Hovland et al. (Toxicol Pathol 35[5]: 676-692, 2007) report  on cerebellar pathology in monkeys after exposure to high  concentrations of glial cell line-derived neurotrophic factor  (GDNF). Loss of Purkinje cells was seen in four animals. The  authors note that pathology could have resulted either from exposure to GDNF or to withdrawal from GDNF They then state  that the small number of affected animals precludes a definitive  conclusion as to the pathogenesis. I beg to differ. It is shown here that the data firmly support withdrawal from high doses as the mechanism.

First I marshal the facts. There was intention to treat forty-five monkeys; however, nine of them had CSF concentrations of GDNF that were never above zero. These nine were therefore never exposed to GDNF, presumably because of pump or catheter malfunction. Inclusion of these animals would make the argument below even stronger, but they are excluded because they were not exposed. Therefore, there are toxicity data only for thirty-six animals. In eleven of these thirty-six animals there was deliberate withdrawal from GDNF (the so-called Recovery group), and a further eight of thirty-six monkeys were inadvertently withdrawn (presumably because of catheter migration).

All four animals with cerebellar lesions belonged to the high dose subgroup of fifteen monkeys. Six of these fifteen animals had been withdrawn (five deliberately, one inadvertently) and all four of the affected animals were in this subgroup of six.

In summary, four of thirty-six animals had lesions. These animals belonged exclusively to the subgroup of fifteen animals exposed to high doses, and within this group of fifteen, they also belonged exclusively to the subgroup of six animals withdrawn from GDNF (i.e., four of these six animals had lesions). Thus, lesions were seen only in animals that had first been exposed to high doses and then withdrawn from these high doses. Lesions were not seen in any other group of animals.

I first ask, "If exposure at medium or low doses was sufficient to cause lesions, what is the probability that the four animals  with lesions coincidentally belonged to the high-dose subgroup and not to the other two groups?"

Of thirty-six animals exposed, there were fifteen animals in the high-dose subgroup, so this probability is just:

which is 0.023. Thus, beyond reasonable scientific doubt, pathology was associated only with treatment at the highest dose.

However, it is still not known if lesions were caused by exposure to these high doses (i.e., direct GDNF toxicity) rather than withdrawal from these high doses. If it was merely exposure, then the fact that lesions were seen exclusively in the subgroup of six withdrawn animals would have to be a coincidence, since all fifteen animals were equally exposed. The probability of this coincidence occurring is just

which is 0.011.

Thus, beyond reasonable scientific doubt, pathology is the result of withdrawal, and not only that, but withdrawal only from the highest doses of GDNF. It is associated neither with exposure to these high doses, nor withdrawal from lower doses.

It is worth noting that this finding is in accord with experiments that show that the p75 neurotrophin receptor induces Purkinje cell autophagy only during withdrawal from high concentrations of neurotrophins (Flores-McClure et al., J Neurosci 24[19]: 4498-4509, 2004). It is also worth noting that, because of the relatively small volume of CSF in monkeys, all doses used in this toxicology study would likely be associated with CSF concentrations far higher than in humans

Michael Hutchinson, M.D., Ph.D