Transcript Slide 1

#16 – Flexible D-R Model for MOA Including
Multiple Pathways - Acrylamide
• Goal: How to model low-dose cancer risk if two
MOAs are operating in different dose ranges
• Method: Evaluate MOAs using modified Hill criteria;
choose models based on MOAs
– Focus is thyroid tumors, considered relevant to humans
– Genotoxicity (mainly low-dose region) and growth
stimulation from thyroid hormone (higher dose)
– Fit d-r model that captures the MOAs
• Demonstration: not a Gold Standard
• Team: Hertzberg, Dourson, Allen, Vincent, Haber
Method for D-R Modeling
of Multiple MOAs
• Evaluate MOAs using modified Hill criteria
– Increase in key processes -> increase in tumors
– Focus is thyroid tumors, considered relevant to humans
– Genotoxicity (mainly low-dose region) and growth
stimulation from thyroid hormone (higher dose)
• Choose models based on MOAs
– Low dose genotoxicity
• Low dose data: shallow slope, linear, no threshold
– Higher dose growth stimulation
• Threshold, higher slope
Input
Acrylamide/Glycidamide PBPK/PD
PBPK-1
Acrylamide
(AA)
PBPK-2
Glycidamide
(GA)
metabolism
urine
1st order
metabolism
Input
urine
Hemoglobin
Adducts
DNA
Adducts
1st order
metabolism
repair
turn over
PD
PBPK-3
Acrylamide metabolite
(AA-GS)
PBPK-4
Glycidamide metabolites
(GA-GS2 + GA-GS3)
urine
urine
3
Modeling Options
• Multiple models
– Best if have MOA-specific, independent data sets
• Single model with MOA threshold
– Statistically similar to multiple models, but simultaneous
model fitting with joined models
– Estimation of break point requires more data
• Single model with desired slope properties
– Flat and no threshold at lowest dose
– Higher slope at upper doses
Example Models
19 Dose groups but only
7 distinct doses to
suggest curvature
Results
• Low dose region
– health-protective, linear cancer slope factor (SF) of
0.030 (mg/kg-day)-1
– also applies to higher doses
• Higher dose region
– Reference Dose (RfD) in the range of 0.05 to 0.02
mg/kg-day
Remaining Issues
• Risk Communication?
– What does dose>RfD (e.g., =0.06) imply re cancer risk?
• What experimental design is the minimum for direct
estimation of the transition dose?
– Need change in curvature, not just in response
• Can nonlinear mutagenicity be biologically modeled?
Multistage model fitted to pooled-all thyroid tumor
data, showing little change in slope between the low
and high dose regions.
Probit model fitted to pooled-all thyroid tumor data, showing
differing slopes between the low and high dose regions.