Background
Bone morphogenetic proteins play an important role in the proliferation, differentiation, and induction of mesenchymal stem cells as well as in the embryogenesis [
1]. BMP-2, currently one of the best-studied bone morphogenetic proteins, is applied increasingly in orthopedic surgery [
2]. The working group of Jennissen developed a method for the surface modification which allows a subsequent biocoating by proteins, e.g., BMP-2 [
3]. The basis of this principle underlies the production of an ultra-hydrophilic surface on titanium implants with chromosulfuric acid, which leads to a so-called inverse lotus effect, which causes a reduction of the contact angle to < 10° [
4]. The procedure has already been published several times in detail and is patented [
3,
5‐
8]. By the addition of anchor molecules, BMP-2 can be bound and retains its biological activity [
6,
9]. Thus, the prerequisite for adequate protein binding is the prior treatment with chromosulfuric acid. Previous studies showed that the treatment with chromosulfuric acid alone has an osseointegrative potential [
10]. BMPs as differentiation factors are suspected to promote malignant degeneration when used in humans [
11].
Gap models have been used since 1980 and earlier in orthopedic research, allowing the investigation of bone repair under special conditions with different questions [
10,
12‐
15].
The animal of our choice, the Göttingen minipig, has been proven to be a reliable animal for examining implants, especially because of the similar characteristics of the mesenchymal stem cells [
16], as well as because of a comparable lamellar bone structure [
17].
Histomorphometry, as well as the polychrome labeling, finds application in the evaluation of bone remodeling processes [
10]. For fluorescence microscopy, in the sense of intravital staining, combinations of different fluorochromes lead to a better resolution of ossification fronts by synergistic effects [
18], and this could be shown, e.g., for tetracycline and xylenol orange [
19]. For some substances, no spectral analysis is needed, which would require an immense effort for the evaluation [
20,
21].
The gold standard in orthopedic research is the histomorphometry, particularly for the assessment of osseointegration, which allows a quantitative assessment on the basis of an established standardized nomenclature since the 1980s [
22‐
24].
The aim of the study was to investigate the osseointegrative force of different titanium surfaces at different postoperative times in respect to a covalently bound layer made of BMP-2 additionally. The application in an animal bone gap-healing model should reveal the potential for osseoinduction and osseoconduction.
Methods
The animal study was approved by the ethical committee of the Regierungspräsidium Karlsruhe, Abteilung 3 - Landwirtschaft, Ländlicher Raum, Veterinär - und Lebensmittelwesen, Karlsruhe, Germany, under the file no. 35-9185.81/G-27/05.
Preparation of implants
Cylindrical dumbbell-shaped implants were used with a total length of 30 mm, a gap length of 15 mm, and a resulting gap after implantation with a height of approximately 1.28 mm as reported by Seidling et al. [
10]. The surface structure was created on one hand, by the application of titanium plasma spray (TPS group), and on the other hand, by modification with a chromosulfuric acid alone (CSA group) and subsequent covalently bound rhBMP-2 (BMP-2 group). This method was developed and patented by the working group of Jennissen (University of Duisburg-Essen) [
3,
7]. The modification is carried out by a treatment with chromosulfuric acid for 60 min at high temperature and subsequent cleaning with EDTA and water [
3,
25]. This process leads to an ultra-hydrophilic surface by thickening the oxide layer with improved binding capacity [
4]. After this treatment, BMP-2 can be covalently bound stable to the surface and retains its biological activity [
25,
26].
The titan implants with the TPS layer in the middle part were produced by the company DOT (Rostock, Germany) and described in more details by Seidling et al. [
10]. Modified in the manner described above, they were provided sterile and ready for implantation by the working group of Jennissen (University of Duisburg-Essen) and the company Morphoplant GmbH (Duisburg, Germany).
Animals and surgery
Twenty-seven Göttinger minipigs (Ellegaard, Dalmose, Denmark) were divided into three groups with nine animals each. In the TPS group, the animals had an average weight of 64.72 kg (SD ± 6.26 kg). In the CSA group, the average weight was 59.5 kg (SD ± 13.85 kg), and in the BMP-2 group, the average weight was 56 kg (SD ± 14.69 kg). The animals were on average 37.7 months old in the mean (SD ± 4.66 months). Thus, the animals were adults, as Christensen had reviewed [
27].
We observed a time period of 4, 8, and 12 weeks postoperatively for each group; so, the animals were accordingly divided into three further subgroups.
Each animal received four implants as described previously by Seidling et al. [
10]. The implants were implanted on both sides, in the cancellous bone of the distal and proximal part of the femur strictly press-fit and under general anesthesia according to Schwarz et al. [
28,
29]. Pre- and postoperative monitoring and housing of the animals were performed in the “Interfakultäre Biomedizinische Forschungseinrichtung (IBF)” of the University of Heidelberg (Germany) up to 14 days postoperatively. They were held for the followed healing period on a farm near Heidelberg until to a few days before the scheduled time point of killing. The animals were fed with “Schweine-Einmastkorn” (17% crude protein) of the company Muskator (Mannheim, Germany). Water was available at any time ad libitum. The animals survived the operation procedure, and there were no perioperative obvious complications observed.
Intravital staining
For the semiquantitative evaluation of the osseointegration, an intra-vital staining was performed with the substances calcein green (20 mg/kg; Fa. Pfizer, Karlsruhe/Germany), xylenol orange (90 mg/kg; Fa. Waldeck GmbH&CoKG, Münster/Germany), and tetracycline (26 mg/kg; Fa. Waldeck GmbH&CoKG Münster/Germany) [
30], according to the scheduled time of application (Table
1) [
10]. The substances were administered by injection into the neck muscles of the animals in the postoperative period.
Table 1
Day of injection and postoperative period for the fluorochromes
4 weeks | Day 9 | Day 16 | Day 24 |
8 weeks | Day 37 | Day 44 | Day 52 |
12 weeks | Day 65 | Day 72 | Day 80 |
Implant separation and preparation
The animals were killed in deep sedation after premedication at the scheduled postoperative time of 4, 8, or 12 weeks, by injection of 20–40 ml KCL 7.45% under strict ECG control. The death was ascertained by asystole, auscultation, and a missing corneal reflex. The procedure is described by Seidling et al. [
10]. Shortly summarized, the preparation of the undecalcified samples was performed according to Donath and Breuner which is a method for embedding and processing of jaw bones with teeth or bone with ceramic and metal implants [
31]. A modified Masson-Goldner staining was used to differentiate between mature bone and immature osteoid [
32]. The sample selection and preparation procedure has already been described by Seidling et al. [
10] and Lehmann et al. [
33] previously. In particular, the very middle specimen of the serial cuts was used.
Evaluation
The histomorphometric evaluation was carried out interactively using a Leica DRC 300 FX-microscope (Fa. Leica Camera AG, Solms, Germany). For this study, a specific evaluation program (“Macro”) has been created with the Leica company based on the Leica Qwin V3 software (Leica Camera AG, Solms, Germany) [
10]. This allows a computer-based assessment on the basis of mosaic images of the complete gap under higher magnification. The samples were sighted on the microscope while using the semiautomatic evaluation program and marking the different types of bony tissues namely osteoid and bone respectively. The created Macro has been validated previously [
34]. The variables of bone and osteoid ongrowth as indicators for the osseoconduction, as well as bone and osteoid volume as indicators for the osseoinduction, were able to be determined due to the use of the staining according to Masson-Goldner [
10,
32].
All slides were checked and then blinded prior to the histomorphometrical evaluation by three independent investigators (ML, RS, EM). A ROI of 5 mm width in the middle of the gap was used according to the prior investigation. All non-evaluable preparations were sorted out. As exclusion criteria, e.g., a partial gap coverage and surface detachment of the TPS surface layer were identified [
10]. The evaluators repeated measurements on a specific specimen for self-supervision [
10].
For the microscopic evaluation with a Leica DRC 300 FX (Leica Camera AG, Solms/Germany; used filters: LeicaI3 = DAPI/Leica N2.1 = Rhodamin/Leica A = FITC) of the polyfluorochrome labelling, semi-quantitative scores were used, which are represented by means of a subjective assessment of the activity in general as an overview (general score), the amount of the observed fluorescence quantitatively for the de novo bone formation (amount score), and the intensity of the observed fluorescence for the bone turnover (intensity score) [
10]. Table
2 shows the scores used and their assessment.
Table 2
Scores for the polyfluorochrome labeling
0 | None | No bone | None |
1 | Uncertain | Minor | Very low intensity |
2 | Minor | Moderate | Minor |
3 | Much | Much | Moderate |
4 | X | X | Much |
5 | X | X | Very much |
Statistics
The data was prepared in cooperation with the Department for Medical statistics, Biomathematics, and Information Processing, Medical Faculty Mannheim of the University of Heidelberg.
The four relevant outcomes of this trial were osteoid ongrowth, bone ongrowth, bone volume, and osteoid volume. Due to the non-normal distribution of the preparations (Kolmogorov test) and missing of homoscedasticity in a Levene’s test, the data experienced a logarythmization after addition of 1 (bone volume and osteoid volume) or 10 (bone ongrowth and osteoid ongrowth). After this transformation, data revealed to be approximately normally distributed with similar variances in the subgroups.
With these data, analysis of variance (ANOVA) with repeated measurement design was performed using SAS release 9.4 (SAS Institute, Cary / North Carolina, USA). The SAS procedure “PROC MIXED” has been applied for two- or three-way ANOVAS where the parameters group (“type of implant”), localization (hip or knee), and postoperative time (4, 8, and 12 weeks) have been considered as fixed factors and animal ID as a random factor. We used these techniques in order to analyze three factors simultaneously and to test for interactions. If necessary, pairwise post hoc comparisons have been done using Tukey’s test.
For the statistical evaluation of the intravital staining, only a descriptive statistics was carried out using Excel for Mac Version 2011 (Microsoft Company, Washington/USA). The level of significance was
α = 0.05. A sample size calculation was performed as described by Seidling et al., previously [
10].
Discussion
The aim of the study was to evaluate the osseointegrative force of rhBMP-2 in a large animal.
We have seen the main effect in the formation of osteoid and bone at the surface of the implant less inside the gap. And the effect of the BMP-2 protein seems to be pronounced more for the growth of osteoid as a precursor of bone than at mature bone. Seen from a temporal perspective, one can assume that the appearance of osteoid is close to the recruitment of preosteoblasts. In the present study, we found an indication of new osteoid formation at the beginning after 4 weeks which we interpret as osteoinduction. This can be correlated to the explanations of Albrektsson and Johannson [
35]. They point out that the osteoinduction may start immediately after injury (in the present study the implantation) with a high activity over the first weeks as the recruitment of preosteoblasts may not be evident until some weeks later. This may explain also our results that we did not see any further effects in the gap in terms of osteoinduction later on.
Several chemotactic properties have been demonstrated for BMP-2 which is discussed as responsible for attracting corresponding effector cells [
36‐
39]. Presumably, this property contributes to the influence of osteoconduction after long-term fixation of the BMP-2 to surfaces [
5]. Since the significant differences were found mainly in the immature bone-like osteoid volume and osteoid ongrowth in the present study, it can be assumed that the effects of BMP-2 have not yet been completely assessed as the observation time was limited to 12 weeks. Further investigations should be performed to show effects at a later time.
The used BMP-2 coating was covalently bound to the titanium TPS surface by prior treatment with chromosulfuric acid and binding via anchoring molecules and maintains a contact angle of 0–8° for 50 days [
25]. The biological activity is reported as still evident [
25] shown in cell culture by detection of alkaline phosphatase as an indicator of bone induction by BMP-2. Thus, the results of the present study can close the gap between in vitro and in vivo assessment together with preliminary studies with the same bioactive layer [
40]. Sumner and coworkers made a study to analyze the influence of different dosage of BMP-2 on top of a HA/TCP. They found a 2–3.5-fold increase of bone formation in the gap which was 3 mm in height [
13]. They compared a previously reported group with HA/TCP to recent groups with locally delivered rhBMP-2 in the left humeri and implants with solely buffer in the right humeri. Effects on the contralateral site were observed with an enhanced bone ingrowth. So, Sumner and colleagues hypothesized that not the BMP-2 but rather factors released by the regeneration process are responsible for this reported contralateral effect. Thus, a systemic effect could be discussed using BMP-2 as implant coating. In the present study, the control groups were separate pigs with only BMP-2 or CSA modification or none (TPS). Thus, an effect on control implants which were set in the same animal was excluded by the study design in the present trial.
As previously described, the implanted specimens were subjected to a treatment with chromosulfuric acid at high temperatures prior to biocoating with BMP-2 via anchor molecules, a process known as “surface enhancement” [
41]. It has already been shown that this has a positive effect on osseoinduction as well as on osseoconduction [
10,
40]. This effect, however, appears to be higher after addition of BMP-2. But, the present results show markable differences between BMP-2 covalent and CSA, which are all not significant post hoc. Considering the results of the paired comparison of CSA and TPS in particular as described by Seidling and coworkers [
10], significant differences were seen for the variables osteoid ongrowth and (12 weeks) osteoid volume (4 weeks). Thus, the effect does not appear to be due solely to the surface modification.
The used gap model allows the differentiation between the osteoinductive and the osteoconductive properties of a layer on an implant in the bone [
10]. Gap models are known as models for healing and turnover research in bone [
14,
42,
43]. It is also known as a model to examine some sealing effects between the implant and bone [
12] when a gap results between the surface of an implant and the bone stock which should be filled by bone. A sealing effect can also be an important feature of a BMP-2 coating in primary treatment with artificial joints or even with revision surgery when a loss of the bone stock came up and gaps between implant and bone are likely [
10,
44,
45]. At the other hand, a gap could be seen as a carrier system like a potential reservoir between implant and bone for BMP enhancing osteogenesis, e.g., for the treatment of nonunions after bone fractures by local delivery of the protein [
46,
47]. We additionally observed a significant difference for the implant site (hip vs. knee) and the variable bone volume in the three-way ANOVA. To what extent, there is an interaction between the site of implantation (knee or hip) and the surface could only be observed. Since it was not content of the present study, further observations are needed to clarify this observation.
Proteins need appropriate carrier systems when bound at the surface of an implant to keep their biological activity [
3,
7,
41,
45,
48]. Several bonding techniques are described [
49]. The specific technique for bonding the BMP-2 at the TPS surface of the used implants is described by Jennissen and his group [
3,
41] and was performed in the present study.
Beneath the choice of the appropriate time point the gap model may have disadvantages for the analysis of the osteoinduction properties of the BMP-2 coating of implants. The BMPs were detected when the bone was implanted into the tissue like the musculus rectus abdominis or quadriceps or in bone [
50]. Looking at the situation in the gap model where alloplastic material is used at the implant site blood supply has to be missed in consequence.
A perfect animal for all types of bone research may not exist as there is a high variation in, e.g., bone turnover [
51]. Especially in small animals, the bone turnover seems to be much higher than in humans or large animals [
51]. As previously described, the lamellar bone structure of Göttingen minipigs is nearly similar to humans, as well as the remodeling sequences [
17,
52]. Furthermore, the bone marrow and marrow stem cells of Göttingen minipigs have similar characteristics to humans, especially in proliferation and morphology [
16]. Not least, there is a high experience with this type of animals in orthopedic research, and they are cost-effective and easy to handle [
53]. Nevertheless, none of the common study animals like dogs, pigs, or rats is fully comparable with humans [
52].
The analysis of the intravital staining revealed bone turnover over the whole observation time. None of the used scores was able to distinguish between the BMP-2 and the TPS group clearly as the values laid close together combined with wide standard deviations for both groups. The wide variances may also be caused by the differing results of the examinations of the three observers (data not shown) reflecting the individual character of the evaluation. However, the activity of bone turnover was not constant in the scores over the observation time. The “general score” and the “amount score” seemed to diverge at the end of the observation time, whereas the “intensity score” shows an undulating behavior.
However, the semi-quantitative nature of the assessment must be considered. In particular, the used scores are highly dependent on the investigator. Furthermore, the light-emitting behavior of the substance in the mineralization front is quite different and depends above all on the processing and handling of the preparations [
20]. Pautke et al. [
20] observed no bleaching effect for the substances xylenol orange, calcein green, and rolitetracycline during storage in the dark and room temperature over a period of 6 months. Only the substance calcein blue showed a disappearance of the fluorochrome effect after 6 months [
20]. Due to the experimental setup, the temporal distance between implantation and evaluation was more than 6 months; so, a bleaching cannot be safely excluded. Comparing both techniques, the histomorphometry and the intravital staining, we guess that the histomorphometry is more effective than the intravital staining in the assessment of osseointegration of implants, not least because of the low intra- and interobserver reliability.
In general, some drawbacks of BMP-2 have to be considered when used. According to the basic characteristic of the protein, ectopic bone formation may be possible after implantation in some case reports we have found in literature [
54‐
57]. However, the working group of the present study was able to exclude a side effect in this case in terms of new bone formation at another site of the treated animals [
33].
A review, published by Carragee et al. [
11], brought BMP-2 into disrepute in the application to humans. Carragee et al. described a clear study design bias with an actually 10–50-fold increased complication rate compared to the previously published data [
11]. There was a 40% increased risk of adverse events compared to the conventional autologous bone graft, as well as an increased risk of malignant disease in the absence of benefit of the BMP-2 [
11,
58,
59]. Since rhBMP-2 is a derivative of a differentiation factor in the mesenchymal development, malignant degeneration is not unlikely [
54,
60]. A participation in tumor angiogenesis, as well as an inhibition of tumor cell apoptosis, could also be demonstrated [
39].
Thus, the harmlessness of BMP-2 should be shown in further studies as this question was not the rationale of the present study dealing with the osseointegration.
Acknowledgements
We want to thank the Department of Statistics and Empirical Research of the Medical Faculty of Mannheim of the University of Heidelberg for the given support in statistics, especially C. Weiss, J. Brade, and S. Büttner. Special thanks to E. Forsch, who supported the surgery and mainly the general anesthesia of the animals. We acknowledge financial support by Deutsche Forschungsgemeinschaft within the funding program Open Access Publishing, by the Baden-Württemberg Ministry of Science, Research and the Arts and by Ruprecht-Karls-Universität Heidelberg. We acknowledge the Fa. DOT GmbH (Rostock, Germany), for the implant preparation. The implants were provided in a collaborative initiative between the Institut für Physiolgische Chemie, AG Biochemische Endokrinologie, Universität Duisburg-Essen, Hufelandstr. 55, D-45122 Essen, and Morphoplant GmbH, Universitätsstr. 136, D-44799 Bochum, Germany by the joint working group consisting of S. Lüers, C. Seitz, S. Madenci, M. Laub, and H.P. Jennissen.