Convection-enhanced delivery of botulinum toxin serotype A into the nonhuman primate cisterna magna and hippocampus

OBJECTIVE: Botulinum toxin serotype A (BoNT/A) was reported to raise the seizure threshold when injected into the seizure focus of a kindled rodent model. Delivering BoNT/A to the nonhuman primate (NHP) central nervous system via convection-enhanced delivery (CED) has not been performed. The objective of this study was to determine the toxicity and distribution characteristics of CED of BoNT/A into the NHP hippocampus and cisterna magna. METHODS: Escalating BoNT/A doses were delivered by CED into the NHP hippocampus (n = 4) and cisterna magna (n = 5) for behavioral and histological assessment and to determine the highest nonlethal dose (LD0) and median lethal dose (LD50). Hippocampal BoNT/A was coinfused with Gd-albumin, a surrogate MRI tracer. Gd-albumin and radioiodinated BoNT/A (125I-BoNT/A) were coinfused into the hippocampus of 3 additional NHPs to determine BoNT/A distribution by in vivo MRI and postmortem quantitative autoradiography. Scintillation counting of CSF assessed the flow of 125I-BoNT/A from the hippocampus to CSF postinfusion. RESULTS: LD0 and LD50 were 4.2 and 18 ng/kg, and 5 and > 5 ng/kg for the NHP hippocampus and cisterna magna, respectively. Gd-albumin and 125I-BoNT/A completely perfused the hippocampus (155-234 mm3) in 4 of 7 NHPs. Fifteen percent of BoNT/A entered CSF after hippocampal infusion. The MRI distribution volume of coinfused Gd-albumin (VdMRI) was similar to the quantitative autoradiography distribution of 125I-BoNT/A (VdQAR) (mean VdMRI = 139.5 mm3 [n = 7]; VdQAR = 134.8 mm3 [n = 3]; r = 1.00, p < 0.0001). No infusion-related toxicity was identified histologically except that directly attributable to needle placement. CONCLUSIONS: Gd-albumin accurately tracked BoNT/A distribution on MRI. BoNT/A did not produce CNS toxicity. BoNT/A LD0 exceeded 10-fold the dose administered safely to humans for cosmesis and dystonia.