Whole Genome Approaches to Complex Kidney Disease
February 11-12, 2012 Conference Videos

Ethnicity and Community: Impact of Genetic Findings and Disclosing Results
Malia Fullerton, University of Washington

Video Transcript

1
00:00:00,000 --> 00:00:07,400
SARA HULL: So, now I’m going to introduce Stephanie Malia Fullerton who’s an Associate Professor of Bioethics in Humanities at the

2
00:00:07,400 --> 00:00:14,333
University of Washington School of Medicine. She holds adjunct positions in the departments of genome sciences and epidemiology and is a co-

3
00:00:14,333 --> 00:00:24,433
investigator of the University of Washington Center for Genomics and Healthcare Equality. Her work explores scientists’ understandings of

4
00:00:24,433 --> 00:00:31,599
human genetic variation and its relationship to disease risk, the use of racial and ethnic constructs and the conduct and interpretation of

5
00:00:31,600 --> 00:00:39,500
genetic research, and the responsible incorporation of genomic methodologies into broader programs of health disparities research.

6
00:00:39,500 --> 00:00:48,666
She’s also very involved in exploring participant perspectives on data sharing, research use, and result return in the context of genomics. And so,

7
00:00:48,666 --> 00:01:06,466
she’s going to add an additional nuance to this conversation, talking with us about ethnicity and community and its impact on these issues.

8
00:01:06,466 --> 00:01:16,799
STEPHANIE FULLERTON: Thank you very much, Sara, and thank you so much to the organizers for inviting me to participate, and thank you to all

9
00:01:16,800 --> 00:01:21,966
of you who are here in the audience. Maybe it’s just because on West Coast time but I’m normally sound asleep at this time on Sunday morning. So,

10
00:01:21,966 --> 00:01:31,299
I’m very pleased that you’re here and I hope that I will be able to keep you awake with some of my thoughts. I want to sort of preface this by saying,

11
00:01:31,300 --> 00:01:41,533
as Sara mentioned, I am involved in a range of research ethics-related research within and around the conduct of genomic research, and I’m

12
00:01:41,533 --> 00:01:50,033
used to expounding from the position of having actual data. I think in this domain where we’re thinking in particular about the implications of

13
00:01:50,033 --> 00:01:58,699
findings from exome and whole genome-level research for communities, we’re actually in a bit of a data void at the moment; we don’t have much

14
00:01:58,700 --> 00:02:07,300
good data. And so, most of what I’m going to be talking to you about today is speculation and basically a call for more research as you all move

15
00:02:07,300 --> 00:02:15,100
forward doing the very important science that you’re trying to do. So, just by way of a road map, I’m going to try to do three things in today’s

16
00:02:15,100 --> 00:02:23,166
presentation. I’ll keep an eye on the time here. First, I want to…oh, and I wanted to say at the outset, so we’ve had a genetic counselor talk to

17
00:02:23,166 --> 00:02:29,499
us about informed consent, we had a lawyer talk to us about return of results, and now you’re going to have a population geneticist talk to you

18
00:02:29,500 --> 00:02:36,166
about community, because that’s what my primary training is actually, in human population genetics, and that sort of colors my take on this

19
00:02:36,166 --> 00:02:43,066
topic and I wanted to sort of say that at the outset. I’m going to talk a little bit about the nature of the findings that are anticipated from whole

20
00:02:43,066 --> 00:02:50,699
exome and genome sequence analysis and its implications for recruitment and result return as it impacts communities. I’m going to talk a little bit

21
00:02:50,700 --> 00:02:58,166
more about another aspect of the kinds of anticipated results that might be generated, focusing in particular on the population

22
00:02:58,166 --> 00:03:06,632
distribution of those anticipated findings, and talk a little bit about the possible implications for study design and communication with communities, as

23
00:03:06,633 --> 00:03:13,766
well as the ultimate translation of the information that we’re going to be generating by these approaches. And then in the last few minutes of

24
00:03:13,766 --> 00:03:22,766
my presentation, begin to kind of very briefly address potential strategies that we might wish to employ in addressing these challenges that are

25
00:03:22,766 --> 00:03:36,732
posed by this new scale of genomic science. Okay. This was basically based on a review of science as I was aware of it prior to yesterday’s

26
00:03:36,733 --> 00:03:43,766
presentations. I tried very hard not to fiddle with these slides, but fortunately I think they’re commensurate with a lot of stuff we heard

27
00:03:43,766 --> 00:03:49,232
yesterday. So, what are we going to expect to find from whole exome and whole genome sequencing? As we’ve already heard, there’s

28
00:03:49,233 --> 00:04:00,166
going to be a deluge—a Niagara Fall, a fire hose, all those water analogies—of genetic information that’s going to be generated. As we also know,

29
00:04:00,166 --> 00:04:07,966
with the switch to sequencing, particularly in the exonic regions of the genome, the vast majority of the genetic variation that’s going to be

30
00:04:07,966 --> 00:04:15,866
identified is going to be rare. So, we’re moving out and away from this focus on common genetic variations that might be shared across

31
00:04:15,866 --> 00:04:24,132
populations and we need to know, looking at variations in any given population, are going to be typically in the fractions of percent, certainly less

32
00:04:24,133 --> 00:04:32,299
than half a percent, and many of the data we heard about yesterday, less than a tenth of a percent in a population relative frequency. And in

33
00:04:32,300 --> 00:04:39,233
addition and very relevantly and importantly to Ben’s presentation that we just heard, a significant fraction of the variation that is going to

34
00:04:39,233 --> 00:04:47,966
be identified, particularly in the exonic regions—not all of it but a significant amount of it—is going to be functionally relevant, with functional

35
00:04:47,966 --> 00:04:58,499
relevance variably and incompletely and imperfectly defined and understood. So, let’s just stop and think about those two aspects of the

36
00:04:58,500 --> 00:05:04,933
kinds of information that are likely to be generated from whole exome and whole genome sequencing. What are some of the implications?

37
00:05:04,933 --> 00:05:16,666
As we heard yesterday, in a turn to looking at rare variation and trying to understand the phenotypic effects of rare variation, we are

38
00:05:16,666 --> 00:05:25,632
going to need to increasingly look at very large numbers of research participants. You thought we needed large sample sizes in GWAS; it’s

39
00:05:25,633 --> 00:05:34,099
going to be even more acute to have large populations as we turn to this brave, new world of exome and whole genome sequencing. And

40
00:05:34,100 --> 00:05:42,866
there’s a really big problem from a research ethics standpoint and the really big problem here is that the populations that we have available to

41
00:05:42,866 --> 00:05:54,432
us and that we have studied for decades now in the genome sciences are very skewed towards populations of European ancestry. Work that was

42
00:05:54,433 --> 00:06:00,399
done by Anna Need and David Goldstein and then subsequently commented on by Carlos Bustamante and colleagues last summer

43
00:06:00,400 --> 00:06:09,266
describing the population distribution of the GWAS studies that have been conducted to date, 96% of GWAS studies have been conducted on

44
00:06:09,266 --> 00:06:19,466
populations of European descent, 4% on non-European samples. So even as we need to be increasing sample sizes, it is not going to be

45
00:06:19,466 --> 00:06:28,832
possible if we are interested in understanding diverse contributions to genetic risk to simply go to our freezers and start with what we have.

46
00:06:28,833 --> 00:06:35,433
There’s going to be an acute need for the recruitment of diverse populations from varied population backgrounds. What are some other

47
00:06:35,433 --> 00:06:45,833
implications? A very interesting one. I don’t really have any direct data that I can show you but, we know this. This is a feature from population

48
00:06:45,833 --> 00:06:51,333
genetics and it’s rather different from the situation with common genetic variants. As we increase the number of the people that we studied, our

49
00:06:51,333 --> 00:07:02,133
problem is going to get harder, not easier, because as sample size increases we’re going to identify more and more and more rare variation,

50
00:07:02,133 --> 00:07:06,999
okay? This is not the case with common variations: the more you look, the more you see what you already know is there and it easier it

51
00:07:07,000 --> 00:07:14,900
becomes to make sense of it. When we switch to looking at rare sequences, rare variations, there’s going to be a significant signal-to-noise problem

52
00:07:14,900 --> 00:07:22,500
and we’re going to have to figure out how to sort that out. And although, as we heard yesterday, there’s some very really neat-o cool statistical

53
00:07:22,500 --> 00:07:31,966
ways of making sense of an increasing statistical power surrounding, in particular, the analysis of an aggregate of rare variations, it’s not at all clear

54
00:07:31,966 --> 00:07:40,266
how we’re going to make effective clinical use of this information. We’re moving away from thinking about genetic risk variations as something that

55
00:07:40,266 --> 00:07:48,532
we might be able to kind of use from a broad public health context, moving much more into the clinical domain, and it’s not exactly clear how

56
00:07:48,533 --> 00:07:55,199
we’re going to make use of this information, so I just want to kind of flag those. We can about that more in questions-and-answers. So, those are

57
00:07:55,200 --> 00:08:05,333
some implications of the fact that we’re going to identify a lot of very rare variation. Another feature of the data, as I indicated, is that we’re

58
00:08:05,333 --> 00:08:16,866
going to identify a very large number, a very large number—hundreds and likely, thousands—of variations which our algorithms tells us are

59
00:08:16,866 --> 00:08:28,566
functionally relevant, and this is not in total, this is per person, okay? And that’s a very interesting and important scientific problem but it makes this

60
00:08:28,566 --> 00:08:35,199
business of thinking about the return of information to research participants that we know there’s a lot of interest in and excitement

61
00:08:35,200 --> 00:08:42,000
about currently in the genomics community, enormously complicated, because it’s been talked about and, as we heard in a very nice

62
00:08:42,000 --> 00:08:50,733
presentation by Dr. Solomon yesterday, variant effects are going to need to be understood and prioritized with regard to their clinical salience

63
00:08:50,733 --> 00:08:58,733
prior to communicating individual findings to research participants. And as Ben has already indicated, and I don’t want to spend a lot of time

64
00:08:58,733 --> 00:09:08,166
on this but actually, we don’t quite know how to do this yet. There’s a lot of disagreement in the bioethics community. Even in the clinical genetics

65
00:09:08,166 --> 00:09:16,132
community I have come to understand there are proposals on the table. This particular paper was alluded to in Dr. Solomon’s presentation yesterday

66
00:09:16,133 --> 00:09:23,466
by Berg and colleagues, a way of actually sort of categorizing or binning possible types of incidental findings that might be generated from

67
00:09:23,466 --> 00:09:36,466
whole exome and whole genome approaches, which basically relies on a joint adjudication of the clinical utility and validity of this information in

68
00:09:36,466 --> 00:09:40,366
conjunction with whether we have some prior knowledge or understanding of whether variants are deleterious or presumed deleterious, and

69
00:09:40,366 --> 00:09:51,632
depending on how information falls out in terms of those two bits of kind of decision making, this gives us an indication…let me see if I can figure

70
00:09:51,633 --> 00:10:04,666
out how to do this…never mind. This gives us an indication of which findings we might likely wish to return. This already is complicated and begins

71
00:10:04,666 --> 00:10:12,199
to start giving me a headache as someone who’s interested in this issue. What has been very sobering to realize is you put three medical

72
00:10:12,200 --> 00:10:21,900
geneticists in a room and they will not even agree with regard to the clinical utility of a lot of this information, and it gets very tricky. So, there’s

73
00:10:21,900 --> 00:10:29,366
going to be a lot of work to be done before we can actually meet, even potentially as understood currently, ethical obligations with regard to return,

74
00:10:29,366 --> 00:10:36,266
and I just wanted to make the point that functional relevance, particularly as we were talking about it yesterday where we were talking about sort of

75
00:10:36,266 --> 00:10:44,666
computer-based algorithms which will tell us about the likely effect on protein function, even those algorithms do not necessarily agree with

76
00:10:44,666 --> 00:10:52,166
each other. Functional relevance and an adjudication of functional relevance which might possibly be algorithmically tractable, is not the

77
00:10:52,166 --> 00:11:02,399
same as clinical utility. As I’ve already said, there’s a lot of disagreement in the medical-genetics community about clinical utility. We came

78
00:11:02,400 --> 00:11:09,933
face-to-face with this issue in the context of deliberative work as part of the Electronic Medical Records and Genomics Research Network that

79
00:11:09,933 --> 00:11:17,633
I’m involved in. We have a paper that is about to come out next week in Genetics in Medicine

80
00:11:22,766 --> 00:11:24,866
talking about our experience with deliberating on findings in the context of that research network, and these were incidental findings generated in

81
00:11:24,866 --> 00:11:29,799
the context of genome-wide association studies, not exome or genome scale investigations—sequencing investigation—where we determined

82
00:11:29,800 --> 00:11:37,333
after much discussion and debate and deliberation that there were four possible classes of findings that might rise to the level of return,

83
00:11:37,333 --> 00:11:45,766
and then after we finally came to that agreement and a consensus of the network, people took that back to their individual sites, looked very carefully

84
00:11:45,766 --> 00:11:53,432
at medical record data for the individuals who were affected by the particular genotypes, and where invariably at every single site and due to a

85
00:11:53,433 --> 00:12:00,199
confluence of factors including informed consent, including the understandings of the institutional review board, and other considerations,

86
00:12:00,200 --> 00:12:12,233
community interactions, no decision was made to return any finding. And in a classic statement—understatement—that I attribute to my many years

87
00:12:12,233 --> 00:12:20,333
of living in the United Kingdom, we wrote, “Although a criterion of ‘clinical actionability’ suggests a clear threshold for identifying results

88
00:12:20,333 --> 00:12:27,566
that should be considered for return, in practice, the identification of specific clinically actionable finding generated from the eMERGE studies was

89
00:12:27,566 --> 00:12:37,599
not straightforward.” This was a very difficult conversation. It took a very long time to figure this out for four incidental findings and we are about

90
00:12:37,600 --> 00:12:42,200
to enter in a brave, new world where we’re thinking about this now not for four, but for hundreds and possibly thousands of potentially

91
00:12:42,200 --> 00:12:55,766
functionally relevant variants, so lots of food for thought there. Okay. So, that is one class of finding, one way of thinking about the data. The

92
00:12:55,766 --> 00:13:02,932
data is generating many rare variants, the data is generating a fraction of which are going to be likely functionally relevant and may pose

93
00:13:02,933 --> 00:13:10,366
obligations, but exactly how to act on those obligations is going to be extraordinarily difficult to sort out. Where does ethnicity and community

94
00:13:10,366 --> 00:13:17,666
figure in here? Well, this brings me to the second way in which I want to think about the potential nature of the findings likely to be generated from

95
00:13:17,666 --> 00:13:26,299
the whole exome and whole genome data. This is just a snippet of the very considerable data that were reported out of the 1000 Genomes Project,

96
00:13:26,300 --> 00:13:36,433
a project to sort of sequence in considerable detail the whole genomes of a large number of samples collected from around the world and in

97
00:13:36,433 --> 00:13:43,799
data that was reported in 2010 and the emerging data from the Exome Sequencing Project which was talked about yesterday, but unfortunately,

98
00:13:43,800 --> 00:13:52,066
those data are not yet available publicly. We know that when we go in and look at populations of individuals sampled from different parts of the

99
00:13:52,066 --> 00:13:59,466
world and we look in particular in the exome regions of the genes where there’s lots of this rare variation, that what you identify that has

100
00:13:59,466 --> 00:14:07,732
previously been described tends to be more often shared between populations and what is rare and new and being seen for the first time is

101
00:14:07,733 --> 00:14:16,666
invariably population-specific, and this has very profound implications for how we’re going to actually communicate with communities about the

102
00:14:16,666 --> 00:14:27,432
conduct of this research. So, a little bit more on expected findings. Rare variation not only will require large sample sizes and have lots of

103
00:14:27,433 --> 00:14:37,366
functional relevance, but as I just indicated, rare variation will frequently be geographically restricted, will be population-specific, and in

104
00:14:37,366 --> 00:14:47,666
addition and this is something which was not talked about very much yesterday—it’s something that I don’t think has received nearly enough

105
00:14:47,666 --> 00:14:52,799
attention but I want to put it on the table for discussion—rare variants are going to be non-randomly distributed amongst individuals and

106
00:14:52,800 --> 00:15:05,933
families. Some individuals and families will have more of this rare novel variation than others. Why? Because of differences in background,

107
00:15:05,933 --> 00:15:13,433
polymorphic variation that have to do with human evolutionary history, and that are largely a function of genetic ancestry, which as we know

108
00:15:13,433 --> 00:15:22,699
as we saw in data presented yesterday by Suzanne Leal and others, is often correlated with self-described ethnicity. Okay, let’s think about the

109
00:15:22,700 --> 00:15:33,400
ethical implications of some of this stuff. Population-Specificity. Well, population-specificity, I think, is going to be an inevitable feature of much

110
00:15:33,400 --> 00:15:40,233
of the data being generated by these methodologies. We have to be very careful of how we talk about such results because

111
00:15:40,233 --> 00:15:47,333
population-specificity could frequently be misunderstood. It might lead us to overemphasize race- or community-specific genetic

112
00:15:47,333 --> 00:15:55,666
vulnerabilities in preference to other shared factors, and in an interesting way and a way that has not at all been addressed or explored, it

113
00:15:55,666 --> 00:16:02,166
could have some profound effects on the identity of people. For example, one could imagine the situation where an individual of one

114
00:16:02,166 --> 00:16:10,699
self-described ethnicity is told that they harbor a variant that has otherwise been described as specific to a different population, a different

115
00:16:10,700 --> 00:16:19,166
ethnicity, a different region of the world. Let us not underestimate the extent to which this is likely. There are some truly remarkable population

116
00:16:19,166 --> 00:16:26,232
genetic data out here. This is one of the most colorful and exemplary examples that I’m aware of, work by John Novembre and colleagues

117
00:16:26,233 --> 00:16:36,699
looking at patterns of genetic variation as assayed from sequencing in large numbers of people from Europe and where, basically, these

118
00:16:36,700 --> 00:16:44,033
are not…the colored circles have to do with sort of composite estimates of patterns of genetic variation and first and second principal

119
00:16:44,033 --> 00:16:53,099
components, but where there’s a very clear tendency for individuals whose heritage basically hails from particular places in Europe to have

120
00:16:53,100 --> 00:17:02,000
their genetic variation very closely resemble one another. So this propensity to have rare variants mapped very closely at specific geographic

121
00:17:02,000 --> 00:17:14,233
locations, particularly for individuals and families who have been in situ for several generations, is very high and we’re going to need to be thinking

122
00:17:14,233 --> 00:17:18,933
about this carefully as we begin figuring out how we’re going to communicate with participants about the kinds of findings we’re generating. In

123
00:17:18,933 --> 00:17:30,299
addition, population-specificity could, frankly, be not only misunderstood but potentially misused. It could distract us from the consideration of other

124
00:17:30,300 --> 00:17:36,300
relevant public health remedies in the context of the diseases and traits that we’re interested in studying and understanding, and it might open up

125
00:17:36,300 --> 00:17:43,233
the potential for stigmatization or discrimination in particular groups on the basis of the apparent population-specificity of findings that are

126
00:17:43,233 --> 00:17:51,933
indicated. There’s no evidence that this has happened to date, necessarily, but there interesting little suggestions of what possibly

127
00:17:51,933 --> 00:18:03,433
could occur. I just want to make reference to this paper which came out several years ago by Carlos Bustamante’s group, doing a resequencing

128
00:18:03,433 --> 00:18:14,066
study of a very small number of individuals, self-described Europeans or European Americans or African Americans and going in and making some

129
00:18:14,066 --> 00:18:24,799
inferences in reference to an out-species—chimpanzee—about the likely deleterious nature of genetic variation and coming to, at the time, the

130
00:18:24,800 --> 00:18:30,233
surprising and somewhat controversial conclusion that there was proportionately more deleterious genetic variation in European as

131
00:18:30,233 --> 00:18:37,999
opposed to African populations. Now, leaving aside the fact that they came to this conclusion on the basis of analysis of 15 Europeans and 15

132
00:18:38,000 --> 00:18:51,800
Africans, which may or may not be problematic, I think it’s very easy to imagine if the headline had been reversed—if they had found that there was

133
00:18:51,800 --> 00:18:59,900
deleterious variation in Africans than in Europeans—how we might think about the social salience of that, particularly in the context of

134
00:18:59,900 --> 00:19:08,533
really pronounced and profound health disparities which exist in the United States context. There’s just an example here of cancer disease burden

135
00:19:08,533 --> 00:19:17,999
by different racial and ethnic categories and sexes and incidence and death, and it would be very easy to jump to the likely erroneous

136
00:19:18,000 --> 00:19:25,466
conclusion that that deleterious variation is explaining a broader disease burden that many, many researchers believe have a much more

137
00:19:25,466 --> 00:19:36,232
complicated and multifactorial basis and a basis in social determinants of health. So obviously, there’s a need for great caution. That’s the

138
00:19:36,233 --> 00:19:45,499
population-specificity aspect of the data. There’s also this issue of the fact that the ways in which rare variations are going to be found in individuals

139
00:19:45,500 --> 00:19:51,833
that we sample are going to be non-randomly distributed and that indeed, population genetics tells us—and we already have data to support

140
00:19:51,833 --> 00:20:01,299
this observation—that individuals of sub-Saharan African ancestry are likely to have more rare and consequently more novel variation. What does

141
00:20:01,300 --> 00:20:09,733
this mean for us as investigators? Does this make individuals of sub-Saharan African ancestry interesting objects of investigation, or instead,

142
00:20:09,733 --> 00:20:19,133
difficult to study? I don’t know how to answer that question, but the difficult to study explanation has been one of the explanations provided for

143
00:20:19,133 --> 00:20:26,733
the fact that we have such an uneven distribution of samples currently in human genetics and genomics. For sure, there are going to be fewer

144
00:20:26,733 --> 00:20:36,933
definitive results to be communicated to research participants from such population genetic backgrounds in the near term, and this is already

145
00:20:36,933 --> 00:20:45,066
something that we have begun to encounter in the context of clinical genetic testing that is already clinically available. These are data from a

146
00:20:45,066 --> 00:20:52,332
particularly…they’re indicative data but it’s from a relatively small study of women who’ve developed breast cancer early in their lives,

147
00:20:52,333 --> 00:21:02,199
showing the genetic population mutational distribution of BRCA1 and BRCA2 variants and basically—I know this is very hard to see—but

148
00:21:02,200 --> 00:21:09,666
what I want to draw your attention to is the row describing variants of uncertain significance showing that approximately four times as many

149
00:21:09,666 --> 00:21:18,966
variants identified in the BRCA1 and BRCA 2 genes are found to be of uncertain clinical significance in individuals of African American

150
00:21:18,966 --> 00:21:28,932
ancestry, in part because they have simply been less well-studied to date. We can take what we’re experiencing in the clinical domain and

151
00:21:28,933 --> 00:21:35,999
bring that into the research context, add to the fact that we’re going to have a lot more rare and novel variation to being with and that we’re not

152
00:21:36,000 --> 00:21:43,766
going to be able to tell people about, and I think this poses some particularly profound ethical concerns. In addition, and this goes to the

153
00:21:43,766 --> 00:21:53,732
question that I had for Dr. Rich at the end of his presentation yesterday and in particular, his clear indication and I believe that the writing is on the

154
00:21:53,733 --> 00:22:01,099
wall, that the kinds of designs that are going to work best for the interrogation of rare variation turned up in exome and whole genome

155
00:22:01,100 --> 00:22:09,766
sequencing studies are family-based, pedigree-based investigations with the expected return to such study designs and the context of this kind

156
00:22:09,766 --> 00:22:17,599
of research. We’re increasingly going to be looking to the need to recruit large and extended pedigrees from all sorts of populations, and yet

157
00:22:17,600 --> 00:22:25,466
we don’t really know very much about how easy it is. I was very gratified to hear that there are a large number of pedigrees already available from

158
00:22:25,466 --> 00:22:33,099
underrepresented ethnic minority communities in the kidney disease domain; that’s wonderful. More will need to be collected and yet we don’t

159
00:22:33,100 --> 00:22:43,100
know, really, how to think about recruitment or even the communication of findings in a family-based context in minority communities. The most

160
00:22:43,100 --> 00:22:50,933
definitive report that was published a couple of years ago now on the use of family history in clinical research, making some recommendations

161
00:22:50,933 --> 00:22:59,399
for how to do this work well, unfortunately noted that there’s actually very little evidence to suggest the role that race or ethnicity, cultural

162
00:22:59,400 --> 00:23:05,733
background, religious belief or other characteristics might have on one’s willingness or ability to report on their family history, or

163
00:23:05,733 --> 00:23:15,066
presumably, to participate in research that would involve family-based ascertainment. These are just big, huge lacuna in our understanding and

164
00:23:15,066 --> 00:23:21,966
we’re going to have to be grappling with these issues, I think, as we move forward, of necessity in order to do the science that we want to do. So,

165
00:23:21,966 --> 00:23:30,199
just to quickly re-cap on my sort of summary of the data and where it leads us in terms of its ethical implications. The need to study and

166
00:23:30,200 --> 00:23:36,000
understand rare variation is going to, out of necessity, require the recruitment of new—not simply just using what we have to

167
00:23:36,066 --> 00:23:45,699
hand—recruiting new large cohorts. Understanding uncertain functional significance is going to require us to recruit and analyze diverse

168
00:23:45,700 --> 00:23:53,400
cohorts that are adequately powered. I didn’t really talk about that very much but it’s not just enough to sort of include minority representatives

169
00:23:53,400 --> 00:23:59,966
in our research, but we need to have enough of them in order to make robust inferences; so adequately powered to identify local population

170
00:23:59,966 --> 00:24:06,466
and possibly familial effects. Because of the population-specific nature of much of this variation we’re going to have to exercise extreme

171
00:24:06,466 --> 00:24:14,099
care in describing findings that are apparently restricted to particular social strata, and because of the non-random distribution of rare variation

172
00:24:14,100 --> 00:24:20,866
we’re going to have to acknowledge lots of complications inherent to the participation of certain communities. Even as we are inviting the

173
00:24:20,866 --> 00:24:29,799
participation, we’re going to have to make clear that we may not have the same kinds of information to return to them. What’s a researcher

174
00:24:29,800 --> 00:24:40,700
to do in the face of these somewhat uncomfortable facts about where the science is leading us? Well, obviously we need to be paying

175
00:24:40,700 --> 00:24:48,133
attention to the role of reciprocity in research. We’ve had some wonderful discussion already this morning about the role of informed consent

176
00:24:48,133 --> 00:24:57,733
and relationship-building and partnership-building with participants; I am all for that. As we do this work I think we have to acknowledge that the

177
00:24:57,733 --> 00:25:04,066
longer term public health benefits of much of what we’re doing in the near term are unclear and are going to be hard to explain to research

178
00:25:04,066 --> 00:25:16,099
participants. I cannot overestimate enough that altruistic research participation, which is what we all depend upon, is a luxury. It is easy for

179
00:25:16,100 --> 00:25:22,766
certain groups of people and it is not for others and we need to begin to acknowledge that, and rather than talking about populations that are hard

180
00:25:22,766 --> 00:25:29,766
to recruit, we need to be going out of our ways in order to collect the kinds of people and information that are going to make clear that

181
00:25:29,766 --> 00:25:37,232
public health benefits are going to be equitably distributed, while we recognize that this distant promise of public health benefit may not suffice

182
00:25:37,233 --> 00:25:47,399
as a rationale for participation for many communities. Near-term benefits are also likely to vary. The fact that a lot of the individual research

183
00:25:47,400 --> 00:25:52,766
findings generated from research are going to be indeterminant—and they’re going to be differentially indeterminant in particular

184
00:25:52,766 --> 00:25:59,732
communities and families—is important and is going to need to be acknowledged. We’re going to have to start having some honest

185
00:25:59,733 --> 00:26:08,133
conversations about the reason why we don’t know as much about certain groups is because we haven’t studied you to date and we need to

186
00:26:08,133 --> 00:26:15,766
be figuring out other ways to demonstrate reciprocity and to reward compensation that might be something other than returning individual

187
00:26:15,766 --> 00:26:25,166
findings. And then, recognizing that all of this conversation, all of this careful work, all of this engagement with populations and communities is

188
00:26:25,166 --> 00:26:33,166
basically the research community asking participants for our trust, and that if we’re going to ask participants for trust, we need to be

189
00:26:33,166 --> 00:26:46,899
demonstrating trustworthiness, and that is quite difficult and complicated, but I think it’s doable. I am very privileged to be a part of a study out of

190
00:26:46,900 --> 00:26:52,266
Morehouse University call MH-GRID, Minority Health-Genomic Research Infrastructure Development project, led by Gary Gibbons, which

191
00:26:52,266 --> 00:26:58,832
is going to be looking at African Americans, doing exome sequencing in African Americans in the context of hypertension risk and where we’re

192
00:26:58,833 --> 00:27:05,199
going to be doing some ethical research, talking to the participants as part of that study about their attitudes towards and preferences with regard to

193
00:27:05,200 --> 00:27:13,366
participation in exome sequencing research. The data that we have to hand, other data about just general African American attitudes towards

194
00:27:13,366 --> 00:27:20,799
participation in genetic research is that African Americans are as likely as whites to express their willingness to participate, and yet, also are

195
00:27:20,800 --> 00:27:29,333
more likely to report feeling that genetic research is going to possibly result in higher insurance; to not benefit their communities, to promote racism,

196
00:27:29,333 --> 00:27:38,266
and to use minorities as guinea pigs. These perceptions are real; they must be adequately addressed and dealt with in the context of our

197
00:27:38,266 --> 00:27:45,666
research engagements. At the same, we need to recognize that in many cases the greatest barriers to the participation of the communities

198
00:27:45,666 --> 00:27:53,899
that we want to involve may not be due to an inherent mistrust of research, but rather to a lack of access to healthcare and research

199
00:27:53,900 --> 00:27:59,833
opportunities, and so we need to be going out of our way to make ourselves available. And we also need to be thinking about research

200
00:27:59,833 --> 00:28:09,566
participants as true stakeholders in the research process. Based on some work that we did out of a group health cooperative not with ethnic

201
00:28:09,566 --> 00:28:18,132
minority communities—with white, middle-class, well-off, well-insured participants—we’ve come to some conclusions about ways in which we

202
00:28:18,133 --> 00:28:24,833
might or steps we might take in the context of genomic research to enhance respectful engagement. A lot of this has to do with things

203
00:28:24,833 --> 00:28:32,933
that Julie talked about in her presentation this morning of maintaining ongoing communication with participants; thinking of informed consent as

204
00:28:32,933 --> 00:28:41,133
a process; making sure that we stay in regular contact; we tell participants what we’re doing, why we’re doing it, and the ways in which we

205
00:28:41,133 --> 00:28:47,933
hope it will beneficial to themselves and, ultimately, to their broader communities. We need to have ways to provide access to the

206
00:28:47,933 --> 00:28:54,299
information about how samples are being used. We need transparent, accountable oversight processes, which is why some of the information

207
00:28:54,300 --> 00:29:01,466
that Laura Rodriguez talked about last night is so incredibly important. Even as we’re asking people to sign on to have their data be placed in federally

208
00:29:01,466 --> 00:29:09,499
controlled, semi-public access repositories, we need to be explaining to them why that is necessary and how their data are going to be

209
00:29:09,500 --> 00:29:18,766
protected and not misused, and we need to continually provide opportunities for research participants of all stripes to provide direct input on

210
00:29:18,766 --> 00:29:25,299
the stewardship of their data. And I know that’s something that we’re not really used to in the genomic research community, but we need to be

211
00:29:25,300 --> 00:29:34,000
moving in that direction. And then in those cases where such ongoing engagement or re-contact for particular research uses is not feasible, we

212
00:29:34,000 --> 00:29:40,733
need to be developing other effective methods to communicate with the public about responsible realistic study procedures, including thinking

213
00:29:40,733 --> 00:29:50,566
comprehensibly about ways to educate the public about the need to conduct research of this kind prior to any research ask in the line of the

214
00:29:50,566 --> 00:30:03,832
question that we had earlier today. So, not a lot of hard data, I know, mostly speculation, but what I hope you will go home with are three conclusions

215
00:30:03,833 --> 00:30:09,666
that I have come to as I’ve begun to think more deeply about the implications of this research for different ethnic groups and communities,

216
00:30:09,666 --> 00:30:16,699
particularly in the United States context. Whole genome approaches to complex kidney disease is going to require the participation of large and

217
00:30:16,700 --> 00:30:23,966
diverse cohorts. We are not simply going to be able to go our freezers; new recruitment is going to be needed and new recruitment and

218
00:30:23,966 --> 00:30:30,966
engagement strategies are going to be required. “Making sense” of the information that we are going to generate in the context of our research

219
00:30:30,966 --> 00:30:38,766
is not going to be straightforward and we’re going to have to exercise great care in describing those findings and returning results to individuals

220
00:30:38,766 --> 00:30:47,499
and to their families. It’s posing a whole new set of issues, and yet ironically in the way that was pointed out in the question-and-answer at the

221
00:30:47,500 --> 00:30:53,500
end of the day yesterday, in some ways returning us to sets of ethical concerns that we grappled with a long time ago when we were

222
00:30:53,500 --> 00:31:01,833
first working in the realm of linkage analysis. And in my opinion and in the opinion of my colleagues at the Center for Genomics and Healthcare

223
00:31:01,833 --> 00:31:10,466
Equality, ongoing respectful engagement which is attentive to and attends to community-specific concerns is ultimately the way in which we’re

224
00:31:10,466 --> 00:31:19,499
going to enhance research participation and reduce the perception of harm and actual harms involved in the conduct of such research. So,

225
00:31:19,500 --> 00:31:26,000
thank you very much for your attention…just some of the various organizations that I’ve been involved in. I’ve also had a consulting role in the

226
00:31:26,000 --> 00:31:46,600
Exome Sequencing Project as well as MH-GRID as I mentioned. So thank you very much. I wowed you all. Yeah?

227
00:31:46,600 --> 00:31:56,366
ROBERT KLETA: Robert Kleta, University College London. If I may, I just want to make a comment about whole genome approaches, complex

228
00:31:56,366 --> 00:32:05,566
kidney disease and large, large cohorts because my impression here is that you and others yesterday left the community in the room a little bit

229
00:32:05,566 --> 00:32:15,332
under the impression that we can’t move forward for rare diseases and the point I’m trying to make is, I think there’s a little bit of disconnect with rare

230
00:32:15,333 --> 00:32:27,199
variant and rare disease. So what I’m trying to say is, if you do linkage studies, then 4 to 10 samples—even recessive or dominant

231
00:32:27,200 --> 00:32:33,400
fashion—can net significant findings and you have the locus and you find the gene and you know what’s going and that can even be true for

232
00:32:33,400 --> 00:32:42,200
complex kidney diseases. If you go to genome-wide association studies—and I think that’s what now many people in the room think about—then I

233
00:32:42,200 --> 00:32:52,300
take the liberty also of saying actually you just need a well-defined cohort and it can be as complex as it be, but 100 samples are enough to

234
00:32:52,300 --> 00:33:01,133
find the locus and the genes of interest. Even two genes can be involved and can be very complex. So what I’m afraid is happening from the

235
00:33:01,133 --> 00:33:09,999
event of genome-wide association studies and common diseases is it’s in everybody’s mind, oh, you need huge, huge, huge cohorts, and I think it

236
00:33:10,000 --> 00:33:19,133
would just be sad if this paradigm is now continued for complex kidney diseases which may just be complex because of the nature but

237
00:33:19,133 --> 00:33:26,799
not because of the approach that should be taken. So again, to try to make the point, I think nothing of what has been said yesterday and

238
00:33:26,800 --> 00:33:37,566
today is wrong, but there’s a disconnect in understanding using these tools in terms of what do we do and what’s right in terms of the biology

239
00:33:37,566 --> 00:33:50,166
or the theoretics behind it? Again, in my opinion, you need well-defined cohorts but then actually you need small cohorts to move forward. You do

240
00:33:50,166 --> 00:34:00,199
not need 1,000 or 2,000 or 10,000 samples. If you study common disease, where you, indeed have lots of noise and many rare variants playing to

241
00:34:00,200 --> 00:34:08,100
different pathways, then sure. That was the disaster of the past years of genome-wide association studies, but I think this should now be

242
00:34:08,100 --> 00:34:18,066
over and we should understand how we could use the tools. So again, no offense intended. If I would know anything—and I take pride, actually,

243
00:34:18,066 --> 00:34:25,232
being trained here at the National Human Genome Research Institute in genetics—I would now have the message, go home, okay, we have 100 or

244
00:34:25,233 --> 00:34:35,766
200 samples of a disorder I’d like to study. I understand that’s my cohort but I’m saying, look, the first science paper, genome-wide association

245
00:34:35,766 --> 00:34:44,732
study 2005 by Kline and colleagues, 86 samples, I will be presenting tomorrow in the Neptune study our approach to membranous nephropathy,

246
00:34:44,733 --> 00:34:55,633
people in the room probably agree that’s a complex disease. 75 samples we find the first locus, 150 we find both loci. Sorry, I just thought I

247
00:34:55,633 --> 00:35:00,699
want to put this here in front of everybody. Thank you.

248
00:35:00,700 --> 00:35:06,533
STEPHANIE FULLERTON: Yes, thank you, and I would actually really invite those of you who are more sort of up on the methodology to sort of

249
00:35:06,533 --> 00:35:19,433
comment on that point. I will say, though, where I am coming from as a bioethicist who interacts with communities around making kind of the case

250
00:35:19,433 --> 00:35:26,666
for research participation and as someone who is enormously concerned that as we generate genomic knowledge, we do not generate

251
00:35:26,666 --> 00:35:34,699
knowledge which will preferentially benefit certain populations and not others. The fact that from a methodological standpoint we can adopt a

252
00:35:34,700 --> 00:35:43,000
study design which only requires a small number of samples and which will get us definitive information on a small number of genes, does not

253
00:35:43,000 --> 00:35:49,600
begin to address the population burden, the pronounced disparities in kidney disease outcomes that are present in this country. And so

254
00:35:49,600 --> 00:35:58,933
even while such designs as you talk about might be tractable from the point of view of identifying particular rare etiological contributions is not

255
00:35:58,933 --> 00:36:05,999
going to make a dent in that public health problem, and that is the problem that the communities that I interact with care about, right? And so, we have

256
00:36:06,000 --> 00:36:15,166
to start figuring out how to do both at the same time and I think this is a complicated issue.

257
00:36:15,166 --> 00:36:22,966
MALE: That was a lovely talk and I want to bring you back to, I think, what was the title of the talk about ethnicities and communities and populations

258
00:36:22,966 --> 00:36:33,766
and I worked with Jeffrey Kopp on a beautiful study where we identified an important locus for renal disease in HIV infection. When we designed

259
00:36:33,766 --> 00:36:42,099
that study we were concerned about the ethnicities and we asked that all patients—because it was proposed to be African American

260
00:36:42,100 --> 00:36:51,300
patients—have four—I think that’s as many grandparents as you’re entitled to—four African American grandparents. We’ve also been

261
00:36:51,300 --> 00:37:03,066
involved in the FIND in the study that when we started it was particularly looking at different ethnicities because we thought that the

262
00:37:03,066 --> 00:37:10,299
expression of renal disease might be very different, both in its genetic susceptibilities and bases and in its clinical expression in different

263
00:37:10,300 --> 00:37:23,833
populations. Now I’m wondering, as people from the 50s and 40s are dying out and we have a new sort of multi-ethnic character in populations

264
00:37:23,833 --> 00:37:32,766
in the United States, what’s going to happen to our principal component analyses? What’s going to happen to our idea about using identified—self-

265
00:37:32,766 --> 00:37:41,699
identified—ethnic backgrounds when we don’t know what those mean anymore or people are fluid in their identification of ethnicities? So, that

266
00:37:41,700 --> 00:37:45,400
should be a fun question. STEPHANIE FULLERTON: Thank you. I mean,

267
00:37:45,400 --> 00:37:50,000
that’s a great question and we had a preliminary answer and I know you weren’t able to be with us for a lot of the day yesterday. There was a

268
00:37:50,000 --> 00:37:58,433
preliminary answer from our analytical people, and Suzanne Leal in particular, showing the ways in which we use the aggregate genetic

269
00:37:58,433 --> 00:38:06,033
information to say things about population genetic background, and increasingly in the genomics community, we use this information in preference

270
00:38:06,033 --> 00:38:16,699
to self-reported ethnicity. And yes, that’s absolutely relevant and important from the point of view of controlling for background population,

271
00:38:16,700 --> 00:38:28,133
allele frequency differences, it might be confounding and association analysis. It becomes more problematic, and this gets to topics that I

272
00:38:28,133 --> 00:38:38,633
talked about a couple of years when I last spoke at an NIDDK meeting, this whole issue of it is not bags of genes that walk into doctors’ offices, but

273
00:38:38,633 --> 00:38:47,466
it’s people who are ascribed to particular racial and ethnic identities. And so even as we can use the genetic information and people’s backgrounds

274
00:38:47,466 --> 00:38:57,066
to kind of put them in the right bin for our genetic analyses, there’s still very, very hard questions about then how that translates back out to the

275
00:38:57,066 --> 00:39:06,366
people of mixed ethnicity who are making up an increasingly large part of the population, and we had this sort of interesting, from my point of view,

276
00:39:06,366 --> 00:39:16,666
sort of hopeful response that as invariably in this work there is a return to a more family-based ascertainment in investigation, that those

277
00:39:16,666 --> 00:39:25,066
particular problems of population background confounding go away, because we can actually look very precisely at patterns of variation in the

278
00:39:25,066 --> 00:39:35,432
family and that there’s less of a concern about at least controlling for that background population in genetic variation. I don’t exactly know how to

279
00:39:35,433 --> 00:39:44,166
evaluate those claims. I still think they’re very interesting. I still think we’re going to have problems with recruitment and with making sense

280
00:39:44,166 --> 00:39:54,966
of the functional significance of these findings in a body that has a collection of genes and that walks through the world with a particular

281
00:39:54,966 --> 00:40:00,399
understanding of its place in the world, and that’s where it gets very complicated.

282
00:40:00,400 --> 00:40:06,133
MALE: [inaudible question from audience] STEPHANIE FULLERTON: Sure. Yes, exactly, and

283
00:40:06,133 --> 00:40:19,233
it’s very complicated and I think this is…as you know, my preference is to kind of keep self-reported ethnicity as a variable—as a covariate—

284
00:40:19,233 --> 00:40:26,399
in our analyses even while we’re controlling for genetic population background. But still, then how to use this information in a public health context

285
00:40:26,400 --> 00:40:32,333
becomes very, very complicated. Yes? GERJAN NAVIS: Gerjan Navis, Groningen. I

286
00:40:32,333 --> 00:40:42,833
recognize and appreciate the complexity of this self-reported ethnicity but there are relevant biological variables that go along with ethnicity

287
00:40:42,833 --> 00:40:52,899
and that’s lifestyle, and of course, it might be more easy to grasp the lifestyle differences between people of varying ethnicity, whatever

288
00:40:52,900 --> 00:41:09,700
that may be, and that might also give us tools to intervene with, let’s say, risk behavior, and that also relates to the remark that was being made

289
00:41:09,700 --> 00:41:15,366
about use of small, very well-characterized populations because, of course, it’s a lot of work to bring lifestyle and to document it properly and

290
00:41:15,366 --> 00:41:27,232
to document it in a reliable fashion. But it might be a way around this very complicated issue of “what is ethnicity?” At least it’s something you

291
00:41:27,233 --> 00:41:31,966
can measure and that’s actionable. STEPHANIE FULLERTON: Yes, absolutely, and I

292
00:41:31,966 --> 00:41:46,166
wholeheartedly agree. I think we need to be measuring very well, very comprehensively all sorts of factors that might be contributing to

293
00:41:46,166 --> 00:41:56,932
disease incidence, and lifestyle factors are a very important consideration. There is also the very interesting issues of…the interesting issue

294
00:41:56,933 --> 00:42:02,499
is: can we basically take ethnicity and sort of disaggregate it into its component pieces, its genetic pieces, and its lifestyle pieces? And I

295
00:42:02,500 --> 00:42:11,100
know there’s a lot of interest in doing that. I and others are interested in that but would like to retain the social identifiers as well because we

296
00:42:11,100 --> 00:42:18,733
also have this problem, which may be less of a problem in the Netherlands, I’m not sure, but it’s certainly a problem in the United States of racism.

297
00:42:18,733 --> 00:42:27,733
And so, even as we account for and take advantage of information on lifestyle factors and on genes, there still might be value in holding in

298
00:42:27,733 --> 00:42:38,233
play in our analyses the ways in the people walk through the world so that we understand better the ways in which social interactions intersect

299
00:42:38,233 --> 00:42:45,866
with lifestyle and other behaviors, as well as genetic influences. We need to have it all.

300
00:42:45,866 --> 00:42:51,632
GERJAN NAVIS: I think we agree and unfortunately yes, in the Netherlands racism is a problem also and a big concern, yes.

301
00:42:51,633 --> 00:42:55,199
STEPHANIE FULLERTON: Yes, thank you. SARA HULL: Thank you. That was a very

302
00:42:55,200 --> 00:43:05,166
interesting presentation. It was great and some important points are being made about the impact of self-reported ethnicity and race on science

303
00:43:05,166 --> 00:43:13,899
and interpretation and I think you also hinted at the inverse point that some of this research is going to have an impact on how people self-

304
00:43:13,900 --> 00:43:22,333
identify. I just wanted to mention a very interesting and complicated ethics consult that we had recently that I guess I would characterize

305
00:43:22,333 --> 00:43:34,733
as misattributed ethnicity and a finding where a SNP that had once been classified as being found only in a certain ethnic or racial group was

306
00:43:34,733 --> 00:43:43,833
reclassified later on, and whether that kind of result should be communicated back to an individual who is receiving results. It launched a

307
00:43:43,833 --> 00:43:52,099
very interesting conversation about self-perceptions and whether it’s racist to ask such questions in consults, but I think that’s just a

308
00:43:52,100 --> 00:44:02,500
glimpse at the feedback and how this is going to go in both directions and the importance of being aware of the impact of emerging research of the

309
00:44:02,500 --> 00:44:09,966
fact that, when you have initial results, that evolving results are going to change over time as you generate more and more information—your

310
00:44:09,966 --> 00:44:18,666
very first point—and what that’s going to mean for how we communicate with people will have really important identity implications.

311
00:44:18,666 --> 00:44:27,499
STEPHANIE FULLERTON: Yes, thank you.




Date Last Updated: 9/18/2012

General Inquiries may be addressed to:
Office of Communications and Public Liaison
NIDDK, NIH
Building 31, Rm 9A06
31 Center Drive, MSC 2560
Bethesda, MD 20892-2560
USA
Phone: 301.496.3583