National Public Health Institute, Finland
Epi/tox view on Benchmark dose by
Jeff Swartout
www.ktl.fi
Jouni Tuomisto
National Public Health Institute (KTL)
Kuopio, Finland
National Public Health Institute, Finland
What is the research question?
www.ktl.fi
• It should pass the clairvoyant test
• What is the probability of having a response R in
an individual of a large population Pop at time t
after an exposure pattern Exp?
–
–
–
–
–
R: What is the outcome measured?
R: what is the magnitude for a positive response?
What is the target population?
What is the time of observation?
What is the exposure pattern?
National Public Health Institute, Finland
Probability of response R
0.8
P(Disease)
0.7
0.6
Response
0.5
E(R | Exp)
0.4
0.3
0.2
www.ktl.fi
0.1
0
0
50
Exposure (mg/kg/day)
100
National Public Health Institute, Finland
Benchmark dose for R
0.8
Response
0.7
0.6
www.ktl.fi
P(Disease)
E(R | Exp)
0.5
BMR: 0.1
0.4
0.3
0.2
P(Exp | R=Bg+(1Bg)*BMR)
0.1
BMDL
0
0
50
Exposure (mg/kg/day)
100
National Public Health Institute, Finland
Critical assumptions
www.ktl.fi
• The function used reflects reality
• The number of parameters used allows for enough
flexibility
• The values of parameters >= 0
ÆThere is no toxicological or epidemiological
support for exactly these assumptions
ÆAssumptions are more based on conventions than
knowledge about reality
National Public Health Institute, Finland
Frambozadrine
0.9
0.8
E(R | Exp)
P(hyperceratosis)
0.7
0.6
BMR: 0.1
0.5
Male rats
0.4
0.3
Female rats
0.2
www.ktl.fi
0.1
BMDL (pooled data)
0
0
50
100
Frambozadrine exposure (mg/kg/day)
National Public Health Institute, Finland
Questions to be studied (frambozadrine)
• Do we need separate DR relationships for males
and females?
– Study: We do not need separate DRs
– Jouni: If the target population Pop is "rats", then we
MUST pool
www.ktl.fi
• This is NOT a question of statistical deviance
National Public Health Institute, Finland
Parsimonate: mice data sets
0.9
0.8
Tumor incidence
0.7
0.6
B6C3F1
0.5
Crj:BDF1
Crj:BDF1 BMDL-BMD
B6C3F1 BMDL-BMD
0.4
0.3
0.2
www.ktl.fi
0.1
0
0
10
20
30
Metabolized dose (mg/kg/day)
40
50
National Public Health Institute, Finland
Conclusions
www.ktl.fi
• We should use statistical methods that reflect our
understanding of physiological reality
• We should be very concerned if conventions or
statistical limitations deviate us from this goal