Article Text

PDF

Expression of epidermal growth factor and transforming growth factor α in interfacial membranes retrieved at revision total hip arthroplasty
  1. Jing-Wen Xua,b,d,
  2. Jian Maa,
  3. Tian-Fang Lia,
  4. Eero Warisa,
  5. Anne Albertyb,
  6. Seppo Santavirtab,
  7. Yrjö T Konttinena,c
  1. aInstitute of Biomedicine, Department of Anatomy, University of Helsinki, Finland, bDepartment of Orthopaedics and Traumatology, Helsinki University Central Hospital, Helsinki, Finland, cDepartment of Oral Medicine, Helsinki University Central Hospital, dDivision of Rheumatology, Long Island Jewish Medical Center, New Hyde Park, NY, USA
  1. Professor Yrjö T Konttinen, Department of Oral Medicine, Surgical Hospital, Kasarmikatu 11–13, FIN-00029, Helsinki University Central Hospital, Helsinki, Finland Email: yrjo.konttinen{at}helsinki.fi

Abstract

BACKGROUND The interfacial membrane between bone and implant has been shown to be a key tissue in the process of aseptic loosening of total hip arthroplasty. The cells within the interfacial membrane produce numerous inflammatory mediators which, through complex mechanisms, cause periprosthetic osteolysis and aseptic loosening. Both epidermal growth factor (EGF) and transforming growth factor α (TGFα) have similar biological functions. They have been found to stimulate bone resorption.

OBJECTIVE To investigate the presence, cellular localisation, and extent of expression of EGF and TGFα in interfacial membrane retrieved from revision total hip arthroplasty and compare it with that in synovial membrane from primary total hip arthroplasty.

METHODS Ten interfacial membranes and 10 synovial membranes were stained with avidin-biotin-peroxidase complex for EGF and TGFα. The staining process was done using the Lab Vision Autostainer. The results were measured by a semiautomatic VIDAS image analysis system.

RESULTS Immunoreactivity for both EGF and TGFα was found in the endothelial cells of blood vessels, macrophages, and fibroblasts, both in interfacial membranes and synovial membranes. However, the number of EGF (980 (370)) and TGFα (1070 (360)) positive cells per mm2 was greater in interfacial membranes than in the synovial membranes (220 (200), 270 (100); p<0.01).

CONCLUSION It is suggested that owing to their increased expression in interfacial membrane, EGF and TGFα may have an important pathogenetic role in stimulating periprosthetic bone resorption in aseptic loosening of total hip arthroplasty.

  • epidermal growth factor
  • transforming growth factor
  • hip arthroplasty

Statistics from Altmetric.com

When hip prostheses are revised owing to aseptic loosening, a layer of interfacial membrane between bone and prosthesis can almost always be found. Because this membrane can release a variety of inflammatory mediators1-6 and is present where bone is resorbed, it must participate in the bone resorption process and contribute to aseptic loosening. It is generally accepted that wear debris generated from a prosthesis is phagocytosed by macrophages/foreign body giant cells and subsequently activates these cells to release a variety of mediators that can stimulate osteoclastic bone resorption to provoke a cascade leading to aseptic loosening.3 7-13 Other factors, such as mechanical factors, may also have an important role in aseptic loosening.

Epidermal growth factor (EGF) was one of the first cytokines to be identified, originally isolated from male mouse submandibular glands.14 Transforming growth factor α (TGFα) is a secreted polypeptide that was first isolated from murine sarcoma virus transformed fibroblast cultures.15 Both EGF and TGFα bind to the same receptor with similar affinity,16 and biological responses caused by these two cytokines are, in most cases, similar.17 18 EGF and TGFα are local regulators of bone metabolism and stimulate bone resorption.19-22

We investigated the presence, cellular localisation, and extent of expression of EGF and TGFα in interfacial membrane from a loose total hip arthroplasty obtained at revision operation and compared the findings with those in synovial membrane from primary total hip arthroplasty carried out owing to osteoarthritis (OA).

Methods

PATIENTS AND SAMPLES

Ten samples of the interfacial membranes between bone and prosthesis/cement were obtained at revision total hip arthroplasty performed for aseptic loosening at the Department of Orthopaedics and Traumatology of Helsinki University Central Hospital, Finland, between May 1996 and June 1997. In all patients, both the clinical and radiographic appearance suggested aseptic prosthetic loosening. All samples were taken from the place at which osteolysis was located. All the components removed were clinically unstable at the time of revision. Culture for both aerobic and anaerobic organisms was carried out on interfacial membranes at revision surgery, and all results were negative. There were no clinical, roentgenographic, or laboratory signs of infection in any case. Ten patients (six female, four male) with a mean age of 72.7 years (range 44–89) were studied. All total hip arthroplasties had originally been performed to treat primary OA. The mean time between primary arthroplasty and revision was 9.4 years (range 1–25). Table 1 presents clinical and demographic data of the patients.

Table 1

Demographic and clinical data of patients with osteoarthritis operated on for aseptic loosening of total hip arthroplasty from whom interfacial membranes were available for epidermal growth factor and transforming growth factor α analysis

For comparison, 10 synovial membranes were obtained from 10 patients with OA (six female, four male) undergoing primary total hip arthroplasty. Their mean age was 56.3 years (range 38–82).

All fresh samples were frozen in isopentane, embedded in OCT compound (Lab-Tek Products, Division of Miles Laboratories, Elkhart, IN, USA) and stored at −20°C until used for immunohistochemistry.

IMMUNOHISTOCHEMISTRY

Staining was done using the Lab Vision Autostainer (Lab Vision Corporation, Fremont, California, USA).The protocol is described below:

Cryostat sections, 6 μm thick, were fixed in 0.3% H2O2 in absolute methanol to block endogenous peroxidase activity for 30 minutes at 22°C. The sections were then incubated with (a) normal rabbit serum for EGF, normal horse serum for TGFα (Vector Laboratory, Burlingame, CA, USA), diluted 1:50 in 20 mM Tris-HCl, 150 mM NaCl, pH 7.5 (Tris buffered saline (TBS)) for 20 minutes at 22°C; (b) polyclonal goat antihuman EGF IgG antibody (dilution 1:100, British Bio-Technology, Oxford, UK) and monoclonal mouse antihuman TGFα IgG antibody (dilution 1:100, Calbiochem, CA, USA) for 30 minutes at 22°C; (c) biotinylated rabbit antigoat IgG for EGF antibody, biotinylated horse antimouse IgG for TGFα antibody (dilution 1:100, Vector Laboratory, Burlingame, CA, USA) for 30 minutes at 22°C; and (d) avidin-biotin-peroxidase complex (dilution 1:100, Vector Laboratory, Burlingame, CA, USA) for 30 minutes at 22°C. Finally, the peroxidase binding sites were revealed with a combination of 3,3′-diaminobenzidine tetrahydrochloride and H2O2 (DAB substrate kit, Vector Laboratories, Burlingame, CA, USA) for five minutes at 22°C. Between the two steps sections were washed three times in TBS, then dehydrated in a gradual ethanol series, cleared in xylene, and mounted manually with synthetic mounting medium (Diatex, Becker Industrifarg AB, Marsta, Sweden). Negative staining controls were obtained by omission of the primary antibody and use of normal rabbit serum (for EGF) and the same isotype IgG serum (for TGFα) at the same concentration. One section from each specimen was incubated with the substrate only, to exclude the detection of endogenous peroxidase activity in endothelial cells.

Tissue structure and cellularity were examined in haematoxylin and eosin and immunohistochemically stained sections. Ultrahigh molecular weight polyethylene particulate debris was examined by polarised light microscopy. Haematoxylin and eosin staining and immunohistochemistry were performed on serial sections of membrane samples.

QUANTIFICATION OF IMMUNOHISTOCHEMICAL STAINING

Positive cells in immunohistochemical staining were measured by image analysis. The number of EGF and TGFα positive cells was counted in five different high power fields (×400) by a low light charge screen coupled to a 12-bit PC digital image camera (SensiCam, Kelheim, Germany), mounted on a Leitz Diaplan (Wetzlar, Germany) light microscope. This was linked to a semiautomatic soft image analysis and processing system (Soft Imaging Analysis System GmbH, Germany) equipped with the VIDAS 3.0 programme (Soft Imaging Analysis System GmbH, Germany). To estimate the adequate sampling area, the number of EGF and TGFα positive cells was calculated in three representative samples in both study groups (interface, control) to obtain an estimate of the population mean for each. Using the formula:

Embedded Image

it was calculated that if at least five fields were counted, the mean value did not differ significantly (p>0.05) from the population mean (t refers tot statistics, –x = the sample mean, μ = the population mean, SEM = standard error of the mean). Thus reliable morphometric results were obtained by calculating five randomly selected fields in each sample. Results are expressed as the number of EGF or TGFα positive cells per mm2 of tissue.

STATISTICAL ANALYSIS

Statistical software, BMDP-PC 7.01, was used to calculate the mean (SEM) dispersion of the data. Normality of the distributions was checked using Wilk's W test. The significance of differences between means was analysed by thet test.

Results

HISTOPATHOLOGICAL EVALUATION

Tissue structure and cellularity varied within one sample and among tissue samples from different patients. The histopathological features of the interfacial membrane1 (figs 1A and B) were similar to those of synovial membrane from patients with OA (fig 1C). The interfacial tissues consisted mainly of macrophages and fibroblasts. T lymphocytes were found dispersed in the macrophage infiltrates in some specimens. Tissue necrosis was also often seen in the interfacial membranes. Particulate debris detected included ultrahigh molecular weight polyethylene, polymethylmethacrylate, and metal wear. Metal particles were identified as irregularly shaped, fine (2–10 μm), black particles. Small metal particles were often phagocytosed by macrophages/foreign body giant cells, but large particulate debris was often seen in the extracellular tissue (figs 1B and D).Cement particles were usually larger and were surrounded by connective tissue and macrophages (fig 1E). Polyethylene particles were identified as highly birefringent fine fragments when viewed under polarised light (fig 1F). This interfacial membrane was often seen to invade periprosthetic bone tissue.

Figure 1

Immunostaining for epidermal growth factor (EGF) (A) and transforming growth factor α (TGFα) (B) in the synovial-like membrane from a patient with aseptic loosening of total hip arthroplasty. Macrophage staining, in particular, was strong and positive cells were almost restricted to the lining of layer of the interface toward the implant side (arrows). The histological features of this membrane were similar to those of synovial membrane from a patient with osteoarthritis (C). (D) Metal-containing macrophages (small arrows) or foreign body giant cells (larger arrows) were positively stained with EGF antibody. (E) Cement particulate debris, which was oval, was surrounded by macrophages which were positively stained with TGFα antibody. (F) Polyethylene wear particles were surrounded by macrophages which were positively stained with TGFα antibody. (Figures A, B, D, E, F: original magnification ×250, avidin-biotin-peroxidase complex staining, without counterstaining; fig C: original magnification ×250, haematoxylin and eosin staining.)

QUALITATIVE ASSESSMENT OF IMMUNOHISTOCHEMICAL STAINING

In all samples the staining intensity varied between samples, but staining was similar throughout a sample. The slides were read by two of the authors (JWX and TFL).

Based on ordinary histological criteria, EGF was found in all interfacial membranes. EGF was found in macrophages/foreign body giant cells (figs 1A and D), fibroblasts (fig 2A), and vascular endothelial cells (fig 2E). In particular, all blood vessels were EGF positive. The result of TGFα staining was similar to that for EGF. TGFα was also found in all interfacial membranes. Macrophages/foreign body giant cells (figs 1B and E), fibroblasts (fig 2B), and vascular endothelial cells (fig 2F) were found containing TGFα. The staining of TGFα in vascular endothelial cells was relatively weak.

Figure 2

(A) Epidermal growth factor (EGF) and (B) transforming growth factor α (TGFα) positive fibroblasts in the synovial-like membrane from a patient with aseptic loosening of total hip arthroplasty. These microphotographs were of a relatively fibrous area, where the long and slender fibroblast-like cells predominate. In contrast with the EGF and TGFα staining in the synovial-like membrane from a patient with aseptic loosening of total hip arthroplasty, there were only a few EGF (C) and TGFα (D) positive cells in the synovial membrane from patients with osteoarthritis who had undergone primary total hip arthroplasty, and staining was relatively weak. In addition to EGF and TGFα positive macrophages and fibroblasts seen in the synovial-like membrane, EGF (E) and TGFα (F) positive vascular endothelial cells (marked with arrows) were also found in this membrane. No EGF (G) and TGFα (H) positive cells were found in any of the staining controls obtained by using an avidin-biotin-peroxidase complex (ABC) staining sequence, in which the primary antibody was omitted and replaced by same isotype IgG serum. (Original magnification ×250, ABC staining, without counterstaining).

Synovial membranes from patients with OA contained few EGF and TGFα positive cells and staining was relatively weak (figs 2C and D), but with a similar cellular origin and distribution profile as for EGF and TGFα positive cells in the interfacial membranes. Staining control confirmed the specificity of all immunoreactions seen (figs 2G and H).

IMAGE ANALYSIS

The number of EGF and TGFα positive cells was higher in the interfacial membranes than in the synovial membranes (p<0.01; table2). Although the numbers were too small for an adequate statistical comparison, there did not seem to be any particular differences between the type of fixation (with or without cement) or of alloy (TiVAlv CoCrMo) used.

Table 2

Epidermal growth factor (EGF) and transforming growth factor α (TGFα) expression as cells/mm2 tissue analysed in interfacial membrane in loose total hip arthroplasty: comparison of interfacial membrane with synovial membrane from patients with osteoarthritis

Discussion

The results of our study confirm that EGF and TGFα are produced at the interfacial membrane retrieved from a loose prosthesis. This study takes advantage of the staining done using the Lab Vision Autostainer, which makes the process easier and the results more reliable than by manual staining. Histopathological examination has shown that heterogeneity in tissue structure and cellularity within one sample and among tissue samples of different patients do exist in the interfacial membrane retrieved from a loose prosthesis at revision total hip arthroplasty. However, we used representative sampling for morphometric measurement of EGF and TGFα positive cells in this membrane and used statistics to show that the number of EGF and TGFα positive cells in this membrane was higher than in a synovial membrane from a patient with OA. This heterogeneity is probably due to the variable biological, mechanical, and material microenvironments along the interface.23

The evidence has shown that EGF and TGFα stimulate bone resorption—for example, in mouse calvaria in organ culture, through a prostaglandin mediated mechanism.21 24 In contrast, in fetal rat long bones, EGF and TGFα stimulate resorption by mechanisms that are independent of prostaglandin synthesis.19 22 EGF and TGFα also cause an increase of plasma calcium in mice25 and stimulate the formation of osteoclast-like cells in long term marrow cultures.26 The biological activities of EGF and TGFα are similar but not identical. For example, TGFα stimulates bone resorption to a much greater extent than EGF in organ cultures of fetal rat long bones.20 27

Much knowledge has accumulated on the biological mechanism whereby the wear debris leads to aseptic loosening. Macrophages/foreign body giant cells phagocytose wear debris generated from prostheses, and this process in turn releases mediators that stimulate bone resorption. These mediators include prostaglandins, cytokines, metalloproteinases, lysosomal enzymes, nitric oxide, and others.1 3 5 10 28 29 In this study we have shown that both EGF and TGFα were induced in cells, including macrophages/foreign body giant cells, fibroblasts, and endothelial cells in interfacial membranes. The greater number of EGF and TGFα positive cells in interfacial membranes than in synovial membranes suggests that local conditions induce EGF and TGFα, capable of stimulating bone resorption and contributing to prosthetic loosening. EGF and TGFα were found in different types of fixation (cemented and uncemented) and alloys (CoCrMo, TiAlV), suggesting that the same process of induced bone resorption contributes to aseptic loosening in all cases regardless of the differences in the type of fixation and alloy.

A major limitation of the study of interfacial membrane obtained at revision total hip arthroplasty for aseptic loosening is the absence of appropriate control samples. For ethical reasons it was not possible to obtain corresponding control samples from well fixed total hip arthroplasties. Goodman et al have studied cellular profile and cytokine production at interface tissue from well fixed prostheses.6 They concluded that the tissues from well fixed implants cannot be considered as controls, because these implants were painful, dislocating, or showed excessive polyethylene wear. Additionally, the time from implantation to revision for well fixed implants in both the cemented and cementless groups was shorter than for the respective groups with loosening. Thus the tissues surrounding well fixed prostheses were from prostheses in the early stages of failure, before more advanced mechanical and biological processes could produce loosening or osteolysis. Because it is impossible to find control samples which match the histopathological features of the interfacial membrane, we used synovial membrane obtained from patients with OA undergoing primary total hip arthroplasty to serve as control tissue in this study. As the histological features of interfacial membrane, or so called synovial-like membrane,1 are similar to those of synovial membrane from a patient with OA, this seems appropriate.

In conclusion, we suggest that owing to their increased expression in interfacial membrane, EGF and TGFα may have an important pathogenetic role in stimulating periprosthetic bone resorption in aseptic loosening of total hip arthroplasty. Additional research is necessary to improve understanding of EGF, TGFα, and the complex cytokine network in interfacial membrane, which contribute to aseptic loosening of a total hip arthroplasty.

Acknowledgments

We thank Ms Eija Kaila and Mr Reijo Karppinen for skilful technical assistance, and Dr H Pätiälä at the Department of Orthopaedics and Traumatology, Helsinki University Central Hospital, for providing the samples used in this study.

References

View Abstract

Footnotes

  • Supported in part by grants from the Academy of Finland, Helsinki University Central Hospital, the Foundation for Orthopaedic and Traumatological Research in Finland.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.