CORR Insights®: What Is the Diagnostic Accuracy of Alpha-Defensin and Leukocyte Esterase Test in Periprosthetic Shoulder Infection?

Where Are We Now?

Shoulder arthroplasty surgeons, like myself, are currently in a quandary. We have more arthroplasty options than ever before and the ability to improve comfort and function for an expansive range of glenohumeral joint maladies. Accordingly, the number of shoulder replacements being performed has soared in the last decade, a trend that is expected to continue [3]. But we perform these procedures knowing that there is an inescapable risk of infection. More than half of arthroplasties undergoing revision ostensibly for aseptic causes are caused by infection, and most of those harbor the most-problematic pathogen our patients now face: Cutibacterium acnes (C acnes; formerly Propionibacterium acnes) [8, 12].

C acnes is part of the normal skin flora, but it also is an opportunistic pathogen. Patients who present with a painful shoulder arthroplasty are at high risk of having an infection, even if they do not show overt signs of infection. Unfortunately, we lack diagnostic tools that have adequate sensitivity and specificity. Serum markers (erythrocyte sedimentation rate, C-reactive protein, and white blood cell count) are not very helpful [15], and attempts at joint aspiration are often not productive. Even when synovial fluid can be obtained, synovial markers have not been more useful than a cell count of the same fluid. And cultures can add to the confusion, as we struggle to distinguish between false-positive contaminants and false-negative results in the hopes of identifying those patients with infections.

The indicators of infection that have been found reliable for hip and knee arthroplasty have not done as well when studied for their ability to identify infected shoulders. C acnes does not generate the robust response seen with common hip and knee pathogens, presenting stealthily with pain and sometimes-subtle radiographic findings [6, 7]. The slightly painful shoulder arthroplasty that returns for a followup visit, particularly one that did well initially, compels us down a difficult diagnostic and therapeutic path with little information to guide us. Do we simply observe, settle for an attempted arthrocentesis during the clinic visit, do extensive blood work, or take the patient to surgery for an arthroscopic or open biopsy?

Does the present study [5] illuminate the way through these difficult decisions? Not exactly. But now we have two papers that address the utility of synovial alpha-defensin in predicting infection after shoulder arthroplasty—this publication [5], and an earlier one by Frangiamore and colleagues [6]. The earlier paper found a sensitivity and specificity of 63% and 95% in identifying patients that met their criteria for infection. The authors of the current study found an even greater sensitivity and specificity of 75% and 96%. Notably, the present study analyzed alpha-defensin by enzyme-linked immunosorbent assay (ELISA) test as compared to the earlier study which used Synovasure® (Zimmer Biomet, Warsaw, IN, USA). Notably, a recent meta-analysis [1] found the Elisa test superior, with the Synovasure® being no better than other synovial markers such as C-reactive protein, interleukin-6 (IL-6), and leukocyte esterase. In any event, the results reported by Ecker et al are still inferior to the near 100% values reported for the hip and knee [4, 13, 14]. With a sensitivity of 75% (alpha-defensin) and 50% (leukocyte esterase), the findings of this study only support use of these markers as confirmatory for infection when positive, but leave us scratching our heads when they are not.

Where Do We Need To Go?

First, we need to determine whether or not a shoulder is infected. The present state of the art has us hedging our diagnoses with categories like “definite infection”, “probable infection”, and “probable contaminant” [2, 7, 10]. This is understandable considering that currently, we cannot even rely on a positive culture to verify the existence of an infection; there is no gold standard. The consensus criteria of the Musculoskeletal Infection Society [11], used by the present authors, have known subjective elements and other investigators consider it insufficient to confront the challenges associated with shoulder arthroplasty [2, 6, 8]. How can we truly evaluate the integrity of any test as a proxy for an infection without solving this first?

Until we have definitive criteria for identifying the existence of an infection, we need to continue to work towards establishing and implementing consensus criteria. Fortunately, dedicated shoulder surgeons are active in this pursuit as evidenced by the most-recent work from the International Consensus Meeting on Prosthetic Joint Infection [9].

Many questions do not yet have answers, but this is likely to change as we gain a better understanding of C acnes biology. It is clear that some phylotypes are more pathogenic than others, but this is not yet clinically useful. Cytokines in the synovial fluid (primarily IL-6) have shown potential as a marker of infection [7]. Boyle et al found increased pathogenicity in hemolysis producing strains of C acnes [2] but this was contradicted by the findings of other investigators [10]. To date, advanced genetic tools (polymerase chain reaction/next generation gene sequencing) have not been productive but such is the way with evolving technologies.

How Do We Get There?

On the practical side, we need to improve our consistency in how we assess our shoulder arthroplasty patients for infection both in the clinic and the operating room. This begins with a careful history, scrutinizing radiographs, appropriate blood work, and standardizing our approach to the acquisition of synovial fluid samples from patients with a painful shoulder arthroplasty. It is well known that “blind” needle insertion in the glenohumeral joint is inconsistent even in experienced hands. Should we begin confirming our needle placement with fluoroscopy and contrast? Perhaps we should relegate these aspiration attempts to interventional radiology. There are emerging methods to handle small sample sizes that may help us overcome the “dry tap” [7]. Similarly, in the operating room, we need to standardize the material sent for culture. How many specimens from how many individual sites? It is clear that we need to consistently send this material for aerobic and anaerobic culture to be held for a minimum of 14 to 21 days, knowing that one report found that 28 days was necessary to culture the pathogenic organism [12]. Should we send control cultures from the operative field (untouched gauze) as well?

For those patients for whom we do diagnose an infection, there are unfortunately also more questions than answers in regard to the best treatment. Do we use antibiotic therapy and retain the implants? If then, which antibiotic(s), for how long and by what route? Perform a one-stage revision or a two-stage process with a spacer? For a thorough review of these and many related questions, the reader is again referred to the most-recent analysis of the International Consensus Meeting on Prosthetic Joint Infection [9] . Although many of the recommendations include the phrases, “there is insufficient evidence”, “as there is no consensus”, “the limited data suggests”, the relevant questions are well articulated, and some have even been answered.