Periprosthetic joint infection after shoulder arthroplasty
In early postoperative prosthesis infections occurring within three months after index surgery the causative agents are mostly
S. aureus. In delayed infections that mostly appear between three months and two years after index surgery the causative agents are often less virulent bacteria from the normal skin flora, such as
S. epidermidis [
17], or, after shoulder arthroplasty,
P. acnes [
23].
P. acnes was isolated in deep cultures obtained from 42% of shoulders operated with an arthroplasty for osteoarthritis [
16], leading the authors to speculate that this microorganism might be causative for the development of osteoarthritis. This hypothesis was contested by Maccioni et al. who found a low rate of
P. acnes in osteoarthritic shoulders undergoing total shoulder replacement [
18], a discrepancy that is possibly due to differences in the specimen collection technique.
Whatever the cause-effect-relationship between
P. acnes and glenohumeral osteoarthritis, there is little doubt that this microorganism is found in a substantial proportion of cultures obtained during either primary or revision surgery of the shoulder.
P. acnes were found in 24% of shoulders undergoing revision arthroscopy for various reasons [
9], and the same organism were isolated from deep tissue culture specimens obtained during open shoulder surgery [
20], even despite timely antibiotic administration [
19]. Male gender and younger age seem to predispose towards
P. acnes colonization [
13].
Prevention of surgical site contamination with
P. acnes seems difficult since this microorganism persists in the dermis even after surgical site preparation with chlorhexidine [
13,
15]. The efficacy of different antibiotics against
P. acnes is controversially discussed, but a survey of
P. acnes isolates from shoulders over a one-year period indicates that penicillin G as well as cephalosporins display acceptable minimal inhibitory concentrations, a finding that is consistent with earlier reports on the susceptibility of
P. acnes to different antibiotics [
4,
26]. It is equally controversial whether standard incubation times suffice for the detection of
P. acnes, or whether prolonged culture periods are required [
2,
25].
There are three different mechanisms behind the development of periprosthetic joint infection: direct intraoperative contamination, haematogenous dissemination, and reactivation of latent infection [
5]. An interdependent relationship exists between bacterial virulence, a patient’s immune status, and the local wound environment [
11]. A small bacterial inoculum during surgery must not necessarily lead to periprosthetic joint infection [
3,
11], and perioperative intravenous antibiotic prophylaxis aims at eradicating contaminating bacteria from the situs. The absence of systemic antibiotic prophylaxis is a major risk factor for the development of infection after primary hip arthroplasties [
6,
7], and – although we are not aware that similar studies exist with regard to shoulder arthroplasty – the same line of reasoning seem probable even in that joint. Thus, systemic antibiotic prophylaxis is routinely administered prior to shoulder arthroplasty, and in agreement with national routines we used cloxacillin at an adequate dose.
Our observation of a high proportion of
P. acnes in infected shoulder arthroplasties is in line with previous reports. Furthermore, the susceptibility of younger males – i.e., the type of patient that dominated the Eclipse® group – to colonization and subsequent infection may have contributed to the observation of an increased number of infections in the Eclipse® group. Our awareness of periprosthetic infections after shoulder arthroplasty made us follow a quite stringent diagnostic algorithm based on arthroscopy and acquisition of tissue cultures, but this is in agreement with literature describing infection with
P. acnes as a reason for otherwise unexplained pain after shoulder surgery [
9].
Apart from patient-related factors mentioned above another possible explanation for our observation of a higher-than anticipated rate of infections in patients operated with the Eclipse® device may also be found in the material and structure of this implant. The trunnion and cage screw of this implant are manufactured from a titanium alloy that has high biocompatibility and that is considered ostoeconductive. These desired effects are on the one hand related to enhanced bony ingrowth, on the other hand biocompatibility can also be associated with an increased potential of bacterial colonization and biofilm formation [
27]. In vitro investigations show that
P. acnes can adhere to and form biofilm on various titanium alloys, furthermore, that both bacterial adherence and the capacity for biofilm formation are dependent on surface roughness of the material in various bacterial strains, including
P. acnes [
1,
8]. Copper oxide coatings of titanium alloys seem to reduce the density of planktonic
p. acnes and also inhibit the formation of biofilm, but such coatings are still at an experimental stage [
21].
Having said that, the Eclipse® prosthesis is of course not the only implant made of titanium alloys, thus, the susceptibility of this specific implant to infection with P. acnes cannot solely be explained by the type of alloy it is manufactured from. However, the metaphyseal cage screw entirely made of a roughly textured, porous titanium alloy is unique to this type of implant, and – given the findings derived from the in vitro studies cited above – surface roughness and alloy composition exert an important influence on bacterial colonization and biofilm formation. It is tempting to speculate that the excellent biocompatibility of this crucial component of the investigated implant may simultaneously cause it to be more susceptible to colonization with P. acnes. However, further in vitro investigations using this and other types of shoulder implants, including other stemless shoulder prosthesis, are necessary before firm conclusions on this topic can be drawn. Further studies would also specify whether the material or the helical screw might influence the risk of bacterial colonization.
Strengths and weaknesses of the current study
Ours is a retrospective cohort study with all typical weaknesses inherent to such a design. Retrospective data collection from medical charts is always less reliable than prospective registration. On the other hand, by cross-validating our findings with the Swedish Shoulder Arthroplasty Register we know that one patient in our cohort was revised at a different unit, thus, we feel quite confident in the completeness of our follow-up as far as the endpoint revision surgery is concerned. There could be cases with latent infection that have not shown up with complaints at our unit, but there is no reason to assume that the resulting failure to detect infections should be unequally distributed between the two groups of patients, i.e., “Eclipse” and “Control” patients. Another advantage of our study is that we follow an entire cohort of patients operated upon during a defined period of time, with no exclusion of any patient for whatever reasons. Our unadjusted survival analysis with the endpoint “occurrence of infection” indicated a statistically significant difference between the two groups of patients. Since the gender and age proportions differed considerably between “Eclipse” and “Control” patients we performed an adjusted regression analysis, and the relative risk of infection was still 4-fold increased in “Eclipse” compared with “Control” patients. Due to the low number of events (11 infections in the entire cohort of 241 patients) confidence intervals around these risk estimates were wide, and the adjusted risk of infection was no longer statistically significantly different between the two groups. However, the occurrence of 10 infections in 102 “Eclipse” patients compared with 1 infection in the remaining 139 ”Control” cases is a finding that raises suspicions. There was a difference in gender and age proportions between the two groups. There were more men in ”Eclipse” compared to the ”Control” group, and patients in the ”Eclipse” group were younger at the time of index surgery. Therefore, the ”Eclipse” group should rather had fewer comorbidities, potentially reducing the risk of periprosthetic infection compared to the”Control” group. On the other hand, males are more prone to developing P. acnes infections, and the predominance of males in the ”Eclipse” group may make this group more susceptible to periprosthetic joint infection. Taken together, the lack of reliable data on medical comorbidities is a weakness of our study.