Enhancing tissue repair in annulus fibrosus defects of the intervertebral disc: analysis of a bio-integrative annulus implant in an in-vivo ovine model
Aldemar Andres Hegewald,
Department of Neurosurgery, University Medical Centre Mannheim, Heidelberg University, Mannheim, Germany
Department of Neurosurgery, Innsbruck Medical University, Innsbruck, Austria
Correspondence to: Aldemar Andres Hegewald, Department of Neurosurgery, University Medical Centre Mannheim, Heidelberg University, Theodor-Kutzer-Ufer 1–3, 68167 Mannheim, Germany. E-mail: email@example.com
Annulus fibrosus (AF) repair techniques for the intervertebral disc (IVD) are of interest to the spine surgeon because (1) they address the unsolved problem of reherniation through the untreated annulus defect after IVD herniation (Barth et al., 2008) and (2) they facilitate the development of nucleus pulposus replacement techniques for IVD diseases by providing adequate nucleus containment (Heuer et al., 2008). Both are highly relevant health economic issues Sherman et al., 2010).
First attempts of AF closure consisted of simple suturing and gluing techniques (Ahlgren et al., 2000; Heuer et al., 2008) and of the insertion of solid mechanical barriers (Bron et al., 2010; Anulex Technologies, 2012; Intrinsic Therapeutics, 2012). These techniques were only partly successful and were often complicated by implant dislocation. Moreover, there is the concern of long-term effects of solid implants on adjacent tissue structures. New generations of these implants are now in the process of controlled clinical trials.
Advances in material sciences and biotechnological techniques are now encouraging innovative biological repair strategies. Biological scaffold technology plays a major role in providing primary stability and three-dimensional spaces for tissue formation by introducing a whole array of absorbable and non-absorbable biomaterials (Ratner and Bryant, 2004; Chan and Leong, 2008). In addition, bioactive factors, enforcing cell proliferation and matrix production (Gruber et al., 1997, Imai et al., 2007, Gilbertson et al., 2008, Kim et al., 2009, Vadalà et al., 2012) as well as cell chemotaxis (Hegewald et al., 2011), might enhance the AF repair process. Cell-based approaches with multipotent progenitor cells or chondrogenic cell lines are discussed as possible therapeutic options (Kuh et al., 2009; Koepsell et al., 2011) and might open the door for gene therapy approaches (Zhang et al., 2007).
This pilot study reports on the suitability of a cell-free, biointegrative annulus implant to enhance tissue repair in a large box-shaped AF defect by means of an in-vivo ovine model. The potential advantages of this implant are its high porosity and its high bioabsorbability.
The natural repair process of AF defects advances from the outside to the inside, but rarely exceeds the outer third of the defect, leaving the inner regions of the AF unrepaired (Key and Ford, 1948; Smith and Walmsley, 1951; Hampton et al., 1989; Ethier et al., 1994; Melrose et al., 2008). This finding and the authors’ experience in biomechanical studies that AF implants demonstrate a certain amount of bulging under load simulation (Hegewald et al., 2009) lead to the hypothesis of this study that an optimal implantation site for an AF implant is at the inner wall of the AF, allowing some bulging into the AF defect without compromising the spinal canal. At the same time, the AF implant can exert its function as a biological scaffold by enabling repair processes in the inner regions of the AF, otherwise left unrepaired. In this way, the establishment of a more reliable AF repair process with enhanced primary stability and repair tissue strength is expected.
2.1 Annulus implant
A customized triphase AF implant 10 mm wide, 5 mm high and 2 mm thick, consisting of two outer phases of absorbable polyglycolic acid (PGA) nonwoven and a centric phase of a non-absorbable polyvinylidene fluoride (PVDF) mesh was fabricated and assembled. The PGA (Purac Biomaterials, Gorinchem, the Netherlands) was melt spun to multifilament fibres and processed to a non-woven, characterized by a porosity of 80% and an interconnecting pore network, feasible for cell in-growth and matrix build-up. The microstructure of the non-woven consists of short filaments arranged in all directions and bonded together mechanically using needle punching. The stabilizing textile mesh is made of non-absorbable PVDF fibers with a diameter of 112 µm (G. Krahmer GmbH, Buchholz, Germany). This textile mesh enforces primary stability, especially at the fixation points of the implant. The mesh was produced using a double raschel warp knitting machine (DDR 16 EAC/EEC; Karl Mayer Textilmaschinenfabrik GmbH, Obertshausen, Germany). In a final step, the warp knitted PVDF mesh (centric phase) and the doubled PGA non-woven (two outer phases) were assembled by customized needle punching.
2.2 Surgical procedure
Twenty healthy adult merino sheep, aged 4–6 years, mean body weight 82 kg, were operated on within this project with the approval of the local competent authority (Regierungspräsidium Karlsruhe, file number 35-9185.81/G-15/09).
The first two sheep were operated via a dorsal inter-laminar approach. In contrast to the human situs, it was found that the outgoing spinal nerve roots are not protected by a dural sleeve and are therefore very sensitive to mobilization. Both sheep were killed post-operatively because of high-grade hind-limb weakness, and a left lateral retroperitoneal approach was then adopted for the study. Two sheep developed a severe wound infection and had to be killed prematurely. Consequently, 16 sheep were included in the study.
After suitable intramuscular preliminary medication with xylazine (0.1–0.15 mg/kg body weight) and ketamine (10 mg/kg body weight), the sheep was intubated and a large stomach tube applied. The maintenance of narcosis was provided by inhalation anaesthesia with isoflurane (1.5–3 vol %) and supported by intravenous fentanyl (1–3 µg/kg body weight) if required. Perioperative antibiotic prophylaxis was provided with intravenous cefazolin (2 g). Intraoperative monitoring was established with arterial blood pressure measurement and measurement of oxygen saturation.
The sheep was carefully bedded on the right side. Using standard sterile techniques, the lumbar segments L3/4 and L4/5 were accessed via a left retroperitoneal, transpsoas (between psoas major and minor) approach. After fluoroscopic examination for level localization, two box-shaped annulus defects of 3.5 × 3.5 mm were applied with a scalpel to the two intervertebral discs. Protruding nucleus pulposus tissue in the defect was carefully coagulated and the adjacent AF was undermined with a hook instrument. The nucleus pulposus was otherwise left intact and no nucleotomy was performed. A blocked randomization for the annulus implant insertion was then conducted.
Directly before application, the annulus implant was immersed in autologous blood serum for potential bioactivation (Hegewald et al., 2011). The implantation was achieved with an inside-out fixation technique, using non-absorbable sutures (4-0 Prolene®, ETHICON GmbH, Norderstedt, Germany) that were prefixed at the four corners of the annulus implant (Figure 1). With this technique, the annulus implant could be pulled inside the disc and tightly fixed to the inner AF.
2.3 Explantation procedure and provocative pressure testing
Explantation was performed 2, 6 and 12 weeks after surgery with five, five and six sheep, respectively, in each explantation group. For each explanted lumbar spine, block randomization between the IVDs L2/3 and L5/6 was conducted to allocate for: (a) fresh annulus defect, (b) fresh annulus defect with implant and (c) intact control. In the analysis, (a) and (b) are referred to as time-point 0 weeks.
Following this procedure, provocative pressure testing was performed. Under fluoroscopic observation, a customized high-pressure needle was placed centrally into the nucleus pulposus. With an angiography pump, continuous injection (0.5ml/min) of contrast medium was applied to the nucleus pulposus compartment until contrast media leakage was observed fluoroscopically. A pressure transducer was used to continuously register the pressure values.
Tissue processing and staining protocols were mostly based on procedures established by Melrose et al. (2004, 2007).
Directly after provocative pressure testing, individual IVDs were excised with a band saw and fixed in 10% neutral buffered formalin (1 week). Decalcification was performed in several changes (three times a week) of 10% formic acid in 5% neutral buffered formalin with constant agitation (4 weeks). After decalcification, the specimens were rinsed in running tap water (30 min) and equilibrated in several changes of 70% ethanol. To prepare tissue sections, specimens were double-embedded in paraffin–celloidin. Specimens were first dehydrated in graded alcohols and afterwards equilibrated in methyl-benzoate for 24 h. To infiltrate the tissue with celloidin, specimens were incubated in a mixture of celloidin and methylbenzoate. First, they were equilibrated in methylbenzoate containing 1% celloidin, then subsequently in methylbenzoate containing 5% celloidin. After purification with chloroform for 1 hour, the specimens were vacuum-infiltrated and embedded in Paraplast (Paraplast Plus®, Sigma-Aldrich®, Germany). Before cutting, the surface was softened by using softening solution containing ammonia, glycerine, 100% ethanol and distilled water. Afterwards, 10 µm sections were cut with a microtome and fixed on star frost plus (SuperFrost® Plus Objektträger, R. Langenbrinck Labor- und Medizintechnik, Emmendingen, Germany) glass slides. Sections were dried at 75°C on a heating plate (10 min) and then overnight in an oven at 55°C. Deparaffinization was performed in xylene. Thereafter, specimens were rehydrated with graded ethanol washes (100–70% v/v).
The following stains were performed: (1) modified Masson's trichrome, (2) toluidine blue–fast green, both according to modified protocols established by Melrose et al. (2007), (3) FAST (Fast Green, Alcian Blue, Safranin-O, tartrazine) according to modified protocols of Leung et al. (2009). For the Masson's trichrome staining, the sections were first incubated in Bouin's mordant fixative at 60°C (1 h) and afterwards purified in running tap water (10 min). Sections were stained with Weigert-s iron-haematoxylin reagent (30 min); adjacent colour differentiation was performed in 1% HCl in 70% ethanol. Afterwards, sections were purified in running tap water and Scott's tap water substitute (3 min) and stained in Ponceau de Xylidine (8 min). After being washed with distilled water, sections were differentiated in 5% phosphotungstic acid solution (10 min) and counterstained with 1% Fast Green FCF (2 min). Finally, sections were dehydrated in several changes of absolute ethanol and cleared in xylene. With Masson's trichrome, cytoplasm and muscle fibres stained red and collagen fibres stained blue. Somewhat unexpected was the consistent reddish staining of the AF; this was also found in the repair tissue of all 12-weeks specimens. For Toluidine Blue–Fast Green staining, sections were stained in 0.04% Toluidine Blue O (10 min) and then counterstained with 0,1% Fast Green FCF (2 min), purified with several changes of isopropanol and cleared in xylene. With Toluidine Blue–Fast Green staining, proteoglycans are stained blue. For multichromatic FAST (Fast Green, Alcian Blue, Safranin-O, tartrazine) staining, initial staining was performed with Alcian Blue 1% (2.5 min) and Safranin-O 0.1% (3 min). After differentiation with ethanol 50% (1 min) counterstaining was performed with tartrazine 0.25% (10 s) and Fast Green 0.001% (5 min). Finally, sections were dehydrated in absolute ethanol and cleared in xylene. With FAST staining, acidic glycoproteins stained blue, neutral glycoproteins appeared orange and fibrous tissue appeared yellowish. After each staining and differentiation step, sections were briefly rinsed in distilled water and mounted in eukitt (EUKITT®, O. Kindler GmbH, Freiburg, Germany) after the staining procedure was completed.
2.5 Histological analysis
Histological analysis was performed with assistance of the digital image processing software for microscopes Axiovision Version 4.8.1 (Zeiss; AxioVision, Carl Zeiss Microscopy GmbH, Göttingen, Germany). Two investigators analysed a total of 57 intervertebral disc specimens. A mean of 17 (SD 3.7) axial sections from the centre of the intervertebral disc were analysed according to the criteria outlined in (Table 1).
Table 1. Applied histological criteria
FAST, Fast Green, Alcian Blue, Safranin-O, tartrazine, according to modified protocols of Leung et al. (2009)
State of annulus defect
• Annulus defect was defined as closed, when it was found completely sealed with repair tissue and/or annulus implant in all sections analysed
Modified Masson's trichrome (mt), FAST, Toluidine Blue–Fast Green (tb-fg)
• A 8 mm range of interest (ROI) adjacent to both sides of the annulus defect was defined for analysis
mt, FAST, tb-fg
• Moderate annulus damage was defined as few or none delaminations of annular fibres in ROI
• Severe annulus damage was defined as pronounced delamination and disorganization of annular fibres in ROI
Repair tissue thickness
• Three measurements of repair tissue thickness were performed in each axial section: in the midsection of the defect and at both integration zones with the adjacent AF, respectively.
mt, FAST, tb-fg
• A mean value was calculated for each group at each time-point
Supra-annular scar tissue
• measurement of supra-annular scar tissue formation was performed at the thickest part of scar tissue between the outer border of repair tissue and/or implant and psoas muscle fibres, clearly identifiable in Masson's trichrome staining
Repair tissue integration
• Tissue continuity of repair tissue with each adjacent annulus fibrosus fibre was assessed and set into proportion
Toluidine Blue/Fast Green ratios
• Toluidine Blue and Fast Green staining proportions were assessed within the repair tissue and/or implant
• tb > fg (> ⅔), tb ≈ fg, tb < fg(< ⅔)
Cell proliferation cluster
• Nucleus pulposus in and under the defect as well as under the annulus fibrosus 8 mm adjacent to the annulus defect was defined as ROI
• Cell proliferation clusters were defined as clusters of ≥ 3 cells
2.6 Statistical analysis
Data analysis was performed with graphpad prism version 5.0d for Mac OS X (GraphPad Software, San Diego, CA, USA). Quantitative data from both groups at single time-points were compared using the nonparametric Mann–Whitney test (two-tailed). For data comparison at different time-points, the non-parametric Kruskal–Wallis test with Dunn's post test was applied. Categorical variables were analysed with Fisher's exact test (two-sided). The significance level for all the statistical tests was p = 0.05.
Sixteen sheep were included in the study. They did not show any anaesthesia or surgically related complications. There were no device-related or other health-related complications.
Contrast media leakage with provocative pressure testing directly after explantation was observed at a mean pressure of 0.53 MPa, with high variance within the groups. There were no statistical significant differences between groups after 2, 6 and 12 weeks. Similarly, there were no differences within groups at different time-points. The intact control did not show any contrast media leakage up to the pressure limit of 4.8 MPa of the system.
At the time-points 0 weeks and 2 weeks, all 10 specimens of the control defect group demonstrated uncontained herniated nucleus pulposus tissue in the annulus defects (Figure 2c,d). For the treated specimens, the annulus implant consistently provided an effective barrier for herniating nucleus pulposus tissue (Figure 2d,e) with no implant dislocation at all time-points. After 6 weeks and 12 weeks, both groups showed comparable results with regard to closing the annulus defect.
The implantation procedure inflicts annulus damage adjacent to the defect in comparison to sole injury of the annulus (Figure 2f). At later time-points, however, no differences were evident.
After 2 weeks, a homogeneous cell infiltration, including the central parts of the annulus implant, is observed (Figure 3a,b). After 6 weeks, the absorbable component (PGA) of the annulus implant completely degraded and degradation products could only sporadically be found in some specimens (Figure 3c,d). The remaining non-absorbable component (PVDF) was found to be steeped by repair tissue that impressed with directional fibre bundles. After 12 weeks, the density of the extracellular matrix considerably increased and a progressive flattening of the cells could be observed (Figure 3e,f).
Repair tissue thickness was significantly greater in the annulus implant group at all follow-ups (Figure 4a,b,d,e,g). After 2 weeks, the annulus implant group demonstrated a moderate biological integration with the host tissue (Figure 4i). With increasing follow-up, the stability and amount of repair tissue integration improved in both groups (Figure 4c,f,i). No pronounced foreign body reaction or granuloma formation was observed. Repair tissue in both groups showed similar appearances, although the repair tissue in the defect group appeared to be more highly vascularized in most specimens (Figure 4f). There was no difference in the amount of supra-annular scar tissue over the defect site between the two groups (Figure 4h).
At 6 weeks follow-up, the control defect group displayed higher proportions of Toluidine Blue staining that were site-concordant with bluish staining in FAST, indicating high proteoglycan proportions in the repair tissue (Figure 5c). In contrast, at later follow-up, Fast Green staining dominated the repair tissue in both groups (Figure 5a–c), which was site-concordant with yellow-orange staining in FAST, indicating a more fibrous tissue with lower proteoglycan concentrations.
The 6-week follow-up was accompanied by the pronounced appearance of disc cell proliferation clusters in the nucleus pulposus of both groups (Figure 5d–f), indicating a degenerative process. These cell clusters were predominately located at the base of the annulus defect.
Early reports on repair mechanisms of AF defects have already described the formation of a fibrous cap in the outer regions of the defects, leaving the inner regions unrepaired (Key and Ford, 1948; Smith and Walmsley, 1951). Further studies reported on the specific repair characteristics of different defect geometries (Hampton et al., 1989; Ahlgren et al., 1994; Ethier et al., 1994). In this study, large box defects were applied with the rationale that, in the case of human IVD herniation, large AF tissue defects are associated with the highest reoperation rate of up to 21% because of reherniation (Carragee et al., 2003).
In up to 2 weeks follow-up, the annulus implant group showed a clear advantage in terms of providing a mechanical barrier for herniating nucleus pulposus tissue in comparison to the control defect group (Figure 2d). This indicates a good primary stability in the phase preceding stable biological integration of the implant. In this animal model, biological integration of the implant was seen after 6 weeks follow-up (Figure 4i). For human application, in a degenerative AF environment, biological integration might take considerably longer. Primary stability of the annulus implant is therefore likely an important aspect for this therapeutic strategy.
Box defects in canine and caprine models were reported to heal from the periphery inward with fibrous tissue (Hampton et al., 1989; Ethier et al., 1994). Similarly, a consistent repair of the control defects could be observed after 6 weeks (Figure 2d). At all follow-ups, however, a superior repair tissue thickness was shown with the annulus implant (Figure 4g). Although the stability results after 2 weeks are very clear (Figure 2d), the data from this study does not allow the conclusion whether superior repair tissue thickness translates into higher protection against herniation incidents at later time-points. The results, however, indicate our original hypothesis to be correct that repair tissue thickness can be improved by providing a biological scaffold in the inner regions of the AF, which are otherwise left unrepaired. After 2 weeks, a homogenous cell infiltration of the annulus implant was detected (Figure 3a,b). After 6 weeks, directional collagen fibre bundles were observed (Figure 3c,d). The biomechanical forces operating on the repair tissue probably cause the directional alignment.
In this study, contrast media leakage was not able to differentiate reliably between the different groups and time-points. The unexpected observation that even fresh annulus defects would display considerable containment characteristics points to a predominant influence of the nucleus pulposus on contrast media containment. There, macroscopically and histologically (Figure 2c), the migration of nucleus pulposus into untreated annulus defects was observed, probably resulting in a tamping effect with similar sealing characteristics with regard to contrast media as with the annulus implant. Concordantly, histological analysis in the early follow up showed high proteoglycan contents in the defect group, resulting in later fibrous transformation; whereas the implant group appears to already form more fibrous tissue in the early follow-up (Figure 5c). The biomechanical implications of these findings are, however, unclear.
In terms of primary or biological repair techniques for AF defects, two approaches can be identified in the literature: (1) suturing techniques for slit defects (Ahlgren et al., 2000; Heuer et al., 2008, Chiang et al., 2011) and (2) covering or plugging techniques for substantial tissue defects.
Closure strategies of substantial tissue defects are of high clinical relevance. One group attempted the sealing of an AF box defect with a non-absorbable viscoelastic polymer-based biomaterial that was injected into an ovine annulus defect (Brand, 2003). At explantation, the biomaterial was found outside of the AF in 100% of the cases, whereas traces of the biomaterial were located only in 42% in the AF defect. This example illustrates rather dramatically that implant dislocation is one of the major problems in the field and underlines the need for fixation strategies for primary implant stability. Similarly, in another study reporting the use of absorbable collagen–glycosaminoglycan scaffolds, with and without autologous AF cells, plugged into box defects of a caprine model (Saad, 2007), the nucleus was often seen bulging out of the disc and a weak tissue integration of the scaffold in the defect was reported. The additional presence of autologous AF cells did not appear to have any effect. Addressing the problem of fixation, Ahlgren et al. (2000) sutured a fascial autograft over box defects in an ovine model. No significant difference was found between repaired and non-repaired annular strength, evaluated by pressure-volume testing. Ledet et al. (2009) opted for bone screw fixation to affix a small intestinal submucosa patch and plug to the outside of a box defect in an ovine model). This bioactive implant was evaluated 24 weeks after implantation. According to the magnetic resonance imaging results, the implant appears to decelerate the degenerative process in comparison with the control defect. Moreover, a superior mean maximum pressure before disc failure is reported in comparison with the control defect. Unfortunately, 50% of the implant group displayed considerably large bridging osteophytes that would be worrying in a clinical application.
In this study, the implantation site at the inner region of the AF proved to be advantageous in terms of implant dislocation. No critical implant dislocation was observed. As expected, the implants bulged into the defect but never exceeded the outer annulus border. Moreover, no pronounced inflammatory reactions or expansive scar tissue were observed in comparison with the control defect. However, this implantation method is demanding, and multiple attempts for correct suture placement. This procedure inflicts damage to the AF adjacent to the defect (Figure 2f). This underlines the importance of a precise application instrument for the implantation procedure to avoid unnecessary perforation of the adjacent annulus and to allow a practicable and safe workflow in a clinical setting where the spinal canal with its nervous structures has to be passed for implant application. Safety, precision and reproducibility of the application procedure, along with the integration into the surgeons’ workflow, will be the major determinants for acceptance in the surgical community and we expect to see a considerable evolution from the basic technique presented in this study. After 6 weeks, progressive annulus damage could also be observed around the control defects (Figure 2f). This is probably because of the underlying degenerative process affecting both groups equally in this study after applying an annulus defect (Figure 5d–f) (Johnson et al., 2001; Guder et al., 2009). In contrast, the acute damage inflicted with implantation appear to have a moderate healing potential. Thus, at later time-points, no difference of annulus damage between both groups was seen. Therefore, this fixation method is possibly a valuable option, especially in the light of the lack of promising alternatives.
In conclusion, the biointegrative annulus implant investigated showed promising results with regard to biointegration, enhancement of repair tissue and function as a mechanical barrier in an ovine model.
Conflict of interest
Christian Kaps is an employee of TransTissue Technologies GmbH (Berlin, Germany). The other authors declare that they have no additional competing interests.
This study was supported by the Federal Ministry of Education and Research (BioInside 13N9831, 13N9830 and 13N9827). The authors are very grateful to Prof. Lothar Schilling for his technical advice, to Marijas Jurisic and Michaela Endres for their technical assistance, to Thorsten Deichmann and Annahit Arshi from the Institut für Textiltechnik, Faculty for Mechanical Engineering, RWTH Aachen University, for providing the annulus implant, to Prof. Kirsten Schmieder for her organizational support and to Dr James Melrose for sharing some of his excellent histological insights.