Page 327 - 20dynamics of cancer

312 CHAPTER 15

Knudson’s approach was simple: predict how a perturbation to pro-
cess alters outcome. Knudson was particularly successful because he
chose to focus on perturbation to heritable properties of cells, in this
case, perturbation caused by an inherited mutation, and because he
chose to focus on quantitative aspects of the ultimate outcome, the rate
at which cancer occurs at different ages.
Current laboratory studies use the same approach. Those lab studies
analyze genetic causation by comparing the age-onset curves between
different genotypes. If a particular genotype shifts the onset curve to
an earlier age, then one ascribes causation to the genetic differences of
that genotype relative to the matched control. Those lab studies almost
always compare incidence curves in a qualitative way, by simply noting
if the incidence curve of a particular subgroup of animals has shifted to
an earlier age relative to matched controls. They discard all the quan-
titative information about outcome contained in the relative rates of
progression for different groups.
Throughout this book, I have advocated quantitative comparisons of
incidence curves to infer causation. I developed an extensive theoretical
framework from which one can predict how genetic or environmental
perturbations alter incidence curves. Such comparative predictions can
be tested easily in studies of laboratory animals, where the experimenter
can control conditions and treatments for different groups.
I have also advocated comparisons of incidence between subgroups
of humans. Such comparisons provide particularly interesting infor-
mation when the subgroups differ in clearly identified aspects of their
genetics. Knudson’s comparison of inherited and noninherited retino-
blastoma provides one example. In that case, identifying the distinct
subgroups is relatively easy, because the inherited cases have distinc-
tive patterns of tumor formation when compared to the noninherited
cases.
New genomic technologies will soon allow much more refined mea-
surement of genotype in human subjects. With that genetic resolution,
one will be able to compare quantitative aspects of age-incidence curves
between groups with and without certain genetic attributes. Such com-
parisons will allow one to ascribe causation to particular genetic dif-
ferences, and then follow up with analysis of the biochemical processes
associated with those genetic differences. This approach demands quan-
titative evaluation of outcome—the age-incidence curve. I have built the