microbiome complexity and staphylococcus aureus in crs

The LaryngoscopeVC 2012 The American Laryngological,Rhinological and Otological Society, Inc.

Microbiome Complexity and Staphylococcus aureus inChronic Rhinosinusitis

Leah M. Feazel, MS; Charles E. Robertson, PhD; Vijay R. Ramakrishnan, MD; Daniel N. Frank, PhD

Objectives/Hypothesis: The aim of this study was to compare microbiological culture-based and culture-independent(16S rRNA gene sequencing) methodologies for pathogen identification in chronic rhinosinusitis (CRS) patients. We hypothe-sized that bacterial culture and DNA sequencing would yield largely concurrent results, although sequencing would detectgreater bacterial diversity, and the sinonasal microbiomes of CRS patients would differ in composition and diversity comparedwith non-CRS controls.

Study Design: Cross-sectional observational study.Methods: Middle meatus swabs from CRS patients collected during endoscopic sinus surgery were analyzed by both

clinical culture and broad-range analysis of 16S rRNA gene pyrosequences.Results: A total of 21 swab samples from 15 CRS patients and five non-CRS controls were analyzed. One CRS patient

was also swabbed 3 weeks postoperatively due to evidence of purulence during a clinical visit. All subjects had positive bac-terial cultures, with a mean of 2.8 isolates per subject. The most prevalent cultivars were coagulase-negative staphylococci(15/20 specimens, 75%), Staphylococcus aureus (10/20, 50%), and Propionibacterium acnes (6/20, 30%). Among 57,407pyrosequences generated, the most prevalent were from coagulase-negative staphylococci (21/21 specimens, 100%), Coryne-bacterium spp (18/21, 85.7%), P acnes (16/21, 76.2%), and S aureus (14/21, 66.7%). Bacterial diversity correlated withrecent antibiotic use, asthma, prior sinus surgery, and relative abundance of S aureus.

Conclusions: DNA pyrosequencing revealed greater biodiversity than culture, although in most cases culture resultsrepresented a subset of the abundant DNA sequence types. CRS patients were characterized by altered microbial composition(P ¼ .02) and greater abundance of S aureus (P ¼ .03).

Key Words: Bacteria, bacterial sinusitis, human microbiome, sinus culture, middle meatus.Level of Evidence: 2b.

Laryngoscope, 122:467–472, 2012

INTRODUCTIONSinus infections in chronic rhinosinusitis (CRS)

cause significant patient discomfort and increased treat-ment costs. CRS is characterized by >12 weeks of signsand symptoms resulting from inflammation of the para-nasal sinuses.1 The occurrence of bacterial infection inCRS can be either a primary cause or resulting factor ofthe inflammatory disease.2 In primary bacterial CRS,pathogen colonization causes chronic immune upregula-tion and inflammation. Alternatively, an otherwiseinflamed or blocked sinus, resulting from anatomicalobstruction, allergies, or mucosal dysfunction, can pro-mote the ideal environmental conditions for secondarypathogenic bacterial growth. Microbial colonization of

the sinuses in CRS often is characterized by the pres-ence of polymicrobial bacterial biofilms, possibly due toabnormal innate immune function in the patient.2–4 Bio-films can protect pathogen subpopulations from bothhost immune systems and antibiotic therapies andmay thus result in poor outcomes with conventionaltreatments.5

Previous studies suggest that a small suite of bacte-ria is common in the paranasal sinuses of subjects withCRS, whereas the sinuses of healthy individuals arelightly colonized by commensal microorganisms. Com-monly observed species include aerobes and anaerobestypical of the nares, oropharynx, and mouth.6–8 How-ever, conventional culture methods may under-representthe bacterial diversity of sinonasal specimens, particu-larly those that harbor complex microbial communities.In contrast, molecular methods for bacterial identifica-tion that do not require axenic cultivation, such aspolymerase chain reaction (PCR), DNA sequencing, andmicroarrays have the potential to elucidate a broaderunderstanding of the complex microbial ecology that istypical of the human body.9 The goal of this pilot studywas to characterize the composition and diversity ofmicrobial communities that colonize the paranasalsinuses of CRS patients via both conventional cultureand culture-independent molecular methodology. Wehypothesized that bacterial culture and DNA

From the Division of Infectious Diseases (L.M.F., D.N.F.); Departmentof Otolaryngology-Head and Neck Surgery (V.R.R.) and; and the MicrobiomeResearch Consortium (D.N.F.), University of Colorado School of Medicine,Aurora, Colorado; and the Department of Molecular, Cellular, andDevelopmental Biology (C.E.R.), University of Colorado, Boulder,Colorado, U.S.A.

Editor’s Note: This Manuscript was accepted for publicationSeptember 26, 2011.

This work was supported in part by NIH grant HG005964 (D.N.F.).The authors have no other funding, financial relationships, or conflictsof interest to disclose.

Send correspondence to Daniel N. Frank, PhD, Mail Stop B168,12700 E. 19th Avenue, RC2, Aurora, CO 80045.E-mail: [email protected]

DOI: 10.1002/lary.22398

Laryngoscope 122: February 2012 Feazel et al.: Microbiome Complexity and S aureus in CRS

467

pyrosequencing would yield largely concurrent results,although more diversity would be observed throughsequencing, and the sinonasal microbiomes of CRSpatients would differ markedly in both composition anddiversity compared with non-CRS controls.

MATERIALS AND METHODS

Study Design and PopulationThis cross-sectional study was approved by the institutional

review board of the University of Colorado (protocol number 10-1437). CRS was diagnosed in patients according to the 2007Adult Sinusitis Guidelines and was initially managed medicallywith a minimum of 3 weeks of antibiotics and topical intranasalsteroids prior to the decision for surgery.10 Asthma was diag-nosed either clinically based on history of wheeze, cough,shortness of breath and chest tightness or on spirometryresponse with bronchodilator and methacholine administration.11

The diagnosis of allergic rhinitis was made by positive skin pricktesting or clinically by the classic constellation of symptoms ofnasal congestion, sneezing, rhinorrhea, and pruritus. Swabswere collected through convenience sampling of patients under-going outpatient endoscopic sinus surgery (ESS) at University ofColorado Hospital between January 2011 and April 2011. Swabspecimens were obtained during ESS with ESwabs (COPANDiagnostics, Inc., Murrieta, CA) for hospital laboratory cultureand CultureSwabs (BD, Franklin Lakes, NJ) for DNA extraction.Swabs were endoscopically guided to sample areas of mucopuru-lence when present, or into the middle meatus when no grossdisease was encountered. Samples were sent for routine aerobic/anaerobic culture at the hospital laboratory and placed on iceand frozen at �80�C until DNA extraction.

DNA-Based TechniquesTotal genomic DNA was purified from swab heads by a

phenol:chloroform, bead-beating protocol (complete details formolecular methods are described in the Supplemental Methods).All DNA extraction and PCR steps were performed in a HEPA-filtered laminar flow hood that was decontaminated by ultra-violet light. Quantitative PCR (QPCR) assay used previouslypublished oligonucleotide primers for total bacteria (16S rRNAgene, FAM reporter)12 and Staphylococcus aureus (femA gene,TET reporter).13 A multiplex PCR assay for S aureus-specificmethicillin-resistant genes (mecORF)14 was used to test formethicillin-resistant S aureus (MRSA), in specimens that werepositive for S aureus by QPCR and sequence.

Amplicons of the 16S rRNA gene (�500 bp; primers27FYMþ3 and 515R)15,16 were generated via broad-range PCR(30–36 cycles) using 50-barcoded reverse primers.17 Pooled ampli-cons were provided to the Center for Applied Genomics at theUniversity of Toronto for pyrosequencing on a 454/Roche Life Sci-ences GS-FLX instrument using titanium chemistry (Roche LifeSciences, Indianapolis, IN). The software used for sequence pol-ishing and 16S rRNA gene sequence analysis consisted of thefunction-specific tools that were considered to be the standardsin the academic community at the onset of this project. Themean trimmed sequence length was �340 bp. Genus-level taxo-nomic calls were produced by the ribosomal database project(RDP) Classifier, which performs naı̈ve Bayesian taxonomic clas-sification versus a training set.18 Species level taxonomyprecision was obtained via BLAST (Basic Local AlignmentSearch Tool, http://blast.ncbi.nlm.nih.gov/Blast.cgi)19 against adatabase of sequences obtained from Silva version 10420 taggedas isolates, and reported results demanded at least 99% sequenceidentity over 95% of sequence length. Standard ecological indices

of diversity, richness, and evenness were conducted with theBIODIV software tool (www.phyloware.com) embedded withinthe sequence analysis pipeline.21

Statistical AnalysisThe R statistical package (Institute for Statistics and

Mathematics, Wien, Austria) was used for all statistical analy-ses.22 Differences in the compositions of microbiomes (i.e.,operational taxonomic unit [OTU] distributions taken as awhole) between CRS patients and non-CRS controls were meas-ured by the Morisita-Horn index using the adonis function ofthe R package vegan, which performs a nonparametric multipleanalysis of variance with adjustment for covariates.22,23 A labelpermutation test with 10,000 replicates was used to assign a Pvalue to the difference in OTU counts between groups of sam-ples. Demographic characteristics were compared using analysisof variance (ANOVA) for continuous outcomes and Fisher exacttests for categorical outcomes including OTU prevalence. Per-cent OTU abundance data were logit-transformed to a linearscale for ANOVAs. All tests of null hypotheses were evaluatedat a ¼ .05.

RESULTSTwenty consecutive patients undergoing ESS

between January 2011 and April 2011 were included inthis study (Table I, Fig. 1). Fifteen patients were diag-nosed with CRS and five patients served as non-CRScontrols. One patient was swabbed both at the time ofsurgery and 3 weeks postoperatively during a clinic visitdue to evidence of continued purulence (specimens 03Aand 03D respectively). Therefore, a total of 21 specimensfrom 20 subjects were analyzed. Frank pus and/or thickmucus was present in five CRS patients (33%) and onecontrol (20%). Out of the 15 patients with CRS, only two(13%) had polyps noted on endoscopy (mean Lund-Mackay preoperative computed tomography score, 9.5 64.9). With so few polyp patients, this clinical feature wasnot included in the statistical analyses. The controlsincluded patients undergoing ESS for tumor (n ¼ 1),fungal ball (n ¼ 2), silent sinus syndrome (n ¼ 1), andseptoplasty (n ¼ 1). None were immunocompromised orknown to have cystic fibrosis. Swabs in control patientswere taken from the side opposite of disease, from aradiographically and endoscopically normal appearing

TABLE I.Patient Demographics and Clinical Characteristics.

CRS Non-CRS

Subjects, no. 15 5

Age (6SD), yr 45 (615.8) 56 (620.0)

Gender (% female) 10 (66.7) 3 (60)

Allergies 12 (80) 1 (20)

Antibiotics <12 weeks prior (%)* 6/16 (37.5) 1 (20)

Asthma (%) 4 (26.7) 0

Polyps (%) 2 (13.3) 0

Prior ESS (%) 7 (46.7) 0

Pus (%) 5 (33.3) 1 (20)

*Sixteen total specimens were included because chronic rhinosinusi-tis (CRS) patient 3 was on antibiotics between the two sampling events.

ESS ¼ endoscopic sinus surgery.


468

middle meatus. All subjects and controls used intranasalsteroid sprays for at least 1 month prior to surgery.

Standard hospital cultures were positive for all sub-jects and controls, with an average of 2.8 isolatesreported per subject (range, 1–5). One specimen from acontrol subject (patient 17) was not sent for culture,therefore culture results were available for 19 of 20 sub-jects and 20 of 21 specimens. There was no statisticaldifference between the number of isolates obtained fromCRS patients and controls (data not shown). The mostcommon cultivars were coagulase-negative streptococci(15/20 cultures, 75%), S aureus (10/20, 50%), and Propio-nibacterium acnes (6/20, 30%). For many specimens (9/20, 45%), culture results were incompletely speciated,with broad phenotypic groups reported (e.g., mixed anae-robes or Gram-negative rods).

Microbes present in middle meatus specimens werecomprehensively surveyed through pyrosequencing andphylogenetic analysis of bacterial 16S rRNA genes. Atotal of 57,407 high-quality pyrosequencing reads weregenerated for all 21 specimens (median, 1,485 reads perspecimen; interquartile range, 682–3,296). The most

prevalent DNA sequence types detected were coagulase-negative staphylococci (21/21 specimens, 100%), Coryne-bacterium spp (18/21, 85.7%), P acnes (16/21, 76.2%),and S aureus (14/21, 66.7%). For 20 of 21 specimens(95.2%) the majority of sequences (>50%) belonged toknown or suspected upper respiratory bacterial genera(Fig. 1).

In the majority of specimens, bacteria identified byclinical culture were also identified in the sequence dataset (Fig. 1). In 13 of 20 specimens analyzed by culture(65%), more prevalent sequence types (>15% of sequen-ces) corresponded to the organisms identified by culture.For example, the sequence library from patient 5 con-tained more than 52% S aureus and 34% Staphylococcusepidermidis sequences, which were both identified byculture. However, nonspecific culture reports such asmixed GNR were difficult to correlate with sequencedata in a meaningful way. Some of the unresolved cul-ture results may, however, represent species detected bysequence (e.g., the moderate mixed anaerobes of patient12 could include Prevotella spp, which comprised 75% ofthe total sequences from that specimen). Although the

Fig. 1. Comparison of bacterial culture and 16S rRNA sequencing results in chronic rhinosinusitis (CRS) cases and controls. Boxed cellsrepresent genera or species identified by culture. The figure summarizes the distribution of human-commensal and/or airway pathogenicbacterial taxa detected by phylogenetic analysis of pyrosequencing reads, clustered into genus-level groups (because of its potential signif-icance to CRS, Staphylococcus aureus is separated from other staphylococci). Values in the table represent percent abundances of partic-ular taxa for each subject (columns sum to 100%). Other refers to bacterial genera not commonly associated with human health or diseaseand to sequences that could not be assigned to the genus level of taxonomy. m ¼ mixed anaerobes by culture not further identified; p ¼culture identified Peptostreptococcus not seen in sequences; c ¼ culture identified Citrobacter freundii not seen in sequences; ESS ¼ en-doscopic sinus surgery; SSS ¼ silent sinus syndrome; septo ¼ septoplasty; b ¼ boxed cells with bold numbers indicate positive result byculture. *Not sent for culture.


469

dominant sequence types frequently corresponded to cul-ture results, in seven of 20 (35%) of the specimens, thecultured organisms were rare components of the corre-sponding sequence libraries (<15% of sequences). Forexample, the specimen from patient 8 cultured Strepto-coccus, Neisseria, and Proprionibacterium (each of whichwere <1% of sequences), but did not culture Stenotro-phomonas (>78% of sequences).

In the case of four isolates, cultivation of micro-organisms identified species not detected in thesequence data set. In one specimen (patient 13), the cul-ture result of Peptostreptococcus was not identified bysequence. However, the related genera Finegoldia, Pepti-nophilus, and Anaerococcus were present in thesequence data, and these organisms have morphologiessimilar to peptostreptococci. The other discordant resultsincluded two P acnes isolates (patients 9 and 18). Inthese specimens, sequences of other anaerobic bacilliwere detected including Actinomycetes, Rothia, and Pre-votella. The final case of discordance was the cultivationof a Citrobacter fruendii isolate (patient 16). No sequen-ces of other citrate-using genera were identified fromthe specimen, although numerous other members of theEnterobacteraceae family were present in the sequencelibrary. Thus, DNA pyrosequencing provided a broader,more incisive, and oftentimes more sensitive catalog ofmicrobes resident in the middle meatus than didculture.

Different kinds and quantities of bacteria inhabitedthe middle meatuses of CRS and non-CRS patients(Fig. 1). A nonparametric multiple ANOVA test withlabel permutation (see the Methods section) identifiedsignificantly different overall distributions of 16S rRNAsequences recovered from the two groups of subjects (P¼ .02). Although the total bacterial composition variedsignificantly, neither richness nor evenness indicesreached statistical significance between cases and con-trols (Table II, Supplemental Fig. 1). CRS patients were,on average, characterized by fewer bacterial types (34vs. 49 genus-level OTUs in CRS vs. non-CRS; P ¼ .23)and less even distributions of genus-level sequence types(P ¼ .09) compared with controls (Supplemental Fig. 1).

In addition to community-level disparities betweenCRS patients and controls, one species in particular,S aureus, was selectively enriched in CRS. Cultivationand pyrosequencing methods for S aureus detection gavebroadly concordant results. Culture results identifiedS aureus in 10 of 16 CRS specimens (63% prevalence)

and zero of five non-CRS controls, whereas DNA sequen-ces representative of S aureus were detected in 11 of 16CRS patient specimens (69% prevalence) and three offive (60%) non-CRS controls. Although the prevalence ofS aureus sequence detection in the cases and controlswas similar, the abundance of S aureus (when present)in CRS patients was higher than in non-CRS patients(mean, 38% vs. 4.6% S aureus sequences; range, 0.7–95.7% vs. 0.2–13.3%). This trend did not reach statisticalsignificance (P ¼ .07), due at least in part to the smallpopulation size (15 cases and 5 controls).

To better understand the distribution and abun-dance of S aureus in the middle meatus of CRS and non-CRS populations, swabs from 22 additional patients andcontrols, along with the initial sample set, were screenedusing a quantitative PCR method specific for S aureusDNA (i.e., enumeration of femA genes). Nine samples(from six CRS cases and three controls) were excludeddue to failure of PCR amplification with pan-bacterialprimers, indicating low biomass. A total of 34 PCR-posi-tive specimens were analyzed: 26 CRS and eight non-CRS. S aureus DNA was detected in 19 of 26 (73.1%)CRS cases (including all 10 culture-positive samples)with an average of 14.8% abundance (range, 0.003–100%), when normalized to the total bacterial signal. Incontrol patients, two of eight (25%) had S aureus ampli-fication, with an average of 0.4% abundance (range,0.06–0.75%). This larger patient population resulted insignificant statistical differences between CRS patientsand controls for both prevalence (P ¼ .03) and meanabundance (P ¼ .03) of S aureus in middle meatusspecimens.

For cases in which S aureus was detected by DNAsequencing, a further nonquantitative PCR assay for a Saureus-specific methicillin-resistant gene (mecA) wasconducted. This assay detected mecA in one specimen(patient 2), the only patient diagnosed with MRSA byculture.

To identify clinical factors influencing the observeddifferences in microbial community diversity and S aureusabundance between CRS cases and controls, we per-formed univariate ANOVA for several candidatedemographic factors (Table II). In this analysis, antibi-otic history and asthma were the most significantfactors associated with biodiversity (both richness andevenness) and S aureus abundance. Both antibiotic expo-sure and asthma were associated with reduced microbialdiversity and increased S aureus abundance, as

TABLE II.Clinical and Microbiological Factors Associated With Middle Meatus Biodiversity by Univariate Analysis*

Diagnosis† Allergies Antibiotics AsthmaPriorESS Pus

Staphylococcusaureus‡

Richness .23 .71 .06 .01 .05 .22 .005

Evenness .09 0.35 .02 .01 .19 .55 .0008

Staphylococcus aureus‡ .15 .19 .08 .001 .12 .68 —

*Table summarizes P values from univariate analysis of variance. Columns are predictor factors and rows are outcome variables.†Chronic rhinosinusitis (CRS) versus non-CRS.‡Percent abundance determined by Staphylococcus aureus 16S rRNA pyrosequencing reads, normalized to total reads per specimen.ESS ¼ endoscopic sinus surgery.


470

measured by specific QPCR. Because S aureus abun-dance also was strongly correlated with both richnessand evenness, it is likely that the effects of antibioticsand asthma on biodiversity were mediated by their pro-motion of higher S aureus abundances in the middlemeatus.

DISCUSSIONThe results of our microbiological analysis of middle

meatus specimens from CRS and non-CRS patients indi-cate that clinical culture and culture-independentanalysis of 16S rRNA pyrosequences are broadly consist-ent, but that the sequence-based approach both providesfiner phylogenetic resolution to identify microorganismsand detects greater biodiversity than does culture.Although sequencing generally identified the samepathogens as culture, the converse was not the case, asculture failed to detect potentially pathogenic membersof many genera (Fig. 1). Most notably, the genus Steno-trophomonas, a group of difficult-to-treat, biofilm-forming microbes related to Pseudomonas spp, wasobserved at high relative abundance in specimens fromboth of the fungal non-CRS cases (patients 8 and 16), butwas not cultivated from either individual. Additionally,the diversity of anaerobic genera (e.g., Propionibacterium,Prevotella) in particular was not demonstrated by cultiva-tion, with most specimens reported as containing mixedanaerobes. Thus, these cases demonstrate the utility ofDNA detection in conjunction with cultivation for theidentification of difficult-to-culture bacterial pathogens.

In three of the four discordant events in which cul-tured isolates were not detected by pyrosequencing, thecause may have been inaccurate identification ofisolates, because other morphologically and/or physio-logically similar species were identified by sequence.However, in the case of one isolate, Citrobacter freundii,failure of detection/identification by DNA sequencing ispossible because no other citrate-using genera weredetected in the sequence library. This could occur if theisolate were a relatively rare member of the sinonasalmicrobiota and the depth of pyrosequencing was insuffi-cient to detect such rare microbes. Alternatively, thiscould be a case of contamination during cultivation.Despite these few discordant isolates, sequencing gener-ally identified the cultured species in addition to manyspecies not detected by cultivation.

Comparison of rRNA gene sequences from CRScases and non-CRS controls revealed that middle meatusmicrobiomes differed significantly in composition. Thesecommunity-level differences could result from repeatedantibiotic therapies or pathogen-induced alterations inthe normal microbiota of the middle meatus (i.e., dysbio-sis). Antibiotic treatment in the 12 weeks prior tosurgery reduced genus-level richness significantly, andspecimens from these patients had more uneven com-munities, suggesting that a few organisms werenumerically dominant in the community. Because of thechronic nature of CRS, antibiotic therapy is likely to bea significant driver of change in the sinonasal micro-biomes of CRS patients. One species in particular, S

aureus, was highly prevalent and abundant in CRSpatients, whereas it was a minor component in non-CRScontrols. These results clearly warrant further investiga-tion to assess their generalizability to a wider patientpopulation. The presence of S aureus is noteworthy inCRS due to the increasing incidence and awareness ofMRSA carriage in the nares, which is a significant con-tributor to surgical site infections and slow mucosalhealing following ESS as a result of biofilm forma-tion.24,25 This study population had only one MRSA case(confirmed by PCR), which was successfully treated in aculture-directed manner of combined oral and topicalantibiotic therapy.

The use of middle-meatus swabs for DNA-basedbacterial assays is appropriate for detection of multiplebacterial species, including anaerobes, which may beundetected when swabs are used solely for culture.26

Based on available cultivation-based studies, the micro-biology of the middle meatus correlates well with that ofthe sinuses, however these experiments have not beenrepeated with pyrosequencing technology.9 This is apotential weakness of the current study, and sequence-based analyses currently are underway. Furthermore,the bacterial communities observed in this study of themiddle meatus, both by culture and DNA sequence, dif-fered from those previously reported in studies of theanterior nares.27 For example, representatives of thegenus Corynebacterium, a common microbe of the skinand anterior nares, were cultivated only once from themiddle meatus, and comprised only �3% of total sequen-ces versus the 8% to 20% reported in our previous studyof the nares.27 This indicates that swabs of the anteriornares would not be appropriate as a replacement of mid-dle meatus swabs in research studies or clinicalinvestigations of CRS microbiology, as has been noted inthe literature previously,28 and swabs were not contami-nated by microbiota of the anterior nares duringinsertion/retraction from the middle meatus or sinuses.

Although sequencing may be superior to clinicalculture in some respects, the method currently requiresmore time and labor than does microbial culture, thuslimiting its clinical applicability. In contrast, QPCRassay of specific pathogens is very rapid (several hours)and comparably sensitive to culture, but is limited todetecting known species. A potential solution to thisproblem would be the design of a PCR panel assay ormicroarray for all suspected CRS pathogens, thus givingthe benefits of broad-range DNA detection with thenecessary rapidity for clinical applications. Finally,because antibiotic sensitivity testing in most practicalsituations currently requires testing of axenic cultivars,clinical culture is unlikely to be replaced by purely mo-lecular methods in the near term.

CONCLUSIONThe results of this preliminary study indicate that

bacterial culture and 16S rRNA gene pyrosequencing ofspecimens from CRS patients and controls yield largelycongruent results, although significantly more diversity,particularly of anaerobic groups, was observed by


471

sequencing than culture. Additionally, CRS patients hadsignificantly different microbial communities than non-CRS controls, with more abundant occurrence of S aur-eus. Prior antibiotic therapy may be a driver of theobserved microbiome alterations. Taken together, theseresults indicate that complex bacterial communities mayplay a significant role in the etiology of CRS.

AcknowledgmentsWe thank Dr. Sergio Pereira (University of Toronto)

for assistance with pyrosequencing, and Dr. Todd Kingdom(University of Colorado-Denver) for helpful editorial com-ments that strengthened the article.

BIBLIOGRAPHY

1. Meltzer EO, Hamilos DL, Hadley JA, et al. Rhinosinusitis: establishingdefinitions for clinical research and patient care. J Allergy Clin Immu-nol 2004;114(6 suppl):155–212.

2. Ooi EH, Wormald P-J, Tan LW. Innate immunity in the paranasal sinuses:a review of nasal host defenses. Am J Rhinol 2008;22:13–19.

3. Foreman A, Psaltis AJ, Tan LW, Wormald P-J. Characterization of bacte-rial and fungal biofilms in chronic rhinosinusitis. Am J Rhinol Allergy2009;23:556–561.

4. Tan L, Psaltis A, Baker LM, et al. Aberrant mucin glycoprotein patternsof chronic rhinosinusitis patients with bacterial biofilms. Am J RhinolAllergy 2010;24:319–324.

5. Lewis K. Persister cells and the riddle of biofilm survival. Biochemistry(Mosc) 2005;70:267–274.

6. Brook I. Microbiology of acute and chronic maxillary sinusitis associatedwith an odontogenic origin. Laryngoscope 2005;115:823–825.

7. Bhattacharyya N. Bacterial infection in chronic rhinosinusitis: a controlledpaired analysis. Am J Rhinol 2005;19:544–548.

8. Stephenson M-F, Mfuna L, Dowd SE, et al. Molecular characterization ofthe polymicrobial flora in chronic rhinosinusitis. J Otolaryngol HeadNeck Surg 2010;39:182–187.

9. Feazel LM, Frank DN, Ramakrishnan VR. Update on bacterial detectionmethods in chronic rhinosinusitis: implications for clinicians andresearch scientists. Int Forum Allergy Rhinol. DOI:10.1002/alr.20071.

10. Rosenfeld RM, Andes D, Bhattacharyya N, et al. Clinical practice guide-line: adult sinusitis. Otolaryngol Head Neck Surg 2007;137(3 suppl):S1–S31.

11. Sistek D, Tschopp J-M, Schindler C, et al. Clinical diagnosis of currentasthma: predictive value of respiratory symptoms in the SAPALDIAstudy. Eur Respir J 2001;17:214–219.

12. Nadkarni MA, Martin FE, Jacques NA, Hunter N. Determination of bacte-rial load by real-time PCR using a broad-range (universal) probe andprimers set. Microbiology 2002;148:257–266.

13. Francois P, Pittet D, Bento M, et al. Rapid detection of methicillin-resist-ant Staphylococcus aureus directly from sterile or nonsterile clinicalsamples by a new molecular assay. J Clin Microbiol 2003;41:254–260.

14. Huletsky A, Giroux R, Rossbach V, et al. New real-time PCR assay forrapid detection of methicillin-resistant Staphylococcus aureus directlyfrom specimens containing a mixture of staphylococci. J Clin Microbiol2004;42:1875–1884.

15. Frank DN, Feldman RA. Molecular-phylogenetic strategies for characteri-zation of uncultured pathogens. In: Tibayrenc M, ed. Encyclopedia of In-fectious Diseases: Modern Methodologies. Hoboken, NJ: Wiley-Liss;2007. p. 57–68. DOI:10.1002/9780470114209.ch4.

16. Lane D. 16S/23S DNA sequencing. In: Stackebrandt E, Goodfellow M, eds.Nucleic Acid Techniques in Bacterial Systematics. New York, NY: JohnWiley and Sons; 1991: 115–175.

17. Frank DN. BARCRAWL and BARTAB: software tools for the design andimplementation of barcoded primers for highly multiplexed DNAsequencing. BMC Bioinformatics 2009;10:362.

18. Wang Q, Garrity GM, Tiedje JM, Cole JR. Naı̈ve Bayesian classifier forrapid assignment of rRNA sequences into the new bacterial taxonomy.Appl Environ Microbiol 2007;73:5261–5267.

19. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. Basic local align-ment search tool. J Mol Biol 1990;215:403–410.

20. Pruesse E, Quast C, Knittel K, et al. SILVA: a comprehensive onlineresource for quality checked and aligned ribosomal RNA sequence datacompatible with ARB. Nucleic Acids Res 2007;35:7188–7196.

21. Frank D. XplorSeq: a software environment for integrated managementand phylogenetic analysis of metagenomic sequence data. BMC Bioinfor-matics 2008;9:420.

22. Development Core Team. R: a language and environment for statistical com-puting. Vienna, Austria: Institute for Statistics and Mathematics;2008:1(10).

23. Oksanen J, Simpson G, Solymos P. vegan. 2008. Available at: http://vegan.r-forge.r-project.org/. Accessed August 8, 2011.

24. Parvez N, Jinadatha C, Fader R, et al. Universal MRSA nasal surveil-lance: characterization of outcomes at a tertiary care center and impli-cations for infection control. South Med J 2010;103:1084–1091.

25. Jervis-Bardy J, Foreman A, Field J, Wormald PJ. Impaired mucosal heal-ing and infection associated with Staphylococcus aureus after endo-scopic sinus surgery. Am J Rhinol Allergy 2009;23:549–552.

26. Mantovani K, Rodrigues Dde O, Tamashiro E, et al. Comparing differentmeth-ods used to collect material for a microbiological evaluation of patients withchronic rhinosinusitis. Braz J Otorhinolaryngol 2010;76:321–325.

27. Frank DN, Feazel LM, Bessesen MT, et al. The human nasal microbiotaand Staphylococcus aureus carriage. PLoS ONE 2010;5:e10598.

28. Nadel DM, Lanza DC, Kennedy DW. Endoscopically guided cultures inchronic sinusitis. Am J Rhinol 1998;12:233–241.


472

Copyright of Laryngoscope is the property of Wiley-Blackwell and its content may not be copied or emailed to

multiple sites or posted to a listserv without the copyright holder's express written permission. However, users

may print, download, or email articles for individual use.

microbiome complexity and staphylococcus aureus in crs

Documents