Nanopore 16S Amplicon Sequencing Enhances the Understanding of Pathogens in Medically Intractable Diabetic Foot Infections

One of the most serious complications of diabetes is foot ulcers (1,2). Patients with diabetes have a 12–25% lifetime risk of developing a foot ulcer, which often becomes infected (3,4). Diabetic foot infections (DFIs) are the most common cause of diabetes-related hospital admissions (5). Diabetic foot ulcers usually heal very slowly because of several factors, including diabetes-associated microvascular disease, peripheral neuropathy, progressive changes of bony structure in the foot, impaired host immune response, and involvement of bacteria that often exist within the wound and are resistant to antimicrobial treatments (1,6). When foot ulcers and infections fail to heal, critical complications, such as osteomyelitis, occur (7–9). Diabetic foot osteomyelitis occurs in 44–68% of patients who are admitted to the hospital because of DFI and is the leading cause of foot amputation in these patients (3).

Molecular-based methods have been increasingly applied for pathogen identification because of recent advances in sequencing technologies (10,11). The 16S rDNA amplicon sequencing is particularly useful for the detection of bacteria (12,13) and has many advantages over conventional culture studies. As well as a rapid turnaround time, 16S sequencing is capable of the detection of unculturable bacteria and polymicrobial infection at once (14,15). Nanopore sequencing is one of the new-generation sequencing technologies produced by Oxford Nanopore Technologies (ONT; Oxford, U.K.). It is carried out by predicting nucleotide sequences from the electrical current patterns that are affected by the bases passing through the nanopore. Nanopore sequencing has many advantageous characteristics, including a simple library preparation procedure that could be extremely useful for clinical metagenomics (16,17). Nanopore sequencing enables real-time analysis of reads and long-read sequencing, which can be very useful for rapid pathogen detection (10,18), and is applicable for bacterial detection by 16S amplicon sequencing (17,19,20). Thus, nanopore 16S amplicon sequencing is being applied to the detection of various bacterial infections, including bacterial meningitis, brain abscess, pneumonia, and others (21–24).

In the current study, we performed nanopore 16S amplicon sequencing on the tissues obtained during surgery from patients with medically intractable DFI. We investigated whether nanopore 16S amplicon sequencing is capable of pathogen identification in patients with DFI and compared its efficacy with conventional culture studies.

Among the patients who visited the orthopedics department of Seoul National University Hospital (SNUH) between June 2018 and December 2019, those with medically intractable DFI requiring surgical debridement or amputation were included. Medically intractable DFI was defined as failure of conservative treatment, including dressing, off-loading techniques such as total contact cast, and antibiotics therapy for at least 1 month. Cases combined with osteomyelitis, which necessitated copious elimination of infected bone and soft tissues, were included. Patients having foot ulcers with uncertain infections or patients who refused consent were excluded. Debrided tissues were obtained during surgery from the deep central portion of lesions, including bone and adjacent soft tissue, by a skillful orthopedic surgeon (D.Y.L.). Then, samples for culture studies were directly placed in blood culture bottles for aerobic and anaerobic culture and were promptly transported to the microbiology laboratory. Samples for sequencing were stored in sterile tubes from the operating room and kept at 4°C before the experimental procedures. The study was approved by the institutional review board of SNUH (IRB No. 1806-032-949), and informed written consent was obtained from all patients.

The DNA was extracted from the surgical specimens using the PureLink Genomic DNA Mini Kit (Invitrogen, Carlsbad, CA) following the manufacturer’s instructions. For each sample, the 16S rDNA PCR was performed as described previously (23). In brief, the full length of 16S rDNA was amplified by PCR using a bacterial 16S rDNA PCR kit (Takara, Tokyo, Japan). The 16S rDNA primer mix (Takara) was added to the genomic DNA. PCR was then performed with an initial denaturation at 94°C for 1 min followed by 35 cycles at 94°C for 30 s, 55°C for 30 s, and 72°C for 1 min, with a final extension at 72°C for 3 min. All PCRs were performed in a C1000 Touch Thermal Cycler (Bio-Rad, Hercules, CA). A negative control (distilled water) and a positive control (bacterial genomic DNA) were included in every PCR. The PCR products were electrophoresed on a 1.5% agarose gel containing 0.05 μL/mL RedSafe (Intron Biotechnology, Seoul, South Korea) and were visualized using a Bio-Rad Gel Doc EZ Imager. When the negative control demonstrated a PCR-positive band, contamination was suspected, and the PCR was repeated from the initial step.

When the result of the 16S rDNA PCR was positive, sequencing libraries were prepared from the PCR products using the Rapid Barcoding Sequencing Kit (SQK-RBK004; ONT). The input DNA was end repaired and A-tailed using the Ultra II End Prep Enzyme (New England Biolabs [NEB], Hertfordshire, U.K.) incubated at 20°C for 5 min and at 65°C for 5 min. The end-prepared DNA was purified with AMPure XP (Beckman Coulter, High Wycombe, U.K.), and the DNA was eluted in nuclease-free water followed by ligation with a 1D adapter using Blunt/TA Ligase Master Mix (NEB) at room temperature for 10 min. The 1D adapter DNA purification was achieved with Adapter Binding Buffer (ONT) using the magnetic stand, and the DNA library was eluted with elution buffer (ONT). The presequencing mix was loaded onto an R9.5 flow cell (FLO-MIN107) in a mix of running buffer with fuel mix and library loading buffer (ONT). Finally, sequencing was performed for 2 or 3 h, and base calling was performed using MinKNOW software.

During or after sequencing, the sequenced reads were analyzed by the cloud-based Metrichor/EPI2ME platform (Metrichor Ltd., Oxford, U.K.). The 16S analysis workflow of EPI2ME was used, which is designed to Basic Local Alignment Search Tool (BLAST) base-called reads against the National Center for Biotechnology Information 16S bacterial database. Generated reads were classified to certain bacteria at the species level on the basis of the percent coverage and identity. The list of the bacteria was arranged in descending order according to the number of aligned reads, and the pathogens from the top of this list were determined by clinicians. The species identification within a certain genus was determined on the basis of the number of aligned reads, with the one with the largest number being selected as the answer.

The nanopore sequencing throughput is affected by many factors, which include the number of active pores remaining in the flow cell, DNA purity and integrity, library concentration, and sequencing time (25). Therefore, the absolute number of the aligned reads does not reflect the absolute abundance of certain bacteria. However, in cases of polymicrobial infection, the relative abundance of each bacterium can be estimated by comparing the number of reads aligned to each bacterium within a single sequencing run. The relative abundance score of certain bacteria (A) was calculated by the aligned read counts divided by the read counts of the most abundant bacteria (number of reads aligned to certain bacteria [A]/number of reads aligned to the most abundant bacteria).

The data sets generated and/or analyzed during the current study are available from the corresponding authors upon reasonable request.

A bacterial culture study and 16S rDNA sequencing were both performed on 54 samples obtained from 45 patients with medically intractable DFI (Table 1). In every case, antibiotics were administered before the collection of the samples. More than 80% of the cases (44 of 54, 81.5%) turned out to be caused by polymicrobial infections (Fig. 1A). A total of 92 bacteria belonging to 18 genera were isolated in 41 patients by conventional culture studies. The bacteria most frequently identified by culture studies were Staphylococcus (n = 18) followed by Streptococcus (n = 16), Escherichia (n = 10), and Enterococcus (n = 9). A total of 290 bacteria belonging to 43 genera were revealed by 16S sequencing. The bacteria most frequently identified by 16S sequencing were also Staphylococcus (n = 25) followed by Finegoldia (n = 23), Prevotella (n = 21), Streptococcus (n = 21), Anaerococcus (n = 19), and Bacteroides (n = 15) (Table 2 and Fig. 2).

All bacteria isolated by culture studies were identified by 16S sequencing except for one case. In this patient (patient 7, Table 1 and Fig. 2), Staphylococcus aureus, a frequent source of contamination, was only isolated in the culture studies. Overall, the 16S sequencing revealed the presence of many anaerobes that were not isolated by culture studies from the samples.

The bacteria that were frequently missed by the culture studies were Prevotella (n = 20), Finegoldia (n = 19), Anaerococcus (n = 18), and Bacteroides (n = 15). Bacteroides, Peptoniphilus (n = 12), Fusobacterium (n = 10), Campylobacter (n = 7), Haemophilus (n = 6), Parvimonas (n = 6), Peptostreptococcus (n = 6), and Porphyromonas (n = 5) were only identified by 16S sequencing and not by culture studies (Table 2 and Fig. 2).

After 16S sequencing, multiple bacteria were revealed in the majority of cases (44 of 54, 81.5%). In addition, the sequenced reads were aligned to a single bacterium in 10 of 54 cases, leading to the diagnosis of monomicrobial infections (illustrative case in Fig. 1B). Thus, the 16S sequencing was capable of differentiating polymicrobial infections from monomicrobial infections.

According to the results of the culture studies, polymicrobial infections were only suspected in 32 cases, and a single bacterium was isolated in 18 cases. Among these 18 cases, 10 turned out to be polymicrobial infections by 16S sequencing. In four cases, no bacteria were cultivated despite the confirmation of bacterial infections by 16S sequencing. Two of these four cases actually had polymicrobial infections that were only revealed by 16S sequencing (Table 1 and Fig. 2). In summary, the 16S sequencing was superior to the culture studies in differentiating polymicrobial infections from monomicrobial infections.

In cases of polymicrobial infections, the sequenced reads were aligned to multiple bacteria (illustrative case in Fig. 1C). Assuming that the 16S PCR would amplify the sequence of each bacterium at a similar rate, the bacterium with the largest number of aligned reads could be considered as the most abundant pathogen in the sample. In addition, the relative abundance of certain bacteria was assessed by comparing the number of reads aligned to each bacterium.

The 16S sequencing was useful for monitoring the remaining pathogen. In some cases, 16S sequencing was repeated after prolonged antibiotic treatment. The composition of the pathogens was changed in the follow-up 16S sequencing (illustrative case in Fig. 1D).

Among the 44 polymicrobial infections revealed by 16S sequencing, the dominant pathogens were frequently missed by the culture studies. In 23 of 44 (55%) cases, the most abundant pathogen revealed by 16S sequencing was not isolated in the culture studies. The most frequently missed dominant pathogens were Prevotella (n = 6) and Bacteroides (n = 6) followed by Anaerococcus (n = 3) and Parvimonas (n = 2) (Fig. 3 and Table 2).

Moreover, the majority of the pathogens isolated by the culture studies were located in the lower ranks of the list of pathogens (illustrative case in Fig. 1C). Within the cases with polymicrobial infections, the culture studies isolated 88 bacteria. Among these, 33 (37.5%) bacteria isolated from the culture studies were actually below third place in the 16S sequencing (Table 1).

The 16S sequencing was much faster than the culture studies. Since the reduction in turnaround time was not the primary objective of this study, the 16S sequencing was conducted only during weekday working hours. When the surgery was performed during nighttime or during the weekend, samples were stored at 4°C until the experimental procedures. Nevertheless, 16S sequencing revealed the results earlier than the culture studies in most of the cases. In 50 of 54 (92.6%) cases, the 16S sequencing was faster than the culture studies (Table 1). In most of the cases, the sequencing was run for 2 h; however, when sequencing and analysis were performed simultaneously, the results were obtained within 30 min of sequencing. This finding implies that even shorter sequencing time (<1 h) would be sufficient for pathogen identification, thus shortening the turnaround time. At best, the turnaround time could be reduced to 6 h from DNA extraction to the analysis of the reads (23).

We performed nanopore 16S amplicon sequencing in surgical specimens from patients with medically intractable DFI. The 16S sequencing successfully identified all the bacteria isolated by the conventional culture studies, was more sensitive than the culture studies, and was capable of detecting multiple anaerobes that were not cultivated by the culture studies. Moreover, 16S sequencing was particularly useful in cases of polymicrobial infections. In some cases of polymicrobial infections, the dominant pathogens were only identified by 16S sequencing and not by culture studies. In addition, in most of the cases, 16S sequencing was faster than conventional culture studies.

Nanopore 16S sequencing was more sensitive than conventional culture studies in identifying pathogens. A substantial number of pathogens were only identified by 16S sequencing and not by culture studies. The 16S sequencing was particularly useful for detecting pathogens that are difficult or impossible to cultivate, especially when antibiotics were given before sample acquisition. This was in line with previous studies that showed that 16S sequencing is more sensitive than culture studies for pathogen detection in DFIs (1,3,26,27). Moreover, in most DFIs, antibiotics are prescribed before the debridement surgery (28–30), which would be another reason for the lower diagnostic yield of the culture studies (31–33). Thus, 16S sequencing would be very useful for pathogen identification in DFIs.

Nanopore 16S sequencing was particularly useful for the detection of polymicrobial infections. In most of the cases of polymicrobial infections, the culture studies only isolated one or two bacteria from the various lists. Certain bacteria that are difficult to cultivate were frequently missed by the culture studies. Moreover, some bacteria grow dominantly during the culture procedure, so the results may not reflect the true bacterial composition within the sample (34–36). Even when culture studies can isolate multiple bacteria from a single specimen, it is hard to obtain quantitative data from culture studies. In addition, 16S sequencing could give information on the relative abundance of certain bacteria among the listed pathogens. Although the abundance and virulence of certain bacteria are not related to each other, these data would be considerably useful during the medical treatment of patients with DFI. When follow-up samples can be obtained regularly, antibiotics could be adjusted to target the predominant pathogen according to 16S sequencing results.

Because of nanopore 16S sequencing, we demonstrated that the majority of the medically intractable DFIs are caused by polymicrobial infections. Staphylococcus and Streptococcus were the most frequently isolated pathogens in medically intractable DFIs. Meanwhile, 16S sequencing revealed that Finegoldia, Prevotella, Anaerococcus, and Bacteroides are highly prevalent in medically intractable DFIs and seldom cultivated in culture studies. Previous studies using molecular techniques have also demonstrated that chronic DFIs are largely caused by anaerobes and polymicrobial infections (1,37). Johani et al. (38) conducted 16S amplicon sequencing in the intraoperative bone specimens of 20 patients with diabetic foot osteomyelitis. Their results revealed that 70% had polymicrobial infections, with Corynebacterium being the leading cause.

Additionally, nanopore 16S sequencing was even faster than culture studies for the detection of pathogens in DFIs. Although we did not put maximum effort into reducing the turnaround time in this study, the 16S sequencing method revealed the pathogen much earlier than the culture studies in most cases. At best, we believe that the turnaround time of 16S sequencing from DNA extraction to pathogen identification could be reduced to 6–9 h. This is comparable to our previous reports performed with other types of clinical samples (23). Nanopore sequencing is suitable for small, rapid sequencing tests; thus, rapid turnaround time will be appropriate for case-by-case applications.

Accurate diagnosis of pathogens could help to improve the poor prognosis of DFIs. Several factors are related to the poor prognosis of DFI, including the presence of microvascular disease, an impaired host immune response, and the occurrence of multilayered microbial communities within the wound known as biofilm (6,34,39). Empiric antibiotics are widely prescribed in the early stages of DFI on the basis of available clinical and epidemiological data (40–43); however, the standard treatment of DFIs involves debridement of the necrotic tissue and antimicrobial treatment targeting the pathogens isolated by culture-dependent methods (28,29,44). Although the need for surgery is caused by various factors (i.e., poor circulation, abscess, osteomyelitis) and not only by the wrong choice of initial antibiotics, we should endeavor to search for an improved method to choose empiric antibiotics. Since the clinical management of osteomyelitis is not universally well defined, use of metagenomics may impart additional wisdom not currently appreciated in clinical practice. By using nanopore 16S sequencing, it will be possible to investigate whether prescribing different empiric antibiotics on the basis of the types of pathogens from the early stage of DFIs can affect patient outcomes. If the comprehensive list of pathogens can be monitored consecutively, it will help to establish precision antimicrobial therapeutics in patients with DFI. Even without the exact susceptibility data of the pathogen, antibiotics can be adjusted according to susceptibility data obtained from the community, or sometimes antibiotics can be narrowed down depending on the remaining pathogens. The efficacy of an antibiotics adjustment strategy that is based on 16S sequencing in patients with DFI needs to be investigated.

There are several limitations to this study. Because the current study was performed in a single institution, the results could have been biased by several factors. There is a possibility of selection bias since SNUH is a tertiary referral hospital. In addition, the expertise of the diagnostic laboratory in culture studies may affect the observations. Moreover, 16S sequencing cannot provide information about antibiotics susceptibility, and the abundance of certain pathogens is not directly related to their virulence. Nevertheless, obtaining accurate information about pathogens in a timely manner with 16S sequencing will help in managing patients with DFI. The prospective application of 16S sequencing for making refined adjustment of antibiotics in DFIs and their impact on prognosis should be investigated in the near future.

In conclusion, nanopore 16S sequencing was particularly useful for pathogen identification in medically intractable DFI. Because of superior sensitivity over culture studies, a greater variety of bacteria, including unculturable anaerobes, was revealed by the nanopore 16S sequencing with a shorter turnaround time. Therefore, nanopore 16S sequencing should be applied more widely in the management of DFIs, and its effect on the outcome of patients should be evaluated in subsequent studies.

Funding. This work was supported by a National Research Foundation of Korea (NRF) grant funded by the Ministry of Science and ICT, Republic of Korea (NRF-2019R1A2C4070284).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. J.M., D.Y.L., and K.C. contributed to the conception and design of the study. J.M. and K.C. wrote the manuscript. N.K., S.-T.L., K.-H.J., and D.O.L. contributed to the acquisition and analysis of data. H.S.L. contributed to the literature search and figure generation. K.-I.P. and S.K.L. contributed to the data interpretation. All authors reviewed the manuscript for scholarly content and accuracy and gave approval for the final draft. D.Y.L. and K.C. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.