Differential gene expression in the periprosthetic membrane: lubricin as a new possible pathogenetic factor in prosthesis loosening

Noninvasive diagnostic of periprosthetic joint infection by urinary peptide markers: A preliminary study

Previous immunohistochemical analyses revealed altered protein expression in the periprosthetic membranes of patients with periprosthetic joint infection (PJI). Proteins are degraded to peptides that may pass the blood-kidney barrier depending on their size. The aim of this study was to evaluate if PJI can be diagnosed based on the urinary peptide excretion pattern. Thirty patients undergoing removal of their hip or knee prostheses due to septic or aseptic loosening were enrolled. Specimen sampling was performed according to the MusculoSkeletal Infection Society criteria. A urinary sample was analyzed before surgery using capillary electrophoresis coupled with mass spectrometry. Peptides with differential urinary excretion between groups were used to establish a multimarker model. A total of 137 peptides were differentially excreted between the septic and aseptic groups. The majority of identified peptide markers were fragments of structural extracellular matrix proteins, potentially due to their origin from the periprosthetic membrane. A marker model with 83 peptides achieved the best diagnostic performance for diagnosing PJI with a sensitivity of 95%, a specificity of 90%, and an area under the curve of 0.96. The altered protein expression pattern in the periprosthetic membrane of PJI is most likely reflected in excreted urinary peptides. Thus, the diagnosis of PJI by urinary peptide markers seems to be a reasonable approach.

Transcriptome-Wide High-Density Microarray Analysis Reveals Differential Gene Transcription in Periprosthetic Tissue From Hips With Chronic Periprosthetic Joint Infection vs Aseptic Loosening

BACKGROUND

Differentiating between periprosthetic hip infection and aseptic hip prosthesis loosening can be challenging, especially in patients with chronic infections. This study used whole-genome microarray analysis to investigate the transcriptomes of periprosthetic hip tissues to identify genes that are differentially transcripted between chronic periprosthetic hip infection and aseptic hip prosthesis loosening.

METHODS

In this pilot study, a total of 24 patients with either chronic periprosthetic hip infection (n = 12) or aseptic hip prosthesis loosening (n = 12) were analyzed. Periprosthetic hip infection was diagnosed based on modified criteria of the Musculoskeletal Infection Society. To evaluate differences in gene transcription, whole-genome microarray analysis was performed on the mRNA of periprosthetic tissue.

RESULTS

Microarray analysis revealed differential gene transcription in periprosthetic hip tissue affected by chronic hip infection vs aseptic hip prosthesis loosening. A total of 39 genes had area under the curve values greater than 0.9 for diagnosing chronic periprosthetic hip infection; 5 genes had annotations relevant to infection and metabolism. The 39 genes also included 7 genes that were differentially transcribed but that have no apparent connection to immune response processes plus 27 genes with unknown function.

CONCLUSION

Differences in gene transcription profiles might represent novel diagnostic targets that can be used to differentiate between chronic periprosthetic hip infections and aseptic hip prosthesis loosening. Secondary metabolites of differentially transcripted genes might serve as easily accessible markers for detecting chronic periprosthetic joint infection in future.

Diagnostic accuracy of bone turnover markers as a screening tool for aseptic loosening after total hip arthroplasty

Aseptic loosening is the most common cause of prosthesis failure after total hip arthroplasty (THA). We measured serum cross-linked carboxyterminal telopeptide of type I collagen (ICTP), tartrate-resistant acid phosphatase 5b (TRAP5b), dickkopf-1 (dkk-1) and sclerostin; and urinary α isomer of C-terminal cross-linked telopeptide of type I collagen (αCTX-I) to investigate their potential diagnostic value detecting aseptic loosening after THA. Biomarkers were measured in 24 subjects with aseptic loosening of THA versus 26 control subjects without loosening after THA. Serum ICTP in the loose group (7.04 ng/mL) was higher than controls (5.15 ng/mL), (p = 0.0007). ROC analysis demonstrated that a serum ICTP >5.5 ng/L had a sensitivity of 91% and specificity of 69% for detecting aseptic loosening (area under ROC curve = 0.77, p = 0.0001), resulting in a positive predictive value (PPV) of 73% and a negative predictive value (NPV) of 90%. Serum TRAP5b in the aseptic loosening group (4.17U/L) was higher than controls (3.44 U/L), (p = 0.03). A serum TRAP5b >2.46 U/L had sensitivity of 100% and a specificity of 31% to detect aseptic loosening (AUC 0.67, p = 0.031), resulting in a PPV of 57% and a NPV of 100%. Serum dkk-1, serum sclerostin and urinary αCTX-I were not elevated in subjects with aseptic loosening (p>0.05). Serum ICTP and TRAP5b show potential utility as screening biomarkers for excluding aseptic loosening, because of their ability to discriminate individuals without disease. Our finding of elevated ICTP, generated by the action of matrix metalloproteinases, suggests a role for this group of endopeptidases in aseptic loosening.

Lubricin and smooth muscle α-actin-containing myofibroblasts in the pseudomembranes around loose hip and knee prostheses

The objective was to evaluate the presence and distribution of the lubricating and anti-adhesion glycoprotein lubricin and cells containing the contractile isoform smooth muscle α-actin (SMA) in pseudomembranes around loose hip prostheses. Periprosthetic tissue was obtained at revision arthroplasty of eight aseptic, loose hip implants, and for comparison three loose knee prostheses. Immunohistochemical analysis was performed in 3 zones: zone 1, within 300μm of the edge of the implant-tissue interface; zone 2, between zones 1 and 3; zone 3, within 300μm of the resected/trimmed edge. The presence of lubricin was extensive in all samples: (1) as a discrete layer at the implant-tissue interface; (2) within the extracellular matrix (ECM); (3) intracellularly. There was significantly more high grade (>50%) lubricin surface staining at the implant-tissue interface compared with the resected edge. While there was also a significant effect of location of high grade ECM lubricin staining, there was no significant effect of implant type (i.e. hip versus knee). All but two hip pseudomembrane samples showed the presence of many SMA-containing cells. There was a significant effect of location on the number of SMA-expressing cells, but not of implant type. These findings might explain why the management of loose prosthesis is so challenging.

Lubricin in human breast tissue expander capsules

Capsular contraction is the most common complication of breast reconstruction surgery. While presence of the contractile protein alpha smooth muscle actin (α-SMA) is considered among the causes of capsular contraction, the exact etiology and pathophysiology is not fully understood. The objective of this study was to investigate the possible role of lubricin in capsular formation and contraction by determining the presence and distribution of the lubricating protein lubricin in human breast tissue expander capsules. Related aims were to evaluate select histopathologic features of the capsules, and the percentage of cells expressing α-SMA, which reflects the myofibroblast phenotype. Capsules from tissue expanders were obtained from eight patients. Lubricin, at the tissue-implant interface, in the extracellular matrix, and in cells, and α-SMA-containing cells were evaluated immunohistochemically. The notable finding was that lubricin was identified in all tissue expander capsules: as a discrete layer at the tissue-implant interface, extracellular, and intracellular. There was a greater amount of lubricin in the extracellular matrix in the intimal-subintimal zone when compared with the tissue away from the implant. Varying degrees of synovial metaplasia were seen at the tissue-implant interface. α-SMA-containing cells were also seen in all but one patient. The findings might help us better understand factors involved in capsule formation.

Adsorption of albumin and sodium hyaluronate on UHMWPE: a QCM-D and AFM study

The biotribological properties of artificial joints, in particular the efficiency of the lubrication, strongly determine their lifetime. The most commonly used artificial joints combine a metallic or ceramic part articulating against a ultra high molecular weight polyethylene (UHMWPE) counterface, and are lubricated by the periprosthetic fluid. This fluid contains several macromolecules, namely albumin and sodium hyaluronate (NaHA), that are known to be involved in the lubrication process. There are several studies in the literature concerning the interaction of the referred macromolecules with ceramic or metallic prosthetic materials. However, to our knowledge, information about their binding to the polymeric surface is practically inexistent. The objective of this work is to contribute to clarify the role played by albumin and NaHA on the biolubrication process, through the investigation of their interaction with the UHMWPE surface. The study involves adsorption measurements using a quartz crystal microbalance with dissipation (QCM-D), the characterization of the adsorbed films by atomic force microscopy (AFM) and wettability determinations. Albumin was found to adsorb strongly and extensively to the polymer, while NaHA led to a very low adsorption. In both cases rigid films were obtained, but with different morphology and porosity. The high binding affinity of the protein to the polymer was demonstrated both by the results of the fittings to Langmuir and Freundlich models and by the values of the adhesion forces determined by AFM. In the simultaneous adsorption of albumin and NaHA, protein adsorption is predominant and determines the surface properties.

Low dose metal particles can induce monocyte/macrophage survival

Aseptic loosening results in pain, loss of function, and ultimately prosthetic joint failure and revision surgery. The generation of wear particles from the prosthesis is a major factor in local osteolysis. We investigated the effects of such wear particles on the survival of monocytes and macrophages, populations implicated in wear particle-driven pathology. Particles from titanium aluminum vanadium (TiAlV) and cobalt chromium (CoCr) alloys were generated in-house and were equivalent in size (0.5-3 microm) to those seen in patients. Human CD14(+) monocytes and murine bone marrow-derived macrophages (BMM) were treated with TiAlV and CoCr particles in vitro, and cell survival was assayed. Both particles increased monocyte and macrophage survival in a dose-dependent manner, with an optimal concentration of around 10(7) particles/mL. Conditioned media from particle-treated BMM also increased macrophage survival. Studies with antibody blockade and gene-deficient mice suggest that particle-induced BMM survival is independent of endogenous CSF-1 (M-CSF), GM-CSF, and TNFalpha. These data indicate that wear particles can promote monocyte/macrophage survival in vitro possibly via an endogenous mediator. If this phenomenon occurs in vivo, it could mean that increased numbers of macrophages (and osteoclasts) would be found at a site of joint implant failure, which could contribute to the local inflammatory reaction and osteolysis.

Gene expression in endoprosthesis loosening: chitinase activity for early diagnosis?

The aim of the study was to identify markers for the early diagnosis of endoprosthesis loosening, for the differentiation between wear particle-induced and septic loosening and to gather new insights into the pathogenesis of endoprosthesis loosening. Gene expression profiles were generated from five periprosthetic membranes of wear particle-induced and five of infectious (septic) type using Affymetrix HG U133A oligonucleotide microarrays. The results of selected differentially expressed genes were validated by RT-PCR (n = 30). The enzyme activity and the genotype of chitinase-1 were assessed in serum samples from 313 consecutive patients hospitalized for endoprosthesis loosening (n = 54) or for other reasons, serving as control subjects (n = 259). Eight hundred twenty-four genes were differentially expressed with a fold change greater than 2 (data sets on http://www.ncbi.nlm.nih.gov/geo/ GSE 7103). Among these were chitinase 1, CD52, calpain 3, apolipoprotein, CD18, lysyl oxidase, cathepsin D, E-cadherin, VE-cadherin, nidogen, angiopoietin 1, and thrombospondin 2. Their differential expression levels were validated by RT-PCR. The chitinase activity was significantly higher in the blood from patients with wear particle-induced prosthesis loosening (p = 0.001). However, chitinase activity as a marker for early diagnosis has a specificity of 83% and a sensitivity of 52%, due to a high variability both in the disease and in the control group.

Expression profiling reveals alternative macrophage activation and impaired osteogenesis in periprosthetic osteolysis

Interactions between periprosthetic cells and prosthetic wear debris have been recognized as an important event in the development of osteolysis and aseptic loosening. Although the ability of wear debris to activate pro-inflammatory macrophage signaling has been documented, the full repertoire of macrophage responses to wear particles has not been established. Here, we examined the involvement of alternative macrophage activation and defective osteogenic signaling in osteolysis. Using real-time RT-PCR analysis of periprosthetic soft tissue from osteolysis patients, we detected elevated levels of expression of alternative macrophage activation markers (CHIT1, CCL18), chemokines (IL8, MIP1 alpha) and markers of osteoclast precursor cell differentiation and multinucleation (Cathepsin K, TRAP, DC-STAMP) relative to osteoarthritis controls. The presence of cathepsin K positive multinuclear cells was confirmed by immunohistochemistry. Reduced expression levels of the osteogenic signaling components BMP4 and FGF18 were detected. Expression levels of TNF-alpha, IL-6, and RANKL were unchanged, while the anti-osteoclastogenic cytokine OPG was reduced in osteolysis patients, resulting in elevated RANKL:OPG ratios. In vitro studies confirmed the role of particulate debris in alternative macrophage activation and inhibition of osteogenic signaling. Taken together, these results suggest involvement in osteolysis of alternative macrophage activation, accompanied by elevated levels of various chemokines. Increased recruitment and maturation of osteoclast precursors is also observed, as is reduced osteogenesis. These findings provide new insights into the molecular pathogenesis of osteolysis, and identify new potential candidate markers for disease progression and therapeutic targeting.

Proposal for a histopathological consensus classification of the periprosthetic interface membrane

AIMS

The introduction of clearly defined histopathological criteria for a standardised evaluation of the periprosthetic membrane, which can appear in cases of total joint arthroplasty revision surgery.

METHODS

Based on histomorphological criteria, four types of periprosthetic membrane were defined: wear particle induced type (detection of foreign body particles; macrophages and multinucleated giant cells occupy at least 20% of the area; type I); infectious type (granulation tissue with neutrophilic granulocytes, plasma cells and few, if any, wear particles; type II); combined type (aspects of type I and type II occur simultaneously; type III); and indeterminate type (neither criteria for type I nor type II are fulfilled; type IV). The periprosthetic membranes of 370 patients (217 women, 153 men; mean age 67.6 years, mean period until revision surgery 7.4 years) were analysed according to the defined criteria.

RESULTS

Frequency of histopathological membrane types was: type I 54.3%, type II 19.7%, type III 5.4%, type IV 15.4%, and not assessable 5.1%. The mean period between primary arthroplasty and revision surgery was 10.1 years for type I, 3.2 years for type II, 4.5 years for type III and 5.4 years for type IV. The correlation between histopathological and microbiological diagnosis was high (89.7%), and the inter-observer reproducibility sufficient (85%).

CONCLUSION

The classification proposed enables standardised typing of periprosthetic membranes and may serve as a tool for further research on the pathogenesis of the loosening of total joint replacement. The study highlights the importance of non-infectious, non-particle induced loosening of prosthetic devices in orthopaedic surgery (membrane type IV), which was observed in 15.4% of patients.

[Inflammatory reactions in the wear particle induced and infectious periprosthetic membrane of loosened hip- and knee endoprostheses: pathogenetic relevance of differentially expressed genes cd9, cd11b, cd18, cd52 and pdgfrbeta]

BACKGROUND

A previous cDNA-microarray analysis described constantly differentially expressed genes in wear particle induced and infectious SLIM (synovial-like interface membrane). This study aims to validate the cDNA microarray data in order to approve differences of the gene expression profiles of RNA and proteins.

METHODS

Tissue from 16 wear particle induced and 20 infectious periprosthetic membranes were analyzed by RT-PCR and immunohistology with regard to the expression of inflammatoric associated genes.

RESULTS

RT-PCR showed the genes cd9, cd11b, cd18, cd52 as well as pdgfrbeta in interface membranes. In the wear particle induced membrane the immunohistochemical analysis showed a significantly weaker gene expression of PDGFRbeta, whereas the differential overexpression of CD9, CD11b and CD52 was confirmed. For CD18, there was no difference in expression between wear induced and infectious periprosthetic tissue.

CONCLUSION

Different pathomechanisms, which are reflected by different gene expression profiles, might produce different types of periprosthetic membranes. By RT-PCR and immunohistochemical analysis the micro array data of the genes cd9, cd11b, cd52 and pdgfrbeta could be validated. Identifying the gene products of cd9, cd11b and cd52 in blood or tissue may help to differentiate between wear induced and infectious loosening.

Expression of CD9, CD11b, CD18, CD52 and PDGFR-beta in the interface membrane of loose endoprostheses

The number of endoprosthesis revision operations is increasing. In a previous analysis using cDNA microarrays, differentially expressed genes were detected in the wear-particle-induced membrane and in the infectious periprosthetic membrane. This study aims at validation of these gene expression profiles in order to find genes that are potentially relevant for the pathogenesis, diagnosis, or therapy of endoprosthesis loosening. Tissue samples from 16 wear-particle-induced and 20 infectious periprosthetic membranes were analyzed by reverse transcription polymerase chain reaction (RT-PCR), in situ hybridization, and immunohistochemistry with regard to the expression of CD9, CD11b, CD18, CD52, and platelet-derived growth factor (PDGFR)-beta. RT-PCR demonstrated cd9, cd11b, cd18, cd52, and pdgfr-beta RNA in all samples. Macrophages and multinuclear giant cells in the wear-particle-induced membranes showed intense immunohistochemical staining for CD9, CD11b, and CD52. The staining of PDGFR-beta was stronger in the infectious membranes, whereas there were no differences for CD18. Using RT-PCR and immunohistochemical analysis, the cDNA-microarray data of cd9, cd11b, cd52, and pdgfr-beta could be validated, whereas the differential expression of cd18 was not confirmed. The potential relevance of these genes for prosthesis loosening is discussed.

Strategies using functional genomics in rheumatic diseases

For functional genomics of inflammatory disorders and infection, rheumatic diseases offer unique features to analyse the transition from infection to chronic inflammation, autoimmunity and immunopathology, both systemic and tissue specific. The diseases are frequent and of considerable socio-economic impact. Well-defined cohorts of patients are available. The tissues and cells involved are readily accessible for molecular analysis. Both genetic predisposition and infection are involved in the aetiopathogenesis of rheumatic diseases. The number of susceptibility and severity genes has been estimated to be at least 30, but only few of them have been identified so far. There is an urgent need for developing new therapies adapted to genetic risk and based on a functional genetic and molecular understanding of chronic inflammation. It is evident that gene analysis in inflammatory rheumatic diseases will not only be beneficial for the large number of patients involved, but will also lead to a better understanding of other inflammatory disorders, thereby possibly leading to novel diagnostic and therapeutic strategies in this important group of disorders.

[Proposal for the classification of the periprosthetic membrane from loosened hip and knee endoprostheses]

After 10 years, loosening of total joint endoprostheses occurs in about 3 to 10 percent of all patients, requiring elaborate revision surgery. A periprosthetic membrane is routinely found between bone and loosened prosthesis. Further histomorphological examination allows determination of the etiology of the loosening process. Aim of this study is the introduction of clearly defined histopathological criteria for a standardized evaluation of the periprosthetic membrane. Based on histomorphological criteria and polarized light microscopy, four types of the periprosthetic membrane were defined: periprosthetic membrane of wear particle type (type I), periprosthetic membrane of infectious type (type II), periprosthetic membrane of combined type (type III), periprosthetic membrane of indifferent type (type IV). Periprosthetic membranes of 268 patients were analyzed according to the defined criteria. The correlation between histopathological and microbiological diagnosis was high (89%, p<0,001), the inter-observer reproducibility was sufficient (95%). This classification system enables a standardized diagnostic procedure and therefore is a basis for further studies concerning the etiology of and pathogenesis of prosthesis loosening.

The microenvironment around total hip replacement prostheses

The metal stem of the totally replaced hip carries load and resists fatigue, but it is electrochemically corroded. Metallic atoms act as haptens, induce type 1 T-helper cells/Th1-type immune responses and enhance periprosthetic osteolysis. Stiff metal implants, which do not have the same elasticity as the surrounding bone, cause stress shielding. Cyclic loading and lack of ligamentous support lead to mechanical and ischemia reperfusion injury and particle formation from bone, polymethylmethacrylate, and porous implant surfaces, which accelerate third-body polyethylene wear. Surgical injury and micromotion induce the formation of a fibrous capsule interface. Type-B lining cells produce lubricin and surface-active phospholipids to promote solid-to-solid lubrication but may loosen the implant from bone. The pumping action of the cyclically loaded joint and synovial fluid pressure waves dissect the implant-host interface and transports polyethylene particles and pro-inflammatory mediators to the interface. Hyaluronan induces formation of a synovial lining like layer. Because of its localization close to bone, foreign body inflammation at the interface stimulates osteoclastogenesis and peri-implant bone loss. Metal-on-metal and ceramic-on-ceramic pairs might minimize third body wear, but can lead to high-impact load of the acetabulum. Diamond coating of a metal-on-polyethylene couple might solve both of these problems. The basic biomaterial solutions allow good mechanical performance and relatively long life in-service, but surface modifications (porous coating, hydroxyapatite, diamond, bioglass, and others) may facilitate performance of the implant and improve the biomaterial and body interfaces.

Perspectives and limitations of gene expression profiling in rheumatology: new molecular strategies

The deciphering of the sequence of the human genome has raised the expectation of unravelling the specific role of each gene in physiology and pathology. High-throughput technologies for gene expression profiling provide the first practical basis for applying this information. In rheumatology, with its many diseases of unknown pathogenesis and puzzling inflammatory aspects, these advances appear to promise a significant advance towards the identification of leading mechanisms of pathology. Expression patterns reflect the complexity of the molecular processes and are expected to provide the molecular basis for specific diagnosis, therapeutic stratification, long-term monitoring and prognostic evaluation. Identification of the molecular networks will help in the discovery of appropriate drug targets, and permit focusing on the most effective and least toxic compounds. Current limitations in screening technologies, experimental strategies and bioinformatic interpretation will shortly be overcome by the rapid development in this field. However, gene expression profiling, by its nature, will not provide biochemical information on functional activities of proteins and might only in part reflect underlying genetic dysfunction. Genomic and proteomic technologies will therefore be complementary in their scientific and clinical application.

Could rheumatoid arthritis result from an abnormal T cell response towards lubricin/superficial zone protein?

The lack of some suppressor T cells (including TCD4+CD25+(high) positively selected first in thymic medulla) specific to a restricted set of autoantigens may be the common link between all patterns of rheumatoid arthritis. In other words, instead of a 'peak' of TCD4+ effector T cells common to all patients with rheumatoid arthritis (which has so far never been demonstrated), a 'hole' in TCD4+CD25+(high) responses towards a limited set of autoantigens responsible for the normal maintenance of tolerance within the joints may be shared by many patients with rheumatoid arthritis. The hallmark of this disorder is the involvement of tissues subjected to friction stress bathed in a lubricating fluid (rheumatoid nodules and bursae, tendinous sheaths, pleura, pericardium, sclera, and joints covered by hyaline cartilage). Consequently, autoantigens shared by all forementioned places may be better candidates than autoantigens restricted to the hyaline cartilage (like collagen II). Tenosynovitis, bursitis and rheumatoid nodules can herald rheumatoid arthritis, and rheumatoid pericarditis is very frequent at the histological level. Lubricin and superficial zone protein (SZP), which are closely related products of the megacaryocyte stimulating factor (MSF) gene, are among the best candidate autoantigens for such a positive selection of suppressor T cells. Lubricin is responsible for most of the lubricating properties of synovial fluid, and SZP (expressed by the superficial articular chondrocytes from diarthrodial cartilages and lining cells of synovial villi) also shares lubricating and cytoprotective properties. Moreover, the expression of lubricin is very probable in pericardium and pleura, and can be induced by friction stress. Although this mucinous glycoprotein may already share close similarities at the antigenic levels with mucins previously demonstrated in Hassall's corpuscles of the thymus, evidence for the ectopic expression of lubricin/SZP within normal human thymus may further support this hypothesis. The prenatal positive selection within the thymus of a functional pool of TCD4+CD25+(high) clones specific for most peripheral tissues is critical (at least in mice) for the quality of tolerance for the rest of the organism's lifespan. Therefore, a poor expression of lubricin/SZP early in life within the human thymus may also favour a lack of suppressor T cells specific to tissues bathed with synovial fluid, i.e. the onset of rheumatoid arthritis later on in life. As studies of human thymus long before the onset of rheumatoid arthritis are hampered for obvious reasons, studies of murine thymus could be a first step. In as much as the human counterpart of lubricin is expressed in the thymic medulla of mice, the generation of knocked-out mice for its expression within the thymus could be one of the best models to test the above hypothesis. The stimulation of TCD4+CD25+(high) clones specific for immunodominant epitopes from the joints/synovial fluid (belonging perhaps to lubricin or SZP) could help restore a normal balance between effector T cells and suppressor T cells in rheumatoid arthritis patients.