With ageing of society as well as the advances of modern medicine and modern orthopedics, the volume of joint replacements that are performed has risen tremendously and will continue to increase [26, 27]. Inevitably, the number of complications to treat will augment simultaneously. PJI is one of the most common complications necessitating re-intervention and is associated with high morbidity and mortality. A recently published article by Kurtz et al. found PJI to be associated with a 5-year overall survival of only 71.7% for the knee, and 67.2% for the hip, respectively, putting PJI in the second leading place overall in terms of mortality rate, surpassed only by cancer [1].
Treatment of PJI aims to eradicate the infection while salvaging life and limb. In very few cases, though, therapy may fail in spite of adequate intervention, which in most cases comprises (staged) revision arthroplasty alongside pathogen-directed antibiotic coverage. In this event, the otherwise in the context of trauma applied principle of “life before limb” may have to be extended to the situation of PJI and infection control by AKA may have to be considered. George et al. [28] recently published that the incidence of AKA related to PJI in the USA almost quadrupled between 1998 and 2013. Notably, AKA in this context multiplies mortality [15, 29]. Furthermore, low functional status has to be anticipated with at best half of the surviving patients reaching the ability to walk [14, 15].
In view of the subject matter’s grave implications and its relevance to a growing number of patients, this study was conducted with the objective to find out whether or not there are patient- or treatment-associated elements that would indicate those patients at risk of ending up with AKA. And if so: are these elements accessible to treatment or could their recognition otherwise change our medical decision-making?
Previously published literature concentrated on identifying comorbidities and other patient characteristics, as well as perioperative influencing factors that are commonly associated with the development of PJI. Various comorbidities have been found linked to PJI, i.e. chronic cardiovascular, pulmonary and renal conditions, preoperative anemia, diabetes, depression and psychoses, obesity, rheumatologic disease, metastatic tumor, male gender, and higher comorbidity scores [1, 30, 31], but also the exposure to previous surgery [32]. Furthermore, measures such as antibiotic prophylaxis, surgical site preparation and operating room environment, improved control of post-operative glycaemia, appropriate management of malnutrition, preoperative anemia, and smoking cessation have been proven to minimize risk of PJI [1, 3, 31, 33]. In our daily practice of primary TKA implantation, we routinely assess individual patient risk and implement named measures in order to minimize infection rates. Nevertheless, PJI cannot always be avoided and we then find ourselves having to treat complex patients presenting this challenging complication. Using the earlier mentioned protocols, treatment is in most cases successful. Publications concerning risk factors for PJI treatment failure, and specifically influencing factors of AKA in the context of PJI, are sparse. Kurtz et al. identified male gender, heart disease and higher Charlson comorbidity index as the most strongly associated patient related factors with PJI treatment failure [1]. Son and colleagues’ results overlapped partially with these findings. Yet, his group proposed that economical motives, region of residence, as well as surgical volume of the respective institution further played a role [13].
This current study confirms that patients with a more severe comorbid profile are at higher risk of having to undergo AKA than those in better general health. Amongst our study population, the distribution of ASA scores between the AKA and the LS subgroup differed significantly (p = 0.009). In order to quantify and thereby illuminate the relationship between ASA and AKA, comparative analysis was undertaken. For reasons of statistical feasibility we looked particularly at all cases classed ASA 2 and 3 – in other words patients with mild systemic disease without functional limitations and patients with severe systemic disease presenting functional limitations. While only one out of 27 ASA 2 patients failed limb salvage, 10 out of 36 ASA 3 patients ended up with AKA, yielding a significant difference in prevalence of 4% versus 28%. ASA 3 patients were almost 12 times more likely to experience AKA than the rest (OR = 11.92). Notably, no statistic difference was found when using the Charlson comorbidity index, another tool that is commonly used for risk stratification in orthopedic surgery [13, 23, 34, 35]. While the discrepancy of our results concerning ASA and Charlson score was somewhat surprising, we learned that other authors have made similar observations. Various researchers have stated that the ASA system may be a better predictor for adverse events than the Charlson score [36, 37] which, despite its formulaic collection of number and severity of conditions, seems to lack clinical applicability.
Alcohol abuse was identified as an additional statistically significant variable. While this seems logical, the validity of this finding may be questioned due to the low number of patients.
Within our patient collective, 100 % of AKA cases were anemic at time of admission to hospital for treatment of PJI, whereas only 58 % of the LS patients (p = 0.022) presented hemoglobin parameters inferior to the threshold values defined by the World Health Organization (WHO) which are 120 g/dl for women and 130 g/dl for men [24]. The average hemoglobin was 99.9 ± 15.1 g/dl in the amputees, as opposed to 118.2 ± 19.9 g/dl in the controls (p = 0.011). Etiologically, we assume anemia of chronic disease to be the most common type of anemia amongst the studied patient collective. This assumption applies with regard to both groups. Differentiated laboratory testing for sub-diagnosis of anemia was not performed. A correlation to the duration of symptoms of infection was not demonstrated and the duration of symptoms did not significantly differ between the two groups. Notably, preoperative anemia has previously been found to increase the risk to develop PJI in patients undergoing total joint replacement [38]. Subsequently, a 2017 study by Lu et al. demonstrated - by means of a multivariate regression model directed to evaluate the effect of anemia in the context of septic revision surgery - that anemia was associated with an increased risk of complications, amongst them persistence of local deep infection, sepsis and septic shock [39]. As described above, treatment of these septic complications may, in turn, sometimes necessitate AKA in order to eliminate the primary infection site. Thus, our study illustrates one of the possible consequences of Lu’s findings and thereby underpins his conclusion to regard preoperative anemia as an important clinical risk factor in patients with PJI. We deduce that early recognition of the problem will allow for timely introduction of adequate causative treatments such as supplementation of vitamins, iron, or erythropoietin, amelioration of comorbidities that are commonly associated with anemia, and/or allogeneic blood transfusions [40]. Nevertheless, at this point we can neither imply that better pre- and perioperative blood management would have a positive impact on the outcome, nor that non-responding to treatment of anemia in patients with PJI could be an indicator of limb salvage failure.
On a qualitative level, it should be remarked that some of the amputees presented circumstances of sorts that may have rendered them particularly prone to infectious complications. Immunosuppression, as a consequence of disease and/or immunosuppressive therapy was, whilst no specific scoring was applied, widely observed amongst the AKA patients. Named by way of example, there was a polytoxicomaniac with HIV and hepatitis A/B/C in whom a monomicrobial PJI caused by candida albicans was diagnosed. As has been found, drug abuse is a predisposing factor for fungal PJI [41]. Moreover, as we have previously published, patients with illicit drug abuse and concomitant HIV and/or hepatitis present a catastrophic incidence of AKA and arthrodesis in case of knee PJI [42]. Several other patients were severely immunocompromized, one of which acutely with myelodysplastic syndrome which was diagnosed shortly after TKA implantation, and was treated by Azacitidine and steroids (3). By the time of presentation at our institution, this patient was septic and his infection had extensively disseminated from the prosthetic knee joint into the soft tissues of the lower leg leading to non-vital musculature and thus the impossibility of local infection control. Speaking of which, the importance of the integrity of the periarticular soft tissues in the context of knee joint replacement and treatment of PJI has been reported before; regarding this subject, authors have described poor wound conditions, preexisting scars or dystrophic skin as relevant risk factors for potentially devastating complications including the need for amputation [43, 44]. In our patient cohort, case 4, a young patient, who had initially received primary plastic reconstruction simultaneously with the implantation of a megaprothesis for treatment of an open fracture, and who continued to have bad soft tissue conditions, exemplifies this.
With respect to additional surgeries, we found that the frequency differed between AKA and LS patients. Statistical analysis showed a trend towards an elevated number of minor reoperations involving the skin and subcutaneous layer in the AKA group. Furthermore, we observed that the majority of the amputees had to endure an unreasonable seeming total number of surgeries - a median of eight, but up to 24 in one case. Five of them started out with an attempt of implant retention (DAIR) that later failed. In retrospect, the burden of first having to bear such a high number of surgeries in conjunction with intravenous antibiotic therapy in a hospital environment, and to then nonetheless having to undergo AKA, an intervention associated with a 1-year mortality of at least 50 to 60% [15–17, 29], seems very high. Our observations also prompt the consideration of whether DAIR is an adequate strategy for patients with severe comorbidities and immunocompromise, even in case of acute PJI. Such patients may be candidates for being assigned to the staged revision arthroplasty route of treatment, regardless of the reported symptom duration. Motivated by both this formal investigation and our clinical experience, we henceforth consider the idea of AKA as a sort of definite treatment option at an earlier stage. Along the same lines, Khanna et al., who have assessed patient satisfaction following AKA for chronically infected TKA, recommended discussing the option of AKA after a maximum of six surgeries. His group had found a high percentage of satisfaction among their seven amputees; six of which would have chosen to have the amputation sooner, given the choice [18].
This study has a number of limitations. Firstly, the study population is relatively small and heterogeneous, while the amount of examined influencing factors is vast. This is, on the one hand, due to the fact that the prevalence of AKA for treatment of PJI remains a rarity despite its aforementioned increase. On the other hand, the inclusion of a large spectrum of parameters was indispensible by reason of the complexity of the affected patients. While statistic significance and trend should therefore be viewed with some reservation, our numeric results appear nonetheless corresponding to clinical observations, clinical experience and common sense. Beyond statistics, we were also able to offer detailed qualitative clinical information and laboratory results. Consequently, we consider our results valuable to future treatment of PJI patients. A multi-center study with a larger study population could allow for confirmation of our conclusions. Secondly, we were not able to enclose the entirety of patients treated for PJI at our institution during the examination period due to incomplete patient records in the earlier years. While we included AKA from 2005 to 2015, the LS control group consisted of patients from 2009 to 2015. Hence, no prevalence of successful revision TKA, versus joint fusion, versus AKA was recordable. Thirdly, as a result of the study’s retrospective nature, the accuracy of our data is limited to the information recovered from the institution’s medical records.