The application of nanopore targeted sequencing in the diagnosis and antimicrobial treatment guidance of bloodstream infection of febrile neutropenia patients with hematologic disease

Abstract Traditional microbiological methodology has limited sensitivity, detection range, and turnaround times in diagnosis of bloodstream infection in Febrile Neutropenia (FN) patients. A more rapid and sensitive detection technology is urgently needed. Here we used the newly developed Nanapore targeted sequencing (NTS) to diagnose the pathogens in blood samples. The diagnostic performance (sensitivity, specificity and turnaround time) of NTS detection of 202 blood samples from FN patients with hematologic disease was evaluated in comparison to blood culture and nested Polymerase Chain Reaction (PCR) followed by sanger sequence. The impact of NTS results on antibiotic treatment modification, the effectivity and mortality of the patients under the guidance of NTS results were assessed. The data showed that NTS had clinical sensitivity of 92.11%, clinical specificity of 78.41% compared with the blood culture and PCR combination. Importantly, the turnaround time for NTS was <24 h for all specimens, and the pre‐report time within 6 h in emergency cases was possible in clinical practice. Among 118 NTS positive patients, 98.3% patients' antibiotic regimens were guided according to NTS results. There was no significant difference in effectivity and mortality rate between Antibiotic regimen switched according to NTS group and Antibiotic regimen covering pathogens detected by NTS group. Therefore, NTS could yield a higher sensitivity, specificity and shorter turnaround time for broad‐spectrum pathogens identification in blood samples detection compared with traditional tests. It's also a good guidance in clinical targeted antibiotic treatment for FN patients with hematologic disease, thereby emerging as a promising technology for detecting infectious disease.


| INTRODUC TI ON
Febrile Neutropenia (FN) is considered as a serious clinical condition with high risk of morbidity and mortality attributed to the immune system deficiency which can not effectively fight against pathogens. 1,2 Patients with hematologic disease have a higher incidence of developing FN due to myelo-suppressive after chemotherapy and concurrent haematopoietic dysfunction, and the fatality rate is extremely increasing once the FN occurs. 2 As bloodstream infections in FN patients can rapidly progress into respiratory and circulatory failure, it should be regarded as a medical emergency, requiring immediate antibiotic treatment. 3 Early and effective microbial diagnosis is essential for guiding precise and successful antibiotic therapy administration and thus prolonging the survival of the patients.
Blood culture is regarded as the gold standard for etiological diagnosis of bloodstream infection, but a low clinical sensitivity especially in some pathogens difficult to grow in culture and a long incubation cycle of about 48 hours greatly impedes its guidance on precise pathogen diagnosis. 4 PCR is a specific, rapid, and economic technology for detecting specific microorganism. However, it has small detection range and is unable to test unknown pathogens and precisely analyse amplified nucleic acid sequences. 5 With the rapid development of sequencing technology, the metagenomic next generation sequencing (mNGS) technology has emerged as a novel culture-independent techniques and has been applied to improve the sensitivity and specificity in identifying pathogens, but it still has some shortcoming such as high costs, long turnaround times and low sensitivity because it needs deeper sequencing and more extensive bioinformatics to interpret. 6,7 Thus, a more targeted sequence technology is urgently needed in clinic.
Unlike previous sequencing, Nanopore targeted sequencing (NTS) was developed by amplifying 16 s rRNA gene (for bacteria), ITS1/2 gene (for fungi), and specific gene (for virus) and using nanopore sequencing platform to sequence the amplified marker genes. Sequencing of the phylogenetic marker genes is a popular approach for identifying microbial species. 8 level than the partial 16 s rDNA. 11 Our NTS is designed to amplify the 16 s rDNA and ITS1-2 because there is no limitation of reads length in Nanopore. The limit of detection (LOD) by NTS has been estimated to be 25 CFU/mL with a mock community composed of six pathogens in our previous article. 12 Compared with mNGS, NTS requires less sequencing data and bioinformatic resources in theory. Furthermore, the sensitivity of NTS is also increased and the cost and turnaround time are reduced at the same time. An inhouse bioinformatic analysis pipeline was enough to diagnose the infectious pathogens by mapping the sequencing results with the constructed databases. 5

| NTS methodology
All DNA extraction and DNA amplification methods were performed based on previously reported method with some modifications. 5,17,18 Briefly, 1.5 ml of EDTA-whole blood samples were centrifuged at 800 × g for 10 min at room temperature. Then, the lower part of  Table S1. Amplification of marker genes was performed and the PCR product of clinical samples, two ETC and two Tris-EDTA buffer (no-template control, NTC) were batched in one sequencing library and the library was sequenced using Oxford Nanopore GridION X5 (Appendix S1).

| Bioinformatics methodology
The bioinformatics method is described in the Appendix S1. For pathogen determination, four controls, including two ETC and two NTC, were designed for filtering out contaminants from NTS laboratory process and from human normal flora during sampling based on previously published method. 7 A reportable list of clinical pathogens was set up referred to published papers which apply next generation sequencing to identify the pathogens in sepsis (kariu sdx.com/kariu s-test/patho gens). 19 The detailed logic of filtering potential laboratory contaminants was performed based on previously reported method 7,19 ( Figure S2). If the antibiotic was neither covering pathogens detected by NTS nor switched rationally, the patients would be classified into NTS results not be considered group.

| Statistical analysis
The sensitivity, specificity, positive predicted value (PPV), and negative predicted value (NPV) were calculated as previously described by comparing the NTS results with the composite clinical diagnosis. 21 Statistical significance was set at a p value of <0.05. Proportional outcomes were compared using the χ 2 test or Fisher's exact test. Continuous variables were compared using the Student's t-test or Wilcoxon signed-rank test. The optimal longterm negative control fold change (LNC-FC) and dynamic negative control fold change (DNC-FC) threshold were calculated by ROC with maximizing the Jordan index. Data analyses were performed using the R software (www.r-proje ct.org).

| Sample and Patient Characteristics
The clinical characteristic of enrolled patients was shown in Table 1.

| Distribution of pathogens population detected by NTS
To assess the validity of NTS for detecting a broad range of infecting To check whether the number of read assignments of NTS was related to the time from sample receipt to report of blood cultures, we analysed 20 specimens positively identified by both NTS and blood culture. As shown in Figure 1E, there was no significant correlation between the read assignments number and the sample receipt to report time (Spearman, p = 0.31). To further check whether blood culture positive or negative had correlation with read assignments number, we divided all the cases with NTS detected K. pneumonia or E.coli into positive or negative groups according to blood culture results, the pathogens from blood culture positive samples did not show a higher read abundance than the samples with blood culture negative samples ( Figure 1F), which indicated that the read assignments number of NTS might not directly indicate the precise content of active pathogens in blood.

| Comparison of diagnostic performance between NTS and culture
Among all cases, 30 (14.85%) cases were positive detected by blood culture and 128 (63.36%) cases were detected positive by NTS (Figure 2A,B). To verify the accuracy of NTS, nested PCR followed by sanger sequencing were tested on 133 cases who are either positive detected by NTS or by blood culture or by both (Figure 2A).

| Turnaround time in clinical practice
The time from sample collection to the final report was within 24 h for NTS, which was significantly shorter than culture methods ( Figure 3A). In some emergency cases, the data sequenced for 1 hour could be basically saturated and sufficient for subsequent bioinformatics analysis to identify the pathogens ( Figure 3B,C), which allowed the clinicians to administer targeted antibiotics on the same day without waiting for the completion of an 8 h sequencing.

| Impact of NTS results in clinical antimicrobial treatment
To assess the impact of NTS results on clinical antimicrobial treatment, we classified the NTS positive (for bacteria and fungi) patients (n = 118) into three groups according to the impact of NTS results in clinical antibiotic regimen modification as explained in method section ( Figure 4A). The anti-infection effectivity and mortality rate were analysed between 'Antibiotic regimen switched according to NTS group' and 'Antibiotic regimen covering pathogens detected by NTS group'. As seen in Figure 4B, the patients whose antibiotic regimen adjusted based on NTS results could acquire similar effectivity and mortality with the patients whose empirical antibiotic regimen covered NTS results from the beginning.
As seen from two typical cases in Figure.      which lead to the omission of these rare pathogens. In our future studies, more systematic detection and quality control, like wholeassay internal normalization and batch controls will be brought in to improve the sensitivity and specificity of NTS. 19 Besides, fast host depletion methods will be introduced to reduce the influence of human DNA, 25 and targeted enrichment methods will be adopted to improve the sensitivity of specific pathogens in bloodstream infection. 26 Fouthly, NTS is its indiscriminately detecting all DNA from active, non-active pathogens and fragments of cfDNA disintegrated by pathogens. Therefore, in the future, RNA sequencing might be adopted to detect active pathogens so as to better guiding the use of antibiotic in clinic. 27 29 First, filter has been applied in closely related microorganisms and only highly trusted species is kept. 10 Second, all high-similar complexes have been identified by analysing all standard sequences and the similar homologous species are reported as a complex in clinical report, which has been adopted in other literature. 19 In conclusion, NTS is a relative efficient and sensitive method which enable simultaneous detection and same-day reporting of a huge quantity of pathogens in blood samples. This approach makes up the gap between highly targeted PCR-based methods and resource-intensive sequencing-based mNGS method. NTS is recommended for monitoring intensive patients during treatment because of its accuracy, comprehensiveness, and rapidity.

CO N FLI C T O F I NTE R E S T
There was no conflict of interest. No external funding was received.

DATA AVA I L A B I L I T Y S TAT E M E N T
The raw sequencing data can be obtained through the National Omics Data Encyclopedia with the accession number (https://www. biosi no.org/node/proje ct/detai l/OEP00 3341).