If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Surgical site infection is a major source of morbidity in patients undergoing pancreatic head resection and is often from organisms in intraoperative bile duct cultures. As such, many institutions use prolonged prophylactic antibiotics and tailor based on bile duct cultures. However, standard cultures take days, leaving many patients unnecessarily on prolonged antibiotics. Nanopore sequencing can provide data in hours and, thus, has the potential to improve antibiotic stewardship. The present study investigates the feasibility of nanopore sequencing in intraoperative bile samples.
Patients undergoing pancreatic head resection were included. Intra-operative bile microbial profiles were determined with standard cultures and nanopore sequencing. Antibiotic recommendations were generated, and time-to-results determined for both methods. Organism yields, resistance patterns, antibiotic recommendations, and costs were compared.
Out of 42 patients, 22 (52%) had samples resulting in positive standard cultures. All positive standard cultures had microbes detected using nanopore sequencing. All 20 patients with negative standard cultures had negative nanopore sequencing. Nanopore sequencing detected more bacterial species compared to standard cultures (10.5 vs 4.4, p < 0.05) and more resistance genotypes (10.3 vs 2.7, p < 0.05). Antimicrobial recommendations based on nanopore sequencing provided coverage for standard cultures in 27 out of 44 (61%) samples, with broader coverage recommended by nanopore sequencing in 13 out of 27 (48%) of these samples. Nanopore sequencing results were faster (8 vs 98 hours) than standard cultures but had higher associated costs ($165 vs $38.49).
Rapid microbial profiling with nanopore sequencing is feasible with broader organism and resistance profiling compared to standard cultures. Nanopore sequencing has perfect negative predictive value and can potentially improve antibiotic stewardship; thus, a randomized control trial is under development.
Infectious complications are a major source of morbidity in patients undergoing pancreatic head resection. Up to 25% of patients experience postoperative surgical site infection (SSI) after pancreatic surgery, and these have shown to be correlative with organisms identified in bile at the time of surgical resection in up to 60% of cases.
Patients undergoing biliary instrumentation have particularly high rates of bacteriobilia; 2019 Enhanced Recovery After Surgery guidelines recommend routine use of intraoperative bile cultures in these patients.
Since 2015, our institution has obtained bile duct cultures on patients undergoing pancreatic head resection. Patients are started on a prophylactic course of antibiotics tailored in the postoperative setting based on intraoperative bile culture susceptibility data. While intraoperative bile cultures and prophylactic antibiotics may reduce infectious complications, they are limited by the turnaround time of culture and antibacterial susceptibility testing, which take days and occasionally weeks to obtain, creating a window for resistant organisms to propagate.
Furthermore, in the clinical microbiology laboratory, antibacterial susceptibility testing is not routinely performed on cultures yielding large numbers of different bacteria. Within the clinical setting, rapid identification of pathogens and profiling of antimicrobial resistance (AMR) may be a tool to mitigate suboptimal antimicrobial therapies, which can potentially exacerbate morbidity and mortality as well as lead to increased hospital cost in the treatment of associated SSIs.
The ideal assay for bile testing is capable of producing accurate microbial information in an expedited fashion,
allowing for the administration of tailored antibiotics in the early postoperative window or for early cessation of antimicrobial prophylaxis if no growth is identified, potentially reducing complications from antibiotic exposure. In addition to providing rapid and accurate results, this ideal test would be cost-effective. We recently demonstrated that culturing aspirated bile as opposed to traditional bile duct swabs significantly increased organism yield and allowed for metagenomic sequencing of samples.
Metagenomic sequencing allows for the detection of almost all known microorganisms simultaneously from clinical samples and removes biases associated with the current culture techniques. Moreover, sequenced reads can be utilized to identify useful pathogenic traits, such as AMR genotypes. Nanopore sequencing (NS) is of particular interest due to its direct, real-time performance that enables microbial identification to be achieved within minutes of the sequence run being initiated, reducing microbial detection time to as low as 6 hours.
Given the potential clinical benefit of metagenomic sequencing to analyze the microbiome of bile, we performed the present study to assess the feasibility and utility of metagenomic sequencing in patients undergoing pancreatic head resection.
Patient demographics and data availability
Patients undergoing pancreatic head resection at a single institution were prospectively enrolled in this Mayo Clinic Institutional Review Board-approved study from April 2020–October 2020. Patient characteristics at surgery and postoperative culture results were gathered through electronic medical record review. This study was deemed exempt from informed consent, and patient identifying information was stored on the institutional server for access by designated study members only. NS was only available to research staff and was not provided to the clinical team.
Bile collection and standard cultures
Bile was collected via bile duct aspiration and bile duct swab. Bile duct aspiration was performed with a 16-gauge needle and 3 cc syringe prior to bile duct division, and aspirated material aseptically transferred into sterile aerobic and anaerobic transport vials. For bile duct swab cultures, any indwelling stent, if present, was removed after duct division. A swab (Becton Dickinson BBL CultureSwab with liquid Stuart’s media [Becton Dickinson, Franklin Lakes, NJ]) was inserted into the cut end of the duct for the collection of bile. Samples were concurrently transported to the clinical microbiology laboratory. Swab samples were inoculated into thioglycolate broth (thio), chocolate blood agar (CBA), tryptic soy agar with 5% sheep blood (BAP), eosin-methylene blue agar and Columbia CNA agar with 5% sheep blood, colistin and nalidixic acid (CNA). Bile fluid aspirate samples were aseptically removed from the transport vial and one aliquot of fluid placed onto each of chocolate blood agar, BAP, eosin-methylene blue agar, and CNA culture media for aerobic culture. Aerobic culture plates were incubated at 35°C in 5% CO2 and examined daily for growth for up to 5 days. Anaerobic culture media were inoculated in parallel and included CDC anaerobe 5% sheep blood, phenylethyl alcohol blood agar, laked Brucella blood agar with kanamycin and vancomycin and a BAP used as an aerotolerance screen. Additionally, up to 1 mL (0.1 mL, minimum) was placed into an anaerobic thioglycolate (thio) vial. The anaerobic thio and aerotolerance BAP were incubated at 35°C in an ambient atmosphere. Anaerobe-specific agar media were placed into GasPak™ (Becton Dickinson, Franklin Lakes, NJ) jar with an AnaeroPack™ and Mitsubishi™ RT anaero-indicator (Remel, San Diego, CA). Anaerobic culture plates were incubated for up to 7 days and the anaerobic thio for up to 14 days. After a sample was aliquoted for standard culture (SC), 1 mL of remaining bile was snap-frozen using liquid nitrogen and collected by the study team for metagenomic analysis. SCs were considered positive if any organisms were present. Samples were considered resistant if resistance to any antibiotic was reported in any microorganism.
DNA was extracted from the bile duct aspirates using a phenol-chloroform method. In brief, 1 mL of bile was centrifuged for 30 minutes at 4oC. The supernatant was discarded, and the pellet was resuspended in 500 μL of buffer A (5M NaCl, 1M Tris, 0.5M EDTA), 500 μL of phenol chloroform, and 210 μL of SDS. Two rounds of bead beating (60 seconds at a speed of 60 m/s) using a FastPrep 24 5G (MP Biomedicals, Solon, OH) was performed to lyse cells. The sample was centrifuged for 3 minutes (14,000 rpm, 4oC) to separate the genomic DNA from cellular proteins, lipids, and other debris. The aqueous phase was transferred to a sterile 1.5 mL Eppendorf tube and mixed with 500 μL of phenol chloroform. Centrifugation was repeated. The aqueous DNA solution was incubated for 1 hour on ice with 600 μL of isopropanol and 60 μL of sodium acetate to allow for DNA precipitation. Following DNA precipitation, the DNA sample was centrifuged for 20 minutes (14,000 rpm, 4oC) to generate a DNA pellet. The supernatant was removed, and the DNA pellet was washed in 500 μL of ethanol. The ethanol was removed, and the DNA pellet was resuspended in 200 μL of PBS. DNA purification was performed using the QIAamp DNA mini kit (Qiagen, Venlo, the Netherlands). Using the NEBNext Microbiome Enrichment Kit, 100–200 ng of extracted genomic DNA underwent host DNA depletion. AMPure XP beads (Agencourt, Beverly, MA) at a ratio of 1.8 beads to sample were used to purify host- genomic DNA, and DNA quantification was performed using a Qubit 2.0 fluorometer (Thermo Fisher Scientific, Waltham, MA) and the Qubit dsDNA high sensitivity kit (Thermo Fisher Scientific, Waltham, MA).
Library preparation and sequencing
Library preparation was performed using the Rapid PCR Barcoding Sequencing Kit (SQK-RPB004, Oxford Nanopore Technologies , Oxford, United Kingdom) according to the manufacturer’s protocol. The procedure took approximately 3 hours per 10 samples. The constructed library was loaded into an R9 Flow cell. Sequencing was performed for 72 hours on a GridION sequencer (Oxford Nanopore Technologies, Oxford, United Kingdom). Low-quality reads (phred score <7.0) were filtered out during the sequencing, and high accuracy base-calling was performed on the filtered reads using MinKNOW (20.06.9).
Microbial identification using NS
Bacterial identification was performed with Centrifuge software (1.0.4) using a customized dataset of bacterial, fungal, and viral genomes belonging to known human commensals and pathogens as well as the human genome. Taxa were determined using an in-house script, and bile samples were declared microbial positive when 500 or more microbial reads were detected following alignment. Additionally, a 50 reads per million (RPM) ratio (RPM-r), defined as RPM-r = RPMsample / RPMnegative control, was used as an additional minimum bacterial read threshold to reduce biases caused by different sequence depth and mitigate concerns regarding potential microbial contamination, as described in previous studies.
Identification of antimicrobial resistance using NS
NS detection of AMR genes was performed using Resfinder 4.0 and the default alignment settings (≥80% identity over ≥60% of the length of the target gene). AMR genes associated with bacteria not detected in bile samples using NS were removed from further analysis. The Wilcoxon test was used to compare the detection of AMR using SC and NS.
SC and NS microbes and antibiotic susceptibility data were de-identified for both sample type and patient. The results from the 2 groups were randomized and provided to 2 blinded, independent providers, a faculty member in the Division of Infectious Diseases and a surgical clinical pharmacist. Microbial coverage recommendations were made based on de-identified data, and recommendations were compared. Recommendations were categorized as correlative, broader with NS, or broader with SC. Samples were considered correlative if identical antibiotic recommendations were made for SC and NS samples. Any addition of antibiotics or broadening of coverage was considered broader with NS or SC.
Time-to-results for SC were calculated as the time of specimen collection to the time of final species identification and reporting of antibiotic susceptibility results in the electronic medical record. Only results from positive SC were used for comparison as cultures are incubated for a fixed period before being reported as having no growth. Aerobic cultures are incubated for up to 5 days, and anaerobic cultures are incubated for up to 14 days; these times were not included. Time-to-results for NS were calculated from the time of specimen collection to analysis of first results, occurring approximately 15 minutes into the sequence run.
Cost of assays was determined as the cost of sequencing reagents versus clinical cost. Cost reduction estimate from antibiotic stewardship was estimated as a 4-day reduction of antibiotic costs from the 5-day standard antibiotic prophylaxis as NS would allow for the cessation of antibiotics by postoperative day 1. Meaning these patients would otherwise receive an additional 4 days of antibiotics without NS. Correlation with SC, time-to-diagnosis, and cost reduction were performed using simple statistical analysis, with χ2 analysis testing for binary variables and Student’s t-test for continuous variables.
Growth of organism in SC correlated with organism detection in NS
A total of 44 patients who underwent pancreatic head resection were included in this prospective clinical study. Forty-two patients were included in the final cohort, with 2 patients excluded for inadequate DNA extraction as both samples failed to generate a cell pellet following the initial centrifugation step, and our aim was to detect cellular DNA. Patients underwent 1 of 2 operations, pancreaticoduodenectomy (n = 33) or total pancreatectomy (n = 9). The median age at the time of surgery was 61 (interquartile range [IQR]) 22–71) years. Biliary stents were present at the time of operation in 52% (22/42) of patients. The indication for surgery was neoplastic process in 83% (35/42) of patients, with 67% (28/42) undergoing resection for pancreatic ductal adenocarcinoma. Baseline patient characteristics are summarized in Table I.
SC were positive in 52% (22/42) of patients. Of these 22 patients with positive cultures, 95% (21/22) had biliary stenting prior to operation. A total of 22 samples were positive for bacterial DNA by NS. In the 20 patients with negative SC, all 20 also had negative NS, demonstrating perfect negative predictive value for the standard of care SC currently used.
Microbial detection results show higher organism yield with NS
Microbial identification using SC and NS revealed that, on average, NS detected twice as many bacterial species in the bile samples compared to SC. Overall, SC identified an average of 4.4 bacterial species per microbial positive sample (range = 1–9 species), whilst NS identified an average of 10.6 bacterial species per microbial positive sample (range = 4–18). This difference was statistically significant (p = 5.06e-6). Comparison of the bacterial species detected by the 2 techniques revealed that an average of 1.5 bacterial species uniquely detected using SC, 7.6 bacterial species uniquely detected by NS, and 2.9 species detected by both techniques. In 45.5% of the bile samples, all the bacterial species detected using SC were also detected using NS (Figure 1).
Microbial identification using SC and NS revealed that the bile samples were dominated by members of the Bifidobacterium, Enterococcus, Klebsiella, Lactobacillus, and Streptococcus genera. Analysis of the bacterial species uniquely detected using SC found that in the majority of incidences where cultured bacteria went undetected using NS, the 2 techniques were detecting different species of Enterococcus (clinical samples 101, 108, 110, 127, and 133), Lactobacillus (clinical samples 126, 127, and 142), and Bifidobacterium (138 and 141). In 2 cases, cultured Staphylococcus aureus was not detected by NS, and in 1 case each, cultured Citrobacter freundii, Pseudomonas aeruginosa, and Streptococcus mitis were missed.
With regards to bacterial species detected using NS, high abundance species (defined as bacterial species with a microbial relative abundance greater than 5%) were typically detected in both SC and NS (Figure 1). Bacterial species not detected by SC included anaerobic species (Atopobium rimae, Campylobacter showae, Fusobacterium nucleatum, Leptotrichia trevisanii, Mogibacterium timidum, Prevotella denticola, and Veillonella dispar) and bacterial species difficult to distinguish phenotypically, such as lactobacilli species and members of the Streptococcus mitis group (Streptococcus mitis, Streptococcus oralis, Streptococcus infantis, Streptococcus parasanguinis) (Figure 1).
Although the primary aim of the study was to compare rates of bacterial detection, SC identified fungal species in 9 of the bile samples, thus enabling the potential for NS in fungal detection to be assessed (Table II). On average, NS detected 772 Candida reads in the bile samples that were culture-positive for Candida species (range = 10–2523) compared to an average of 0.66 reads that were detected in the Candida negative culture samples (range = 0–6). Additionally, the teleomorph of Candida krusei, Pichia kudriavzevii, was detected in sample 134, and Saccharomyces cerevisiae was detected in 3 of the bile samples (Table II).
Table IIDetection of fungal species using standard culture and Nanopore sequencing
AMR phenotypes were detected in 21 of the 22 microbial positive samples using SC (range = 0–6 phenotypes per sample), and following removal of potential false-positive NS detected AMR genes in 19 of 22 samples, corresponding to an average of 10.3 predicted AMR phenotypes (range = 0–20 phenotypes per sample).
Comparison of AMR detected in NS compared to SC revealed that the number of predictive AMR phenotypes detected using NS was significantly higher than the number of AMR phenotypes observed by SC (average of 10.3 predictive phenotypes vs 2.7 observed phenotypes, P = .0002). The most commonly detected phenotypes included resistance to aminoglycosides (amikacin, gentamicin, streptomycin, tobramycin), beta-lactams (amoxicillin, ampicillin, aztreonam, cefazolin, cefepime, cefotaxime, cefoxitin, ceftazidime, ceftriaxone, cephalothin, piperacillin), lincosamides (clindamycin, lincomycin), macrolides (azithromycin, erythromycin, telithromycin), streptogramins (dalfopristin, pristiniamycin ia, pristiniamycin iia, quinupristin, virginiamycin m, virginiamycin s), and tetracyclines (doxycycline, minocycline, tetracycline).
In 5 samples, there was a perfect correlation between AMR detected using SC and NS; this included 1 sample that was negative for AMR using both SC and NS. Differences in AMR phenotypes detected using the 2 methods were typically a result of NS detecting AMR phenotypes that were not assessed using SC and NS detecting different cephalosporin resistance phenotypes compared to SC (68% of samples). Additionally, SC detected several AMR phenotypes that were not detected by NS. This included ampicillin resistance (detected in 18 samples using SC and 11 samples using NS) and cefazolin resistance (detected in 15 samples using SC and detected in 0 samples using NS). Analysis of the Resfinder database revealed that cefazolin resistance went undetected using NS as a result of the database not containing resistance genes for the first-generation cephalosporin. Genes conferring resistance to cephalothin (another first-generation cephalosporin) and several third and fourth-generation cephalosporins (ceftriaxone, ceftazidime, cefepime) were detected, and it is likely that in samples where these genes were detected, cefazolin resistance was also present.
Antimicrobial recommendations were made for NS and SC results in the 22 patients with both assays positive for microbes (Table III). Correlation was seen in 32% (14/44) samples. Thirty-nine percent of samples (17/44) were associated with broader antimicrobial recommendations based on SC, and 30% (13/44) with broader recommendations based on NS.
A total of 16% (7/44) of patients experienced SSI in the first 30 days postoperatively. Biliary instrumentation was performed in 57% (4/7) of patients who experienced SSI, and 57% (4/7) had positive intraoperative SC and NS. Cultures were performed and were positive for growth in the postoperative infectious complications in 4 out of 7 patients. Organisms detected in the postoperative infection cultures correlated with both SC and NS in 50% (2/4) of cases.
Median time to final culture results in patients with positive SC was 98 hours (IQR 80–152 hours). With regard to time-to-results for the NS technique, it took approximately 8 hours to prepare the bile samples and generate usable results. This included approximately 4.5 hours to extract and quantify genomic DNA from bile, 2.5 hours to prepare and run the Rapid PCR Barcoding protocol, 10 minutes to quantify the PCR amplicons, and 30 minutes to prepare and load the sequencing library into the flow cell. Within 15 minutes into the sequence run, bacterial species could be determined, and within 1 hour AMR genes were detected, with the majority of AMR genes identified within 6 hours of the sequence run.
DNA extracted from the bile samples was sequenced in batches of 5 samples per sequence run. The total cost to perform DNA extraction, deplete host DNA, and sequence the sample using NS was approximately $165 when sequencing was performed in batches of 5 (Table IV). Costing was reduced by bulk purchasing of the flow cells (R9), reducing price of flow cell from $900 per cell to $475 per cell, and the use of the Rapid PCR Barcoding kit (SQK-RPB004), which enabled multiple samples to be sequenced concurrently. Microbial identification and detection of AMR genes were performed using in silico analysis, and as the tools used are all freely available, the analysis did not contribute towards the total cost of using NS. In contrast, the cost of performing aerobic and anaerobic culture analysis on the bile samples was $38.49. This included a charge of $16.43 to perform the aerobic culture analysis and $22.06 to perform the anaerobic culture analysis. The difference in the cost of NS vs SC was $126.51, which was partially offset by a reduction in antibiotic cost in those with negative NS.
Preliminary results from this clinical study demonstrated that microbial DNA sequencing of bile samples from patients undergoing pancreatic head resection is not only feasible but produces highly correlative results with SC. NS produced results in an average time of 8 hours from sample collection to result availability compared to 98 hours for SC, with 100% of positive SC demonstrating microbial DNA by NS. This difference in time-to-results is clinically significant, particularly with regards to the rapid identification of AMR genes and detection of anaerobic bacterial species and fungal species. Rapid results have the potential to limit inappropriate empirical antimicrobial coverage, decrease associated morbidity and mortality, decrease hospital length of stay, and lower associated hospital costs. The findings and implications of this study in the management of patients undergoing pancreas surgery in the modern era are discussed below.
The described NS assay had a perfect negative predictive value for SC, which means that it can potentially have immediate clinical impact as it can reduce the number of patients on prophylactic antibiotics unnecessarily. A mounting problem in the world of microbiology is that of increasing resistant organisms known as “superbugs.” The World Health Organization reports that AMR is one of the biggest threats facing humanity, and the 2019 Centers for Disease Control and Prevention report on Antibiotic resistance threats in the United States estimated that 2.8 million people are infected with AMR bacteria annually, resulting in 35,000 deaths.
If antibiotics can be deemed unnecessary or targeted to patients' specific organisms using metagenomics by the time they reach the hospital floor, days of antibiotic therapy may be avoided entirely. Our current practice includes a 5-day course of prophylactic antibiotics in patients undergoing pancreatic head resection, which is completed in patients with no microbial growth, as aerobic cultures are incubated for 5 days and anaerobic cultures are incubated for 14 days. The NS had perfect correlation with SCs regarding microbial detection. Therefore, patients with no detected microbes on NS can likely have antimicrobial prophylaxis stopped within 8 hours of bile acquisition that will improve antibiotic stewardship and reduce the contribution to superbug development. Additionally, NS has been previously demonstrated to be a tool for real-time monitoring of AMR gene transfer in the hospital setting,
As such, biliary obstruction requires preoperative biliary instrumentation and drainage to not only avoid life-threatening complications, such as cholangitis, but also to provide duct patency for the duration of neoadjuvant chemotherapy. While this increase in biliary instrumentation in the perioperative setting is oncologically necessary, it is associated with an increased rate of SSI and bacterobilia.
This is supported by findings in our own patient population in which we have seen a strong correlation with biliary stenting and positive cultures. This information may help define the patient population who truly require prolonged perioperative antibiotics as patients with and without biliary stenting have clearly distinct SC and NS patterns, with the vast majority of patients without instrumentation having sterile cultures. We have recognized this pattern in additional institutional studies; thus, we are currently in the process of reviewing our institutional data to optimize our clinical practice regarding these distinct patient populations.
Our NS data also supports the notion of gut microbiota as a source of infection as identified organisms (Bifidobacterium, Enterococcus, Klebsiella, Lactobacillus, and Streptococcus) are genera commonly associated with the gut. These genera have also been observed to be the dominant bacteria in the hepatic duct after pancreatic head resections
; identification of these organisms suggests that the translocation of microorganisms from the gut into the bile duct is the source of bile microbes and is likely due to the biliary stenting exposing the physiologically sterile bile tract to the intestinal tract. By improving the administration of antibiotics to target bacterobilia early using NS, postoperative infection risk associated with biliary instrumentation and stenting may be mitigated. Additionally, patients with biliary stents have higher numbers of organisms present, which may be identified with NS as demonstrated in this study as NS has shown to significantly increase the yield of species detected compared to SC. Whether or not this will identify organisms found in SSIs remains unproven. In the present study, we demonstrated only 40% of patients had organisms that would implicate bacteriobilia as a source of infection. This is not so much a failure of either of the assays but rather a causation of infection that was not contaminated bile. This rate is lower than seen in our previous work in which we demonstrated a higher correlation,
but this is likely due to a small sample size of only 5 cultures.
Although NS has a higher up-front cost, it has the potential to reduce healthcare costs through a reduction in SSI and through a reduction of complications from prolonged antibiotics. SSI is the most common cause of hospital-acquired infection, with an estimated yearly cost to the US healthcare system of $3.5–$10 billion dollars.
In patients undergoing major pancreatic operations, the cost of SSI is likely closer to the upper limit of this range, given the frequent need for interventional procedures and associated imaging. Furthermore, these patients are recovering from major operations, leaving them in a severely compromised physiologic state given the magnitude of pancreatic head resection. For these patients, an SSI can lead to clinical decline and death, making prevention of utmost importance. The enormity of the financial impact and risk of morbidity makes the <$130 difference in cost between the 2 assays nominal. The prevention of a single SSI would cover the cost of thousands of assays. Furthermore, the cost reduction due to an improvement in antibiotic stewardship through reduction of prophylactic antibiotics alone (1 day vs standard 5) offsets the additional cost of NS entirely in patients with negative NS as antibiotics can be stopped within 8 hours from the time of bile collection.
One limitation of this technique is the lack of true standardization. We are in the process of developing protocols, which we will aim to publish for more universal application of techniques, which will further assist in making this technique easier to adapt at institutions. There are standardized sequencing kits but no standardization in terms of DNA extraction methods. Researchers will often optimize the extraction protocols to suit their needs, and whilst there has been a significant number of papers describing its clinical applications (predominately proof-of-principle studies on rapid identification of pathogens and AMR genes in clinical samples
, NS has not been formally approved for clinical use. In regard to availability and translating to existing labs, once the sequencer has been received in the lab, setting up the technology is fairly straightforward. DNA extraction can be automated to ensure standardization within the lab/hospital setting, library preparation and sequencing is typically easy to perform (degree of difficulty will vary depending on the sample type), the reagents can be purchased in bulk and a delivery schedule arranged, and the sequencer takes up very little space (the minion is approximately the size of a mobile phone whilst the gridION is slightly larger than a regular printer).
At this time, NS data is not made available to the clinical team as this preliminary data was gathered to assess feasibility and correlation with current standard of care. Therefore, the impact of providing this data to the clinical team remains unknown. However, the clinical antimicrobial recommendations suggest that changes in practice could occur if NS is used. Our results demonstrate that antimicrobial recommendations based on NS would provide coverage that would be correlative or broader than SC in 61% of samples. This would mean that 39% of patients would have narrower antimicrobial recommendations based on NS. It is unclear what the implications of this data are as we do not know which assay is necessarily more clinically relevant. While interpreting these data, one should keep in mind that SC is not necessarily the ideal “gold standard” to use for microbial profiling; NS techniques have been shown to be less biased and more complete. What is clear is that NS leads to differences in antibiotic recommendations, but until NS techniques are evaluated in a randomized controlled trial, the actual clinical implication of these results is unknown as we will not know if the narrower antibiotic recommendations in the 39% of patients will be as efficacious as SC recommendations in the prevention of SSI. It is our future aim to answer these questions, and, given these promising preliminary data, we are in the process of developing a randomized controlled trial to give us answers on the clinical impact of these techniques. However, one of the greatest challenges with regards to this research is the abundance of data made known by this assay. In the blinded AMR recommendations part of this study, we attempted to tackle the issue of too much data by only informing the independent providers of the high abundant species (relative abundance greater than 5% of total microbial reads sequenced) and AMR encoded by the high abundant species. Moving forward in the future trial, we plan to remove any well-established commensal organisms, such as the Bifidobacteria species from the list provided to the clinicians. We also plan on providing AMR results based on what antibiotics are clinically in use and have the additional AMR results available if required. Care teams will be provided with either SC data or given early NS data by postoperative day 1 as supplemental data in addition to SC. This study will assess the ability of early NS data to reduce the rate of SSI in the postoperative period. Additional future directions involve expanding the types of samples analyzed with NS. This technique can be applied to all clinical samples (ie, blood, body cavity/drain fluid, urine) to gather microbial information; thus, it may have healthcare implications in general. While we do not advocate the use of NS as a stand-alone assay, we envision an algorithm for clinical use after validation which can be seen in Figure 2.
One of the strongest aspects of this assay is its rapid time to actionable results. In some cases, this could be as early as when the patient leaves the post-anesthesia care unit and arrives on the hospital floor. One can imagine how the application of this to other hospital fluids would potentially have a major impact. For example, pan-culture results in hours for septic patients would revolutionize critical care. However, while this innovative yet feasible assay is promising, this study has several limitations. First, this assay needs to be validated in a large cohort of patients. The aim of this study is to create preliminary data, which has been accomplished and now can be verified with additional testing. Second, there were differences in antibiotic recommendations. These differences may reduce the use of unnecessary antibiotics and thus reduce the risk of inappropriate antimicrobial use. However, select patients may still have delayed broadening of coverage until an SC result, and this may additionally lead to an increase in changes in antibiotics throughout the hospital course. And finally, the clinical impact is unknown as providers do not have results available, but this will become evident with a randomized controlled trial.
In conclusion, metagenomic analysis with NS of bile can rapidly define microbial profiles in a cost-effective manner. Given promising preliminary results, the development of a randomized controlled trial is underway to elucidate the clinical implications of this assay in patients undergoing pancreatic head resection. The impact of this assay on the treatment of infection, while presently unknown, may set in motion a new treatment paradigm in all clinical samples.
Funding received from Mayo Clinic Center for Individualized Medicine , Transform the Practice Award.
Conflict of interest/Disclosure
The authors have no related conflicts of interest to declare.
The Impact of bile duct cultures on surgical site infections in pancreatic surgery.