Well, thank you all for the opportunity to tell you more about my studies.
I guess at this point, I can't really say "introduce them," because we've
discussed quite a bit yesterday. But hopefully I can contribute a
little bit more. Is there a pointer here?
So, where I'd like to start out, is I think that a lot of people, when
they talk about genetics, they think of a gene for alcoholism or a gene
for some disease, and I want to start out by presenting how geneticists
who study complex diseases, ones where there isn't a one-to-one correspondence
with a gene, envision the problem. And in my first slide here --
I hope everybody will bear with the slide and the detail that's involved
here -- but what we envision going on is, what we call the phenotype, which
would just be a characteristic, like you're alcoholic or you're not, and
some people would be affected by this condition and other people would
not be affected by this condition. And psychiatrists and qualified
clinical people would divide the people according to set criteria.
But that's very, sort of, crude classification, and underlying that crude
classification, you'd have something called liability. And this would
reflect a lot of different contributions from genes, maybe at one locus,
maybe at many loci, and environmental factors. And when I talk about
environment, I'm not talking about, like the type of environment that ecologists
look at how many days a year there was snow or something like that, but
all kinds of things: where you went to school; what hospital you went to;
how much money you make; whether you're a male or a female, and the like.
And regarding this liability, we envision that there's a distribution
in the population, that people differ in terms of their liability (and
by that I mean their risk for developing alcoholism in this case, or the
disease), and that if we look at all people in the population, we might
get something like this upper blue line here, and that some portion of
the population would be considered affected, and it would be all those
people with liability that was greater than 1, in this case, if you had
a 2, 3 or 4, something of the like. Now, there are a lot of factors
that would contribute to it, and genes would be some of the possible types
of factors. And so we would have a genetic locus, and it would have
two alleles, for this simple example, a capital disease allele and lower
case, what would be not-disease allele. And that there would be different
genotypes represented down here along the horizontal axis of the graph,
and the mean liability for people with different genotypes is different
-- and so I'm just drawing this out so you can see the differences, but
in the one case, the red line would go for the disease homozygotes, and
they have a higher mean than the heterozygotes, who have a higher mean
than the homozygotes for the non-disease-rendering allele. But when
we look at this threshold here, by which we decide who's affected and not,
we can see that no matter what your genotype is, there would be some chance
that you'd be affected and some chance that you wouldn't be affected.
And the people with the non-disease-rendering genotype would be quite a
bit less likely to be affected; people who are heterozygotes would be more
likely, and people who are the other homozygote would be even more likely
to be affected. But the point is, if I look at who's affected and
who's not affected, I can't tell you what their genotype is. And
that's the way we envision complex diseases. And what we want to
know about is what contributes to liability? Which of the genes are
there? Here I've drawn out one. So what are the differences,
say, in people in this yellow line in liability? Well, it might be
what other genes they have. Or it might be how far they were able
to go in school. Or it might be how old they were when somebody set
up a crack lab in their neighborhood, or any number of different kinds
of things. But the point is, just because you have one gene or genotype
or another, there's a lot of variability.
So with that kind of background and conceptualization presented here,
I'd like to tell you a little bit now what I call affected and why I do
it. Well, I'm interested in alcoholism. Well, alcoholism is a vague
term, because there are a lot of different definitions of alcoholism, and
sometimes when we talk in sort of a vernacular sense, we might say, "Oh,
I think my father's alcoholic because he drinks on weekends," or something
like that. Or you might even have problems. But that's still
different from other people's definitions of alcoholism, and what I've
gone with, since I'm a geneticist by training, and can't do everything,
is called the DSMIIIR Alcohol Dependence Criteria. And the DSM is
the Diagnostic and Statistical Manual of Mental Disorders. As you
would imagine, it's out in several editions, and some of them are revised.
[laughter] At the time this study was set up, this was the current
version. There's a DSM IV that's out now; it's not terribly different,
especially with respect to alcoholism. It covers all kinds of disorders,
like depression, anxiety, phobias, schizophrenia, you name it.
But I've also, there are different DSM diagnoses. And one's called
alcohol abuse, and that sort of describes a recurrent maladaptive use of
it, but you could be diagnosed with alcohol abuse if you had several DUIs
on your record. Now, that's a softer diagnosis, and I think it's
one that's likely to maybe have cultural specificity, because police may
pull over people of different races at different times. But if you
live in the state of California, and you get pulled over for a DUI, you'll
be sent to alcohol treatment. And so it's, these types of things
aren't independent. Alcohol dependence reflects an inability to stop.
And so what we're talking about is addiction, talking about people who
can't stop drinking, we're talking about people -- if drinking isn't a
disease, these people do have alcohol-induced psychosis; they're at high
risk for liver diseases, such as cirrhosis of the liver, it's severe.
Now, the way they make the diagnosis is you have a sort of a menu of
about nine types of conditions, and you need three or more of them.
They have to last for at least a month, or repeatedly over a longer period
of time. And they include things like the person takes more, longer
than they intended. So we have to talk to people and find this sort
of thing out, because it's not how much you drink. If I intend to
drink two drinks, but I find that I end up drinking four drinks, I would
fulfill that criterion. Whereas if somebody else in the room intended
to drink four drinks and drank four drinks, they wouldn't. So there's
not a direct -- when we look at the alcohol dependence, there's not a direct
relationship with how much you drink, and that's an important thing.
It involves desire to control use, how much time you spend obtaining alcohol,
using it, or recovering from its effects; whether you're undergoing intoxication
or withdrawal during major obligations like you show up at work drunk or
with a hangover or worse. Do you give up activities -- so there
is a social function, there's a dance on Friday night. I didn't go
because I got too drunk before it started. Do you keep using alcohol?
You've been to a marriage counselor, you know your wife's unhappy, or your
husband's unhappy, with your drinking, but you keep doing it anyway.
Or your doctor tells you that your liver's in bad shape, and that if you
keep drinking you're going to die, yet you still do it. Tolerance:
does it take more to get the same effect, or are you getting less effect
when you drink? Withdrawal symptoms: they're defined in terms of
the DSM in some detail, and then the question is, do I take, do I drink
so I don't go into withdrawal? So a lot of these people ... what
we're talking about is addiction. I just wanted to establish that.
Now, how do we -- who do we study? It was a Southwestern tribe:
I think we've talked about some of the inadequacies of hiding exact tribal
names, but it was the approach that we chose. I just want to point
out that this is a population -- all the people speak the same language;
they have a traditional culture; they trace their biological heritage,
their ancestors, mothers, fathers, grandparents, back a long time.
They live on three nearby reservations; they were pre-Columbian inhabitants
of the Southwest. Where they lived when the first Spanish got there
is where they live today. And that's an important thing: if you go
to another state where there's a large Native American population like
Oklahoma, people have been moved around, in very dispersed, different parts
of the country, and that can affect their well-being. There's been
sustained European contact since about 1600; there's relatively low non-Indian
admixture. Now, there's a lot of admixture between tribes.
We've asked people about it, [and] a lot of people are married in from
one tribe or another tribe, or whatever. We've reviewed some records
that have, demographic records, that have been kept, and come up with something
close to 5%. We've looked at some genetic markers, and it gives us
the same answer.
But, despite all of this sort of cohesiveness of the population, what
we find is that currently there's not a lot of awareness of traditional
culture. So people, most people, many people don't know ceremonies,
haven't practiced ceremonies, don't know the language. But there
still are people who do as well. But many, many people don't.
But we think that this sort of involvement in your society -- whether you
want to call it sociology, or cultural anthropology or whatever -- is probably
an important factor in what contributes to those distributions in liability.
And that's what we're analyzing. We're not trying to fit a genetic
interpretation on everything we see. We want to know what's contributing,
whatever it is.
All of our subjects were adults over 21 years old; I think that may
be a serious limitation of this study, because a lot of these adults when
you talk to them, had severe alcohol problems when they were as young as
15 years old. And the fact that getting through the ethics of informed
consent is a lot easier with adults than with children, the important events
for this disorder may be happening when you're young. And that's
something that we have to consider in future studies. They had to
be eligible for membership in the tribe, which included a quarter or more
heritage. Almost all the people were full-blooded American Indians,
but as I mentioned, a lot of, there's a lot of mixing among tribes.
They had to be members of large pedigrees; that's because we had a genetic
study, and genetics requires analyzing family data. And we talked
to people in the community who were sort of considered matriarchs or patriarchs
to help us identify them. And then we recruited people independent
of any clinical histories. We wanted to have an idea about what happens
if you look at the population. We had to sort of insure this: we
had these very large families, and we talked to every single person in
the family before we moved to the next family, to make sure that we weren't
biased, you know, by talking to people who wanted to talk about it because
they didn't have the disease, or did have the disease, or for whatever
reason. We wanted to have complete coverage.
About 95% of the people -- well, about 3% of the people that we contacted
chose not to participate in the study, so we had a very high rate of participation.
There were another -- we oftentimes gave people an opportunity to make
up their mind. We didn't come to their door one day and say, "This
is what we're doing, would you like to do it today, yes or no, tell me
right now?" We said, "This is what we're doing, we'll come back and
ask you about it." And when the study ended, there were some people
that hadn't made up their minds yet.
Okay, now how do we make these diagnoses for this trait I was talking
about? As I mentioned, they were made from the DSM-IIIR, but how
do you know if somebody fulfills those criteria? Well, you have to
use a questionnaire -- psychologists call them instruments; I can't do
that. I think if you have a piece of paper with questions on it,
it's a questionnaire. [laughter] Instrument if it's a ruler
or a test tube, or something that you can talk about, talk about instrument.
But somebody else used the word 'reified' today, and there's some reification
among psychologists and psychiatrists. But it's something that's
called the Schedule for Effective Disorders and Schizophrenia, Lifetime
Version. It's something that's been established; it's been used,
it's been verified for replicability, repeatability, and the like.
Maybe not perfectly, but it's also not something that's ad hocly imposed.
Now, the questionnaires were filled out by an interviewer, so the person
who's being interviewed doesn't have the responsibility of doing that and
feeling like they have choices A, B, C or D and none of them are appropriate;
their reactions can be recorded. The interviewer was a cross-cultural
psychologist, so he has a Ph.D. in psychology. In this study, almost
600 people were interviewed. The psychologist lived on the reservation
for five years for this. So we've talked about cultural involvement
-- you know, is this via helicopter or I forget what the analogy was --
it wasn't by helicopter. This guy spent a lot of time there.
He'd lived on other reservations before he started this project, and got
to know the community quite well. The government paid his salary
the whole time, so there's been some question: will the government pay
for people to do it right? I can't, I don't disburse funds, but it
has happened. Now, he does the interviews, then they were sent out
to two people to read the interviews and make evaluations whether the person
had a disorder or didn't have a disorder, and then there was a conference
that was held for all of the cases of disagreement, where they reviewed
it with a psychiatrist who had done cross-cultural work. The [inaudible
proper name] concordance was very high: they agreed 93% of the time,
and then felt satisfied in these consensus conferences to resolve the other
What did we find? And this is where we're going to find some
shocking results. We looked at 329 women and 253 men, with the average
ages in the mid to late 30s. Now, that's a big difference in the
sex ratio. However, it is consistent with the United States census
for this population; and so there's a tremendous attrition of males, and
we see that reflected in our sample. Among the women, when we talk about
alcohol dependent, it's not: were they alcohol dependent at the time, but
whether ever in their life they had qualified for a diagnosis. So
about half the women had, half of them hadn't, but many more men had, about
6 out of 7 versus about 1 out of 7. So it's a very, very, very prevalent
problem in this population.
The next thing that we wanted to know was what sort of family patterning
was there involving this? And so we divided people up by degrees
of relationship. So these folks here shared half of their genes,
and that would include relationships like mother/child, father/child, full
siblings. A quarter of their genes would include things like, relationships
like aunt/nephew, half-siblings, grandparent/grandchild. Third would
be an eighth of their genes, and it would be roughly equivalent to first
cousins or the like. So as -- if we look here, at the line for the
females, what we find is that there's considerable correlation for the
first-degree relatives, for the second-degree relatives, and it's diminishing
in the second degree, and for the third degree, and when you get out to
fourth degree, there's no discernible correlation.
These p-values relate to something that's called statistical significance;
if it's below .05, by convention that's considered statistically significant.
If it's above .05, the interpretation is, well, you could get results like
that by just random chance, if there were no differences by degree
of relationship. And you can see for the females it's very low for
the first degree, it's statistically significant for the second and third,
not for the fourth degree. For the males, though, we really don't
hit statistical significance anywhere. So the family patterning
-- does a female relative predict alcohol, if she's alcoholic, does it
predict alcoholism in her female relatives, her daughters or whatever?
Yes. They -- oh, time. Okay. The important -- there's
an important family patterning. I'm going to have to pick up here
a little bit. What I want to show you though, is, I've also been
studying another American Indian population in the Southeast. In
this population here, what you find is that alcoholism is very low, and
that there's tremendous population variability. So that if I study
one population, I can't superimpose that pattern onto another population.
The next thing I want to show is that if we look at two studies of alcohol
dependence in Euro-Americans, what we find is that the so-called heritability
or the family patterning is higher in males than in females, whereas in
Native Americans, it's higher in females than in males. And so if
I go to the books on it, it says that the genetics on alcoholism is high
in males but not in females. Yet, if I go to a population they haven't
studied, I find it's the reverse way around. Now, I don't want to
imply that all of the family patterning is due to genes, because it could
be due to shared environment, and it's hard to separate those things out.
Nonetheless, there is an important family patterning, and it's something
important to look at.
We've done some linkage analyses, where we've looked at a lot of random
locations across chromosomes. Each dot here represents a particular
location on a chromosome that stretches, they stretch out over chromosome
11, in units called centimorgans. We find strong evidence for predisposing
gene at the top end of the chromosome, on the PR. We find some interesting
evidence on chromosome 4, using a similar sort of approach. We don't
find evidence like that everywhere we look. If we look at chromosome
1, it's quite flat, there's virtually no evidence for a gene harboring
I'm going to quickly go ahead. What's, what are the key elements
of research success? Well, you've got to explain it. Why is
it important to the researchers and potential participants? When
I went to the Southeastern tribe, I told them it was important to my research
program, because I wanted to look at people who didn't have big problems
so I could help figure out what made you healthy as well as what made you
disorders, have a disorder, so I asked them for their help. You have
to explain potential risks and benefits. This conference has been
very heavily weighted towards population sorts of levels, but there are
risks to individuals. Like in the course of giving these interviews,
it may cause psychiatric conditions to reoccur, and it can actually be
quite traumatic. We've dealt, we're equipped, shall I say, with what
to do in case that happens, who the people can go to for help. You
have to protection to participants and communities. In my more recent
studies, I don't see the names of the people, even though I'm the principal
investigator on the studies. You have to tell people what the likely
outcomes are going to be: it's going to tell us general knowledge, specific,
or is it going to result in specific applications? Generally, we
tell them it's general knowledge. Then I think it's also important
to explain the technical soundness. I don't think it's, we haven't
talked about the ethics of how important it is to have a research program
that 's likely to be successful, but when you're asking these questions
about whether you can find a gene or association between a gene and an
environment, or a disease [tape stops]