Introduction

Sepsis is among the most common causes of mortality for hospitalized patients worldwide, and its incidence is steadily increasing1,2. So far, blood-culture is commonly known as the gold standard for the detection of microbial pathogens in the bloodstream. This method, however, possesses some intrinsic limitations, as it is laborious and can only identify microbes that grow under optimized culture conditions3,4. Volumes of 20–30 ml of blood are required for aerobic and anaerobic microbial cultures5,6, such volumes may be a challenge to obtain from elderly or neonatal patients. The classical method is also difficult in establishing the diagnosis for patients who uptake antibiotics before7. In clinical practice, only 14–30% of sepsis cases are clinical-relevantly diagnosed by blood culture8,9. Therefore, it is essential to establish a rapid diagnostic procedure to complement, or even improve the performance of blood culture approaches.

Polymerase Chain Reaction (PCR) based techniques can sense small amounts of pathogen’s DNA directly from blood samples within 3–6 hours, thereby supporting a treatment afterward10. However, PCR-based diagnosis may significantly be thwarted by human DNA abundance and DNA fragments which are evolutionary conserved between humans and bacteria11,12,13,14,15. Recently, we demonstrated as a proof of concept that the depletion of human DNA from blood samples of septic patients prior to the PCR based diagnostic steps enhances the detection threshold up to 10 CFUs of E. coli per ml. By this assay, we showed an enhanced detection of up to 40% compared to 23% of blood cultures16. It was a promising result providing a new tool for septic diagnosis of Infectious disease specialists. However, in clinical practice, we realized that the group-specific primers/probes designed for Gram-negative, Gram-positive and Enterobacteriacea bacteria acquired unsatisfied performance. As a consequence, we omitted the group-specific screening and subjected the DNA-depleted specimens directly as samples for individual genus-specific real-time PCR assays which target common sepsis-causative pathogens, namely Escherichia coli, Klebsiella pneumoniae, Pseudomonas aeruginos, Acinetobacter baumanii, Neisseria meningitidis, Staphylococus sp, Staphylococus aureus, Streptococus spp., Streptococcus suis, S. pneumonia, Enterococcocus spp., Fusobacterium spp. and Bacteriodes spp. (Fig. 1). This modification now translated into a prototype sepsis diagnostic kit (so-called Sepsis@Quick) and is intellectually protected under Vietnamese national filed patent number 17556. In addition, during our clinical practice, we also documented some patients who were treated with antibiotics prior to blood-sampling for laboratory diagnosis, often got negative blood-culture results while their corresponding Sepsis@Quick based diagnostics were positive. Therefore, it prompted us to conduct this prospective observational study to comparatively evaluate the clinical uses of our novel Sepsis@Quick kit and conventional blood cultures for the diagnosis of bloodstream infection.

Figure 1
figure 1

Scheme of study design and workflow. (A) Former MCLB-1 based stepwise realtime PCR protocol including Group-specific screening by PCR reactions targeting bacterial 16SrRNA gene to differentiate Gram-positive, Gram-negative and Enterobacteriaceae groups. Samples positive in the screening assay were subjected to genus-specific real-time PCR reactions to detect 13 most common sepsis causative pathogens. (B) the current protocol of Sepsis@quick diagnostics kit was subjected to individual to genus-specific real-time PCR of E. coli, K. pneumoniae, P. auriginosa, A. baumannii, N. meningitidis, Staphylococus sp, S aureus, Streptococus sp. S. suis, S. pneumonia, Enterococcocus sp, Fusobacterium sp and Bacteriodes sp.

Full size image

Results

Characteristics of the study cohort

The characteristics of the study population according to the Sepsis-3 classification were shown in Table 1. In total, 144 patients were included: central nervous system infection (CNS, 24), pneumonia (34), intra-abdominal infection (39), urinary tract infection (18), skin-soft tissue infection (16) and unknown source (13). The average SOFA score was 7(4–10) (median (IQ)) and average APACHE II score was 17.2 ± 7.2. Among 144 study subjects, 125 patients had serological lactate measurements with average concentrations (median (IQ)) of 3.4 (2.1–6.2) mmol/L. 43 patients received continuous renal replacement therapies (CRRT); 81 patients were treated with antibiotics before blood-sampling for detecting pathogens and 74 out of those patients acquired antibiotic revision latter. Most of them were treated by antibiotics longer than 1 week before hospitalized (1 to 10 days) and cephalosporin III were more frequent used. 62 patients were diagnosed as septic shock and 64 patients had other comorbidities, including cancer, diabetes mellitus, HIV/AIDS, hypertension, congestive heart failure, coronary disease, chronic renal failure, chronic hepatic failure, chronic alcohol abuse and chronic obstructive pulmonary disease. The most frequent pathogens detected were Escherichia coli, Klebsiella pneumoniae, Streptoccocus spp. and Neisseria meningitidis; the time-frame for blood culture reading was 48 hours, whereas it took 4 hours only for Sepsis@Quick to establish the diagnosis.

Table 1 Clinical and laboratory characteristics of the study cohort.
Full size table

Capacity of pathogenic detection of Sepsis@Quick and blood culture

By using Sepsis@Quick test, 83 out of 144 recruited samples (57.6%) were identified pathogens including (E. coli (37), K. pneumoniae (15), N. meningitidis (6), S. suis (5); S. pneumoniae (5), S. aureus (1), Staphyloccocus sp (1), Streptoccocus spp. (10), Bacteriodes spp (1), Enterococcocus spp (3), A. baumannii (2). In addition, three out of 83 (3.6%) samples were detected with polymicrobial infection by Sepsis@Quick (Fig. 2). The conventional blood culture approach achieved positive colonies from 49 cases (34%); consisting of E. coli (24), K. pneumonia (12), N. menintigitis (1), S. suis (4); S. pneumonia (1), S. aureus (1), S. mitis (1), S. sangius (1), A. hydrophilia (1), A. sobria (1), M. morgani (1) and A. baumannii (1), (Fig. 2); 43 out of 49 culture-positive cases (87.8%) (Fig. 3) were also detected by Sepsis@Quick. E. coli and K. pneumonia were the most frequently detected pathogens by both approaches. Detailed patients’ characteristics according to Sepsis@Quick and blood culture results are shown in Table 1.

Figure 2
figure 2

Numbers of individual pathogens detected by blood culture and Sepsis@quick: Upper panel shows the numbers positive cases or number of polymicrobial infection detected by blood culture or sepsis@quick, lower panel shows number of individual microbial pathogens detected by blood culture or sepsis@quick.

Full size image
Figure 3
figure 3

Diagnostic performance and concordance portion between the two methods: Total 144 patient blood samples were recruited into this current study, of which, Sepsis@Quick identified 83/144 (57.64%) positive. Blood culture approach detected 49 cases (34%) positive and 43 out of 49 culture-positive cases (87.8%) were co-detected by Sepsis@Quick and only 40 cases (27.78) were not diagnosed.

Full size image

Influence of prior antibiotic treatment on diagnostic performance of blood culture and Sepsis@Quick

Our data revealed that prior initiation of antibiotic treatment significantly reduced positivity rates acquired by blood cultures (from 42.9% to 27.2%; P = 0.049), while diagnostics results achieved by Sepsis@Quick did not differ between naïve patients (57.1%) and antibiotics-treated patients (69.1%) (P = 0.92), (Table 2).

Table 2 Influence of previous treatment with antibiotics on diagnostics performance of Sepsis@quick and blood culture.
Full size table

Sepsis@quick influences better to the revision of antibiotics therapies

Admitted patients who were diagnosed with either severe sepsis or septic shock would be treated with empirical broad-spectrum antibiotics such as cephalosporin III or/and fluoroquinolon or/and aminoglycoside, carbapenem → cephalosporin III or carbapenem or/and fluoroquinolon or/and aminoglycoside) before or without the identification of causative microbial pathogens. And once, causative pathogens were identified, the previously indicated antibiotics would be revised. In our study cohort, we documented that the revisions of antibiotic therapies happened to 74 cases. As seen in Table 3, sepsis@Quick provided positive diagnostics for 45 cases, whereas blood culture only detected microbial pathogens only in 26 scenarios (P = 0.0017); Especially, amongst 20 cases for whom, the antibiotic de-escalation were indicated, Sepsis@Quick identified positive results for 15 cases, which is significant higher that detected by blood culture (8 cases, P = 0.025).

Table 3 Influence of Sepsis@quick on revision of antibiotic administration.
Full size table

Sepsis@Quick diagnosis performance reduces sepsis-related mortality

Our data reveals that even though the SOFA score17 (most comprehensive criteria for organ failure evaluation of sepsis patients) was not significantly different between the two subgroups (Sepsis@quick (+), n = 83 versus (Sepsis@quick (−), n = 61) or between two subgroups (BC(−)/Sepsis@quick (+) n = 40) versus (BC(−)/Sepsis@quick (−) n = 55), (Supplementary Fig. 1) the mortality rates were clearly distinguishable between of Sepsis@quick (+) (n = 83, 37.3%) versus Sepsis@quick (−) (n = 61, 54.1%), p = 0.046 and especially in patient group with blood culture negative, the use of sepsis@quick really associated with the reduction of sepsis related mortality rate (BC(−)/Sepsis@quick (+) n = 40, 30%) versus (BC(−)/Sepsis@quick(−) n = 55, 58.2%) p = 0.007 (Fig. 4).

Figure 4
figure 4

Relationship between sepsis-related mortality rate and diagnostic performance of Sepsis@quick: Two left bars show comparisons of mortality rate between patient group carrying sespsis@quick positive result versus patient group with sespsis@quick negative diagnosis. Two right bars show comparisons of mortality rate between patient group of blood culture negative but sepsis@quick positive versus patient group with blood culture negative and sepsis@quick negative (double nengative). Both comparison were performed by Chi-square test.

Full size image

Discussion

Currently, blood culture and DNA based PCR methods are the main tools for identification of sepsis causative pathogens. However, while blood culture is time-consuming and has other drawbacks, including the difficulty to apply optimal growth conditions for unknown or/and fastidious pathogens, large volumes of blood required for aerobic and anaerobic cultures5,6, impairment of cultures due to pre-treatment with antibiotics, the direct identification of pathogens by PCR is difficult due to low concentrations of bacteria in blood samples and the strongly inhibitory effect of human DNA. Hence, in daily clinical practice, diagnostics results gained by the two methodologies are not frequently consistent.

Our research group and Oliver Liesenfeld et al. have already demonstrated a proof-of-concept that by reducing the amount of human DNA from septic patients’ blood helps to significantly enhance the detection limit of the downstream PCRs in the diagnosis of blood stream infection15,16. In this study, we modified a previous assay to formulate a prototype diagnosis kit called Sepsis@Quick and evaluate the clinical performance of this novel kit against that of conventional blood culture for identification of sepsis causative pathogens.

Our data showed that Sepsis@Quick testing was superior to blood cultures. With the use of Sepsis@Quick, more pathogens and even polymicrobial infection were identified. The Sepsis@Quick’s performance was faster and less influenced by previous antibiotic treatment. Moreover, positive results achieved by Sepsis@Quick in a shorter diagnostic time provided additional awareness that helps clinicians to revise the previous use of antibiotics or/and drive the changes in therapeutics practices properly; thereby indirectly reducing the sepsis-related mortality rate. In the present study, blood cultures were able to establish a diagnosis for 34% of recruited patients; this is in the range gained by other studies18,19. In contrast, Sepsis@Quick could provide diagnoses for 57.6% of the study group. In addition, more than 87% of blood-culture positive cases were confirmed by the Sepsis@Quick system. In other studies, results obtained by another PCR-based system (SeptiFast) were quite variable while the SeptiFast assay had a higher diagnostic performance compared to blood culture20,21,22,23,24. Whereas, both methods displayed either similar sensitivities or the SeptiFast assay even acquired less favorability with respect to blood culture13,19,25,26.

Despite the recommendations from the Surviving Sepsis Campaign on archiving blood samples for culture prior to treatment with broad-spectrum antibiotics27, about 50–70% of septic patients receive antibiotics before blood samples are collected for culture28. Under antibiotic pressure, it is hard for some bacteria to form colonies, hence reducing the effectiveness of blood cultures, considering this and keeping in mind that blood-culture normally provides a diagnosis after 48 hrs or even longer, there would be no result available after four hours upon patient recruitment if only the blood-culture approach is used in clinical practices. In our study, the identification of sepsis-causing pathogens by Sepsis@Quick kit was finalized as quickly as four hours but our molecular approach is an open system that allows the users to economically customize it with a number of pathogens wanted relevant to given local medical settings. Users of the Sepsis@Quick system do not need to process patient samples in batches mode but can conduct the assay for individual patients upon the collection of samples, ensuring the availability of Sepsis@Quick results in four hours; to this point, other similar studies also support our findings. In the previous studies, the SeptiFast kit was used to detect the pathogens significantly earlier than blood cultures19,20. However, we are the first to demonstrate that positive results provided by PCR-based approaches help to decrease sepsis-associated mortality. In another aspect, although, Sepsis@Quick identify a narrow range of detectability than SeptiFast does, SeptiFast is a closed system with a given number of detectable pathogens13, whereas Sepsis@Quick is an open system, allowing ones to customize for the detection of pathogens relevant for given medical settings29.

Nevertheless, there are several limitations in the present study: it was a single center observational study including a high proportion of sepsis shock patients (n = 62.4%) who might carry higher densities of bacterial loads in the peripheral blood; this might limit the generalizability of the findings. However, the results are consistent with those generated by other investigators, who also compared the performance of PCR versus blood culture for identification of sepsis causative pathogens20,22,23,24, suggesting that these findings may apply to other patient cohorts as well. Our current version of Sepsis@Quick assay cannot detect fungi and other bacterial pathogens such as Aspergillus fumigatus, Candida spp., Burkholderia spp., A. hydrophilia, A. sobria, M. morgani, E. aerogenes/cloacae, M. morganii and S. marcescens. This may partially explain the lack of DNA amplification in about 14% of episodes, whereas blood culture was still able to show growth. Hence, inclusion of more pathogens into the future forms of Sepsis@Quick is needed to improve the pathogenic detection spectrum.

Conclusions

The novel PCR-based Sepsis@Quick system provides faster detection and identifies more bacterial pathogens than the conventional blood culture, particularly in patients who undergo previous antibiotics treatment, hence reducing sepsis-related mortality.

Methods

Study cohort and patient blood sampling

A total of 144 blood samples from naïve and septic patients pre-treated with antibiotics were collected at the 108 Military Central Hospital, Vietnam. Diagnoses of sepsis were confirmed by physicians in accordance with an acute increase in Sequential Organ Failure Assessment (SOFA) scores by two or more as defined by the Sepsis-3 guidelines17. Further characteristics of the patients (age, sex, comorbidity, immunosuppression, the SOFA scores and signs of organ dysfunction were documented at admission (Table 1)17.

Study design

All methods used in this study were in accordance with the relevant guidelines and regulations and were approved by the institutional review board and an independent Ethics Committee of the 108 Military Central Hospital, Hanoi, Vietnam. Informed written consent was obtained from all study patients. Twelve milliliters of venous blood collected in duplicates from all patients were subjected to both blood culture and the Sepsis@Quick test in order to identify the causing pathogen(s) (Fig. 1).

Preparation of reagents under aseptic conditions

For decontamination of the PCR master mix, 8-methoxypsoralen was dissolved in dimethylsulfoxide (DMSO) (Sigma-Aldrich, St. Louis, MO, USA). For decontamination of the real-time PCR master mix reagents, 25 µg/mL of 8-methoxypsoralen (Sigma-Aldrich) and 10 min of exposure to UV irradiation at 366 nm over a distance of three cm were applied30. Components included in the Sepsis@Quick kit were also decontaminated by UV irradiation at 280 nm over a distance shorter than five cm30. Working places, as well as all accessories and tools, routinely undergo weekly decontamination by spraying with DNA-ExitusPlus™ solution (AppliChem, Darmstadt, Germany).

Enrichment of bacterial DNA

Enrichment of bacterial DNA for downstream PCR analysis was performed similarly to our previous study with minor modifications16. In brief, 2 volumes of 1.2 ml peripheral venous blood samples obtained from every sepsis patients recruited were added to an equal volume of mammalian cell lysis buffer (MCLB-1, containing Na2CO3 and Triton-X100) for three minutes at 37 °C to allow for shearing of human chromatin into DNA fragments. After the incubation step, an equal volume of neutralization buffer (NB buffer containing Tris–HCl) was added in order to prevent further cell lysis. The samples were then centrifuged at 5000 g for 5 min. Supernatants were discarded and bacterial pellets were reconstituted in 200 μl of 1x Tris buffered saline (TBS, pH 7.6) and used for isolation of bacterial DNA by the Automatic Nucleic Acids Extraction System (Sacace Biotechnologies, Como Italy). Eluted DNA was reconstituted in 200 ul 25 mM Tris–HCl pH8.0 for downstream PCR analysis (MCLB-1 and NB buffer are included in the Sepsis@Quick kit).

Real-time PCR conditions

The real-time PCR assay mixtures consisted of 7.5 µl Taqman real-time PCR master mix (Qiagen, Hilden, Germany), 5 µl of DNA template, 5 pmol of primers and 0.2 pmol of probes (included in the Sepsis@Quick kit). Reactions were run in the Stratagene M3000p (San Diego, CA, USA) device with a pre-incubation step at 50 °C for 15 min, initial denaturation at 95 °C for 5 min, followed by 45 cycles of 95 °C for 15 sec and 60 °C for 60 sec.

Automatic bacterial culture

Two volumes of collected blood samples from each patient were subjected for blood culture in the BD BACTECTM 9120 system (Becton Dickinson, New Jersey, USA) at 36 °C ± 0.5 °C for 48 hours. Once the growth occurred in both tubes, identification of bacterial species was performed using the VITEK® 2 automated system (BioMérieux, Marcy-l’Étoile, France).

Statistical analysis

All statistical analyses were performed using the SPSS software (v19). Values are given as numbers with percentages and median with interquartile where appropriate. Chi-square or Fisher’s exact tests were performed to compare categorical variables and positivity rates of the different diagnostic test systems. Mann-Whitney U or Kruskal–Wallis was used to compare two or more means, respectively. Sensitivity and specificity of the Sepsis@Quick system were calculated by using the blood culture positive result or Sepsis-3 criteria as reference17. The odds ratio and 95% confidence intervals (CIs) were calculated in order to estimate the association of negative results of both systems with previous exposure to antibiotics; The level of statistical significance was set at 0.05 and all tests were 2-tailed.