Biomedical Engineering Reference
In-Depth Information
11.3
UPDATING THE MOLECULAR DISEASE MODEL
For personalized genomic medicine to be successful, one must be able to cor-
relate specifi c genomic and clinical profi les with potentially effective therapies.
Unfortunately, there are so many possible genomic-level variations that every
patient could potentially represent an entirely new disease. For example, Lee
et al. [2, p. 473] reported “
50,000 high - confi dence single nucleotide variants”
and “530 somatic single nucleotide variants in [one lung cancer tumor], includ-
ing [392] in coding regions, as well as 43 large-scale structural variations”;
All
this from a single patient's tumor
! One hopes that the huge number of possible
mutations can be categorized into a handful of functional subtypes based upon
the molecular drivers of the disease (i.e., biochemical pathways), and so into
a handful of therapeutic approaches, but we cannot know at the outset, of
course, what the subtypes are and how to treat them.
Thus, to carry this program forward, it is necessary to obtain genomic infor-
mation from a wide range of individuals—indeed, potentially a huge number
of them and probably multiple samples from each patient in order to assess
intratumor variability. Moreover, in order to determine treatment effi cacy, one
needs to correlate the details of this massive body of genomic information
with information about how individuals with different genomic profi les are
treated and how they respond.
The molecular disease model summarizes some of this complexity into
actionable subtypes in terms of tumor profi les, pathways, and so on, but it is
only useful when it has something to offer and when it is correct. Determining
when the disease model is incomplete is easier than determining when it is
wrong. It is
incomplete
when no actionable treatments can be computed from
the model. This defi cit is addressed by the
personalized virtual biotech
meth-
odology described in the next section. The model is
wrong
when information
is forthcoming regarding potential treatments but when the patient fails to
respond to a treatment selected from this set (i.e., their cancer progresses).
Whether a particular patient's disease is progressing more under one treat-
ment than under another, and taking into account the near infi nitude of vari-
ables that may cause particular individuals with otherwise genomically (and
functionally) identical tumors to respond differently to treatment, is a very
diffi cult problem which we address through a
macroscale N - of - 1 adaptive trial
strategy
, described in the following section.
>
11.3.1
Personalized Virtual Biotech
When no effective treatment hypotheses can be found in the disease model,
the personalized virtual biotech methodology provides a means to effi ciently
carry out ad hoc research to discover potential novel approaches to a particu-
lar patient's disease. The idea of virtualizing therapy discovery is not new; large
pharmaceutical companies often operate projects that are nearly completely
Search WWH ::
Custom Search