Multicentre Experience on the Diagnostic Yield of PCR Testing for Bacterial Pathogens in Sterile Site Infections

Abstract

Background:

Infections in sterile body sites are serious despite their low incidence. Accurate diagnosis is crucial for effective antimicrobial management. This study assessed the diagnostic yield of multiplex polymerase chain reaction (PCR) in identifying bacterial pathogens in sterile site specimens other than blood and cerebrospinal fluid.

Methods:

Bacterial pathogen PCR panels were independently developed and validated by the laboratories at the Hamilton Regional Laboratory Medicine Program (HRLMP) in Canada and Sidra Medicine in Qatar. Retrospective culture and PCR data for the periods of July 2022 to November 2023 and September 2021 to February 2023 were extracted from the laboratory information systems of HRLMP and Sidra Medicine, respectively. The diagnostic yield of PCR between different groups was compared using the McNemar test or chi-square test.

Results:

Validation studies showed 100% sensitivity for PCR assays in both laboratories, with varying specificity due to the detection of additional pathogens by PCR. Combining post-implementation data from both laboratories, 38.7% of 512 specimens were PCR-positive for target organisms, compared to 6.1% by culture. While the diagnostic yield of PCR was significantly higher than that of culture in both adult and paediatric populations (p < .001), HRLMP data indicated a significantly higher diagnostic yield of PCR in the paediatric population compared to adults (64.7% versus 17.4%; p < .001). The most commonly PCR-detected pathogens were Streptococcus pneumoniae, Streptococcus pyogenes, and Staphylococcus aureus, with pleural fluid being the most frequently positive specimen type.

Conclusion:

This study supports using PCR alongside culture to enhance pathogen detection and improve the management of sterile site infections, particularly in paediatric patients.

Introduction

The morbidity, mortality, and health care costs associated with infections of normally sterile body sites, such as blood, body fluids, and tissues, are substantial despite their low incidence rates relative to other infections (1, 2, 3). Treatment, prognosis, and outcomes of these infections heavily depend on identifying the underlying cause of the disease. While bacterial infection is the predominant cause of such infections, the diagnostic yield of routine bacterial culture may be poor due to antibiotic treatment initiation before specimen collection. Syndromic molecular test panels are now widely available for detecting pathogens in respiratory infections, gastrointestinal infections, and central nervous system infections such as meningitis and encephalitis (4,5). However, experience with sterile site specimens for infectious syndromes such as endocarditis, pleural effusion, and bone and joint infections is limited. In this study, we assessed the diagnostic yield of laboratory-developed polymerase chain reaction (PCR) for bacterial pathogens on sterile site specimens other than blood and cerebrospinal fluid (CSF) in adult and paediatric populations, utilizing data from two different laboratories.

Materials and Methods

Bacterial pathogen PCR panels were independently developed by the laboratories at Hamilton Regional Laboratory Medicine Program's (HRLMP's) regional virology and molecular laboratory in Hamilton, Canada (serving children and adults in five acute care hospitals) and at Sidra Medicine, a tertiary care children's and women's hospital in Doha, Qatar. The HRLMP PCR assay includes Streptococcus pneumoniae, Staphylococcus aureus, Streptococcus pyogenes, and the Streptococcus anginosus group as pathogen targets. Additionally, specimens from pediatric patients were tested for Kingella kingae. The Sidra PCR panel includes all these pathogens except the Streptococcus anginosus group, but it includes additional pathogens: methicillin-resistant Staphylococcus aureus (MRSA), Streptococcus agalactiae, Haemophilus influenzae, Klebsiella pneumoniae, Pseudomonas aeruginosa, Escherichia coli, and Acinetobacter baumannii (Supplemental Table 1). The target pathogens in each PCR panel were selected by a group of medical microbiologists and infectious disease physicians based on current knowledge of the most common bacterial species implicated in infections of normally sterile sites, such as pleural effusion and empyema, acute osteomyelitis, infectious arthritis, and endocarditis (6, 7, 8, 9). The primer and probe sequences for different PCR assays are listed in Supplemental Tables 2 and 3.

PCR testing was performed according to the standard operating procedures in the respective laboratories. Briefly, total nucleic acids were extracted from specimens on the bioMerieux easyMAG (bioMerieux) at HRLMP or the EZ1 Advanced XL (Qiagen) at Sidra. The HRLMP assay was performed by two multiplex PCRs on a Rotorgene system (Qiagen). At Sidra Medicine, PCR testing was performed in an array format (10) with a combination of eight single-plex and multiplex reactions on the ABI 7500 Fast (Thermo Fisher) or QuantStudio 5 (Thermo Fisher) real-time PCR systems.

At HRLMP, for all pathogens except K. kinage, validation study included 221 specimens (138 synovial fluid and 83 pleural fluid) that were tested by standard bacteriological culture. An additional 39 synovial fluid specimens were assessed for K. kingae only. Specimens used for validation were either culture positive for one or more pathogens targeted by the PCR assay, or culture positive for other bacterial species not targeted by PCR, or negative for bacterial growth after 5 days of incubation. At Sidra, validation study included a total of 195 specimens that included original clinical specimens (n = 53), external quality assessment (EQA) samples (n = 7), and simulated specimens (n = 135) prepared by making a suspension of clinical isolates (both target and non-target bacterial species) in Tris-EDTA (ethylenediaminetetraacetic acid) buffer. Original clinical specimens include pleural fluid (n = 16), synovial fluid (n = 7), tissue (n = 8), peritoneal fluid (n = 4), abscess fluid (n = 5), and other fluids (n = 13). Apart from the bacterial species that are targeted by PCR, the non-target bacterial species assessed include Staphylococcus capitis, Staphylococcus ludgensis, Stenotrophomonas maltophilia, Streptococcus lutiensis, Streptococcus oralis, Streptococcus constellatus, Streptococcus anginosus, Serratia liquefaciens, Proteus mirabilis, Micrococcus luteus, Enterobacter cloacae, Enterococcus faecalis, Enterococcus faecium, Candida albicans, Candida glabrata, and coagulase-negative Staphylococcus aureus.

After implementation, all sterile site specimens (n = 232) underwent testing by culture and PCR at Sidra Medicine. At HRLMP, only specimens negative by culture after 24 hours with special requests from clinicians were subjected to PCR (n = 280). All cultures were performed according to the standard operating procedures established in the respective laboratories and followed the guidelines of the Clinical Laboratory Standards Institute (CLSI). Retrospective data for pleural fluid, synovial fluid, tissue, and other body fluids that underwent simultaneous testing by culture and PCR during the periods of July 2022 to April 2024 and September 2021 to April 2024 were extracted from the laboratory information systems of HRLMP and Sidra, respectively. The study was performed as part of a laboratory quality improvement project. Only laboratory data along with patient age and gender were collected, without any patient identifying information. The diagnostic yields of PCR and culture in different populations were assessed using the chi-squared test, and the diagnostic yields of PCR versus culture on matched samples were assessed using the McNemar test. Independent group t test (two-tailed) was used to compare mean cycle threshold (Ct) values between two different groups of specimens.

Results

The performance characteristics of the PCR assays for different organisms were established by independent validation studies in both laboratories (Supplemental Table 1). Briefly, the limit of detection of PCR for different target organisms ranged from 2.1 × 10³ to 2.7 × 10³ genome equivalents per mL at HRLMP and from 4.6 × 10¹ to 2.1 × 10⁴ genome equivalents per mL at Sidra. The intra-assay and inter-assay precision were determined to be 100% in both laboratories. With respect to standard culture methods, the sensitivity of the PCR assays for detecting different organisms was 100% in both laboratories. However, the specificity of the PCR assays against culture ranged from 78% to 99.1% at HRLMP and from 83% to 100% at Sidra, depending on the target organism. All discrepant results were verified by a second molecular test developed in the laboratory (data not shown). All culture-positive test results were concordant with PCR results. However, PCR detected a large number of additional pathogens compared to culture, particularly in cases of S. pneumoniae, S. aureus, and S. pyogenes (Supplemental Table 4).

At Sidra, a total of 232 specimens from 203 paediatric patients (mean age 5.9 [SD 4.3] years, 45% female) were assessed after implementation of multiplex PCR for bacterial pathogens. At HRLMP, a total of 195 specimens from 172 adult patients (mean age 67 [SD 16] years, 46% female) and 85 specimens from 75 paediatric patients (mean age 6.0 [SD 4.1] years, 48% female) were assessed. Overall, combining data from two centres, 31 out of 512 specimens from 450 patients (6.1%) were positive by culture, and 198 (38.7%) were positive by PCR for the target organisms. While a large number of additional pathogens were detected by PCR, all specimens that tested positive for the target pathogens by culture were also identified by PCR. Other organisms not targeted by PCR grew in a total of 19 specimens (3.7%) out of 512 specimens, combining data from both sites. At Sidra, where all samples were simultaneously tested by both methods, only 9 (3.9%) out of 232 specimens were positive for PCR non-targeted pathogens by culture. These bacteria include coagulase-negative staphylococci (CoNS); viridans group Streptococci; groups B, C, and G Streptococci; Candida species; and anaerobic organisms.

The diagnostic yield of PCR was significantly higher than that of culture in all populations (p < .001) (Table 1) despite the differences in the testing algorithms. At Sidra Medicine, where all specimens were simultaneously tested by PCR and culture, 46.9% of specimens were positive for one or more pathogens by PCR, compared to only 9.1% by culture (p < .001) (Table 1). At HRLMP, diagnostic yield of PCR on specimens that were negative by culture at 24 hours was significantly higher in the paediatric population compared to adults (64.7% versus 17.4%; p < .001). However, during the same period, the diagnostic yields of culture were not significantly different (p = .98) for adult versus paediatric populations (data not shown). The most common pathogens detected by PCR were S. pneumoniae (49.5% of all PCR-positive results), S. pyogenes (21.8% of all PCR-positive results), and S. aureus (13.6% of all PCR-positive results) (Table 2) and pleural fluid was the most common specimen type where a pathogen was detected (62.6% of all PCR-positive results) (Table 1). Using a subset of PCR-positive specimens, we also assessed the Ct value distribution for culture-positive and culture-negative specimens, irrespective of pathogen type, age group, or testing site. The mean Ct value for target pathogens in culture-positive specimens was 27.8 (SD 5.8), compared to 32.8 (SD 4.9) in culture-negative specimens, a difference that was statistically significant (p < .01) by an independent samples t test (Supplemental Figure 1).

Table 1:

Diagnostic yield of PCR compared to culture in adult versus paediatric patients by specimen type

	Percentage positive (no.)
	Adults (HRLMP)		Children (HRLMP)		Children (Sidra)		Total by specimen
Specimen	Culture	PCR	Culture	PCR	Culture	PCR	Culture	PCR
Pleural fluid	4 (25)	36 (25)	9.1 (33)	84.8 (33)	10.6 (132)	65.9 (132)	9.5 (190)	65.2 (190)
Synovial fluid	0 (41)	9.75 (41)	0 (12)	25 (12)	3.8 (26)	30.8 (26)	1.3 (79)	19 (79)
Tissue*	0 (37)	10.8 (37)	0 (6)	16.7 (6)	6.25 (16)	12.5 (16)	1.7 (59)	11.9 (59)
Other†	6.5 (92)	18.5 (92)	0 (34)	67.6 (34)	8.6 (58)	20.7 (58)	6 (184)	28.3 (184)
Total by site and age group	3.6 (195)	17.4 (195)	3.5 (85)	64.7 (85)	9.1 (232)	46.9 (232)	6.1 (512)	38.7 (512)

^*Most tissues samples at HRLMP are from shoulder, ankle, knee, and hip surgeries, and tissue samples at Sidra are lymph node, bone, ankle, or splenic biopsies

^†Other specimen type includes peritoneal fluid, pericardial fluid, abscess, etc

PCR = Polymerase chain reaction; HRLMP = Hamilton Regional Laboratory Medicine Program

Table 2:

Diagnostic yield of PCR compared to culture in adult versus paediatric patients by pathogen type

	Adult (HRLMP), no. (%); n = 195		Children (HRLMP), no. (%); n = 85		Children (Sidra), no. (%); n = 232
Organism	Culture	PCR	Culture	PCR	Culture	PCR
S. pneumoniae	1 (0.5)	5 (2.6)	0	30 (35.3)	5 (2.1)	67 (28.8)
S. pyogenes	0 (0)	7 (3.1)*	3 (3.5)	21 (24.7)	4 (1.7)	17 (7.3)
MSSA	3 (1.0)	10 (5.1)	0	2 (2.4)	5 (2.2)	9 (3.9)
MRSA†	-	-	-	-	4 (1.7)	7 (3.0)
S. anginosus gr.	3 (1.5)	12 (5.6)	0	1 (1.2)	-	-
S. agalactiae	-	-	-	-	0 (0)	0 (0)
K. kingae	-	-	0	1 (1.2)	0 (0)	2 (0.9)
H. influenzae	-				1 (0.4)	7 (3.0)
E. coli		-	-	-	4 (1.7)	3 (1.3)
K. pneumoniae	-				2 (1.7)	3 (1.3)
P. aeruginosa		-	-	-	1 (0.4)	2 (0.9)
Total	7 (3.6)	34 (17.4)	3 (3.5)	55 (64.7)	*28 (12.1)	*117 (50.4)

^*Pathogen counts are higher than number (no) of PCR-positive specimens because some samples were positive for more than one target

^†MRSA and MSSA were separately counted

MRSA = Methicillin-resistant Staphylococcus aureus; MSSA = Methicillin-sensitive Staphylococcus aureus

Discussion

Proper identification of causative pathogens is particularly important for appropriate antimicrobial management of invasive infections in sterile body sites. While most common organisms infecting sterile body fluids are not difficult to culture (11), in many cases it is challenging to recover viable organisms for identification because the patient has already been empirically treated with broad-spectrum antibiotics. Molecular methods for detecting microbial pathogens are increasingly replacing or complementing conventional methods in clinical settings. Studies have shown the benefits of implementing multiplex molecular panels in detecting pathogens associated with various infectious disease syndromes, including faster results, higher detection rates, shorter hospital stays, and reduced antibiotic use (12).

Our results suggest that the detection rate of bacterial pathogens is greatly increased by PCR, particularly in the paediatric population. Among all the specimens tested, detection rates were highest in pleural fluid specimens and for pathogens such as S. pneumoniae, S. pyogenes, and S. aureus across all populations. This suggests that the application of a multiplex bacterial PCR panel could be particularly useful in managing paediatric pleural effusion and empyema by enabling the early implementation of appropriate antimicrobial therapy.

One limitation of our study is that the test referral criteria at Sidra and HRLMP were not identical. While all sterile site specimens at Sidra were simultaneously tested by PCR and culture, at HRLMP, only culture-negative samples after 24 hours were referred for PCR testing. As a result, the bacteriological culture positivity rate at Sidra (9.1%) was higher than in both the adult and paediatric populations at HRLMP (3.6% and 3.5%, respectively). Despite this, the difference between culture and PCR detection was substantial across all populations.

Another limitation is that the multiplex PCR panels used at the two sites were different. Nevertheless, we observed very similar results when comparing bacterial PCR and culture for patients with sterile site infections in two independent populations. In this study, we were unable to assess the impact of pre-test antibiotic treatment on test results or the clinical value of utilizing PCR for these infections. However, a follow-up study is currently underway to investigate these research questions.

While bacteriological culture is an agnostic test that can detect most viable bacteria present in a specimen, multiplex PCR is limited to detecting only selected pathogens. Despite this, our culture data from Sidra indicate that only a small proportion of these infections (3.9%) were caused by bacterial pathogens not targeted by the PCR assay. The low proportion of these organisms may be related to the known epidemiology of these infections (6-9) but could also be due to the poor sensitivity of culture. In such cases, agnostic molecular tests, such as 16S ribosomal RNA (rRNA) gene sequencing, may improve the diagnostic yield in sterile site infections. However, it should be noted that quantitative PCR (qPCR)–based assays have higher sensitivity, are much easier to perform in a clinical setting, and offer faster turnaround times compared to 16S rRNA sequencing. A comparison of the diagnostic yield between qPCR-based multiplex assays and 16S rRNA gene sequencing in sterile site infections may provide empirical evidence to support the optimal diagnostic approach for these infections.

In conclusion, our results suggest that PCR testing should be used alongside culture to improve the diagnosis and management of sterile site infections, especially those caused by predominant streptococcal and staphylococcal pathogens. PCR testing for other pathogens may be guided by patient demographics and epidemiological data specific to the patient populations.

Funding Statement

No funding was received for this article.

Contributors:

Conceptualization: MR Hasan, P Tang, P Jayaratne, A Perez-Lopez, D Leto, D Yamamura, M Smieja; Methodology: A Sharma, P Jayaratne, C Rutherford, S Sundararaju, M Suleiman; Validation: A Sharma, P Jayaratne, C Rutherford, S Sundararaju, M Suleiman; Formal Analysis: MR Hasan; Investigation: A Sharma, P Jayaratne, C Rutherford; Resources: A Perez-Lopez, D Leto, D Yamamura, M Smieja; Data Curation: MR Hasan; Writing – Original Draft: MR Hasan; Writing – Review & Editing: MR Hasan, P Tang, P Jayaratne, C Rutherford, A Perez-Lopez, D Leto, D Yamamura, M Smieja; Supervision: M Suleiman, M Smieja.

Ethics Approval:

Ethics approval was not required for this article.

Informed Consent:

N/A

Registry and the Registration No. of the Study/Trial:

N/A

Data Accessibility:

N/A

Funding:

No funding was received for this article.

Disclosures:

The authors have nothing to disclose.

Peer Review:

This article has been peer reviewed.

Animal Studies:

N/A