceptible allele consensus sequence for each VGS species by determining the gene sequence in VGS strains that were fully susceptible to all of the tested fluoroquinolones. Polymorphisms in these strains (compared with S. pneumoniae) were considered non-determinative. When aligned to the QRDRs of S. pneumoniae, non-determinative polymorphisms were observed in all four genes/proteins that were distinct for individual VGS species (Supplementary Figs. S1 and S2). Supplementary Fig. S2 related to this article can be found, in the online version, at http://dx.doi.org/10.1016/j.ijantimicag. 2014.01.031. Establishing consensus (i.e. susceptible) sequences for each VGS species allowed for the assessment of determinative polymorphisms in the QRDRs among the fluoroquinolone-resistant strains. Consistent with data from E. coli and S. pneumoniae, all but one of the quinolone resistance determinative polymorphisms were located in the genes encoding either GyrA or ParC (Fig. 2), with
S81A (n = 1)
E85K (n = 2)
E85G (n = 1)
S81F (n = 42)
S81Y (n = 15)
C
ParC Determinative Mutations (n = 73)
D83G D83N S79N (n = 1) (n = 3) (n = 1) D83Y (n =2) S79C W92C (n = 1) (n = 1) S79A (n = 1) S79Y S79R (n = 20) (n = 11)
S81L (n = 9)
S79F S79I (n = 18) (n = 14)
60
Determinative polymorphism pattern Both GyrA and ParC ParC Alone GyrA Alone None
40
P < 0.05
20
in ac ox ifl ox M
C
ip
ro
vo
flo
flo
xa
xa
ci
ci
n
n
0
Le
Next we sought to use the MLSA species data to evaluate the relationship between fluoroquinolone susceptibility and VGS species. By MSLA, there were 11 distinct VGS species causing disease in this cohort (Supplementary Fig. S1). Fluoroquinolone resistance was observed for all VGS species/groups with the exception of both Anginosus group strains (Fig. 1). Streptococcus mitis and Streptococcus oralis strains were more likely to be fluoroquinolone non-susceptible compared with other VGS species (P < 0.001 by 2 analysis). This difference is likely accounted for by the fact that patients with S. mitis and S. oralis infections were more likely to be receiving fluoroquinolone prophylaxis compared with other patients (P < 0.002 by 2 analysis). Thus, we conclude that fluoroquinolone non-susceptibility is not limited to a particular VGS species or group. Supplementary Fig. S1 related to this article can be found, in the online version, at http://dx.doi.org/10.1016/j.ijantimicag. 2014.01.031.
B
GyrA Determinative Mutations (n = 70)
N on e
3.2. Quinolone resistance according to viridans group streptococci species
A
Number of isolates
required to inhibit 90% of the isolates) ≥32 g/mL]; levofloxacin, 69% (MIC50 ≥ 32 g/mL; MIC90 ≥ 32 g/mL); and moxifloxacin, 60% (MIC50 = 3 g/mL; MIC90 = 16 g/mL) (Fig. 1). As expected, patients receiving fluoroquinolone prophylaxis were significantly more likely to be infected with a VGS strain non-susceptible to one or more fluoroquinolones (74/79; 94%) compared with patients not receiving fluoroquinolone prophylaxis (16/36; 44%) (P < 0.001). However, the finding that 16 of 36 patients were infected with fluoroquinolone non-susceptible VGS without receiving prophylaxis led us to examine their records. We found that only 5 of those 16 patients had received a fluoroquinolone in the 6 months preceding their VGS bacteraemia episode. Therefore, nearly one-half of the VGS strains causing bacteraemia in patients not receiving fluoroquinolone prophylaxis were non-susceptible to one or more fluoroquinolones, suggesting that these patients may have acquired fluoroquinolone-resistant VGS strains rather than having such strains selected for by fluoroquinolone exposure. Supplementary Table S1 related to this article can be found, in the online version, at http://dx.doi.org/10.1016/j.ijantimicag. 2014.01.031.
3
Type of fluorquinolone prophylaxis Fig. 2. Determinative polymorphisms observed in this study and the relationship with the type of fluoroquinolone prophylaxis received at the onset of viridans group streptococci (VGS) bacteraemia. (A) Specific GyrA polymorphisms observed, including the number of strains with each polymorphism. (B) Specific ParC polymorphisms observed, including the number of strains with each polymorphism. (C) Depiction of the number of strains (y-axis) having either no determinative polymorphisms or determinative polymorphisms in GyrA alone, in ParC alone or in both GyrA and ParC, stratified by the type of prophylaxis being administered (x-axis). The three strains with GyrA and ParE polymorphisms were included in the much larger GyrA + ParC group given that ParC and ParE together make up topoisomerase IV. P-Values refer to 2 analysis, which showed that patients receiving ciprofloxacin prophylaxis were more likely to have ParC polymorphisms alone compared with patients receiving either levofloxacin or moxifloxacin prophylaxis.
the lone exception being the D435N polymorphism in ParE [7]. This polymorphism was observed in three strains (one S. oralis, one Streptococcus salivarius and one Streptococcus vestibularis), always in conjunction with a GyrA determinative polymorphism. Seven strains (six S. mitis and one S. oralis) contained a determinative polymorphism only in ParC and were mainly resistant to ciprofloxacin and susceptible to moxifloxacin (Supplementary Table S2). The most common pattern of determinative QRDR polymorphisms observed was alterations in both GyrA and ParC, which was associated with high-level resistance to all fluoroquinolones tested (Supplementary Table S2). The vast majority of determinative GyrA polymorphisms involved S81, with only three strains
Please cite this article in press as: Sahasrabhojane P, et al. Species-level assessment of the molecular basis of fluoroquinolone resistance among viridans group streptococci causing bacteraemia in cancer patients. Int J Antimicrob Agents (2014), http://dx.doi.org/10.1016/j.ijantimicag.2014.01.031
G Model
ARTICLE IN PRESS
ANTAGE-4285; No. of Pages 5
P. Sahasrabhojane et al. / International Journal of Antimicrobial Agents xxx (2014) xxx–xxx
4
A
B 2 2
S. mitis
1 3
1 3
4 4 0 .0 1
0 .0 1
S. oralis 5 5
= No determinative polymorphism = S79A = S79F = S79C = S79Y = S79N + D83N = S79I = D83N = S79R = W92C = D83Y = D83G
= No determinative polymorphism = S81F = S81Y = S81L = S81A = E85K
Fig. 3. Identical determinative polymorphisms of the ParC and GyrA proteins can be found in Streptococcus mitis and Streptococcus oralis strains that are closely related by multilocus sequence analysis (MLSA). Panels (A) and (B) show phylogenetic relationships among S. mitis and S. oralis strains as described for Supplementary Fig. S1. Numbers 1–5 refer to clusters of strains as determined by having 0 or 1 nucleotide difference over the 3063 bp concatenated MSLA sequence. (A) Fluoroquinolone resistancedetermining polymorphisms in ParC are shown as per the legend. Note that all strains in Clusters 2–5, including all six strains in Cluster 3, have the same ParC polymorphisms. Panel (B) is the same as panel (A) except that GyrA polymorphisms are shown.
having a polymorphism at E85 (Fig. 2A). Similarly, S79 was the main site of determinative ParC polymorphisms, with six strains having a polymorphism at D83 and one strain having a W92C polymorphism (Fig. 2B). Of note, there have been contradictory reports regarding the role of ParC K137 polymorphisms in the development of fluoroquinolone resistance in S. pneumoniae [7]. In the current data set, ParC K137 polymorphisms were ubiquitous in fluoroquinolonesusceptible VGS of the Sanguinis, Salivarius and Anginosus groups, demonstrating that ParC polymorphisms at K137 do not contribute to VGS fluoroquinolone resistance. Supplementary Table S2 related to this article can be found, in the online version, at http://dx.doi.org/10.1016/j.ijantimicag. 2014.01.031. A real-time PCR assay has been developed to detect the presence of the C-to-T substitution that results in the S81F GyrA polymorphism and the C-to-A substitution that results in the S79Y ParC polymorphism in S. pneumoniae [9]. Although these substitutions were frequently observed in the fluoroquinoloneresistant VGS strains in this study, together these two mutations accounted for less than one-half of the determinative GyrA and ParC polymorphisms in this cohort, indicating that a broader array of molecular changes will need to be targeted in order to fully capture the range of determinative polymorphisms underlying fluoroquinolone resistance in VGS (Fig. 2 and Supplementary Table S3). The nucleotide changes underlying determinative polymorphisms in GyrA and ParC are summarised in Supplementary Table S3.
Supplementary Table S3 related to this article can be found, in the online version, at http://dx.doi.org/10.1016/j.ijantimicag. 2014.01.031. 3.4. The determinative polymorphism pattern in GyrA and ParC can be linked to the type of fluoroquinolone prophylaxis being received Nearly 70% of the current cohort developed VGS bacteraemia during fluoroquinolone prophylaxis, providing an unprecedented opportunity to study the relationship between fluoroquinolone receipt and the mechanism of streptococcal fluoroquinolone resistance. In vitro studies in S. pneumoniae have established that different fluoroquinolones have a distinct predilection to select for mutations in ParC compared with GyrA, but such bias has not been demonstrated clinically in streptococci [10]. In the current data set, the presence of determinative polymorphisms in GyrA or ParC was statistically different depending on the type of fluoroquinolone prophylaxis administered (Fig. 2C). Specifically, patients receiving ciprofloxacin were more likely (P < 0.05 by 2 analysis) to have polymorphisms in ParC alone compared with patients receiving levofloxacin or moxifloxacin. Polymorphisms in both GyrA and ParC were present in all isolates from patients receiving moxifloxacin prophylaxis and in the vast majority of patients receiving levofloxacin (Fig. 2C). Two strains with ParC polymorphisms alone and three strains with both GyrA and ParC polymorphisms (Fig. 2C) were isolated from patients not receiving prophylaxis. In these five patients,
Please cite this article in press as: Sahasrabhojane P, et al. Species-level assessment of the molecular basis of fluoroquinolone resistance among viridans group streptococci causing bacteraemia in cancer patients. Int J Antimicrob Agents (2014), http://dx.doi.org/10.1016/j.ijantimicag.2014.01.031
G Model ANTAGE-4285; No. of Pages 5
ARTICLE IN PRESS P. Sahasrabhojane et al. / International Journal of Antimicrobial Agents xxx (2014) xxx–xxx
there was no record of fluoroquinolone receipt in the 180 days preceding their infection onset, suggesting that these patients may have acquired fluoroquinolone-non-susceptible VGS in the absence of direct fluoroquinolone exposure. 3.5. Combination analysis of GyrA and ParC polymorphisms and multilocus sequence analysis data in Streptococcus mitis and Streptococcus oralis strains Because the S. mitis and S. oralis strains in this study appeared to have a propensity to be fluoroquinolone non-susceptible and comprised the majority of isolates, we next sought to combine the MLSA and GyrA/ParC QRDR data for strains of these two species. Most S. mitis and S. oralis strains differed by ≥20 nucleotides from their closest neighbour over the 3063 concatenated MLSA nucleotide sequence. However, five non-epidemiologically related clusters of S. mitis and S. oralis strains were observed (Fig. 3), where isolates within each cluster were essentially genetically identical (0 or 1 bp difference, >99.9% identity) by MLSA, including one cluster involving six patients in which S. mitis strains were isolated over a greater than 1-year period (Cluster 3; Fig. 3A). For four of the five clusters (Clusters 2–5), the strains within each cluster had the same S79 ParC determinative polymorphism (Fig. 3A). The exception was Cluster 1, where one strain was fluoroquinolone-susceptible and the other strain was fluoroquinolone-resistant containing a ParC S79R polymorphism. From a GyrA standpoint, strains from two of the five clusters had the same S81 GyrA polymorphism (Fig. 3B). For Cluster 3, there were two distinct GyrA variations observed, with five strains sharing the same GyrA polymorphism (Fig. 3B). Review of patient data did not reveal an epidemiological link between patients harbouring strains from the same clusters, as such strains were isolated from patients in different hospital units and over a relatively long period of time (data not shown). This discovery was surprising because, based on limited data, it has been thought that each patient carries a mixture of genetically unique fluoroquinolone-susceptible VGS strains where at least one becomes resistant to fluoroquinolones following antimicrobial exposure [8,11]. In this scenario, one would expect to see no relationship between genetic relatedness by MLSA and ParC polymorphism unless the genetic background of the strain predisposes the strain to develop a particular ParC polymorphism. The apparently random distribution of the 11 distinct ParC polymorphisms across S. mitis and S. oralis strains (see Fig. 3) does not appear to support a genetic predisposition hypothesis. Rather, the most likely explanation for these data appears to be that a small subset of S. mitis and S. oralis strains with pre-existing polymorphisms in ParC have been acquired by distinct patients. This finding is similar to recently published findings of clonal dissemination of fluoroquinolone-resistant E. coli [12]. In S. pneumoniae, ParC polymorphisms have limited detectable impact on bacterial fitness [13]. Moreover, S. pneumoniae strains with pre-existing ParC determinative polymorphisms, while testing in the susceptible range for fluoroquinolones, can rapidly acquire GyrA determinative polymorphisms following fluoroquinolone exposure leading to fluoroquinolone treatment failure [14]. Thus, data in the current study indicate that at least some of the patients may be exogenously acquiring VGS strains with pre-existing ParC polymorphisms that then develop into fully fluoroquinolone-resistant VGS with both ParC and GyrA determinative polymorphisms. 4. Conclusion The emergence of fluoroquinolone resistance among streptococci is a major public health concern, particularly as genetic
5
exchange of resistance determinants between VGS and S. pneumoniae is known to occur [15]. Previous studies of fluoroquinolone resistance in streptococci have mainly been conducted at either the surveillance level or by studying the emergence of resistance under laboratory conditions [4,5,7]. Prior to this study, patientspecific clinical data on fluoroquinolone resistance, particularly with regard to previous fluoroquinolone exposure, were scant [1,8]. By combining molecular epidemiology, targeted gene sequencing and patient-specific clinical observations, we have generated several new insights into the epidemiology and mechanisms of VGS fluoroquinolone resistance, which lay the groundwork for more extensive investigation into fluoroquinolone non-susceptibility among streptococci in general. Funding: This work was supported by an internal grant from MD Anderson Cancer Center (Houston, TX) to SAS. Competing interests: None declared. Ethical approval: The research was conducted in accordance with the Declaration of Helsinki and was approved by the MD Anderson (MDA) Institutional Review Board [PA12-0230]. References [1] Razonable RR, Litzow MR, Khaliq Y, Piper KE, Rouse MS, Patel R. Bacteremia due to viridans group streptococci with diminished susceptibility to levofloxacin among neutropenic patients receiving levofloxacin prophylaxis. Clin Infect Dis 2002;34:1469–74. [2] Bishop CJ, Aanensen DM, Jordan GE, Kilian M, Hanage WP, Spratt BG. Assigning strains to bacterial species via the internet. BMC Biol 2009;7: 3. [3] Maeda Y, Murayama M, Goldsmith CE, Coulter WA, Mason C, Millar BC, et al. Molecular characterization and phylogenetic analysis of quinolone resistancedetermining regions (QRDRs) of gyrA, gyrB, parC and parE gene loci in viridans group streptococci isolated from adult patients with cystic fibrosis. J Antimicrob Chemother 2011;66:476–86. [4] Rodríguez-Avial I, Rodríguez-Avial C, Culebras E, Picazo JJ. Fluoroquinolone resistance among invasive viridans group streptococci and Streptococcus bovis isolated in Spain. Int J Antimicrob Agents 2007;29: 478–80. [5] Brueggemann AB, Coffman SL, Rhomberg P, Huynh H, Almer L, Nilius A, et al. Fluoroquinolone resistance in Streptococcus pneumoniae in United States since 1994–1995. Antimicrob Agents Chemother 2002;46:680–8. ˜ [6] González I, Georgiou M, Alcaide F, Balas D, Linares J, de la Campa AG. Fluoroquinolone resistance mutations in the parC, parE, and gyrA genes of clinical isolates of viridans group streptococci. Antimicrob Agents Chemother 1998;42:2792–8. [7] Patel SN, McGeer A, Melano R, Tyrrell GJ, Green K, Pillai DR, et al. Susceptibility of Streptococcus pneumoniae to fluoroquinolones in Canada. Antimicrob Agents Chemother 2011;55:3703–8. [8] Prabhu RM, Piper KE, Litzow MR, Steckelberg JM, Patel R. Emergence of quinolone resistance among viridans group streptococci isolated from the oropharynx of neutropenic peripheral blood stem cell transplant patients receiving quinolone antimicrobial prophylaxis. Eur J Clin Microbiol Infect Dis 2005;24:832–8. [9] Page S, Vernel-Pauillac F, O’Connor O, Bremont S, Charavay F, Courvalin P, et al. Real-time PCR detection of gyrA and parC mutations in Streptococcus pneumoniae. Antimicrob Agents Chemother 2008;52:4155–8. [10] Pestova E, Millichap JJ, Noskin GA, Peterson LR. Intracellular targets of moxifloxacin: a comparison with other fluoroquinolones. J Antimicrob Chemother 2000;45:583–90. [11] Wisplinghoff H, Reinert RR, Cornely O, Seifert H. Molecular relationships and antimicrobial susceptibilities of viridans group streptococci isolated from blood of neutropenic cancer patients. J Clin Microbiol 1999;37: 1876–80. [12] van Hees BC, Tersmette M, Willems RJ, de Jong B, Biesma D, van Hannen EJ. Molecular analysis of ciprofloxacin resistance and clonal relatedness of clinical Escherichia coli isolates from haematology patients receiving ciprofloxacin prophylaxis. J Antimicrob Chemother 2011;66:1739–44. [13] Rozen DE, McGee L, Levin BR, Klugman KP. Fitness costs of fluoroquinolone resistance in Streptococcus pneumoniae. Antimicrob Agents Chemother 2007;51:412–6. [14] Varon E, Houssaye S, Grondin S, Gutmann L. Nonmolecular test for detection of low-level resistance to fluoroquinolones in Streptococcus pneumoniae. Antimicrob Agents Chemother 2006;50:572–9. [15] Ip M, Chau SSL, Chi F, Tang JL, Chan PK. Fluoroquinolone resistance in atypical pneumococci and oral streptococci: evidence of horizontal gene transfer of fluoroquinolone resistance determinants from Streptococcus pneumoniae. Antimicrob Agents Chemother 2007;51:2690–700.
Please cite this article in press as: Sahasrabhojane P, et al. Species-level assessment of the molecular basis of fluoroquinolone resistance among viridans group streptococci causing bacteraemia in cancer patients. Int J Antimicrob Agents (2014), http://dx.doi.org/10.1016/j.ijantimicag.2014.01.031
ARTICLE IN PRESS International Journal of Antimicrobial Agents xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
International Journal of Antimicrobial Agents journal homepage: http://www.elsevier.com/locate/ijantimicag
Short Communication
Species-level assessment of the molecular basis of fluoroquinolone resistance among viridans group streptococci causing bacteraemia in cancer patients ˜ a , Luis Velazquez a , Miguel Saldana ˜ a, Pranoti Sahasrabhojane a , Jessica Galloway-Pena a b a,c,∗ Nicola Horstmann , Jeffrey Tarrand , Samuel A. Shelburne a
Department of Infectious Diseases, MD Anderson Cancer Center, Houston, TX 77030, USA Department of Laboratory Medicine, MD Anderson Cancer Center, Houston, TX 77030, USA c Department of Genomic Medicine, MD Anderson Cancer Center, Houston, TX 77030, USA b
a r t i c l e
i n f o
Article history: Received 11 November 2013 Accepted 30 January 2014 Keywords: Streptococcus spp. Fluoroquinolone Molecular epidemiology Bacteraemia Neutropenia
a b s t r a c t Viridans group streptococci (VGS) are a major cause of bacteraemia in neutropenic cancer patients, particularly those receiving fluoroquinolone prophylaxis. In this study, we sought to understand the molecular basis for fluoroquinolone resistance in VGS causing bacteraemia in cancer patients by assigning 115 VGS bloodstream isolates to specific species using multilocus sequence analysis (MLSA), by sequencing the quinolone resistance-determining regions (QRDRs) of gyrA, gyrB, parC and parE, and by testing strain susceptibility to various fluoroquinolones. Non-susceptibility to one or more fluoroquinolones was observed for 78% of isolates, however only 68.7% of patients were receiving fluoroquinolone prophylaxis. All but one of the determinative QRDR polymorphisms occurred in GyrA or ParC, yet the pattern of determinative QRDR polymorphisms was significantly associated with the fluoroquinolone prophylaxis received. By combining MLSA and QRDR data, multiple patients infected with genetically indistinguishable fluoroquinolone-resistant Streptococcus mitis or Streptococcus oralis strains were discovered. Together these data delineate the molecular mechanisms of fluoroquinolone resistance in VGS isolates causing bacteraemia and suggest possible transmission of fluoroquinolone-resistant S. mitis and S. oralis isolates among cancer patients. © 2014 Elsevier B.V. and the International Society of Chemotherapy. All rights reserved.
1. Introduction Infection is a leading cause of morbidity and mortality in neutropenic cancer patients, who are often given prophylactic antimicrobials in an attempt to decrease infection rates. Fluoroquinolones are the major class of antibiotics used for infection prophylaxis, particularly levofloxacin and ciprofloxacin [1]. However, widespread use of fluoroquinolone prophylaxis has led to increasing emergence of antimicrobial-resistant organisms [1]. Viridans group streptococci (VGS) are a large category of genetically
∗ Corresponding author at: Department of Infectious Diseases and Department of Genomic Medicine, MD Anderson Cancer Center, Unit 1460, 1515 Holcombe Boulevard, Houston, TX 77030, USA. Tel.: +1 713 792 3629; fax: +1 713 792 5381. E-mail addresses: [email protected], [email protected] (S.A. Shelburne).
heterogeneous streptococci that often cause serious infections in cancer patients, in part because of their propensity to develop fluoroquinolone resistance [1]. Although VGS encompass at least 15 distinct species capable of causing disease in humans, commonly used phenotypic and genotypic techniques often fail to accurately classify VGS strains to species level [2]. Multilocus sequence analysis (MLSA) was recently developed and is now considered a new ‘gold standard’ for VGS species identification as it has been the most successful approach to date at resolving species clusters for this taxonomically challenging group of organisms [2]. Previous studies of fluoroquinolone resistance in VGS have employed methodologies known to be problematic for proper VGS species identification [3,4]. Therefore, data that include accurate species identification when considering the specific genetic mechanisms underlying fluoroquinolone resistance and its epidemiology in particular VGS species are needed. In order to address this, we
http://dx.doi.org/10.1016/j.ijantimicag.2014.01.031 0924-8579/© 2014 Elsevier B.V. and the International Society of Chemotherapy. All rights reserved.
Please cite this article in press as: Sahasrabhojane P, et al. Species-level assessment of the molecular basis of fluoroquinolone resistance among viridans group streptococci causing bacteraemia in cancer patients. Int J Antimicrob Agents (2014), http://dx.doi.org/10.1016/j.ijantimicag.2014.01.031
G Model
ARTICLE IN PRESS
ANTAGE-4285; No. of Pages 5
P. Sahasrabhojane et al. / International Journal of Antimicrobial Agents xxx (2014) xxx–xxx
2
% of isolates non-susceptible to indicated fluoroquinolone
Ciprofloxacin
Levofloxacin
Moxifloxacin
100 80 60 40 20
S. S. m i in S tis fa ( nt . o ra N= Sa is ng /au lis 67) (N st ui r Sa nis ali =22 s ) liv gr ( ar ou N = iu p 4 s ( gr N = ) ou 16 ) To p ( ta N = l( 4) N = 11 S. 5) S. m i in S. tis fa n or (N Sa tis = a ng /au lis 67) (N st ui r Sa nis ali =22 liv gr s (N ) ar ou = iu p 4 s ( gr N = ) ou 16 ) To p ( ta N = l( 4) N = 11 S. 5) S. m i in S. tis fa n or ( N Sa tis = a ng /au lis 67) (N st ui r Sa nis ali =22 liv gr s (N ) ar ou = iu p 4 s ( gr N = ) ou 16 ) To p ( ta N = l( 4) N = 11 5)
0
VGS Species/Groups Fig. 1. Fluoroquinolone susceptibilities among the different viridans group streptococci (VGS) species causing bacteraemia in cancer patients. The percentage of strains that are non-susceptible to ciprofloxacin, levofloxacin and moxifloxacin is designated by species, with some VGS isolates placed into groups for ease of depiction. The number of isolates per species/group is indicated at the bottom of the graph. The two Anginosus group strains, which were both fully susceptible to all tested fluoroquinolones, are not included for space purposes.
took advantage of a large sample of VGS bacteraemia isolates and patient-specific clinical data and combined these resources with the improved speciation technique of MLSA. 2. Materials and methods 2.1. Bacterial strains, culture conditions, minimum inhibitory concentrations (MICs) and species determination VGS bloodstream isolates were collected from unique patients at MD Anderson Cancer Center hospital (Houston, TX) from 15 April 2011 to 31 December 2012. Clinical and microbiological data, including receipt of fluoroquinolones, were abstracted from electronic medical records using a standardised data collection form. Initial identification of VGS strains was performed using standard microbiological techniques. Strains were grown in Todd–Hewitt broth with 0.2% yeast extract (Becton, Dickinson, and Co., Sparks, MD) at 37 ◦ C with 5% CO2 . The MIC of various fluoroquinolones was determined by Etest (bioMérieux, Marcy l’Étoile, France). MICs were interpreted according to Clinical and Laboratory Standards Institute (CLSI) criteria as follows: for ciprofloxacin and levofloxacin, sensitive ≤2 g/mL, intermediate 4 g/mL and resistant ≥8 g/mL; and for moxifloxacin, sensitive ≤1 g/mL, intermediate 2 g/mL and resistant ≥4 g/mL. VGS species determination and phylogenetic analyses were performed using MLSA as previously described [2]. 2.2. Sequencing of gyrA, gyrB, parC and parE quinolone resistance-determining regions (QRDRs) Amplification and sequencing of gyrA, gyrB, parC and parE loci in each of the VGS strains was done using previously published primers and protocols [3]. Delineation of the QRDRs corresponded to the following amino acid residues from Escherichia coli: 46–172 for GyrA; 371–512 for GyrB; 35–157 for ParC; and 398–483 for ParE [5]. QRDRs were compared with the consensus sequence for
Streptococcus pneumoniae (National Center for Biotechnology Information) given that a consensus sequence, or susceptible allele, has not been established for each VGS species. Polymorphisms in the VGS QRDRs were considered determinative if the polymorphisms had been previously described as contributing to fluoroquinolone resistance, or if the polymorphism was present in non-susceptible VGS strains with no other polymorphisms known to contribute to fluoroquinolone resistance [1,3,6–8]. Polymorphisms not meeting these criteria were considered non-determinative. 2.3. Statistical analysis The relationship between QRDR polymorphisms and type of fluoroquinolone prophylaxis was determined using 2 analysis. The statistical significance of the differences between the rates of fluoroquinolone non-susceptibility among distinct VGS species/groups was analysed by Fisher’s exact test. All tests of significance were two-sided, and statistical significance was defined at P ≤ 0.05. Statistical analysis was performed using IBM SPSS Statistics for Windows v.19.0 (IBM Corp., Armonk, NY). 3. Results and discussion 3.1. Clinical characteristics of infected patients and fluoroquinolone susceptibility of viridans group streptococci isolates During the study period, 115 unique patients with VGS bacteraemia were identified; 93 (80.9%) were neutropenic and 79 (68.7%) were receiving fluoroquinolone prophylaxis at the time of infection onset. Of the 79 patients receiving fluoroquinolone prophylaxis, 57 (72%) were receiving levofloxacin, 17 (22%) ciprofloxacin and 5 (6%) moxifloxacin (Supplementary Table S1). Rates of VGS non-susceptibility to each of the tested fluoroquinolones were as follows: ciprofloxacin, 78% [MIC50 (MIC required to inhibit 50% of the isolates) ≥32 g/mL; MIC90 (MIC
Please cite this article in press as: Sahasrabhojane P, et al. Species-level assessment of the molecular basis of fluoroquinolone resistance among viridans group streptococci causing bacteraemia in cancer patients. Int J Antimicrob Agents (2014), http://dx.doi.org/10.1016/j.ijantimicag.2014.01.031
G Model ANTAGE-4285; No. of Pages 5
ARTICLE IN PRESS P. Sahasrabhojane et al. / International Journal of Antimicrobial Agents xxx (2014) xxx–xxx
3.3. The majority of fluoroquinolone-resistant viridans group streptococci strains have determinative polymorphisms in GyrA and ParC Since there are limited data available regarding the genetic mechanisms of fluoroquinolone resistance among invasive VGS [1], we next sequenced the QRDRs of gyrA, gyrB, parC and parE in all 115 VGS bacteraemia isolates. First we attempted to define the susceptible allele consensus sequence for each VGS species by determining the gene sequence in VGS strains that were fully susceptible to all of the tested fluoroquinolones. Polymorphisms in these strains (compared with S. pneumoniae) were considered non-determinative. When aligned to the QRDRs of S. pneumoniae, non-determinative polymorphisms were observed in all four genes/proteins that were distinct for individual VGS species (Supplementary Figs. S1 and S2). Supplementary Fig. S2 related to this article can be found, in the online version, at http://dx.doi.org/10.1016/j.ijantimicag. 2014.01.031. Establishing consensus (i.e. susceptible) sequences for each VGS species allowed for the assessment of determinative polymorphisms in the QRDRs among the fluoroquinolone-resistant strains. Consistent with data from E. coli and S. pneumoniae, all but one of the quinolone resistance determinative polymorphisms were located in the genes encoding either GyrA or ParC (Fig. 2), with
S81A (n = 1)
E85K (n = 2)
E85G (n = 1)
S81F (n = 42)
S81Y (n = 15)
C
ParC Determinative Mutations (n = 73)
D83G D83N S79N (n = 1) (n = 3) (n = 1) D83Y (n =2) S79C W92C (n = 1) (n = 1) S79A (n = 1) S79Y S79R (n = 20) (n = 11)
S81L (n = 9)
S79F S79I (n = 18) (n = 14)
60
Determinative polymorphism pattern Both GyrA and ParC ParC Alone GyrA Alone None
40
P < 0.05
20
in ac ox ifl ox M
C
ip
ro
vo
flo
flo
xa
xa
ci
ci
n
n
0
Le
Next we sought to use the MLSA species data to evaluate the relationship between fluoroquinolone susceptibility and VGS species. By MSLA, there were 11 distinct VGS species causing disease in this cohort (Supplementary Fig. S1). Fluoroquinolone resistance was observed for all VGS species/groups with the exception of both Anginosus group strains (Fig. 1). Streptococcus mitis and Streptococcus oralis strains were more likely to be fluoroquinolone non-susceptible compared with other VGS species (P < 0.001 by 2 analysis). This difference is likely accounted for by the fact that patients with S. mitis and S. oralis infections were more likely to be receiving fluoroquinolone prophylaxis compared with other patients (P < 0.002 by 2 analysis). Thus, we conclude that fluoroquinolone non-susceptibility is not limited to a particular VGS species or group. Supplementary Fig. S1 related to this article can be found, in the online version, at http://dx.doi.org/10.1016/j.ijantimicag. 2014.01.031.
B
GyrA Determinative Mutations (n = 70)
N on e
3.2. Quinolone resistance according to viridans group streptococci species
A
Number of isolates
required to inhibit 90% of the isolates) ≥32 g/mL]; levofloxacin, 69% (MIC50 ≥ 32 g/mL; MIC90 ≥ 32 g/mL); and moxifloxacin, 60% (MIC50 = 3 g/mL; MIC90 = 16 g/mL) (Fig. 1). As expected, patients receiving fluoroquinolone prophylaxis were significantly more likely to be infected with a VGS strain non-susceptible to one or more fluoroquinolones (74/79; 94%) compared with patients not receiving fluoroquinolone prophylaxis (16/36; 44%) (P < 0.001). However, the finding that 16 of 36 patients were infected with fluoroquinolone non-susceptible VGS without receiving prophylaxis led us to examine their records. We found that only 5 of those 16 patients had received a fluoroquinolone in the 6 months preceding their VGS bacteraemia episode. Therefore, nearly one-half of the VGS strains causing bacteraemia in patients not receiving fluoroquinolone prophylaxis were non-susceptible to one or more fluoroquinolones, suggesting that these patients may have acquired fluoroquinolone-resistant VGS strains rather than having such strains selected for by fluoroquinolone exposure. Supplementary Table S1 related to this article can be found, in the online version, at http://dx.doi.org/10.1016/j.ijantimicag. 2014.01.031.
3
Type of fluorquinolone prophylaxis Fig. 2. Determinative polymorphisms observed in this study and the relationship with the type of fluoroquinolone prophylaxis received at the onset of viridans group streptococci (VGS) bacteraemia. (A) Specific GyrA polymorphisms observed, including the number of strains with each polymorphism. (B) Specific ParC polymorphisms observed, including the number of strains with each polymorphism. (C) Depiction of the number of strains (y-axis) having either no determinative polymorphisms or determinative polymorphisms in GyrA alone, in ParC alone or in both GyrA and ParC, stratified by the type of prophylaxis being administered (x-axis). The three strains with GyrA and ParE polymorphisms were included in the much larger GyrA + ParC group given that ParC and ParE together make up topoisomerase IV. P-Values refer to 2 analysis, which showed that patients receiving ciprofloxacin prophylaxis were more likely to have ParC polymorphisms alone compared with patients receiving either levofloxacin or moxifloxacin prophylaxis.
the lone exception being the D435N polymorphism in ParE [7]. This polymorphism was observed in three strains (one S. oralis, one Streptococcus salivarius and one Streptococcus vestibularis), always in conjunction with a GyrA determinative polymorphism. Seven strains (six S. mitis and one S. oralis) contained a determinative polymorphism only in ParC and were mainly resistant to ciprofloxacin and susceptible to moxifloxacin (Supplementary Table S2). The most common pattern of determinative QRDR polymorphisms observed was alterations in both GyrA and ParC, which was associated with high-level resistance to all fluoroquinolones tested (Supplementary Table S2). The vast majority of determinative GyrA polymorphisms involved S81, with only three strains
Please cite this article in press as: Sahasrabhojane P, et al. Species-level assessment of the molecular basis of fluoroquinolone resistance among viridans group streptococci causing bacteraemia in cancer patients. Int J Antimicrob Agents (2014), http://dx.doi.org/10.1016/j.ijantimicag.2014.01.031
G Model
ARTICLE IN PRESS
ANTAGE-4285; No. of Pages 5
P. Sahasrabhojane et al. / International Journal of Antimicrobial Agents xxx (2014) xxx–xxx
4
A
B 2 2
S. mitis
1 3
1 3
4 4 0 .0 1
0 .0 1
S. oralis 5 5
= No determinative polymorphism = S79A = S79F = S79C = S79Y = S79N + D83N = S79I = D83N = S79R = W92C = D83Y = D83G
= No determinative polymorphism = S81F = S81Y = S81L = S81A = E85K
Fig. 3. Identical determinative polymorphisms of the ParC and GyrA proteins can be found in Streptococcus mitis and Streptococcus oralis strains that are closely related by multilocus sequence analysis (MLSA). Panels (A) and (B) show phylogenetic relationships among S. mitis and S. oralis strains as described for Supplementary Fig. S1. Numbers 1–5 refer to clusters of strains as determined by having 0 or 1 nucleotide difference over the 3063 bp concatenated MSLA sequence. (A) Fluoroquinolone resistancedetermining polymorphisms in ParC are shown as per the legend. Note that all strains in Clusters 2–5, including all six strains in Cluster 3, have the same ParC polymorphisms. Panel (B) is the same as panel (A) except that GyrA polymorphisms are shown.
having a polymorphism at E85 (Fig. 2A). Similarly, S79 was the main site of determinative ParC polymorphisms, with six strains having a polymorphism at D83 and one strain having a W92C polymorphism (Fig. 2B). Of note, there have been contradictory reports regarding the role of ParC K137 polymorphisms in the development of fluoroquinolone resistance in S. pneumoniae [7]. In the current data set, ParC K137 polymorphisms were ubiquitous in fluoroquinolonesusceptible VGS of the Sanguinis, Salivarius and Anginosus groups, demonstrating that ParC polymorphisms at K137 do not contribute to VGS fluoroquinolone resistance. Supplementary Table S2 related to this article can be found, in the online version, at http://dx.doi.org/10.1016/j.ijantimicag. 2014.01.031. A real-time PCR assay has been developed to detect the presence of the C-to-T substitution that results in the S81F GyrA polymorphism and the C-to-A substitution that results in the S79Y ParC polymorphism in S. pneumoniae [9]. Although these substitutions were frequently observed in the fluoroquinoloneresistant VGS strains in this study, together these two mutations accounted for less than one-half of the determinative GyrA and ParC polymorphisms in this cohort, indicating that a broader array of molecular changes will need to be targeted in order to fully capture the range of determinative polymorphisms underlying fluoroquinolone resistance in VGS (Fig. 2 and Supplementary Table S3). The nucleotide changes underlying determinative polymorphisms in GyrA and ParC are summarised in Supplementary Table S3.
Supplementary Table S3 related to this article can be found, in the online version, at http://dx.doi.org/10.1016/j.ijantimicag. 2014.01.031. 3.4. The determinative polymorphism pattern in GyrA and ParC can be linked to the type of fluoroquinolone prophylaxis being received Nearly 70% of the current cohort developed VGS bacteraemia during fluoroquinolone prophylaxis, providing an unprecedented opportunity to study the relationship between fluoroquinolone receipt and the mechanism of streptococcal fluoroquinolone resistance. In vitro studies in S. pneumoniae have established that different fluoroquinolones have a distinct predilection to select for mutations in ParC compared with GyrA, but such bias has not been demonstrated clinically in streptococci [10]. In the current data set, the presence of determinative polymorphisms in GyrA or ParC was statistically different depending on the type of fluoroquinolone prophylaxis administered (Fig. 2C). Specifically, patients receiving ciprofloxacin were more likely (P < 0.05 by 2 analysis) to have polymorphisms in ParC alone compared with patients receiving levofloxacin or moxifloxacin. Polymorphisms in both GyrA and ParC were present in all isolates from patients receiving moxifloxacin prophylaxis and in the vast majority of patients receiving levofloxacin (Fig. 2C). Two strains with ParC polymorphisms alone and three strains with both GyrA and ParC polymorphisms (Fig. 2C) were isolated from patients not receiving prophylaxis. In these five patients,
Please cite this article in press as: Sahasrabhojane P, et al. Species-level assessment of the molecular basis of fluoroquinolone resistance among viridans group streptococci causing bacteraemia in cancer patients. Int J Antimicrob Agents (2014), http://dx.doi.org/10.1016/j.ijantimicag.2014.01.031
G Model ANTAGE-4285; No. of Pages 5
ARTICLE IN PRESS P. Sahasrabhojane et al. / International Journal of Antimicrobial Agents xxx (2014) xxx–xxx
there was no record of fluoroquinolone receipt in the 180 days preceding their infection onset, suggesting that these patients may have acquired fluoroquinolone-non-susceptible VGS in the absence of direct fluoroquinolone exposure. 3.5. Combination analysis of GyrA and ParC polymorphisms and multilocus sequence analysis data in Streptococcus mitis and Streptococcus oralis strains Because the S. mitis and S. oralis strains in this study appeared to have a propensity to be fluoroquinolone non-susceptible and comprised the majority of isolates, we next sought to combine the MLSA and GyrA/ParC QRDR data for strains of these two species. Most S. mitis and S. oralis strains differed by ≥20 nucleotides from their closest neighbour over the 3063 concatenated MLSA nucleotide sequence. However, five non-epidemiologically related clusters of S. mitis and S. oralis strains were observed (Fig. 3), where isolates within each cluster were essentially genetically identical (0 or 1 bp difference, >99.9% identity) by MLSA, including one cluster involving six patients in which S. mitis strains were isolated over a greater than 1-year period (Cluster 3; Fig. 3A). For four of the five clusters (Clusters 2–5), the strains within each cluster had the same S79 ParC determinative polymorphism (Fig. 3A). The exception was Cluster 1, where one strain was fluoroquinolone-susceptible and the other strain was fluoroquinolone-resistant containing a ParC S79R polymorphism. From a GyrA standpoint, strains from two of the five clusters had the same S81 GyrA polymorphism (Fig. 3B). For Cluster 3, there were two distinct GyrA variations observed, with five strains sharing the same GyrA polymorphism (Fig. 3B). Review of patient data did not reveal an epidemiological link between patients harbouring strains from the same clusters, as such strains were isolated from patients in different hospital units and over a relatively long period of time (data not shown). This discovery was surprising because, based on limited data, it has been thought that each patient carries a mixture of genetically unique fluoroquinolone-susceptible VGS strains where at least one becomes resistant to fluoroquinolones following antimicrobial exposure [8,11]. In this scenario, one would expect to see no relationship between genetic relatedness by MLSA and ParC polymorphism unless the genetic background of the strain predisposes the strain to develop a particular ParC polymorphism. The apparently random distribution of the 11 distinct ParC polymorphisms across S. mitis and S. oralis strains (see Fig. 3) does not appear to support a genetic predisposition hypothesis. Rather, the most likely explanation for these data appears to be that a small subset of S. mitis and S. oralis strains with pre-existing polymorphisms in ParC have been acquired by distinct patients. This finding is similar to recently published findings of clonal dissemination of fluoroquinolone-resistant E. coli [12]. In S. pneumoniae, ParC polymorphisms have limited detectable impact on bacterial fitness [13]. Moreover, S. pneumoniae strains with pre-existing ParC determinative polymorphisms, while testing in the susceptible range for fluoroquinolones, can rapidly acquire GyrA determinative polymorphisms following fluoroquinolone exposure leading to fluoroquinolone treatment failure [14]. Thus, data in the current study indicate that at least some of the patients may be exogenously acquiring VGS strains with pre-existing ParC polymorphisms that then develop into fully fluoroquinolone-resistant VGS with both ParC and GyrA determinative polymorphisms. 4. Conclusion The emergence of fluoroquinolone resistance among streptococci is a major public health concern, particularly as genetic
5
exchange of resistance determinants between VGS and S. pneumoniae is known to occur [15]. Previous studies of fluoroquinolone resistance in streptococci have mainly been conducted at either the surveillance level or by studying the emergence of resistance under laboratory conditions [4,5,7]. Prior to this study, patientspecific clinical data on fluoroquinolone resistance, particularly with regard to previous fluoroquinolone exposure, were scant [1,8]. By combining molecular epidemiology, targeted gene sequencing and patient-specific clinical observations, we have generated several new insights into the epidemiology and mechanisms of VGS fluoroquinolone resistance, which lay the groundwork for more extensive investigation into fluoroquinolone non-susceptibility among streptococci in general. Funding: This work was supported by an internal grant from MD Anderson Cancer Center (Houston, TX) to SAS. Competing interests: None declared. Ethical approval: The research was conducted in accordance with the Declaration of Helsinki and was approved by the MD Anderson (MDA) Institutional Review Board [PA12-0230]. References [1] Razonable RR, Litzow MR, Khaliq Y, Piper KE, Rouse MS, Patel R. Bacteremia due to viridans group streptococci with diminished susceptibility to levofloxacin among neutropenic patients receiving levofloxacin prophylaxis. Clin Infect Dis 2002;34:1469–74. [2] Bishop CJ, Aanensen DM, Jordan GE, Kilian M, Hanage WP, Spratt BG. Assigning strains to bacterial species via the internet. BMC Biol 2009;7: 3. [3] Maeda Y, Murayama M, Goldsmith CE, Coulter WA, Mason C, Millar BC, et al. Molecular characterization and phylogenetic analysis of quinolone resistancedetermining regions (QRDRs) of gyrA, gyrB, parC and parE gene loci in viridans group streptococci isolated from adult patients with cystic fibrosis. J Antimicrob Chemother 2011;66:476–86. [4] Rodríguez-Avial I, Rodríguez-Avial C, Culebras E, Picazo JJ. Fluoroquinolone resistance among invasive viridans group streptococci and Streptococcus bovis isolated in Spain. Int J Antimicrob Agents 2007;29: 478–80. [5] Brueggemann AB, Coffman SL, Rhomberg P, Huynh H, Almer L, Nilius A, et al. Fluoroquinolone resistance in Streptococcus pneumoniae in United States since 1994–1995. Antimicrob Agents Chemother 2002;46:680–8. ˜ [6] González I, Georgiou M, Alcaide F, Balas D, Linares J, de la Campa AG. Fluoroquinolone resistance mutations in the parC, parE, and gyrA genes of clinical isolates of viridans group streptococci. Antimicrob Agents Chemother 1998;42:2792–8. [7] Patel SN, McGeer A, Melano R, Tyrrell GJ, Green K, Pillai DR, et al. Susceptibility of Streptococcus pneumoniae to fluoroquinolones in Canada. Antimicrob Agents Chemother 2011;55:3703–8. [8] Prabhu RM, Piper KE, Litzow MR, Steckelberg JM, Patel R. Emergence of quinolone resistance among viridans group streptococci isolated from the oropharynx of neutropenic peripheral blood stem cell transplant patients receiving quinolone antimicrobial prophylaxis. Eur J Clin Microbiol Infect Dis 2005;24:832–8. [9] Page S, Vernel-Pauillac F, O’Connor O, Bremont S, Charavay F, Courvalin P, et al. Real-time PCR detection of gyrA and parC mutations in Streptococcus pneumoniae. Antimicrob Agents Chemother 2008;52:4155–8. [10] Pestova E, Millichap JJ, Noskin GA, Peterson LR. Intracellular targets of moxifloxacin: a comparison with other fluoroquinolones. J Antimicrob Chemother 2000;45:583–90. [11] Wisplinghoff H, Reinert RR, Cornely O, Seifert H. Molecular relationships and antimicrobial susceptibilities of viridans group streptococci isolated from blood of neutropenic cancer patients. J Clin Microbiol 1999;37: 1876–80. [12] van Hees BC, Tersmette M, Willems RJ, de Jong B, Biesma D, van Hannen EJ. Molecular analysis of ciprofloxacin resistance and clonal relatedness of clinical Escherichia coli isolates from haematology patients receiving ciprofloxacin prophylaxis. J Antimicrob Chemother 2011;66:1739–44. [13] Rozen DE, McGee L, Levin BR, Klugman KP. Fitness costs of fluoroquinolone resistance in Streptococcus pneumoniae. Antimicrob Agents Chemother 2007;51:412–6. [14] Varon E, Houssaye S, Grondin S, Gutmann L. Nonmolecular test for detection of low-level resistance to fluoroquinolones in Streptococcus pneumoniae. Antimicrob Agents Chemother 2006;50:572–9. [15] Ip M, Chau SSL, Chi F, Tang JL, Chan PK. Fluoroquinolone resistance in atypical pneumococci and oral streptococci: evidence of horizontal gene transfer of fluoroquinolone resistance determinants from Streptococcus pneumoniae. Antimicrob Agents Chemother 2007;51:2690–700.
Please cite this article in press as: Sahasrabhojane P, et al. Species-level assessment of the molecular basis of fluoroquinolone resistance among viridans group streptococci causing bacteraemia in cancer patients. Int J Antimicrob Agents (2014), http://dx.doi.org/10.1016/j.ijantimicag.2014.01.031
