3.1 Patient Characteristics and mNGS Results
A total of 55 patients were enrolled , including 31 males and 24 females, with an average age of 45 years (11-74 years). Among them, 33 patients (33/55=60.0%) had underlying diseases, including 9 cases of acute lymphoblastic leukemia, 9 cases of acute myeloid leukemia, 5 cases of lymphoma, 3 cases of myelodysplastic syndrome, 3 cases of autoimmune anemia, 2 cases of aplastic anemia, 1 case of chronic myeloid leukemia and 1 case of vasculitis. A total of 36 patients (36/55 = 65.5%) were clinically diagnosed with mixed pulmonary infections, and 19 patients (19/55 = 34.5%) were diagnosed with non-mixed infections (single infections or infections with unknown pathogens) (See Supplementary Table 1, Additional File 1).
Pulmonary biopsy and bronchoalveolar lavage fluid were collected for mNGS. The report provided specific sequencing reads of all microorganisms with valid data that can be detected in specimen. Propionibacterium acnes, Micrococcus luteus, Malassezia globosa, Lactococcus lactis, and Saccharomyces were not interpreted as pathogens, as they were known normal flora of the skin or respiratory tract.
3.2 Comparison of mNGS and Conventional test in the Diagnosis of Mixed Pulmonary Infection
3.2.1 Comparison of Diagnostic Performance for Differentiating Mixed Infection From Non-mixed Infection
In 55 patients with pulmonary infection, the comparison of mNGS and conventional test is presented in Table 2. Mixed pulmonary infection was defined when two or more infectious pathogens were indentified. The sensitivity and specificity of diagnosing mixed pulmonary infection by mNGS were 97.2% (95%CI:83.8-99.9%) and 63.2% (95%CI:38.6-82.8%) respectively,with NPV and PPV being 92.3%(95%CI:62.1-99.6%) and 83.3% (95%CI:68.0-92.5%). The sensitivity and specificity of diagnosing mixed pulmonary infection by conventional diagnostic testing were 13.9% (95%CI:5.2-31.0%) and 94.7% (95%CI:71.9-99.7%) respectively, with NPV and PPV being 36.7% (95%CI:23.8-51.7%) and 83.3% (95%CI:36.5-99.1%).
Table 2 Performance of metagenomic next-generation sequencing (mNGS) and conventional testing in diagnosis of mixed pulmonary infection.
Diagnostic testing
|
Sensitivity % (95% CI)
|
Specifificity % (95% CI)
|
PPV % (95% CI)
|
NPV % (95% CI)
|
Conventional laboratory-based diagnostic testing
|
13.9
(5.2-31.0)
|
94.7
(71.9-99.7)
|
83.3 (36.5-99.1)
|
36.7
(23.8-51.7)
|
mNGS
|
97.2
(83.8-99.9)
|
63.2
(38.6-82.8)
|
83.3 (68.0-92.5)
|
92.3
(62.1-99.6)
|
3.2.2 Concordance Between mNGS and Conventional test
In our results, mNGS and conventional test were both positive for mixed infection diagnose in 5 cases(5/55=9.1%). A total of 37 cases (37/55=67.3%) were positive for mixed infection by mNGS only ,7 of them were false positives. There was 1 case(1/55=1.8%)negative for mixed infection by mNGS only, and this case was false negative. mNGS and conventional diagnostic testing were both negative for diagnosing mixed infection in 12 cases(12/55=21.8%), one of them were false negative (Figure 1 A).
mNGS and conventional test were both positive for pathogens detection (single infection and coinfection) in 26 (26/55=47.3%) cases and were both negative in 4 (4/55=7.3%) cases. A total of 24 cases were positive for pathogens detection by mNGS only (24/55=43.6%) and 1(1/55=1.8%)were positive by conventional diagnostic testing only . Within the 26 double-positive cases, results of mNGS and conventional tests were completely matched in 3 cases and were totally mismatched in 6 cases. Among 3 completely matched cases, only one was mixed infection (NO.31), the others were single infections. The remaining 17 cases were found to be partially matched, where at least 1 detected pathogen was overlapped between mNGS and conventional tests (Figure 1 B).
3.3 Comparison of mNGS and Conventional test in the Pathogens Detection
3.3.1 The Eficiency of mNGS in Negative Cases Identified by Conventional test
Of 28 cases (NO.1-14,NO.35-48; Table 1,3) which had negative results by conventional test, mNGS analysis produced negative results in 4 cases, produced monomicrobial detection in 5 cases and polymicrobial detection in 19 cases. Among 24 patients positive for pathogens, mNGS identified 17 species of pathogens. The most frequent detected pathogen was Human cytomegalovirus (12 cases), followed by Pneumocystis jirovecii (5 cases), Ralstonia insidiosa(5 cases), Acinetobacter baumannii (5 cases), Pseudomonas aeruginosa (4 cases). Fungi were reported positive from 13 patients. There were 5 cases (NO.1, NO.4, NO.13, NO,36, NO.46) positive for Pneumocystis jirovecii by mNGS. Human cytomegalovirus and Pneumocystis jirovecii were detected in NO.1 and NO.4. Human cytomegalovirus, Pneumocystis jirovecii and Rhizopus microsporus were detected in NO.13. Pneumocystis jirovecii and Torque teno virus were detected in NO.36. NO.46 was positive for Pneumocystis jirovecii. There were another 2 specimen (NO.3,NO.6) contained Aspergillus fumigatus in combination with other pathogens. NO.11 was identified as mixed fungal infection (Aspergillus niger and Candida albicans) by mNGS, in combination with Acinetobacter baumannii . Other specimen contained Rhizopus delemar (NO.14), Aspergillus oryzae (NO.12,NO.48),Cryptococcus neoformans (NO.10) and Rhizopus oryzae (NO.7) (See Supplementary Table 1, Additional File 1). In conclusion, for negative cases identified by conventional test, mNGS raised the efficiency of mixed pulmonary infection diagnosis.
3.3.2 The Efficiency of mNGS in Positive Cases for Single Infection Identified by Conventional test
Of 21 cases (21/55=38.2%) which were identified as single infection by conventional test, mNGS results were consistent with conventional tests in 2 cases (NO.51 and NO.53). The pathogen detected were Pseudomonas aeruginosa and Acinetobacter baumannii in NO.51 and NO.53, respectively. Another 13 results were partially matched, among which 6 cases were positive for Pneumocystis jirovecii (NO.16,NO.50),Staphylococcus epidermidis (NO.17),Acinetobacter baumannii (NO.19),Pseudomonas aeruginosa (NO.21) and Klebsiella pneumoniae (NO.24) by culturing. However, mNGS detected more pathogens. Specimen form Patient NO.22 and NO.28 had positive GM test results. Besides Aspergillus, mNGS also detected Haemophilus parainflfluenzae and Human cytomegalovirus in NO.22 and NO.28, respectively. The cryptococcal capsular polysaccharide antigen test had positive results in NO.20, and mNGS results reported mixed infection of Cryptococcus neoformans and Pneumocystis jirovecii. Among 3 cases which were positive for Mycobacterium tuberculosis by Xpert MTB, 2 cases were positive for Mycobacterium tuberculosis and Human cytomegalovirus by mNGS, and the other was positive for Mycobacterium tuberculosis and Torque teno virus. The histopathological result of NO.15 detected mold hyphae, with evidence including alveolar septal fibrous tissue hyperplasia, inflammatory exudate necrosis, silk-like structure and positive PAS results. Results of mNGS were positive for Pneumocystis jirovecii and Aspergillus oryzae (Table 3). In summary, mNGS complemented the diagnosis of mixed pulmonary infection.
Table 3 Results obtained in the analysis of respiratory specimens of patients.
|
No. Patients (%)
|
Positive for mixed infection by mNGS
|
42 (76.4)
|
Positive for single infection by mNGS
|
8(14.5)
|
Negative for pathogen by mNGS
|
5(9.1)
|
Positive for mixed infection by conventional test
|
6(10.9)
|
Positive for single infection by conventional test
|
21(38.2)
|
Negative for pathogen by conventional test
|
28(50.9)
|
Among the other 6 patients, results of conventional test were paradoxical with that of mNGS. In Patient NO.18 , a 36-year-old woman with no underlying disease, Cryptococcus neoformans identified by cryptococcal capsular polysaccharide antigen test was not detected by mNGS, whereas mNGS reported Klebsiella pneumoniae, Pseudomonas aeruginosa, Haemophilus parainflfluenzae and Aspergillus fumigatus. In addition, histopathology and culturing results were negative. In Patient NO.23, a 24-year-old man with acute myeloid leukemia, fungus identified by culture was not detected by mNGS, whereas mNGS reported Klebsiella pneumoniae, Human cytomegalovirus and Rhizomucor pusillus. Rhizomucor pusillus is a thermophilic fungus that lives in hot environment and can infect humans and animals. It can cause necrosis of infected tissues and invade nervous system. It is commonly found in lungs of immunocompromised patients , so it was also interpreted as infectious pathogen. In Patient NO.27, a 46-year-old woman with lymphoma, Staphylococcus epidermidis identified by culturing was not detected by mNGS, whereas mNGS reported Human cytomegalovirus and Pneumocystis jirovecii .In Patient NO.29, a 63-year-old woman with no underlying disease, Aspergillus identified by GM test was not detected by mNGS, whereas mNGS reported Haemophilus parainflfluenzae and Pseudomonas aeruginosa. In Patient NO.52, a 26-year-old man with no underlying disease, Aspergillus identified by GM test was not detected by mNGS, whereas mNGS reported Acinetobacter baumannii and Cryptococcus neoformans . Patient NO.54 is a 68-year-old man with no underlying disease. The culture result was Flavobacterium indologenes, whereas mNGS reprted Cryptococcus neoformans only (See Supplementary Table 4, Additional File 2).
3.3.3 The Efficiency of mNGS in Positive Cases for Mixed Pulmonary Infection Identified by Conventional test
Of 6 cases (6/55=10.9%) which were identified as mixed infection by conventional test, mNGS results were consistent with conventional tests in Patient NO.31. In 4 out of the 6 cases, results of mNGS and conventional tests were partially matched. Patient NO.30, a 62-year-old man with no underlying disease, had culturing positive results for Acinetobacter baumannii and Pneumocystis jirovecii . Besides, human cytomegalovirus nucleic acid test was positive. In addition to above 3 pathogens, mNGS also detected Aspergillus fumigatus. The possible reason for the absence of Aspergillus fumigatus in culturing was the limited incubation duration. Specimen from Patient NO.32, a 63 year old woman with acute lymphoblastic leukemia, was positive for Pseudomonas aeruginosa by culturing and positive for Aspergillus by GM test, whereas mNGS reported Aspergillus fumigatus ,Pseudomonas aeruginosa and Streptococcus pneumoniae . Patient NO.34 was a 59-year-old man with no underlying disease. Human cytomegalovirus identified by mNGS was not detected by conventional test. Aspergillus identified by GM test was not detected by mNGS. Patient NO.33 was a 56-year-old man with no underlying disease. Pneumocystis jirovecii was positive by culturing. GM test and human cytomegalovirus nucleic acid test were positive. mNGS identified Pneumocystis jirovecii and human cytomegalovirus, but not Aspergillus. Patient NO.55, a 60-year-old man with positive T-spot result, was positive for Acinetobacter baumannii by culturing .But mNGS result of NO.55 was negative (Table 3). In conclusion, even in the specimen where conventional tests identified mixed pulmonary infection, mNGS still played an important role, because it has the ability to identify both common and rare pathogens without any prior hypothesis.
3.4 Pathogens Detected by mNGS
In 55 specimen, 5 species of pathogens (Mycobacterium abscessus, Rhizopus, Haemophilus parainflfluenzae, Rhizomucor pusillus, Streptococcus pneumoniae) were identified by mNGS, but not by conventional tests; however, Flavobacterium indologenes was only detected by the conventional test. Among 55 specimen, the most frequently detected pathogen by mNGS was Human cytomegalovirus, followed by Pneumocystis jirovecii, Pseudomonas aeruginosa, Acinetobacter baumannii, Klebsiella pneumoniae and Aspergillus fumigatus . mNGS reported 19 (19/55 = 55.88%) mixed infections containing Human cytomegalovirus. Human cytomegalovirus and Pneumocystis jirovecii were the most commonly detected co-pathogens in the group of polymicrobial pulmonary infection which were detected in 7 cases. In addition, Human cytomegalovirus often co-occurred with Pseudomonas aeruginosa (4 cases),Aspergillus fumigatus (3 cases),Klebsiella pneumoniae (3 cases) and Acinetobacter baumannii (3 cases). Pneumocystis jirovecii was second in frequency of detection , which was reported in 13 cases (23.6%). Pseudomonas aeruginosa often coexisted with Klebsiella pneumoniae and this combination was detected in 5 cases which were the second most common co-pathogens in mixed pulmonary infection (Table 5).
Table 5 Human cytomegalovirus,Pneumocystis jirovecii,Pseudomona aeruginosa,Klebsiella pneumoniae ,Acinetobacter baumannii and Aspergillus fumigatus among 55 patients
Pathogen
|
No. (%)
|
No. Occurrences with
|
Human cytomegalovirus
|
Pneumocystis jirovecii
|
Pseudomonas aeruginosa
|
Klebsiella pneumoniae
|
Acinetobacter baumannii
|
Aspergillus fumigatus
|
Human cytomegalovirus
|
22 (40.0)
|
—
|
7
|
4
|
3
|
3
|
3
|
Pneumocystis jirovecii
|
13 (23.6)
|
7
|
—
|
1
|
0
|
1
|
1
|
Pseudomonas aeruginosa
|
11 (20.0)
|
4
|
1
|
—
|
5
|
2
|
1
|
Klebsiella pneumoniae
|
8 (14.5)
|
3
|
0
|
5
|
—
|
1
|
2
|
Acinetobacter baumannii
|
8 (14.5)
|
3
|
1
|
2
|
1
|
—
|
2
|
Aspergillus fumigatus
|
7 (12.7)
|
3
|
1
|
1
|
2
|
2
|
—
|