Early application of metagenomics next-generation sequencing may significantly reduce unnecessary consumption of antibiotics in patients with fever of unknown origin

Abstract

Background: Metagenomic next-generation sequencing (mNGS) is a novel nucleic acid method for the detection of unknown and difficult pathogenic microorganisms, and its application in the etiological diagnosis of fever of unknown origin (FUO) is less reported. We aimed to comprehensively assess the value of mNGS in the etiologic diagnosis of FUO by the pathogen spectrum and diagnostic performance, and to investigate whether it is different in the time to diagnosis, length of hospitalization, antibiotic consumption and cost between FUO patients with and without early application of mNGS.

Methods: A total of 149 FUO inpatients underwent both mNGS and routine pathogen detection was retrospectively analyzed. The diagnostic performance of mNGS, culture and CMTs for the final clinical diagnosis was evaluated by using sensitivity, specificity, positive predictive value, negative predictive value and total conforming rate. Patients were furtherly divided into two groups: the earlier mNGS detection group (sampling time: 0 to 3 days of the admission) and the later mNGS detection group (sampling time: after 3 days of the admission). The length of hospital stay, time spent on diagnosis, cost and consumption of antibiotics were compared between the two groups.

Results: Compared with the conventional microbiological methods, mNGS detected much more species and had the higher negative predictive (67.6%) and total conforming rate (65.1%). Patients with mNGS sampled earlier had a significantly shorter time to diagnosis (6.05+/-6.23 vs. 10.5+/-6.4 days, P < 0.001) and days of hospital stay (13.7+/-20.0 vs. 30.3 +/-26.9, P < 0.001), as well as a significantly less consumption (13.3+/-7.8 vs. 19.5+/-8.0, P < 0.001) and cost (4543+/-7326 vs. 9873 +/- 9958 China Yuan [CNY], P = 0.001) of antibiotics compared with the patients sampled later.

Conclusions: mNGS could significantly improve the detected pathogen spectrum, clinical conforming rate of pathogens while having good negative predictive value for ruling out infections. Early mNGS detection may shorten the diagnosis time and hospitalization days and reduce unnecessary consumption of antibiotics.

Keywords: Consumption of antibiotics; Fever of unknown origin; Metagenomic next-generation sequencing; Pathogen detection; Pathogen spectrum.

Fig. 1

Flowchart of the study design

Fig. 2

Pathogen spectrum of pathogens. (A) The distribution of pathogens detected by culture and CMTs. (B) The distribution of pathogens detected by metagenomic next generation sequencing (mNGS). Each bar is divided into two parts, indicated by different colors. Verified represents pathogens detected by traditional detection methods (culture and CMTs) or mNGS that were verified by clinical diagnosis, while unverified represents pathogens detected that were not supported by clinical diagnosis

Fig. 3

Types of samples used for pathogen detection

Fig. 4

Comparison of the diagnostic performance of different methods. (A) Comparison of the diagnostic performance of the three methods in all samples. (B) Comparison of the diagnostic performance of the three methods in blood and bone marrow samples. Differences in significance between mNGS and culture, as well as mNGS and CMTs have been labeled. Significance was calculated by chi-square test. * represents p < 0.05, ** represents p < 0.01, and *** represents p < 0.001

Fig. 5

Effect of different sampling time for mNGS detection on (A) time spent on diagnosis, (B) length of hospitalization, (C) Medication costs per capita, and (D) defined daily doses (DDDs). * represents p < 0.05, ** represents p < 0.01, and *** represents p < 0.001