Honda et al. (2019): Updated Armitage et al. (2014) Model

Gregory S. Honda

July 13, 2018

Please send questions to

from “Using the concordance of in vitro and in vivo data to evaluate extrapolation assumptionst”

Gregory S. Honda, Robert G. Pearce, Ly L. Pham, R. W. Setzer, Barbara A. Wetmore, Nisha S. Sipes, Jon Gilbert, Briana Franz, Russell S. Thomas, John F. Wambaugh

PLoS ONE 14(5): e0217564.

The authors thank James Armitage for providing a revised version of his model (


Linking in vitro bioactivity and in vivo toxicity on a dose basis enables the use of high-throughput in vitro assays as an alternative to traditional animal studies. In this study, we evaluated assumptions in the use of a high-throughput, physiologically based toxicokinetic (PBTK) model to relate in vitro bioactivity and rat in vivo toxicity data. The fraction unbound in plasma (\(f_{up}\)) and intrinsic hepatic clearance (\(Cl_{int}\)) were measured for rats (for 67 and 77 chemicals, respectively), combined with fup and \(Cl_{int}\) literature data for 97 chemicals, and incorporated in the PBTK model. Of these chemicals, 84 had corresponding in vitro ToxCast bioactivity data and in vivo toxicity data. For each possible comparison of in vitro and in vivo endpoint, the concordance between the in vivo and in vitro data was evaluated by a regression analysis. For a base set of assumptions, the PBTK results were more frequently better associated than either the results from a “random” model parameterization or direct comparison of the “untransformed” values of \(AC_{50}\) and dose (performed best in 51%, 28%, and 21% of cases, respectively). We also investigated several assumptions in the application of PBTK for in vitro-in vivo extrapolation (IVIVE), including clearance and internal dose selection. One of the better assumptions sets-restrictive clearance and comparing free in vivo venous plasma concentration with free in vitro concentration-outperformed the random and untransformed results in 71% of the in vitro-in vivo endpoint comparisons. These results demonstrate that applying PBTK improves our ability to observe the association between in vitro bioactivity and in vivo toxicity data in general. This suggests that potency values from in vitro screening should be transformed using IVIVE to build potentially better machine learning and other statistical models for predicting in vivo toxicity in humans.

HTTK Version

This vignette was created with httk v1.9. Although we attempt to maintain backward compatibility, if you encounter issues with the latest release of httk and cannot easily address the changes, historical versions of httk are available from:

Prepare for session

R package knitr generates html and PDF documents from this RMarkdown file, Each bit of code that follows is known as a “chunk”. We start by telling knitr how we want our chunks to look.

Clear the memory

It is a bad idea to let variables and other information from previous R sessions float around, so we first remove everything in the R memory.


eval = execute.vignette

If you are using the RMarkdown version of this vignette (extension, .RMD) you will be able to see that several chunks of code in this vignette have the statement “eval = execute.vignette”. The next chunk of code, by default, sets execute.vignette = FALSE. This means that the code is included (and necessary) but was not run when the vignette was built. We do this because some steps require extensive computing time and the checks on CRAN limit how long we can spend building the package. If you want this vignette to work, you must run all code, either by cutting and pasting it into R. Or, if viewing the .RMD file, you can either change execute.vignette to TRUE or press “play” (the green arrow) on each chunk in RStudio.

# Set whether or not the following chunks will be executed (run):
execute.vignette <- FALSE

Load the relevant libraries

If you get the message “Error in library(X) : there is no package called ‘X’” then you will need to install that package:

From the R command prompt:


Or, if using RStudio, look for ‘Install Packages’ under ‘Tools’ tab.


Updated Armitage 2014 Top Panel Figure 3

This produces an updated version of the top panel of Figure 3 from the paper Armitage, 2014

armitage.dt <- copy(armitage_input)
armitage.dt[,well_number:=384] %>% 
  .[,option.bottom:=TRUE] %>% 
  .[,option.plastic:=TRUE] %>% 
  .[,Tsys:=37] %>% 
  .[,Tref:=298.15] %>% 
  .[,FBSf:=0.1] %>% 

# Make sure we have data for the chemical:
armitage.dt <- subset(armitage.dt,casrn%in%get_cheminfo())
# These chemicals are missing water solubilities:
armitage.dt <- subset(armitage.dt,!(casrn%in%c("71751-41-2","34590-94-8","141517-21-7")))

armitage.dt2 <- armitage_estimate_sarea(tcdata = armitage.dt)
armitage_output1 <- armitage_eval(tcdata = armitage.dt2[,ac50:=50])
armitage_output2 <- armitage_eval(tcdata = armitage.dt2[,ac50:=1])
armitage_output3 <- armitage_eval(tcdata = armitage.dt2[,ac50:=0.001])

Plot results

armitage_output <- rbind(armitage_output1[,xfill:="50 \U00B5M"],
                         armitage_output2[,xfill:="1 \U00B5M"],
                         armitage_output3[,xfill:="1 nM"])
armitage_output[,xfill:=factor(xfill,levels=c("50 \U00B5M","1 \U00B5M","1 nM"))]

ggplot(armitage_output) +
  geom_point(aes(x=gkow,y=log10(ccells/MW*1e9),color=xfill),shape=1) +
  labs(color=expression("AC"["50"]),x="log P", y = "log(ng chemical/g cells)") +