On Tue, 30 May 1995, David Curtis wrote:
> By "complex" I meant "not exhibiting Mendelian patterns of inheritance".
> So either incomplete penetrance, or phenocopies, or both. Locus
> heterogeneity (as in early-onset Alzheimer's) doesn't really count in
> this case because you only find out about it after you've got linkage.
OK, but it should include more than incomplete penetrance and phenocopies.
Most often we are probably dealing with multiple loci. Their effects may
be additive or non-additive (epistatic). On top of this, environmental
effects either add to or interact with the genotypic effects. For many
disorders, e.g., schizophrenia, I would expect considerable etiological
heterogeneity. So very different G X E systems may be important in
different subtypes. It is easy to see why there has been little success
in finding genes for complex phenotypes.
> I think breast cancer is OK,
> because even if the BRCA1 and BRCA2 mutations are completely Mendelian (are
> they?) it's a common disease which isn't generally Mendelian and so in the
> analysis one has to cope with the situation that one is never really sure
> which individuals and which families represent true genetic cases.
A mutation (or a gene, or an allele) will always be "Mendelian." An
exception might arise when there is selection on gametes, zygotes or
embryos--i.e., segregation distortion (but this is probably not so
common). Of course, BRCA mutations are Mendelian, but breast cancer in
families segregating BRCA genes do not follow precisely a mendelian
pattern. If I recall, penetrance of BRCA1 is about 85%, not 100%, and
families segregating BRCA1 can have breast cancers in members who don't
carry BRCA1.
"Complex phenotype" is difficult to define. There is obviously a
multidimensional continuum of complexity. (Some dimensions would include
things like: extent of etiological heterogeneity, extent of genetic
heterogeneity, "penetrance" [a concept that disappears when all has been
revealed], additivity of locus effects and of genetic effects, etc.)
I believe the most important thing is to try to understand the disease
well enough to know how to select families where a stronger relation
between the disease and a single locus will be observed. In Alzheimer's
disease, diabetes and breast cancer, families were selected for study if
they included early onset cases. A little primitive as a selection
criterion, but it seems to be much better than no selection at all.
I do think that success is highly dependent on our ability to select
samples with a high prevalence of a subtype that shows a strong
single-gene effect. In effect, in "mendelian" disorders such samples were
always available without selection.
> So I was looking for nice examples to give those pessimists who doubt we
> will ever find anything using a linkage strategy.
I am not one of those pessimists. Let us all know when you find new
examples.
Mike
Michael B. Miller, M.S., Ph.D.
Department of Psychiatry (Box 8134)
Washington University School of Medicine
4940 Children's Place, St. Louis, MO 63110
office phone: (314) 362-9428 FAX: (314) 362-9420
home phone (with voice-mail): (314) 645-1863
Email: mbmiller at sirronald.wustl.edu
WWW Homepage: ftp://sirronald.wustl.edu/pub/mbmiller/mike.htm
