Current understanding of cellular and molecular events in intervertebral disc degeneration: implications for therapy

Transcription

1 Journal of Pathology J Pathol 2002; 196: Published online 11 February 2002 in Wiley InterScience ( DOI: / path.1050 Review Article Current understanding of cellular and molecular events in intervertebral disc degeneration: implications for therapy A. J. Freemont*, A. Watkins, C. Le Maitre, M. Jeziorska and J. A. Hoyland Musculoskeletal Research Group, The Medical School, Stopford Building, University of Manchester, Manchester M13 9PT, UK * Correspondence to: Professor A. J. Freemont, Musculoskeletal Research Group, The Medical School, Stopford Building, University of Manchester, Manchester M13 9PT, UK. tony.freemont@man.ac.uk Received: 17 September 2001 Accepted: 12 October 2001 Abstract Until recently, material removed from the intervertebral disc (IVD) at surgery consisted either of loose bodies from the centre of the IVD or discal tissue displaced (prolapsed) into the intervertebral root or spinal canals. This material is best regarded as a by-product of disc degeneration and therefore not representative of the disease process itself. Recent advances in surgical techniques, particularly anterior fusion, in which large segments of the anterior part of the IVD are excised with the anatomical relationships between different components intact, have generated material that can be investigated with modern molecular and cell biological techniques. This is an important area of study because degeneration of the lumbar IVDs is associated, perhaps causally, with low back pain, one of the most common and debilitating conditions in the West. Degeneration carries implications of inevitable progression of wear-and-tear associated conditions. Modern research on human IVD tissue has shown that this is far from the case and that disruption of the micro-anatomy described as degeneration is an active process, regulated by locally produced molecules. The exciting consequence of this observation is the possibility of being able to inhibit or even reverse the processes of degeneration using targeted therapy. Copyright # 2002 John Wiley & Sons, Ltd. Keywords: intervertebral disc degeneration; angiogenic molecules; neurogenic molecules; cytokines Chronic low back pain and the IVD Approximately 70% of the population will experience low back pain (LBP) during their lives [1] and it accounts for about 15% of all sickness leave in the UK [2]. Although LBP constitutes an important public health issue, little is known of its pathogenesis. The organs that together make up the spine and its accessory structures are so numerous and complex that the causes of LBP are undoubtedly multifactorial. However, clinical, interventional, and imaging studies on volunteers and patients [3 5] and therapeutic trials [6] have produced evidence implicating mechanical or traumatic disorders of the intervertebral disc (IVD) in up to 40% of cases. In almost every case when IVD tissue has been examined from patients with LBP, it has shown the features typical of discal degeneration (see below). Structure and function of the IVD The three components of the IVD are: (i) The end-plate This structure consists of a layer of cartilage, resembling articular cartilage that covers the central parts of the inferior and superior cortical bone surfaces of the vertebral bodies. Ex vivo vertical loading experiments of isolated spinal segments show that this is the weakest part of the spinal column [7]. (ii) Nucleus pulposus (NP) The space between the end-plates of adjacent vertebrae is filled by the nucleus pulposus, consisting of chondrocytes within a matrix of type II collagen and proteoglycan, mainly aggrecan [8]. Surprisingly little is known of the structure of the NP or the biology of its cells. The type II collagen fibres are not believed to give the same level of order to the structure or the same degree of mechanical stability to the matrix as in articular cartilage. The proteoglycans are hydrophilic, causing the NP to swell. The swelling pressure of the NP proteoglycans is constrained by the end-plates above and below and the annulus fibrosus (see below) around the periphery. On H&E sections, the NP appears homogeneous and pale lilac-blue, consistent with its complement of proteoglycans. In polarized light, it exhibits little birefringence. (iii) Annulus fibrosus (AF) This comprises dense sheets of highly orientated collagen fibres (mainly type I but also types II and III) in which are cells with the morphology and phenotype of fibroblasts. Our experience is that when cultured in alginate they can be transformed into chondrocytes. Functionally, the annulus fibrosus is a very Copyright # 2002 John Wiley & Sons, Ltd.

2 Current understanding of cellular and molecular events in IVD degeneration 375 strong ligament binding together the outer rims of adjacent vertebrae. On H&E sections, it exhibits fairly uniform eosinophilia and in polarized light, its constituent alternating bands of highly orientated collagen fibres are clearly seen. The IVD is a joint and, as such, allows movement between bones, in this case adjacent vertebral bodies. In its normal fully hydrated state, each IVD permits small degrees of flexion, extension, lateral bending, and twisting, but is resistant to compressive loads. When summated, the small movements at each IVD permit great mobility of the entire spine. Degeneration of the IVD It is believed that many of the disorders of the IVD are mechanical in origin. The erect spine is subjected to a range, rate, and degree of loading that arguably it is poorly designed to sustain. As a consequence, it suffers mechanical damage, which, in a high proportion of individuals, results in a pattern of morphological and histological changes, well characterized from autopsy studies, which together constitute discal degeneration. With the exception of the outer third of the AF, the normal adult IVD is avascular and aneural; there is a relatively clear demarcation between the AF and the NP; and the end-plate forms an intact layer of cartilage overlying cortical bone. In degeneration, the NP is disrupted, with changes in the proportion and types of proteoglycans and collagens (making the NP more eosinophilic), a reduction in the total number of lacunae containing viable chondrocytes, but the formation of chondrocyte clusters in many lacunae, just as in osteoarthritis of articular cartilage, which may even increase the total number of cells present in the diseased tissue. In addition, there is breakdown of the matrix with the formation of permeative slit-like spaces. There is often also disruption of the collagen fibre arrays in the AF, traumatic damage to the end plate, and vessel and nerve ingrowth into the inner AF and NP. Current understanding of the biology of connective tissues would causally implicate alterations in the function of local cells in these events. Because man is the only obligate bipedal vertebrate, replicating human disorders of the IVD in other animals has proved problematic. As a consequence, there are no satisfactory animal models of human IVD disease, which has inhibited the study of this group of disorders. An alternative strategy is to study human tissue removed from cadavers or from patients at surgery. This has been facilitated by the development of modern tissue-probing techniques [9 12] that can be performed on histological sections of human tissue, such as immunohistochemistry, in situ hybridization, in situ RT-PCR, and lectin histochemistry, as well as by new surgical approaches to the management of IVD disease. The recent advent of an anterior approach to the diseased IVD to facilitate disc replacement or spinal fusion has generated large wedges of discal tissue for study, often, in the case of spinal fusion, from multiple levels. These are a valuable source of material because they consist of primary diseased tissue in which the components of the IVD are in the same anatomical relationship to one another as in vivo. There has also been a trend towards pre-operative discography, in which IVDs are injected with a radioopaque medium. This is performed under aware state anaesthesia so that the patient can report if discography recapitulates symptoms. Either pain level or non-pain level (or both) IVDs may be removed depending on the clinical indications, particularly in multi-level fusion surgery, and this gives the pathologist the opportunity to relate findings to symptomatology. In scientific terms, studies of random samples of human tissue cannot be as satisfactory as controlled experiments. Nevertheless, they have generated useful data from which has come the realization that the origins of IVD degeneration are neither as passive nor as inevitable as the term implies. This has, in turn, opened the way for a re-evaluation of the management of IVD disease and LBP. Recent advances in the understanding of disc degeneration The problems of controls Two major issues have dictated the pace and direction of research into the processes of IVD degeneration: the adequacy or otherwise of control material and the difficulty in defining a recognized starting point from which to assess degenerative change. Tissue acquisition Until recently, difficulty in accessing relatively intact normal human IVD, other than from cadavers, has meant that a greater emphasis has been placed on those aspects of the IVD that do not change significantly in the first few hours after death, such as matrix chemistry and disc biomechanics. As spinal surgery, particularly spinal reconstruction for metastatic malignancy, has advanced, this situation has changed and with it, opportunities have developed to obtain an improved understanding of the biology of the IVD cells in discal degeneration. Ageing and load The nature of collagenous tissues means that their structure changes with time and with the organism s age. Continuing post-synthetic changes in the structure of many matrix molecules and alteration in the nutritional status of these poorly vascularized tissues with age lead to modifications in the collagen and proteoglycan composition of the IVD [13]. As the incidence of discal degeneration also increases with age, distinguishing normal ageing from disease [14] becomes paramount. This is complicated by the high

3 376 A. J. Freemont et al. frequency of disc degeneration at some spinal levels (e.g. L3 4, L4 5, and L5 S1), making the definition of normality problematic. In almost every connective tissue there are cells that respond to differential loading by changing their biology [15] and the IVD is no different [16]. Lifestyle, body weight, and age are major factors that influence the load environment of the normal IVD. In patients with LBP, pain and stiffness also play a part, through the natural protective reaction of abnormal prolonged contraction of specific muscle groups. Such factors further complicate the starting point from which to assess IVD degeneration. Altered matrix composition and integrity When compared with age- and sex-matched controls, there are profound alterations in the cell biology of chondrocytes in degenerate IVD. Normal discal chondrocytes are characterized by expression of type II collagen and proteoglycans [17] and we have derived evidence that this is regulated by the master chondroregulatory gene SOX-9 (unpublished observations). In discal degeneration, chondrocyte synthesis of matrix molecules changes differentially with the degree of degeneration [18], leading to an increase in the synthesis of collagens I and III and decreased production of aggrecan. Exactly what effect these changes have on IVD function is unknown. Furthermore, the regulation of matrix turnover is deranged, affecting both synthesis and degradation [19]. In degeneration, there is a net increase in matrix-degrading enzyme activity over natural inhibitors of such activity, which leads to loss of discal matrix. Particular attention has been paid to the role of matrix metalloproteinases (MMPs) in these processes [20 23], making them a potential target for therapy designed to inhibit discal degeneration. More recently, following work on the mechanism of aggrecan degradation in articular cartilage, interest has grown in the possible role of aggrecanases in IVD degeneration [19]. Aggrecan has two cleavage sites, one acted upon by MMPs and the other by members of a group of enzymes called the ADAMs family, after their hybrid function A Disintegrin And Metalloproteinase. In fact, the aggrecanases are two enzymes, ADAMTS4 and 5, that in addition to their disintegrin and metalloproteinase function also have ThromboSpondin motifs. From the pathological perspective, these studies have largely been undertaken, not by looking for the enzymes themselves, but through the application of antibodies targeted at the breakdown product of aggrecan formed by enzyme action at the specific site [19]. Reduced cell number The reduction in chondrocytes that typifies IVD degeneration has been ascribed to apoptosis [24]. Indeed, there is some evidence of a dose-dependent relationship between apoptosis and excessive load (lifestyle, body weight, age) [25]. As in other chondroid tissues, nitric oxide has been implicated in the induction of apoptosis [26]. Nerve and blood vessel ingrowth Although the normal adult IVD is avascular and aneural, nerves [27] and blood vessels [28,29] grow into diseased IVD. One avenue of investigation has been the local production of angiogenic and neurogenic molecules within degenerate IVD. Expression of the potent angiogenic factor vascular endothelial growth factor (VEGF) [30] has been demonstrated within the IVD. We have just completed an investigation of the active (b) chain of nerve growth factor (NGF) expression by cells in the IVD. NGF is synthesized by blood vessels in discs showing nerve ingrowth. Furthermore, the small nerves adjacent to the vessels express the high-affinity receptor for NGF, Trk-A. The implication of these data is that while either angiogenesis or neuronogenesis could be targets for therapy, angiogenesis drives nerve ingrowth and may be the more important from a therapeutic perspective. Cytokines as regulators of disease processes Recently there has been growing interest in the possible role of cytokines in regulating the connective tissue degradation, nerve and vessel ingrowth, and macrophage accumulation that characterize IVD degeneration. A number of cytokines have been implicated, including TNF, IL-1, PDGF, VEGF, IL-6, IGF-1, TGF-b, EGF, FGF, and IL-10, amongst others [23,30,31]. Of these, IL-1 is particularly interesting, because in other settings there is evidence that it is involved in cartilage homeostasis [32], largely through its ability to switch chondrocytes from anabolism to catabolism, inducing cartilage breakdown at molecular and morphological levels [33,34]. It has also been shown to be a regulator of angiogenesis in joints and non-articular cartilage, possibly by acting as a promoter of the potent angiogenic factor VEGF [35,36]. In the IVD, both its isoforms, IL-1a and IL-1b have been shown [37 39] to increase proteoglycan release and degradative enzyme production. IL-1b also increased production of MMPs and pain mediators, such as the eicosanoid prostaglandin E2, by human IVD cells [40]. IL-1 production by human IVD cells has also been reported [41]. Although not by any means conclusive, these studies implicate IL-1 both in matrix degradation and in pain. In addition, there is evidence that IL-1 could mediate some of the other characteristics of IVD disease, such as nerve and vessel ingrowth [42]. If IL-1 is the regulator of cartilage catabolism, the transforming growth factor beta (TGF-b) superfamily is the regulator of cartilage anabolism. Although there have been few experiments investigating the presence or effects of TGF-b on discal cartilage metabolism,

4 Current understanding of cellular and molecular events in IVD degeneration 377 Nishida et al. [43] induced a two-fold increase in proteoglycan synthesis by rabbit nucleus pulposus cells following injection of rabbit discs with an adenoviral vector carrying a human TGF-b transgene. This will undoubtedly turn out to be a key experiment in this area. There is a considerable amount still to be discovered about the role of cytokines in discal degeneration. However, these pioneering experiments hint at possibilities that have not previously been considered, both for improving understanding of the processes leading to IVD degeneration and for its management. The exciting aspect of this work is that the interest in the cytokine biology of the IVD comes at a time when advancing technology makes it possible to advance the study more rapidly and comprehensibly than has hitherto been possible. For instance, we have undertaken cytokine microarray analysis of cdna derived from RNA extracted from diseased human IVD. The results, showing expression of at least ten different cytokines working through two separate intracellular signalling pathways, are an indication of the extent of cytokine activity in the diseased IVD. The question still remains as to what initiates cytokine-mediated events and how this might be linked to the association between load and disc degeneration. Clinical implications of these studies Currently, the mainstays of treatment for damaged IVD are either symptomatic medical therapy or surgery, both of which have limited success. The novel data that are starting to accumulate hold out the possibility of managing IVD degeneration by normalizing the cell biology of the NP and the AF. A working hypothesis is that discal degeneration is an active process, induced by abnormal load and mediated by cytokines. We and others are currently examining the possibility of using genes introduced into target cells to produce proteins within the degenerate disc that will create a chemical environment that is conducive to restoring cell function towards normality. This is helped by the relatively immune privileged and hypovascular nature of the IVD, which allows the use Normal Degenerate Cytokine mediated events Abnormal load Cytokine Synthesis Changed matrix synthesis Degradative enzyme synthesis NP Neuronogenic and angiogenic cytokines Cell proliferation Slit Changed chemical environment Figure 1. The pattern of cellular events that lead from normality to degeneration as currently understood. Abnormal load (usually repetitive) sometimes associated with an abnormal chemical environment leads to cytokine synthesis by the chondrocytes within the nucleus pulposus. Either through direct cytokine-mediated events (e.g. angiogenesis and nerve ingrowth) or through secondary phenomena, the changes that histopathologists recognize as degeneration [e.g. cell proliferation (and apoptosis), matrix degradation, and changes in the balance between collagen and proteoglycan synthesis] occur

5 378 A. J. Freemont et al. of viral vectors for gene delivery whilst minimizing the risk of an immune response against infected cells and of dispersal of the gene product widely throughout the body [43,44]. The problem still remains of establishing a normal load environment within the damaged IVD in which genetically engineered cells might thrive, but combinations of physiotherapy and/or modern implant technology, with the potential to fashion support devices with biodegradable smart matrices that act as scaffold, load normalizer and cell stimulator, make this an opportune time to be a connective tissue pathologist with a personal interest in the management of back pain. Acknowledgements We wish to acknowledge the support of the Wellcome Trust (SHoWCASe awards /Z/99 and /Z/0), the Arthritis Research Campaign (Project grant F0531, ICAC grant F0551), the MRC (Co-operative Group Grant G ), and the joint Research Councils (MRC, BBSRC, EPSRC) UK Centre for Tissue Engineering (34/TIE13617). References 1. Macfarlane GJ, Thomas E, Croft PR, Papageorgiou AC, Jayson MI, Silman AJ. Predictors of early improvement in low back pain amongst consulters to general practice: the influence of premorbid and episode-related factors. Pain 1999; 80: Jayson MI. The Lumbar Spine and Back Pain (3rd edn). Churchill Livingstone: Edinburgh, Kelgren JH. The anatomical source of back pain. Rheumatol Rehab 1977; 16: Lam KS, Carlin D, Mulholland RC. Lumbar disc high-intensity zone: the value and significance of provocative discography in the determination of the discogenic pain source. Eur Spine J 2000; 9: Luoma K, Riihimaki H, Luukkonen R, Raininko R, Viikari- Juntura E, Lamminen A. Low back pain in relation to lumbar disc degeneration. Spine 2000; 25: Barrick WT, Schofferman JA, Reynolds JB, et al. Anterior lumbar fusion improves discogenic pain at levels of prior posterolateral fusion. Spine 2000; 25: Holmes AD, Hukins DWL, Freemont AJ. End plate displacement during compression of lumbar vertebra-disc vertebra segments and the mechanism of failure. Spine 1993; 18: Taylor JR, Scott JE, Cribb AM, Bosworth TR. Human intervertebral disc acid glycosaminoglycans. J Anat 1992; 180: Freemont AJ, Hampson V, Tilman R, Goupille P, Taiwo Y, Hoyland JA. Gene expression of matrix metalloproteinases 1, 3, and 9 by chondrocytes in osteoarthritic human knee articular cartilage is zone and grade specific. Ann Rheum Dis 1997; 56: Hoyland JA, Mee AP, Baird P, Braidman IP, Mawer EB, Freemont AJ. Demonstration of estrogen receptor mrna in bone using in situ reverse-transcriptase polymerase chain reaction. Bone 1997; 20: Freemont AJ, Byers RJ, Taiwo YO, Hoyland JA. In situ zymographic localisation of type II collagen degrading activity in osteoarthritic human articular cartilage. Ann Rheum Dis 1999; 58: Baris C, Carter DH, Freemont AJ, Thorp BH, Braidman IP. A method for immunofluorescent localisation of oestrogen receptors in bone sections from an egg-laying poultry strain. Avian Pathol 1998; 27: Scott JE, Bosworth TR, Cribb AM, Taylor JR. The chemical morphology of age-related changes in human intervertebral disc glycosaminoglycans from cervical, thoracic and lumbar nucleus pulposus and annulus fibrosus. J Anat 1994; 184: Repanti M, Korovessis PG, Stamatakis MV, Spastris P, Kosti P. Evolution of disc degeneration in lumbar spine: a comparative histological study between herniated and postmortem retrieved disc specimens. J Spinal Disord 1998; 11: Fujisawa T, Hattori T, Takahashi K, Kuboki T, Yamashita A, Takigawa M. Cyclic mechanical stress induces extracellular matrix degradation in cultured chondrocytes via gene expression of matrix metalloproteinases and interleukin-1. J Biochem (Tokyo) 1999; 125: Handa T, Ishihara H, Ohshima H, Osada R, Tsuji H, Obata K. Effects of hydrostatic pressure on matrix synthesis and matrix metalloproteinase production in the human lumbar intervertebral disc. Spine 1997; 22: Chelberg MK, Banks GM, Geiger DF, Oegema TR Jr. Identification of heterogeneous cell populations in normal human intervertebral disc. J Anat 1995; 186: Antoniou J, Steffen T, Nelson F, et al. The human lumbar intervertebral disc: evidence for changes in the biosynthesis and denaturation of the extracellular matrix with growth, maturation, ageing, and degeneration. J Clin Invest 1996; 98: Roberts S, Caterson B, Menage J, et al. Matrix metalloproteinases and aggrecanase: their role in disorders of the human intervertebral disc. Spine 2000; 25: Kanemoto M, Hukuda S, Komiya Y, Katsuura A, Nishioka J. Immunohistochemical study of matrix metalloproteinase-3 and tissue inhibitor of metalloproteinase-1 human intervertebral discs. Spine 1996; 21: Liu J, Roughley PJ, Mort JS. Identification of human intervertebral disc stromelysin and its involvement in matrix degradation. J Orthop Res 1991; 9: Nemoto O, Yamagishi M, Yamada H, Kikuchi T, Takaishi H. Matrix metalloproteinase-3 production by human degenerated intervertebral disc. J Spinal Disord 1997; 10: Kang JD, Georgescu HI, McIntyre-Larkin L, Stefanovic-Racic M, Donaldson WF 3rd, Evans CH. Herniated lumbar intervertebral discs spontaneously produce matrix metalloproteinases, nitric oxide, interleukin-6, and prostaglandin E2. Spine 1996; 21: Gruber HE, Hanley EN Jr. Analysis of aging and degeneration of the human intervertebral disc. Comparison of surgical specimens with normal controls. Spine 1998; 23: Lotz JC, Colliou OK, Chin JR, Duncan NA, Liebenberg E. Compression-induced degeneration of the intervertebral disc, an in vivo mouse model and finite-element study. Spine 1998; 23: Kohyama K, Saura R, Doita M, Mizuno K. Intervertebral disc cell apoptosis by nitric oxide: biological understanding of intervertebral disc degeneration. Kobe J Med Sci 2000; 46: Freemont AJ, Peacock TE, Goupille P, Hoyland JA, O Brien J, Jayson MI. Nerve ingrowth into diseased intervertebral disc in chronic back pain. Lancet 1997; 350: Kauppila LI. Ingrowth of blood vessels in disc degeneration. Angiographic and histological studies of cadaveric spines. J Bone Joint Surg Am 1995; 77: Yasuma T, Arai K, Yamauchi Y. The histology of lumbar intervertebral disc herniation. The significance of small blood vessels in the extruded tissue. Spine 1993; 8: Tolonen J, Gronblad M, Virri J, Seitsalo S, Rytomaa T, Karaharju EO. Platelet-derived growth factor and vascular endothelial growth factor expression in disc herniation tissue: an immunohistochemical study. Eur Spine J 1997; 6: Igarashi T, Kikuchi S, Shubayev V, Myers RR Volvo Award winner in basic science studies: Exogenous tumor necrosis factor-alpha mimics nucleus pulposus-induced neuropathology. Molecular, histologic, and behavioral comparisons in rats. Spine 2000; 25: Cawston T, Billington C, Cleaver C, et al. The regulation of MMPs and TIMPs in cartilage turnover. Ann N Y Acad Sci 1999; 878: Nishida Y, D Souza AL, Thonar EJ, Knudson W. Stimulation

6 Current understanding of cellular and molecular events in IVD degeneration 379 of hyaluronan metabolism by interleukin-1alpha in human articular cartilage. Arthritis Rheum 2000; 43: van Den Berg WB. The role of cytokines and growth factors in cartilage destruction in osteoarthritis and rheumatoid arthritis. Z Rheumatol 1999; 58: Gerber HP, Vu TH, Ryan AM, Kowalski J, Werb Z, Ferrara N. VEGF couples hypertrophic cartilage remodeling, ossification and angiogenesis during endochondral bone formation. Nature Med 1999; 5: Jackson JR, Minton JA, Ho ML, Wei N, Winkler JD. Expression of vascular endothelial growth factor in synovial fibroblasts is induced by hypoxia and interleukin 1beta. J Rheumatol 1997; 24: Shinmei M, Kikuchi T, Yamagishi M, Shimomura Y. The role of interleukin-1 on proteoglycan metabolism of rabbit annulus fibrosus cells cultured in vitro. Spine 1988; 13: Maeda S, Kokubun S. Changes with age in proteoglycan synthesis in cells cultured in vitro from the inner and outer rabbit annulus fibrosus. Responses to interleukin-1 and interleukin-1 receptor antagonist protein. Spine 200 ; 25: Rannou F, Corvol MT, Hudry C, et al. Sensitivity of annulus fibrosus cells to interleukin 1 beta. Comparison with articular chondrocytes. Spine 2000; 25: Kang JD, Stefanovic-Racic M, McIntyre LA, Georgescu HI, Evans CH. Toward a biochemical understanding of human intervertebral disc degeneration and herniation. Contributions of nitric oxide, interleukins, prostaglandin E2, and matrix metalloproteinases. Spine 1997; 22: Takahashi H, Suguro T, Okazima Y, Motegi M, Okada Y, Kakiuchi T. Inflammatory cytokines in the herniated disc of the lumbar spine. Spine 1996; 21: Lindholm D, Heumann R, Meyer M, Thoenen H. Interleukin-1 regulates synthesis of nerve growth factor in non-neuronal cells of rat sciatic nerve. Nature 1987; 330: Nishida K, Kang JD, Gilbertson LG, et al. Modulation of the biologic activity of the rabbit intervertebral disc by gene therapy: an in vivo study of adenovirus-mediated transfer of the human transforming growth factor beta 1 encoding gene. Spine 1999; 24: Nishida K, Kang JD, Suh JK, Robbins PD, Evans CH, Gilbertson LG. Adenovirus-mediated gene transfer to nucleus pulposus cells. Implications for the treatment of intervertebral disc degeneration. Spine 1998; 23: