Sparse NCA Calculations
Sparse noncompartmental analysis (NCA) is performed when multiple individuals contribute to a single concentration-time profile due to the fact that there are only one or a subset of the full profile samples taken per animal. A typical example is when three mice have PK drawn per time point, but no animals have more than one sample drawn. Another typical example is when animals may have two or three samples during an interval, but no animal has the full profile.
Sparse NCA Setup
Sparse NCA is setup similarly to how normal, dense PK sampling is
setup with PKNCA. The only difference are that you give the
sparse option to PKNCAconc(), and in your
interval calculations, you will request the sparse variants of the
parameters. As of the writing of this vignette, the only sparse
parameter for calculation is sparse_auclast. Any of the
non-sparse parameters will be calculated based on the mean profile of
the animals in a group.
The example below uses data extracted from Holder D. J., Hsuan F., Dixit R. and Soper K. (1999). A method for estimating and testing area under the curve in serial sacrifice, batch, and complete data designs. Journal of Biopharmaceutical Statistics, 9(3):451-464.
# Setup the data
d_sparse <-
data.frame(
id = c(1L, 2L, 3L, 1L, 2L, 3L, 1L, 2L, 3L, 4L, 5L, 6L, 4L, 5L, 6L, 7L, 8L, 9L, 7L, 8L, 9L),
conc = c(0, 0, 0, 1.75, 2.2, 1.58, 4.63, 2.99, 1.52, 3.03, 1.98, 2.22, 3.34, 1.3, 1.22, 3.54, 2.84, 2.55, 0.3, 0.0421, 0.231),
time = c(0, 0, 0, 1, 1, 1, 6, 6, 6, 2, 2, 2, 10, 10, 10, 4, 4, 4, 24, 24, 24),
dose = c(100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100)
)Look at your data. (This is not technically a required step, but it’s good practice.)
library(ggplot2)
ggplot(d_sparse, aes(x=time, y=conc, group=id)) +
geom_point() +
geom_line() +
scale_x_continuous(breaks=seq(0, 24, by=6))
Data Setup Note
Sparse NCA requires that subject numbers (or animal numbers) are given, even if each subject only contributes a single sample. The reason for this requirement is that which subject contributes to which time point changes the standard error calculation. If all individuals contribute a single sample, a simple way to handle this is by setting a column with sequential numbers and giving that as the subject identifier: