Patients undergoing lung transplantation at Sichuan Provincial People’s Hospital from October 2018 to June 2022 were included in this study. The inclusive and exclusive criteria for donor lungs were described in our previous study (18), and listed as follows.

Donor lungs inclusion criteria: (a) Age < 60 years old, smoking history < 20 packs/year. (b) No chest injury. (c) Continuous mechanical ventilation < 1 week. (d) PaO₂ > 300 mmHg (FiO₂ = 100%, PEEP = 5cm H₂O). (e) X-ray or CT shows that the lung field is relatively clear. (f) No abscess secretion was found through bronchoscopy in the lung bronchus.

Donor lungs exclusion criteria: (a) Age > 60 years old, smoking history > 20 packs/year. (b) Chest trauma and lung contusion. (c) Continuous mechanical ventilation > 1 week. (d) PaO₂ < 300 mmHg (FiO₂ = 100%, PEEP = 5cm H₂O). (e) X-ray or CT shows that the lung field is infected. (f) There are purulent secretions at bronchoscopy in the donor’s lower airways. (g) The percentage of white blood cells, neutrophils, C-reactive protein, and procalcitonin increases gradually compared with the situation at the onset of the disease. (h) The donor’s body temperature is higher than normal. (i) Blood culture is positive.

Basic information about the enrolled recipients, including age, sex, primary indications for a lung transplant, types of lung transplantation (bilateral or unilateral), and infection status within one month following the operation, was recorded. Prognostic information on the enrolled patients’ antimicrobial use, mechanical ventilation, and ICU hospitalization was recorded in detail.

In most lung transplant centers in China, timely bronchofiberscopy after surgery is a routine examination aiming to clean the airway secretions through a bronchofiberscope, which helps to avoid obstructing the small airway and reduce pathogens. Therefore, airway secretions were absorbed from the deep airway by bronchofiberscope two hours after the operation and sent for traditional microbial culture and mNGS for pathogen detection immediately. In the following days, within one month, airway secretions or BALFs were collected for microbial culture every few days, depending on the actual conditions in recipients. Microbial culture for the above samples was conducted in our hospital. The yielded pathogen spectrum was analyzed and compared between these two methods. The incidence of infection within one month and the occurrence time in these recipients were determined. The causative agents for infection were compared with the pathogenic microorganisms reported by mNGS in airway secretions collected within two hours to evaluate the role of mNGS in forewarning potential pathogens.

The whole process of mNGS was completed by Genoxor Medical Science and Technology Inc. (Shanghai, China). The airway secretions samples were stored at 4°C and sent for mNGS detection within 24h. These steps included pre-treatment, DNA extraction, library construction, sequencing, bioinformatic analysis, and interpretation of data (19). A 1.5ml microcentrifuge tube containing 0.6ml of sample, enzyme, and 1.0g of glass beads (0.5mm) was attached to a horizontal platform on a vortex mixer and agitated vigorously at 2,800–3,200 rpm for 30 min. Then DNA in airway secretions samples was extracted using the TIANamp Micro DNA Kit (DP316, Tiangen Biotech) according to the manufacturer’s instructions. After DNA concentration and purity detection, the libraries were constructed undergoing DNA fragmentation, end-repair, adapter ligation, and PCR amplification. DNA library concentration was measured by Qubit 2.0. An Agilent 2100 test achieved quality control of the DNA libraries. After being pre-quantified by qRT-PCR, quality-qualified libraries were sequenced on the NextSeq™ 550DX platform in SE-75 sequencing type according to the manufacturer’s instructions.

Bioinformatics analysis of the mNGS data was performed according to the procedure described in a previous study (20). Raw data (raw reads) were subjected to a quality control process for trimming adapter sequences and removing low-quality tails, reads, and connector sequences using Trimmomatic v0.36 (21). The obtained high-quality and adequate data are called clean reads. Reads mapping to the human genome GRCh37 were removed using the calibration software Bowtie v2.2.6 (22), and the remaining were called unmapped reads (microbial reads). All the microbial reads were deposited in the database under the Sequence Read Archive (SRA) accession number PRJNA932550. Unmapped% refers to the proportion of microbial reads in the clean reads. Duplicated reads introduced in the PCR step were deleted using FASTX-Toolkit, Fulcrum, FastUniq, and CD-HIT-DUP tools (23). Subsequently, Kraken v2.0.9-beta (24) was adopted for the taxonomic classification of microbial reads, with a microbial genome database in NCBI constructed using 51543 genomes of about 27000 species (ftp://ftp.ncbi.nlm.nih.gov/genomes/) (25). The number of reads in the Kraken classification report was further estimated by the Bayesian algorithm named Bracken to produce species-level abundance estimates (26). The estimates of the percentage relative abundance of each species were computed using the reads per kilobase of transcript per million mapped reads (RPKM), a normalization method for mNGS reads, and RPKM was calculated using the formula: gene reads/[the total mapped reads (millions) × genome length (KB)] (27).

The mNGS assay was employed for detecting microorganisms, including bacteria, viruses, fungi, and parasites, and a positive result will be judged if it satisfies any of the following criteria described previously (17). 1) The relative abundance of bacteria (excluding M. tuberculosis complex) and fungi was greater than 30% at the genera level; 2) Virus detection was considered when the stringent map read number (SMRN) was ≥3. 3) For M. tuberculosis complex, at least one number of reads should be aligned to the reference genome at the species or the genus level. However, a positive mNGS finding did not invariably indicate the presence of causative pathogens. Microorganisms detected with mNGS were categorized into colonized, putative, and pathogenic microorganisms. It would be the clinician’s responsibility to determine the putative pathogens and pathogenic microorganisms through comprehensive clinical assessments. In the pathogenic spectrum analysis, the proportion of the pathogenic species, the detection frequency, was calculated with a formulation: the number of samples in which a particular species was detected/the total number of samples.

Before and after the lung transplant, the infectious risk and status of the recipients were monitored. The suspicion and diagnosis of infection were based on several clinical symptoms, including body temperature, computed tomography, etiological examination, and immune indicators. In the infected recipients, the putative pathogens and pathogenic microorganisms were judged based on a comprehensive analysis of clinical data, including the number of reads for mNGS, the clinical presentations, radiologic manifestations, conventional detection findings, clinical epidemiology, and the treatment effect of the antibiotic therapy. The putative pathogens or pathogenic microorganisms could be ascertained if the two clinicians approved. Further discussion by senior clinicians is needed in case of a significant disagreement between the first two clinicians. Then, the targeted antibiotic therapy was formulated to fight against infection, and a favorable outcome further confirmed the causative agent. The consistency of mNGS with the causative agents in the infected recipients was evaluated at the species level.

Descriptive statistics were computed for the overall samples and stratified by the positive pathogen detected by mNGS on airway secretions samples. Mean ± standard deviation (SD) or median (interquartile range, IQR) was used for describing the continuous variables. Chi-squared or Fisher’s Exact test was used to compare the two groups’ differences. The significance level was set at 0.05. All statistical analyses were performed using the GraphPad software 8.0.

From October 2018 to June 2022, 40 patients received lung transplant surgery in our hospital, and 38 eligible patients were included for the final analysis. Two recipients were excluded because of death quickly without any microbial culture result. Basic information of these patients was provided in Supplementary Table 1 . Of all 38 lung transplant recipients, the mean age was 58.13 years (ranges 33-70), including 33 (86.84%) males. The most common primary disease was COPD (19, 50%), followed by interstitial lung disease (18, 47.37%), with the addition of one patient with pneumosilicosis. In terms of the lung transplant types, 23 (60.53%) underwent bilateral transplantation and 15 (39.47%) unilateral transplantation. In the 38 recipients, new-onset infection within one month occurred in 26 (68.42%). These clinical characteristics were recorded and demonstrated in Table 1 . After lung transplant, the initial antibiotic regimens frequently include Sulbactam/Cefopcrazone and Piperacillin Sodium/Tazobactam Sodium. Immunosuppressant regiments comprise cyclosporin A, tacrolimus, and methylprednisolone.

38 airway secretions samples from 38 lung transplant recipients were collected within two hours after surgery and simultaneously sent for etiological examination by traditional microbial culture and mNGS. The study design is illustrated in Figure 1 . The detecting results of the two methods in each patient were provided in Supplementary Table 1 . This supplementary material also included detailed information concerning each sample’s sequencing number of reads (raw reads, clean reads, clean reads/raw reads, unmapped reads, and unmapped %), as well as the putative pathogens in each patient and their relative abundance. It demonstrates that the raw reads range from 4M to 57M, with an average of 20M; most ratios of clean reads to raw reads are above 90%. Unmapped% refers to the proportion of microbial reads in the clean reads, ranging from 0.69% to 79.21%.

143 kinds of pathogenic microorganisms were found in 35 (92.11%, 35/38) airway secretions specimens using mNGS, while the detection rate by microbial culture was 26.31% (10/38) (P<0.0001) ( Table 2 ). Statistically, mNGS identified pathogenic microorganisms at the level of species or genus, which were further classified into five types, including bacteria (72.73%), fungi (13.29%), virus (11.89%), mycoplasma (1.4%), and parasites (0.7%) ( Figure 2A ). When analyzed at the species level, S. pneumoniae (28.95%) and H. parainfluenzae (23.68%) were the top two bacteria, followed by S. aureus (21.05%), S. pseudopneumoniae (21.05%), K. pneumoniae (21.05%), and A. baumannii complex (21.05%) ( Figure 2B ). C. albicans (21.05%) was the most dominant fungi detected with mNGS. Human betaherpesvirus 5 (18.42%) was the most prevalent virus. Seven pathogenic microorganisms were detected through the traditional culture method in 10 airway secretions samples. K. pneumoniae was detected in three cases (7.89%); S. aureus was detected in two samples (5.26%) ( Figure 2C ). The other bacteria include A. baumannii and A. ursingii, and fungi like C. parapsilosis were detected in one sample (2.63%).

Figure 3 illustrates the results of etiological identification by mNGS and traditional culture and the information on causative agents in recipients infected within one month. Within one month, 26 (68.42%) of the 38 recipients got infected, and the median time of new-onset infection was 9 days, ranging from 3 to 25 days. Among the 26 infected recipients, 10 (38.46%) got infected within one week following the lung transplant operation, and infection in 7 (26.92%) cases occurred within one to two weeks. The remaining 7 (26.92%) and 2 (7.69%) got infected within two to three weeks and three to four weeks, respectively ( Table 3 ). Consequently, infection onset within one week was the highest, and more than half (65.38%) of recipients developed an infection within two weeks. The drug sensitivity of the pathogens was also examined through microbial culture and demonstrated in Figure 3 . Multiple drug resistance was observed in S. aureus (case 1), A. baumannii (case 8), S. maltophilia (case 11), and B. multivorans (case 15).

The consistency of two hours airway secretions-mNGS and microbial culture results with the causative agents in the infected recipients was determined at the species level and illustrated in Figure 3 . In 9 (34.62%, 9/26) infected recipients (patient 1, 2, 3, 5, 7, 8, 12, 13, 26), their causative agents were detected by mNGS in advance (in the airway secretions collected within two hours), who got an infection at the 3rd, 3rd, 3rd, 5th, 5th, 5th, 8th, 8th, and 25th day, respectively, after lung transplant ( Figure 3 ). Except for an infection caused by A. fumigatus on the 25th day, the median time of infection occurring in the rest 8 recipients was 5 days following the operation. Namely, most of them (6, 66.67%) were infected within one week, 2 (22.22%) cases suffered between one to two weeks, and 1 (11.11%) at three to four weeks ( Table 4 ). A decreased trend was observed in consistency, along with the prolonged infection time.