A toxicologist looks at Developmental Neurotoxicity Testing

1
A toxicologist looks at Developmental
Neurotoxicity Testing - costs and
benefits
David Ray, (an experimental neurotoxicologist) Uni
versity of Nottingham Medical School, Queens
Medical Centre, Nottingham, United Kingdom
2
There are many sorts of testing

• To be sure that there is no adverse effect
• But where do we search, and when do we stop
  looking?
• To characterise an identified adverse effect
• Relatively easy once it has been found
• Nature of the effect may change as dose is
  reduced
• To establish a safe exposure level (NOAEL)
• Harder for the more variable end points
• Need to be certain that end point is the most
  sensitive
• Always expensive, as requires many groups
• To understand the mechanism of toxicity
• Requires individually tailored test systems
• Helps greatly with extrapolation to man

3
How does developmental neurotoxicity differ from
ordinary neurotoxicity?

• Adult neurotoxicity
• c. 2,000 known chemicals, and as many types of
  target cell
• most chemicals target a sub-set of less than 10
  of these
• cell types, and most chemicals target
  different sub-sets!
• Developmental toxicity is even worse, as new
  targets are
• seen that are not present in adults (e.g.
  migration)
• The functional consequences are however similar
  to those of adult neurotoxicants, although they
  may show up only later in life.
• Adverse effects may be subtle
• Normally we solve this by increasing dose until
  a robust
• effect is seen, but maternal toxicity may
  limit scope
• Pathology more difficult as usually no cellular
  reaction

4
What sort of test system should we use?

• One that detects all possible adverse effects
  that might be seen in man
• The only way to do this is to expose a large
  human
• population.
• If we have learned one thing from lead, methyl
• mercury, and valproate it is that we really must
  stop doing this!
• All non-human test systems are a compromise
• so any false negatives put humans at risk
• even a good system that is too costly to apply
  to
• all candidate chemicals also puts humans at
  risk

5
An example of a problem with test sensitivity
Schizophrenia in humans usually develops only in
the late teens or early twenties. The mechanism
is not understood, but animal models are of some
value. Would a schizophrenia-like condition
caused by developmental exposure to chemicals be
detectable by the current animal protocols?
The US developmental neurotoxicity testing
protocol requires spontaneous motor activity
monitoring on postnatal days 13, 17, 21, and 60
(2) plus tests of auditory startle habituation
and associative learning around days 21 and day
60. European requirements are similar.
So, are 21 and 60 days old late enough to show up
any schizophrenia-like effects in rats?
6
1) Effect of 7 day ventral hippocampal lesions on
the locomotor response of rats to challenge doses
of MK-801
at 65 days
at 150 days old
Challenge at 35 days



mg/kg
mg/kg
mg/kg
? sham lesion at 7 days ? lesion at 7 days
Data of Al-Amin,H.A. et al. (2001) Biol
Psychiat.. 49528-539.
i.e. the neonatal damage is not detectable at 35
days, and not fully expressed even at 65 days
old.
7




Data of Bhardwaj et al. (2004) Neuropharmacol.
4685-94.
i.e. damage did not change receptor density at 35
days, but did at 56 days of age.
8
So, caution!
The lesion-based schizophrenia models suggest
that the final 60 day DNT measure may not be late
enough to be sensitive to delayed effects.
So we may come rather close to finishing our
standard DNT measures too early, but what about
the sensitivity of our tests?
Pharmaceutical companies evaluate
anti-schizophrenic drugs in rats with tests
specifically designed to be difficult, and so
sensitive to subtle changes. Such tests require
far too much time to set up and to evaluate.
9
An example of a small effect
Maternal haloperidol at postnatal days 3-6
Spontaneous activity of NMRI mice was
significantly decreased at postnatal day 38 after
4 daily doses of 0.5 mg/kg to the dam.
10
that can be clearer with a more focused test
When the same mice were challenged with
apomorphine at day 60 there was a clearer
increase in the stereotypy score (others have
shown similar effects at 0.5 mg/kg x 20 days or
1.5 mg/kg x 4 days)
Thus when the mechanism is known, it is not hard
to find a test that will give a strong positive
result. However, this does not help much when the
mechanism is not known!
11

• So what is the way forward?
• For screening
• I have heard of no alternative screening system
  that could hope to answer open-ended questions
  such as is this substance likely to be without
  any sort of adverse action on the developing
  mammalian brain?
• I know of no human developmental neurotoxicant
  would fail to have been flagged up as a potential
  problem by a simple screen for adult rat
  neurotoxicity at higher dose levels.
• I also think that there are none that would not
  be detected (at some dose) by current DNT
  protocols.

12
But theory says we should be careful about using
adult neurotoxicity as a pre-screen

• The developing brain has target processes not
  seen in the adult (e.g. neuronal migration, glial
  guidance)
• Hence there is the possibility that an agent may
  be a developmental neurotoxicant yet have no
  adverse effect on the adult brain.
• However most neurotoxicants have multiple
  mechanisms of action, and none have yet been
  found that act exclusively on the developing
  brain so far!
• It may well be that alternative systems could
  provide a more secure trigger for in vivo DNT
  testing though.

13
An unexpected surprise with chlorpyrifos

• Studies by Slotkin at al. have shown that, as
  well as inhibiting acetylcholinesterase,
  chlorpyrifos reduces cell division (DNA
  synthesis) in the brain
• The effect occurs both in cholinergic and
  non-cholinergic cells
• It requires doses that (perhaps coincidentally)
  inhibit acetylcholinesterase to a significant but
  sub-symptomatic extent, but the effect was
  discovered independently
• At 1 mg/kg/d over post-natal days 1-4 in rats,
  there were marked changes in serotonergic
  synaptic markers even out at 5 months old.
• Females and males showed very different effects

14

Slotkin Seidler (2005) Dev. Brain Res.
158115-119
15

• However
• These findings did not alter the safe threshold
  of exposure for chlorpyrifos, and so far DNT
  studies have led to only 3 (unspecified)
  reductions in regulatory safe exposure limits
• Hence the pragmatic response of many overseas
  academic toxicologists is that the present system
  works well without DNT, and that the EPA scheme
  is an over-elaborate waste of money. This is
  unreasonable - but not without some truth!
• Retrospective evaluation by ILSI of known human
  developmental neurotoxicants suggests that
  (academic) rodent DNT tests would have identified
  them, although not even one chemical causing DNT
  in man has been evaluated using the full US EPA
  regulatory protocols
• and the EPAs animal tests are slow and
  expensive.

16

• A way forward for better evaluation of identified
  developmental neurotoxicants
• Current DNT testing uses catch-all methods
  that say little about what the mechanism of the
  problem is or whether it is relevant to man, and
  so do not inform regulatory decisions.
• Defined in vitro or other assays are the best
  way of answering questions about mechanisms. Thus
  new valproate analogues would be far better
  screened by an assay for histone deacetylases
  Gurvich et al., (2005) FASEB J 199 1166-1168
  than by conventional animal DNT studies.
• Such studies are quick, humane, efficient, and
  exploit the best modern neuroscience. Although
  too specific to enable us to say that anything
  passing them is safe, they are ideal for
  informing extrapolation of ambiguous conventional
  test results to man and for prioritising
  conventional testing.

17

• Problems with the current screening tests
• False positives can be generated by the multiple
  measures needed to cover the many sorts of
  adverse effects that can be produced, and animals
  are wasted in lower dose groups if the higher
  dose is negative.
• It can be hard to distinguish primary
  developmental neurotoxicity from secondary
  effects of threshold-level maternal toxicity
  because
• The brain is reasonably well protected from
  malnutrition in utero, but not from disturbance
  of placental blood flow or poor nursing
• Maternal stress in rats leads to lasting changes
  in brain-derived neurotrophic factor in the next
  generation

18
The effects of maternal stress
Pregnant rats were restrained for 45 minutes a
day from E14 to delivery. Brain-Derived
Neurotrophic Factor message in the progeny was
measured at 90 days old
Fumagalli et al. (2004) Eur. J. Neurosci.
201348-1354
19
We could improve follow ups to standard
developmental neurotoxicology studies by

• Using a simpler initial testing protocol with
  fewer dose groups, and only going on to determine
  a NOAEL if a (possibly false) positive is seen at
  a high dose
• Confirming positives and the NOAEL with
  individually tailored and interpretable end
  points, e.g. pharmacological challenge, as
  suggested in vitro determined mechanisms
• Using better methods for gross morphology, such
  as MRI
• Better differentiating developmental delay from
  irreversible effects

20
Attitudes of academics versus regulators

• Academic
• Emphasis on novelty
• new agent
• new hazard
• new method
• Little toxicokinetics
• Postnatal / adolescent
• Sex differences are interesting
• Emphasis on continued funding!

• Regulator (screener)
• Emphasis on practicality or consistency
• Applicable to multiple agents
• Validated outcomes
• Ideally the same everywhere
• Emphasis on toxicokinetics
• Sex differences are confounders
• Focus on prenatal
• Fixed endpoints

Slightly adapted from Slotkin (2004)
Neurotoxicol. 25631-640
21

• The human cost of DNT can be smaller than that of
  other developmental factors
• In a WHO study of a smelter town in Yugoslavia,
  lead exposure and social factors were measured
• Together these explained 41-44 of the variance
  in overall IQ. The 7-year lead exposure explained
  2.8-4.2 of the variance in IQ, and a change in
  lifetime blood lead from 10 to 30 micro/dl
  corresponded to a decrease of 4.3 IQ points.
  However, the IQ/lead association was small
  relative to that seen with socio-economic
  deprivation.
• So DNT is not the only thing that we need to take
  action on!

Wasserman et al.(1997) Environ. Health Persp.
105 956-62.
22
Conclusions 1/2 (alternative tests)

• DNT is a multimodal process, so tests that
  evaluate only a few mechanisms cannot in
  themselves avoid false negatives, or the
  potential for false reassurance that will harm
  human health
• Cell-based tests have very limited applicability
  to the critical later stages of nervous system
  development (where target cell innervation and
  neuronal function become important)
• although intact lower vertebrates should prove
  very useful for this, once their tests are
  validated
• Any useful screening system must have
  bioactivation capacity
• Alternative tests often give better mechanistic
  information
• Any mechanism-based test that gives a positive
  result below the level that is adverse (e.g. the
  NTE test - with no neuropathy below 70
  inhibition) can be challenging for regulators!

23
Conclusions 2/2 (animal tests)

• Regulatory animal tests cannot reliably detect
  subtle effects although using higher doses can
  compensate to some extent
• Only components of the EPA tests have been
  validated against known human developmental
  neurotoxicants, and the full tests involve too
  many dose groups
• Triggering DNT testing only if there is potential
  for adult neurotoxicity is a pragmatic approach,
  but would be more reassuring if it was also
  triggered by in vitro DNT positives
• For chemicals where little basic toxicology is
  available, even a fully in vitro test for DNT
  would not avoid animal testing for other forms of
  toxicity, so it is better to refine and integrate
  than to replace
• Academics need to understand that regulators are
  not backward - but face different, often real
  world, problems.

24
Disclosures slide Personal interests Consultanc
y ZBL Bioplasma AG Non-personal
interests Funding for pyrethroid research
Bayer, DuPont, Zeneca My thanks to our kind
hosts to Dr. Bala Muhammad for his hard work to
the members of the ILSI developmental
neurotoxicology panel for trying to educate me
about DNT and to the MRC for financial support.