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

European Studies and Repositories: Opportunity and Challenge
Gerjan Navis, University Medical Center Groningen, The Netherlands

Video Transcript

1 00:00:00,000 --> 00:00:10,766
JEFFREY KOPP: So, our next speaker, Gerjan Navis, from Groningen, Netherlands, is a professor in genetics, nephrology and urology at

2
00:00:10,766 --> 00:00:19,399
the academic hospital there. Her role is to think about the same kinds of things that Linda Kao was just talking about, now in the context of

3
00:00:19,400 --> 00:00:27,866
European or possibly other international cohorts, and in fact, she’s worked closely with several cohorts including the PREVEND study of diabetic

4
00:00:27,866 --> 00:00:37,366
disease, hypertension, Global Blood Pressure Genetics Consortium, and she’s been involved in a large number of GWAS consortia similar to

5
00:00:37,366 --> 00:00:43,499
what Linda Kao was talking about. So with that, I’ll welcome her and look forward to her talk.

6
00:00:43,500 --> 00:00:50,266
GERJAN NAVIS: Well thank you, Dr. Kopp, for your kind introduction. Also, thank you for the invitation to be part of this challenging

7
00:00:50,266 --> 00:01:03,299
conference. Actually, it’s humbling as a simple nephrologist to be scheduled among such excellent speakers. But yes, I’m here and I will do

8
00:01:03,300 --> 00:01:14,300
my best. I’m not a professor of genetics but I’m a professor of nephrology and one of the things I’ve been doing over the last ten years is try to

9
00:01:14,300 --> 00:01:25,433
grasp the meaning of genetics for clinical nephrology. Actually, this is the Groningen University Hospital where I’m located and this is

10
00:01:25,433 --> 00:01:38,499
the area where Groningen is located, and actually, this area has characteristics of a founder population. It might be not Iceland but

11
00:01:38,500 --> 00:01:46,633
probably it’s somewhere between Iceland and the United States in terms of founder characteristics. Relatively rural, Groningen is the

12
00:01:46,633 --> 00:01:55,733
only really big town and this is what it looks like these days. We have a really good winter, ice everywhere so you can speed skate out in the

13
00:01:55,733 --> 00:02:04,699
open, which is one of the nicest and most exciting things one can do, and what you see here is that half of the population, including the

14
00:02:04,700 --> 00:02:18,966
army, is cleaning the ice for the other half to be able to skate. One of the things that we have in this part [---] from where I come from is a long

15
00:02:18,966 --> 00:02:28,399
journey through all the freezing towns, 200 kilometers long. It’s only there when there’s ice all over the place, when the ice is safe all over. It’s

16
00:02:28,400 --> 00:02:40,133
being issued one in each 20 years or so and last week we would have one. Well, it has been cancelled, unfortunately, because the

17
00:02:40,133 --> 00:02:49,899
temperature is going up. But nevertheless, making long journeys on the ice is one of the most exciting things one can do. It’s also very slippery

18
00:02:49,900 --> 00:03:00,333
and there’s a metaphor because now this weekend I cannot be on the ice—I’m here—and I’m on another slippery, long but exciting journey,

19
00:03:00,333 --> 00:03:12,266
namely, the journey of finding out how genetics and genomics can help to resolve the big issues in nephrology. Well, I’ll discuss some of the

20
00:03:12,266 --> 00:03:21,132
experience and share with you some of the experiences we had in the last ten years in the journey from clinical nephrology to GWAS and

21
00:03:21,133 --> 00:03:30,999
back, tell you about some of the resources we built along the way, because we started off with zero, and share with you some of the lessons

22
00:03:31,000 --> 00:03:42,566
we learned along the way. Well, let’s first agree on the big issues in nephrology. I think I’ll start off with those ones, and it’s in my view—my very

23
00:03:42,566 --> 00:03:49,766
biased view—the prevention of progressive renal function loss and the prevention of its complications. As long as we agree on that, at

24
00:03:49,766 --> 00:04:00,899
least we have a starting point. What did we expect from genetics? What could be its potential? First of all, the early identification of

25
00:04:00,900 --> 00:04:11,933
high risk subjects, leading to better risk stratification which could modify clinical practice in terms of better allocation of preventive and

26
00:04:11,933 --> 00:04:20,999
therapeutic resources. The other thing, basic science: identification of novel pathways or dissection of known pathways of disease that

27
00:04:21,000 --> 00:04:30,066
could give us better targets for intervention, and finally, providing a rationale for individualized therapy and thus reduce the burden of treatment

28
00:04:30,066 --> 00:04:40,999
with the same or even more health benefits. That would be the big expectations on genetics. Well, these were issues we discussed when we

29
00:04:41,000 --> 00:04:49,533
starting with reaching it, which is a European investigator-driven network that started with commissions that recruited geneticists and

30
00:04:49,533 --> 00:04:59,566
epidemiologists, established in 2004, with members all over Europe in an essentially bottom-up approach. Our mission was stated as:

31
00:04:59,566 --> 00:05:08,466
promote the translation of the advancements in genetics to clinical benefit for the renal patients and do this in terms of expertise development and

32
00:05:08,466 --> 00:05:17,799
sharing expertise and data, from the fact that we realized actually we knew nothing; we had nothing, we could nothing, and we wanted to

33
00:05:17,800 --> 00:05:28,700
work on infrastructure for genetic research in nephrology. The issue at that time, as it is now, was that there were a lot of association data

34
00:05:28,700 --> 00:05:40,533
around that poorly replicated and it also poorly translated into the clinic, and already at that time we realized—as everybody else—that quick and

35
00:05:40,533 --> 00:05:52,099
dirty answers will not work in complex disease and that genetics will not provide magic bullets in complex disorders. Essentially, there are two

36
00:05:52,100 --> 00:06:00,966
approaches to address this and have been addressed already this morning: first of all, to increase the numbers in the genetic epidemiology

37
00:06:00,966 --> 00:06:12,766
approach that allows GWAS by ever larger studies up to tens and hundreds of thousands, so sample sizes unheard of in any nephrology

38
00:06:12,766 --> 00:06:21,532
study; replication, replication, replication and meta-analysis, that’s one approach essentially. The other one is to refine the phenotype—the

39
00:06:21,533 --> 00:06:28,699
clinical approach that we are familiar with because we do the clinic everyday—with a careful definition based on pathophysiology on

40
00:06:28,700 --> 00:06:40,833
knowledge of renal disease and analyze for gene-environment interaction to dissect so-called contradictory results. And finally, in science

41
00:06:40,833 --> 00:06:52,799
when there is not a single methodology as a Holy Grail, you combine data from different sources. People in sociology knew it all the time, they call it

42
00:06:52,800 --> 00:07:03,266
triangulation. If you find the same thing for three different sources that are a little shaky on their own account, you can believe your results, the

43
00:07:03,266 --> 00:07:15,366
simplified phenotype for the large numbers, the enriched phenotype for the small numbers, and combination of data. Of course, this was where

44
00:07:15,366 --> 00:07:27,832
the bottle-neck was because nephrology was not up to that and perhaps still not is, so what we set as our aim was provide an infrastructure for

45
00:07:27,833 --> 00:07:37,333
large scale genetic studies in nephrology by facilitating network-structured collaboration between existing cohorts to optimize the use of

46
00:07:37,333 --> 00:07:47,699
available resources; second, facilitate prospective data-harmonization for future studies; and finally, establish multidisciplinary

47
00:07:47,700 --> 00:07:59,733
collaboration through creating an environment where you can keep up with rapid developments in the field. And luckily, we acquired EU funding

48
00:07:59,733 --> 00:08:10,133
for the GENECURE projects—2.3 million—on the topic “Novel Approaches to Identify Genetic Determinants of Accelerated Atherosclerosis in

49
00:08:10,133 --> 00:08:18,333
Kidney Disease.” We had nine partners in seven EU countries and it was a three-year project that was slightly extended. So this was the

50
00:08:18,333 --> 00:08:30,899
GENECURE project, and for the first time, we had money, the GENECURE had as its aims to provide infrastructure by setting up a phenotype bank, a

51
00:08:30,900 --> 00:08:41,700
DNA bank and web-based access, and providing statistical expertise. We also wanted to do some science, identify risk genes—hypothesis-free

52
00:08:41,700 --> 00:08:51,333
and hypothesis-driven—but also use this as a learning process for collaborative studies and data sharing at the sites that we were not yet

53
00:08:51,333 --> 00:09:01,966
familiar with. So, feed the knowledge of the study and all the things that we did wrong into infrastructure building. Actually, with the project

54
00:09:01,966 --> 00:09:13,332
finished, we succeeded in most of it, only we did not succeed in establishing a DNA bank because of several practical, legal, and ethical issues. In

55
00:09:13,333 --> 00:09:24,833
some countries it’s not allowed to ship DNA, for instance. We also did not succeed in really a web-based depository for data sharing. There

56
00:09:24,833 --> 00:09:35,599
were practical issues, there were issues of ethical…there were ethical problems and also intellectual property issues. But nevertheless, we

57
00:09:35,600 --> 00:09:46,366
could establish infrastructure, summarized here, which I could call the GENECURE legacy, and this infrastructure is not only for EU partners, it’s

58
00:09:46,366 --> 00:09:56,866
open to everybody, so also to you. First, we established an online Renal Biobank Catalogue which gives an overview of existing biobanks of

59
00:09:56,866 --> 00:10:05,932
renal cohorts that have clinical data and DNA available and there’s no more than 20 cohorts included with covering the full range of CKD,

60
00:10:05,933 --> 00:10:15,599
hemodialysis, and transplantation. So, this is an online access tool to find cohorts for collaboration and replication. It’s a data catalogue

61
00:10:15,600 --> 00:10:28,633
only and no data sharing yet. We could develop this very efficiently by collaboration with the P3G Project in Canada because they are experts in

62
00:10:28,633 --> 00:10:39,766
this field—I’ll show that in a minute—and we did this together with them. The second tool that we developed was an Online Renal DataSHaPER

63
00:10:39,766 --> 00:10:51,399
which, essentially, is a minimal dataset which is a harmonization tool for new renal biobanks to facilitate the sharing of data, because everybody

64
00:10:51,400 --> 00:11:06,433
who has tried to collaborate between cohorts knows how annoying and time-consuming it is when datasets are not harmonized. This is what it

65
00:11:06,433 --> 00:11:18,466
looks like. If you go to the P3G website, you’ll find the Biobank Catalogue that was originally developed for general population cohorts, and the

66
00:11:18,466 --> 00:11:27,766
Renal Catalogue was actually the first of the disease-based P3G catalogues. So, you can go through the website and click and then find the

67
00:11:27,766 --> 00:11:45,532
Renal Biobank Catalogue, and you can also join it. So, it’s easy and convenient. The second tool, the Online Renal DataSHaPER, again, was developed

68
00:11:45,533 --> 00:11:55,099
by P3G originally for general population cohorts and then, together with us, they developed the Renal DataSHaPER, minimal datasets for CKD,

69
00:11:55,100 --> 00:12:08,600
hemodialysis, peritoneal dialysis, and transplantation; again, by the same website. So, we now have these infrastructural tools for

70
00:12:08,600 --> 00:12:19,733
facilitating exchange and data sharing and we can upscale studies and form larger studies and it’s open for use by the renal community, and this

71
00:12:19,733 --> 00:12:30,499
is Marion Verduijn, who did most of the work. Of course, these are very simple infrastructural tools and much more has to be done, and the

72
00:12:30,500 --> 00:12:41,033
infrastructural development is also still ongoing in terms of refinement of data harmonization, increasing interoperability of datasets, legal

73
00:12:41,033 --> 00:12:51,433
issues, ICT for data sharing, etc., and this is now being done by the EU-project BIO-SHARE which is disease-overriding, not only kidney disease but

74
00:12:51,433 --> 00:13:03,033
at least including kidney disease, so that’s good for us; also diabetes is in it. So, this is ongoing. Now, what’s good is also that during the

75
00:13:03,033 --> 00:13:13,066
development of the Renal DataSHaPER we were also in the process of establishing two large CKD cohorts in Europe: the PSI BIND-NL cohort in the

76
00:13:13,066 --> 00:13:24,232
Netherlands, and the German CKD cohort. So, we set up these datasets according to the form of the DataSHaPER and it’s now also being adopted

77
00:13:24,233 --> 00:13:34,766
by a third, the CKD-REIN in France, so it’s operational and in the future we will have…this will facilitate a lot of the work that are planning to

78
00:13:34,766 --> 00:13:46,799
do. So now, about these large CKD cohorts because I was asked to tell you something about cohorts and resources, but I thought it would be

79
00:13:46,800 --> 00:13:57,333
very boring to give a long list of cohorts, so I just picked out a few ones that I’m very familiar with and that also offer prospects and also illustrate

80
00:13:57,333 --> 00:14:06,866
some of the concepts that might be of relevance for you also. We now have three large CKD cohorts in Europe that are based on routine

81
00:14:06,866 --> 00:14:18,099
nephrology care; routine in terms of dedicated nephrology care. Thanks to the DataSHaPER, we have a more or less comparable data structure

82
00:14:18,100 --> 00:14:29,200
and minimal data set; it’s a CKD-REIN. The target was 3,000—start to include now—and we’ll have a follow-up for 5 years, which is, they are

83
00:14:29,200 --> 00:14:39,566
located in France. The German CKD is 5,000; inclusion will be completed by April and we’ll follow them for 10 years, and there is the

84
00:14:39,566 --> 00:14:49,732
BIND-NL cohort, which is coordinated by me. We now have over 2,000 patients included and this is slightly different from the other two because we

85
00:14:49,733 --> 00:14:58,966
will not stop the inclusion; we will start to include new patients and the follow-up will also continue. These three cohorts have a strategic alliance for

86
00:14:58,966 --> 00:15:07,832
independent replication and data sharing, upscaling, and of course, it’s all long-term investments, but I think this will be a tremendously

87
00:15:07,833 --> 00:15:20,099
important resource in the future. To give some more detail on our own cohort…so as I said, it’s based on routine nephrology care of the eight

88
00:15:20,100 --> 00:15:30,566
UMCs across the Netherlands and all CKD patients not on dialysis are eligible. The decision model is consensus, of which I’m responsible,

89
00:15:30,566 --> 00:15:41,266
and what makes it easier is that the rest is all men so that makes it easier for me to be the one who is responsible of the consensus, actually. But,

90
00:15:41,266 --> 00:15:51,832
we have a generic minimal data set, biobank and DNA bank, and we have enrichment profiles for APKD and primary and secondary glomerular

91
00:15:51,833 --> 00:16:03,966
diseases. This is good for science but it has also a financial background because it’s routine care, and you do not usually do a CRP in a patient with

92
00:16:03,966 --> 00:16:16,432
diabetic nephropathy of kidney size in a patient with diabetic nephropathy, but you do so in APKD. So, we can budget this patient care and

93
00:16:16,433 --> 00:16:29,666
nevertheless have a nice and dedicated dataset. The inclusion I already mentioned; it will continue as part of routine care. The study description will

94
00:16:29,666 --> 00:16:39,299
appear shortly, and that’s important because the discussion of, let’s say, controls—healthy controls—was discussed also here with

95
00:16:39,300 --> 00:16:44,933
collaboration with LifeLines/PREVEND which has general population cohorts in Groningen and the northern Netherlands for general population

96
00:16:44,933 --> 00:16:58,933
controls, and we are very much open to collaboration with cohorts abroad. Now I switch to the general population cohorts. I mentioned

97
00:16:58,933 --> 00:17:11,566
PREVEND already. PREVEND has a follow-up of 15 years now and was a cohort enriched for albuminuria because the aim was to detect and

98
00:17:11,566 --> 00:17:22,432
determine incidence of chronic kidney disease. We still have 6,000 patients in the cohorts but it turned out, after 15 years of experience, that

99
00:17:22,433 --> 00:17:34,566
starting off with 8,500 healthy subjects, you end up with surprisingly little patients that need a nephrologist. So that’s good news, but not for

100
00:17:34,566 --> 00:17:43,032
these studies, of course, because we are nephrologists who want to learn something about those patients that have a true risk ending up

101
00:17:43,033 --> 00:17:50,899
with a nephrologist. So now, we have something even better, which is LifeLines, and this is a tremendous cohort. This is again, the northern

102
00:17:50,900 --> 00:18:03,000
Netherlands. The area is relatively rural, as I said already, founder characteristics, a million-and-a-half people, and we are now sampling this

103
00:18:03,000 --> 00:18:13,400
general population by general population practices and regional health centers into a large data bio DNA bank and the target is to sample

104
00:18:13,400 --> 00:18:24,166
10% of the population, which amounts to 165,000 participants. And what might be especially interesting for this audience is that this is a 3-

105
00:18:24,166 --> 00:18:36,766
generation design starting with the middle-aged and finding the father or the mother and finding children; so, 3-generation designs. We have over

106
00:18:36,766 --> 00:18:49,199
80,000 included now, also follow-up in the first included patients and GWAS available in over 11,000 and we will continue with GWAS; we

107
00:18:49,200 --> 00:19:07,100
have funding for 20,000. So, this is a very unique resource and it contains a tremendously rich phenotype because this is multi-disease/multi-

108
00:19:07,100 --> 00:19:19,266
domain. It has a rich list of questionnaires, many domains, family histories, quality of life, lifestyle and health behavior, drugs, eating habits,

109
00:19:19,266 --> 00:19:29,632
psychological datasets, psychological questionnaires, a lot of physical measurements with a really exact phenotype, including blood

110
00:19:29,633 --> 00:19:42,866
pressure, ankle-brachial index and cognition tests. We have biochemistry. Of course, we have the creatinine. We have a huge biobank, we have

111
00:19:42,866 --> 00:19:53,532
an event registration and we have the 24-hour urine collection and a first morning void. I marked the 24-hour urine collection in red because I think

112
00:19:53,533 --> 00:20:05,266
this is a particularly rich resource, if you are able to instruct your patients to collect properly, and that is because a lot of nutritional factors can be

113
00:20:05,266 --> 00:20:17,632
derived from those; I’ll show you that in a minute. By now, this is the largest collection of 24-hour urine collection in the world. So taken together,

114
00:20:17,633 --> 00:20:29,499
LifeLines, which is very open to collaboration and the collaboration is free when it applies to GWAS and for other studies a small fee is requested, it

115
00:20:29,500 --> 00:20:46,500
is a superior design for genetic studies, it’s very strong in nutritional data, it’s strong for renal data, it’s strong for diverse renal phenotypes. Now,

116
00:20:46,500 --> 00:20:59,333
after this brief shift to the cohorts…to those phenotypes because what phenotype, what data, do we need to have impact on clinical

117
00:20:59,333 --> 00:21:13,533
nephrology? And this was raised in discussion after Linda Kao’s talk already, and now we’ll go back to the clinic of nephrology. This shows you

118
00:21:13,533 --> 00:21:24,666
a cartoon that depicts lifetime risk in CKD, and what it shows you, essentially—this is overall risk in arbitrary units and this is the course of

119
00:21:24,666 --> 00:21:32,132
kidney disease—you start with common risk factors equal to the general population, then you get uremia-related risk factors, then you get

120
00:21:32,133 --> 00:21:41,533
common to dialysis, and treatment-related factors are superimposed, but your uremia-related risk decreases and you get transplanted and your

121
00:21:41,533 --> 00:21:52,266
risk profile changes again. What this essentially says is that driving forces behind the pathology change over time. So, that means if you have the

122
00:21:52,266 --> 00:21:59,866
cut-off here, you have a very different gene environmental profile than when you have the cut-off here, and to make things even more

123
00:21:59,866 --> 00:22:10,666
complicated, is the proportion of patient surviving. Of patients with CKD class III, 90% die before they enter into dialysis. We do know that people

124
00:22:10,666 --> 00:22:18,666
in dialysis have a very tremendously elevated cardiovascular risk and there are definitely genetic risk factors into that, but to end up in

125
00:22:18,666 --> 00:22:28,199
dialysis, they survived—this increased—cardiovascular risk during the preceding period. So, this makes it extremely complicated and this is

126
00:22:28,200 --> 00:22:38,933
also a setting where you can expect many contradictory results. So, what do we need to generate data that will impact the clinical

127
00:22:38,933 --> 00:22:48,466
nephrology? We need a longitudinal perspective; not only a longitudinal perspective, we need longitudinal data and we need those data over a

128
00:22:48,466 --> 00:22:56,999
duration of follow-up that’s relevant to the natural course of renal disease, which is typically decades. So, we need high-quality phenotype

129
00:22:57,000 --> 00:23:09,733
data and we also need the data that cover the complexity of the process. And I’ll now show you something on the refinement of phenotype

130
00:23:09,733 --> 00:23:19,599
because a very detailed phenotype, of course, conflicts with sample size, but at the same time it’s the best long-term investment you can make.

131
00:23:19,600 --> 00:23:31,600
To make it a little bit provocative, quick and dirty phenotypes will lead us nowhere, but that’s probably too provocative. This was a phenotype

132
00:23:31,600 --> 00:23:42,733
that brought my group a Lancet paper: rate of renal function loss dependent on ACE genotype—DD genotype—more rapid rate of renal function

133
00:23:42,733 --> 00:23:53,866
loss during therapy and it was interpreted at a time of therapy resistance; rate of renal function loss. But everybody who looks at this data sees

134
00:23:53,866 --> 00:24:07,132
that there’s something fishy; the starting point is different. The DD started with a lower GFR than the other genotype groups. This was through

135
00:24:07,133 --> 00:24:17,266
GFR, by the way, so these were very hard and solid data, very detailed follow-up, but we didn’t trust our data or we didn’t trust our interpretation,

136
00:24:17,266 --> 00:24:28,399
actually. And then we went back to the patient records and now we have this advantage of patients that patients are not very mobile—they

137
00:24:28,400 --> 00:24:36,400
stick to their same region where they live—and we found—and these are creatinine data—we found back the data over the preceding four

138
00:24:36,400 --> 00:24:45,200
years and now the picture is totally different because before they were put on treatment, the DDs had a more rapid rate of renal function loss

139
00:24:45,200 --> 00:24:58,333
and they were actually the only ones who had a treatment benefit; totally different. Actually, the PIs on the first paper were very ashamed when we

140
00:24:58,333 --> 00:25:08,399
found out this, so this is why there are 12 years between the 2 papers, but it shows you that you have to be very critical on your phenotype. When

141
00:25:08,400 --> 00:25:14,733
the phenotype is rate of renal function loss, your patients have to be certified for rate of renal function loss and you have to do effort to

142
00:25:14,733 --> 00:25:25,999
document rate of renal function loss. The other thing we did…and these are all association data but I include them to show the concepts because

143
00:25:26,000 --> 00:25:34,266
they might equally apply to GWAS, because that’s also association. The issue on the former slides was response to [---], but everybody, every

144
00:25:34,266 --> 00:25:46,032
nephrologist who has treated a patient with [---], should know that the main determinant of the response of [---] is sodium intake and we did

145
00:25:46,033 --> 00:25:55,633
several studies on it. But this study shows, because from that concept we started looking at the response to [---] dependent on ACE genotype

146
00:25:55,633 --> 00:26:02,699
and dependent on urinary sodium excretion. Apart from our own data there were very few groups that documented it, so we went to

147
00:26:02,700 --> 00:26:11,166
Austrian colleagues and we found that, indeed, this antiproteinuric response was heavily dependent on sodium, whereas in iACE, it was

148
00:26:11,166 --> 00:26:20,899
not, but what was not in the paper—sorry for that—here were the Austrians with the ACE bind and all the sodium consumed and here were the

149
00:26:20,900 --> 00:26:30,833
Dutch, and it’s cross-sectional with sodium intake [---] cross-sectional later. Years later, we succeeded in prospectively confirming that, now

150
00:26:30,833 --> 00:26:40,499
published in the Journal of Hypertension. So, there’s a strong environmental component to this phenotype, which is sodium status, and actually,

151
00:26:40,500 --> 00:26:51,566
there were 12 years between the 2 studies. Why were we so persistent? Because we had these data showing yet another phenotype of the ACE

152
00:26:51,566 --> 00:27:03,699
ID genotype blood pressure response to Ang1, studied in the same subjects two times—low sodium and high sodium. With high sodium you’ll

153
00:27:03,700 --> 00:27:14,966
find the phenotype that you anticipate. When you have high ACE then the more pronounced the response to Ang1 DD, ID and II. But when you put

154
00:27:14,966 --> 00:27:25,432
the same subjects on low sodium, the phenotype is fully abolished. All phenotypes in the ACE ID genotype and involved that I studied are

155
00:27:25,433 --> 00:27:33,499
abolished by low sodium diets, but sodium, I think, is hardly, if ever, consistently considered in studies on ACE genotypes and everybody lost

156
00:27:33,500 --> 00:27:44,433
interest. That might be justified, but nevertheless, it illustrates something very important and that is: if you neglect relevant environmental factors, this

157
00:27:44,433 --> 00:27:56,099
is a source of conflicting results and also nihilism because nobody’s interested in the genotype. But it’s also—and much more important—it’s a missed

158
00:27:56,100 --> 00:28:09,900
opportunity to identify intervention strategies by targeting this permissive environmental factor. I think that’s a serious issue. The other thing about

159
00:28:09,900 --> 00:28:19,900
refined phenotyping, properly documenting your environmental factors, comes from a study that we did in GeneCure in the Necosad dialysis

160
00:28:19,900 --> 00:28:30,033
population replicated in Swedish dialysis cohorts and it relates to the fact that if you have signs of inflammation as a dialysis patient, your mortality

161
00:28:30,033 --> 00:28:44,299
tends to be very high, but with a breakup by inflammation, CRP above 10 and the CCR5 deletion to 32 genotype, we found that this

162
00:28:44,300 --> 00:28:58,566
particular genotype protects against inflammation-associated mortality, and this is really a relevant difference—40% survival and 60% survival—and

163
00:28:58,566 --> 00:29:08,566
this is complemented by in vitro data showing that this mutation is associated with absence of the receptor on the inflammatory cells that blunt the

164
00:29:08,566 --> 00:29:19,832
inflammatory response and shift of the Th2/Th1 balance to Th2. I show this for the conceptual things that we can learn. It shows the potency of

165
00:29:19,833 --> 00:29:30,433
gene-environment analysis as a dissection tool, because without…because, let’s say, the genetic study tells us that the CCR5 is important in

166
00:29:30,433 --> 00:29:38,099
inflammation-associated mortality in dialysis patients. It shows the importance of clinical knowledge where we find phenotyping and it

167
00:29:38,100 --> 00:29:47,433
could guide personalized medicine. A CRR5 blocker in inflamed dialysis patients? There is a CCR5 blocker around; it’s being used in HIV

168
00:29:47,433 --> 00:29:58,299
patients to prevent virus entering into cells. So essentially, this is about feeding clinical knowledge into genetic studies. I’ve shown this

169
00:29:58,300 --> 00:30:07,233
for association studies; we should do it for gene, also. But will this lead to datasets? Because what I’m essentially saying is: we need all data

170
00:30:07,233 --> 00:30:17,366
and we need them longitudinally, so we want everything. So, will this lead to datasets that are simply not affordable when we want them? Well,

171
00:30:17,366 --> 00:30:29,432
there’s a step we see around it, because in datasets from dedicated clinical care, most clinically relevant factors will be available as they

172
00:30:29,433 --> 00:30:38,499
are collected as being part of best practice. You do it anyway. So just biobank it, make a registry, and you have the data. So, infrastructure for

173
00:30:38,500 --> 00:30:46,766
genetic studies, based on cohorts in a dedicated and protocolized clinical routine, might prove an affordable and relevant strategy to move

174
00:30:46,766 --> 00:30:59,599
forward, and I think we are trying to do that know with PSI BIND Initiative. Well, this is more or less a summary of my journey through the world of

175
00:30:59,600 --> 00:31:11,533
biobanking, which was exciting, confusing sometimes, but it has provided us, let’s say, with knowledge and let’s say with interests that are

176
00:31:11,533 --> 00:31:21,199
together open to collaboration, also, with cohorts from abroad. We are very open to collaboration and a lot of this work I could do thanks to all my

177
00:31:21,200 --> 00:31:31,666
GeneCure friends. Thank you. ERWIN BOTTINGER: Thank you for providing

178
00:31:31,666 --> 00:31:44,799
such an insightful talk about bringing clinical care data to the research world. One question that I have for you—this is Erwin Bottinger, Mt. Sinai,

179
00:31:44,800 --> 00:31:56,866
being also a biobanker—is to what extent do you have your clinical data represented in electronic health records, which is clearly a tremendous

180
00:31:56,866 --> 00:32:05,832
advantage in terms of knowledge for presentation and data mining for the kind of work that you are undertaking.

181
00:32:05,833 --> 00:32:18,166
GERJAN NAVIS: Actually, it is a very important question. This PSI BIND-NL grant that we got was aimed just to provide such an electronic-based

182
00:32:18,166 --> 00:32:30,299
registry of the clinical care data. This proved an enormous effort. Actually, this was a four-year project. It was extended to a five-year project

183
00:32:30,300 --> 00:32:39,166
because there were not only technical hurdles but even legal hurdles, so the law in the Netherlands was changed to be able to do so,

184
00:32:39,166 --> 00:32:49,032
and actually even by now we have to support an electronic system by hand because it’s not completely ready, but let’s say, we are optimistic

185
00:32:49,033 --> 00:33:01,166
that in a year from now or so it will be completely electronic, but it was very difficult. In the Netherlands this was all in the context of the aim

186
00:33:01,166 --> 00:33:13,166
of a national electronic patient data set that has actually now been discarded by the government because there are too much technical hurdles,

187
00:33:13,166 --> 00:33:27,632
also privacy hurdles, but this PSI Initiative is meant to pave the way and also to, let’s say, to make all the errors that have to be resolved for

188
00:33:27,633 --> 00:33:34,966
better EPD—electronic patient data—record system later on.

189
00:33:34,966 --> 00:33:46,566
ERWIN BOTTINGER: A follow-up question if I may. You’re consenting these individuals to participate in these longitudinal projects and, you know, in

190
00:33:46,566 --> 00:33:55,432
the United States there are tremendous issues in terms of the uniformity of the consent, and in particular, the consent explicitly allowing genetic

191
00:33:55,433 --> 00:34:09,399
testing such as GWAS and in depositing these data in public data domains. Do you have consent structure that is basically permitting this?

192
00:34:09,400 --> 00:34:20,066
GERJAN NAVIS: We do this all under informed consent, so it’s under the Dutch law you should have consent for this. So how it goes in practice

193
00:34:20,066 --> 00:34:27,866
is, the patient comes to the clinic and we see this is a patient that will stay because he’s got kidney disease, so on the second or third visit we ask

194
00:34:27,866 --> 00:34:40,932
informed consent. Almost two in three give consent and the rest is, let’s say, not being followed this way. We will test in retrospect for

195
00:34:40,933 --> 00:34:56,066
the representativity of the cohorts that does consent. But yes, we are legally…we should have this informed consent and it’s also the only

196
00:34:56,066 --> 00:35:07,599
real difference between routine care and this current databanking/biobanking status that will persist because we need it. So, we need

197
00:35:07,600 --> 00:35:18,533
personal for it because this informed consent form is terribly long, including terrifying statements on genetics, so I’m actually amazed

198
00:35:18,533 --> 00:35:31,266
that it does not scare off all our patients. But let’s say, if we devote a lot of time to it by a nurse practitioner and explain about it, it still works. So

199
00:35:31,266 --> 00:35:40,532
my worry is: how we can keep up these nurse practitioners by the time that we are supposed to do it all over clinical care?

200
00:35:40,533 --> 00:35:47,499
ANDREY SHAW: Hi. Andrey Shaw, from Washington University. So, this is just a general question for these large cohorts of CKD patients:

201
00:35:47,500 --> 00:35:55,233
how easy is it to break these out into pathological descriptors like FSGS or NPGN, etc.?

202
00:35:55,233 --> 00:36:05,466
GERJAN NAVIS: Let’s say for this PSI BIND cohort, this is the nephrology care of the University Center. So, we get a diagnosis based

203
00:36:05,466 --> 00:36:13,066
on our routine care. If we need a renal biopsy, we do a renal biopsy. That’s not because of the cohort, that’s because of clinical care. If you

204
00:36:13,066 --> 00:36:21,599
have, let’s say, 5 grams of proteinuria and erythrocytes in the urinalysis, we will do a renal biopsy and have a diagnosis. We do have,

205
00:36:21,600 --> 00:36:31,333
however, a big category that we call CKD not otherwise specified, where we have no good clinical reason to do a biopsy.

206
00:36:31,333 --> 00:36:35,599
ANDREY SHAW: But these datasets can actually…you could extract just that data.

207
00:36:35,600 --> 00:36:46,333
GERJAN NAVIS: Yeah. Actually, the way we have organized it…and I think that’s a big difference with the way we made data sets

208
00:36:46,333 --> 00:36:53,866
formerly. Formerly when you wanted a data set and you had the money, you hired a bunch of ladies typing data into a database on it, usually

209
00:36:53,866 --> 00:37:03,866
Lexo[---]. What we now have made is an ICT application that extracts the dataset that we want from the so-called electronic patient records that

210
00:37:03,866 --> 00:37:15,866
are not ready yet, but that’s a strategy and it works with some, let’s say, hands-on support.

211
00:37:15,866 --> 00:37:26,932
WENDY HOY: Thank you for your lovely presentation. My name is Wendy Hoy and I’m from Queensland in Australia. I’m just alerting you all

212
00:37:26,933 --> 00:37:37,466
that we have begun, in Queensland, a similar process that you describe; it’s now in its third year. We call it CKD Queensland. Now,

213
00:37:37,466 --> 00:37:49,799
Queensland is a population of 4.6 million which is very much multi-ethnic, as is all of the Australian population, and it has one health provider, which

214
00:37:49,800 --> 00:37:59,800
is Queensland Health. All the nephrologists in Queensland are largely employees of Queensland Health, and we have set up a

215
00:37:59,800 --> 00:38:11,933
collaboration, which is universal, with all the renal service provider in Queensland to inform their patients, when they see them for their first

216
00:38:11,933 --> 00:38:22,266
consultation in the renal practice system, of the opportunity to join a cohort, become part of a registry which is longitudinal, and to contribute to

217
00:38:22,266 --> 00:38:33,266
a biobank. And so far, we’ve had no refusals for the first 1,000 patients over the first 6 months. We figure there are about 15,000 patients in that

218
00:38:33,266 --> 00:38:43,266
public renal system—there would be a few more in the private system—and we are very interested in collaboration with the international

219
00:38:43,266 --> 00:38:53,066
programs that are watching progression and looking at genetics, amongst other things. We have described this in, I imagine, it’s the same

220
00:38:53,066 --> 00:39:01,399
NDT volume that you have written for, the World Kidney Day this year, and look forward to hearing about the other programs that are happening

221
00:39:01,400 --> 00:39:04,866
internationally. Thank you.

GERJAN NAVIS: Well, thank you very much for

222
00:39:04,866 --> 00:39:10,399
that.

JEFFREY KOPP: You may have said this and I

223
00:39:10,400 --> 00:39:14,066
may have missed it. What percent of your patients give consent?

224
00:39:14,066 --> 00:39:16,966
GERJAN NAVIS: Approximately two-thirds. JEFFREY KOPP: Two-thirds.

225
00:39:16,966 --> 00:39:24,766
GERJAN NAVIS: And let’s say between…so that might seem…at least it’s much less than in north Queensland, apparently, but we do get phone

226
00:39:24,766 --> 00:39:33,132
calls from patients from non-university hospitals asking for a possibility to join. Actually, what scares them off

227
00:39:33,133 --> 00:39:36,599
is the terrible informed consent form. JEFFREY KOPP: Check Wendy’s consent. Is

228
00:39:36,600 --> 00:39:47,200
there any…are there differences that you can see between the third who say no and the two-thirds who say yes? Demographic variables,

229
00:39:47,200 --> 00:39:51,800
medical diagnosis… GERJAN NAVIS: We have not formally evaluated

230
00:39:51,800 --> 00:40:02,866
that for the moment but the impression is, no. AFSHIN PARSA: Hi. Afshin Parsa from University

231
00:40:02,866 --> 00:40:09,066
of Maryland. I very much enjoyed your talk and especially the points you made about a G-by-E environment. I think that’s an area that’s

232
00:40:09,066 --> 00:40:17,732
understudied. Regarding the databases that you have, I was wondering how many are existing now that would have longitudinal phenotypes

233
00:40:17,733 --> 00:40:26,466
with a genotype that’s already existent. So for example, I know the German CK and the CKD line still don’t have genotyping and hopefully in a few

234
00:40:26,466 --> 00:40:33,132
years, but currently within the Netherlands and LifeLines, do you have an idea of how many are longitudinal renal phenotypes?

235
00:40:33,133 --> 00:40:39,933
GERJAN NAVIS: Well, at least for LifeLine and PREVEND we do have them, but as I emphasize, those are general population cohorts, and

236
00:40:39,933 --> 00:40:48,933
unfortunately very few end up in the hands of nephrologists. But of course, PREVEND has the longest follow-up but…well, if mentioned, the

237
00:40:48,933 --> 00:40:59,866
smallest number of true, real patients and not CKD3. There are a lot with CKD3 but they don’t end up with a nephrologist. LifeLines has now a

238
00:40:59,866 --> 00:41:14,332
follow-up of 5 years and has a much larger number, but there’s only in LifeLines 11,000 now and it will be 20,000 in 6 months, but only 2

239
00:41:14,333 --> 00:41:26,166
sampling points. So, there is a problem, because well, all these issues that we discussed on what is progressive CKD if you have only creatinine?

240
00:41:26,166 --> 00:41:40,099
So, the Germans indeed they have the DNA, but they don’t have the genotyping. We have the DNA and we have no GWAS in the PSI BIND-NL, so

241
00:41:40,100 --> 00:41:48,966
we still need the money. So, if anyone in the audience has, let’s say, a grant proposal and needs us as cohorts, here am I ready to

242
00:41:48,966 --> 00:41:58,232
collaborate. So for the moment, we will have to do with LifeLines and PREVEND. The real CKD cohorts, generally, have no GWAS.

243
00:41:58,233 --> 00:42:05,099
AFSHIN PARSA: And then, just in terms of the refined phenotype, that’s something to look forward to and I think you were alluding to. We

244
00:42:05,100 --> 00:42:13,033
just looked at our primary look at the CRIC GWAS, so that’s the Chronic Renal Insufficiency Cohort of 4,000 people. Even when we stratified by the

245
00:42:13,033 --> 00:42:20,566
different subgroups, we do have within subgroups of 700-800 people genome-wide significant hits when we have refined

246
00:42:20,566 --> 00:42:29,499
phenotypes of longitudinal follow-up in diabetics and non-diabetics. So, I think that approach might continue to give results that are stronger than

247
00:42:29,500 --> 00:42:34,633
what you saw in the population-based cohorts. GERJAN NAVIS: Yeah. I think that’s the way to

248
00:42:34,633 --> 00:42:41,033
go. We need cohorts with, let’s say, real CKD patient with a frequent follow-up sampling to be sure that these are these are the very people

249
00:42:41,033 --> 00:42:52,566
who lose their renal function. So, that was why I also showed those data from the ACE phenotypes in the Lancet in ’96. Those slope data

250
00:42:52,566 --> 00:43:01,499
were obtained with four sampling points per year and we had a decent slope because they had true CKD progression in disease, and that would

251
00:43:01,500 --> 00:43:12,433
be the informative cohorts for, let’s say, the population that the nephrologists care for. So I think, actually, that if we look too much to the

252
00:43:12,433 --> 00:43:17,599
general population, we are answering quite different questions than

253
00:43:17,600 --> 00:43:21,466
when we stuck to the patients that are in our nephrology hands anyway.




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