A G Smith

1
Human risk assessment of dioxins and
PCBs uncertainties and mechanistic complexities

• A G Smith
• MRC Toxicology Unit
• Leicester University

2
Environmental polyhalogenated dioxins,furans and
dioxin-like PCBs
Dioxins
Furans
Biphenyls
Human exposure episodes noted from 1940s First
shown experimentally as toxic in the 1970s
3
Dioxins etc continue to be of public concern

• In humans exposure is of 3 types
• Accidental high exposures (eg Seveso)
• Exposed workers in pesticide plants etc
• Exposure in food (main UK source)

Experimental toxicity massively studied now
there is so much information it can be difficult
to see the wood for the trees
4
Dioxin exposure in humans
COT

• Chloracne- most proven association but no dose
  response relationship
• Hepatic toxicity- increase in liver enzymes in
  plasma
• Cardiovascular diseases- in occupational exposure
• Diabetes- no consistent findings
• Reproductive- high exposure, changes in sex
  ratio endometriosis variable
• Neurological/psychological- possible but variable
• Neurobehavioural- possible
• Immunological- no consistent findings
• Cancer- regarded as probable human carcinogen but
  conclusion based on overall picture from
  experimental and human evidence-still contentious
  

5
Chloracne is the foremost proven consequence of
dioxin for humans occurs at acute high exposure
(favoured by some Ukranians)
Can occur through oral exposure Rodriguez-Pichardo
et al 1991
Most species are resistant Rabbits and some mice
can be made to respond
6

• May persist for many years, despite fall in
  dioxins,
• and not just chloracne
• but also other effects such as liver damage

7
Dioxin toxicity in animals
COT

• Lethality- LD50s vary from 1-5000 µg/kg
• Wasting syndrome- but also oedema
• Hepatic toxicity- includes porphyria
• Immunotoxicity- reduction in humoral response and
  splenic and thymus atrophy
• Thyroid hypertrophy
• Reproductive- deceased fertility, decreased sperm
  counts, effects on testes and prostate,
  teratogenicity
• Genotoxicity- mostly negative does not bind to
  DNA nonmutagenic
• Cancinogen- Thyroid and liver tumours, promoter
  of skin and liver cancer

8
Toxicity in mice

• Mouse strains were known to respond markedly
  differently to polycyclic aromatic hydrocarbons
  such as 3-methylcholanthrene, benzoapyrene and
  dibenzahanthracene biochemically and in cancer.
• This was postulated to be mediated by a receptor
  the Aromatic (Aryl) Hydrocarbon Receptor.
• Soon observed that 2,3,7,8-tetrachlorodibenzo-p-di
  oxin (TCDD) was the most high affinity ligand for
  this protein and overcame resistance in mice that
  did not respond to 3-MC etc.
• The AH receptor is a ligand activated
  transcription family of the PAS family. Occurs
  in most species (including rats and humans) but
  in some does not bind ligands.
• Subsequent studies demonstrate that most
  toxicities are mediated by AHR. Polymorphisms
  can affect efficiency of binding and thus
  susceptibility.

9
A key discovery was that dioxin (TCDD) was the
most powerful ligand for the Aryl Hydrocarbon
Receptor (first identified with chemicals such
as benzoapyrene and 3-MC)
How the myriad of toxicities occurs is far from
clear
10
Risk assessment of dioxins and dioxin-like PCBs
based on

• Relative affinities for AHR and dioxin-like
  action
• Human exposures (some controversial)
• Animal studies

11
Quantitation of AHR-mediated toxicities

• Experimentally, most aspects of dioxin toxicity
  and biochemical responses have been shown to
  depend on AHR using resistant and knockout mice
• Capabilities of dioxins, dibenzofurans and
  dioxin-like PCBs to bind to the AHR generally
  show structure-activity relationships similar to
  elicitation of biochemical and toxic responses
• Using TCDD as the gold standard this allows
  Toxic Equivalence Factors (TEFs) to be assigned
  to different congeners
• In turn this allows TEQs to be calculated for
  complex mixtures

12
Examples of TEFs

• 2,3,7,8-TCDD
• 1,2,37,8-PeCDD
• OCDD
• 2,3,7,8-TCDF
• PCB 77 (3,4,3,4)
• PCB 126 (3,4,5,3,5)
• PCB 169 (3,4,5,3,4,5)

• 1
• 0.1
• 0.0001
• 0.1
• 0.0001
• 0.1
• 0.01

Sum of these allows estimation of TEQs
These get adjusted with more data
13
Toxicodynamics and toxicokinetics

• These chlorinated chemicals vary considerably
• in their tissue distribution and absorption
• metabolism and elimination
• Half-lives in mice and rats may only be weeks but
  TCDD 7.5 years in man. OCDD is 120 years in man.
• In terms of risk assessment, these have to be
  taken into account.

14
COT assessment

• To develop a risk assessment from the known
  hazard information, effects in animals or humans
  have to be selected that are relevant to humans
  at low doses.
• Is a threshold likely?
• With dioxins it was considered that most of the
  toxicity was mediated by the AHR i.e. a threshold
  was probably involved, and thus a NOEL
  uncertainty factor approach was suitable.
• Find the lowest dose that gives a NOEL or if not
  a LOEL has to be identified.
• What is the criteria for exposure/dose level?

15
What to choose for human study to provide a
NOEL/LOEL?

• It was concluded that the available human data
  was not sufficiently rigorous for establishment
  of a tolerable daily intake.
• Epidemiological did not reflect the most
  sensitive population seen in animal studies.
• Too many confounding factors in exposure
  assessments to be sure due to dioxins.
• Importantly, exposure of humans did not reflect
  UK situation where most likely from food.

http//www.foodstandards.gov.uk/committees/cot/sum
mary.htm
16
Animal studies

• Because COC had decided that the mechanisms of
  cancer (whatever these may be) were threshold
  based, all toxicological endpoints could be
  examined to find the most sensitive for TCDD
  toxicity and this would also cover increased
  cancer risk.
• In fact, very few studies could found that were
  in accord with modern criteria for identifying
  NOEL/LOEL. Of course many others were extremely
  good science for mechanistic and susceptibility
  interpretations but not suitable here.
• The most sensitive endpoints appeared to be on
  the developing reproductive systems of male rat
  fetuses exposed in utero.
• Despite inconsistencies between studies on some
  endpoints, it was considered that effects on
  sperm production and morphology represented the
  most sensitive effects that could be used for
  deriving a Tolerable Daily Intake.
• Sperm reserve in men is much less than the rat
  and thus may be highly relevant to humans.

17
What studies to use for TDI?

• 3 studies on sperm quality were available but
  none perfect
• A variety of exposure routes and endpoints and
  what dioxin levels present in tissues.

18
Use of body burden

• Rodents require higher doses (100-200-fold) to
  reach the same equivalent body burdens as in
  humans on exposure to food etc (differences in
  toxicokinetics etc).
• A consensus view is that body burden is the more
  appropriate parameter for comparison between
  species.
• The data of Hurst et al (2000) has given the
  distribution of TCDD in maternal and fetal tissue
  on Gestation Day 16 after single dose on D15 and
  chronic dosing before mating. This allows
  toxicodynamic and toxicokinetic estimates of
  maternal v fetal levels depending on dose route
  and timing.
• Using the two lowest doses a ratio of 2.5 was
  calculated to be used in estimates of body burden
  from single dose and subchronic exposure in
  other dosing studies.

19
Stolen from JC Larsen and Andy Renwick!
20
Calculation of TDI

• The study of Faqi et al (1998) was chosen as the
  most suitable for estimation of TDI although some
  problems and no NOEL but the lowest LOEL i.e. a
  loading dose then a maintenance dosing regime.
• Using the previous factors the subcutaneous
  dosing dosage regimen of Faqi et al was converted
  to a steady state maternal burden on GD16 at the
  LOEL.
• This was estimated as 33 ng/kg bw.

21

• With the study of Faqi et al as the most suitable
  available for TDI estimation. An uncertainty
  factor of 1 was used for interspecies differences
  in toxicokinetics because of using body burdens
  not dose.
• An uncertainty factor of 1 was used for
  interspecies differences and human variability in
  toxicodynamics on the basis that rats may be more
  sensitive than humans but the most sensitive
  humans may be as sensitive as rats.
• An uncertainty factor of 3.2 for human
  variability in toxicokinetics to allow for
  increased accumulation in the most susceptible
  individuals (for dioxins with ½ lives less than
  TCDD).
• An uncertainty factor of 3 to allow for use of
  LOEL rather than NOEL
• Thus total of 9.6 (3 x 3.2 x 1 x 1) uncertainty
  factor

22

• Using the overall uncertainty factor of 9.6 and
  the calculated maternal steady-state body burden
  from the study of Faqi et al (LOEL 33 ng/kg/bw)
  gives a tolerable human equivalent maternal body
  burden of 3.4 ng/kg/bw.
• Putting this into daily intake (pg/kg/day)
• body
  burden(pg/kg bw) x ln2
• 
  bioavailability x ½ life in days
• 3400 x
  0.693
• 0.5 x
  2740 (7.5years)
• 1.7
  pg/kg/day

23

• This TDI was rounded to 2 pg WHO TEQ/kg bw per
  day
• based on developing male reproductive system
  and maternal body burden
• WHO (1998) 1-4 pg WHO TEQ/kg bw per day
• SCF 14 pg WHO TEQ/kg bw per week
• JECFA 70 pg WHO TEQ/kg bw per month
• COT considered is adequate to protect against
  cancer and cardiovascular effects.
• UK consumer levels are falling but TDI is
  near the the value for the average consumer and
  lower than 97.5 percentile

24
Is this it? End of story

• There actually many uncertainties
• Dose-additivity is fundamental to the TEF idea
  and a reasonable idea however we are still far
  from sure that this applies in the complex
  mixtures pertinent to human exposures. Dose
  reponses.
• TEFs are mostly derived from animal data, are
  they appropriate for humans? e.g. carcinogenicity

25
The AH receptor

• There is a lot we do not understand about the AH
  receptor. Ligand binding activities may not be
  all. Human and rat polymorphisms may affect
  action as a transcription factor in ways we do
  not understand yet.
• Other PAS factors such as HIF and ARNT may have
  crucial roles in mediating toxicity.
• What about interaction between ligands?
• Dependence on AHR is not the same as
  susceptibility

26
Most sensitive individuals

• 7. What does this mean. Experimental studies are
  starting to show modifier genes of Ahr. Similar
  effects in most human diseases thought to be
  monogenic.

27
Chloracne occurs at acute high exposure
Some evidence for genetic susceptibility
Hairless mice are sensitive Interaction between
Hr gene and Ahr
28
What are the critical events leading to toxicity?
8. We need to pursue fundamental mechanisms to
really understand the risks. What is dependent on
a threshold? What influence does CYP1A2 hepatic
expression have?
29
Other AHR ligands

• What about non chlorinated ligands such as PAHs
  and natural ligands. Should we not ignore
  them?
• What about chemicals such as hexachlorobenzene
  and brominated diphenyl ethers? All in
  environment. Many toxic actions are similar to
  dioxin-like PCBs.
• Consideration of these aspects may cause TDI to
  be exceeded but ?

30
Thanks

• To Diane Benford and ex colleagues on COT