Introduction

Administration of quinolone antibacterial prophylaxis during the neutropenic period of bone marrow or peripheral blood stem cell transplantation (PBSCT) for hematologic malignancy reduces the incidence of gram-negative bacterial infections [1–9]. Gram-positive bacteria are the most common organisms (29–74%) isolated in breakthrough bacteremia in patients receiving quinolone prophylaxis [10, 11]; viridans group streptococci are second only to coagulase-negative staphylococci [11–13]. Chemotherapy-induced oral mucositis is considered the major source of viridans group streptococcal bacteremia in this patient population [13–15]. Viridans group streptococcal bacteremia can lead to life-threatening conditions such as septic shock, acute respiratory distress syndrome, and endocarditis [10–12, 15–18].

Breakthrough bacteremia with quinolone-resistant viridans group streptococci during quinolone prophylaxis is of concern [19, 20]. We reported a cluster of cases of levofloxacin-resistant viridans group streptococcal bacteremia among autologous PBSCT recipients receiving levofloxacin prophylaxis [19]. The isolates also had reduced susceptibility to gatifloxacin and moxifloxacin. These findings prompted a switch to the combination of gatifloxacin and penicillin as antibacterial prophylaxis in an attempt to reduce the incidence of viridans group streptococcal bacteremia. Other hematology programs have taken similar measures [10, 21]. Gatifloxacin was chosen for its superior in vitro activity against viridans group streptococci (compared to older quinolones) [22–25]. Quinolones inhibit DNA gyrase (topoisomerase II) and topoisomerase IV, which are tetramers composed of GyrA/GyrB and ParC/ParE subunits, respectively [26]. Mutations in the quinolone resistance-determining region (QRDR) of gyrA at codons 81 (Ser) or 85 (Glu) and/or parC at codons 79 (Ser) or 83 (Asp) and efflux confer resistance to quinolones in viridans group streptococci [27, 28].

We hypothesized that the prior episodes of quinolone-resistant viridans group streptococcal bacteremia in our autologous PBSCT patients receiving quinolone prophylaxis resulted from selection of these organisms in their oropharynx.

The purpose of this study was to prospectively monitor for the emergence of quinolone resistance in viridans group streptococci recovered from the oropharynx among neutropenic hematology patients receiving quinolone prophylaxis. Quinolone-resistant viridans group streptococci were phenotypically and genotypically characterized.

Patients and methods

Patient selection

Volunteer patients ≥18 years who were expected to undergo autologous or allogeneic bone marrow transplantation or PBSCT at the Mayo Clinic, Rochester, MN, USA, were enrolled from 19 July 2002 to 15 May 2003. Before transplantation and initiation of quinolone prophylaxis and after obtaining informed consent, the subject's oropharynx was swabbed with a BBL CultureSwab Collection & Transport System (Becton Dickinson, Sparks, MD, USA). A repeat oropharyngeal culture was obtained approximately 1 week after commencing quinolone prophylaxis. Follow-up oropharyngeal cultures were not obtained if the subject did not undergo transplantation or developed febrile neutropenia before receiving 7 days of quinolone prophylaxis, or if a quinolone was not prescribed.

Prophylaxis with a quinolone and penicillin VK was initiated during conditioning chemotherapy 2 days prior to transplantation and continued through neutrophil engraftment (absolute neutrophil count [ANC] >500 cells/mm3) or development of febrile neutropenia. Subjects with multiple myeloma received two courses of quinolone prophylaxis: first, during the neutropenic period (ANC ≤500 cells/mm3) of stem cell mobilization, and second, during transplantation. Patients were enrolled and sampled only during their first course of quinolone prophylaxis for either stem cell mobilization or transplantation.

Subjects were monitored for infectious complications during the neutropenic period after transplantation while receiving quinolone prophylaxis. The Mayo Clinic Institutional Review Board approved this study.

Oral antimicrobial prophylaxis regimen

From October 2001 until March 2003, gatifloxacin 400 mg once daily was administered to most patients. After this time, moxifloxacin 400 mg once daily was substituted (due to a hospital formulary change). Penicillin VK 500 mg twice daily was included in the prophylaxis regimen during transplantation, but not during stem cell mobilization. Patients received the quinolone alone for a minimum of 7 days after chemotherapy for stem cell mobilization during the expected period of neutropenia. Fluconazole was used as antifungal prophylaxis and valacyclovir as antiviral prophylaxis for the initial 30 days post-transplant. Study subjects were seen daily by the transplant service.

Conditioning chemotherapy

Patients with non-Hodgkin's lymphoma typically received carmustine, etoposide, cytarabine, melphalan, and dexamethasone conditioning chemotherapy. Patients with multiple myeloma or amyloidosis received high-dose melphalan. Allogeneic stem cell transplantation included total body irradiation. For patients with multiple myeloma, stem cell mobilization consisted of high-dose cyclophosphamide and granulocyte-macrophage colony stimulating factor or granulocyte colony stimulating factor. All patients had a tunneled central venous catheter.

Definitions

Neutropenia was defined as an ANC of ≤500 cells/μl. Duration of neutropenia was the number of days the ANC was ≤500 cells/μl. Febrile neutropenia was defined as an ANC of ≤500 cells/μl and an oral temperature of ≥38°C.

Isolation and identification of viridans group streptococci

Each oropharyngeal swab was inoculated onto duplicate sheep blood agar plates and one trypticase soy agar plate supplemented with 5% lysed horse blood and 1 μg/ml of gatifloxacin. After 20–24 h of incubation at 37°C in a 5% CO2 incubator, unique colonies resembling viridans group streptococci were subcultured and classified by morphological and biochemical criteria into the five major groups of viridans group Streptococcus species: mitis, anginosus, salivarius, sanguinis, or mutans [29].

Antimicrobial susceptibility testing

MIC values were determined by broth microdilution for ciprofloxacin (USP, Rockville, MD, USA), levofloxacin (Pharmaceutical Research Institute, Spring House, PA, USA), gatifloxacin (Bristol-Myers Squibb, Plainsboro, NJ, USA), moxifloxacin (Bayer Pharmaceutical Division, West Haven, CT, USA), garenoxacin (Bristol-Myers Squibb, Plainsboro, NJ, USA), penicillin (Sigma Chemical, St. Louis, MO, USA), erythromycin gluceptate (USP), and vancomycin (Sigma) and interpreted in accordance with the guidelines of the Clinical and Laboratory Standards Institute [30, 31]. In order to perform an exploratory analysis of susceptibilities, the following breakpoints (in μg/ml) were applied: S≤2, R≥8 for ciprofloxacin, and S≤1, R≥4 for gatifloxacin, moxifloxacin, and garenoxacin, for which there are no established Clinical and Laboratory Standards Institute breakpoints. Gemella morbillorum American Type Culture Collection 27825 was used as a quality control strain for susceptibility testing.

Identification of mutations in the quinolone resistance-determining region

Isolates recovered before quinolone use and demonstrating reduced susceptibility to levofloxacin or ciprofloxacin (MIC≥8 μg/ml), as well as all isolates recovered during quinolone use, underwent QRDR mutation testing. DNA was prepared for amplification with either the IsoQuick Nucleic Acid Extraction Kit (ORCA Research, Bothell, WA, USA) or with alkaline wash as described previously, with some modifications [32]. For alkaline wash, several fresh colonies were inoculated into 7 ml of Todd–Hewitt broth and grown overnight at 37°C in a 5% CO2 incubator. The bacterial suspension was centrifuged at 4,000×g for 25 min, and the pellet was resuspended in 500 μl of alkaline wash solution (0.05 M sodium citrate, 0.5 M sodium hydroxide), transferred to a 1.5-ml microcentrifuge tube, and incubated at room temperature for 20 min. The suspension was centrifuged at 14,000×g for 5 min, and the pellet was resuspended in 500 μl of Tris–HCl buffer (0.5 M, pH 8.0). The suspension was centrifuged again at 14,000×g for 5 min, and the pellet was resuspended in 100 μl of sterile water in a screw-cap microcentrifuge tube and placed in a boiling water bath for 10 min. The tube was centrifuged for 10 min at 14,000×g and the supernatant used for polymerase chain reaction (PCR).

PCR amplification of gyrA was performed either with primers 5′CAGGKGATGTYATGGGTA3′ (forward) and 5′GAAGCATTTCCARRGCAA3′ (reverse), which amplify a 163 base pair region of gyrA, or with previously published primers [33]. For parC, primers 5′CAGCGYCGKATTCTTTATTC3′ (forward) and 5′TTCKGTRTCRTCAAAGTTCC3′ (reverse), which amplify a 287 base pair region of the QRDR of parC, were used.

PCR was performed in a total reaction volume of 50 μl containing 5 μl of extracted DNA, 1× PCR buffer containing 1.5 mM MgCl2, 200 μmol/l of each deoxyribonucleotide (Boehringer Mannheim, Indianapolis, IN, USA), 0.4 μmol/l of each primer, 1.25 U of AmpliTaq Gold DNA polymerase (Applied Biosystems, Foster City, CA, USA), and 10 μl of 50% glycerol. The following cycling parameters were used: initial preheating for 10 min at 95°C, followed by 40 cycles of 1 min at 94°C, 1 min at 46°C for gyrA or 50°C for parC primers, and 1 min at 72°C. Amplification was confirmed by running the amplified product on a 1.5% agarose gel with 0.5 μg/ml of ethidium bromide. DNA sequencing was performed by ABI Prism Big Dye Terminator cycle sequencing with the ABI Prism 377 automated DNA sequencer (Applied Biosystems). Sequence data were analyzed using Sequencher 3.1.1 (GeneCodes, Ann Arbor, MI, USA). Codon positions 81 and 85 of gyrA and 79 and 83 of parC were analyzed for mutations in comparison with previously published QRDR sequences of viridans group streptococci [19, 33, 34].

Statistical methods

Differences in antimicrobial susceptibility of viridans group streptococci recovered before and during prophylaxis with a quinolone were analyzed using the two-tailed Fisher's exact test or chi-square test with the statistical software package JMP Version 5.0.1.2 (SAS Institute, Cary, NC, USA). A p value of <0.05 was considered statistically significant.

Results

Patient characteristics

Forty-eight patients were enrolled. The median patient age was 55 years (range, 23–72 years); 62% were men, and all were Caucasian. Seventeen (35%) patients had non-Hodgkin's lymphoma, 17 (35%) had multiple myeloma, nine (19%) had amyloidosis, two (4%) had Hodgkin's lymphoma, one (2%) had acute lymphoblastic leukemia, one (2%) had chronic lymphocytic leukemia, and one (2%) had POEMS (polyneuropathy, organomegaly, endocrinopathy, monoclonal gammopathy, and skin changes) syndrome.

Twenty-five (52%) enrolled patients had received an antimicrobic within the month preceding their first oropharyngeal culture; however, only five had received a quinolone in this time frame. Forty-three patients underwent PBSCT, three underwent bone marrow transplantation, and two had no transplant performed. Of the 46 transplanted patients, 45 received a quinolone; one received cefixime as antibacterial prophylaxis (because of a potential drug interaction with an antiarrhythmic agent).

Microbiological follow-up investigations

Forty-seven of 48 (98%) patients had viridans group streptococci recovered from the initial oropharyngeal swab. Thirty-eight of the 46 (83%) transplanted patients had a follow-up oropharyngeal culture while receiving antibacterial prophylaxis. Thirty-three of these 38 (87%) patients were receiving a gatifloxacin-based regimen (with or without penicillin) at the time the follow-up throat swab was obtained (Table 1). Eight transplanted patients did not have a follow-up culture because no quinolone was used (n=1), the patient was missed at follow-up (n=2), or the patient developed early febrile neutropenia (n=5).

Table 1 Follow-up microbiological results and patient outcome after hematopoietic stem cell transplantation
Full size table

Twenty of 38 (53%) patients who had a follow-up swab had viridans group streptococci recovered (Table 2). The median duration of quinolone use at the time of repeat oropharyngeal culture was 8 days (range, 6–11 days).

Table 2 Species groups of viridans group streptococci isolated from patients before and after quinolone prophylaxis
Full size table

Selection of quinolone resistance

There was a statistically significant decrease in quinolone susceptibility among viridans group streptococcal isolates recovered during quinolone prophylaxis versus those recovered before quinolone prophylaxis (Table 3); 70 versus 11% were susceptible to ciprofloxacin (p<0.0001), 89 vs. 30% were susceptible to levofloxacin (p<0.0001), 89 versus 37% were susceptible to gatifloxacin (p<0.0001), and 91 versus 41% were susceptible to moxifloxacin (p<0.0001). These findings were both similar and statistically significant when the analysis included only the 20 patients who had isolates recovered both before and during quinolone use (data not shown). Among the available quinolones, moxifloxacin or gatifloxacin had lower MIC90 values than levofloxacin or ciprofloxacin. There was no statistically significant difference in penicillin or erythromycin susceptibility prior to versus during prophylaxis (Table 3). Only isolates from S. mitis and S. sanguinis species groups demonstrated resistance to quinolones both before and during quinolone prophylaxis (Table 4).

Table 3 Antimicrobial susceptibility of viridans group streptococci recovered from 47 patients before and 20 patients after commencing quinolone prophylaxis
Full size table
Table 4 Susceptibility patterns of different species groups before and after quinolone prophylaxis
Full size table

Presence of quinolone resistance-determining region mutations in quinolone-resistant isolates

QRDR regions were successfully amplified in 21 of 40 (53%) isolates for gyrA and in 36 of 40 (90%) isolates for parC. The following mutations were found in GyrA: Ser 81→Tyr (n=3) or Phe (n=6), and/or Glu 85→Lys (n=2). Mutations found in ParC were Ser 79→Arg (n=4), Ile (n=3), Phe (n=4), or Tyr (n=2), and/or Asp 83→Asn (n=2) or Glu (n=4). Most isolates that were not susceptible to levofloxacin had at least one QRDR mutation in either gyrA or parC. One S. mitis isolate with an MIC value of 16 μg/ml for levofloxacin, 8 μg/ml for gatifloxacin, and 4 μg/ml for moxifloxacin had a one-step mutation in GyrA (Ser 81→Phe) only. For three isolates (one S. mitis, two S. sanguinis) with levofloxacin MIC values of 16–32 μg/ml, two-step mutations involving GyrA (Ser 81→Phe or Tyr) and ParC (Ser 79→Ile, Tyr, or Arg) were detected. In one S. mitis isolate that was highly resistant to levofloxacin, gatifloxacin, and moxifloxacin (MIC 128 μg/ml), three QRDR mutations were detected (GyrA: Ser 81→Phe, Glu85→Lys; ParC: Ser79→Phe).

Clinical outcomes

No episodes of viridans group streptococcal bacteremia occurred. Ten episodes of breakthrough bacteremia, two gram-negative and eight gram-positive, occurred during antibacterial prophylaxis (Table 1).

Discussion

Our study revealed two important clinical findings. First, we found that quinolone-resistant viridans group streptococci were frequently recovered from the oropharynx of patients after only a brief period of exposure to quinolones (primarily gatifloxacin) despite the superior in vitro activity of gatifloxacin against viridans group streptococci compared to levofloxacin or ciprofloxacin. Second, despite the presence of these bacteria in the oropharynx of our study subjects, no episodes of breakthrough bacteremia with viridans group streptococci developed during quinolone–penicillin prophylaxis during the neutropenic period after transplantation.

There have been many published reports of breakthrough bacteremia with quinolone-resistant viridans group streptococci during quinolone prophylaxis [10, 17, 19]. One recently published study has characterized the prevalence of quinolone resistance in viridans group streptococci colonizing the oropharynx of hematology patients during quinolone prophylaxis [35]. In this relatively small (20 study subjects) study, the investigators reported 0% (first week) and 20% (second week) prevalence rates of levofloxacin resistance (MIC≥8 μg/ml) among viridans group streptococci isolated during levofloxacin prophylaxis. In contrast, we observed a much higher prevalence of levofloxacin resistance (59%) among viridans group streptococci recovered after just 6–11 days of quinolone prophylaxis (Table 3) [10, 17, 19]. Methodological differences between this study and ours may explain these discrepancies. The investigators performed twice weekly surveillance throat cultures during levofloxacin prophylaxis [35]. In the eight patients with neutropenic fever, levofloxacin was continued and empiric antimicrobial therapy, typically imipenem, was added. It is unclear in what percentage of patients surveillance throat cultures were performed while levofloxacin was being administered with empiric antimicrobial therapy for febrile neutropenia. This practice may have impacted the recovery of viridans group streptococci and resulted in an underestimation of the prevalence of quinolone resistance. We identified and attempted to determine the presence of QRDR mutations in our viridans group streptococcal isolates, which was not done in their study. The majority (96%) of our patients underwent stem cell transplantation, in contrast to approximately half of their patients. Notably, both studies reported a similar percentage of bloodstream infections associated with resistance to the quinolone administered (Table 1).

We found low rates of susceptibility to available quinolones (11–41%) among viridans group streptococci isolated during quinolone prophylaxis (Table 3). Gatifloxacin or moxifloxacin did not prevent the selection of resistance in vivo. The specific gyrA and parC mutations identified in our quinolone-resistant viridans group streptococcal isolates are consistent with those previously reported [27, 28, 36, 37].

During chemotherapy-induced neutropenia, the use of a quinolone with poor in vitro activity against viridans group streptococci has been found to be a risk factor for breakthrough viridans group streptococcal bacteremia. Gatifloxacin and moxifloxacin have better in vitro activity than levofloxacin against viridans group streptococci [36, 38]. However, no definite conclusion can be drawn regarding the role gatifloxacin (or moxifloxacin) played in the absence of breakthrough bacteremia with viridans group streptococci because of the addition of penicillin. Several studies have suggested, however, that the addition of penicillin to a quinolone-based prophylaxis regimen reduces the incidence of viridans group streptococcal bacteremia [1, 10, 16, 21].

There are limitations to our study. First, it was not meant to determine the efficacy of a quinolone–penicillin prophylaxis regimen in preventing viridans group streptococcal bacteremia. The small sample size, along with the fact that consecutive patients undergoing transplantation were not enrolled, may have potentially resulted in an underestimation of the incidence of breakthrough bacteremia with viridans group streptococci; however, the selection of quinolone resistance should not have been affected. Despite the small sample size, we were able to find a statistically significant difference in quinolone susceptibility after quinolone exposure. Second, there were a variety of antibacterial prophylaxis regimens used, including penicillin, which may have affected our findings. Despite this, we observed a high prevalence of quinolone-resistant isolates. Third, the duration of persistent colonization with quinolone-resistant viridans group streptococci was not determined. It was not feasible to repeat oropharyngeal cultures after cessation of quinolone prophylaxis, given that most patients were placed on broad-spectrum intravenous anti-infective agents for the management of febrile neutropenia (Table 1). Colonization may be expected to persist for at least a month [39]. Fourth, there were difficulties in amplifying both gyrA and parC in many isolates; to our knowledge, one primer set cannot amplify these genes in all five viridans streptococcal species groups [27, 28, 34, 36, 38].

In summary, this study demonstrates that administration of a quinolone to a neutropenic hematology patient for a brief period of time, as short as 1 week, selects for quinolone-resistant viridans group streptococci in the oropharynx.