This study investigates an important and hitherto unaddressed issue of patient synovial fluid composition and its impact upon implant wear and risk of failure. Furthermore, comparisons were made between patient synovial fluid and the accepted lubricant for validation.
NMR spectroscopy distinguished small molecule differences between HSF and 25BCS. HSF samples contained an altered lipid profile, with lower levels of choline and 3-hydroxyisobutyrate. HSF metabolic products lactate, glucose and creatine were also reduced. Amino acids alanine, valine, glutamine and leucine were elevated.
When correlating the wear testing studies to the NMR spectroscopy of the fluids, significance was detected for both macromolecules and small molecules. However, all wear correlations disappear once the macromolecule signal is suppressed. Therefore, it is possible that these significant small molecules: amino acids (glutamine, glycine, threonine, asparagine, histidine, proline and tyrosine), glucose and citrate are bound to macromolecules.
Correlations to wear were seen with the NMR peaks from GAG and the N-acetyl groups. Prior reports have demonstrated these signals originate from proteoglycan breakdown in synovial fluid(40–44). Therefore, it is hypothesised that a proteoglycan breakdown and possibly GAG is directly correlated to material wear.
Metabolic differences between 25BCS and HSF
A large variation in osteoarthritic HSF small molecule composition was seen. Previously comparisons had only studied macromolecules(45). The variation in HSF composition demonstrated by the PCA analysis was consistent with other NMR descriptions of the osteoarthritic populations(44).
Synovial fluid is formed from an endocapillary filtrate of blood serum(46). Therefore, HSF and 25BCS fluids are expected to share composition and be different. This is relevant in scenarios of inflammation, like osteoarthritis, when there is increasing capillary wall permeability. Scenarios driving selective capillary wall permeability, active transport to or from the synovial fluid or metabolic processes occurring within the joint itself could cause differences between the fluids(47).
The presence of active joint inflammation reduces the oxygen tension and yields an anaerobic state(48, 49). Animal studies of osteoarthritic synovial fluid compared to healthy controls found an elevation in anaerobic metabolism, fatty acid chain oxidation and proteoglycan degradation(50). The free oxygen of bovine and human arterial blood is similar(51, 52). Therefore, it was predicted that osteoarthritis HSF samples would be anaerobic compared to a non-diseased 25BCS sample. Despite this, 25BCS samples had consistently higher levels of lactate but lower pyruvate. This may be due to a lack of pyruvate production from glycolysis with concomitant consumption to lactate.
It is possible the industrial processes of harvesting, filtering (> 0.2µm), and packaging commercially available 25BCS have ongoing cellular metabolism thereby increasing anaerobic metabolites. However, due to centrifugation prior to storage, HSF purposefully will not. This is so HSF will metabolically mirror the in vivo human joint as best as possible. It was not possible to mimic the processes of 25BCS production in this study.
The HSF samples contained higher concentrations of glucose and creatine. Furthermore, the lipid peaks in 25BCS provided stronger NMR peaks from the (CH2)n moiety(41). Higher levels of choline and 3-hydroxyisobutyrate in the 25BCS samples suggest a destruction of complex lipid species and fatty acid metabolism(50). Alternatively, it could indicate a high level of gluconeogenesis.
The presence of proteoglycan in the HSF is supported, although not confirmed, by the observation of the N-acetyl moiety (2.03ppm). A larger N-acetyl signal has been indicative of proteoglycan destruction, including hyaluronan(50). Free amino acids alanine, glutamine, leucine and valine were elevated in the HSF compared to the 25BCS model, possibly due to proteinolysis. Changes in valine have been reported due to prolonged low temperature storage, but it remains unclear if this is the case(53).
Isopropyl alcohol was found within the 25BCS samples. This contaminant has been seen from skin swabbing but the presence within the 25BCS cannot be explained. The effect upon lubrication is not known.
The 25BCS model fluid contained more acidic anaerobic metabolites: lactate, pyruvate and 3-hydroxybutyrate. Macromolecule surface binding of lipid and hyaluronan has been shown to be effective in reducing implant wear(54, 55). It is possible that the acid metabolites could affect the surface binding of charged species, protein conformation or protein aggregation, thereby limiting lubrication. In addition, the acidic species could propagate corrosion wear of the implant(56). Whilst synovial fluid contains an inherent buffering capacity, the measurement of the pH before and after wear testing has shown the potential for disturbance with a consistent increased alkalinity(9).
The correlation between implant wear and the metabolic profile
When testing wear using HSF and relating the AWD back to the fluid composition by NOESYPR1D NMR spectroscopy, several small and macromolecules were significant. However, the CPMG spectra universally showed no significance, presumably due to the suppression of macromolecule signals. The loss of significant small molecules when changing from the NOESYPR1D spectra to the CPMG implies that small molecules are bound or complexed in fast exchange with macromolecules. This effect has been described in plasma with binding to human serum albumin and the potential exists within HSF due to its large proteoglycan and GAG content(57).
Several amino acids positively correlated with measured CoCrMo wear but their change in response to joint disease is known to be variable. The design of this sudy was limited by the use of HSF only from OA patients with no control samples available. The reports of these metabolites in synovial fluid suggest that their detection decreases with osteoarthritis, although, glycine, asparagine, proline and tyrosine are reported to increase. There are no reports of changing levels of lysine or histidine with joint disease in HSF. The significance of these metabolites is unclear. Cartilage NMR spectroscopic studies have suggested alteration of glycine levels correlating with proteoglycan destruction and collagen destruction(50).
The presence of significant N-acetyl (δ = 2.02ppm) and GAG (δ = 7.98ppm) peaks indicates that GAG plays a role in lubrication of this specific wear test. HSF is known to contain several GAG molecules including: chondroitin sulphate, keratan sulphate and hyaluronan. It has been postulated that these molecules provide lubrication in the native joint; hyaluronan joint supplementation by injection being an accepted osteoarthritis treatment(58–60). It is indicated that one shoulder of the N-acetyl peak (δ = 2.02ppm) is significant, rather than the whole peak, suggesting a specific contributor. It is not possible using these methods to specifically identify which proteoglycan species is relating to CoCrMo wear. However, it seems clear that GAG macromolecule destruction is responsible.
The significance of increasing glucose, citrate and pyruvate with CoCrMo wear may be related to metabolic stress and changing energy utilisation within the samples due to joint disease. The level of glucose has been shown to both increase and decrease in joint disease(50, 61). In addition, citrate and pyruvate have also been demonstrated to consistently increase with joint disease and this change was attributed to anaerobic metabolism(50). Whilst these metabolites alter in joint disease and relate directly to metabolic stress, it is unclear what the correlation with prosthetic wear is attributable to. It is possible that these metabolites may confound with another lubricating component of HSF, increased in metabolic stress and/or advanced joint disease. Alternatively, the anaerobic environment of the joint may have a direct relationship with the lubrication properties of the HSF due to alteration of the pH or molecular interactions. However, this seems unlikely since the other detectable anaerobic metabolic markers such as lactate and formate were not significant.
An alternative hypothesis is that the glucose and citrate may be bound to a significant lubricating macromolecule within the HSF likely to be protein and GAG in origin as evidenced by our findings.
We aim to study the macromolecular structure of the significant products that play a key role in implant wear. In addition, we aim to substitute alternative bearing surfaces such as metal-on-poly, ceramic-on-ceramic or ceramic-on-poly, to investigate if the same metabolites are key to measured wear.