Evaluation of a multiplex PCR kit for detection of 17 respiratory pathogens in hospitalized patients.

Background: Rapid pathogen identification is critical for optimizing diagnosis and treatment of infectious diseases. Multiplex polymerase chain reaction (PCR) is a sensitive, broad-spectrum molecular detection technique that is simple and rapid to perform. It is capable of simultaneously screening for multiple pathogens within a short time range. Here, we designed and evaluated a multiplex PCR kit for the identification of 17 common respiratory pathogens in clinical samples from hospitalized patients.
Methods: A total of 452 samples from hospitalized patients, including 242 respiratory and 210 non-respiratory samples, were analyzed for 13 bacteria and 4 fungi by a multiplex fluorescent PCR kit. The diagnostic performance of the kit was assessed by considering routine microbiology as the reference standard.
Results: The overall positivity rate of the multiplex PCR kit was 86.9%, much higher than that noted on routine microbiology (56.9%). Furthermore, the co-infection detection rate was also significantly higher than that noted on routine microbiology (69.5% vs. 15.0%). Compared with routine microbiology, kit sensitivity was >90% for detection of most target bacteria, with a negative predictive value (NPV) of >99%, especially for detection of Pseudomonas aeruginosa, Klebsiella pneumoniae, Acinetobacter baumannii, Stenotrophomonas maltophilia, Enterococcus faecium, Enterococcus faecalis and Escherichia coli. The kit was noted to be particularly superior in identifying Stenotrophomonas maltophilia and Streptococcus pneumoniae as compared to routine microbiology. The multiplex PCR kit was noted to be less sensitive (33.3-59.6%) and more specific (93.9-100.0%) for detection of mycobacteria and fungi. Conclusions: Our multiplex PCR kit offers a rapid and sensitive diagnosis of common bacterial pneumonia, although sensitivity for mycobacteria and fungi warrants enhancement. Further optimization includes minimizing false positivity and increasing relevance to clinical application. 2022 Journal of Thoracic Disease. All rights reserved.

Introduction

Pulmonary infections are the largest cause of human disease burden and carry high morbidity and mortality (1). Crude mortality was estimated at 20–60% for hospitalized patients with severe community-acquired pneumonia (2-4) and 30–70% for hospital-acquired pneumonia (5). Rapid identification of causative pathogens and prompt initiation of effective antimicrobial therapy are crucial for the prognosis of pneumonia patients, with significantly increased mortality reported among patients receiving delayed or ineffective treatment (6). In China, in-hospital mortality rates for diagnosed and undiagnosed patients are 21.7% and 25.9%, respectively (2).

Diagnosis of most bacterial and fungal infections depends on successful microbial or fungal culture, a process that takes about 48–72 h, prior to which patients can only receive empirical therapy. In addition, the low sensitivity of routine microbiology (approximately 44.2%) (7) that frequently limits patients to broad-spectrum treatment not only worsens prognosis but also increases the risk of adverse effects and promotes antibiotic resistance (8,9).

Multiplex polymerase chain reaction (PCR) offers rapid detection and precise identification of a large number of respiratory viruses by amplifying genomic fragments (10). A prior study reported a 30–50% increase in respiratory virus identification when using multiplex PCR as compared to antibody and culture methods (11). However, relatively few studies have investigated the use of multiplex PCR for the diagnosis of bacterial and fungal infections. Here, we designed and assessed a multiplex fluorescent PCR kit capable of detecting 17 respiratory pathogens (13 bacteria and 4 fungi) simultaneously in one PCR panel. We collected 452 respiratory and non-respiratory samples from hospitalized patients to evaluate the clinical performance of this kit and compared the results with those of routine microbiological tests. We present the following article in accordance with the STARD reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-22-544/rc).

Methods

Multiplex fluorescent PCR kit

The 17 respiratory pathogens targeted for detection by our multiplex PCR kit were as follows: Pseudomonas aeruginosa, Klebsiella pneumoniae, Acinetobacter baumannii, Staphylococcus aureus, Stenotrophomonas maltophilia, Streptococcus pneumoniae, group A streptococcus, Haemophilus influenzae, Enterococcus faecium, Enterococcus faecalis, Escherichia coli, Mycobacterium tuberculosis, atypical mycobacteria, Candida albicans, Mucor, Aspergillus and Penicillium marneffei. Target sequences corresponding to these pathogens were searched in the National Center of Biotechnology Information (NCBI); retrieved sequences were aligned using specialized software to select conserved regions with higher homology as targets for pathogen detection and thus ensure greater specificity. Corresponding primer and probe sequences were designed according to relevant pathogen conserved sequences (). The pre-designed multiplex fluorescent PCR kit panel included the primer and probe reaction mixture (4 µL) as well as nucleic acid amplification reaction solution (16 µL). The nucleic acid amplification reaction solution used as quantitative real time PCR (qPCR) reaction premix consisting of Tris (0.05–0.1 M), KCl (100–200 mM), MgCl2 (1–5 mM), dNTPs (0.1–0.8 mM) and Taqman (0.1–1 U/µL).

Table 1

Primers and probes used for the detection of 17 pathogens

Pathogen	Forward primer (5'–3')	Reverse primer (5'–3')	Probe (5'–3')
Bacteria
Pseudomonas aeruginosa	TCGCAGACCAAGGACAAGCT	TTGCCCATCTCCTGTTCCA	TCTCCTCCGAGGTGAAGACCGCG
Klebsiella pneumoniae	TCCCCTTTGCCGTGAATAATC	GCGGCATACGCTGCTGTAT	CCCCGGTGGTCACCATTTCGG
Acinetobacter baumannii	TGCGACACAACTCGACGTTT	AATCTAGCACGACCTGACCATAGAC	TTTAAACCGATTGATTTGTCGCCGATCTTT
Staphylococcus aureus	TGATGGCTTTGAAGTAGTTTT	CACGATTCGAATAGTAAACATAA	TGCAGCAAGCCTTTTCTCTAAAATT
Stenotrophomonas maltophilia	CATGGCCAACGAAGAAAAGC	TGCAGGGTGTGGGTCACTT	TACGGCGTGCAGTTCCACCCG
Streptococcus pneumoniae	CCTCGTTGAAGCAATGGTGC	CCCAACAAGTGAATCACCAACA	TGGCATGGGCATGGTTGGTTTGGT
Group A streptococcus	CTCCTGGTGATCCCATACCAA	CTCCTGGTGATCCCATACCAA	TCCCACAAAGTCAGCACTGCTTAGACCA
Haemophilus influenzae	CCCAACAAGTGAATCACCAACA	TTCGCACATGAGCGTCAGTA	ACCGAAGGCGAAGGCAGCCCC
Enterococcus faecium	CAATGCTGCTTTGATACGAGTGT	AAGTTCTTGTCCGTGTTGACTTCA	CAGTGATCACGCCGTCTTTCAAAGGAA
Enterococcus faecalis	ACTTTGGTGTTGTTGAAGGTTTAATG	CCTTTAGGATGTGGTCCGTCTAA	CTATCCACGCTTACACAGGTGACCAAATGA
Escherichia coli	GCTGCTGTGGCGTCAAACT	GACCTACATGAGTGATTGCCTGAA	TTTTCACCAGGCGCAGACTTGCTGT
Mycobacterium tuberculosis	GCATCTGGCCACCTCGAT	GCCGCCAACTACGGTGTTTA	CCCTCACGGTTCAGGGTTAGCCACA
Atypical mycobacteria	TCCCGGGCCTTGTACACA	CCACTGGCTTCGGGTGTTA	CGCCCGTCACGTCATGAAAGTCG
Fungi
Candida albicans	ACCTGAAGTTTTACAATCAGCAACA	TGCTCGTAGCATTATCTATGCCTTA	TTACCAGCAGAATCAAAATGCACTTGACCA
Mucor	AGTACTTTGAAAAGAGAGTTAAACAG	GCTGATTAACTTCAAGTCAGTCT	TCGCAACCGACTCCATTAAGAACACCA
Aspergillus	GGCCGACAACAGCGTCAT	TCTGCTTGGCGGTGATGTAA	ATGTGAAATTGCCAAGAGGGAAGCATTTG
Penicillium marneffei	TGAAATTGTTCCTGCTCATGGT	CCACTCCCGTCGTAAATGTGT	TTCCTGGCATCCCTGTCAGCCATT
Internal reference	CTTCAGCATGGCGGTGTTT	CCGCGCAGAGCCTTCA	CAGATTTGGACCTGCGAGCGGG

The plasmid used as a positive control for kit sensitivity evaluation was purchased from Shanghai Generay Biotech Co., Ltd. Target pathogen plasmids were mixed and diluted with Tris-EDTA to 100 copies/µL, 10 copies/µL, 2 copies/µL, 1 copy/µL, and 0.5 copies/µL, and then detected using the PCR kit. For sensitivity testing, a positive result was determined if the cycle threshold (Ct) value was ≤38 and the lowest copy number for which both tests were positive was considered as test sensitivity. Results revealed the minimum copy number of each target pathogen detected to have been 0.5 copies/µL and 1 copy/µL ().

Table 2

Sensitivity assessment of the multiplex PCR kit for the 17 pathogens

Pathogen	100 copies/μL			10 copies/μL			2 copies/μL			1 copies/μL			0.5 copies/μL
	Value 1 (Ct)	Value 2 (Ct)		Value 1 (Ct)	Value 2 (Ct)		Value 1 (Ct)	Value 2 (Ct)		Value 1 (Ct)	Value 2 (Ct)		Value 1 (Ct)	Value 2 (Ct)
Bacteria
Pseudomonas aeruginosa	28.67	28.54		31.69	30.67		33.44	33.25		34.22	36.31		34.71	35.87
Klebsiella pneumoniae	27.15	27.30		30.51	29.44		32.41	32.17		33.87	33.25		35.38	32.66
Acinetobacter baumannii	29.95	29.88		33.27	32.99		36.06	35.59		35.57	–		37.12	36.71
Staphylococcus aureus	28.89	29.01		32.84	32.04		34.37	34.90		36.73	35.43		36.13	36.17
Stenotrophomonas maltophilia	29.11	29.12		32.23	32.14		34.86	35.65		37.62	35.76		37.67	–
Streptococcus pneumoniae	29.37	29.47		33.16	30.06		34.54	35.46		35.24	37.19		–	–
Group A streptococcus	27.59	27.60		31.32	30.83		33.17	33.28		35.00	33.79		35.71	35.50
Haemophilus influenzae	30.64	30.91		35.48	35.06		38.78	35.12		34.82	–		–	36.95
Enterococcus faecium	30.28	30.43		33.16	34.57		34.79	35.90		–	36.17		36.91	–
Enterococcus faecalis	30.14	30.50		34.95	33.36		34.96	35.23		36.94	36.45		–	–
Escherichia coli	28.97	29.25		32.22	32.37		36.10	34.39		35.91	34.88		–	–
Mycobacterium tuberculosis	28.51	28.52		31.51	32.61		35.18	34.64		34.89	36.10		–	36.82
Atypical mycobacteria	27.91	27.52		30.87	31.05		34.35	34.69		34.46	34.36		33.90	35.97
Fungi
Candida albicans	26.350	28.848		30.651	33.690		34.477	36.970		36.164	34.976		36.517	36.377
Mucor	27.431	27.457		30.505	31.309		35.603	33.693		35.324	35.225		35.571	36.694
Aspergillus	25.937	25.928		27.874	31.236		32.483	32.921		33.645	34.419		34.821	34.613
Penicillium marneffei	27.053	27.564		32.355	31.663		34.128	33.295		33.570	35.512		36.055	36.943

PCR, polymerase chain reaction; Ct, cycle threshold value.

Samples

According to the number of samples required for positive validation and negative controls, a total of 483 samples (452 of them eligible) were discontinuously collected from infectious disease patients admitted to the Department of Pulmonary Medicine at Zhongshan Hospital, Fudan University between December 2020 and October 2021. Specimens of sufficient volume for two repeated PCR tests (>1 mL) and had at least one available test result on routine microbiological testing were included. The 452 eligible samples included 242 respiratory samples and 210 non-respiratory samples. Respiratory samples consisted of 206 sputum and 36 lower respiratory tract samples (endotracheal aspirates, bronchoalveolar lavage fluid and lung tissue). Non-respiratory samples consisted of 62 excrement (feces and urine) and 148 other body fluid and tissue samples (pleural effusion, ascitic fluid, cerebrospinal fluid, bile, pericardial fluid). After sample collection, approximately 1 mL of specimen was aliquoted per sample and stored at −80 ℃ for multiplex PCR testing within one week. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). This study was approved by the Medical Ethics Committee of Zhongshan Hospital, Fudan University (approval No. B2019-249R) and written informed consent was obtained from all patients.

Nucleic acid extraction

After homogenization, enzyme digestion or sample centrifugation, homogenous samples (300 µL) were transferred to a sample lysis plate and subjected to automated nucleic acid extraction using a nucleic acid extraction kit (Cat. #TQ-BG-001; BioGerm, Shanghai, China) and a BG-Abot-96 automated nucleic acid extraction system (BioGerm) according to manufacturer instructions.

Multiplex fluorescent PCR detection of respiratory pathogens

Nucleic acid extracted from all 452 specimens was subjected to multiplex amplification using our multiplex fluorescent PCR kit for respiratory pathogen detection. All PCR reactions were carried out using 5 µL of resuspended deoxyribonucleic acid (DNA), 4 µL of primer/probe mixture and 16 µL of nucleic acid amplification reaction solution. Optimal cycling conditions for nucleic acid amplification were initially 5 min at 95 ℃ followed by 40 cycles at 95 ℃ for 10 s and 55 ℃ for 40 s using an ABI 7500 (Invitrogen, Waltham, USA) device. It took a total of 80 min to complete pathogen detection. A positive result was defined by Ct values of ≤35 or 35–38 confirmed on repeat testing. Clinical information and routine microbiological tests results were not provided to researchers performing and evaluating multiplex PCR testing.

Routine microbiological result collection

All routine microbiological tests were performed according to requests of clinicians based on relevant standard operating procedures. Routine microbiological tests including bacterial and fungal smears and cultures, mycobacterial acid-fast smear and culture, and Aspergillus culture; data were collected from medical records upon conclusion of multiplex PCR testing. Clinical data and multiplex PCR test results were not available to performers of routine microbiological examinations.

Statistical analyses

Statistical analyses were performed using SPSS version 20 (IBM, Armonk, USA) and MedCalc version 20.027 (MedCalc Ltd. Ostend, Belgium). Statistical comparisons were analyzed using the chi-squared test. Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) were analyzed for each PCR target considering routine microbiology as the reference standard (smears and cultures are the traditional diagnostic criteria for the 17 pathogens detected by our kit) (12).

Results

Routine microbiology results

Conventional bacterial smears and cultures were performed using 442 samples for detection of Pseudomonas aeruginosa, Klebsiella pneumoniae, Acinetobacter baumannii, Staphylococcus aureus, Stenotrophomonas maltophilia, Streptococcus pneumoniae, group A streptococcus, Haemophilus influenzae, Enterococcus faecium, Enterococcus faecalis and Escherichia coli (). Conventional fungal smears and cultures were performed on 405 samples to identify Candida albicans, Mucor, Aspergillus and Penicillium marneffei (). A total of 145 samples were evaluated using acid-fast smear and culture for M. tuberculosis. Aspergillus cultures were performed on 93 samples (). Presence of one or more pathogens was noted in 56.9% (257/452) of all samples; the positivity rate in respiratory and non-respiratory samples was 74.0% (179/242) and 37.1% (78/210), respectively.

Table 3

Positive rates and numbers of pathogens detected by routine microbiology from respiratory samples or non-respiratory samples

Pathogen	Positive rate
	All samples (n=452)	Respiratory samples (n=242)	Non-respiratory samples (n=210)
Bacteria
Pseudomonas aeruginosa	15.8% (70/442)	22.8% (54/237)	7.8% (16/205)
Klebsiella pneumoniae	6.8% (30/442)	8.4% (20/237)	4.9% (10/205)
Acinetobacter baumannii	19.7 (87/442)	23.6% (56/237)	15.1% (31/205)
Staphylococcus aureus	5.4% (24/442)	7.2% (17/237)	3.4% (7/205)
Stenotrophomonas maltophilia	3.6% (16/442)	5.1% (12/237)	2.0% (4/205)
Streptococcus pneumoniae	0% (0/442)	0% (0/237)	0% (0/205)
Group A streptococcus	0% (0/442)	0% (0/237)	0% (0/205)
Haemophilus influenzae	1.4% (6/442)	2.5% (6/237)	0% (0/205)
Enterococcus faecium	3.6% (16/442)	1.7% (4/237)	5.9% (12/205)
Enterococcus faecalis	2.3% (10/442)	1.3% (3/237)	3.4% (7/205)
Escherichia coli	2.0% (9/442)	0.8% (2/237)	3.4% (7/205)
Mycobacterium tuberculosis	6.2% (9/145)	8.6% (8/93)	1.9% (1/52)
Atypical mycobacteria	5.5% (8/145)	8.6% (8/93)	0% (0/52)
Fungi
Candida albicans	11.9% (48/405)	19.1% (43/225)	2.8% (5/180)
Mucor	0% (0/405)	0% (0/225)	0% (0/180)
Aspergillus	6.5% (6/93)	6.6% (6/91)	0% (0/2)
Penicillium marneffei	0% (0/405)	0% (0/225)	0% (0/180)

Among the 442 samples studied using bacterial smear and culture, Acinetobacter baumannii and Pseudomonas aeruginosa were the most frequently noted pathogens among both respiratory and non-respiratory samples (). Acinetobacter baumannii was found in 23.6% (56/237) of respiratory and 15.1% (31/205) of non-respiratory samples, while Pseudomonas aeruginosa was found in 22.8% (54/237) of respiratory and 7.8% (16/205) of non-respiratory samples (). Klebsiella pneumoniae was the third most frequently detected pathogen (8.3%) in respiratory samples (), while Enterococcus faecium was the third most frequently detected pathogen (5.9%) in non-respiratory samples (). Notably, neither Streptococcus pneumoniae nor group A streptococcus was noted among the 442 samples evaluated (; ). Similarly, Haemophilus influenzae and atypical mycobacteria were not detected among the 210 non-respiratory samples (). On fungal smear and culture, Candida albicans was found in 43 respiratory and five non-respiratory samples, while Aspergillus was found in six respiratory samples ().

Figure 1

Numbers and types of pathogens detected by routine microbiology or multiplex PCR kit from studied samples. (A) Numbers and types of pathogens detected by routine microbiology or multiplex PCR kit from respiratory samples; n=242. (B) Numbers and types of pathogens detected by routine microbiology or multiplex PCR kit from non-respiratory samples; n=210. PCR, polymerase chain reaction; MTB, Mycobacterium tuberculosis; GAS, group A streptococcus; AMB, atypical mycobacteria.

Multiplex PCR results

The overall positive rate for multiplex PCR detection exceeded that of routine microbiology (86.9%; 393/452); overall positive rates in respiratory and non-respiratory specimens were 94.2% (228/242) and 78.6% (165/210), respectively. Furthermore, the most frequently detected pathogens on multiplex PCR were Stenotrophomonas maltophilia in both respiratory (65.7%; 159/242) and non-respiratory (42.4%; 89/210) samples (). In respiratory samples, the other most frequently detected pathogens were Acinetobacter baumannii (56.6%; 137/242), Pseudomonas aeruginosa (51.2%; 124/242), Klebsiella pneumoniae (48.8%; 118/242) and Staphylococcus aureus (27.7%; 67/242) (; ). Among non-respiratory samples, the other four most frequently noted pathogens were Acinetobacter baumannii (40.5%; 85/210), Klebsiella pneumoniae (24.3%; 51/210), Pseudomonas aeruginosa (21.9%; 46/210) and Enterococcus faecalis (16.2%; 34/210) (; ). Streptococcus pneumoniae was found in 36 samples using multiplex PCR but not routine microbiology ().

Table 4

Positive rates and numbers of pathogens detected by multiplex PCR from respiratory samples or non-respiratory samples

Pathogen	Positive rate
	All samples (n=452)	Respiratory samples (n=242)	Non-respiratory samples (n=210)
Bacteria
Pseudomonas aeruginosa	37.6% (170/452)	51.2% (124/242)	21.9% (46/210)
Klebsiella pneumoniae	37.4% (169/452)	48.8% (118/242)	24.3% (51/210)
Acinetobacter baumannii	49.1% (222/452)	56.6% (137/242)	40.5% (85/210)
Staphylococcus aureus	21.9% (99/452)	27.7% (67/242)	15.2% (32/210)
Stenotrophomonas maltophilia	54.9% (248/452)	65.7% (159/242)	42.4% (89/210)
Streptococcus pneumoniae	8.0% (36/452)	14.0% (34/242)	1.0% (2/210)
Group A streptococcus	0.9% (4/452)	1.7% (4/242)	0% (0/210)
Haemophilus influenzae	7.7% (35/452)	12.0% (29/242)	2.9% (6/210)
Enterococcus faecium	11.1% (50/452)	9.5% (23/242)	12.9% (27/210)
Enterococcus faecalis	16.8% (76/452)	17.4% (42/242)	16.2% (34/210)
Escherichia coli	11.7% (53/452)	13.2% (32/242)	10.0% (21/210)
Mycobacterium tuberculosis	1.1% (5/452)	2.1% (5/242)	0% (0/210)
Atypical mycobacteria	0.4% (2/452)	0.4% (1/242)	0.5% (1/210)
Fungi
Candida albicans	11.5% (52/452)	18.2% (44/242)	3.8% (8/210)
Mucor	0.2% (1/452)	0% (0/242)	0.5% (1/210)
Aspergillus	0.9% (4/452)	1.2% (3/242)	0.5% (1/210)
Penicillium marneffei	0.2% (1/452)	0.4% (1/242)	0% (0/210)

PCR, polymerase chain reaction.

No atypical mycobacteria were detected in respiratory samples and no Mycobacterium tuberculosis or group A streptococcus were detected in non-respiratory samples (). Multiplex PCR results revealed 48 cases of fungi in respiratory samples, including 44 Candida albicans, 3 Aspergillus and 1 Penicillium marneffei, as well as 10 cases in non-respiratory samples, including 8 Candida albicans, 1 Mucor and 1 Aspergillus ().

Comparison of routine microbiology and multiplex PCR kit pathogen detection

Not only was the overall positive rate of pathogen detection on multiplex PCR significantly higher than that of routine microbiology, but the rate of multiple organisms detection was also markedly greater. As shown in , 155 respiratory (64.0%) and 66 non-respiratory (31.4%) samples were found to be positive for three or more pathogens on multiplex PCR, significantly greater than the eight respiratory (3.3%) and four non-respiratory (1.9%) samples evaluated using routine microbiology. Moreover, no specimens were positive for five or more pathogens on routine microbiology; multiplex PCR revealed 63 respiratory and 17 non-respiratory samples to be positive for five or more pathogens ().

Figure 2

Number of significant organisms detected per sample by routine microbiology or multiplex PCR kit. (A) Number of significant organisms detected per respiratory sample by routine microbiology or multiplex PCR kit. (B) Number of significant organisms detected per non-respiratory sample by routine microbiology or multiplex PCR kit. PCR, polymerase chain reaction.

The overall proportion of agreement (concordant positive/all positive results on routine microbiology) among multiplex PCR and routine microbiology for detection and identification of the aforementioned bacteria was found to be 95.8%. The sensitivity of multiplex PCR was >90% for most target bacteria; NPVs were >99%, especially for Pseudomonas aeruginosa, Klebsiella pneumoniae, Acinetobacter baumannii, Stenotrophomonas maltophilia, Enterococcus faecium, Enterococcus faecalis and Escherichia coli (). Specificity and PPV were lower as multiplex PCR detected more pathogens per sample and revealed more positive samples as compared to routine microbiology (). Test result consistency (Table S1) also revealed that multiplex PCR had higher sensitivity for detection of most bacteria and thus positivity coincident rates were relatively low. Streptococcus pneumoniae and group A streptococcus were not detected on routine microbiology but were found in 36 and four samples, respectively, on multiplex PCR; as such their relevant specificities were high but sensitivities could not be calculated (). However, lower sensitivity (33.3–59.6%) and higher specificity (93.9–100.0%) for detection of fungi and mycobacteria was noted using our multiplex PCR kit (). Although the number of fungi detected using multiplex PCR was close to that of routine microbiology, the concordant rate of positivity between the two methods was poor.

Table 5

Pathogen-specific performance of PCR tests as compared with routine microbiology as the gold standard in respiratory samples or non-respiratory samples

Pathogen	All samples (n=452)					Respiratory samples (n=242)					Non-respiratory samples (n=210)
	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)		Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)		Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)
Bacteria
Pseudomonas aeruginosa	97.1	72.8	40.2	99.3		98.1	61.7	43.1	99.1		93.8	83.6	32.6	99.4
Klebsiella pneumoniae	96.7	66.5	17.4	99.6		95.0	54.8	16.2	99.2		100.0	79.5	20.0	100.0
Acinetobacter baumannii	95.4	61.1	37.6	98.2		96.4	54.1	39.4	98.0		93.5	68.4	34.5	98.3
Staphylococcus aureus	87.5	81.6	21.4	99.1		88.2	76.8	22.7	98.8		85.7	86.9	18.8	99.4
Stenotrophomonas maltophilia	93.8	46.7	6.2	99.5		100.0	35.6	7.6	100.0		75.0	59.2	3.5	99.3
Streptococcus pneumoniae	NA	92.5	0	100.0		NA	86.9	0	100.0		NA	99.0	0	100.0
Group A streptococcus	NA	99.1	0	100.0		NA	98.3	0	100.0		NA	100.0	100.0	100.0
Haemophilus influenzae	83.3	93.1	14.3	99.8		83.3	89.6	17.2	99.5		NA	97.1	0	100.0
Enterococcus faecium	87.5	91.5	28.0	99.5		75.0	91.4	13.0	99.5		91.7	91.7	40.7	99.4
Enterococcus faecalis	90.0	84.5	11.8	99.7		100.0	83.3	7.1	100.0		85.7	85.9	17.6	99.4
Escherichia coli	100.0	89.8	17.0	100.0		100.0	87.2	6.3	100.0		100.0	92.9	33.3	100.0
Mycobacterium tuberculosis	33.3	100.0	100.0	95.8		25.0	98.8	66.7	93.3		0	100.0	NA	98.1
Atypical mycobacteria	0	100.0	NA	94.5		0	100.0	NA	91.4		NA	100.0	100.0	100.0
Fungi
Candida albicans	59.6	93.9	56.0	94.6		58.1	90.1	58.1	90.1		75.0	97.7	42.9	99.4
Mucor	NA	100.0	100.0	100.0		NA	100.0	100.0	100.0		NA	100.0	100.0	100.0
Aspergillus	50.0	100.0	100.0	96.7		50.0	100.0	100.0	96.6		NA	100.0	100.0	100.0
Penicillium marneffei	0	100.0	NA	99.3		NA	100.0	100.0	100.0		NA	100.0	100.0	100.0

PCR, polymerase chain reaction; PPV, positive predictive value; NPV, negative predictive value; NA, not available.

Discussion

In recent years, multiplex PCR has emerged as a simple, rapid and highly sensitive method for detection of respiratory pathogens (10,13). Here, we designed a multiplex fluorescent PCR kit for the identification of 17 respiratory pathogens simultaneously and evaluated its performance using 452 clinical samples obtained from inpatients. Sensitivity, specificity, PPV and NPV were analyzed for each PCR target considering routine microbiology as the gold standard.

Based on fluorescent PCR principles, we designed specific primers and Taqman probes for pathogen detection using a fluorescent PCR instrument. Our kit was designed to simultaneously detect 13 bacteria (Pseudomonas aeruginosa, Klebsiella pneumoniae, Acinetobacter baumannii, Staphylococcus aureus, Stenotrophomonas maltophilia, Streptococcus pneumoniae, group A streptococcus, Haemophilus influenzae, Enterococcus faecium, Enterococcus faecalis, Escherichia coli, Mycobacterium tuberculosis, and atypical mycobacteria), as well as four fungi (Candida albicans, Mucor, Aspergillus, and Penicillium marneffei) in one PCR panel. Compared with routine microbiology testing, which usually takes at least 48–72 h to obtain the culture result and requires higher sample quality, the multiplex PCR kit can complete the detection of 17 pathogens in only 80 min. Our kit is thus useful for the auxiliary diagnosis and epidemiological surveillance of respiratory pathogens.

The overall positive rate of pathogen detection (86.9% vs. 56.9%) and the detection rate of multiple organisms on multiplex PCR were both significantly higher than those of routine microbiology. Medical record review revealed detection of relevant pathogens on routine microbiology after significant illness progression; multiplex PCR was capable of detecting pathogens rapidly and with greater sensitivity. Clinical use of multiplex PCR thus offers a greater possibility of earlier diagnosis and treatment initiation.

In this study, clinical data analyzed were obtained from patients suffering nosocomial infections. In agreement with relevant prior literature (14), our routine microbiology findings revealed Acinetobacter baumannii, Pseudomonas aeruginosa and Klebsiella pneumoniae to have been the three predominant pathogens in respiratory samples, while Acinetobacter baumannii, Pseudomonas aeruginosa and Enterococcus faecium were found to have been the three predominant pathogens in non-respiratory samples. However, the pathogen most frequently detected by multiplex PCR in both respiratory and non-respiratory specimens was Stenotrophomonas maltophilia, and even it was found in more than half of respiratory specimens. As recent literature has reported the incidence of hospital-acquired Stenotrophomonas maltophilia infections to be on the increase (15), particularly among immunocompromised patients (16), its high rate of detection here warrants attention. Although Stenotrophomonas maltophilia is not highly virulent, it has emerged as an important nosocomial pathogen associated with a high mortality among patients suffering bacteremia that ranges from 14% to 69% (15,17,18). Inappropriate initial empirical antibiotic therapy is one of the risk factors for poor outcomes in the setting of Stenotrophomonas maltophilia pneumonia (18). Thus, the high sensitivity of multiplex PCR in Stenotrophomonas maltophilia identification is bound to assist clinicians in formulating more effective treatment regimens earlier and avoiding unreasonable antibiotic administration. Greater detection of Streptococcus pneumoniae on multiplex PCR is similarly important. Although one of the most common causes of pneumonia (19,20), the positive rate of Streptococcus pneumoniae detection on routine microbiological culture remains relatively low due to its fastidious nature (21,22). None of the 452 samples evaluated in this study were positive for Streptococcus pneumoniae on routine culture, but multiplex PCR successfully revealed Streptococcus pneumoniae in 36 samples. This advantage of multiplex PCR is bound to help clinicians more accurately establish diagnoses and perform relevant drug susceptibility testing.

Our results revealed that multiplex PCR has a high overall agreement with routine microbiology regarding the detection of bacteria commonly associated with pneumonia. Here, the detection sensitivity for most target bacteria was >90%, with NPVs >99%. One potential challenge of clinically implementing multiplex PCR use is that it detects additional bacteria not reported on routine microbiology (23). The number of bacteria detected by multiplex PCR is more than twice that detected using routine microbiology. Although this phenomenon underscores the high sensitivity of multiplex PCR, it remains necessary to distinguish possible causes of false positive results. First, some of the discordant results (i.e., multiplex PCR positive but culture-negative) found in this study might have been caused by administration of antimicrobial therapy prior to sampling (13). Because nucleic acids persist in vivo for some time, their presence does not necessarily signify ongoing infection. Another challenge posed by nucleic acid diagnostics for respiratory pathogen detection is the difficulty of distinguishing between colonization and infection (13,24). The presence of non-colonizing bacteria may certainly indicate an infectious etiology, but some respiratory pathogens such as Streptococcus pneumoniae and Candida albicans commonly colonize the upper respiratory tract and thus may present a diagnostic dilemma (25,26). It is thus important to establish a relationship between organism load and clinical context based on factors such as pathogen colonization characteristics to accurately determine etiology (27,28). In the further research plan, we will explore to distinguish infection from contamination or colonization by defining different-level cut-off values.

This study was not without its limitations. Due to insufficient relatively small number of samples, not enough cases of several uncommon pathogens were available for investigation. Thus, a larger-scale study is required to confirm our findings. Furthermore, since routine microbiology testing is not the true gold standard for pathogen detection, considering routine tests as the gold standard for comparison with multiplex PCR may obscure true multiplex PCR effectiveness. Future comparisons based on final diagnoses established by clinicians and documented patient outcomes are warranted.

Conclusions

In summary, our multiplex PCR kit designed to identify 17 respiratory pathogens possesses the advantages of high sensitivity and rapid turn-around. Thus, it is capable of aiding rapid identification of infectious etiologies. Importantly, the lower sensitivities of our kit regarding detection of mycobacteria and fungi require further study and enhancement. In addition, awareness of the clinical significance of potential false-positive results is critical when using our kit in practice. This, in turn, requires comprehensive judgment based on specific clinical manifestations and disease conditions of individual patients to accurately determine both etiology and treatment options (29). A large-scale, prospective study is required to confirm our findings and further explore the clinical effects of our multiplex PCR kit on patient management.

The article’s supplementary files as