GDNF-induced cerebellar toxicity: A brief review

Abstract

Recombinant-methionyl human glial cell line-derived neurotrophic factor (GDNF) is known for its neurorestorative and neuroprotective effects in rodent and primate models of Parkinson's disease (PD). When administered locally into the putamen of Parkinsonian subjects, early clinical studies showed its potential promise as a disease-modifying agent. However, the development of GDNF for the treatment of PD has been significantly clouded by findings of cerebellar toxicity after continuous intraputamenal high-dose administration in a 6-month treatment/3-month recovery toxicology study in rhesus monkeys. Specifically, multifocal cerebellar Purkinje cell loss affecting 1-21% of the cerebellar cortex was observed in 4 of 15 (26.7%; 95% confidence interval [CI]: 10.5-52.4%) animals treated at the highest dose level tested (3000μg/month). No cerebellar toxicity was observed at lower doses (450 and 900μg/month) in the same study, or at similar or higher doses (up to 10,000μg/month) in subchronic or chronic toxicology studies testing intermittent intracerebroventricular administration. While seemingly associated with the use of GDNF, the pathogenesis of the cerebellar lesions has not been fully understood to date. This review integrates available information to evaluate potential pathogenic mechanisms and provide a consolidated assessment of the findings. While other explanations are considered, the existing evidence is most consistent with the hypothesis that leakage of GDNF into cerebrospinal fluid during chronic infusions into the putamen down-regulates GDNF receptors on Purkinje cells, and that subsequent acute withdrawal of GDNF generates the observed lesions. The implications of these findings for clinical studies with GDNF are discussed.

Intermittent convection-enhanced delivery of GDNF into rhesus monkey putamen: absence of local or cerebellar toxicity

“…This hypothesis is consistent with the observed absence of cerebellar toxicity with IT (monthly) and ICV (biweekly and monthly) bolus administration of GDNF at doses of 2800–9333 µg/4 weeks given for 3–6 months (Hovland et al 2007; Luz et al 2016). The hypothesis was considered sufficiently robust to allow the selection of a single dose level for the present study and avoid an unnecessary use of animals.…”

“…This suggests that there was significant (and, probably due to rapid mixing of the aCSF and GDNF fractions of the infusate in the interstitial space, greater than anticipated) drug leakage into the CSF compartment. As a result, five animals had GDNF C CSF values > 1700 pg/mL at both time points and thus, were as close as possible with intermittent delivery to the hypothesized GDNF C CSF pattern underlying the cerebellar lesions associated with continuous IPu delivery (Luz et al 2016). The absence of cerebellar toxicity under these conditions is presumably due to the short (34 h) half-life of GDNF in CSF (Luz et al 2016), which leads to rapid elimination of the drug from CSF as evidenced by the fact that all pre-dose CSF samples were clear of GDNF.…”

“…In addition, in that study, an increased inflammatory reaction to the foreign protein was seen at the infusion site and along the catheter track with the same dose. Although the lower doses tested in the continuous dosing study were not associated with toxicologically relevant findings, and a 4-week brain dose of 840 µg GDNF ( C i : 0.2 µg/µL) was defined as the NOAEL, the cerebellar toxicity with the high dose was of significant concern (in particular because it was not clinically monitorable and did not produce MRI signals) and contributed to a temporary halt of the GDNF clinical development program in PD in 2004 (Luz et al 2016). …”

“…Samples from outside the CNS were preserved in 10% neutral buffered formalin (Medical Chemical, Torrance, CA, USA) or 4% paraformaldehyde (American MasterTech Scientific). In view of the large database available from previous toxicology studies (Luz et al 2016), these samples were archived but not analyzed. CNS samples (22 coronal brain slices including 3 cerebellar slices; cervical, thoracic and lumbar spinal cord; cervical, thoracic and lumbar dorsal root ganglia; and trigeminal ganglia) were initially preserved in 4% paraformaldehyde and then transferred to 10% neutral buffered formalin within 4 days.…”

“…Altogether, 11 studies involving 205 animals were performed, including 6 studies testing the intracerebroventricular (ICV) route of administration ( N  = 86), 1 intrathecal (IT) study ( N  = 17) and 4 intraputamenal (IPu) studies ( N  = 102); the findings from these studies have recently been reviewed (Luz et al 2016). …”