Diagnostic performance and clinical impact of blood metagenomic next-generation sequencing in ICU patients suspected monomicrobial and polymicrobial bloodstream infections

Abstract

Introduction: Early and effective application of antimicrobial medication has been evidenced to improve outcomes of patients with bloodstream infection (BSI). However, conventional microbiological tests (CMTs) have a number of limitations that hamper a rapid diagnosis.

Methods: We retrospectively collected 162 cases suspected BSI from intensive care unit with blood metagenomics next-generation sequencing (mNGS) results, to comparatively evaluate the diagnostic performance and the clinical impact on antibiotics usage of mNGS.

Results and discussion: Results showed that compared with blood culture, mNGS detected a greater number of pathogens, especially for Aspergillus spp, and yielded a significantly higher positive rate. With the final clinical diagnosis as the standard, the sensitivity of mNGS (excluding viruses) was 58.06%, significantly higher than that of blood culture (34.68%, P<0.001). Combing blood mNGS and culture results, the sensitivity improved to 72.58%. Forty-six patients had infected by mixed pathogens, among which Klebsiella pneumoniae and Acinetobacter baumannii contributed most. Compared to monomicrobial, cases with polymicrobial BSI exhibited dramatically higher level of SOFA, AST, hospitalized mortality and 90-day mortality (P<0.05). A total of 101 patients underwent antibiotics adjustment, among which 85 were adjusted according to microbiological results, including 45 cases based on the mNGS results (40 cases escalation and 5 cases de-escalation) and 32 cases on blood culture. Collectively, for patients suspected BSI in critical condition, mNGS results can provide valuable diagnostic information and contribute to the optimizing of antibiotic treatment. Combining conventional tests with mNGS may significantly improve the detection rate for pathogens and optimize antibiotic treatment in critically ill patients with BSI.

Keywords: blood culture; bloodstream infection; intensive care unit; metagenomics next generation sequencing; polymicrobial infection.

Figure 1

A schematic of the study profile.

Figure 2

Distribution of pathogens identified by blood mNGS and blood culture. (A) Gram-negative bacteria. (B) Gram-positive bacteria. (C) fungi. (D) Viruses. **P<0.01 and ****P<0.0001. CMV, Cytomegalovirus; EBV, Epstein-Barr virus; TTV, Torque Teno virus; HHV, human herpes virus; HBV, Hepatitis B virus; JCPyV, JC polyomavirus.

Figure 3

The positivity distribution of blood mNGS and blood culture. (A) The positive rates of mNGS and blood culture for all cases (n=162). ****P<0.0001. (B) The positive rates of mNGS and blood culture for all cases (n=162) after viruses detected by mNGS were excluded. ****P<0.0001. (C) Pie chart demonstrating the positivity distribution of mNGS and blood culture for all tests. The double-positive samples were further categorized as completely matched, partly matched (at least one overlap of pathogens was observed) and mismatched.

Figure 4

The infection types and responsible pathogens. (A) ROC curve of mNGS, blood culture and mNGS+culture with reference to final clinical diagnosis. (B) Pie chart demonstrating the infectious type, and the co-infection was further categorized as bacteria-bacteria, bacteria-fungi, bacteria-virus and fungi-fungi. (C) the distribution of microbial species that involving in single infection and co-infection.

Figure 5

The adjustment of empirical antimicrobial treatment. (A) Pie chart illustrating the coverage of initial empirical antibiotics usage. (B) the adjustment of empirical antibiotics usage in each group. (C) the basis for the adjustment of empirical antibiotics usage.