Next-generation sequencing diagnostics of bacteremia in septic patients

Abstract

Background: Bloodstream infections remain one of the major challenges in intensive care units, leading to sepsis or even septic shock in many cases. Due to the lack of timely diagnostic approaches with sufficient sensitivity, mortality rates of sepsis are still unacceptably high. However a prompt diagnosis of the causative microorganism is critical to significantly improve outcome of bloodstream infections. Although various targeted molecular tests for blood samples are available, time-consuming blood culture-based approaches still represent the standard of care for the identification of bacteria.

Methods: Here we describe the establishment of a complete diagnostic workflow for the identification of infectious microorganisms from seven septic patients based on unbiased sequence analyses of free circulating DNA from plasma by next-generation sequencing.

Results: We found significant levels of DNA fragments derived from pathogenic bacteria in samples from septic patients. Quantitative evaluation of normalized read counts and introduction of a sepsis indicating quantifier (SIQ) score allowed for an unambiguous identification of Gram-positive as well as Gram-negative bacteria that exactly matched with blood cultures from corresponding patient samples. In addition, we also identified species from samples where blood cultures were negative. Reads of non-human origin also comprised fragments derived from antimicrobial resistance genes, showing that, in principle, prediction of specific types of resistance might be possible.

Conclusions: The complete workflow from sample preparation to species identification report could be accomplished in roughly 30 h, thus making this approach a promising diagnostic platform for critically ill patients suffering from bloodstream infections.

Keywords: Circulating nucleic acids; Diagnostics; Next-generation sequencing; Sepsis.

Figures

Fig. 1

Distribution of cfDNA concentrations over different patient groups and time points. a Comparison of cfDNA concentrations between healthy volunteers (V), septic patients at the onset of sepsis (S T0), and non-infected patients following major abdominal surgery (P). b Alterations in cfDNA concentrations of septic patients’ plasma samples collected over the observational period of the trial. Samples were obtained at sepsis onset (T0), after 24 h (T1), 4 days (T2), 7 days (T3), 14 days (T4), and 28 days (T5). c Comparison of cfDNA concentrations in patients undergoing major abdominal surgery without evidence of infection. Blood samples from the postoperative group were collected prior to surgery (T0), immediately following the end of the surgical procedure (T1), and 24 h later (T2)

Fig. 2

Rationale of the SIQ score and SIQ plot. a Outline for obtaining a SIQ score and SIQ plot. Total cfDNA is isolated from a patient’s plasma and sequenced. From sequencing results, human cfDNA are removed after mapping and only unmapped reads are further processed. From these unmapped reads, microbial species are classified and reads are normalized, counted, and sorted by their abundance. For each species obtained from a patient, results are compared with likewise processed samples of uninfected controls, exemplified for microbial species X, which is found in the patient’s sample as well as in most control samples and, therefore, represents a contaminant. However, species Y is found in high abundance only in the patient’s sample and in none of the controls and, therefore, receives a high significance and consequently a high SIQ score, indicated by the radius of its data point in the SIQ plot. b Distribution of normalized counts for each species found in the plasma sample of patient S9 at the onset of sepsis (T0). Only the most abundant species, Enterobacter cloacae, was labeled. c Distribution of the normalized counts for E. cloacae for all samples analyzed. Red, septic patients; blue, controls (elective surgery and healthy volunteers). Only sample S9 with the most abundant E. cloacae reads was labeled. d Distribution of the normalized counts of Propionibacterium acnes for all samples. Red, septic patients; blue, controls (elective surgery and healthy volunteers). e SIQ plot integrating abundance and significance of all species for patient S9 at the onset of sepsis (T0). Coordinates of the data points (species) are the relative abundance (log2) on the x-axis and the significance expressed as 1 − p value on the y-axis. The dashed line marks a p value of 0.05. Data points with log2 > 0 and p value <0.05 are labeled. The SIQ score of a species in the respective sample is integrated as the radius of the data point

Fig. 3

Time course SIQ analyses compared with conventional clinical microbiology data for two patients. a Time course of patient S10. A 68-year-old male patient presented with a tumor of his stomach with the need for a gastrectomy. Following the surgical procedure the patient suffered from septic shock due to severe pneumonia without any evidence of an anastomosis insufficiency. Staphylococcus aureus was shown to be the dominant organism in different secretions (e.g., tracheal secretion, abdominal wound swab, blood culture, etc.). In addition, pneumonia was shown to be accompanied (respectively boosted) by reactivation of herpes simplex virus type 1 (HSV1) in tracheal secretions. Following a prolonged weaning phase, the patient was then able to be discharged to the normal ward 6 weeks after the onset of septic shock. In this figure, the antibiotic treatment regime, SIQ scores for species identified via NGS/SEPseq, and cfDNA concentrations of the respective plasma samples are plotted for the trial period of 28 days. Pertinent (clinical microbiology) laboratory results are marked using arrows to indicate the day the clinical specimen was obtained. Abbreviations: BC blood culture, CVC central venous catheter, TS tracheal secretion, HSV herpes simplex virus, CIP ciproflocaxine, MTZ metronidazole, MEM meropenem, VAN vancomycin, CFG caspofungin, FLX flucloxacillin, FLC fluconazole, ACV aciclovir, AFG anidulafungin, TGC tigecycline. Anti-infectives are displayed as antibacterial antibiotics, antimycotics, and antivirals in light grey, black, and dark grey, respectively. The relative amount of bacteria found by conventional clinical microbiology is indicated with plenty (p), medium (m), or scarce (s). (For a detailed list of the anti-infective abbreviations, see Table S5.) b Time course of patient S60. Following a complicated course of perforated sigmoid diverticulitis, a 70-year-old female patient presented for reconstruction of bowel continuity. In the postoperative phase the patient developed septic shock due to bowel leakage with the need for surgical revision. Abdominal wound swabs were shown to be positive for Escherichia coli and Enterococcus faecium. One day later the patient suffered from a second septic hit due to perforation of the colon with the need for surgical colectomy and construction of a stump by Hartmann. Afterwards the patient suffered from another septic hit due to an insufficiency of the stump by Hartmann. Accordingly, one further explorative laparotomy was performed and an intensive abdominal lavage was initiated. In the further course of the septic disease the patient developed a fourth septic hit due to ventilator-associated pneumonia triggered by E. coli, Stenotrophomonas maltophilia, and Klebsiella pneumoniae. Following a prolonged weaning phase the patient was then able to be transferred to the intermediate care ward after 3 months of ICU treatment. Ultimately, the patient could be discharged from hospital another 2 weeks later. Pertinent (clinical microbiology) laboratory results are marked using arrows to indicate the day the clinical specimen was obtained. Abbreviations: BC blood culture, CVC central venous catheter, TS tracheal secretion, BAL bronchoalveolar lavage, HSV1 herpes simplex virus 1, IPM imipenem, LZD linezolid, CFG caspofungin, ACV aciclovir, TZP piperacillin-tazobactam, CTX cotrimoxazol, CAZ ceftazidime. Antibacterial antibiotics are colored in light grey. The relative amount of bacteria found by conventional clinical microbiology is indicated with plenty (p), medium (m), or scarce (s). (For a detailed list of the anti-infective abbreviations, see Additional file 9: Table S5)

Fig. 4

Genome coverage and resistance profile of a patient infected with vancomycin-resistant E. faecium (VRE). a Mean genome coverage of approximately 0.3 of the E. faecium genome (2.8 Mb). b Table with hits to the CARD database. The CARD/GenBank accession number is listed as well as the alias gene name, gene coverage calculated from read length ratio to gene length, number of reads mapped to this gene, and the respective organism to which the sequence is assigned