Background
Pulmonary infection is a leading cause of death and morbidity worldwide [
1]. The risk of mixed pulmonary infection is high, especially in immunocompromised patients such as patients with hematological malignancies. Mixed pulmonary infection was defined when two or more infectious pathogens were indentified. Compared to patients with monomicrobial pulmonary infection, patients with polymicrobial pulmonary infection may have different antibiotic spectrums and more severe clinical manifestation. Diagnosis of polymicrobial infection must be as accurate as possible, because combined treatment has many potential side effects [
2]. However, fast and accurate infection diagnose is challenging due to the limitations of current conventional tests in terms of sensitivity, speed and spectrum for pathogen detection [
3,
4].
Next-generation sequencing (NGS), also termed high-throughput or massively parallel sequencing, is a technology that allows for thousands to billions of DNA fragments to be simultaneously and independently sequenced. The applications of NGS in clinical microbiological testing are manifold, including metagenomic NGS (mNGS), which allows for an unbiased detection of pathogens [
5]. When applied to clinical practice, Qing Miao et al. reported that the sensitivity and specificity of mNGS for diagnosing infectious disease were 50.7 and 85.7% respectively. mNGS outperformed culture-depend methods, especially for the detection of
Mycobacterium tuberculosis (MTB), viruses, anaerobes and fungi. Furthermore, mNGS is less affected by prior antibiotic exposure [
6]. In addition, mNGS could offer an improved detection of pulmonary infectious pathogens in lung biopsy tissues, with potential benefits in speed and sensitivity [
7]. However, reports on the use of mNGS in mixed pulmonary infection remain scarce. Most studies on mNGS focused on the diagnosis of single infection.
In this study, we evaluated the performance of this approach in the diagnosis of mixed pulmonary infection. The mNGS results were compared with those from conventional laboratory-based diagnostic methods. Our results indicated that mNGS benefited the efficiency of co-pathogens detection.
Discussion
mNGS offers the possibility of fast pathogen identification without a prior hypothesis of the target. Theoretically, given sufficiently long sequencing lengths, multiple hits to the microbial genome, and a well-annotated reference database, nearly all microorganisms can be uniquely identified [
11]. This retrospective study for the first time reported the sensitivity and specificity of mNGS in the diagnosis of mixed pulmonary infection.
Compared to conventional tests, the sensitivity of mNGS was significantly higher (97.2% vs 13.9% of conventional tests;
P < 0.01), while the specificity of mNGS was lower (63.2% vs 94.7% of conventional tests;
P = 0.07). For infectious disease diagnosis, Qing Miao et al. reported mNGS increased the sensitivity rate by approximately 15% in comparison with that of culturing (50.7% vs 35.2%;
P < 0.01), while the specificity rate of mNGS was comparable with that of culture (89.1% vs 85.7% vs;
P = 0.39) which is inconsistent with our data [
6] .This may be due to the fact that false positive rate of mNGS was high in our results, which was 16.7% (95% CI: 7.5–32.0%).
According to our data, mNGS had a broader spectrum for pathogen detection than conventional tests. Most patients (60.0%) enrolled in this study were immunocompromised because of hematological malignancies, and the efficacy of routine culturing (i.e., growth in media) in pathogen detection was hampered by early administration of broad spectrum or prophylactic antimicrobial drugs. The presence of fastidious or slow growing pathogen also limited the sensitivity of culturing-based methods [
5]. Application of mNGS improved the diagnosis sensitivity of pulmonary fungal infections. mNGS identified fungi in 31 (56.4%) out of 55 cases, of which only 10 (18.2%) cases were positive for the same fungi by conventional tests. Qing Miao et al. systematically compared detection by mNGS and culturing in a pairwise manner and found that mNGS had superior feasibility in detecting fungi (OR, 4.0 [95% CI, 1.6–10.3]; P<0.01) [
6]. In our results,
Rhizopus identified in 3 cases by mNGS was not detected by any conventional tests. Henan Li et al. reported that tissues were usually homogenized in a glass grinder and used for smear and culture in the clinical microbiology laboratory, and this grinding procedure may affect the isolation of
Zygomycetes (such as
Rhizopus and
Mucor). The mNGS analysis doesn’t require this grinding procedure, and identified more
Zygomycetes than culturing-based method [
7]. The number of cases positive for
Aspergillus identified by conventional tests (9 cases, 16.4%) was less than the number of cases identified by mNGS (14 cases, 25.5%).
Aspergillus culturing is time-consuming with low positive rate. The time required for smear to check fungi is short, but operators are supposed to have higher abilities to identify fungi among the same genus. The GM test is highly recommended in the diagnosis of
Aspergillus [
12]. However, there are many controversies in the application of GM test, such as: 1) sensitivity and specificity are varied in different diseases; 2) special types of diseases and patient status can lead to false positive results.
The results of this study indicated that mNGS covered more bacteria. The mixed infection of
Pseudomonas aeruginosa and
Klebsiella pneumoniae in this study was the second (5 cases, 13.9%) common combination. The positive rate of
Pseudomonas aeruginosa and
Klebsiella pneumoniae by mNGS was higher than that by culturing. The positive rate of other bacteria such as
Mycobacterium tuberculosis, Acinetobacter baumannii and Haemophilus parainflfluenzae by mNGS was also higher than that by culturing. However, Toma et al reported that, compared with sequencing, culturing-based method is able to identify the vast majority (74%) of bacterium-associated pneumonia [
13]. The inconsistence between our study and Toma’s might result from the low immune functions of most patients in this study. The use of prophylactic or broad-spectrum antibiotics made bacterial culture even more difficult.
In this study, the underlying diseases of 23 patients were hematological malignancies with low immune functions. Thus, pathogens of mixed infections in these patients might be different from those in the general population.
Human cytomegalovirus was the most commonly detected pathogen in the study, which occurred in 19 cases of mixed infection. Of these 19 patients, only 2 patients were positive for
Human cytomegalovirus by conventional tests, and 17 patients were positive by mNGS. We also detected mixed infections of
Human cytomegalovirus and
Pneumocystis jirovecii in 7 patients, which was the most common combination of pathogens. Immunocompromised patients are susceptible to infection by these pathogens.
Human cytomegalovirus is a common β-herpesvirus that infects most of the adult population. It remains predominantly dormant after primary infection, and is relatively innocuous in healthy adults [
14]. However, in patients with immune dysfunction or immunosuppression, such as acquired immune deficiency syndrome (AIDS) patients, organ transplantation recipients, and patients in the intensive care unit (ICU) [
15],
Human cytomegalovirus infection may cause serious end-stage diseases, such as leukopenia, hepatitis, nephritis, interstitial pneumonia, gastrointestinal disease and even death [
16,
17].
Pneumocystis jirovecii was an early indicator of the human immunodeficiency virus (HIV) epidemic and occurred in 70–80% of AIDS patients [
18]. There is an increasing population of susceptible non-HIV-infected patients, including those with solid malignancies, solid organ transplantation and the recipients of hematopoietic stem cell transplantation, patients receiving immunosuppressive therapies for autoimmune and inflammatory conditions and those with genetic primary immune deficiency disorders [
19]. A national study over the decade 2000–2010 showed an increase in incidence of
Pneumocystis jirovecii infection, and the largest population associated with
Pneumocystis jirovecii were those suffering from underlying hematological malignancy [
20]. The difficulty in isolating and culturing
Pneumocystis jirovecii has hindered both diagnosis and research. Several methods using various coculture cell lines were described but failed to attain widespread use [
21‐
24]. The application of mNGS is a promising method for the fast and accurate detection of
Pneumocystis jirovecii.
Our study had several limitations. In our study, the most common co-pathogens in mixed infections were human cytomegalovirus and Pneumocystis jirovecii, while the majority of patients were immunocompromised, which may lead to biased conclusions. Moreover, our mNGS tests were delivered to the commercial laboratory rather than an microbiology laboratory in hospital, which might sacrifice sensitivity rate because of reduced viability due to increased turnaround time.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.