*3.6. Antimicrobial Therapy*

Results from additional NGS analysis affected therapy in 20% of all cases (5/25) (Table 3, Figure 1). Identification of clinically relevant pathogens by NGS led to the initiation of four antimicrobial therapies in three cases:


Besides clinical factors, NGS further contributed to the continuation of empirical antimicrobial therapy in three cases. Considering the lack of antimicrobial susceptibility testing, empirical antimicrobial treatment was assumed to be appropriate after NGS provided the only identification of causative pathogens in blood samples. NGS did not substantially contribute to the termination of therapy in any of these cases.

**Table 3.** Contribution of NGS to the optimization of antimicrobial therapy. NGS results led to the initiation of four targeted therapies in three cases. Aciclovir was administered in two cases after HSV-1 was detected by NGS and confirmed by PCR. Vancomycin was started in two cases following the detection of E. faecium by NGS. In three cases, following pathogen detection by NGS, ongoing empiric therapy was considered appropriate and was continued accordingly. BAL: bronchoalveolar lavage; BC: blood culture; EBV: Epstein–Barr virus; HSV-1: herpes simplex virus type 1; NGS: next-generation sequencing; PCR: polymerase chain reaction.


**Figure 1.** Diagnostic and therapeutic contribution of additional NGS considering BC positivity. BC failed to detect relevant pathogens in 22 of 25 cases. In six of these cases, additional NGS led to further diagnostic tests or contributed to the optimization of antimicrobial therapy. BC: blood culture; BSI: bloodstream infection; NGS: next-generation sequencing.

## **4. Discussion**

This retrospective study investigated the diagnostic performance and potential impact on antimicrobial therapy of additional NGS pathogen diagnostics in COVID-19 ICU patients. NGS was able to detect three times as many bacterial or fungal pathogens in blood samples compared to BC alone (*p* = 0.05). Moreover, NGS detected a viral reactivation in 28% of patients, which would have remained undiscovered by standard diagnostic alone. Based on these results, NGS directly led to the initiation of targeted antimicrobial therapy in 12% of all cases and contributed to the continuation of appropriate therapy in 12% of cases, considering the overall clinical context. NGS led to additional diagnostic procedures, which potentially altered therapy.

In this study, bacterial BSIs were diagnosed by BC analysis in only 12% of cases despite corresponding clinical symptoms. This may be due to the limitations faced by classical culture-based methods: (1) low sensitivity, especially for slow-growing and fastidious organisms, (2) interference due to prior antibiotic exposure, (3) contamination often led to false positive results, and (4) long turnaround time, as standard culture methods typically require 12–36 h for positive signaling and up to 72 h for accurate pathogen identification, including antimicrobial susceptibility testing [13–15].

NGS analysis detected ten additional bacteria as well as one fungus compared to BC in this study cohort. Furthermore, six bacteria were detected in the bloodstream, which otherwise would have been regarded as localized infections (i.e., isolates were only detected in routine non-blood samples). Previous studies reported a 1.5–5.2-fold increase in sensitivity by NGS in patients with suspected sepsis compared to BC [13,17,18]. Our results are in line with these previous studies and demonstrate a higher positivity rate of NGS in the detection of bacteria or fungi compared to BC analysis, in COVID-19 patients (36% vs. 12%, *p* = 0.05).
