Improving the microbiological diagnosis of fracture-related infection and prosthetic joint infection through culturing sonication fluid in Bactec blood culture bottles

Abstract

Background

Sonication of surgically removed implants appears to optimize the microbiological diagnosis in orthopedic implant-associated infections (OIAI). However, reports of infection with negative cultures can still reach high rates. A study evaluating the inoculation of sonication fluid into blood culture bottles (SFBCB) in patients with fracture-related infection (FRI) and prosthetic joint infection (PJI) is necessary. This study compared the accuracy SFBCB over the conventional sonication fluid cultures (CSFC) and tissue culture (TC).

Methods

Consecutive patients who underwent implant removal surgeries due to any reason had their implants sonicated according to standardized method. Definitions of PJI and FRI were based upon criteria by European Bone and Joint Infection Society (EBJIS) and Metsemakers, respectively. Between three to five intraoperative tissue samples were processed. The implant`s sonication fluid was seeded onto sheep blood agar, chocolate agar, thioglycolate broth and on tryptic soy broth for CSFC, while was also inoculated into blood culture bottles and incubated in the automated system during 5 days for SFBCB.

Results

Overall, 74 patients were analyzed, of which 57 with OIAI (48 FRI and 09 PJI) and 17 aseptic failures (03 arthroplasties and 14 osteosynthesis). Interestingly, SFBCB demonstrated significantly higher sensitivity compared to CSFC (96.5% [95% CI, 88–100] vs. 78.9% [95% CI, 66–89], p = 0.004), and to TC (96.5% [95% CI, 88–100], vs. 57.9% [95% CI, 44–71], p < 0.001), whereas there were no significant differences in specificity between the three methods.

Conclusion

In comparison to CSFC and TC, SFBCB improved sensitivity for diagnosing OIAI without compromising specificity.

Introduction

Prosthetic joint infection (PJI) affects approximately 1 to 2% of primary arthroplasties and 10% of revision arthroplasties, while fracture-related infection (FRI) occurs in 1 to 5% of closed fractures and over 30% of open fractures [1, 2]. These statistics underscore the profound social, clinical, and economic implications associated with these complications. Additionally, these conditions are expected to increase over the next decades due to surgeries performed on high-risk patients, an aging population, comorbidities, and immunosuppression potentially resulting in functional impairment and long-term disabilities [3]. Currently, adhering to the criteria established by the European Bone and Joint Infection Society (EBJIS) and Metsemakers is likely most appropriate to the diagnosis of PJI and FRI, respectively [4, 5]. In common, these definitions highlight the importance of a meticulous microbiological diagnostic workup, which encompasses precise culture of tissue samples and sonication of explanted hardware. Furthermore, they establish thresholds to enhance specificity, and emphasize the timing of culture incubation, ensuring detection of microorganisms with low virulence, such as Cutibacterium acnes. For instance, a recent published study concluded that a cutoff point for sonicate fluid (SF) culture positivity of ≥ 20 colony forming unit /10 mL is suitable for hip and knee PJI diagnosis [6].

Despite its critical importance, intra-operative tissue culture (TC) proves inadequate in achieving microbiological diagnosis in up to 60% of cases, particularly among patients using systemic antibiotics and those with chronic infections [7–9]. This challenge is closely associated to the prevalence of biofilm-forming pathogens in orthopedic implant-associated infections (OIAI) [10]. The resilient nature of biofilms poses a formidable obstacle to conventional diagnostic methods, as it constitutes a physical barrier that harbors bacteria in a sessile form (with slow growth and low virulence), protecting against the immune system and antibiotic action. Indeed, in late-stage PJI, the greatest bacterial inoculum is likely concealed within biofilm [11]. Therefore, the utilization of sonication promotes the detachment of biofilm from the implant surface, facilitating microbiological detection and enhancing culture sensitivity [9, 12, 13]. Reports have indicated a heightened diagnostic sensitivity, reaching up to 80%, when sonication is incorporated to the conventional microbiological workup particularly in cases involving recent antibiotic usage [14–17]. A recent study published by our group underscores the significance of conducting a minimum of three TC in conjunction with sonication culture, resulting in a combined sensitivity of 98.1% and specificity of 95.3%, utilizing EBJIS criteria for PJI diagnosis [9].

Despite the enhanced microbial recovery, the overall yield of sonication fluid culture may still be susceptible to the effects of diminished microbial inoculum [17]. Studies have detected microbial DNA via Polymerase Chain Reaction in synovial fluid, even in cases where cultures yield negative results, highlighting potential limitations of culture media in bacterial recovery [18, 19]. While a limited number of studies have shown that inoculating sonication fluid into blood culture bottles (SFBCB) significantly improved the diagnosis of OIAI, relevant questions remain unanswered [17]. This involves assessing the diagnostic validity of SFBCB in patients with FRI and applying recently published diagnostic criteria for PJI and FRI [4, 5]. In this context, this present study sought to assess the performance of culturing sonication fluid in conventional agar, thioglycolate broth (TG), and tryptic soy broth (TSB) (CSFC) in comparison to SFBCB and TC for diagnosing OIAI. While previous research primarily concentrates on assessing the diagnostic precision of CSFC and SFBCB in PJI patients, our investigation takes a broader perspective by extending our analysis to encompass patients with FRI.

Materials and methods

Study design and population

The study took place from October 2021 to March 2023 at the Department of Orthopedics and Traumatology of two large tertiary academic hospitals in São Paulo, Brazil. During this period, consecutive patients who underwent total or partial removal of implants (prosthetic joint or internal fixation devices) for any indication had their implants subjected to sonication. In total, 74 patients were analyzed after the exclusion of three patients: 12 with arthroplasties and 62 with various sizes of internal fixation devices (osteosynthesis). Subjects were excluded if evident contamination occurred during implant removal, transportation, or processing of retrieved implants, if the container size for the implant was insufficient, or if fewer than two tissue culture samples were available. Informed consent was obtained from each patient, and the study was approved by the local Institutional Review Board (CAAE: 46277421.1.0000.5479, number: 4.713.409).

Study definitions

For diagnosing PJI, a more straightforward criterion was applied, as defined by the EBJIS. Diagnosis was confirmed if at least one of the following criteria was met: (i) Sinus tract with evidence of communication to the joint or visualization of the prosthesis, (ii) synovial fluid white blood cell count above 3,000 cells/µl, or polymorphonuclear leukocytes exceeding 80%, or a positive immunoassay or lateral-flow test for alpha-defensin, (iii) ≥ two positive samples of intraoperative (fluid and tissue) with the same microorganism or > 50 CFU (colony forming units) /ml of any organism through sonication (without concentration method), and (iv) presence of ≥ five neutrophils in ≥ five high-power fields (400x) in histological analysis, or presence of histologically visible microorganisms [4].

In cases of FRI, the consensus definition published by Metsemakers et al. was considered, in which diagnosis required the presence of at least one of the following criteria: (i) Fistula, sinus or wound breakdown (with communication to the bone or the implant), (ii) purulent drainage from the wound or presence of pus during surgery, (iii) phenotypically indistinguishable pathogens identified by culture from at least two separate deep tissue/implant (including sonication-fluid) specimens taken during an operative intervention, (iv) presence of microorganisms in deep tissue taken during an operative intervention, as confirmed by histopathological examination using specific staining techniques for bacteria or fungi [5]. Notably, for the evaluation of the culture techniques in this study, all criteria involving culture were omitted.

Sample collection

Periprosthetic tissue samples were collected and processed following the methodology outlined by Peel TN et al. [20]. A minimum of two samples were carefully obtained using different sterile materials from areas displaying evident signs of inflammation at the implant-bone interface membrane during surgical procedures. Each tissue specimen was individually placed into properly labeled sterile containers to maintain sterility. Each tissue specimen was individually placed into appropriately labeled sterile containers to preserve sterility. These containers were designated for subsequent microbiological and histological analysis at the microbiology laboratory. Concurrently, implant removal was conducted under aseptic conditions by the surgeon and then transferred into sterile polyethylene containers with tightly sealed lids. The packages were exclusively opened within the confines of the operating room. Upon preparation, the implants were immersed in up to 250 ml of Ringer Lactate solution within the container, which was subsequently hermetically sealed, appropriately labeled, and dispatched to the laboratory. Upon arrival at the laboratory, all samples underwent culturing within a 6-hour timeframe to uphold the integrity and viability of the specimens [10].

Tissue culture (TC)

The homogenized tissue sample in saline solution was distributed onto anaerobe blood sheep agar plates (BioMérieux, Marcy l’Étoile, France) under both aerobic and anaerobic conditions, as well as MacConkey agar (KASVI^®, Sao Jose dos Pinhais, Brazil) plates under aerobic conditions. Additionally, it was introduced into both TSB (KASVI^®, Sao Jose dos Pinhais, Brazil) and TG (Oxoid Ltd., Thermo Fisher Scientific, Boston, MA, USA), each under aerobic and anaerobic conditions, respectively. Aerobic cultures were sustained at a temperature of 37 °C for 7 days, whereas anaerobic cultures were allowed to incubate for a period of 14 days to ensure thorough microbial evaluation. TSB and TG were seeded onto sheep blood agar plates (BioMérieux, Marcy l’Étoile, France) [17, 26].

Sonication of retrieved implants

In the laboratory, the container containing the implant was agitated on the Vortex-Genie 2 for 30 s. Subsequently, it was submitted to a 5-minute sonication using the BactoSonic 14.2 (Bandelin GmbH, Berlin, Germany) at a frequency of 40 ± 2 kHz, with a power of 0.22 ± 0.04 W/cm². Finally, the container was returned to the Vortex for an additional 30 s, following the technique described by Trampuz et al. [15].

Conventional sonication fluid culture (CSFC)

Aliquots of 10 µL of the SF were inoculated on aerobic and anaerobic blood sheep agar and chocolate agar plates. Aerobic cultures were incubated at a temperature ranging from 35 to 37 °C for 7 days, while anaerobic cultures were incubated for 14 days. The remaining 4 ml of the SF was directly inoculated into 10 ml of TG and TSB, and incubated for 14 days and 7 days, respectively. If turbidity was observed in the broth, 10 µL samples were transferred onto aerobic and anaerobic blood sheep agar as well as chocolate agar plates. Daily inspections were conducted on all plates, with a positive test result indicated by microbial growth exceeding 50 CFU per plate [15, 26, 28].

Sonication fluid inoculated into blood culture bottles (SFBCB)

For this analysis, 10 ml aliquots of the SF were added to each aerobic and anaerobic BACTEC™ PLUS blood culture bottle equipped with an antimicrobial removal system. The cultures were incubated in the automated blood culture system for 5 days. Upon obtaining positive cultures, a 10 µL aliquot was seeded onto sheep blood agar plate for bacterial growth (33–34, 36).

Bacterial identification and susceptibility test

Microbial identification was conducted using the matrix-assisted laser ionization-desorption-time-of-flight (MALDI-TOF MS) technique using a Microflex LT spectrometer and the BiotyperTM 3.3 software package (Bruker DaltonicsTM, Billerica, MA, USA). Antimicrobial susceptibility testing was performed following the latest standards and cutoff points established by the Brazilian Committee on Antimicrobial Susceptibility Testing and the European Committee on Antimicrobial Susceptibility Testing (BrCAST/EUCAST, 2022) [21]. Minimum inhibitory concentrations (MICs) for antimicrobial agents were determined using the agar dilution technique. The S. aureus ATCC 29,213 strain served as a quality control. For vancomycin and polymyxin B, MIC determinations were carried out using broth microdilution as recommended. Quality control was again performed using the S. aureus ATCC 29,213 strain and Enterococcus faecalis ATCC 29,212.

Statistical analysis

The data were collected via the REDcap^® platform. Patient’s demographic characteristics were presented as frequencies and percentages or as mean and standard deviation. Qualitative variables were compared using either Fisher’s exact test or chi-square, as appropriate Continuous variables were compared using Student’s t-test for normally distributed data or the Mann-Whitney U test for non-normally distributed data. Sensitivity, specificity, positive predictive value, and negative predictive value of the diagnostic techniques were compared using the McNemar test with R software version 4.2.1 (2022-06-23 ucrt), the “DTComPair” package, and the “sesp.mcnema” function. Sensitivities and specificity of the techniques in different subgroups were compared using the proportion comparison test. The confidence interval considered was 95%, and the adopted significance level was 5%. Data were analyzed using SPSS software for Windows version 19.0 and R Studio version 2023.12.1.

Results

Patient and implants characteristics

Overall, 77 patients who underwent implant removal surgery were included in the study. Exclusion of three patients occurred, as two of the implants did not fit the container dimensions and one was contaminated during surgical retrieval process. Thus, the final study cohort comprised 74 patients, each corresponding to a single implant. Of these, 77% (57/74) were diagnosed with OIAI, with 48 cases of FRI and 9 cases of PJI, while 23% (17/74) had aseptic failures, consisting of 14 osteosynthesis and 3 arthroplasties cases. Using the criteria outlined by EBJIS and Metsemakers, 75% (9/12) and 77.4% (48/62) were regarded as PJI and FRI, respectively. The hip was the most common site for PJI (55.6%, 5/9), while lower limb fractures accounted for 66.1% (41/62) of FRI cases. A comparison of variables between patients with OIAI and aseptic failure (AF) in the study population, including demographic parameters, types of arthroplasties and locations, types of osteosynthesis, time elapsed between implant insertion and removal, clinical findings of OIAI, and radiological signs of implant loosening, is summarized in Table 1. Predictably, prolonged antibiotic administration, typically prescribed when suspicion of infection is elevated, was significantly more common in patients meeting any confirmatory infection criteria (84.4% vs. 5.9%, p < 0.001).

Table 1.

Comparison of variables and categories between patients with OIAI and aseptic failure

Variable and categories	EBJIS + METSEMAKERS#		p-Value
	Positive N (%)	Negative N (%)
Sex
Male	35/45 (77.8)	10/45 (22.2)	0.848 1
Female	22/29 (75.9)	7/29 (24.1)
Age (years) (mean [SD])	45.3 (18.3)	45.8 (21.3)	0.9323
Weight (mean [range])	73.4 (23.0–120.0)	74.9 (45.0–115.0)	0.7673
Height (mean [range])	168.2 (117.0–186.0)	166.6 (153.0–186.0)	0.382 4
BMI (means [SD])	25.8 (5.6)	26.8 (5.6)	0.567 4
Implant types
Internal fixation device	48/62 (77.4)	14/62 (22.6)	1.02
Prosthetic joint	9/12 (75.0)	3/12 (25.0)
Prosthetic joint sites
Hip	5/7 (71.4)	2/7 (28.6)
Knee	2/3 (66.7)	1/3 (33.3)	0.824 1
Shoulder	2/2 (100)	0/2 (0)
Primary arthroplasty cause
Fracture/trauma	6/7 (85.7)	1/7 (14.3)
RA@	1/1 (100.0)	0/1 (0.0)	0.689 1
Osteoarthritis	1/2 (50.0)	1/5 (50.0)
Unknown	1/2 (50.0)	1/2 (50.0)
Bone fixation device site
Humerus	2/3 (66.7)	1/3 (33.3)	0.255 1
Radius	1/2 (50.0)	1/2 (50.0)
Ulna	2/5 (40.0)	3/5 (60.0)
Femur	9/13 (69.2)	4/13 (30.8)
Tibia	22/23 (95.7)	1/23 (4.3)
Fibula	3/3 (100.0)	0/3 (0)
Vertebra	2/3 (66.7)	1/3 (33.3)
Foot	4/6 (66.7)	2/6 (33.3)
Hip	3/4 (75.0)	1/4 (25.0)
Internal fixation device
Rod	6/10 (60.0)	4/10 (40.0)	0.393 1
Plate	37/44 (84.1)	7/44 (15.9)
Screw	4/6 (66.7)	2/6 (33.3)
Kirschner wire	1/2 (50)	1/2 (50)
Time between implant insertion and removal (months) (mean [range])	33.3 (1.0-240.0)	39.9 (1.0–228.0)	0.076 3
CLINICAL CRITERIA
Local pain
Yes	44/55 (80.0)	11/55 (20.0)	0.349 2
No	13/19 (68.4)	6/19 (31.6)
Local erythema
Yes	32/37 (86.5)	5 /37(13.5)
No	25/37 (67.6)	12/37 (32.4)	0.053 1
Local edema
Yes	27/31 (87.1)	4/31 (12.9)
No	30/43 (69.8)	13/43 (30.2)	0.080 1
Fever
Yes	18/19 (94.7)	1/19 (5.3)
No	39/55 (70.9)	16/55 (29.1)	0.054 2
Implant loosening
Yes	10/12 (83.3)	2/12 (16.7)
No	47/62 (75.8)	15/62 (24.2)	0.721 2
Prolonged antibiotic use
Yes	31/32 (96.9)	1/32 (3.1)
No	26/42 (61.9)	16/42 (38.1)	< 0.001 1

1 - Chi-square test / 2 - Fisher’s exact test / 3 - Student’s t-test / 4 - Mann-Whitney test. EBJIS = European Bone and Joint Infection Society Definition. BMI = Body Mass Index. ^@RA = Rheumatoid arthritis. SD = Standard deviation *The criteria from EBJIS and Metsemarkers that involve culture were omitted for the evaluation of culture techniques

Accuracy comparison of microbiological diagnostic techniques

In total, 45.9% of subjects (34/74) had positive TC (33 OIAI and 1 AF), while CSFC and SFBCB were positive in 66.2% (49/74) (45 OIAI and 4 AF) and 77.0% (57/74) (55 OIAI and 2 AF) of subjects, respectively (P < 0.001). Among 57 patients diagnosed with OIAI (48 FRI and 09 PJI), the sensitivity of CSFC and SFBCB were both significantly higher compared to TC (78.9%, 45/57 versus 57.9%, 33/57, P = 0.011) (96.5%, 55/57 versus 57.9%, 33/57, P < 0.001), respectively. Comparison amongst sonication fluid methods demonstrated a significantly higher sensitivity when SF is inoculated in BCB compared to conventional methodology (96.5% versus 78.9%, P = 0.004). On the other hand, no statistical difference was observed on the specificity in the comparison of the three methods (Table 2). Worthy of notice, SFBCB showed a great positive and negative predictive values of 96.5% (88–100) and 88.2% (64–99), whereas on TC, positive predictive value (PPV) and negative predictive value (NPV) was 97.1% (85–100) and 40% (25–57), respectively (Table 3). Five or more tissue samples were collected in 74.4% of subjects (Table 4).

Table 2.

Sensitivity and specificity comparison between TC, SFBCB, and CSFC^a

COMPARISON	SENSITIVITY b		SPECIFICITY b
	VALUES (%)	P-VALUE	VALUES (%)	P-VALUE
TC and SFBCB	57.9 and 96.5	< 0.001	94.1 and 88.2	0.564
TC and CSFC	57.9 and 78.9	0.011	94.1 and 76.5	0.180
SFBCB and CSFC	96.5 and 78.9	0.004	88.2 and 76.5	0.317

TC = Tissue culture; SFBCB = Sonicate fluid in blood culture bottles; CSFC = Conventional sonicate fluid culture

^aComparison is among all subjects of study (n = 74). ^bSensitivity and Specificity comparisons were performed using the McNemar test

Table 3.

Performance description of TC, SFBCB, and CSFC

	Sensitivity		Specificity		PPV		NPV
Diagnostic method	% (no. detected/total)	95% CI	% (no. detected/total)	95% CI	% (no. detected/total)	95% CI	% (no. detected/total)	95% CI
TC	57.9 (33/57)	44–71	94.1 (16/17)	71–100	97.1 (33/34)	85–100	40.0 (16/40)	25–57
SFBCB	96.5 (55/57)	88–100	88.2 (15/17)	64–99	96.5 (55/57)	88–100	88.2 (15/17)	64–99
CSFC	78.9 (45/57)	66–89	76.5 (13/17)	50–93	91.8 (45/49)	80–98	52.0 (13/25)	31–72

TC = Tissue culture; SFBCB = Sonicate fluid in blood culture bottles; CSFC = conventional sonicate fluid culture; PPV = positive predictive value; NPV = negative predictive value; CI = confidence interval

Table 4.

Distribution of periprosthetic tissue samples collected per patient

Periprosthetic tissue samples collected	Number of patients N (%)
3	13/74 (17.6)
4	5/74 (6.8)
5	49/74 (66.2)
Above 5	7/74 (9.4)

^aPeriprosthetic tissue samples were collected according to the methodology outlined by Peel TN et al

Microbiological results

Microbiological yield was higher for SF inoculated in TSB compared to TG and agar plates. SF cultures in TSB yielded more Gram-positive cocci (GPC), whereas TG broth yielded more Gram-negative bacilli (GNB). Gram-positive cocci and Gram-negative bacilli were isolated in 65.6% and 34.4% of all CSFC, respectively (Table 5). Considering TC, CSFC, and SFBCB, 173 microorganisms were isolated, and Gram-positive cocci and Gram-negative bacilli accounted for 70.5% and 29.5%, respectively. Staphylococcus aureus was isolated in TC, SFBCB, and CSFC, in 47.4%, and 39.7%, and 41.9%, respectively. Staphylococcus epidermidis was isolated in 13.2%, 13.7%, and 11.3%, respectively. SFBCB presented a superior performance compared to TC in identifying microorganisms such as S. epidermidis, Enterobacter cloacae, Staphylococcus warneri, Staphylococcus haemolyticus, Enterococcus faecalis, Streptococcus agalactiae and Pseudomonas aeruginosa. Furthermore, SFBCB was the unique method capable of identifying certain bacteria including Corynebacterium spp, Staphylococcus cohnii, Klebsiella aerogenes, Enterococcus faecium, Dermobacter hominis and Acinetobacter ursingii. Notably, blood culture bottles yielded negative cultures in only 24.3% of cases. The distribution of isolated microorganisms for each technique is detailed in Table 6.

Table 5.

Microbiological findings of CSFC in 74 orthopedic implant-associated infection according to type of culture

		Value
Characteristic		SF in agar plates		SF in TG		SF TSB
Total no. of microorganisms isolated		35		41		46
No. (%) of cases by no. of microorganisms detected
1	29/74 (39.2)		32/74 (43.2)		38/74 (51.4)
≥ 2	3/74 (4.1)		4/74 (5.4)		4/74 (5.4)
0	42/74 (56.8)		38/74 (51.4)		32/74 (43.2)
No. (%) of GPCa		25/35 (71.4)		24/41 (58.5)		31/46 (67.4)
Staphylococcus aureus		17		17		19
Staphylococcus epidermidis		3		3		4
Staphylococcus haemolyticus		2		2		2
Othersa		3		2		6
No. (%) of GNBb		10/35 (28.6)		17/41 (41.5)		15/46 (32.6)
Pseudomonas aeruginosa		2		5		2
Klebsiella pneumoniae		1		3		3
Serratia marcescens		1		1		1
Enterobacter cloacae		5		5		5
Proteus mirabilis		1		2		2
Othersb		0		1		2

^a GPC: Gram positive cocci; ^b GNB: Gram-negative bacilli; SF = Sonication fluid; TG = thioglycolate broth; TSB = Tryptic soy broth

^a = Staphylococcus capitis; Staphylococcus warneri; Streptococcus canis; Enterococcus faecalis; Micrococcus luteus; Bacillus cereus

^b = Acinetobacter baumannii

Table 6.

Comparative assessment of microbial retrieval using different culture techniques

Microorganisms identified	TC (n = 38)	SFBCB (n = 73)	CSFC (n = 62)
Staphylococcus aureus	18/38 (47.4%)	29/73 (39.7%)	26/62 (41.9%)
Staphylococcus epidermidis	5/38 (13.2%)	10/73 (13.7%)	7/62 (11.3%)
Enterobacter cloacae	2/38 (5.3%)	6/73 (8.2%)	6/62 (9.8%)
Klebsiella pneumoniae	4/38 (10.6%)	4/73 (5.5%)	3/62 (4.9%)
Pseudomonas aeruginosa	1/38 (2.6%)	3/73 (4.0%)	5/62 (8.1%)
Acinetobacter baumannii	3/38 (7.9%)	2/73 (2.7%)	2/62 (3.2%)
Staphylococcus warneri	1/38 (2.6%)	3/73 (4.0%)	1/62 (1.6%)
Staphylococcus capitis	1/38 (2.6%)	1/73 (1.4%)	1/62 (1.6%)
Proteus mirabilis	0/38 (0%)	2/73 (2.7%)	2/62 (3.2%)
Staphylococcus haemolyticus	0/38 (0%)	1/73 (1.4%)	2/62 (3.2%)
Streptococcus agalactiae	1/38 (2.6%)	2/73 (2.7%)	0/62 (0%)
Enterococcus faecalis	0/38 (0%)	1/73 (1.4%)	2/62 (3.2%)
Bacillus cereus	0/38 (0%)	1/73 (1.4%)	2/62 (3.2%)
Serratia marcescens	0/38 (0%)	1/73 (1.4%)	1/62 (1.6%)
Streptococcus canis	0/38 (0%)	1/73 (1.4%)	1/62 (1.6%)
Staphylococcus cohnii	0/38 (0%)	1/73 (1.4%)	0/62 (0%)
Streptococcus dysgalactiae	1/38 (2.6%)	0/73 (0%)	0/62 (0%)
Klebsiella aerogenes	0/38 (0%)	1/73 (1.4%)	0/62 (0%)
Enterococcus faecium	0/38 (0%)	1/73 (1.4%)	0/62 (0%)
Acinetobacter ursingii	0/38 (0%)	1/73 (1.4%)	0/62 (0%)
Corynebacterium	0/38 (0%)	1/73 (1.4%)	0/62 (0%)
Dermabacter hominis	0/38 (0%)	1/73 (1.4%)	0/62 (0%)
Micrococcus luteus	0/38 (0%)	0/73 (0%)	1/62 (1.6%)
Candida parapsilosis	1/38 (2.6%)	0/73 (0%)	0/62 (0%)
Negative culturea	40/74 (54.1%)	18/74 (24.3%)	25/74 (33.8%)

The “n” refers to the total count of microorganisms detected in each culture type: TC = Tissue culture; SFBCB = Sonicate fluid in blood culture bottles; CSFC = conventional sonicate fluid culture

^aThe percentage of negative culture considered all the patients as the total

Susceptibility analysis

Overall, Gram-negative microorganisms exhibited heightened rates of resistance to most antimicrobials tested. Notably, all GNB demonstrated 100% resistance to cefoxitin, over 90% resistance to amoxicillin/clavulanic acid and piperacillin/tazobactam, and more than 80% to ciprofloxacin and cefepime, as shown in Fig. 1A and B. Conversely, Gram-positive bacteria presented resistant to penicillin (78,1%), followed by ciprofloxacin (71%), levofloxacin (70.8%). Interestingly, 36.5% of the Staphylococcus aureus were considered multidrug-resistant (MDR) for being non-susceptible to ≥ 1 agent in ≥ 3 antimicrobial categories. Furthermore, 50% of the Staphylococcus aureus isolates were considered MRSA.

Fig. 1.

Antibiotic Susceptibility Profile of Gram-Positive (A) and Gram-Negative Bacteria (B) from tissue and sonication fluid cultures in orthopedic implant-associated infections, shows high rates of resistance to most antimicrobials tested. Gram-positive bacteria (A) display higher rates of resistance β-lactams, quinolones, erythromycin and clindamycin. Gram-Negative Bacteria (B) exhibit a multidrug-resistant pattern

Discussion

OIAI present a significant challenge due to the predominance of biofilm-forming pathogens, which frequently resist conventional antibiotic treatments effective against planktonic pathogens [10]. This challenge is exacerbated by the increasing occurrence of implant-associated infections caused by multidrug-resistant (MDR) pathogens, complicating treatment regimens [22]. While antibiotic therapy guided by prior pathogen identification and susceptibility tests leads to higher cure rates and prevents further development of antimicrobial resistance, microbiological identification based on periprosthetic tissue culture may remain negative in up to 60% of cases OIAI, particularly among patients receiving antibiotic therapy [7, 8]. Importantly, a study by Mortazavi et al. found that failure to identify bacteria in cases of PJI predicts failure of two-stage exchange arthroplasty of the knee, with an odds ratio of 4.5 [23].

Numerous studies on diagnosing OIAI conducted across European and North American medical centers have demonstrated tissue culture sensitivity exceeding 60% [10, 24, 25]. On the other hand, our findings repeatedly revealed a TC sensitivity less than 60% (57.9%), consistent with the results of other studies [26, 27]. In the present study, tissue culture positivity fell below 60%, even though at least five or more tissue samples were collected in 74.4% of subjects studied. Conversely, a recent study assessed the accuracy of different numbers of specimens and diagnostic cut-offs for microbiological testing of deep-tissue specimens in patients with FRI. By collecting at least five deep-tissue specimens and using the diagnostic cut-off of at least two indistinguishable microorganisms, the tissue culture sensitivity was 68% [28]. Despite recent advances in the microbiological workup of OIAI based on traditional tissue cultures, its yield remains suboptimal.

Therefore, the implant sonication technique has proven to be a successful tool in overcoming the limitations of tissue culture yield, as demonstrated in several studies where SF culture yield was higher than conventional TC for diagnosing OIAI [9, 12, 13, 29]. It is currently validated that combination of three to five deep-tissue specimens with sonication method improves the diagnosis of OIAI [4, 5, 9, 28]. Nevertheless, diagnostic accuracy of OIAI remains challenging and further optimization of microbiological yield is necessary. Culturing synovial fluid, and sonication fluid into BCB to increase sensitivity of OIAI cases has been assessed by few groups, mostly for diagnosing PJI [17, 30–32]. Synovial fluid inoculated into BCB produces better results than growth on agar plates, and it has long been validated as a gold standard method [30]. Additionally, in a meta-analysis of only 4 studies analyzing 118 PJI cases, SFBCB increased the microbiological diagnosis with a pooled sensitivity and specificity of 0.85 and 0.86, respectively [31].

However, the diagnostic validity of SFBCB especially for FRI, and its application with the newly accepted definitions for PJI and FRI, remain unclear. The present study assessed 57 OIAI (48 FRI and 09 PJI) and SFBCB significantly increased the microbial diagnosis of OIAI compared to TC and CFSC, reaching higher sensitivity rates (96.5%). We speculate that the higher volume (up to 10 ml) of SF added to each growth medium of both aerobic and anaerobic BACTEC™ PLUS BCB, which contain an antimicrobial removal system is most likely to have influenced these results, as it allows immediate microbial growth after inoculation [33–35]. Interestingly, similar results were obtained by Portillo et al., in which 100% of sensitivity was achieved using SFBCB, although only 21 cases of FRI were reported [36].

Shen et al. and Toyama et al. also assessed the accuracy of SFBCB in PJI patients, reporting sensitivities of 85% and 80%, respectively [31, 36]. To our knowledge, no other study has compared the accuracy of SFBCB with CSFC and TC among FRI patients. Previous investigations of diagnostic accuracy of SFBCB have utilized both the BD Bactec system and BACT/ALERT^® blood culture systems, which are traditionally employed to shorten the duration of incubation [31, 35, 37]. While this study used the BACTEC™ PLUS BCB system, we believe that these systems are equivalent, and it does not seem to have influenced our results over other studies.

Interestingly microbiological results were obtained in our study. Firstly, SFBCB detected a higher number of pathogens and yielded negative cultures less frequently than TC and CSFC, consistent with the results of a meta-analysis evaluating PJI cases [31]. Secondly, bacterial yield was slightly higher for SF inoculated in TSB compared to TG and agar plates, while SF inoculated in TG broth yielded more GNB. TG is a complex enriched nonselective microbial growth medium widely used to support the growth of a broad range of aerobic and anaerobic microorganisms, while it is not inherently selective for either GPC or GNB. Albeit we cannot explain this awkward finding, GNB accounted for nearly one third of our isolates, which is likely to have influenced our results. Furthermore, several bacteria were isolated solely using SFBCB, some of which (Corynebacterium spp, S. cohnii, E. faecium, D. hominis and A. ursingii) are likely to be regarded as contamination. Although we found no significant difference in specificity between the three microbiological methodologies, SFBCB method lacks the possibility of colony counting threshold to define a true positive culture, therefore potentially limiting its specificity [17, 35]. It is worth mentioning that a recently published study investigated the sonication fluid cutoff point necessary to provide clinically significant culture results in PJI patients concluding that a cutoff point of ≥ 20 CFU/10 mL resulted in 99.8% of specificity [6]. Thirdly, in our study, polymicrobial infections were more frequently identified by SFBCB compared to CSFC and TC. Considering that biofilms create favorable conditions for polymicrobial infections, the higher rates of polymicrobial infection identified by SF methods are likely explained by their capacity to dislodge bacteria protected within the biofilms [16]. In contrast, the results of other investigator revealed statistically significant superiority of TC over SF cultures in detecting polymicrobial PJI (97.0% versus 67.0%) and TG broth yielded higher rates compared to BCB [38]. The influence of antimicrobial therapy up to 14 days before surgery on the positivity of the cultures has been investigated in numerous studies, with evidence suggesting a likely decrease in TC sensitivity while not impacting the sensitivity of SF culture [12, 15, 17, 35]. In the present study, previous antibiotic intake did not affect microbiological yield of SF cultures.

We recognize the limitations of our study. It is important to highlight that the hospitals included in this research do not routinely conduct histological analysis of periprosthetic tissues, which may have misclassified few cases as aseptic instead of PJI or FRI. Inoculating SFC into BCBs impedes the quantification of colony counts, therefore affecting the specificity of the test. It is essential to acknowledge the heightened risk of contamination associated with sonication compared to tissue culture due to the additional handling required by the technique. Therefore, strict adherence to the protocol is essential to mitigate the risk of contamination. Finally, in this study we have not compared the length of time to microorganism detection between tests. The rapid bacterial growth associated with this technique could prove highly valuable in ensuring a faster microbiological diagnosis and patient treatment, as already highlighted in the Portillo study [35].

In conclusion, our findings suggest that sonication techniques may offer a promising approach for detecting microorganisms within biofilms compared to traditional tissue culture methods. Furthermore, the enhanced sensitivity observed with sonication fluid culture in blood culture bottles compared to conventional sonication fluid culture underscores its potential as a valuable diagnostic tool. These insights need further exploration to optimize diagnostic accuracy and guide clinical practice.

Author contributions

All authors contributed to the study conception and design. Conceptualization: Anderson Xarif Bogarin Velasquez, Mauro José Salles; Methodology: Anderson Xarif Bogarin Velasquez, Giselle Burlamaqui Klautau, Mariana Neri Lucas Kurihara, Ingrid Nayara Marcelino Santos, Laura Batista Campos, Mayara Muniz de Andrade, Mauro José Salles; Formal analysis and investigation: Ícaro Santos Oliveira, Thomas Stravinkas Durigon, Laís Sales Seriacopi; Writing - original draft preparation: Anderson Xarif Bogarin Velasquez, Mariana Neri Lucas Kurihara, Mauro José Salles; Writing - review and editing: Mariana Neri Lucas Kurihara, Anderson Xarif Bogarin Velasquez, Mauro José Salles; Funding acquisition: Mauro José Salles, Ícaro Santos Oliveira, Thomas Stravinkas Durigon, Laís Sales Seriacopi; Resources: Mauro José Salles; Supervision: Mauro José Salles.

Funding

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication. This work was supported by Fundo de Amparo à Pesquisa (FAP) of Faculdade de Ciências Médicas da Santa Casa de São Paulo (No: 2021/06314-3) and Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) (No: 006/2021).

Data availability

In the manuscript, there is no raw data such as nucleic acid sequences, protein sequences, genetic maps, SSR, expression data, etc. Nevertheless, all the metaphases pictures are in possession of the authors and available for the reviewers or for submissions in any database if necessary.

Declarations

Competing interests

The authors have no competing interests to declare that are relevant to the content of this article.