What is the potential of oligodendrocyte progenitor cells to successfully treat human spinal cord injury?

It is possible to classify 'stem cells' into two broad categories, embryonic and somatic; the latter term including all stem cells present in the adult as well as those found in foetal and neonatal tissues [27]. Embryonic stem cells are derived from the human blastocyst and maintain the ability to differentiate into any cell type (pluripotency) [28], whilst somatic stem cells are able to differentiate into a limited number of cell lineages (multipotency). It is this ability to differentiate into a variety of cell types, as well as secrete growth factors, that has gained interest from the field of spinal cord injury. It is hoped that stem cells can be used to replace neural cells lost during the injury process and stimulate a repair environment [29]. A number of stem cell types including mesenchymal stem cells, olfactory ensheathing cells and neural stem cells have been considered for use in SCI. However, further discussion of the use of these cell types is outside the scope of this article and is well reviewed elsewhere [27, 29‐33].

The cell type currently being trialled in human patients, and the focus of this article, is the oligodendrocyte progenitor cell. These cells are derived from human embryonic stem cells by culturing them in conditions to drive commitment to an oligodendroglial lineage [34, 35]. Once purified, up to 95% of these cells display markers characteristic of oligodendroglial cells - GalC, RIP and O4 - and morphologically resemble oligodendroglial cells [26]. These progenitors can then be injected into the spinal cord where, in a demyelinating environment, they further differentiate into oligodendrocytes - the myelinating cells of the central nervous system [26]. In the human SCI lesion site, it is hoped that OPCs will work as a "combination therapy" [29] - phenotypically replacing lost oligodendrocytes and hence remyelinating axons that have become demyelinated during SCI, as well as secreting neurotrophic factors to establish a repair environment in the lesion [29].

One of the key premises of using OPC transplants for treatment of SCI is that demyelination is a major contributor to the pathogenesis of the injury. Clearly, in order for remyelinating therapies to succeed there must be appropriate targets - i.e. naked, intact axons - in which function is restored to some degree following treatment. Somewhat surprisingly, the existence of such targets following human SCI is controversial and casts doubt on much of the fundamental science underlying OPC treatment. Whilst a number of animal studies [36, 37] have demonstrated myelin loss with sparing of the associated axon - so-called 'primary demyelination' - the extent to which this occurs following human injury is ambiguous and a number of post mortem studies of human SCI have suggested that it may not occur to a significant degree [20, 22, 38].

It may be possible to explain the observed difference in the extent of primary demyelination between post mortem human samples and animal models. Experimental data demonstrates that axons need myelin for trophic support - mice missing two major myelin proteolipids, PLP and DM-20, typically show axonal swelling and degeneration [39]. Therefore any naked axons in humans may degenerate if they are not provided with support from oligodendrocytes and hence wouldn't be detected post mortem. Additional mouse data has supported this theory by showing a close correlation between myelin status and axonal survival [40].

Nonetheless, there are a number of inconsistencies with this explanation. Firstly, investigations in rats have identified chronically demyelinated axons [37], suggesting that some axons are capable of surviving without oligodendrocyte support. Additionally, were this explanation valid, then one would expect to see naked axons in acute or intermediate stage post mortem lesions, before they had time to degenerate. However, this is not reported to be the case [20, 22, 38].

In summary, the role of demyelination is far from clear. Whilst some animal studies [36, 37] suggest that naked, primary demyelinated axons - the target for OPC treatment - exist, post mortem human studies do not corroborate this [20, 22, 38]. This disparity could be due to the explanation given above, however, as discussed, this has a number of caveats. Nonetheless there could be other reasons for the conflicting results between human and animal studies such as the fact that post mortem samples - particularly from acute or intermediate phase lesions - necessarily come from injuries associated with high mortality and hence represent a more severe lesion. As suggested by Guest et al (2005), primary demyelination is more likely in less severe injury where axons can survive but oligodendrocytes cannot [38]. Perhaps, as suggested by Rowland et al, emerging magnetic resonance imaging technologies such as magnetization transfer and diffusion tensor imaging which allow the structural integrity of tissue to be imaged in live patients will clarify the situation in humans [19].

Overall, despite the prominent role for demyelination suggested in animal studies, a lack of consensus on the importance of this process in humans creates uncertainty regarding how successfully OPCs can treat human SCI. Indeed, even if remyelination is a successful strategy, there are many barriers to clinical improvement. For example, glial scar formation can create a physical barrier to OPCs whilst a number of molecules can have an inhibitory effect on OPCs. For example, both chondroitin sulphate proteoglycans (CSPGs) and TNF-α found within the scar have been shown to reduce OPC growth [41, 42], whilst other studies have shown that bone morphogenic protein (BMP) produced by astrocytes can reduce OPC differentiation into oligodendrocytes and promote differentiation into astroctyes [43].

OPC transplantation is a dual strategy, aiming not just to allow remyelination but also providing trophic support and a repair environment [44]. Spinal cord injury sets in motion a plethora of repair mechanisms, including endogenous remyelination and increased expression of a number of neurotrophic factors such as transforming growth factor-β2 (TGF-β2) and brain-derived neurotrophic factor (BDNF) [45, 46]. Such up-regulation is thought to contribute to neuroprotection and even axonal sprouting [47]. However, axonal sprouting and endogenous repair is often abortive - due, in part, to an insufficient repair environment to overcome inhibitory mechanisms or promote maintained regeneration [48]. Moreover, although endogenous remyelination is often seen following SCI, it is usually functionally and anatomically incomplete [48‐50]. It is therefore hoped that the grafting of a large number of OPCs soon after injury will create a 'repair environment' and allow these processes to develop further.

A host of evidence is weighing up in favour of this theory - OPCs are capable of producing numerous neurotrophins including midkine, TGF-β2 and BDNF, contributing to a repair environment [51] and even promoting neurite outgrowth of rat sensory neurons in vitro [51]. In vivo studies (discussed in more detail below) have also suggested that OPC transplantation can significantly alter lesion pathogenesis and influence gene expression towards an uninjured pattern [52]. Indeed, in one experimental system, exposure of media conditioned by OPCs and oligodendrocytes alone was enough to increase the survival and axonal lengthening of neurons [53]. Therefore it is possible that even in the absence of remyelination, OPC transplantation may bring clinical improvement by providing trophic support and creating a regenerative environment, preventing the cellular damage and apoptosis seen in the secondary injury phase.

Despite this, eliciting neural regrowth is an ambitious goal as despite early optimism, neurotrophin based therapies, which aim to induce a repair environment, have been largely unsuccessful [54‐56] and harnessing this process has remained elusive. Moreover, sprouting can also be pathogenic as new connections made by sensory afferents can lead to neuropathic pain [57] and autonomic dysreflexia [58]. Further, the inflammatory milieu following SCI is extremely complex and so the timing of any acute graft needs to be extremely carefully considered [59]. Hence whilst attempting to provide a supportive environment is a reasonable approach, it may be difficult to achieve in practice.

In summary, the potential benefit of remyelination remains controversial due to disagreements in the role of this process in human SCI. In addition, even if a remyelinating strategy is the correct one, there are a number barriers to success remaining including scar formation, inhibitory molecules and the complexity of the inflamed lesion. The latter also makes generation of a repair environment somewhat complex. These factors leave challenges to be overcome in the development of this therapy.

Despite much debate surrounding the underlying principles of the potential of OPC transplants, a number of whole body in vivo experiments have been conducted to assess OPC-based therapy. The most important of these in the public domain, cited on the announcement of the clinical trial [4], is that of Keirstead et al (2005) [44]. The authors successfully demonstrated that injection of OPCs into rats with a thoracic contusion SCI seven days post-injury led to remyelination and restoration of some locomotor function. A similar study into cervical SCI [52], conducted by the same group, obtained similar results - demonstrating that OPC transplantation can improve forelimb motor function as well as altering lesion pathogenesis; shown by increased white and grey matter sparing, decreased cavitation and altered gene expression [52]. This, too, was claimed to be "proof-of-concept" for the OPC clinical trials in a separate press release [60]. Subsequently, other groups have also shown neurological improvements following OPC injection into rodent models of SCI, further validating this approach [61‐63]. Whilst these papers [44, 52] undoubtedly offer strong evidence for the promise of OPC therapy, there are a number of considerations to be made before these results can be applied to human therapy.

Firstly, contusion injury was produced by an 'impactor' - a device which transiently delivers a specific force to the posterior of the spinal cord [64]. It is questionable how accurately this models human SCI where a variety of injury mechanisms often occur anteriorly [65], followed by prolonged compression, spinal fractures [66], haemorrhage and inflammation [67] and serious systemic injuries. Secondly, all SCIs were elicited under anaesthesia, followed by sterile wound closure and antibiotics [44, 52]. Anaesthesia is known to have neuroprotective properties [68, 69], and may therefore alter the injury outcome, whilst antibiotics prevent infection, a factor that can exacerbate human SCI [9]. Thirdly, all animals were surgically prepared before injury - the paravertebral muscles were dissected and the spinal laminae removed. This procedure is known to reduce secondary disease processes such as swelling and oedema, altering disease pathogenesis [70]. Furthermore, these studies used locomotor function as a guide for neurological improvement but made no attempt to examine sensory or autonomic function - something that any treatment of SCI needs to address. Additionally, in the 2005 study, locomotion was assessed, in part, using the Basso Beattie and Bresnahan (BBB) scale [44], the sensitivity and reproducibility of which has been criticised [71]. Finally it is of interest to note that the authors of the 2005 study declare "no potential conflict of interest" [52] despite their work being funded by the company running the clinical trial, and being named by the latter as 'collaborators' [60]. This is probably of no consequence, but increased transparency would likely bestow greater confidence in the results obtained.

Whilst it is possible to criticise the extrapolation of animal data to humans, particularly in the context of SCI [72], one must bear in mind that whilst small-animal models may be flawed, they are still an important tool and live in vivo experiments are a crucial element of preclinical investigation. In spite of the aforementioned conflict of interest, the studies were rigorously conducted with appropriate controls and blinding of investigators where necessary. The fact that significant positive results were obtained, for both thoracic and cervical injury, provides a reasonable basis for development of this treatment. Whilst historically, promising animal data has often translated poorly into clinical success [72], it remains to be seen whether this will be true in this case.