The diagnostic value of blood sample NGS in patients with early periprosthetic joint infection after total hip arthroplasty

Abstract

The diagnostic value of next‐generation sequencing (NGS) of blood samples from patients with periprosthetic joint infection (PJI) after total hip arthroplasty (THA) was evaluated by comparing it with drainage fluid NGS and bacterial culture. The study was designed as a retrospective diagnostic test. Thirty‐six infected patients were diagnosed with PJI according to the Musculoskeletal Infection Society (MSIS) criteria and 57 volunteers were included in our study. NGS and bacterial culture were chosen to detect PJI after THA. Blood samples and drainage fluid were collected for NGS, and the drainage fluid, which was collected at the same time as the NGS drainage fluid sample, was used for bacterial culture. The primary outcomes of interest were sensitivity, specificity, and accuracy. In the infection group, 31 patients showed positive results by blood sample NGS, 33 patients showed positive results by drainage fluid NGS, and 17 patients showed positive bacterial culture results. In the control group, the results of 2 blood sample NGS, 16 drainage fluid NGS, and 3 bacterial cultures were positive. The sensitivity, specificity, and accuracy of the blood sample were 0.86, 0.96, and 0.92, respectively. The sensitivity, specificity, and accuracy of the drainage fluid samples were 0.92, 0.72, and 0.80, respectively. The sensitivity, specificity, and accuracy of bacterial culture were 0.47, 0.95, and 0.79, respectively. The study demonstrated that both the sensitivity and specificity of NGS were higher than those of bacterial culture, regardless of the kind of sample. Compared with drainage fluid NGS, the sensitivity of blood sample NGS was slightly lower (0.86 vs 0.92), but blood sample NGS showed higher specificity (0.96 vs 0.72). In total, the diagnostic value of blood sample NGS was superior to that of drainage fluid NGS and bacterial culture. The majority of infected patients could be identified by blood sample NGS. Moreover, because of its high specificity, blood sample NGS can not only detect infectious bacteria but also distinguish infectious from non‐infectious bacteria, which is dramatically different from using drainage fluid NGS.

Abbreviations

ACC

accuracy

CBC

complete blood count

CRP

c‐reactive protein

ESR

erythrocyte sedimentation rate

FN

false negative

FP

false positive

LR+

positive likelihood ratio

LR‐

negative likelihood ratio

MSIS

musculoskeletal infection society

NPV

negative predictive value

NGS

next‐generation sequencing

PCR

polymerase chain reaction

PJI

periprosthetic joint infection

PV+

positive predictive value

PV‐

negative predictive value

PPV

positive predictive value

Sen

sensitivity

Spe

specificity

THA

total hip arthroplasty

TN

true negative

TP

true positive

WBC

white blood cell

1. INTRODUCTION

Periprosthetic joint infection (PJI) is the most devastating complication after total hip arthroplasty (THA) ¹ , ² and one of the leading causes of revision after THA. ³ Published evidence suggests that the incidence of PJI after THA surgery is approximately 1% ~ 2%, with an increase in the aging population being anticipated. ⁴ Currently, although many methods exist to diagnose PJI, the definite diagnosis of PJI is still difficult for clinicians. ⁵ In clinical practice, it is difficult to accurately determine whether PJI exists and which infective pathogen causes PJI. ⁵ , ⁶ Bacterial culture is the standard method for the diagnosis of PJI, but the sensitivity is highly inconsistent (58%–95%) among different studies. ⁷ Serological markers are also common technology for the clinical diagnosis of PJI. For instance, C‐reactive protein (CRP) is a commonly used marker ⁸ ; however, it has demonstrated limited sensitivity from 67% to 85.1%. ⁹ , ¹⁰ Moreover, serological markers are not useful in the identification of infectious pathogens. ¹¹ For example, the erythrocyte sedimentation rate (ESR) and CRP could both increase whether there was a gram‐positive or gram‐negative bacterial infection, which would not help clinicians identify the kind of bacteria. Obviously, the common disadvantage of these techniques is low sensitivity. In addition, it is difficult to culture special microorganisms, such as fungi and mycobacteria. ¹² To improve this disadvantage, NGS was introduced into the diagnosis of PJI.

NGS is a novel method that can identify all nucleic acids in a given sample within a short time period. NGS has been shown to substantially increase the detection rate of bacteria. According to the report from Yin et al., the sensitivity of NGS to diagnose PJI was approximately 93.3%; moreover, NGS showed a much higher diagnostic value in culture‐negative cases (87.5%). ⁹ , ¹³ Although NGS has demonstrated high sensitivity, it still has inherent disadvantages. At present, in most studies using NGS to diagnose PJI, the samples mainly come from incision drainage and intraoperative samples. ¹⁴ The drainage fluid sample could be obtained preoperatively. However, the NGS results of drainage fluid samples are easily affected by unrelated resident bacteria, ¹ which therefore limits the specificity of the drainage fluid NGS detection methods. Intraoperative samples have higher specificity and sensitivity. ¹⁵ However, obtaining samples from surgery is invasive to the patient. Furthermore, the NGS results of intraoperative samples cannot provide any evidence that could help make clinical decisions prior to debridement surgery and provide guidance in the use of antibiotics before PJI debridement. Compared with drainage fluid samples and intraoperative samples, blood is an ideal sample. Normally, the bloodstream is sterile. ¹⁶ Because bacteria can enter the bloodstream from the infected lesion, the infection can be confirmed as long as the bacteria are detected in the blood. However, the sensitivity of blood bacterial culture is relatively low. ¹⁷ , ¹⁸ In a study on the diagnosis of typhoid fever, Mogasale reported that the blood culture sensitivity was 66%. ¹⁹ In addition, some pathogens, such as mycobacterium or fungus, are difficult to detect by blood culture.

Based on its high sensitivity, NGS can detect pathogens correctly, even if there is an extremely low level of bacteria. The specificity of blood sample NGS would be, predictably, high. Therefore, we chose blood sample NGS detection before PJI debridement to determine the pathogens that are present. However, how sensitive is blood sample NGS? At present, very few studies have explored the diagnostic value of blood sample NGS, which is worth being evaluated in detail. Therefore, this study was conducted to evaluate the diagnostic value of blood sample NGS. Both the sensitivity and specificity of blood sample NGS were expected to be high.

2. PATIENTS AND METHODS

2.1. Study design

The study is designed as a retrospective diagnostic study. The diagnostic criteria according to the Musculoskeletal Infection Society (MSIS) were regarded as the “gold standard”. ²⁰ , ²¹ The participants were divided into an infection group and a control group according to the MSIS criteria. In this study, the diagnostic value of blood sample NGS was precisely evaluated according to some parameters, such as sensitivity, specificity, and accuracy.

2.2. Participants

Infected patients undergoing THA who were diagnosed with early PJI from October 2018 to September 2021 were included in the study. Note that, although there was no evidence demonstrating that the risk of infection was different between cemented and uncemented prostheses in THA, ²² it should be clarified here that all the patients in this study received uncemented hip prostheses. There was adequate evidence to identify the existence of infection for each infected patient diagnosed with PJI. The inclusion criteria of the patients were as follows: (1) the patients were diagnosed with PJI according to the MSIS criteria. (2) the infection was identified within 30 days. (3) Drainage fluid was continuously present (4) The infection was confirmed through the result of bacterial culture, for which we used at least three separate drainage fluid samples or intraoperative samples. The exclusion criteria of the patients were as follows: (1) infected lesions present in other parts of the body; (2) additional antibiotics therapy had been given to treat the infection after 24 hours postoperative, at when the first sample had not been collected; and (3) suspected but undiagnosed infection. According to the inclusion and exclusion criteria, 36 infected patients were included in the study. Meanwhile, 57 volunteers were included as controls. The volunteers also underwent THA but were not diagnosed with PJI after surgery. The initial sample size calculation assumed that in the infection group, the sensitivity of blood sample NGS, drainage fluid NGS and bacterial culture were 80%, 90%, and 50%, respectively. When the significance level and power of the test were set to 0.05 and 0.8, respectively, there were 14 infected patients and 14 controls. Therefore, the sample size in the study was considered to be sufficient. This study was approved by the Institutional Review Board of the Third Hospital of Hebei Medical University and was conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all individuals before they participated in the study.

2.3. Clinical evaluation

The demographic characteristics of all patients were retrospectively collected from medical records. The maximum temperature of the patients after infection was recorded and used for analysis in the infection group. In contrast, the maximum temperature three days after surgery was recorded in controls. This was because fever is the most common complication after surgery, and fever in the first three days after surgery was considered to be associated with the absorption of necrotic tissue and hematomas. ²³ For the control group, the body temperature after three days postoperative was less affected by the surgery, which was suitable for comparison with the infection group. So, the body temperature at that time point for each individual in control group was used for further analysis in this study. For the patients who were infected, additional complete blood count (CBC), CRP, and ESR tests were performed. In the infection group, the first test results after infection were recorded for analysis. The CBC, CRP, and ESR were routinely tested on the 1st and 4th days after surgery. The test results on the 4th day were recorded in the controls. Generally, the drainage fluid was obtained for bacterial culture at least three times before debridement and was cultivated for 14 days each time. The result of the patient's first bacterial culture, in which the sample was obtained before using antibiotics was used for further analysis in the study. Intraoperative samples were collected for bacterial culture to evaluate the accuracy of the test results. In contrast, the drainage fluid taken on the 3rd day, when the incision dressing was changed, was used for bacterial culture in the controls. In addition, according to a relative report, the surgical time, the volume of bleeding, and the use of bone grafting could all increase the risk of infection. ²⁴ , ²⁵ , ²⁶ Therefore, the above information associated with surgery was collected based on the medical records for THA.

2.4. Sampling and next‐generation sequencing

In this study, both blood samples and drainage fluid samples obtained from each participant were collected for NGS detection. The drainage fluid sample for NGS detection was simultaneously collected when the drainage fluid sample was collected for bacterial culture. For the patients with infections, another blood sample was collected for NGS when the blood sample was collected for the first additional CBC, CRP, and ESR testing. In the control group, another blood sample for NGS testing was collected at the same time that the blood samples were collected for routine CBC, CRP, and ESR testing, which were performed on the 4th day after THA surgery.

Both the acquisitions of blood sample and drainage fluid sample were followed standard operating procedures (SOPs). Fasting venous blood sample was obtained from patient in the morning in the treatment room, which was disinfected by ultraviolet light for at least 1 hour. Median cubital vein was selected as blood sampling sites, which were sterilised at least three times (15 cm around the puncture point) with cotton balls dipped with 0.2% iodin and 75% ethanol. Tourniquet was fastened before the skin puncture and released after the success of puncture. Then the blood sample was obtained and saved in a sterilised vacuum blood collection tube without anticoagulation. Drainage fluid sample was obtained in the same treatment room. First, the dressing of the incision was removed. Then the incision was sterilised with cotton balls dipped with 0.5% iodophor three times. After the dry of sterilised fluid, a sterilised swab was inserted into the incision to obtain the drainage fluid sample. During this procedure, the local tissues around the incision might be compressed in order to obtain enough drainage fluid volume. Subsequently, the swab was saved in a sterilised collection tube.

The collected samples were diluted with 0.5 mm glass beads separately. After the diluted samples were centrifuged, the TIANamp Micro DNA Kit was used for DNA extraction according to the manufacturer's instructions. By fragmenting DNA, repairing end‐performing phosphorylation, adaptor ligation, and Polymerase Chain Reaction (PCR) amplification, extracted DNA was used for the construction of DNA libraries. An Agilent 2100 was used to evaluate the quality of the DNA libraries. Qualified DNA libraries were sequenced on the BGISEQ‐50 platform. High‐quality reads were obtained by removing low‐complexity reads and eliminating human host sequences. Such high‐quality reads were aligned to microbial genome databases, including bacterial, viral, fungal, and parasite databases. ²⁷ , ²⁸ , ²⁹ , ³⁰

2.5. Outcome

The primary outcomes of the study were the sensitivity and specificity of blood sample NGS. The secondary outcomes included the accuracy, Youden's index, positive predictive value (PV+), negative predictive value (PV−), positive likelihood ratio (LR+), and negative likelihood ratio (LR−). The parameters of both drainage fluid NGS and bacterial culture were simultaneously calculated and compared with those of blood sample NGS.

2.6. Statistical analysis

Statistical analyses were performed using SPSS 26.0 statistical software for Windows (IBM, Armonk, NY) and Excel 2021 for Windows (Microsoft Corporation, Seattle, WA). Continuous variables are expressed as the mean ± standard deviation, and categorical variables are expressed as frequencies. Sensitivity, specificity, accuracy, positive predictive value, negative predictive value, positive likelihood ratio, and negative likelihood ratio were calculated to indicate the diagnostic value of the blood sample NGS. McNemar tests were used to identify the differences in sensitivities, specificity, and accuracy between the groups. Pearson's chi‐squared test was used to identify the differences in positive predictive value, negative predictive value, positive likelihood ratio, and negative likelihood ratio between the groups. A P value less than 0.05 was considered to be significant.

3. RESULTS

3.1. Demographic information and clinical features

In this study, the mean age of the patients was 48.66 ± 13.70 years (range from 23 to 74 years), 64 patients (68.8%) were men, and the mean body mass index was 26.85 ± 2.83 kg/m² (range from 22.28 to 31.79 kg/m²). Based on the MSIS standard, the infection and control groups included 36 and 57 patients, respectively. There was no significant difference in the baseline characteristics, such as age, sex, body mass index, alcohol consumption, steroid use, and rheumatoid status, between the groups. Detailed information is provided in Table 1.

TABLE 1.

Demographic characteristics of the patients undergoing total hip arthroplasty

Characteristics		Infection (n = 36)	Control (n = 57)	Total (n = 93)	Test statistics	P
Age (years)		48.86 ± 12.70	48.63 ± 14.41	48.66 ± 13.70	−0.036	0.972 a
Sex	Male	23 (63.9%)	41 (71.9%)	64 (68.8%)	0.665	0.415 b
	Female	13 (36.1%)	16 (28.1%)	29 (31.2%)
Body mass index		26.88 ± 2.88	26.84 ± 2.82	26.85 ± 2.83	−0.126	0.900 a
Smoking	Yes	11 (30.6%)	13 (63.8%)	24 (25.8%)	0.692	0.406 b
	No	25 (69.4%)	44 (77.2%)	69 (74.2%)
Alcohol	Yes	18 (50.0%)	25 (43.9%)	43 (46.2%)	0.335	0.563 b
	No	18 (50.0%)	32 (56.1%)	50 (53.8%)
Steroid	Yes	9 (25.0%)	7 (12.3%)	16 (17.2%)	2.506	0.113 b
	No	27(75.0%)	50 (87.7%)	77 (82.8%)
Rheumatoid	Yes	4 (11.1%)	6 (10.5%)	10 (10.8%)	0.008	0.929 b
	No	32 (88.9%)	51 (89.5%)	83 (89.2%)

^a Mann–Whitney U test.

^b Chi‐square test.

Meanwhile, no statistically significant difference was noted in the surgical time, bleeding, or bone grafting between the groups. However, most of the inflammatory parameters were different between the infection group and the control group, except for the WBC count. The mean body temperature of the infection group was 38.34 ± 0.85°C, while that of the control group was 37.68 ± 0.95°C (P < 0.001). The neutrophil counts of the infection group (83.48 ± 11.51 × 10⁹/L) were higher than those of the control group (69.43 ± 15.85 × 10⁹/L) (P < 0.001). The CRP level was significantly higher in the infection group (100.17 ± 29.35 mg/L) than in the control group (82.08 ± 33.04 mg/L) (P = 0.014). The ESR level was also significantly higher in the infection group (100.83 ± 27.19 mm/h) than in the control group (77.37 ± 30.39 mm/h) (P = 0.001). In addition, in the infection group that had a total of 36 patients, there were 22 patients with a final diagnosis of gram‐positive bacterial infection, 10 patients with gram‐negative bacterial infection, two patients with mycoplasma infection, one patient with gram‐positive bacilli infection, and one patient with mycobacterium infection. Detailed information is shown in Table 2.

TABLE 2.

Clinical features of the patients undergoing total hip arthroplasty

Characteristics		Infection (n = 36)	Control (n = 57)	Total (n = 93)	Test statistics	P
Pathogen	Gram+ cocci	22 (61.1%)
	Gram+ bacilli	1(2.8%)
	Gram− bacilli	10 (28.8%)
	Mycobacterium	1 (2.8%)
	mycoplasma	2 (5.55%)
Surgical time		50.28 ± 18.90	48.25 ± 20.19	49.03 ± 19.62	−0.460	0.645 a
Bleeding		252.78 ± 105.52	215.79 ± 123.26	230.11 ± 117.52	−1.461	0.144 a
Bone grafting	Yes	5 (13.9%)	7 (12.3%)	12 (12.9%)	0.051	0.822 b
	No	31 (86.1%)	50 (87.7%)	81 (87.1%)
Temperature		38.34 ± 0.85	37.68 ± 0.95	37.94 ± 0.97	−2.732	0.006 a
WBC count		9.02 ± 2.74	8.52 ± 2.76	8.71 ± 2.75	−0.907	0.364 a
Neutrophils		83.48 ± 11.51	69.43 ± 15.85	74.87 ± 15.83	−4.271	<0.001 a
CRP		100.17 ± 29.35	82.08 ± 33.04	89.09 ± 32.72	−2.445	0.014 a
ESR		100.83 ± 27.19	77.37 ± 30.39	86.45 ± 31.23	−3.277	0.001 a

Abbreviations: CRP, c‐reactive protein; ESR, erythrocyte sedimentation rate; WBC, white blood cell.

^a Mann–Whitney U test.

^b Chi‐square test.

3.2. Results of next‐generation sequencing

Compared with traditional bacterial culture, more infected patients could be identified correctly by NGS regardless of the kind of sample. The blood sample NGS results of 31 patients were positive, and those of the other 5 patients were negative. The results of NGS from drainage fluid were positive in 33 patients and negative in 3 patients. Only 17 patients were positive for bacterial culture, and 19 patients were negative.

In the control group, blood sample NGS and bacterial culture detected more negative results than drainage fluid NGS. Fifty‐five patients had negative blood sample NGS results among 57 patients, and only 2 patients had positive results. NGS of the drainage fluid showed 41 negatives and 16 false positives. Traditional bacterial culture showed 54 negative and 3 positive results. Detailed information regarding the clinical features is illustrated in Table 3.

TABLE 3.

Diagnostic capabilities of blood NGS, drainage fluid NGS and bacterial culture

	FP	TP	TN	FN	ACC	Sen	Spe	PPV	NPV	+LR	−LR
blood NGS	2	31	55	5	0.92	0.86	0.96	0.94	0.92	24.54	0.14
drainage fluid NGS	16	33	41	3	0.80	0.92	0.72	0.67	0.93	3.27	0.12
Test statistics a					0.309	7.622	0.809	8.139	0.082	313.140	4.483
P1					0.017	0.625	0.001	0.004	0.775	<0.001	0.034
bacterial culture	3	17	54	19	0.79	0.47	0.95	0.85	0.74	8.97	0.56
Test statistics b					0.101	0.380	0.115	0.380	0.105	43.819	363.649
P2					0.004	0.650	1.000	0.003	1.000	<0.001	<0.001

Note: Comparison between drainage fluid NGS and bacterial culture, P < 0.05.

Abbreviations: ACC, accuracy; FP, false positive; FN, false negative; +LR, positive likelihood ratio; −LR, negative likelihood ratio; NPV, negative predictive value; PPV, positive predictive value; Sen, sensitivity; Spe, specificity; TP, true positive; TN, true negative.

^a Comparison between blood NGS and drainage fluid NGS.

^b Comparison between blood NGS and bacterial culture.

3.3. Diagnostic value of next‐generation sequencing

As expected, due to the elimination of false positive results, blood sample NGS has the highest specificity, positive predictive value, and positive likelihood ratio. In contrast, despite the highest sensitivity, the specificity of drainage fluid NGS, as well as the positive predictive value and the positive likelihood ratio, were relatively low due to the high false‐positive rate. In detail, compared with drainage fluid NGS, blood sample NGS has higher specificity (0.96 vs 0.72, P = 0.001), positive predictive value (0.94 vs 0.67, P = 0.004) and positive likelihood ratio (24.54 vs 3.27, P < 0.001), as well as a higher accuracy (0.92 vs 0.80, P = 0.017). Although the sensitivity (0.86 vs 0.92, P = 0.625) of blood sample NGS appeared to be slightly lower than that of drainage fluid, no significant difference was found.

In addition, compared with traditional bacterial culture, the sensitivity (0.86 vs 0.47, P = 0.650), positive predictive value (0.94 vs 0.85, P = 0.003) and negative predictive value (0.92 vs 0.74, P = 1.000) of blood sample NGS were much higher than those of bacterial culture. Detailed information regarding the diagnostic value of the three detection methods is also shown in Table 3.

4. DISCUSSION

Traditionally, the identification of infectious pathogens depends on the isolation and culture of bacteria. However, this method has a relatively low sensitivity and long culture time. Currently, some novel detection methods, such as qPCR, NGS and 16S ribosomal DNA identification based on molecular technology were introduced and applied to identify infectious pathogens. ³¹ , ³² Nucleic acid fragments in samples could be directly detected and then retrieved in the biological databases. The infectious pathogens could be rapidly identified by these novel culture‐independent methods. Among them, qPCR and 16S ribosomal DNA identification used to be the most cost‐effective and widely used molecular biology methods, which could detect specific gene segments. Nevertheless, these methods could only identify one or some certain genus or species of the bacteria but not differentiate two closely related bacteria in the same genus or species, such as Shigella spp. and Escherichia coli. ³³ Compared with the above methods, NGS can make a sequencing of all the nucleic acid fragments in one single test and provide more comprehensive and in‐depth information. This would make an effort in making diagnosis of infectious pathogens and provide a reference for the clinical application of antibiotics. In this study, a blood sample NGS was used to make a diagnosis of PJI, and it revealed an excellent sensitivity and specificity.

In summary, positive results of blood sample NGS were commonly observed in patients with PJI, indicating that the blood of patients with PJI indeed contained trace bacteria or at least contained fragments of bacteria that could be detected by NGS, which is an extremely sensitive method. The diagnostic value of blood sample NGS was then evaluated. As hypothesized previously, the results of this study demonstrated that the sensitivity of blood sample NGS was sufficient for the diagnosis of PJI (86%), particularly when combined with clinical manifestations of infection, such as elevated CRP levels and increased incision drainage. In addition, the more important finding of this study was that, due to the elimination of background bacteria and compared with traditional drainage fluid NGS, the false positive result of blood sample NGS was dramatically decreased. In this case, the major disadvantage of NGS, which is that a healthy individual could be easily misdiagnosed as an infected individual, had been overcome by the modified sampling method. Therefore, it is reasonable to believe that NGS of blood samples might be an ideal alternative to NGS of drainage fluid since the sensitivities of these two diagnostic methods was similar. However, false positive results were rarely identified in blood sample NGS.

In fact, it is difficult to determine whether a patient is suffering from infection or not. There are several diagnostic criteria for infection and PJI, such as the guidelines raised by Spanish Society of Infectious Diseases and Clinical Microbiology (SEIMC), ³⁴ the instruction from The American Academy of Orthopaedic Surgeons (AAOS) ³⁵ and the criteria from Workgroup of the MSIS. According to the study of Pulido et al, MSIS criteria were considered to be both sensitive and specific for the accurate diagnosis of PJI, and which was commonly used in several similar studies. ³⁶ So, in this study, the MSIS criteria were used for clinical diagnosis of PJI. Huang et al. studied the diagnostic value of NGS for PJI, collecting preoperative synovial fluid as research samples. They reported that compared with bacterial culture, NGS possessed higher sensitivity but lower specificity. ³⁷ Furthermore, in a study that included 65 revision arthroplasties and 17 primary arthroplasties, Tarabichi et al. collected intraoperative samples for NGS testing. They reported that NGS was more sensitive than bacterial culture (89.3% vs 60.7%), but NGS was less specific than culture (73.0% vs 97.3%). ³⁰ Therefore, the traditional NGS test was more suitable for identifying the kind of bacteria causing the infection after the patients were diagnosed with PJI. The unique characteristic that distinguished this study from previous studies was that the use of blood samples was first introduced in NGS for the identification of PJI. Similar to our study, the study of Sarah et al. suggested that bacteria could be isolated from blood samples obtained from patients with pneumonia, and the positive rate of blood culture was approximately 50.4%. In their study, it was observed that the most common bacteria were Streptococcus pneumoniae (33%), followed by S. aureus (22%) in the blood samples of patients with pneumonia. ³⁸ In a study assessing the value of blood culture in diabetic foot osteomyelitis, Letertre‐Gibert et al. reported that the blood culture was positive in 15.8% of patients. The most common bacteria in the blood samples collected from patients was Streptococcus sp. ³⁹ Benkler et al. collected microbiological data on deep sternal wounds and found that the results of 40% of the blood cultures were consistent with those of wound cultures, and the most commonly cultured bacteria from blood cultures was S. aureus. ¹⁴ The above studies suggested that bacteria did exist in the bloodstream of infected patients and could even be found by bacterial culture using blood samples. However, the bacterial cultures from blood samples only showed positive results when a certain amount of bacteria was present in the blood. However, the high sensitivity of NGS makes it possible to find extremely low levels of bacteria, especially from blood samples. In addition, according to the above studies, it could be speculated that gram‐positive bacteria were more easily detected in the blood of infected patients. One of the characteristics of PJI is that gram‐positive bacteria are still the main kind of bacteria that cause PJI, such as S. aureus and coagulase‐negative staphylococci. ⁴⁰ Meanwhile, our study demonstrated that there were 22/36 patients with a final diagnosis of a gram‐positive bacterial infection in the infection group. The gram‐positive bacteria often migrated to the bloodstream, which resulted in the viable detection of infective bacteria from the blood samples of patients with PJI.

Two major questions should be answered when managing a patient who is suspected to suffer from PJI: whether the patient is infected or not and which kind of bacteria caused the infection. Traditional drainage fluid NGS has a limited effect on determining whether a patient is infected. When performing NGS from drainage fluid from a healthy individual, false positive results are commonly found. ⁴¹ , ⁴² A low PPV means that only 67% of the positive results of drainage fluid NGS were correct. Therefore, when bacteria were found by drainage fluid NGS in a patient, he or she sometimes cannot be diagnosed as an infected individual. In this situation, the diagnosis of PJI must be established by combining NGS with other clinical and laboratory evidence. In contrast, the specificity of blood sample NGS was much higher than that of drainage fluid NGS. This means that healthy individuals are only rarely misdiagnosed as infected individuals. Therefore, considering the acceptable sensitivity, blood sample NGS might be more suitable for determining whether a patient was infected or not. The second question is whether the correct pathogenic bacteria can be identified by the diagnostic method. In this study, a positive result was defined when the bacteria that was identified by the diagnostic method was the same as the bacteria causing the infection. Therefore, the true positive rate, namely, the sensitivity, which means that in infected patients, the possibility of infected bacteria being correctly identified, was more important in the identification of infectious bacteria. Due to the highest level of sensitivity, drainage fluid NGS might be more suitable for such a circumstance (to determine the kind of infectious bacteria). In summary, blood sample NGS was more suitable for identifying whether an individual was infected. If the infection status was confirmed in a certain patient, NGS of the drainage fluid could help detect the kind of infectious bacteria.

Some limitations must be considered in this study. First, the sample size was relatively small. Therefore, the statistical power is relatively weak. Second, although in several similar studies the clinical diagnoses that were determined by MSIS criteria were considered the “gold standard” diagnoses, it is uncertain whether these patients were truly infected. Additionally, the bacterial species and genera determinations were not mentioned according to the MSIS standard. To eliminate potential misdiagnoses, only patients who were confirmed to have infection (the same bacteria isolated from at least three separate drainage fluid samples or the bacterial culture results preoperatively were consistent with those intraoperatively) and in whom the infectious bacteria were determined were included in this study. Consequently, some patients with mild to moderate PJI were excluded. This might lead to an overestimation of the sensitivity because the bacteria might only enter the bloodstream in serious infections. Third, a benefit analysis was not performed for NGS. Forth, PJI might be caused by some rare pathogen, such as mycobacterium, mycoplasma, and chlamydia. In this study, there were only three patients infected by mycobacterium and mycoplasma. Therefore, the diagnostic value of NGS when used for the diagnosis of PJI caused by rare pathogens was not evaluated. Finally, patients with subacute and chronic PJI were excluded from this study, which might decrease the impact of the results.

5. CONCLUSION

This study confirmed that blood sample NGS helped identify whether an infection existed and which was the culprit pathogen causing the infection following THA. Due to the elimination of false positive results, the specificity and positive predictive value of blood sample NGS were dramatically increased compared with those of drainage fluid NGS. As a minimally invasive diagnostic method, drainage fluid NGS could be recommended for patients who are suspected to suffer from PJI after THA.

CONFLICT OF INTEREST

The authors declare that there is no conflict of interest.

Zhang B, Li M, Liu Y, et al. The diagnostic value of blood sample NGS in patients with early periprosthetic joint infection after total hip arthroplasty. Int Wound J. 2023;20(4):961‐970. doi: 10.1111/iwj.13943

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.