Endotoxin Accumulation during Particle-induced Osteolysis Is Due to Systemic Endotoxin Rather than Surgical Contamination

*Tatro, JM;*Goldberg, VM; **Stewart, MC; +*Greenfield, EM *Case Western Reserve University and University Hospitals, Cleveland, OH; ** Department of Veterinary Clinical Medicine, University of Illinois, Champaign-Urbana, IL [email protected] INTRODUCTION: Many laboratories have shown that bacterial endotoxin increases the biological responses induced by orthopaedic wear particles [1]. For example, removal of adherent endotoxin from titanium particles significantly reduces their ability to stimulate cytokine production and osteoclast differentiation in vitro as well as osteolysis in the murine calvarial model [1]. A balance between endotoxin accumulation and endotoxin clearance controls the steady-state level of endotoxin surrounding orthopaedic wear particles in the murine calvarial model [2]. However, it has not been determined whether the endotoxin accumulation is due to the accumulation of systemic endotoxin or to bacterial contamination during particle implantation. The results of the current study show that the endotoxin accumulation is not due to bacterial contamination. METHODS: Commercially pure titanium (cpTi) particles with adherent endotoxin were obtained from Johnson Matthey, (75% < 6.5um [3], 36 Endotoxin Units /10 particles). Virtually “endotoxin-free” Ti particles of the same size and shape as particles with adherent endotoxin were prepared as in [4]. Osteolysis was measured seven days after implantation of particles or PBS, as a negative control, by computer assisted histomorphometry of contact microradiographs using the quantitative version of the murine calvarial model [5]. All animal experiments were approved by our IACUC. All mice were 6-8 week old females. Fibrous tissue layers overlying the calvaria containing implanted particles were aseptically collected at sacrifice, stored in endotoxin-free tubes at -20oC, and homogenized in endotoxin-free water with homogenizers rendered endotoxin-free by baking at 260oC for 18-20 hours. The concentrations of endotoxin in the retrieved tissues and in the original particle suspensions were measured by the high sensitivity version of the spectrophotometric Limulus Amebocyte Lysate assay with the modification described in [6] to reduce false positives due to β-glucan-like molecules. All data are presented as means + SEM. Statistical analyses were by ANOVA with Fisher’s Protected LSD post-hoc tests. RESULTS: Two types of experiments were performed to determine whether endotoxin that accumulates after implantation of “endotoxinfree” particles is derived from bacterial contamination. First, we determined whether endotoxin accumulation occurs rapidly after implantation, which would be consistent with bacterial contamination. Endotoxin significantly accumulated in mice 7 days after implantation with “endotoxin-free” cpTi particles but did not accumulate after 20 minutes or 2 hours (Figure 1A, N=11). To validate these results, it was necessary to document that a substantial proportion of the implanted particles were recovered at the early time points. For this purpose, cpTi particles were retrieved from the fibrous tissue overlying the calvaria using a nitric acid digestion method that does not detectably alter the size or shape of cpTi particles [7]. These measurements show that 6080% of the cpTi particles were recovered (Fig. 1B, N=11). Together, these results rule out the possibility that a large initial bolus of bacterial contamination accounts for the observed endotoxin accumulation. An alternative possibility is that a low level of bacterial contamination might lead to substantial bacterial growth during the seven days between particle implantation and recovery. However, in multiple studies of particle-induced osteolysis, involving over 1200 mice, we have had only six cases of detectable infection. The infected mice were always excluded from further analysis. To more directly assess the possibility of bacterial growth, endotoxin accumulation was measured following particle implantation in germ-free conditions. For this purpose, germ-free mice (Taconic, Germantown, New York) were shipped and continuously housed in germ-free isolator units and all experimental manipulations were performed in the isolator units. These mice lack detectable bacteria as assessed by culture of feces, bedding, and drinking water. Cultures also showed that the isolator and all instruments were free of bacteria. Although these mice are germ-free, they are exposed to endotoxin through commercial mouse feed, drinking water, and bedding. These materials are sterilized prior to use but contain dead bacteria. Consequently, germ-free and normal flora rodents have similar plasma endotoxin levels [8]. The control groups for these experiments consisted of normal flora mice that were housed and implanted with “endotoxin-free” titanium particles in standard conditions. Similar levels of endotoxin accumulation were observed in the germ-free and control mice (Fig. 2A, N=7). Taken together, the time course experiments and the germ-free mice experiments show that the observed endotoxin accumulation is unlikely due to bacterial contamination during particle implantation. Interestingly, similar levels of osteolysis were also observed in the germ-free and normal flora mice (Fig. 2B, N=14). DISCUSSION: These results show that endotoxin accumulation during particle-induced osteolysis is due to systemic endotoxin rather than surgical contamination. Therefore, the balance between accumulation and clearance of endotoxin may regulate the rate of osteolysis in the murine calvaria model as well as in patients with aseptic loosening. In patients, systemic endotoxin [9, 10] likely derives from intestinal flora, minor infections or bacteremias following dental procedures. The possibility that endotoxin accumulation also occurs in patients with aseptic loosening is supported by the recent finding that endotoxin exists in periprosthetic tissue from a subset of patients with aseptic loosening despite the absence of clinical or microbiological signs of infection [11]. ACKNOWLEDGEMENT: Supported by NIH RO1 AR43769 to EMG