Open Forum Infectious Diseases MAJOR ARTICLE Molecular Identification of Bacteria in Intra-abdominal Abscesses Using Deep Sequencing 1,a 1,b 1,2,c 1 1 1,2 Andrew Kozlov, Lorenzo Bean, Emilie V. Hill, Lisa Zhao, Eric Li, and Gary P. Wang 1 2 Division of Infectious Diseases and Global Medicine, Department of Medicine, University of Florida College of Medicine, Gainesville, Florida; Infectious Diseases Section, Medical Service, North Florida/South Georgia Veterans Health System, Gainesville, Florida Background. Intra-abdominal abscesses are localized collections of pus, which generally arise from a breach in the normal mucosal defense barrier that allows bacteria from gastrointestinal tract, and less commonly from the gynecologic or urinary tract, to induce inflammation, resulting in an infection. The microbiology of these abscesses is usually polymicrobial, associated with the primary disease process. However, the microbial identity, diversity and richness in intra-abdominal abscesses have not been well characterized, due in part to the difficulty in cultivating commensal organisms using standard culture-based techniques. Methods. We used culture-independent 16S rRNA Illumina sequencing to characterize bacterial communities in intra-abdom- inal abscesses collected by percutaneous drainage. A total of 43 abscess samples, including 19 (44.2%) Gram stain and culture-nega- tive specimens, were analyzed and compared with results from conventional microbiologic cultures. Results. Microbial composition was determined in 8 of 19 culture-negative samples and 18 of 24 culture-positive samples, iden- tifying a total of 221 bacterial taxa or operational taxonomic units (OTUs) and averaging 13.1 OTUs per sample (interquartile range, 8–16.5 OTUs). Microbial richness for monomicrobial and polymicrobial samples was significantly higher than culture-negative samples (17 and 15.2 OTUs vs 8 OTUs, respectively), with a trend toward a higher microbial diversity (Shannon diversity index of 0.87 and 1.18 vs 0.58, respectively). Conclusions. e b Th acterial consortia identified by cultures correlated poorly with the microbial composition determined by 16S rRNA sequencing, and in most cases, the cultured isolates were minority constituents of the overall abscess microbiome. Intra- abdominal abscesses were generally polymicrobial with a surprisingly high microbial diversity, but standard culture-based techniques failed to reveal this diversity. es Th e data suggest that molecular-based approaches may be helpful for documenting the presence of bacteria in intra-abdominal abscesses where standard cultures are unrevealing, particularly in the setting of prior antibiotic exposure. Keywords. Intra-abdominal abscesses; 16S RNA sequencing; Illumina sequencing; Microbial diversity; Bioinformatics. e de Th velopment of intra-abdominal abscesses is a consequence Recent studies suggest that molecular interactions within these of inflammatory responses to endogenous microflora that gain diverse communities may increase the virulence of known access to a normally sterile site, resulting in local inflammation pathogens in a synergistic manner  However, standard cul- and the formation of pus. Abscesses develop as a result of either ture-based studies have shown that only 11%–18% of intrac- direct extension of normal polymicrobial endogenous flora into erebral abscesses , 11%–40% of liver abscesses [7, 8], and a normally sterile body site or secondarily through perforation 44% of splenic abscesses  are polymicrobial. In some cases, or laceration. If left untreated, abscesses may lead to bacteremia antibiotic exposure prior to drainage of abscesses likely reduces and cause significant morbidity and mortality . Abscesses can the yield of bacteria recovery and/or alters the microbial profile arise at any location within the human body, and each abscess determined by cultures . collection is associated with unique characteristics. These e o Th ptimal management of large intra-abdominal and pel- abscesses are generally not associated with a single organism but vic abscesses is drainage followed by adjunctive antimicrobial reflect diverse ecological niches, whether in chronic wounds , therapy, which together achieve a success rate of up to 70%– periodontal disease , or pulmonary infections . 80% [11, 12]. Broad-spectrum antimicrobial therapy with activ- ities against Gram negatives and anaerobes is oen admin ft stered as empiric therapy , as the microbiology of intra-abdominal Received 15 November 2017; editorial decision 18 January 2018; accepted 20 January 2018. Present affiliations: Andrew Kozlov, University of Pennsylvania, Philadelphia, Pennsylvania; abscesses is thought to largely reflect the endogenous flora at b c Lorenzo Bean, Weill Medical College of Cornell, New York City, New York; Emilie Hill, body sites near the location of the abscesses. In the acidic envir- University of Texas Southwestern, Dallas, Texas onment of the stomach, which is hostile to most bacteria except L.B., A.K., and E.V.H. contributed equally to this work. Correspondence: G. P. Wang, M.D., Ph.D., Division of Infectious Diseases and Global for Helicobacter spp., the total bacterial counts are much lower in Medicine, Department of Medicine, University of Florida College of Medicine, 1600 SW Archer the stomach compared with the colon . However, the micro- Road, DSB D2-14E, Gainesville, FL 32610 (firstname.lastname@example.org). bial community composition and structure of intra-abdominal Open Forum Infectious Diseases Published by Oxford University Press on behalf of Infectious Diseases Society of America 2018. abscesses are oen n ft ot characterized in detail, in part because This work is written by (a) US Government employee(s) and is in the public domain in the US. cultivation and identification of anaerobic organisms are DOI: 10.1093/ofid/ofy025 Bacterial Composition of Intra-abdominal Abscesses • OFID • 1 Downloaded from https://academic.oup.com/ofid/article-abstract/5/2/ofy025/4823433 by Ed 'DeepDyve' Gillespie user on 16 March 2018 labor-intensive and some organisms may be uncultivable . dominated by anaerobic organisms, and that standard clin- u Th s, the diversity and microbial ecology of intra-abdominal ical cultures significantly underestimate their overall micro- abscesses remain poorly understood and have not been well bial diversity. Culture-based microbial composition correlated characterized, and the relative abundance of different microbes poorly with the abscess microbiome determined by 16S rRNA in abscesses occurring at various body sites is not known. sequencing, which may be helpful to confirm the presence of Furthermore, to what extent the standard microbiologic cul- bacteria in intra-abdominal abscesses where standard cultures tures correlate with microbiome composition determined by are unrevealing, especially in the setting of prior antibiotic culture-independent sequencing is not known. administration. Recent advances in sequencing technology have greatly con- METHODS tributed to our understanding of the human microbiome and have led to an increasing appreciation of the role of the endogenous Ethics Statement microflora in human biology [16–21]. Based on culture-based De-identified clinical specimens from the Clinical Microbiology methods, the number of bacterial species in abscesses has been Laboratory at University of Florida Health (Gainesville, FL) reported to vary from 2 to 6 . e Th predominant anaerobes were used for this study. The study was approved by University frequently cultured include the Bacteroides fragilis (B. fragilis) of Florida Institutional Review Board. group, Prevotella spp., Porphyromonas spp., Peptostreptococcus Samples and DNA Purification spp., and Clostridium spp., and the most commonly isolated aer- As part of routine clinical care for abscess samples received in obic and facultative bacteria are the family of Enterobacteriaceae anaerobe transport tubes, blood agar plate (BAP), chocolate and Enterococci spp. Brook and Frazier examined 52 intra-ab- (CHOC), MacConkey (MAC), bacteriodes bile esculin (BBE)/ dominal abscess specimens retrospectively and found that 34 kanamycin-vancomycin laked blood (KVLB), Brucella, and specimens (65%) were mixed anaerobic and aerobic infections anaerobic Thiglycolate broth were inoculated in the anaerobic and 47 (90%) were polymicrobial with an average of 3.7 isolates chamber and incubated at 37°C under anaerobic condition. per specimen (on average, 2.1 were anaerobes and 1.6 were fac- Two microscopic slides were made from the abscess samples ultative anaerobes or aerobes) . The most frequently cultured inside the anaerobic chamber, Gram stained, and microscop- anaerobes included Peptostreptococcus spp., B. fragilis group, ically examined to report the presence of polymorphonuclear Clostridium spp., and Prevotella spp., and the most commonly leukocytes and bacteria. The plates were examined daily for isolated aerobic and facultative bacteria included Escherichia coli, the presence of colonies over the course of 5 days. Facultative Enterococci spp., and Staphlococcus aureus. In a subsequent study, anaerobe and strict anaerobe were determined by compar- they analyzed 22 intra-abdominal abscesses from diverticu- ing the growth of specific bacteria under both aerobic and litis and found that 17 (77%) were mixed anaerobic and aerobic anaerobic conditions. Gram stains were carried out for each infections and 19 (86%) were polymicrobial, with an average of specific bacteria to determine bacterial morphology, color, 3.3 isolates per specimen (on average, 1.7 were anaerobes and 1.6 and spore-forming ability. Distinct colonies were subcultured were facultative anaerobes or aerobes) . The most frequently to obtain pure colonies for further anaerobe identification. cultured anaerobes were Bacteroides spp., Peptostreptococcus spp., Anaerobe identification was performed using RapID ANA and Clostridium spp., and the most commonly isolated aerobic and/or mass spectrometry. and facultative bacteria were E. coli and Streptococcus spp. It is For each sample, results of Gram stain and bacterial cul- now known that more than 5000 bacterial species reside in the ture and the body location of percutaneous drainage by the gastrointestinal tract  and more than 700 reside in the oral Interventional Radiology Service were recorded. Genomic cavity , which include many previously unidentified or un- DNA (gDNA) was extracted using the PSP Spin Stool DNA cultivable organisms. Thus, culture-based methods may under - Kit according to the manufacturer’s instructions (STRATEC estimate the number of organisms present in abscesses , Biomedical, Berlin, Germany). The concentration of purified raising the possibility that abscesses of endogenous origin may be DNA was quantified using a Nanodrop spectrophotometer more diverse than previously thought. (ThermoScientific, Carlsbad, CA). To our knowledge, the microbial composition of intra-ab- 16S rRNA Illumina Sequencing and Bioinformatics Analysis dominal abscesses has not been examined using culture-inde- The V1-V3 hypervariable region (~500 bp) of 16S rRNA gene pendent methods and directly compared with microbiologic segment was amplified using barcoded polymerase chain reac- cultures obtained for clinical indications. Here, we used 16S tion (PCR), and PCR products were gel purified, pooled, and rRNA deep sequencing to determine the microbial compos- paired-end sequenced at 2 × 300 bp using the MiSeq Reagent ition of intra-abdominal abscesses drained percutaneously Kit v3 on an Illumina MiSeq instrument. Methods for PCR, by interventional radiology and compared with Gram stain Illumina sequencing, and bioinformatic analysis are described and culture reported by clinical labs. We show that intra-ab- in detail in the Supplementary Data. dominal abscesses are generally polymicrobial in nature, 2 • OFID • Kozlov et al Downloaded from https://academic.oup.com/ofid/article-abstract/5/2/ofy025/4823433 by Ed 'DeepDyve' Gillespie user on 16 March 2018 RESULTS or (2) Gram stain demonstrated a single morphology but cul- ture was negative. We defined a sample as polymicrobial if (1) Sample Characteristics and 16S rRNA Sequence Analysis culture grew more than 1 microorganism or (2) Gram stain A total of 113 clinical samples drained by interventional demonstrated multiple distinct morphologies but culture radiologists at UF Health were available from the Clinical was negative. Thus, 16 of 24 (66.7%) were classified as pol- Microbiology Laboratory. We selected 43 samples that met ymicrobial and 8 (33.3%) were monomicrobial. A total of 18 the following criteria: (1) results of Gram stain and micro- Gram stain/culture-positive samples (13 polymicrobial and 5 biologic culture were available, (2) the sample was from an monomicrobial samples) yielded 16S rRNA gene amplification intra-abdominal location, and (3) the duration from the time products (75%). of sample collection to storage at −70°C was less than 72 All 26 samples that yielded amplification products were deep hours (all samples were stored at 4°C while undergoing cul- sequenced using the Illumina MiSeq platform, yielding 919 669 ture and sensitivity testing per clinical routine). A 72-hour reads aer fi ft ltering and quality control, with a mean of 30 656 threshold was selected based on an analysis that demonstrated reads per sample (interquartile range, 26 086–37 051). From 26 minimal variations in the relative proportions of major op- samples, we identified a total of 221 distinct operational taxo- erational taxonomic units (OTUs) at 4°C over 72 hours nomic units (mean, 13.1 OTUs per sample; interquartile range, (Supplementary Figure 1). Of the 43 samples, 24 (55.8%) 8–16.5 OTUs). The mean number of OTUs was 17 (range, were positive by Gram stain and/or culture and 19 (44.2%) 12–23) for monomicrobial samples and 15.2 (range, 2–32) for were negative by both Gram stain and cultures (Figure 1). All polymicrobial samples. In comparison, the mean number of 43 samples were subjected to V1-V3 16S rRNA gene ampli- OTUs for Gram stain/culture-negative samples was 8 (range, fication by barcoded PCR. Of these, 17 (40%) failed repeat- 2–18) (Figure 2A). The microbial richness (ie, the number of edly to yield an amplification product—these included 6 OTUs) for monomicrobial and polymicrobial samples was sig- Gram stain/culture-positive samples and 11 Gram stain/cul- nificantly higher than culture-negative samples (17 and 15.2, ture-negative samples. Thus, these 17 samples were excluded respectively, vs 8; P < .05, unpaired Student t test). Combining from subsequent analysis. As samples have been previously monomicrobial and polymicrobial samples, microbial rich- de-identified, patient-level data including antibiotic history ness for Gram stain/culture-positive samples was significantly were not available for analysis. Culture results for the 6 Gram higher than for Gram stain/culture-negative samples (15.7 vs 8; stain/culture-positive samples that failed PCR amplification P < .05, unpaired Student t test). No significant difference was are shown in Supplementary Table 1. observed between monomicrobial and polymicrobial samples. Microbial diversity for both monomicrobial and polymi- Comparison of 16S rRNA Gene Sequencing and Bacterial Culture crobial samples was higher than for culture-negative samples To compare 16S rRNA gene sequencing with conventional cul- (0.87 and 1.18, respectively, vs 0.58). However, these differences tures, we first defined a Gram stain/culture-positive sample as were not statistically significant. Combining monomicrobial monomicrobial if (1) culture yielded a single microorganism 43Abscess Samples 24 Gram stain and/or culture-positive 19 Gram stain and culture-negative 16 polymicrobial 8 monomicrobial 3 PCR 13 PCR 3 PCR 5 PCR 8 PCR 11 PCR negative positive negative positive positive negative 26 samples sequenced and analyzed Figure 1. Samples analyzed in this study. A total of 43 clinical samples were identified that met the inclusion criteria. These included 19 samples (44%) that were negative by both Gram stain and culture and 24 samples (56%) that were positive by Gram stain and/or culture. All 43 samples were subjected to V1-V3 16S rRNA gene amplification. Of the 19 Gram stain/culture-negative samples, 8 samples (42%) were successfully amplified and sequenced using Illumina. Of the 24 Gram stain/culture-positive samples, 18 samples (75%) were amplified and sequenced. Bacterial Composition of Intra-abdominal Abscesses • OFID • 3 Downloaded from https://academic.oup.com/ofid/article-abstract/5/2/ofy025/4823433 by Ed 'DeepDyve' Gillespie user on 16 March 2018 * Clostridium spp. , 5% Sutterella spp. ,7% Escherichia Clostridium spp. , perfringens, 61% 14% Escherichia coli, 10% Caulobacter subvibrioides Monomicrobial Polymicrobial Culture-negative Enterococcus faecium, 96% 2.0 Figure 3. Microbial composition and read abundance, determined by 16S rRNA sequencing in representative monomicrobial samples. (A) Gram stain of this sample showed 4+ polymorphonuclear leukocytes (PMNs) and 2+ Gram (+/-) rods, and 1.5 Escherichia coli was isolated by culture. The dominant operational taxonomic units (OTUs) mapped to Escherichia spp. and E. coli. All organisms with ≥1% frequency were anaerobes or facultative anaerobes. (B) Gram stain of this sample was nega- 1.0 tive, but subculture of the specimen grew coagulase-negative Staphlococci. The dominant OTU was Enterococcus faecium. Blue pie slices denote concordance be- tween 16S rRNA sequencing data and microbiologic cultures, and the pie slices in shades of gray indicate organisms with their read abundance not identified by 0.5 culture. 0.0 detected in high abundance, and several organisms with ≥1% Monomicrobial Polymicrobial Culture-negative in read abundance were mapped to anaerobes or facultative anaerobes. In contrast, for 2 of the 5 monomicrobial samples Figure 2. Microbial richness and diversity by subgroups. (A) The number of op- erational taxonomic units (microbial richness), determined by 16S rRNA sequenc- (Figure 3B; Supplementary Figure 2C), the cultured isolates ing for each of the 3 groups (classified according to culture results), is shown on were not the most abundant organisms in the abscess col- the y-axis. (B) Shannon index (microbial diversity) is shown on the y-axis. Mean lections. For example, in a perihepatic abscess (Figure 3B), Shannon indices were compared. The asterisk indicates P < .05 (unpaired Student t test). coagulase-negative staphylococci (CoNS) was identified by culture (which was likely a skin contaminant). However, 16S rRNA gene sequencing revealed a total of 15 OTUs including and polymicrobial samples, microbial diversity for Gram stain/ Enterococcus faecium (95.5%) and Caulobacter subvibrioides culture-positive samples was higher than for Gram stain/cul- (2.6%) as the dominant OTUs, with Staphylococcus epider- ture-negative samples (1.09 vs 0.58; P = .08) but was not statis- midis, a CoNS, as a minority species, constituting 0.08% of tically significant (Figure 2B). total sequence reads. Monomicrobial Specimens Were Polymicrobial 16S rRNA gene sequencing identified a diverse community Organisms Isolated by Culture Were Generally Minority Populations in Polymicrobial Samples profile in all 5 monomicrobial samples (Table S2), harbor- Overall, the number of OTUs in polymicrobial samples ing many OTUs in each sample (Figure 3; Supplementary (Supplementary Table 3) was not significantly different from Figure 2). The organisms identified by microbiologic culture that of monomicrobial samples (mean, 15.23 vs 17). However, were the dominant OTUs in 3 of the 5 samples (Figure 3A; the range of OTUs varied widely (range, 2–32) (Figure 2A). In Supplementary Figure 2A and B). For example, E. coli iden- contrast to monomicrobial samples, the organism identified by tified by culture in an intra-abdominal abscess (Figure 3A) culture was the dominant OTU in only 1 of 13 polymicrobial was the dominant OTU (71.1%). Interestingly, Clostridium samples (Figure 4A). For the remaining 12 samples, the cultured perfringens (13.7%) and other Clostridium spp. were also 4 • OFID • Kozlov et al Downloaded from https://academic.oup.com/ofid/article-abstract/5/2/ofy025/4823433 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Shannon index Number of OTUs Staphylococcus Bacteroides spp. (eg, Figure 4B; Supplementary Figure 3A–K). Enterococcus spp. spp. , Sequences belonging to the genera Staphylococcus spp., 20% Streptococcus spp., the family Enterobacteriaceae, and other aer- obic Gram-negatives were generally a minority population. For Enterococcus example, E. coli or Enterococcus spp. were identified by culture faecium, in 11 of 13 samples, but either constituted a very small minority 77% of the overall microbial population or were not detected at all by deep sequencing (Supplementary Figure 3). Prevotella spp. Lower Microbial Richness and Diversity in Gram Stain/Culture-Negative Dialister pneumosintes Prevotella denticola, Samples 6% Compared with Gram stain/culture-positive samples, Gram Lactobacillus gasseri, 69% stain/culture-negative samples (Table S4) had a lower number of OTUs (Figure 2). Of the 8 Gram stain/culture-negative Prevotella samples, 5 had a dominant OTU that represented >95% of all veroralis, 16% 16S rRNA sequences—these were Streptococcus spp., B. fra- gilis, Parvimonas spp., Prevotella bivia, and Peptoniphilus spp. Figure 4. Microbial composition and read abundance, determined by 16S rRNA sequencing in polymicrobial samples. (A) Gram stain showed 4+ Gram-positive (Figure 5A; Supplementary Figure 4A–D). The remaining 3 cocci in pairs and chains, 2+ Gram-negative rods, and few Gram-positive rods. samples harbored a single dominant OTU with an abundance Enterococcus spp. and Candida albicans were isolated by cultures. The dominant ranging from 47% to 69%—1 was Fusobacterium (Figure 5B), operational taxonomic units (OTUs) mapped to Enterococcus faecium. (B) Gram stain showed 4+ polymorphonuclear cells and 2+ Gram-variable rods. Escherichia a Gram-negative anaerobe, and 2 belonged to Streptococcus coli was isolated by culture. The dominant OTUs were Lactobaccilus gasseri and spp. (Supplementary Figure 4E and F). Overall, aerobic Gram- Prevotella spp. Blue pie slices denote concordance between 16S rRNA sequencing negative bacilli such as E. coli and Enterobacter comprised a data and microbiologic cultures, and pie slices in shades of gray indicate organisms small proportion of all 16S rRNA gene sequences in Gram stain/ not identified by culture. OTUs with ≥5% abundance are labeled on the pie chart. OTUs with <5% but ≥1% abundance are shown in the side bar. culture-negative samples (Figures 5; Supplementary Figure 4). DISCUSSION organisms were minority populations in the abscess collections, and the dominant OTUs often mapped to Lactobacillus spp. The present study reports the microbial composition (deter- or anaerobes such as Prevotella spp., Fusobacterium spp., and mined by 16S rRNA sequencing) in abscess samples drained Staphylococcus spp. Prevotella bivia 96% Parvimonas spp. Actinomyces spp. Atopobium spp. Prevotella oralis, 14% Granulicetella spp. Prevotellatannerae Streptococcus spp. Streptococcus spp.2 Streptococcus intermedius, Gemella spp. 16% Fusobacterium spp., 52% Figure 5. Microbial composition and read abundance determined by 16S rRNA sequencing in Gram stain and culture-negative samples. (A) The sample was negative by Gram stain and culture, but 16S rRNA sequencing analysis revealed a single dominant operational taxonomic unit (OTU). (B) The sample was negative by Gram stain and culture but sequencing analysis identified 11 OTUs with ≥1% read abundance. All 11 OTUs corresponded to organisms that were anaerobes or facultative anaerobes. OTUs with ≥5% abundance are labeled on the pie chart, and OTUs with <5% but ≥1% abundance are shown in the side bar. Bacterial Composition of Intra-abdominal Abscesses • OFID • 5 Downloaded from https://academic.oup.com/ofid/article-abstract/5/2/ofy025/4823433 by Ed 'DeepDyve' Gillespie user on 16 March 2018 percutaneously and compares with microbiologic cultures responsible for a monomicrobial infection, our sequence data indi- performed for clinical indications. We found that culture-pos- cate a polymicrobial composition. Thus, our results suggest that cul- itive samples had a significantly higher number of OTUs than ture-independent methods may be useful to confirm the presence culture-negative samples, with a trend toward a higher micro- of bacteria in select clinical settings where prior antibiotic usage pre- bial diversity. Of the 5 samples that grew a single organism by cludes successful culturing of organisms from clinical samples. culture, sequencing analysis yielded the same organism as the Our data suggest that culture-based approaches may overesti- dominant population in 3 samples, but other bacteria in lower mate the prevalence of monomicrobial infections, missing much of abundance were also detected. In contrast, in the majority (12 the hidden microbial diversity in abscess collections . Similarly, of 13) of samples that were polymicrobial by culture, the cul- previous studies have suggested the presence of polymicrobial com- tured organisms were not the dominant OTUs. These results position in samples with negative culture results . Here we have suggest that microbial richness and diversity of intra-abdom- demonstrated the ability of 16S rRNA sequencing to characterize inal abscesses are higher than previously thought. Using cul- microbial compositions in both culture-positive and culture-nega- ture-independent 16S rRNA gene sequencing, many taxa that tive samples. All 26 samples that were sequenced had more than 1 were not cultivated were identified, revealing the overall abscess distinct taxa (mean, 13.1 OTUs per sample), and the overall compos- microbiome not appreciated using culture-based techniques. In ition likely more closely resembled the underlying microbial diver- most cases, the consortia of cultured bacteria were not the most sity present in these abscesses. An appreciation of this biodiversity abundant bacteria identified by 16S rRNA gene sequencing. and how the organisms interact with each other, or the “superorgan- However, the data should be interpreted with caution given the ism” , is essential for understanding the development of abscess lack of patient-level data and antibiotic history. formation and ultimately how these organisms interact with the host In samples with positive Gram stain and/or culture, especially to contribute to the pathophysiology of disease. Interspecies com- those that grew multiple organisms in culture, we found that cul- munication has been shown to serve as a prerequisite for certain tured isolates were generally minority constituents of the overall bacterial infections, and pathogen growth in the presence of specific community that is dominated by anaerobes and facultative bacterial niches could modulate gene expression to a more virulent anaerobes. The cultured isolates were frequently aerobic Gram- phenotype . Additionally, minor communities undetected in negatives such as E. coli and Klebsiella spp., but other organ- traditional cultures could contribute to pathogen persistence aer ft isms such as Pseudomonas spp., Enterobacter spp., and group B clinical treatment by releasing antibiotic-inactivating proteins, con- Streptococcus were also observed. This is not surprising as these ferring protection to the community as a whole [33, 34]. organisms generally grow well in cultures and likely outcompete e s Th tandard of care for patients with large intra-abdominal strict anaerobes that may be difficult to preserve during trans- abscesses is percutaneous drainage followed by targeted anti- port and require special care to cultivate in the laboratory, which biotic therapy guided by culture and susceptibilty. However, if are oen n ft ot achievable in standard clinical microbiology labs. standard cultures are unrevealing, treatment with broad-spec- Interestingly, sequences corresponding to these cultured organ- trum antimicrobials is oen im ft plemented, which could lead to isms were oen det ft ected at very low frequencies or not detected undesirable consequences such as antibiotic resistance and com- at all, suggesting that Enterobacteriaceae and Streptococci com- plications such as Clostridium dic ffi ile infection. Knowledge of monly isolated from intra-abdominal abscesses may constitute a the genus or species of bacteria most likely responsible for clin- very small minority of the overall bacterial population. ical disease or virulence of the “superorganism” may allow for Interestingly, samples that were negative by Gram stain and cul- selection of narrower-spectrum antimicrobial agents. Similarly, ture showed significant biodiversity. However, the total number of targeted treatment against an organism isolated by culture may bacterial taxa was generally lower compared with Gram stain/cul- not be appropriate, as the isolated organisms may not be the pri- ture-positive samples. Notably, the prevalence and abundance of mary pathogen. Animal studies of intra-abdominal sepsis have members of the phylum Proteobacteria, including E. coli, which is suggested that coliforms contribute to early sepsis while anaer- known to co-enrich other known pathogens , were low. It’s likely obes are implicated in later stages of abscess formation . This that these samples were obtained from patients who were receiving is consistent with our sequencing results, which showed a dom- antibiotics at the time of abscess collection. Unfortunately, the asso- inance of a variety of uncultured anaerobic organisms. ciated clinical data and antibiotic exposure history were not avail- Culture-independent methods are limited by the inability able. Sample 94 was of particular interest (Supplementary Table 4 to determine antimicrobial susceptibility, leaving the clinician and Supplementary Figure 4F). This specimen was a liver abscess to make treatment decisions based on known microbe char- and was Gram stain/culture-negative. 16S rRNA sequencing ana- acteristics and local antibiogram. Additionally, the clinical lysis revealed a polymicrobial composition dominated by S. interme- significance of bacterial taxa identified by sequencing is diffi- dius, in addition to E. coli and Enterobacter spp. While organisms cult to deduce. Further studies are necessary to understand the in the S. anginosus group (eg, S. intermedius) are oen a ft ssociated dynamic interplay within bacterial communities during abscess with intra-abdominal abscesses including the liver and could be formation. Moreover, sequencing may lend itself to potential 6 • OFID • Kozlov et al Downloaded from https://academic.oup.com/ofid/article-abstract/5/2/ofy025/4823433 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Acknowledgements artifacts of amplifying contaminating environmental microbes We thank members of Wang laboratory for helpful discussion. as can occur in clinical cultures following percutaneous access. Financial support. This work was supported by the Gatorade Trust In our study, we have focused our discussion on taxa that were through funds distributed by the University of Florida, Department of present at ≥1% relative read abundance. Medicine, and in part by the Infectious Diseases Society of America med- ical scholars program to A.K. As this study used universal primers for bacterial small subunit Potential conifl cts of interest. All authors: no reported conflicts of ribosomal RNA, fungal pathogens could not be identified. In 2 interest. All authors have submitted the ICMJE Form for Disclosure of cases, conventional culture identified Candida albicans , 1 with Potential Conflicts of Interest. Conflicts that the editors consider relevant to concomitant normal gut flora and the other with Enterococcus spp. the content of the manuscript have been disclosed. Sequencing of these samples identified multiple bacterial species. References In future studies, fungal rDNA PCR could be used where pyogenic 1. Kaplan GG, Gregson DB, Laupland KB. Population-based study of the epidemi- fungal infection is suspected or when both conventional methods ology of and the risk factors for pyogenic liver abscess. Clin Gastroenterol Hepatol 2004; 2:1032–8. and 16S rRNA sequencing fail to identify a putative pathogen. 2. Han A, Zenilman JM, Melendez JH, et al. The importance of a multifaceted Illumina sequencing of the bacterial 16S rRNA gene provides approach to characterizing the microbial flora of chronic wounds. Wound Repair a semiquantitative measure of microbial compositions in abscess Regen 2011; 19:532–41. 3. Mendes L, Azevedo NF, Felino A, Pinto MG. Relationship between invasion of collections that cannot be deduced using culture-based methods. the periodontium by periodontal pathogens and periodontal disease: a systematic Conventional diagnostics require the presence of viable organ- review. Virulence 2015; 6:208–15. 4. Dickson RP, Erb-Downward JR, Prescott HC, et al. Analysis of culture-dependent isms, which could be compromised by antibiotic use. Moreover, versus culture-independent techniques for identification of bacteria in clinically the drainage and sampling approach, the transit time from the obtained bronchoalveolar lavage fluid. J Clin Microbiol 2014; 52:3605–13. 5. Murray JL, Connell JL, Stacy A, et al. Mechanisms of synergy in polymicrobial bedside to the laboratory, conditions of specimen storage (ie, infections. J Microbiol 2014; 52:188–99. anaerobic storage), the presence of fastidious microorganisms 6. Al Masalma M, Armougom F, Scheld WM, et al. The expansion of the microbi- ological spectrum of brain abscesses with use of multiple 16S ribosomal DNA (eg, obligate intracellular bacteria or obligate anaerobes), and sequencing. Clin Infect Dis 2009; 48:1169–78. antibiotic therapy can all ae ff ct culture yield and biochemical 7. Chen SC, Tsai SJ, Chen CH, et al. Predictors of mortality in patients with pyogenic tests significantly more than culture-independent approaches liver abscess. Neth J Med 2008; 66:196–203. 8. Alvarez JA, González JJ, Baldonedo RF, et al. Single and multiple pyogenic liver [36, 37]. In contrast, the 16S rRNA gene acts as a molecular bar- abscesses: etiology, clinical course, and outcome. Dig Surg 2001; 18:283–8. code, allowing taxa-specific identification of bacteria without 9. Ferraioli G, Brunetti E, Gulizia R, et al. Management of splenic abscess: report on 16 cases from a single center. Int J Infect Dis 2009; 13:524–30. the need for culture. Thus, when culture-based methods fail, 10. Mancini N, Carletti S, Ghidoli N, et al. The era of molecular and other non-cul- metagenomic approaches may be a reasonable alternative for ture-based methods in diagnosis of sepsis. Clin Microbiol Rev 2010; 23:235–51. 11. Cinat ME, Wilson SE, Din AM. Determinants for successful percutaneous image- identifying fastidious or uncultivable organisms . On the guided drainage of intra-abdominal abscess. Arch Surg 2002; 137:845–9. other hand, we note that 17/43 (40%) specimens failed to yield 12. Robert B, Chivot C, Fuks D, et al. Percutaneous, computed tomography-guided drainage of deep pelvic abscesses via a transgluteal approach: a report on 30 cases PCR amplication products. Among them, 11 were Gram stain/ and a review of the literature. Abdom Imaging 2013; 38:285–9. culture-negative samples that likely harbored no bacteria at the 13. Solomkin JS, Mazuski JE, Bradley JS, et al. Diagnosis and management of complicated time the samples were collected. Of the other 6 Gram stain/cul- intra-abdominal infection in adults and children: guidelines by the Surgical Infection Society and the Infectious Diseases Society of America. Clin Infect Dis 2010; 50:133–64. ture-positive samples, the presence of PCR inhibitors may have 14. Pei Z, Bini EJ, Yang L, et al. Bacterial biota in the human distal esophagus. Proc contributed to the failure in PCR amplification. Thus, while Natl Acad Sci U S A 2004; 101:4250–5. 15. Fenollar F, Raoult D. Molecular diagnosis of bloodstream infections caused by the sequence-based approach may be useful, technical or sam- non-cultivable bacteria. Int J Antimicrob Agents 2007; 30(Suppl 1):S7–15. ple-specific factors may be a limitation in select samples. 16. Sibley CD, Church DL, Surette MG, et al. Pyrosequencing reveals the complex polymi- crobial nature of invasive pyogenic infections: microbial constituents of empyema, liver In summary, the current study uncovered a hidden microbial abscess, and intracerebral abscess. Eur J Clin Microbiol Infect Dis 2012; 31:2679–91. diversity in intra-abdominal abscesses and suggests that conventional 17. Woo PC, Lau SK, Teng JL, et al. Then and now: use of 16S rDNA gene sequenc- ing for bacterial identification and discovery of novel bacteria in clinical micro- culture-based methods may selectively isolate aerobic and/or faculta- biology laboratories. Clin Microbiol Infect 2008; 14:908–34. tive aerobes, which are minority constituents of the overall microbial 18. Cai HY, Caswell JL, Prescott JF. Nonculture molecular techniques for diagnosis of community in abscesses in some cases. As the role of a polymicrobial bacterial disease in animals: a diagnostic laboratory perspective. Vet Pathol 2014; 51:341–50. community in abscess formation remains poorly understood, future 19. Wolcott RD, Gontcharova V, Sun Y, et al. Bacterial diversity in surgical site infec- studies should focus on understanding the interplay between various tions: not just Aerobic cocci any more. J Wound Care 2009; 18:317–23. 20. Mishra AK, Dufour H, Roche PH, et al. Molecular revolution in the diagnosis of bacteria in abscess collections and the clinical implications, including microbial brain abscesses. Eur J Clin Microbiol Infect Dis 2014; 33:2083–93. antimicrobial management and treatment response. 21. Salipante SJ, Sengupta DJ, Rosenthal C, et al. Rapid 16S rRNA next-generation sequencing of polymicrobial clinical samples for diagnosis of complex bacterial infections. PLoS One 2013; 8:e65226. 22. Brogden KA, Guthmiller JM, Taylor CE. Human polymicrobial infections. Lancet Supplementary Data 2005; 365:253–5. Supplementary materials are available at Open Forum Infectious Diseases 23. Brook I, Frazier EH. Microbiology of subphrenic abscesses: a 14-year experience. online. Consisting of data provided by the authors to benefit the reader, Am Surg 1999; 65:1049–53. the posted materials are not copyedited and are the sole responsibility of 24. Brook I, Frazier EH. Aerobic and anaerobic microbiology in intra-ab- the authors, so questions or comments should be addressed to the corre- dominal infections associated with diverticulitis. J Med Microbiol 2000; sponding author. 49:827–30. Bacterial Composition of Intra-abdominal Abscesses • OFID • 7 Downloaded from https://academic.oup.com/ofid/article-abstract/5/2/ofy025/4823433 by Ed 'DeepDyve' Gillespie user on 16 March 2018 25. Dethlefsen L, Huse S, Sogin ML, Relman DA. The pervasive effects of an antibiotic on the 32. Duan K, Dammel C, Stein J, et al. Modulation of Pseudomonas aeruginosa human gut microbiota, as revealed by deep 16S rRNA sequencing. PLoS Biol 2008; 6:e280. gene expression by host microflora through interspecies communication. Mol 26. Dewhirst FE, Chen T, Izard J, et al. The human oral microbiome. J Bacteriol 2010; Microbiol 2003; 50:1477–91. 192:5002–17. 33. Brook I. Beta-lactamase-producing bacteria in mixed infections. Clin Microbiol 27. Sibley CD, Church DL, Surette MG, et al. Pyrosequencing reveals the complex polymi- Infect 2004; 10:777–84. crobial nature of invasive pyogenic infections: microbial constituents of empyema, liver 34. Brook I. Microbiology of polymicrobial abscesses and implications for therapy. J abscess, and intracerebral abscess. Eur J Clin Microbiol Infect Dis 2012; 31:2679–91. Antimicrob Chemother 2002; 50:805–10. 28. Pettengill JB, McAvoy E, White JR, et al. Using metagenomic analyses to estimate the 35. Bartlett JG, Onderdonk AB, Louie T, et al. A review. Lessons from an animal consequences of enrichment bias for pathogen detection. BMC Res Notes 2012; 5:378. model of intra-abdominal sepsis. Arch Surg 1978; 113:853–7. 29. Brook I. Microbiology of polymicrobial abscesses and implications for therapy. 36. Baron E, Thomson R. 2011. Specimen collection, transport, and processing: J Antimicrob Chemother 2002; 50:805–10. bacteriology. In: Versalovic J, Carroll K, Funke G, et al., eds. Manual of Clinical 30. Al Masalma M, Armougom F, Scheld WM, et al. The expansion of the microbi- Microbiology. 10th ed. Washington, DC: ASM Press: 228–71. ological spectrum of brain abscesses with use of multiple 16S ribosomal DNA 37. Peters RP, van Agtmael MA, Danner SA, et al. New developments in the diagnosis sequencing. Clin Infect Dis 2009; 48:1169–78. of bloodstream infections. Lancet Infect Dis 2004; 4:751–60. 31. Eberl G. A new vision of immunity: homeostasis of the superorganism. Mucosal 38. Pallen MJ. Diagnostic metagenomics: potential applications to bacterial, viral and Immunol 2010; 3:450–60. parasitic infections. Parasitology 2014; 141:1856–62. 8 • OFID • Kozlov et al Downloaded from https://academic.oup.com/ofid/article-abstract/5/2/ofy025/4823433 by Ed 'DeepDyve' Gillespie user on 16 March 2018
Open Forum Infectious Diseases – Oxford University Press
Published: Feb 1, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera