Diagnostic accuracy of the alpha-defensin lateral flow test in periprosthetic shoulder infection: a retrospective study

Abstract

Background

The aim of this study is to assess the diagnostic value of the alpha-defensin lateral flow (ADLF) test in a group of patients with suspected shoulder periprosthetic joint infections (PJIs) and its usefulness and correlation with other diagnostic tests in the decision-making process.

Methods

This retrospective single-center study was conducted between 2018 and 2023. Alpha-defensin was measured using an ADLF test kit. The final diagnosis of shoulder PJI was confirmed or ruled out based on the Musculoskeletal Infection Society criteria.

Results

After selecting patients based on the exclusion criteria, the final study group comprised 22 patients. Of all patients, the test was positive in five cases. The ADLF test was false-positive in two patients and was false-negative in zero patients. The test was truly negative in 17 cases, resulting in a sensitivity of 100.00% (95% CI, 100.00%–100.00%) and a specificity of 89.47% (95% CI, 76.65%–102.30%). The positive predictive value was 60.00% (95% CI, 39.53%–80.47%) and the negative predictive value was 100.00% (95% CI, 100.00%–100.00%). The accuracy of the α-defensin test was found to be 90.9%.

Conclusions

Although this study has limitations such as small sample size and research design, the ADLF test has proven to be a useful diagnostic tool. It can serve as a supplement to traditional tests, providing additional support to orthopedic surgeons in both preoperative and intraoperative diagnosis, especially in its ability to exclude the possibility of a PJI.

Level of evidence

III.

INTRODUCTION

Infections of shoulder prostheses represent the fifth most common cause of prosthetic revision, with an incidence rate of 1.6%, positioning them as the second leading cause of revision procedures [1-4]. Diagnosis represents a challenge for orthopedists because the most common pathogens are of low virulence (Staphylococcus and Cutibacterium acnes), often delaying diagnosis compared to that of other prosthetic joints [5]. A precise diagnosis of infection is crucial, in order to avoid unnecessary two-stage revisions due to false-positive preoperative tests, as well as to prevent recurrent failures from false-negative results leading to one-step revision [6].

In patients with shoulder prosthesis infection, pain, stiffness, and functional limitation can be the only symptoms present. Preoperative C-reactive protein (CRP) demonstrates a sensitivity of 42% and specificity of 84% at a cutoff value of 10 mg/L, although its accuracy is influenced and reduced by several factors [7]. Synovial fluid analysis has a high incidence of “dry taps” (44%), with sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) insufficient to reliably rule out infection [8].

Bacterial culture, considered the gold standard for diagnosing periprosthetic infections in other joints, shows a sensitivity of 50%–67% in shoulder prosthesis infections, depending on the bacterium, and a specificity of 100% [9]. C. acnes, which is fairly common (38.9%), is the most difficult to identify on culture, often requiring 2 or 3 weeks of incubation [10].

The sensitivities of sonicate fluid (i.e., fluid obtained by sonication of prosthetic components to dislodge bacteria into a liquid for microbiological analysis) vary between 36%–75% when no cutoff values are used, and the specificity is 82%, which is lower than the specificity of bacterial culture on periprosthetic tissues [11,12]. Intraoperative frozen section analysis showed a sensitivity of 47%–57% and a specificity of 100%. Since 50% of infection cases are associated with C. acnes, the criterion of 7–10 polymorphonuclear neutrophil in 5 high-power fields was proposed to increase sensitivity to 72% [12,13]. As a result, new biomarkers have been sought to detect periprosthetic infection with better accuracy in order to provide additional decision-making tools. In cases of painful shoulder prostheses, conclusive results might not be obtained from standard tests before and during surgery that would enable classification according to the Musculoskeletal Infection Society (MSIS) criteria [14,15].

Alpha-defensin has been included among the diagnostic biomarkers of periprosthetic joint infections (PJIs). It is an antimicrobial peptide released into the synovial fluid by neutrophil granulocytes when stimulated by pathogens [16,17]. Its diagnostic accuracy for periprosthetic infections in the hip and knee has been reported to be high in the literature [18,19]. Alpha-defensin can be quantitatively analyzed using enzyme-linked immunosorbent assay (ELISA) in the laboratory, which shows sensitivity values of 63%–75% and specificity of 95%–96%, but has a response time of 4–5 hours and high cost (around $760) [20]. Alternatively, α-defensin can be analyzed qualitatively with an alpha-defensin lateral flow (ADLF) test, which has the advantage of rapid response time (about 10 minutes), making it useful for both preoperative and intraoperative decision-making [21]. Moreover, some recent reports have suggested that there are differences in sensitivity and specificity between qualitative and quantitative analysis of α-defensin [22]. In hip and knee PJIs, qualitative α-defensin testing achieves a sensitivity of 69% (46%–92%) and a specificity of 94% (86%–100%) [21].

To the best of our knowledge, only two reports, Frangiamore et al. [23] and Weigelt et al. [17], have assessed ADLF accuracy in diagnosing shoulder PJI. Currently, the sensitivity values (60%–63%) are not ideal for detecting all infections, but the test remains an adjunct criterion in surgeons’ decision-making processes. There is a lack of data in the literature to support ADLF testing in shoulder PJI.

The aim of this study is to assess the diagnostic value of ADLF testing in a group of patients with suspected shoulder PJI, as well as to evaluate its usefulness and correlation with other diagnostic tests in the decision-making process. The choice of this test is based on its validity described in researches on Knee and Hip PJI, on the rapid response times and on the possibility of having an additional tool needed to make a decision.

METHODS

The study was performed in accordance with the ethical standards of the 1964 Declaration of Helsinki and in compliance with the Health Insurance Portability and Accountability Act (HIPAA) regulation. The study was approved by the Ethical Committee of Fondazione Poliambulanza Istituto Ospedaliero on December 1, 2022 (No. NK5022). Each patient provided written informed consent for enroll­ment in this scientific study.

This retrospective study was conducted between 2018 and 2023. All patients who had undergone periarticular aspiration in the presence of a previous shoulder prosthesis were enrolled, following the internal register of the product used. Demographic data were collected. Alpha-defensin was measured using an ADLF test kit (Synovasure, Zimmer Biomet) when there was suspicion of periprosthetic infection in cases of painful or suspected loosening of shoulder prostheses.

Preoperative or intraoperative shoulder aspiration was performed on all patients with painful shoulder arthroplasty, as well as prior to revision arthroplasty due to early loosening. The test required a minimum of 15 μL of synovial fluid obtained from joint aspiration, which was then mixed with a reagent and applied to ADLF testing. In this study, included cases were classified according to the MSIS criteria before and during surgery, as well as upon completion of postoperative cultures. According to the MSIS criteria [15], the included cases were preoperatively categorized as follows: (1) highly suspected PJI (presence of a sinus tract or at least three minor criteria), (2) no suspected PJI (absence of major criteria or fewer than three minor criteria), or (3) undetermined (incomplete preoperative diagnostics, such as non-interpretable synovial leukocyte counts).

Patients with insufficient synovial fluid aspiration to perform ADLF testing, contamination of synovial fluid with blood, presence of a sinus tract, and/or those who had recently received antibiotic therapy were excluded from this analysis. The ADLF test is a qualitative colorimetric test; therefore, the presence of blood in synovial fluid causes false positives as described in the technical data sheet. The exclusion criteria for performing the test can reduce sample size. A smaller sample size results in less reliable outcomes that are more susceptible to random variation, increasing the risk of selection bias. Therefore, in such cases, it becomes more difficult to draw conclusions that are applicable to larger populations. All other tests, including CRP, synovial fluid analysis, intraoperative biopsy for frozen section analysis, microbiological analysis, and sonicate fluid obtained from the removed prosthesis, were performed routinely. ADLF testing was performed by another trained operator blinded to the outcomes of the bacterial culture and frozen section analysis.

The type of surgery performed was also recorded, specifically whether it was a one-stage revision, two-stage revision, or DAIR (debridement, antibiotics and implant retention). If reimplantation was carried out, the type of implant was assessed, and whether it was an anatomical arthroplasty, reverse arthroplasty, endoprosthesis, or cuff tear arthroplasty. The final diagnosis of shoulder PJI was confirmed or ruled out based on the MSIS criteria, considering the results of microbiological analysis and positive histological findings of acute inflammation from intraoperative biopsies. Fig. 1 summarizes the perioperative steps from clinical and diagnostic evaluation to surgery.

Fig. 1.

Flowchart of diagnostic steps of this study. ADLF: alpha-defensin lateral flow, CT: computed tomography, DAIR: debridement, antibiotics and implant retention.

Statistical Analysis

Continuous variables were reported as mean and standard deviation (SD), while categorical variables were expressed as frequency distributions and percentages. According to the final PJI diagnosis based on the MSIS criteria, the ADLF test was classified as either true-positive, true-negative, false-positive, or false-negative. The diagnostic value of the ADLF test was evaluated by calculating its sensitivity, specificity, positive and NPVs, and accuracy. A 95% CI was determined for these metrics. All statistical analyses were performed using the Stata software package 14.2 (StataCorp.), Multi-Processor version, on the macOS platform. Since this was a retrospective study, no formal sample size calculation was performed and non-parametric tests were used.

RESULTS

From June 2018 to September 2023, 52 consecutive patients underwent shoulder arthrocentesis due to suspected PJI and were subsequently enrolled in the study. Two patients were excluded due to ongoing antibiotic treatment, three due to previous positive culture diagnoses, and 25 patients were excluded due to shoulder aspirate reasons, of which 17 were excluded due to an insufficient amount of aspirate and 8 due to bloody aspirate. The final study group therefore comprised 22 patients. Of these, eight were men (36.4%) and 14 were women (63.6%). The average age at the time of diagnostic arthrocentesis was 77.1 years (SD, 5.7; range, 67–91).

All patients underwent surgery. Specifically, 17 patients (77.3%) underwent prosthesis revision, with six cases involving the implantation of cuff tear arthroplasty (27.3%) and one case involving a resection arthroplasty (4.6%). Two patients (9.1%) underwent removal of the previous implant with the placement of a spacer, two patients (9.1%) underwent the second stage of a two-stage procedure with spacer removal and the implantation of a new shoulder arthroplasty, while one patient underwent a DAIR procedure (4.6%).

Of the 22 Synovasure (Zimmer Biomet) tests performed, 17 cases (77.3%) were negative, consistent with the other tests; in five cases, the Synovasure test was positive (22.7%). Of the five positive cases, two were not considered infected due to the presence of pigmentation around the shoulder, minute pseudotumor, implant loosening and high levels of metal ions in the blood typical of adverse reaction to metal debris, as well as due to superficial surgical wound infection/dehiscence with negative periprosthetic tissues and sonicate fluid bacterial culture. Of the remaining three cases (13.6% of the total sample), one case was consistent with the other tests, all of which were positive; while two cases were not consistent with the other examinations performed.

Specifically, one case had a positive ADLF test, elevated preoperative CRP (22.6 mg/L, cutoff 5 mg/L), a frankly infected macroscopic appearance of the tissues, and negative tissue cultures. The patient was treated with a two-stage procedure. The other case had a positive ADLF test, with negative preoperative and intraoperative tests and negative tissue cultures, but postoperatively, there was a positive culture only after sonication. The patient underwent one-stage revision. Staphylococcus capitis (patient no. 2) and Staphylococcus aureus (patient no. 20) were isolated. Table 1 provides details on the tests performed on the patients.

Table 1.

Tests performed for the diagnosis of PJI

Patient No.	Preoperative CRP (mg/L; cutoff 5.0)	Histology on intraoperative biopsy	Culture on intra-articular fluid	Culture on tissue samples	Culture on prosthesis sonication	Synovasure	PJI according to MSIS criteria
1	3.3	Negative	Negative	Negative	Negative	Negative	Not infected
2	0.5	Negative	Negative	Negative	Positive	Positiveb)	Possible infected
3	22.9	Negative	Negative	Negative	Negative	Negative	Not infected
4	16.6	Negative	Negative	Negative	Negative	Negative	Not infected
5	2.4	Negative	Negative	Negative	Negative	Negative	Not infected
6	44.6	Negative	Negative	Negative	Negative	Negative	Not infected
7	2.7	Negative	Negative	Negative	Negative	Negative	Not infected
8	0.6	Negative	Negative	Negative	Negative	Negative	Not infected
9	7.1	Negative	Negative	Negative	Negative	Negative	Not infected
10	2.8	Negative	Negative	Negative	Negative	Negative	Not infected
11	1.2	Negative	Negative	Negative	Negative	Negative	Not infected
12	45.2	Negative	Negative	Negative	Negative	Positivea)	Not infected
13	9.7	Negative	Negative	Negative	Negative	Negative	Not infected
14	9.0	Negative	Negative	Negative	Negative	Negative	Not infected
15	4.7	Negative	Negative	Negative	Negative	Negative	Not infected
16	22.8	Negative	Negative	Negative	Negative	Positivea)	Possible infected
17	34.9	Negative	Negative	Negative	Negative	Negative	Not infected
18	8.4	Negative	Negative	Negative	Negative	Negative	Not infected
19	13.9	Negative	Negative	Negative	Negative	Negative	Not infected
20	278.4	Positive	Positive	Positive	Positive	Positiveb)	Infected
21	4.0	Negative	Negative	Negative	Negative	Negative	Not infected
22	18.5	Negative	Negative	Negative	Negative	Positivea)	Not infected

PJI: periprosthetic joint infection, CRP: C-reactive protein, MSIS: Musculoskeletal Infection Society.

^a)The positive cases of the ADLF test. The discordant cases—where the ADLF test was positive but the other tests were negative—are numbers 2, 12, 16, and 22.

^b))The only case where all tests showed positive agreement is case number 20.

In the 19 patients without a diagnosis of PJI (84.6%), the most common cause was aseptic glenoid loosening (eight out of 19 cases, 42.1%), followed by recurrent dislocation (five cases, 26.3%), aseptic humeral loosening (three cases, 15.8%), and painful prosthesis without a specific diagnosis (three cases, 15.8%). At the last clinical and radiographic follow-up, at 37.4 months (SD, 26.2; range, 12–61), none of these patients showed any signs of infection. The ADLF test was false-negative in zero patients and false-positive in two patients, resulting in a sensitivity of 100.00% (95% CI, 100.00%–100.00%) and a specificity of 89.47% (95% CI, 76.65%–102.30%). The PPV was 60.00% (95% CI, 39.53%–80.47%) and the NPV was 100.00% (95% CI, 100.00%–100.00%).

The accuracy of the ADLF test was calculated based on the true positives (n=3) and true negatives (n=17) out of the total number of cases (n=22), resulting in a value of 90.9%. Based on the results obtained, the values of sensitivity, specificity, predictive values, and 95% CIs were calculated for each diagnostic test. These data are presented in Table 2.

Table 2.

The values of sensitivity, specificity, and predictive values (95% CIs) for each diagnostic test performed on all 22 patients

	Sensitivity (%)	Specificity (%)	Positive predictive value (%)	Negative predictive value (%)
Synovasure	100.00 (100.00–100.00)	89.47 (76.65–102.30)	60.00 (39.53–80.47)	100.00 (100.00–100.00)
C-reactive protein	66.67 (46.97–86.37)	42.11 (21.47–62.74)	15.38 (0.31–30.46)	88.89 (75.76–102.02)
Histology	33.33 (13.63–53.03)	100.00 (100.00–100.00)	100.00 (100.00–100.00)	90.48 (78.21–102.74)
Culture on tissue samples	33.33 (13.63–53.03)	100.00 (100.00–100.00)	100.00 (100.00–100.00)	90.48 (78.21–102.74)
Culture on prosthesis sonication	66.67 (46.97–86.37)	100.00 (100.00–100.00)	100.00 (100.00–100.00)	95.00 (85.89–104.11)

These results are based on a sample of only 22 patients and are characterized by wide CIs, thus should be interpreted with caution.

DISCUSSION

The key finding of this study is the diagnostic performance of the ADLF test (Synovasure, Zimmer Biomet) for shoulder PJI. In the analyzed samples, ADLF testing demonstrated good intraoperative diagnostic accuracy, correctly identifying two out of three cases of infection at the conclusion of the diagnostic process. Specifically, the test achieved a sensitivity of 100%, specificity of 89.5%, PPV of 60%, and NPV of 100%. Furthermore, the accuracy was notably high at 90.9%. However, it is important to note that the low incidence of infection, often presenting as low-grade cases, limits comprehensive evaluation of the test’s performance. Nevertheless, the ADLF test can be used as an additional parameter to rule out aseptic mobilization or instability during the preoperative and intraoperative period, given the unreliability of other tests.

Caution should be exercised when drawing definitive conclusions. It is important to consider the limitations of this study, which involved small sample sizes and wide confidence intervals. Even if the test shows a sensitivity or NPV of 100%, with a limited sample, this estimate could be due to chance and may not reflect the true performance of the test in a larger population. Sensitivity and NPV values with wide confidence intervals are not reliable, as they could vary significantly.

In a review conducted by Marson et al. [24], which included 11 studies, the pooled sensitivity and specificity of different α-defensin test systems used for diagnosing PJI were assessed. The combined sensitivity of the laboratory α-defensin test was found to be 0.95, with a pooled specificity of 0.97 based on four studies that used a threshold level of 5.2 mg/L. In comparison, the pooled sensitivity for the lateral flow cassette test was 0.85, with a pooled specificity of 0.90. There was a significant difference in sensitivity (P=0.019), but not in specificity (P=0.47). The authors concluded that the lateral flow cassette test demonstrated significantly lower performance, and its pooled results were comparable to those of the leukocyte esterase test [24].

Many authors have studied α-defensin as a biomarker for hip and knee PJIs, and it has been shown to have the highest diagnostic accuracy ratio [25]. This biomarker is released by activated neutrophils, with a sensitivity of 100%, a specificity of 96%, and an accuracy of 96% [26]. Synovial α-defensin is an antimicrobial peptide that can be measured by ELISA or ADLF testing and has been included in the European Bone and Joint Infection Society definition of PJI [27]. Although the ADLF device has been reported to have poor sensitivity, it offers the advantages of being rapid and possessing high specificity for diagnosing PJI [25]. However, an important drawback of this test is its high cost ($93). Additionally, doubts may arise when synovial fluid is not obtainable or is contaminated with blood [25].

The MSIS criteria help categorize whether a painful arthroplasty is infected. A false-negative final outcome leads to a missed diagnosis of infection and therefore one-stage surgical treatment, while a false-positive final outcome leads to overtreatment, that is, unnecessary two-stage surgical treatment. Performing this additional rapid test alongside standard examinations provides an extra piece of information when deciding how to treat a painful prosthesis. This is especially important because infections caused by low-virulence microorganisms, such as those that can occur in shoulder arthroplasty, might not be detected by conventional tests. With reference to the pathogens isolated in the patients included in our study, S. capitis, a coagulase-negative staphylococcus, has documented potential for causing human disease and nosocomial spread. A study by Tevell et al. [28] highlighted that biofilm-forming ability is common in S. capitis, and its ability to invade cells, despite its low virulence potential, may contribute to its persistence within host cells. S. aureus, one of the most frequently isolated organisms in PJI, is a Gram-positive bacterium commonly found as a commensal on the skin and in the upper airways [29]. It can become an opportunistic pathogen when influenced by certain host factors, environmental conditions, or bacterial interactions. Staphylococcus species can form biofilms on foreign materials, encasing themselves in a polymeric matrix. This biofilm formation combined with a slow growth rate provides significant protection against both the host's immune system and antimicrobial treatments [29].

Shoulder PJI remains a diagnostic challenge due to the indolent nature of commonly isolated organisms, such as C. acnes. The most commonly used preoperative and intraoperative predictors of infection perform poorly, especially when compared with those for hip and knee PJI [30,31]. CRP is typically the first biomarker evaluated; however, it has not proven to be a reliable diagnostic test. Altered CRP values are not always suggestive of infection and often do not align with other diagnostic tests, except when CRP levels are extremely high [30].

Slow-growing Gram-positive anaerobic bacteria (SGAB) such as C. acnes are increasingly recognized as causative agents of implant-associated infections (IAIs) in shoulder arthroplasty. Diagnosing SGAB IAIs is difficult due to their non-specific clinical and laboratory findings as well as the highly specific growth conditions required for these bacteria [27]. Infections caused by low-virulence microorganisms can result in false-negative outcomes [32]. In our experience, a case of false-negative tissue culture was observed with S. capitis. Standard diagnostic methods may not reliably detect infections, particularly low-grade ones [33]. In this instance, neither white blood count nor serum CRP level was elevated. No microorganisms were detected in the standard synovial fluid culture; however, detection was successful using blood culture bottles. Studies on C. acnes shoulder PJI have indicated that synovial α-defensin measured by ELISA may play a role in diagnosing C. acnes IAI. However, Weigelt et al. [17] and Unter Ecker et al. [20] found that the ADLF testing failed to identify the only culture-positive C. acnes shoulder PJI in their study.

In contrast, the synovial ADLF test can yield false positives in cases of aseptic inflammatory conditions such as rheumatoid arthritis, calcium pyrophosphate deposition disease, and psoriatic arthritis. Synovial fluid in these cases may show elevated neutrophils, similar to PJI, or adverse local tissue reactions due to failed metal-on-metal bearings or corrosion at baseplate-glenosphere junction due to modularity [33,34]. Superficial surgical wound complications are among the most common reasons for an additional intervention after total shoulder arthroplasty. Patients undergoing revision as treatment of these complications benefit significantly as long as the prosthesis remains in place [35]. The PPV value of this analysis is relatively low compared to the specificity value because there are two false positive cases identified intraoperatively due to inflammation by release of metal ions for fretting damage and corrosion at modular connections and by surgical wound complications.

Therefore, we conclude that the ADLF test alone may lead to underdiagnosis or overdiagnosis, resulting in inappropriate treatment. This conclusion aligns with findings from two other studies available in the literature [17,23] that have a larger sample size. In our brief experience, the ADLF test, considering its limitations, can be included in a multiparametric evaluation and help make surgical decisions; its NPV makes it particularly helpful for ruling out shoulder PJIs. It can be a useful tool in the emergency department for acute assessment of suspected PJI, as well as for intraoperative use during revision surgery without septic suspicion. A potential limitation of this test is the high cost, which does not allow its availability in all hospitals.

This study has some limitations. It was a retrospective single cohort study without a control group, for example ELISA or other biomarkers analysis. The sample size was small, and the study was not multicentric. Furthermore, the test requires effective arthrocentesis, both in terms of sufficient volume and the absence of blood traces. Additionally, stratification for different pathogens and their relative virulence was not performed due to an insufficient sample size and will be necessary in future research. Compared to other studies in the literature, the findings are still relevant. The low incidence of shoulder PJI and the small sample size highlight the need for larger-scale studies to further evaluate the test's performance. Further prospective, multicenter studies are necessary to assess the diagnostic impact of the ADLF test on shoulder PJI and to identify other promising biomarkers both in the serum and in the synovial fluid.

CONCLUSIONS

In conclusion, this retrospective study evaluated the ADLF test's diagnostic value in patients with painful or unstable shoulder arthroplasty. While the ADLF test is helpful in clinical practice, it is insufficient on its own and should be used to complement traditional tests. It is potentially a useful preoperative and intraoperative diagnostic tool, particularly for ruling out infection. Low-grade infections remain challenging, and intraoperative decisions should rely on frozen section analysis, tissue appearance, and preoperative CRP values.