Abstract
Introduction
The early diagnosis of infections in acute-on-chronic liver failure (ACLF) is still difficult. mNGS(metagenomic next-generation sequencing) is a no-bias, sensitive pathogen diagnosis method, and further research on mNGS in ACLF is needed.
Methods
A total of 275 ACLF patients with suspected or confirmed infections were recruited and divided into the mNGS group and the non-mNGS group. Differences between the two groups were assessed.
Results
The 1:1 Propensity score matching (PSM) for balancing the baseline variables produced 86 patients in each group. From these 86 patients in the mNGS group, 134 samples were collected and analyzed. The overall microbiological positive rate (103/134, 76.9%) detected by mNGS was higher than that detected by culture (24/134, 17.9%), particularly for fungi (14.9% vs. 2.2%). The etiological diagnosis rates for pulmonary and thoracoabdominal infections detected by the mNGS method were higher than those of the culture method (47.9% vs. 11.4%; 52.0% vs. 18.4%, respectively). The etiological diagnosis can be confirmed 22.83 ± 26.27 h ahead of time. mNGS testing did not significantly improve 90-day transplant-free survival in the overall cohort (sHR 0.96, 95% CI 0.72–1.27; P = 0.76). In the subgroup where mNGS guided therapy, numerically higher resolution rates were observed for pulmonary (53.8% vs 37.1%), abdominal (63.2% vs 52.6%), and bloodstream infections (66.7% vs 50.0%), though these differences were not statistically significant.
Conclusions
mNGS is a valuable diagnostic tool for ACLF with infections, especially for viruses and fungi. mNGS allows for precise and earlier pathogen diagnosis, enabling timely and targeted anti-infective therapy. mNGS may be associated with improved clinical outcomes in ACLF patients with co-infections, though this potential association requires further validation.
Trial registration
The study was registered on Clinicaltrials.gov (registration number: NCT05740696, release date: February22,2023). Accessible at: https://classic.clinicaltrials.gov/ct2/show/NCT05740696
Supplementary Information
The online version contains supplementary material available at 10.1186/s12876-025-04601-8.
Keywords: Acute, On, Chronic liver failure; Infection; Metagenomic next, Generation sequencing
Introduction
Acute-on-chronic liver failure (ACLF) is an acute hepatic insult induced by various causes that manifests as jaundice and coagulopathy and is characterized by a high 28-day mortality. According to the Asian Pacific Association for the Study of the Liver (APASL) criteria, ACLF is defined as acute hepatic insult in the setting of pre-existing chronic liver disease, irrespective of the presence of cirrhosis or prior decompensation [1]. Immune dysfunction secondary to systemic inflammation is an important cause of susceptibility to infection in ACLF patients [2]. A recent study indicated that the infection prevalence rate is in the range of 38–75% [3]. Bacterial infections (BIs) are both a precipitant and a frequent complication of ACLF [4]. Bacteria are also the most common pathogens of infection in ACLF patients. Apart from bacteria, fungal and nonhepatotropic viral infections such as human cytomegalovirus (CMV) and Epstein-Barr virus (EBV) are not rare in patients with ACLF [5–9]. The common infection types include pneumonia, spontaneous bacterial peritonitis (SBP), bloodstream infection, biliary infection, and urinary tract infection [3, 10].
Infections are often related to a poor prognosis in ACLF [4, 11]. Early diagnosis and timely administration of appropriate antibiotics are crucial. However, the timely and accurate diagnosis of infection is frequently difficult. The causes are multifactorial. First, the symptoms of infection in ACLF patients are often insidious, which makes early identification difficult. In addition, precise etiological diagnosis of infection is also a challenge. Some pathogens, such as viruses and atypical pathogens, for instance Pneumocystis jirovecii, cannot be detected by culture. The low positive rate and long turnaround time of conventional bacterial and fungal culture methods also make early diagnosis more difficult. Therefore, empirical antimicrobial therapy is the major mode of treatment for infections in clinical practice. More precise and faster diagnostic tools are urgently needed.
As a hypothesis-independent, rapid, and sensitive technology, metagenomic next-generation sequencing (mNGS) has been broadly applied in the diagnosis of infectious diseases in recent years [12, 13]. It also has potential clinical value to impact treatment decision-making [14, 15]. Whether mNGS is beneficial for ACLF patients with infections remains to be further explored. This study was designed to systematically evaluate the clinical utility of mNGS in ACLF patients with suspected infections, focusing specifically on its effects on diagnostic yield and therapeutic decision-making, while also examining potential associations with clinical outcomes. This work seeks to provide additional evidence for understanding mNGS implementation in the ACLF population.
Methods
Patient population
This retrospective study analyzed data from the “REgistry Study for Optimal Management of LiVer FailurE in the Chinese Population (RESOLVE-C)” since 2011 (NCT05740696), a longitudinal cohort initiated in 2011(supplementary material). For this analysis, we included ACLF patients with suspected or confirmed infections who were admitted to the Center of Liver Diseases at the First Affiliated Hospital of Xi'an Jiaotong University between January 2019 and July 2023. It is important to clarify that while the registry prospectively collects data, the design and analysis for the present study are strictly retrospective.
The inclusion criteria were as follows: (1) ACLF was diagnosed according to the diagnostic criteria recommended by the Asian Pacific Association for the Study of the Liver (APASL) [1]; and (2) suspected or confirmed infections during hospitalization or at admission according to the infection diagnostic criteria. Patients who had any of the following conditions were excluded: (1) solid organ or hematologic malignancies, such as hepatocellular carcinoma and leukemia; (2) a history of liver transplantation; (3) ongoing use of immunosuppressive medications or coinfection with HIV; (4) patients who died, were discharged, or received liver transplantation within 48 h of admission; (5) patients with incomplete clinical data; and (6) pregnancy.
Patients were divided into the mNGS and non-mNGS groups according to whether mNGS was performed during hospitalization.
Definitions related to infection
The presence of infections was considered both on admission and during hospitalization. The diagnosis of infection was made by two independent investigators from the departments of infectious diseases after reviewing the patients’ complete medical records. The diagnostic criteria for infections were as follows: (1) Pulmonary infection: New radiological pulmonary infiltration with the presence of dyspnea, cough, purulent sputum, pleuritic chest pain, or signs of consolidation; positive findings on auscultation (rales or crepitation) or at least one sign of infection: Core body temperature > 38 °C or < 36 °C, or leukocyte count > 10,000/mm3 or < 4000/mm3 in the absence of antibiotics; (2) Thoraco-abdominal infections: Defined by a polymorphonuclear cell count ≥ 250/mm3 in ascitic fluid or ≥ 1000/mm3 in pleural fluid; (3) Bloodstream infection: Positive blood culture with the exclusion of contaminating pathogens; (4) Catheter related infection: Positive blood and catheter cultures; (5) Biliary infection: 1) Acute cholecystitis: fever with abdominal pain, WBC ≥ 10 × 109/L and gallbladder inflammation or abscess confirmed by ultrasonography or CT; 2)Acute cholangitis: fever, abdominal pain with acute elevation of serum bilirubin, WBC ≥ 10 × 109/L and cholangitis or abscess confirmed by ultrasonography or CT; (6) Urinary tract infection (UTI): Presence of urinary symptoms (frequency, urgency, dysuria), with or without lower abdominal tenderness, costovertebral angle tenderness, or fever, plus at least one of the following: Pyuria (≥ 5 WBC/HPF in men or ≥ 10 WBC/HPF in women), with catheterized patients assessed in combination with urine culture; Clinical diagnosis of UTI or documented response to antimicrobial therapy;(7) Soft tissue/skin infection: Fever accompanied by localized clinical signs of cellulitis (e.g., erythema, warmth, swelling, pain); (8) Suspected infection: fever require ing antibiotics and fulfillment of the following conditions: 1) WBC ≥ 10 × 109/L; 2) CRP ≥ 20 mg/dl and/or PCT ≥ 0.5 ng/ml.
The criteria for infection resolution were as follows. Infections were considered resolved when all clinical signs of infection disappeared and with the presence of (1) pulmonary infection: Resolution of clinical signs and symptoms, along with radiographic improvement and negative control cultures (if a pathogen was identified at diagnosis); (2) thoraco-abdominal infections: polymorphonuclear cell count in ascitic < 250/mm3/pleural fluid < 1000/mm3; (3) bloodstream or catheter infection: negative control cultures after antibiotic treatment; (4) biliary infection: improvement of cholestasis, resolution of clinical symptoms and negative control cultures if positive at diagnosis; (5) urinary infections: normal urine sediment and negative urinary culture; (6) soft tissue/skin infections: normal skin examination and negative culture of skin secretions. Other infections were based on conventional clinical criteria [16].
Adjudication of mNGS findings and clinical impact
The clinical interpretation of mNGS results was conducted through a standardized, three-tiered adjudication framework.
First, each detected microorganism was assigned a grade reflecting its clinical relevance as follows: (1) Definite: consistent with concurrent culture or PCR; (2)Probable: a likely cause of infection based on clinical context; (3)Possible: a potential, but less common, causative agent; (4)Unlikely: deemed a contaminant or colonizer; or (5)False negative: a clinically confirmed infection with a negative mNGS result.Subsequently, the impact of these graded results was evaluated separately for diagnosis and treatment. The diagnostic impact (e.g., providing a faster result, identifying co-infections) was categorized according to the criteria detailed in Supplementary Table S1. The therapeutic impact (e.g., initiation, escalation, or de-escalation of anti-infective therapy) was categorized according to the criteria in Supplementary Table S2 [17].
All adjudications were performed by two independent clinicians, with any discrepancies resolved through consensus.
Metagenomic next-generation sequencing
Samples (including ascites, pleural effusion, BALF, sputum blood, catheter, bone marrow, and bile) were collected into sealed sterile tubes and transported on dry ice immediately to Hugobiotech Co., Ltd (Beijing, China). The DNA was extracted and purified from 200 µL of sample (e.g., plasma, ascites, etc.) according to the manufacturer’s instructions for the QlAamp DNA Micro Kit (50) #56,304. The DNA concentration and quality were checked through Qubit and agarose gel electrophoresis. The DNA was used for library construction (QIAseq™ Ultralow Input Library Kit) and high-throughput sequencing on an Illumina NextSeq platform. Short or low-quality reads were removed from the raw data. To obtain high-quality data, human reads were removed by mapping reads to the human reference genome using SNAP software. The remaining data were aligned to the microbial Genome Database (ftp://ftp.ncbi.nlm.nih.gov/genomes/) using Burrows‒Wheeler Alignment to obtain the final microbial composition of the samples. The database collected microbial genomes from NCBI. A positive mNGS result was given when its coverage ranked in the top 10 of the same kind of microbes and was absent in the negative control [“Notemplate” control (NTC)] or when its ratio of reads per million between the sample and NTC (RPMsample/RPMNTC) > 10 if RPMNTC ≠ 0. In parallel with the samples, negative and positive controls were also set up for mNGS detection using the same procedure and bioinformatics analysis. For viruses, M. tuberculosis, and Cryptococcus, a positive mNGS result was considered when at least 1 unique read was mapped to the species level and absent in the NTC or when RPMNTC ≠ 0 and RPMsample/RPMNTC > 5. The clinical relevance of positive results (i.e., distinguishing true pathogens from colonization or contamination) was determined by integrating the findings with the patient’s concurrent clinical manifestations and laboratory test results.
Conventional culture assay
The culture methods were operated according to routine microbial culture processes, such as colony morphology and conventional biochemical reactions. All procedures were completed by the Clinical Laboratory of the First Affiliated Hospital of Xi’an Jiaotong University. The clinical relevance of positive results (i.e., distinguishing true pathogens from colonization or contamination) was determined by integrating the findings with the patient’s concurrent clinical manifestations and laboratory test results.
Statistical analysis
Analyses were performed with IBM SPSS 27.0 (Chicago, USA) and R software. Propensity score matching (PSM) was employed to mitigate the impact of selection bias and balance confounding variables that could be inferred from the baseline characteristics. A greedy algorithm and nearest neighbor method (caliper was 0.2) were used to match patients in a random order of 1:1 on the PS logarithm. Covariate balance after propensity score matching was assessed using standardized mean differences (SMD) and variance ratios (VR), with absolute SMD < 0.1 and a VR close to 1.0 indicating adequate balance. The MatchIt package in R was utilized for this purpose.
Quantitative data are expressed as the mean ± standard deviation (SD), interquartile range (IQR) or median (range), and categorical data are expressed as frequencies and percentages. The chi-square test was used for categorical variables. Student’s t test, paired t test and the Mann‒Whitney or Kruskal‒Wallis test were used for the comparison of quantitative data. Actuarial probabilities of death or liver transplantation during follow-up were calculated by the Kaplan‒Meier method and compared by the log-rank test. Competing risk regression (Fine-Gray model) was used to assess the association between mNGS and transplant-free survival, treating liver transplantation as a competing risk.
The results are two-tailed. A P value of < 0.05 was considered statistically significant. GraphPad Prism 9.0 (La Jolla, CA) and EXCEL 2022 were utilized to generate graphical representations.
Results
Sample and patient characteristics
Between January 2019 and July 2023, 303 ACLF patients with suspected or confirmed infections were enrolled from the Center of Liver Diseases, the First Affiliated Hospital of Xi’an Jiaotong University. All enrolled patients either underwent etiological testing for suspected infections or had a clinical indication requiring empirical antibiotic therapy. A total of 28 patients were excluded for the following reasons: liver transplantation (n = 5), hepatocellular carcinoma (n = 5), lymphomas (n = 2), leukemia (n = 2), breast cancer (n = 1), hospitalization for less than 48 h (n = 11), or incomplete clinical data (n = 2). Eventually, 275 patients were included in the study (Figure S1). Hemoglobin, leukocyte count, neutrophil percentage, ALT, and AST at baseline were different between the two groups, therefore, 1:1 PSM was performed to balance the baseline characteristics. The primary baseline demographics and disease characteristics of the PSM and raw cohort are summarized in Table 1. After PSM, all measured baseline characteristics demonstrated excellent balance, with SMD below 0.1 and VR approaching 1.0 between the mNGS and control groups (detailed in Supplementary Table S2).
Table 1.
Baseline characteristics of the study population
| PSM(n = 172) | RAW(n = 275) | |||||
|---|---|---|---|---|---|---|
| mNGS Group (n = 86) |
non-mNGS (n = 86) |
P value | mNGS Group (n = 86) |
non-mNGS (n = 189) |
P value | |
| Age (years) | 48.5 ± 13.2 | 48.7 ± 12.9 | 0.680 | 49(40.5–58.25) | 45(37–55) | 0.179 |
| Male, n(%) | 62(72.1%) | 54(62.8%) | 0.193 | 62(72.1) | 139(73.5) | 0.801 |
| Etiology | 0.995 | 0.08 | ||||
| HBV | 42(48.8) | 41(47.7) | 41(47.7) | 127(67.2) | ||
| Alcohol | 10(11.6) | 11(12.8) | 11(12.8) | 11(5.8) | ||
| Autoimmune liver disease | 10(11.6) | 10(11.6) | 10(11.6) | 11(5.8) | ||
| Drug | 7(8.1) | 6(7.0) | 6(7.0) | 9(4.8) | ||
| others | 18(20.9) | 17(19.8%) | 18(20.9) | 31(16.4) | ||
| Complications | ||||||
| Ascites, n (%) | 28 (32.6) | 25(29.1) | 0.620 | 28(32.6) | 53(28.0) | 0.446 |
| Gastrointestinal bleeding, n (%) | 5(5.8%) | 5(5.8%) | 1.000 | 5(5.8) | 6(3.2) | 0.482 |
| Hepatic encephalopathy, n (%) | 12(14.0%) | 9(10.5%) | 0.485 | 12(14.0) | 12(6.3) | 0.038 |
| Laboratory data | ||||||
| Hemoglobin (g/L) | 108.2 ± 25.1 | 108.15 ± 25.9 | 0.777 | 106.0(92.8–127.0) | 122.0(102.5–137.5) | 0.001 |
| Platelet (× 10 ^9/L) | 76.5(51.0–144.8) | 76.5(54.0–106.8) | 0.900 | 76.5(51.0–144.8) | 86.0(56.0–127.5) | 0.604 |
| Leukocytes count (× 10^9/L) | 7.3(5.4–10.4) | 6.8(4.8–10.1) | 0.258 | 7.3(5.4–15.2) | 6.3(4.5 + 12.2) | 0.005 |
| Lymphocyte count(× 10^9/L) | 0.88(0.53–1.38) | 0.98(0.65–1.41) | 0.163 | 0.88(0.52–1.39) | 0.96(0.64–1.40) | 0.179 |
| Neutrophil percentage(%) | 77.3(67.7–84.2) | 74.1(66.1–82.9) | 0.111 | 77.3(67.7–84.2) | 74.1(64.9–80.9) | 0.016 |
| C-reactive protein (mg/L) | 11.8(8.5–21.5) | 11.5(5.4–21.2) | 0.167 | 11.8(8.5–21.6) | 10.4(5.5–19.2) | 0.19 |
| PCT (ug/mL) | 0.59(0.32–1.15) | 0.55(0.30–1.02) | 0.484 | 0.59(0.32–1.15) | 0.53(0.32–0.95) | 0.424 |
| ALT(U/L) | 90(38.5–235.5) | 77.5(32–401) | 0.570 | 90(38.3–241.8) | 167(56–551.50) | 0.028 |
| AST(U/L) | 102.5(54.5–194.0) | 98.5(55.0–328.0) | 0.904 | 102.5(54.3–196.0) | 148(63.0–419.0) | 0.02 |
| Total Bilirubin (μmol/L) | 295.4(182.6–448.4) | 321.9(191.9–428.2) | 0.660 | 295.4(181.6–455.1) | 325(206.4–428.4) | 0.743 |
| ALB (g/L) | 28.9(26.6–32.8) | 29.8(26.1–35.1) | 0.476 | 28.9(26.5–32.8) | 30.4(27.4–35.0) | 0.067 |
| PTA(%) | 42.0(31.0–50.6) | 37.7(28.0–47.0) | 0.094 | 42.0(31.0–50.8) | 38.0(2.1–48.2) | 0.111 |
| INR | 1.83(1.5–2.4) | 2.01(1.65–2.70) | 0.059 | 1.8(1.5–3.1) | 2.0(1.6–2.6) | 0.101 |
| Serum creatinine (μmol/L) | 65.5(48.0–1.03) | 56.0(44.0–81.0) | 0.164 | 65.5(48.0–104.5) | 58.0(46.0–75.0) | 0.144 |
| MELD score | 27(22.25–31.75) | 27(22–30) | 0.626 | 27(22.25–31.75) | 26(23–30) | 0.223 |
| ALSS therapy | 48(55.8%) | 51(59.3%) | 0.643 | 51(59.3) | 108(57.1) | 0.737 |
Data are presented as mean ± standard deviation, median (interquartile range), or number (%). Baseline characteristics were balanced after 1:1 propensity score matching (PSM). The RAW column shows data for all patients before matching, and the PSM column shows data after matching. Categorical variables were compared using the Chi-square test; continuous variables were compared using the Student's t-test or Mann–Whitney U test. ALSS, artificial liver support system; MELD, model for end-stage liver disease; PTA, prothrombin activity; INR, international normalized ratio
In the mNGS group, 134 samples were collected from 86 patients and subjected to both mNGS and culture simultaneously. These samples consisted of ascites (n = 57), pleural effusion (n = 4), BALF (n = 20), sputum (n = 1), blood (n = 42), catheter (n = 1), bone marrow (n = 2), and bile (n = 7). A total of 293 samples were collected for culture alone (without mNGS). For the non-mNGS group, a total of 230 samples were collected from 86 patients. The sample characteristics of the cultures in the two groups are illustrated in Figure S2 .
Table 2.
The clinical impact of mNGS on diagnosis and treatment
| Category | Clinical Impact | Grade | Description | Case number, n(%) |
|---|---|---|---|---|
| Etiological Diagnosis | Positive | D1 | mNGS result was quicker than culture | 4(3.0%) |
| D2 | Co-infection was diagnosed according to mNGS. | 13(9.7%) | ||
| D3 | The detection time window was longer than that for culture | 0(0.0%) | ||
| D4 | Plasma mNGS results contributed to pathogen identification | 42(31.3%) | ||
| No effect | D5 | Negative result of mNGS | 31(23.1%) | |
| D6 | The pathogen detected by mNGS was the same as culture, but not detected earlier than culture. | 5(3.7%) | ||
| D7 | The microbes detected by mNGS were assessed as unlikely pathogens. | 33(24.6%) | ||
| Negative | D8 | BSI pathogen was undetected by mNGS and without suspected pathogen detection | 6(4.5%) | |
| Treatment | Positive | M1 | Initialization of the appropriate antibiotics treatment | 28(20.9%) |
| M2 | Antibiotic escalation | 2(1.5%) | ||
| M3 | Antibiotic de-escalation | 1(0.7%) | ||
| M4 | Confirmed empirical treatment | 12(9.0%) | ||
| No effect | M5 | No adjustment in treatment while the result was positive | 60(44.8%) | |
| M6 | The patient was discharged or dead | 0(0.0%) | ||
| M7 | No adjustment in treatment while the result was negative | 31(23.1%) | ||
| Negative | M8 | mNGS led to unnecessary treatment | 0(0.0%) |
Adjudication results are based on 134 samples. The grading criteria for diagnostic impact (D1-D8) and therapeutic impact (M1-M8) are detailed in Supplementary Tables S1 , y.BSI, bloodstream infection
Detection performance of mNGS
Comparison of the positive rate in ACLF between the mNGS and non-mNGS groups
A total of 134 samples were collected for mNGS and culture simultaneously, and the samples were categorized as thoracoabdominal fluid (n = 61), BALF/sputum (n = 21), blood/catheter/bone marrow (n = 45), and bile (n = 7). The overall positivity rate of mNGS (103/134 76.9%) was significantly higher than that of culture (24/134 17.9%). The positive rates of mNGS in the four types of samples were 80.3%, 57.8%, 100%, and 100%, respectively, higher than those of culture (13.1%, 4.4%, 42.9%, 71.4%). In addition, the positive rates of mNGS for viruses, bacteria, and fungi were 64.9%, 32.8%, and 14.9%, respectively, higher than those of culture (0.0%, 15.7%, 2.2%) (Fig. 1A). The positivity rates of mNGS and culture tests classified by sample and type of pathogen in detail presented the same results (Fig. 1B).
Fig. 1.
mNGS significantly improves the microbiological positive rate. A The overall positivity rate of mNGS (103/134 76.9%) was significantly higher than that of culture (24/134,17.9%). The positive rates of mNGS for virus, bacteria, and fungi were 64.9%, 32.8%, and 14.9%, respectively, significantly higher than those of culture (0.0%, 15.7%, 2.2%). B Sample types were classified as thoraco-abdominal fluid blood/catheter/bone marrow, BALF/sputum, bile. The positive rate of mNGS were 80.3%, 57.8%, 100%, 100%, respectively, higher than culture (13.1%, 4.4%, 42.9%, 71.4%). C Concordance between mNGS and culture results. D 2 × 2 contingency tables comparing the performance of mNGS relative to culture for 134 samples and four types of samples. E Pathogens detected by mNGS and/or culture. The most detected pathogen was Human betaherpesvirus 5 (n = 51). The most detected Gram-negative bacteria was Klebsiella pneumoniae (n = 14). Enterococcus faecium (n = 6) was the most detected Gram-positive bacteria. Aspergillus(n = 8) was the most frequent fungi, including. Corresponding culture only detected partial pathogens. ****: P < 0.0001; ***: P < 0.001; **: P < 0.01; ns: P > 0.05
Concordance between mNGS and culture for pathogen detection
We analyzed the consistency of pathogens identified by mNGS and culture. Overall, the results of mNGS and culture were both positive in 23 (23/134, 17.2%) patients and negative in 30 (30/134, 22.4%) patients. A total of 80 (80/134, 59.7%) patients were positive by mNGS only, but 1 (1/134, 0.7%) patient was positive by culture only. Additionally, for 23 double-positive patients, the results between mNGS and culture were completely consistent in 2 (2/134, 1.5%), partially consistent in 14 (14/134, 10.4%), and completely inconsistent in 7 (7/134, 5.2%) (Fig. 1C). The 2 × 2 contingency tables showed the consistency of the pathogens in the different samples between mNGS and culture (Fig. 1D).
Pathogens detected by mNGS and culture in ACLF
A total of 243 strains of pathogens were identified in 134 patients by mNGS. Viruses and bacteria were the most common pathogens, with 133 strains (133/243 54.7%) and 84 strains (84/243 34.6%). Of the 84 detected bacteria, 28 (33.3%) were gram-positive bacteria, and 56 (66.7%) were gram-negative bacteria. The most detected pathogen was human betaherpesvirus 5 (n = 51). The most commonly detected Gram-negative bacteria were Klebsiella pneumoniae (n = 14). Enterococcus faecium (n = 6) was the most detected gram-positive bacteria. In total, 26 strains (26/243 10.7%) of fungi were detected. Aspergillus (n = 8) was the most detected fungus, including Aspergillus flavus (n = 4) and Aspergillus fumigatus (n = 2) (Fig. 1E).
Application of mNGS in the diagnosis and anti-infection therapy of ACLF
Graded evaluation of mNGS for the diagnosis and treatment of infections in ACLF
The positive impacts of mNGS on diagnosis and anti-infection therapy accounted for 44.0% (59/134) and 32.1% (43/134), respectively. The grading evaluation of diagnosis and treatment is shown in detail in Table 2. Among the positive impacts of the mNGS results, BALF/Sputum, detection of fungi and multipathogens accounted for the highest percentage. The details of the diagnosis and treatment grade are shown in Fig. 2A-D.
Fig. 2.
Graded evaluation of mNGS for the diagnosis and treatment of infections. A The positive rate of impact on the diagnosis of BALF/sputum, thoraco-abdominal fluid blood/catheter/bone marrow, and bile were 76.2%,39.3%,28.9%,71.4%. B The positive rate of impact on the treatment of BALF/sputum, thoraco-abdominal fluid blood/catheter/bone marrow, and bile were 61.9%,32.8%,13.3%,57.1%. C The corresponding positive rates of impact on the diagnosis for detection of viruses, bacteria, fungi, and multi-pathogens were 50.0%, 30.0%, 80.0%, and 89.5%. D The corresponding positive rates of impact on the treatment for detection of viruses, bacteria, fungi, and multi-pathogens were 40.0%, 16.0%, 80.0%, and 71.1%
Impacts of mNGS on diagnosis
In the mNGS group, 48 pulmonary infections, 50 thoracoabdominal infections, 19 bloodstream infections, and 9 biliary tract infections were diagnosed. Four types of infections were more frequently diagnosed in patients in the mNGS group than in the non-mNGS group (55.8% vs. 40.7%; 58.1% vs. 44.2%; 22.1% vs. 7.0%; 10.5% vs. 9.3%) (Fig. 3A). The incidence of urinary tract infection and skin and soft tissue infection in the mNGS groups was 9.3% (8/86) and 1.2% (1/86), respectively, lower than that in the non-mNGS group (16.3%, 14/86; 3.5%, 3/86). mNGS significantly improved the etiological diagnosis rate of pulmonary infections (47.9% vs. 11.4%, P < 0.001) and thoracoabdominal infections (52.0% vs. 18.4%, P < 0.01) (Fig. 3B). A comparison of the 16 concordant mNGS and culture results for bacterial/fungal pathogens showed that mNGS shortened the time to pathogen identification by 22.83 ± 26.27 h. (Fig. 3C). Compared to the non-mNGS group, pulmonary infections, thoraco-abdominal infections, bloodstream or catheter infections caused by bacterial/fungal pathogens were diagnosed 33.11 ± 11.75 h, 57.22 ± 8.751 h, and 64.21 ± 10.03 h earlier in the mNGS, respectively (Fig. 3D).
Fig. 3.
Impacts of mNGS on diagnosis and treatment. A The infections were more frequently diagnosed in patients of mNGS group, including pulmonary infection (55.8% vs 40.7%), thoraco-abdominal infection (58.1% vs 44.2%), bloodstream or catheter infection (22.1% vs 7.0%), biliary tract infection (10.5% vs 9.3%). B The etiological diagnosis rates of mNGS group were 47.9%, 52.0%,79.0% and 44.4%, respectively. The corresponding rate of non-mNGS group were 11.4%, 18.4%, 83.3%, and 12.5%, respectively. C Comparing turnaround time with consistent results between mNGS and culture, the etiological diagnosis can be confirmed 22.83 ± 26.27 h ahead of time (n = 16). D Comparison of turnaround time of mNGS-diagnosed infections with culture-diagnosed infections in different types of infections. The etiological diagnosis of pulmonary infections can be made earlier 33.11 ± 11.75 h. Those of thoraco-abdominal infection and bloodstream or catheter infection can be made earlier 57.22 ± 8.751 h and 64.21 ± 10.03 h, respectively. E The distribution of adjusting anti-infection drugs according to mNGS results. The 43 cases of mNGS results had positive impact on anti-infection treatment. Of those, 72.1% (31/43) resulted in a modification of treatment, including 23.3% (10/31) antibiotic treatment, 23.3% (10/31) antiviral treatment, 18.6% (8/31) anti-fungi treatment, and 7.0% (3/31) multiple treatments
Impacts of mNGS on anti-infection therapy
The mNGS results of 23.1% (31/134) of 27 patients led to the modification of the anti-infection treatment. The original therapies were maintained because the prior empirical medications were appropriate for the pathogens detected in 12 patients. Appropriate anti-infective drugs were applied in 28 patients. The major types of anti-infective drug adjustments included antibiotic treatment, antifungal treatment, and antiviral treatment (Fig. 3E). Twelve patients received ganciclovir antiviral treatments for mNGS-detected human betaherpesvirus 5 (n = 11) and human gammaherpesvirus 4 (n = 1). Five patients received cotrimoxazole for mNGS-detected Pneumocystis jirovecii. Four patients received G + antibiotics for mNGS-detected Enterococcus faecium. The utilization of antimicrobial agents in both groups is summarized in Table S3.
Impact of mNGS on clinical outcome
The overall resolution rates were comparable between the mNGS group (50%, 43/86) and the control group (64%, 55/86), with no statistically significant difference observed (P = 0.065). We assessed the impact of mNGS testing on 90-day mortality using a competing risks regression model (Fine-Gray), treating liver transplantation as a competing event. The 90-day cumulative incidence of mortality was 52.9% in the mNGS group and 54.7% in the no-mNGS group. The competing risk analysis confirmed that this absolute difference of 1.8% was not statistically significant, with no significant association observed between mNGS testing and the risk of mortality (sHR 0.96, 95% CI 0.72–1.27; P = 0.76). As shown in Fig. 4A, the cumulative incidence of death and liver transplantation did not differ significantly between the mNGS and non-mNGS groups in competing risks analysis.
Fig. 4.
Impacts of mNGS on clinical outcome of ACLF with infections: infection resolution and 90-day transplant free survival rate. A Comparison of 90-day transplant-free survival between the mNGS and non-mNGS groups in the overall cohort (analyzed using competing risk regression). B The resolution rates of pulmonary infection, thoraco-abdominal infections, bloodstream or catheter infections, biliary tract infections were 53.8% (7/13), 63.2% (12/19), 66.7% (4/6), 33.3% (1/3) in mNGS group with positive treatment impact, respectively. The corresponding resolution rate in non-mNGS group were 37.1% (13/35),52.6% (20/38),50% (3/6),50% (4/8), respectively(P > 0.05). C The 90-day transplant free survival rate of pulmonary infection, thoraco-abdominal infections, bloodstream or catheter infections, biliary tract infections were 61.5% (7/13),57.9% (11/19), 66.7% (4/6), 33.3% (1/3) in mNGS group with positive treatment impact, respectively. The corresponding survival rate in non-mNGS group were 34.3% (12/35), 42.1% (16/38), 33.3% (2/6),50% (4/8), respectively(P > 0.05)
Given the prevalence of multi-site infections, we next analyzed the subgroup in whom mNGS directly guided therapeutic changes. The mNGS results of 13 pulmonary infection patients had a positive treatment impact, 7 of which (53.8%) achieved resolution. The resolution rates of the thoraco-abdominal infections, bloodstream or catheter infections, and biliary tract infections were 63.2% (12/19), 66.7% (4/6), and 33.3% (1/3) in the mNGS group with a positive treatment impact, respectively. No statistically significant differences in resolution rates were observed for pulmonary (53.8% vs. 37.1%; P = 0.34), thoracoabdominal (63.2% vs. 52.6%; P = 0.57), or bloodstream infections (66.7% vs. 50.0%; P = 0.99) when comparing the mNGS-treatment-positive subgroup to the non-mNGS group, despite numerical trends favoring the former (Fig. 4B). In the mNGS-treatment-positive subgroup, the 90-day transplant-free survival rates for pulmonary (61.5% vs. 34.3%; P = 0.11), thoracoabdominal (57.9% vs. 42.1%; P = 0.28), and bloodstream infections (66.7% vs. 33.3%; P = 0.57) were numerically higher than those in the no-mNGS group, although these differences did not reach statistical significance (Fig. 4C).
Discussion
Many investigations have indicated that infections are common complications closely related to poor prognosis in patients with ACLF [4, 7, 10], yet achieving early and precise diagnosis remains challenging. While predictive models for bacterial infection have been developed [18, 19], metagenomic next-generation sequencing (mNGS) offers a hypothesis-free approach with proven utility in various clinical settings [13, 17, 20–22].
In our study, we evaluated the application of mNGS in ACLF patients with infections. The pathogen spectrum revealed by mNGS was substantially broader than that revealed by conventional culture. This technique has allowed for significant advances in the detection of fungi and viruses and has broadened the detection range of potential pathogens. Using mNGS, Chen et al. [7] elucidated a nonhepatotropic virus (NHV) signature in acutely decompensated cirrhosis that is similar to those observed in sepsis and hematological malignancies. As a special immunosuppressive population, patients with opportunistic infections (e.g., CMV, Aspergillus, Pneumocystis jirovecii) were not uncommon in the ACLF population in our study. Consistent with prior reports, Klebsiella pneumoniae was the most prevalent bacterium, and Gram-negative bacteria were more frequent in ACLF [23]. It is critical to note that the broader pathogen spectrum described here pertains specifically to the technical detection capability of mNGS; the clinical significance and diagnostic impact of these findings are explored next.
For the evaluation of the impact of mNGS results on diagnosis and treatment, we refer to the research criteria of Feng et al. [17]. The detection rates for pulmonary, thoracoabdominal, bloodstream/catheter, and biliary tract infections were higher in the mNGS group than in the non-mNGS group likely because it is often deployed after empirical antibiotic failure in more complex cases. To mitigate this bias, we compared etiological diagnosis rates in clinically confirmed infections, finding that mNGS significantly improved rates for pulmonary (P < 0.001) and thoracoabdominal infections (P < 0.01). Conversely, the lower incidence of urinary tract and skin/soft tissue infections in the mNGS group likely reflects a specimen selection bias, whereby readily diagnosable samples are less often sent for mNGS. The technique also advanced etiological diagnosis by approximately 22 h, a timeframe expected to shorten with wider adoption.
Beyond diagnosis, mNGS also significantly influenced anti-infective management, enabling a shift from empirical to precision treatment. The relatively modest impact of mNGS on the management of viral infections, as compared to bacterial or fungal ones, can be attributed to two primary factors. First, the high sensitivity of DNA-based mNGS frequently detects viruses of uncertain clinical significance (e.g., latent herpesviruses), which necessitates cautious physician interpretation and limits immediate therapeutic changes. Second, the inherent limitation of our DNA-seq approach in detecting common pathogenic RNA viruses (e.g., influenza) precluded the identification of some readily treatable viral pathogens, thereby reducing the overall actionable viral findings.
In our cohort, 11 patients with mNGS-detected CMV received ganciclovir based on clinical assessment despite negative CMV-DNA results, and 8 (72.7%) of them achieved clinical improvement, supporting the value of mNGS-guided intervention in such complex scenarios. Similarly, mNGS may help address the underdiagnosis of infections like PCP, though further studies are needed to clarify prophylaxis indications in high-risk liver failure patients [24].
This study was conducted at a major tertiary hospital in Northwest China with a high standard of empirical antimicrobial therapy. Within this context, prognostic analyses accounting for competing risks demonstrated no significant improvement in clinical outcomes in the overall cohort with mNGS implementation. It has been controversial whether mNGS could improve the prognosis of some infectious diseases [20, 25–27]. However, in the subgroup where mNGS results directly guided therapy, numerical improvements in infection resolution and 90-day survival were observed across pulmonary, thoracoabdominal, and bloodstream infections, though these did not reach statistical significance. Previous studies have also shown that mNGS has good diagnostic performance for pulmonary infections in immunocompromised individuals [13, 28, 29]. The clinical value of mNGS in this complex setting may thus lie in guiding earlier, more precise anti-infective strategies rather than in directly altering overall prognosis. Considering that this may be related to the higher positive rate of bile culture and limited data, those in the patients with bile infections were not improved.
While the substantial cost of mNGS introduces a potential socioeconomic confounder, its advantages in rapid pathogen identification and accurate diagnosis represent clinically meaningful benefits that may contribute to improved infection control and patient outcomes.
However, there are certain limitations in this study. First, its single-center, retrospective design with a limited sample size may harbor residual confounding like patient outcomes despite PSM. Second, RNA sequencing was not included in our mNGS test. Given the well-characterized clinical background of our ACLF cohort, this limitation is unlikely to substantially affect the interpretation of our primary results. However, it is important to note that the detection of hepatotropic viruses may have certain limitations in specific clinical scenarios. Third, while our study focused on comparing mNGS against culture, the influence of other diagnostics like PCR may have been underestimated. However, a validated mNGS assay for respiratory virus detection demonstrated superior overall predictive agreement (97.9%) compared to RT-PCR (95.0%), supporting its competitive diagnostic performance [30].
Conclusion
In summary, mNGS is a valuable diagnostic tool for the ACLF population, especially for viral and fungal infections. More potential pathogens were detected by mNGS than by culture. The etiological diagnosis rate of pulmonary infection and thoracoabdominal infection by mNGS was significantly improved. Meanwhile, mNGS can clarify the etiological diagnosis at least 22 h earlier, allowing for early intervention. The mode of anti-infective treatment has also transformed from empirical treatment to precision treatment. The application of mNGS may be associated with a potential beneficial effect on the clinical outcomes of ACLF patients with coinfection.
Supplementary Information
Acknowledgements
Thanks to the Clinical Research Center of the First Affiliated Hospital of Xi 'an Jiaotong University and the Research Electronic Data Capture database for the support of this study.
Abbreviations
- ACLF
Acute-on-chronic liver failure
- ALSS
Artificial Liver Support System
- APASL
Asian Pacific Association for the Study of the Liver
- mNGS
Metagenomics next-generation sequencing
- NHV
Non-Hepatotropic Virus
- PCP
Pneumocystis jiroveci Pneumonia
- PSM
Propensity Score Matching
Authors’ contributions
Study design, Yushan Liu, Yingli He, Taotao Yan, Yingren Zhao; Data collection, Yushan Liu, Xiaonan Wu, Qijuan Zang, Qiannan Wang, Pan Huang, Yamin Wang, Shuting Zhang, Siyi Liu; Data analysis, Yushan Liu, Qiao Zhang, Juan Li, Chengbin Zhu, Yingli He; Writing − original draft, Yingli He, Yushan Liu. All authors read and approved the final version of the report.
Funding
This work was supported by National Natural Science Foundation of China (81971310); the Project of Infectious Disease Clinical Medical Research Center of Shaanxi Province, China(2020LCZX-01); the Key RaD Program of Shaanxi (2018ZDXM-SF-037, 2020LCZX-01) and Clinical Research Award of the First Affiliated Hospital of Xi'an Jiaotong University, China (No. XJTU1AF2021CRF-006 and 2022-XKCRC-04).
Data availability
The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.
Declarations
Ethics approval and consent to participate
This study was conducted in accordance with the Declaration of Helsinki and the protocol was approved by the Ethics Committee of the First Affiliated Teaching Hospital of Xi’an Jiaotong University. Written informed consent was obtained from the patients and/or the legal guardian of deceased patients for participation.
Consent for publication
Written informed consent was obtained from the patients and/or relatives for publication of this study.
Competing interests
The authors declare that there are no conflicts of interest regarding the publication of this paper.
Footnotes
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Sarin SK, Choudhury A, Sharma MK, Maiwall R, Al Mahtab M, Rahman S, et al. Acute-on-chronic liver failure: consensus recommendations of the Asian Pacific association for the study of the liver (APASL): an update. Hepatol Int. 2019;13(4):353–90. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Wasmuth HE, Kunz D, Yagmur E, Timmer-Stranghöner A, Vidacek D, Siewert E, et al. Patients with acute on chronic liver failure display “sepsis-like” immune paralysis. J Hepatol. 2005;42(2):195–201. [DOI] [PubMed] [Google Scholar]
- 3.Wong F, Piano S, Singh V, Bartoletti M, Maiwall R, Alessandria C, et al. Clinical features and evolution of bacterial infection-related acute-on-chronic liver failure. J Hepatol. 2021;74(2):330–9. [DOI] [PubMed] [Google Scholar]
- 4.Fernandez J, Acevedo J, Wiest R, Gustot T, Amoros A, Deulofeu C, et al. Bacterial and fungal infections in acute-on-chronic liver failure: prevalence, characteristics and impact on prognosis. Gut. 2018;67(10):1870–80. [DOI] [PubMed] [Google Scholar]
- 5.Gupta E, Ballani N, Kumar M, Sarin SK. Role of non-hepatotropic viruses in acute sporadic viral hepatitis and acute-on-chronic liver failure in adults. Indian J Gastroenterol. 2015;34(6):448–52. [DOI] [PubMed] [Google Scholar]
- 6.Hu J, Zhao H, Lou D, Gao H, Yang M, Zhang X, et al. Human cytomegalovirus and Epstein-Barr virus infections, risk factors, and their influence on the liver function of patients with acute-on-chronic liver failure. BMC Infect Dis. 2018;18(1):577. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Li B, Hong C, Fan Z, Cai S, He Q, Lan X, Lai Q, Ji Y, Luo W, Li J et al. Prognostic and therapeutic significance of microbial cell-free DNA in plasma of people with acutely decompensated cirrhosis. J Hepatol. 2023;78(2):322-332. [DOI] [PubMed]
- 8.Chen D, Qian Z, Su H, Meng Z, Lv J, Huang Y, et al. Invasive pulmonary aspergillosis in acute-on-chronic liver failure patients: short-term outcomes and antifungal options. Infect Dis Ther. 2021;10(4):2525–38. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Yang Q, Zhou Z, Yang X, Chen Y, Liu A, Zhang B, et al. Latent cytomegalovirus reactivation in patients with liver failure: a 10-year retrospective case-control study, 2011–2020. Front Cell Infect Microbiol. 2021;11:642500. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Mucke MM, Rumyantseva T, Mucke VT, Schwarzkopf K, Joshi S, Kempf VAJ, et al. Bacterial infection-triggered acute-on-chronic liver failure is associated with increased mortality. Liver Int. 2018;38(4):645–53. [DOI] [PubMed] [Google Scholar]
- 11.Zhai X-R, Tong J-J, Wang H-M, Xu X, Mu X-Y, Chen J, et al. Infection deteriorating hepatitis B virus related acute-on-chronic liver failure: a retrospective cohort study. BMC Gastroenterol. 2020;20(1):320. 10.1186/s12876-020-01473-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Liu H, Zhang Y, Yang J, Liu Y, Chen J. Application of mNGS in the etiological analysis of lower respiratory tract infections and the prediction of drug resistance. Microbiol Spectr. 2022;10(1):e0250221. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Azar MM, Schlaberg R, Malinis MF, Bermejo S, Schwarz T, Xie H, et al. Added diagnostic utility of clinical metagenomics for the diagnosis of pneumonia in immunocompromised adults. Chest. 2021;159(4):1356–71. [DOI] [PubMed] [Google Scholar]
- 14.Xu J, Zhou P, Liu J, Zhao L, Fu H, Han Q, et al. Utilizing metagenomic next-generation sequencing (mNGS) for rapid pathogen identification and to inform clinical decision-making: results from a large real-world cohort. Infect Dis Ther. 2023 ;12(4):1175-1187 10.1007/s40121-023-00790-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Niles DT, Revell PA, Ruderfer D, Marquez L, McNeil JC, Palazzi DL. Clinical impact of plasma metagenomic next-generation sequencing in a large pediatric cohort. Pediatr Infect Dis J. 2022;41(2):166–71. [DOI] [PubMed] [Google Scholar]
- 16.Fernandez J, Acevedo J, Prado V, Mercado M, Castro M, Pavesi M, et al. Clinical course and short-term mortality of cirrhotic patients with infections other than spontaneous bacterial peritonitis. Liver Int. 2017;37(3):385–95. [DOI] [PubMed] [Google Scholar]
- 17.Xu C, Chen X, Zhu G, Yi H, Chen S, Yu Y, et al. Utility of plasma cell-free DNA next-generation sequencing for diagnosis of infectious diseases in patients with hematological disorders. J Infect. 2023;86(1):14–23. [DOI] [PubMed] [Google Scholar]
- 18.Lin S, Yan YY, Wu YL, Wang MF, Zhu YY, Wang XZ. Development of a novel score for the diagnosis of bacterial infection in patients with acute-on-chronic liver failure. World J Gastroenterol. 2020;26(32):4857–65. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Zhang Z, Ma K, Yang Z, Cheng Q, Hu X, Liu M, et al. Development and validation of a clinical predictive model for bacterial infection in Hepatitis B virus-related acute-on-chronic liver failure. Infect Dis Ther. 2021;10(3):1347–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Zhang P, Chen Y, Li S, Li C, Zhang S, Zheng W, et al. Metagenomic next-generation sequencing for the clinical diagnosis and prognosis of acute respiratory distress syndrome caused by severe pneumonia: a retrospective study. PeerJ. 2020;8:e9623. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Yang A, Chen C, Hu Y, Zheng G, Chen P, Xie Z, Fan H, Sun Y, Wu P, Jiang W et al. Application of Metagenomic Next-Generation Sequencing (mNGS) Using Bronchoalveolar Lavage Fluid (BALF) in Diagnosing Pneumonia of Children. Microbiol Spectr. 2022;10(5):e0148822. [DOI] [PMC free article] [PubMed]
- 22.Chen H, Zhang Y, Zheng J, Shi L, He Y, Niu Y, et al. Application of mNGS in the etiological diagnosis of thoracic and abdominal infection in patients with end-stage liver disease. Front Cell Infect Microbiol. 2021;11:741220. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Yang L, Wu T, Li J, Li J. Bacterial infections in acute-on-chronic liver failure. Semin Liver Dis. 2018;38(2):121–33. [DOI] [PubMed] [Google Scholar]
- 24.Franceschini E, Dolci G, Santoro A, Meschiari M, Riccò A, Menozzi M, et al. Pneumocystis jirovecii pneumonia in patients with decompensated cirrhosis: a case series. Int J Infect Dis. 2023;128:254–6. [DOI] [PubMed] [Google Scholar]
- 25.Sun L, Zhang S, Yang Z, Yang F, Wang Z, Li H, et al. Clinical application and influencing factor analysis of metagenomic next-generation sequencing (mNGS) in ICU patients with sepsis. Front Cell Infect Microbiol. 2022;12:905132. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Xie F, Duan Z, Zeng W, Xie S, Xie M, Fu H, et al. Clinical metagenomics assessments improve diagnosis and outcomes in community-acquired pneumonia. BMC Infect Dis. 2021;21(1):352. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Xi Y, Zhou J, Lin Z, Liang W, Yang C, Liu D, et al. Patients with infectious diseases undergoing mechanical ventilation in the intensive care unit have better prognosis after receiving metagenomic next-generation sequencing assay. Int J Infect Dis. 2022;122:959–69. [DOI] [PubMed] [Google Scholar]
- 28.Peng J-M, Du B, Qin H-Y, Wang Q, Shi Y. Metagenomic next-generation sequencing for the diagnosis of suspected pneumonia in immunocompromised patients. J Infect. 2021;82(4):22–7. [DOI] [PubMed] [Google Scholar]
- 29.Chen S, Kang Y, Li D, Li Z. Diagnostic performance of metagenomic next-generation sequencing for the detection of pathogens in bronchoalveolar lavage fluid in patients with pulmonary infections: systematic review and meta-analysis. Int J Infect Dis. 2022;122:867–73. [DOI] [PubMed] [Google Scholar]
- 30.Tan JK, Servellita V, Stryke D, Kelly E, Streithorst J, Sumimoto N, et al. Laboratory validation of a clinical metagenomic next-generation sequencing assay for respiratory virus detection and discovery. Nat Commun. 2024;15(1):9016. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.