Glial-derived neurotrophic factor gene transfer for Parkinson's disease: Anterograde distribution of AAV2 vectors in the primate brain

Abstract

Delivery of neurotrophic factors to treat neurodegenerative diseases has not been efficacious in clinical trials despite their known potency for promoting neuronal growth and survival. Direct gene delivery to the brain offers an approach for establishing sustained expression of neurotrophic factors but is dependent on accurate surgical procedures to target specific anatomical regions of the brain. Serotype-2 adeno-associated viral (AAV2) vectors have been investigated in multiple clinical studies for neurological diseases without adverse effects; however the absence of significant clinical efficacy after neurotrophic factor gene transfer has been largely attributed to insufficient coverage of the target region. Our pre-clinical development of AAV2-glial-derived neurotrophic factor (GDNF) for Parkinson's disease involved real-time image guided delivery and optimization of delivery techniques to maximize gene transfer in the putamen. We have demonstrated that AAV2 vectors are anterogradely transported in the primate brain with GDNF expression observed in the substantia nigra after putaminal delivery in both intact and nigrostriatal lesioned primates. Direct midbrain delivery of AAV2-GDNF resulted in extensive anterograde transport to multiple brain regions and significant weight loss.

Introduction

An important therapeutic innovation with the potential to dramatically inhibit progression of the motor deficit in Parkinson's disease is the delivery of potent neurotrophic factors directly into the degenerated region of the brain. Delivery of a gene therapy vector or therapeutic agent into the brain via direct injection offers an effective method for bypassing the blood brain barrier and targeting a specific anatomical site. However, the large size and complex architecture of the human brain often prevents scalable translation of small animal experimental methods involving direct brain delivery into effective medical procedures. Previous clinical studies investigating the direct delivery of recombinant glial-derived neurotrophic factor (GDNF) protein into the putamen did not yield significant efficacy (Patel and Gill, 2007). The reason for this disappointing outcome is still unclear, but sub-optimal distribution of infused GDNF protein is suspected. The generation of anti-GDNF antibodies in some subjects, along with reported cerebellar toxicity in Rhesus macaques after cessation of prolonged putamen infusion of GDNF, diminished enthusiasm for this approach (Hovland et al., 2007, Lang et al., 2006). In contrast, gene therapy has become a more favored approach primarily because targeting of much lower levels of growth factor only to the nigrostriatal neurons seems to provide a way of harnessing the potency of neurotrophic factors while limiting the potential side effects that might be triggered through inadvertent escape of the neurotrophic factor into other parts of the brain or into peripheral circulation. Adeno-associated viral (AAV) vectors have emerged as the current vehicle of choice for neurological gene transfer with multiple clinical trials completed or initiated. Over 100 patients have received direct cranial delivery of AAV serotype 2 (AAV2) vectors in clinical trials for Parkinson's disease without any serious adverse events attributable to the vector (Christine et al., 2009, Kaplitt et al., 2007, LeWitt et al., 2011, Marks et al., 2008, Marks et al., 2010, Muramatsu et al., 2009, Muramatsu et al., 2010). The recently completed AAV2-GAD (glutamic acid decarboxylase) gene therapy trial for advanced Parkinson's disease is the first phase 2 double-blinded clinical study to demonstrate the efficacy of gene therapy in a neurological disorder (LeWitt et al., 2011).

The exploration of this technology, however, has revealed complex parameters that must be controlled in order to achieve efficacy without unacceptable side effects. The difficulty of achieving the requisite level of transgene expression over sufficient areas of the striatum to exert a positive effect in Parkinson's disease patients was encountered in a 58-patient controlled (sham surgery) phase 2 trial of AAV2-Neurturin (CERE-120). This trial was sufficiently powered to overcome the anticipated placebo effect prevalent in Parkinson's disease studies. In this study, a total of 16 vector injections were made via 8 needle passes that resulted in only about 15% coverage of the putamen, based on post-mortem analysis of 2 participants in the study who had died from unrelated causes (Bartus et al., 2011). This result suggested that perhaps inadequate distribution of vector within the putamen was responsible for the lack of clinical effect. In contrast, AAV2-mediated putaminal expression of aromatic l-amino acid decarboxylase (AADC), the rate-limiting enzyme required for conversion of l-dopa into dopamine, was substantial as shown by PET with the AADC-specific substrate, 6-[¹⁸F]-fluoro-metatyrosine (FMT) (Christine et al., 2009, Eberling et al., 2007, Muramatsu et al., 2010). The use of only two reflux-resistant infusion cannulae per hemisphere in the AAV2-hAADC study enabled pressurized convection-enhanced delivery (CED) that increases parenchymal distribution surrounding the cannula tip. Distribution of the vector infusate in these patients was visualized by postoperative MR imaging within 7 hours of vector delivery and was calculated to cover 22% of the putamen volume (Valles et al., 2010). Although the coverage of the putamen does not appear to be substantially different between the two delivery procedures adopted for these clinical trials, the CED approach required a smaller number of infusions and, therefore, lowered the risk of adverse events associated with intracranial cannula placement. It is, however, evident from these early studies that, without a standardized delivery system that can be implemented in multicenter studies and reliably provide a larger distribution of vector within the target brain structure, it will not be possible to fully evaluate the efficacy of therapeutic gene therapy vectors. Even in the successful AAV2-GAD study where gene transfer was targeted to a relatively small deep brain nucleus, the subthalamic nucleus, it was found that precise targeting was critical to clinical benefits, and no motor improvements were observed when the cannula was not correctly positioned (LeWitt et al., 2011).

Therefore, one of the greatest challenges in the development of therapeutic gene therapy for the treatment neurological disorders is to gain control over the distribution of gene transfer within the central nervous system (CNS). Neurotrophic factors play a critical role in maintaining the CNS and have long been proposed as therapeutic agents for neurodegenerative diseases. However, despite the promise that neurotrophic factors offer for neurodegenerative diseases, including nerve growth factor for Alzheimer's disease (Cattaneo et al., 2008), no neurotrophic factor has established clinical efficacy in a placebo-controlled trial. The challenges in delivering proteins, such as neurotrophic factors, to the brain are evident in the large number of animal and human studies that have focused on developing GDNF, a potent growth factor for dopaminergic neurons, into a treatment for Parkinson's disease (Kirik et al., 2004). Sub-optimal distribution has been blamed for the lack of clinical efficacy in investigations of GDNF or its homolog, Neurturin, to inhibit the progression of motor deficits in Parkinson's disease (Lang et al., 2006, Marks et al., 2010, Nutt et al., 2003). Therefore, a crucial component of our ongoing AAV2-GDNF gene therapy program for Parkinson's disease has been the parallel development of a real-time image-guided neurosurgical drug delivery system. This platform technology significantly enhances the accuracy of targeting deep brain structures and maximizes distribution of vector within the parenchyma surrounding the delivery catheter to ensure optimal coverage of the target structure. Repetitive magnetic resonance (MR) imaging during infusions into the non-human primate (NHP) brain provided real-time feedback of vector distribution that enabled us to optimize the cannula design and prescribe infusion protocols (i.e. infusion rates, volumes and cannula positioning) that minimize the risk of off-target distribution or backflow while maximizing target coverage (Yin et al., 2010, Yin et al., 2011). In collaboration with MRI Interventions (Irvine, CA, USA; formally SurgiVision Inc.), we have developed a reflux-resistant cannula with a 3-mm fused silica distal tip that is compatible with the FDA-approved Clearpoint® intra-operative MR trajectory guidance frame (Richardson et al., 2011a).

An unanticipated consequence of enhanced target coverage has been the repeated observation of AAV2-mediated transduction of neurons in anatomical areas of the primate brain not directly targeted but known to receive axonal projections originating from the site of vector infusion (Kells et al., 2009). A retrospective analysis of our recent studies shows significant evidence for anterograde, but not retrograde, transport of AAV2 vectors in the NHP brain. This anterograde trafficking of AAV2 may be advantageous when the secondary region is also a valid target for gene delivery. However, transgene expression within non-targeted areas of the brain may equally lead to serious adverse events such as the weight loss that occurred after intracerebroventricular delivery of GDNF protein to Parkinson's disease patients (Nutt et al., 2003).