Meta-analysis of sonicate fluid in blood culture bottles for diagnosing periprosthetic joint infection

Abstract

Introduction: Periprosthetic joint infection (PJI) is the most serious complication after arthroplasty, and the diagnosis of PJI is still challenging with modern medical technology. To improve the diagnostic rate, combined diagnostic methods are gradually beginning to be used to diagnose PJI. Sonication is one accurate way to diagnose PJI, but there is minimal research regarding the diagnostic value of sonicate fluid (SF) in blood culture bottles (BCB). Therefore, we evaluated this combined diagnostic method by meta-analysis.

Methods: We searched English publications in electronic databases regarding the use of sonicate fluid in blood culture bottles (SF-BCB) for diagnosing PJI, screened the literature according to inclusion criteria, assessed the quality of the selected literature, and collected information regarding SF-BCB.

Results: This meta-analysis includes 4 studies that evaluated SF-BCB for the diagnosis of PJI. The pooled sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR) and diagnostic odds ratio (DOR) are 0.85 (95% Confidence interval [CI], 0.77 to 0.91), 0.86 (CI, 0.81 to 0.91), 5.34 (CI, 3.13 to 9.11), 0.16 (CI, 0.06 to 0.48) and 39.01 (CI, 9.04 to 168.35), respectively. The area under the curve (AUC) of the summary receiver operating characteristic (SROC) is 0.9186 (standard error, 0.0205).

Conclusion: SF-BCB has great value for the microbiological diagnosis of PJ, especially for patients with prior antibiotic treatment.

Introduction

Early diagnosis and bacterial detection during PJI are the key to successful treatment ¹. However, because there is no gold standard to diagnose the existence of biofilms on the surface of a joint prosthesis, the diagnosis of PJI is difficult ². So far, comprehensive analyses with some effective diagnostic methods seem to be the best options for diagnosing PJI. Sonicate fluid culture (SFC) is a valuable assessment tool for PJI. SFC dislodges the biofilm on the implant surface through low frequency ultrasound, which preserves the activity of the microorganism ³. The high sensitivity and specificity of SFC at different locations with PJI have been reported ^{4, 5}. In knee prostheses, the sensitivity and specificity are 90.6% and 100%, respectively, and in hip prostheses, the sensitivity and specificity are 87% and 100%, respectively. In elbow prostheses, the sensitivity and specificity are 89% and 100%, respectively.

However, at the early stage of PJI, sonication might be not helpful for diagnosis ^{4, 6}. Prieto-Borja L et al. ⁷ found that SFC is not superior to periprosthetic tissue culture and synovial fluid culture in the early stages of PJI. However, the sensitivity of SFC is significantly increased if diagnosis is delayed. In order to improve the accuracy of PJI diagnose, PCR, microcalorimetry and SF-BCB techniques have been published ^{8, 9, 10}.

The culture result of synovial fluid culture in BCB is better than growth on an agar plate, so synovial fluid culture in BCB is already used regularly ¹¹. However, studies regarding the use of SF-BCB for the diagnosis of PJI are uncommon ^{10, 12-15}, and thus, the diagnostic validity of SF-BCB is still unclear. Therefore, we performed a meta-analysis to evaluate the effectiveness of SF- BCB for diagnosing PJI.

Methods

Search strategy

We searched English publications from the electronic databases of PubMed, Medline, and Web of Science until 30 April 2018 using the following medical subject headings (MeSH) or keywords: “periprosthetic joint infection OR prosthetic joint infection OR orthopaedic implant infection”, “sonication fluid OR sonicate fluid OR sonication”, and “blood culture vials OR blood culture bottles OR blood culture system”. We also reviewed the references in eligible studies and the studies that cited the eligible studies in Google Scholar.

Inclusion criteria

The selected articles had to meet the following inclusion criteria:

1. The diagnosis of PJI is based on criteria from the Musculoskeletal Infection Society (MSIS), the European Bone and Joint Infection Society (EBJIS) and the Infectious Diseases Society of America (IDSA) ¹⁷; or strong evidence of infection, such as the presence of a sinus tract, purulence around the prosthesis, acute inflammation in periprosthetic tissue, or microbial growth in the synovial fluid, periprosthetic tissue samples, or SF.

2. The numbers of true-positive (TP), true-negative (TN), false-positive (FP) and false-negative (FN) values are clear, and the computed results of sensitivity and specificity are identical to the article described.

3. Only human studies in English are included.

4. Each of the selected studies contains at least 15 cases.

Quality assessment

Two reviewers independently screened the retrieved clinical studies for inclusion criteria, extracted data from all studies, and conducted a quality evaluation. The methodological quality of the selected studies was evaluated by using the diagnostic accuracy studies tool (QUADAS-2) ¹⁶, which is specifically developed for systematic reviews focusing on diagnostic accuracy. When confronted with disagreements, a third reviewer adjudicated.

Data extraction

The following information in the articles was collected: first author, year of publication, country, enrolment period, number of infected cases, location, whether antibiotics were used, diagnostic criteria or method, sonication procedure, and the sensitivity and specificity of SF-BCB.

Statistical analysis

For the analysis of the diagnostic value of SF-BCB, all statistical analysis are conducted using Meta-Disc software (version 1.4). The specificity, sensitivity, PLR, NLR, DOR and AUC of SROC are assessed. The percentage of the total variation across studies is described by the I² statistic, which indicates the existence of significant heterogeneity when the value exceeds 50%. The value of I² ranges from 0 to 100%, with 0 implying no observed heterogeneity and larger values indicating increasing heterogeneity. The analysis of heterogeneity between studies is conducted using the χ² test. If there is no significant heterogeneity across studies (P>0.1, I²≤50%), the analysis is performed using a fixed-effects model; otherwise, the random effects model (P≤0.1, I²>50%) is used.

Results

Ninety-six articles are identified by the literature search. After literature review of the title, the abstract and the full text of the articles, 4 studies are included in the diagnostic meta-analysis ^{10, 12-14}, including 3 articles from European countries (Germany, Spain, and Greece) and 1 study from China. The flow diagram is shown in Figure 1. As a result, 118 prosthetic joint infection cases are included in the meta-analysis. The detailed characteristics of each study are described in Table 1. The graphical summary of the methodological assessment based on the QUADAS-2 quality assessment for the recruited studies in the meta-analysis is illustrated in Figures 2 and 3.

Figure 1.

Flow diagram of the included studies

Table 1.

The detail characteristics of the included studies

Reference	Country	Enrollment period	Infected Cases	Location	Received antibiotics	Solution for implant	Diagnostic standard†	Vortex	Sonication	Centrifugation	Sensitivity	Specificity
Viktor Janz (2013) 13	Germany	October 2010 - March 2011	23	hip and knee	No	Ringer's solution	P, H, IOF, M	30 seconds	3 min	No	91%	81%
María Eugenia Portillo (2015) 12	Spain	June 2013 - December 2013	18	joint prosthesis	YES	NA‡	C, P, H, IOF	30 seconds+30 seconds*	1 min	No	100%	100%
Hao Shen (2015) 10	China	August 2011 and May 2014	50	Knee 20,hip 30	YES	Ringer's solution	P, H, IOF, M	30 seconds+30 seconds	10 min	10 min	88%	87%
Antonios Stylianakis (2018) 14	Greece	September 2011 and April 2015	27	hip and knee	YES	Ringer's solution	P, H	30 seconds+30 seconds	5 min	30 seconds	62.96%	83.91%

* vortex before and after sonication †C, clinical signs of infection; P, presence sinus tract or purulence around the prosthesis; H, histological examination; IOF, intraoperative finding; M, microbiological or laboratory examination. ‡ NA, not available

Figure 2.

Methodological quality summary

Figure 3.

Methodological quality graph

Significant heterogeneity is found in the sensitivity (I² = 79.9%), specificity (I² = 67.7%), PLR (I² = 46.3%), NLR (I² = 78.2%) and DOR (I² = 72.7%). Thus, the random-effects model is used. No heterogeneity is found in PLR (I² = 46.3%), and the fix-effects model is applied. The pooled sensitivity, specificity, PLR, NLR, and DOR estimates for the detection of PJI using SF-BCB are 0.85 (95% CI, 0.77 to 0.91), 0.86 (CI, 0.81 to 0.91), 5.34 (CI, 3.13 to 9.11), 0.16 (CI, 0.06 to 0.48), and 39.01 (CI, 9.04 to 168.35), respectively (Figure 4, Figure 5, Figure 6, Figure 7 and Figure 8). The SROC plot shows the summary sensitivity and specificity under 95% confidence and prediction regions, with an AUC of 0.9186 (standard error, 0.0205) (Figure 9).

Figure 4.

Forest plots of sensitivity of SF-BCB for PJI diagnosis

Figure 5.

Forest plots of specificity of SF-BCB for PJI diagnosis

Figure 6.

Forest plots of PLR of SF-BCB for PJI diagnosis

Figure 7.

Forest plots of NLR of SF-BCB for PJI diagnosis

Figure 8.

Forest plots of DOR of SF-BCB for PJI diagnosis

Figure 9.

SROC of SF-BCB for PJI diagnosis

Discussion

Various diagnostic methods of PJI have been published, such as erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), synovial fluid leukocyte count and neutrophil percentage, histopathological examination, periprosthetic tissue culture, and sonication ¹⁷. Discovering pathogenic bacteria is undoubtedly the core, but the range of diagnostic accuracy in traditional cultures is 39% - 70%. Moreover, culture-negative results present a challenge for diagnosing PJI. The cure rate of culture-negative PJI is low (69.2%), and serious complications such as amputation or death might occur. The major reason for culture-negative results in PJI is antibiotic usage before sample collection ²⁰. However, sonication could improve diagnostic accuracy in those patients that have already used antibiotics.

Sonication dislodges adherent bacteria from prostheses. This easy and valuable technology is used routinely in diagnosing PJI, and it is also included in the diagnostic criteria of the EBJIS ¹⁸. One meta-analysis of 16 SFC studies ¹⁹ shows the sensitivity is 79% (95% CI = 0.76 - 0.81), and the specificity is 95% (CI = 0.94 - 0.96). SFC is more sensitive than tissue culture, even in patients that received antibiotic therapy up to 14 days prior.

The improved combination method of SF-BCB for the diagnosis of PJI has been reported in recent years, and the sensitivity of SF-BCB is greater than conventional SFC (100% versus 87%). In the case of previous antibiotic treatment before sample collection, the culture sensitivity of SF-BCB is higher than that of SFC ¹². SF-BCB can detect more positive growth than conventional SFC (101 cases versus 51 cases). SF-BCB can reduce culture time, as well; the average culture time in SF-BCB is shorter than that of SFC (2.9 days versus 4.2 days) ¹⁵.

The diagnostic methods of SF-BCB and of synovial fluid into BCB have been reported in the diagnosis of PJI ¹⁰. The sensitivity of SFC and of synovial fluid culture in BCB are 88% and 64%, respectively. In the cases of patients who received antibiotics within the 14 days before surgery, the percentage of positive cultures in SF and synovial fluid are 81% and 52%, respectively. In the cases of patients without antibiotic treatment, the percentage of positive cultures in SF and synovial fluid are 93% and 72%, respectively. SF-BCB has a better capability of detecting microorganisms than synovial fluid in BCB, but the specificity of synovial fluid is better than SF-BCB (98% versus 87%). Another study also states that the percentage of positive cultures in SFC-BCB is higher than in synovial fluid in BCB (44% versus 22%), but the average duration of positive growth in synovial fluid is shortened to 1.8 days, compared with 2.9 days in SF ¹⁵.

Compared to traditional tissue culture, SF-BCB is more sensitive but less specific in patients with previous antibiotic therapy (62.96% to 40.74%, 83.91% to 96.55%) or without previous antibiotic therapy (91% to 75%, 81% to 100%) ^{13, 14}. In recent research, the sensitivity and specificity of tissue culture in BCB are lower than in conventional SFC; whether in the hip and knee (65.1% to 69.8%, 98% to 100%) or in the shoulder and elbow (72.2% to 88.9%, 87.5% to 100%) ²¹.

With the development of molecular biological techniques, PCR has been applied to the diagnosis of PJI. Its advantages include short culture time (< 5 hours), automation, high sensitivity and specificity, and the ability to easily distinguish PJI from aseptic loosening ^{9, 22}. Comparing to SF-BCB, the sensitivity of PCR is 59.26% and the sensitivity of SF-BCB is 62.96%, respectively, and the specificity of PCR and SF-BCB are close ¹⁴.

In our study, the pooled sensitivity is 0.85 (CI = 0.77 - 0.91), the specificity is 0.86 (CI = 0.81 - 0.91), the PLR is 5.34 (CI = 3.13 - 9.11), the NLR is 0.16 (CI = 0.06 - 0.48), and the DOR is 39.01 (CI = 9.04 - 168.35). The AUC value of the SROC curve is 0.9186. SF-BCB is a valuable microbiological diagnosis for PJI with high sensitivity, even in patients who received antibiotic treatment within 14 days of surgery. BCB allows the automated detection of microbial growth technology, and it reduces 60.1% of the work hours for laboratory staff compared with conventional techniques. It is also a cost-saving method ²³.

This study has some limitations. First, there are only 4 articles with cases included in our meta-analysis, although we set the standard at a minimum of 15 cases in each study. Second, in three-quarters of the studies, some patients received antibiotic treatment within 14 days of surgery. This could impact culture results, but it also suggests that SF-BCB has an acceptable sensitivity and specificity after antibiotic therapy. Third, in our study, the diagnostic standards are different, and this might negatively influence diagnostic accuracy. Renz N et al ²⁴ use three definition criteria to assess the value of Alpha defensin lateral flow (ADLF) test for diagnosing PJI, the results of sensitivity for ADLF test are 84%, 67%, 54% with different definition criteria. Fourth, sonication has a non-standard methodology. Some research methods include vortex and sonication (without centrifugation) ^{13, 25}. Some scholars use vortexing, sonication and centrifugation as a standard method ²⁶. Vortexing or centrifugation with sonication could improve the culture result ^{26, 27}. A subgroup of meta-analyses ²⁸ also shows that the sonication method with vortexing or centrifugation has better culture results than without vortexing or centrifugation.

Conclusions

The results of the meta-analysis show that SF-BCB has great value for the microbiological diagnosis of PJI. Its capability of detecting pathogens is better than conventional tissue culture, conventional SF, SF-PCR and synovial fluid in BCB, especially for patients with previous antibiotic treatment.

Abbreviations

PJI

Periprosthetic joint infection

SF

sonicate fulid

BCB

blood culture bottles

SF-BCB

sonicate fluid in blood culture bottles

PLR

Positive likelihood ratio

NLR

Negative likelihood ratio

DOR

Diagnostic odds ratio

CI

Confidence interval

AUC

Area under the curve

SROC

Summary receiver operating characteristic

SFC

Sonicate fluid culture

MeSH

medical subject headings

TP

true-positive

TN

true-negative

FP

false-positive

FN

false-negative

ESR

erythrocyte sedimentation rate

CRP

C-reactive protein

EBJIS

European bone and joint infection society

MSIS

Musculoskeletal Infection Society

IDSA

Infectious Diseases Society of America

ADLF

Alpha defensin lateral flow.