Quantitative Morphology in Kidney Research
February 13-14, 2012 Conference Videos

Panel Discussion: Is it Possible to Accurately Measure Podocyte Number in Clinical Biopsy Material?
Kevin Lemley (Chair), University of Southern California
Behzad Najafian, University of Washington
John Basgen, Charles R. Drew University of Medicine and Science
Jennifer Weil, NIDDK
Susanne Nicholas, University of California, Los Angeles

Video Transcript

1
00:00:00,000 --> 00:00:08,366
JEFFREY KOPP: These one-on-one and one-on-two and three groups of people talking are obviously a large part of why we’re here, so I

2
00:00:08,366 --> 00:00:18,432
hate to cut off these side-bar discussions. And actually, this next session which Kevin is going to lead is intended to continue this, so I hope the

3
00:00:18,433 --> 00:00:25,199
same conversations that were just going on potentially can, and some of the questions that have been asked, can now be directed to our

4
00:00:25,200 --> 00:00:30,966
four participants, plus Kevin.
KEVIN LEMLEY: Kind of in the model that we had

5
00:00:30,966 --> 00:00:39,866
yesterday, everybody’s been warned that if they have a point they want to make they can have a few slides on it, so why don’t we start at the

6
00:00:39,866 --> 00:00:49,299
end? Susanne, did you have any slides? And this is to address, as you see, the topic is: can we, in fact, after hearing everything you’ve heard,

7
00:00:49,300 --> 00:00:57,166
give accurate measures or estimates of podocyte numbers in clinical biopsy material that may be limiting in terms of volume?

8
00:00:57,166 --> 00:01:04,699
SUSANNE NICHOLAS: Good morning. I had to look at my watch to make sure it’s still morning. It’s been such a fruitful morning already of

9
00:01:04,700 --> 00:01:20,066
wonderful information. In terms of addressing the issues that we are faced with today and over the last couple of days, my motivation to learn and to

10
00:01:20,066 --> 00:01:33,432
use stereologic methods really comes from the perspective of the translational scientist. My interest is in identifying novel treatments for

11
00:01:33,433 --> 00:01:46,999
diabetic nephropathy and other chronic kidney diseases and it’s the hope that the data that we obtain from our animal studies will help in better

12
00:01:47,000 --> 00:01:59,566
clinical study design. So, I do endorse and support and appreciate the need to obtain accurate measures of podocyte number as well

13
00:01:59,566 --> 00:02:10,132
as other renal structures, and I’d like to illustrate that with an example. So this is—and we only have three slides so that’s about all I have—this

14
00:02:10,133 --> 00:02:19,633
is an example of a study where we identified a therapy that caused regression of glomerulosclerosis in a diabetic model and this

15
00:02:19,633 --> 00:02:27,966
was marked by a dose-dependent reduction in albumin excretion rate up to 60% and what you can see from these PS stains is that there was a

16
00:02:27,966 --> 00:02:36,299
significant reduction in the PS-positive staining material with the active drug versus the inactive drug that was accompanied by a decrease in

17
00:02:36,300 --> 00:02:53,100
matrix proteins, TGF-beta, PAI-1, NOX4, all of the markers that we associated with an increase in diabetic kidney disease. Using the

18
00:02:53,100 --> 00:03:01,233
fractionator-disector design-based method we were able to count podocytes and show there was a decrease in podocyte density. But this

19
00:03:01,233 --> 00:03:11,833
decrease in podocyte density was not due to decrease in podocyte number but is actually because the therapy prevented the diabetic

20
00:03:11,833 --> 00:03:21,433
kidney disease-induced increase in glomerular volume. And so in fact, in these aged diabetic animals with significant glomerulosclerosis that

21
00:03:21,433 --> 00:03:32,866
we had measured by morphometric analyses, we could see that podocyte number actually did not change. In a more recent study we have

22
00:03:32,866 --> 00:03:45,866
identified a novel antioxidant that has attenuated the kidney swelling and discoloration and deformity that we see in this adenine-induced

23
00:03:45,866 --> 00:03:58,866
model of progressive tubulointerstitial nephropathy, and we see there’s an improvement with the drug we call F1 in the tubulointerstitial

24
00:03:58,866 --> 00:04:09,232
damage as well as in glomerular sclerosis damage. So, this now sets us up to use morphometric analyses to identify changes in

25
00:04:09,233 --> 00:04:22,866
glomerular structure as well as podocyte number and to determine if there’s cross-talk between different cells within the disease. So, using

26
00:04:22,866 --> 00:04:32,566
morphometric analyses can also help us understand some of the mechanisms involved in the disease. And so as we move to designing our

27
00:04:32,566 --> 00:04:41,166
clinical studies then, we appreciate that accurate measure of podocyte number will be necessary and we acknowledge several of the challenges

28
00:04:41,166 --> 00:04:49,499
that have been enumerated over the last couple of days or so. So with that in mind then, I provide you food for thought. I think from our discussions

29
00:04:49,500 --> 00:04:59,600
this morning that many of us or many of you will agree that perhaps one of the things that we may want to do—and some of this may reiterate what

30
00:04:59,600 --> 00:05:09,200
John Bertram said yesterday—is to come up with a consensus or create a consortium that can provide some guidelines for podocyte counting as

31
00:05:09,200 --> 00:05:23,400
well as other measurements within the kidney. There are regulatory steps for other things like biomarkers, etc., before they go to open access

32
00:05:23,400 --> 00:05:34,066
or publication. Maybe this consortium may function as a “central clearinghouse” and provide quality control for every new method that has

33
00:05:34,066 --> 00:05:46,166
been established or is proposed to estimate podocyte number and perhaps this consortium can determine whether there are certain criteria

34
00:05:46,166 --> 00:05:55,999
that these methods should fulfill. Of course, this will depend on the questions that are being asked. We can determine whether these new

35
00:05:56,000 --> 00:06:05,133
methods may need to be validated using design-based methodology and perhaps add a disclaimer or a warning regarding bias and potential

36
00:06:05,133 --> 00:06:16,233
inferences from different protocols that are published. Some of the other things that we do need to do include designing the right studies

37
00:06:16,233 --> 00:06:27,433
such as using pre-implantation biopsies, as some of Kevin’s studies have demonstrated. I mentioned here using whole donor biopsies and

38
00:06:27,433 --> 00:06:41,599
you may gawk at that but it depends on how important we think this really is and perhaps we may be able to satisfy or confirm many of the MRI

39
00:06:41,600 --> 00:06:53,600
approaches to quantification in order to preempt something like this. But importantly, the point that I’d like to make really point at establishing

40
00:06:53,600 --> 00:07:05,766
collaborations: collaborations with labs that can provide the resources, the skill, and the training that are necessary for studies like this that are

41
00:07:05,766 --> 00:07:17,699
design-based; and also to encourage your faculty, young faculty, and technicians to attend workshops that present the principles of

42
00:07:17,700 --> 00:07:29,566
stereology; and probably most importantly, we really need to continue the dialogue. I think this meeting is excellent at continuing this dialogue but

43
00:07:29,566 --> 00:07:38,532
we can’t let it stop here. We were very fortunate a couple of years ago to put on an Inaugural Stereology and Its Application in Kidney Disease

44
00:07:38,533 --> 00:07:49,999
Symposium and that symposium was attended by many of the speakers over the last couple of days, including Wendy Hoy and Behzad and

45
00:07:50,000 --> 00:08:02,733
Kevin, and I submit to you that any of us, any of our institutions, can really put our heads together to continue this dialog under this type of forum.

46
00:08:02,733 --> 00:08:09,566
So with that in mind, I’ll have the next person present their slides.

47
00:08:09,566 --> 00:08:16,766
JENNIFER WEIL: Good morning. I’m Jennifer Weil. I work with Dr. Robert Nelson with the NIDDK in Phoenix, Arizona, where we study diabetic

48
00:08:16,766 --> 00:08:28,299
nephropathy in Pima Indians, and I thought today I would try to give you some practical illustrations of the problems that I encounter. I use a Weibel-

49
00:08:28,300 --> 00:08:39,233
Gomez method along with Kevin Lemley’s guidance and I routinely encounter problems that are not based on equations, not based on

50
00:08:39,233 --> 00:08:48,166
assumptions, but are very practical in nature, and I thought it might be useful to show you some of the practical problems that I encounter every day

51
00:08:48,166 --> 00:08:58,132
and that are not going to be solved by any of the debate that we’ve already had here this morning. So, I’m showing you here an entire glomerulus

52
00:08:58,133 --> 00:09:06,899
from one diabetic Pima Indian and I almost picked this glomerulus at random for the purpose of this presentation because the problems that I show

53
00:09:06,900 --> 00:09:17,233
you here are so widespread, that any glomerulus might have done. So, these people have very large glomeruli, we believe from birth, before they

54
00:09:17,233 --> 00:09:25,533
become diabetic. The glomeruli become bigger as the diabetes progresses, as we’ve talked about this morning, and one of the problems that I

55
00:09:25,533 --> 00:09:34,166
encounter very routinely is that the glomerulus fills the entire Bowman’s space and this leads to lots of problems in interpretation of what is a

56
00:09:34,166 --> 00:09:44,432
podocyte, what is not a podocyte, so I thought maybe we could zoom in and try to understand how the disease and the underlying person in

57
00:09:44,433 --> 00:09:52,433
whom the disease occurs makes these very difficult. So I’m going to address three different areas of this one tuft, starting with number one.

58
00:09:52,433 --> 00:10:02,533
Let’s zoom in. So I’m a little embarrassed because if you all disagree with me, now I’ve gone public with how I made the decision. Using the Weibel-

59
00:10:02,533 --> 00:10:13,266
Gomez method I identify all of the nuclei in the tuft in the cross-section and I label them with a number and what kind of cell I think it is. And so,

60
00:10:13,266 --> 00:10:22,532
here I looked at this along the Bowman’s capsule right along here and I had to decide whether this was a visceral or a parietal epithelial cell, and

61
00:10:22,533 --> 00:10:31,633
moreover I had to decide was this one cell or two cells represented by this one cross-sectional area? You can see that I solved this by calling it

62
00:10:31,633 --> 00:10:41,333
visceral—I decided it was part of the tuft—and I called it just one cell. But I encounter this every day and I have to make a decision about what is

63
00:10:41,333 --> 00:10:50,466
this and maybe there is someone here who could tell me why I’m making the right or the wrong calls, but I don’t know right now whether I am or

64
00:10:50,466 --> 00:11:01,732
I’m not. So, that’s problem number one. Problem number two—and we talked a little bit about this—that podocytes, especially in disease states,

65
00:11:01,733 --> 00:11:12,866
can develop more than one nucleus. So, these are a number of nuclear profiles and my question for you is: how many independent nuclei are

66
00:11:12,866 --> 00:11:23,299
here? How many cells do these profiles represent? This is a thin section in EM and I still can’t really tell how many cells I’m looking at. You

67
00:11:23,300 --> 00:11:34,633
can see, again, I’ll go public and tell you that I decided that there are five profiles and I decided there was…and Mike Mauer said that maybe we

68
00:11:34,633 --> 00:11:41,066
were undercounting these things. I submit that maybe I’m overcounting these things, but I decided that there might be four here and I

69
00:11:41,066 --> 00:11:47,732
decided that the cytoplasm between these two nuclei were continuous enough for me to believe that they were related, but maybe I’m wrong.

70
00:11:47,733 --> 00:11:54,799
Maybe the cytoplasm is continuous here or maybe this nucleus is actually…these are all one cell. Honestly, I don’t know and I have to make a

71
00:11:54,800 --> 00:12:07,533
decision every day: how am I going to count these things? And finally, I’m pretty sure that these cells here are parietal epithelial cells but

72
00:12:07,533 --> 00:12:17,333
they are extending cytoplasm to the tuft. And so, a really critical part of my work is to correlate structure and function. I am looking at albuminuria

73
00:12:17,333 --> 00:12:26,899
and GFR in these subjects, and if I make a mistake and I say these are not podocytes or they’re not contributing to the glomerular filtration

74
00:12:26,900 --> 00:12:39,500
barrier, I’m not going to get good correlation with the amount of albuminuria. In this particular case I

75
00:12:47,233 --> 00:12:49,566
did decide to leave them out. I was not able to sort of jump in and call them cells, but I recognize that by making this decision, I may get poor

76
00:12:49,566 --> 00:12:55,332
correlation with structure and function, which is the whole purpose of my study. So, I appreciate our efforts to come up with better methods but

77
00:12:55,333 --> 00:13:01,399
some of the problems are not going to be solved by what we’ve already discussed here today, and they might be meaningful problems. Thank

78
00:13:01,400 --> 00:13:17,433
you. Dr. Wiggins, you were smiling. Do you have an answer to my problems?

79
00:13:17,433 --> 00:13:25,066
ROGER WIGGINS: [inaudible response].
JENNIFER WEIL: That’s right, can I be replaced? I want to be replaced by an automated system.

80
00:13:25,066 --> 00:13:32,299
AGNES FOGO: Agnes Fogo from Vanderbilt. I think those are beautiful illustrations of conundrums where we have judgments, but I

81
00:13:32,300 --> 00:13:40,300
think the terminology that you used, you were saying, “Is this a visceral epithelial cell or parietal epithelial cell?” and those are actually very good

82
00:13:40,300 --> 00:13:50,800
terms because visceral we should think of as just being something that is touching the tuft, and whether it is a podocyte origin or whether it is a

83
00:13:50,800 --> 00:13:59,166
parietal cell origin is a mechanistic problem but your observation of whether it is on the tuft, that is one that you can say, that this is a visceral

84
00:13:59,166 --> 00:14:09,599
location epithelial cell, and I think that this type of approach will not tell you whether it originated from a transdifferentiated podocyte that

85
00:14:09,600 --> 00:14:20,633
transitioned in situ or a migrating parietal epithelial cell. We have immunohistochemistry evidence confirming earlier studies from [---] and other

86
00:14:20,633 --> 00:14:32,266
people from more than a decade ago, that cells that distinctly are in a visceral location may show loss of podocyte numbers, the de-differentiated

87
00:14:32,266 --> 00:14:39,866
podocytes that Laura Barisoni and Vivette D’Agati first described are, indeed, marking as PECs, but they are visceral; whether they’re podocytes or

88
00:14:39,866 --> 00:14:47,766
not is a different question.
JENNIFER WEIL: I understand and I think that we

89
00:14:47,766 --> 00:14:54,766
have a big problem with nomenclature or what I would almost say is nominalism. How we name something troubles us in terms of its functionality

90
00:14:54,766 --> 00:15:03,199
—the name, the function, the origin—these things are not uniform and makes the discussion much more complicated.

91
00:15:03,200 --> 00:15:12,600
ANDREW RULE: Hi. Andrew Rule, Mayo Clinic. I was just wondering, have you thought of labeling these questionable areas where you don’t know

92
00:15:12,600 --> 00:15:19,033
how many nuclei there are or whether they’re parietal and see what better predicts, what helps us understand physiology better, and maybe that

93
00:15:19,033 --> 00:15:22,333
should guide this somewhat.
JENNIFER WEIL: Yes. I have a medical student

94
00:15:22,333 --> 00:15:30,166
who needed a project and I basically sent him to start to count these kinds of things. I wanted him to start to count parietal cells that were clearly

95
00:15:30,166 --> 00:15:36,532
parietal. I wanted to characterize those parietal cells and I wanted to know how many bridges—if this is a bridge—I also wanted to count the

96
00:15:36,533 --> 00:15:45,233
number of bridges. But as it’s been pointed out by many people here that, without serial sections—Behzad helped me understand this—without

97
00:15:45,233 --> 00:15:54,066
serial sections it’s kind of hard to even know: is this really a bridge that I’m looking at here? It could be just an artifact of the way that the very large

98
00:15:54,066 --> 00:16:01,832
tuft is expanded within Bowman’s space. So I had a student working on it, but these are questions that I think could be better answered in

99
00:16:01,833 --> 00:16:08,433
animal models.
JOHN BERTRAM: Congratulations on your

100
00:16:08,433 --> 00:16:15,266
bravery, anyway, standing up and trying to do that. I don’t know that I might have done that, so I’m very impressed. I had a slide in my talk

101
00:16:15,266 --> 00:16:22,832
yesterday which I took out but it kind of said, step one in trying to count something, whether it’s whatever method we’ve heard about here in the

102
00:16:22,833 --> 00:16:32,433
last day or so, you’ve got to be able to identify it, and I often say if a farmer’s trying to count a sheep or a cow, if he or she can’t tell a sheep

103
00:16:32,433 --> 00:16:41,466
from a cow, he’s kind of got a problem. It gets down to definitions and how you’re going to do that and, you know, as Agnes and others are

104
00:16:41,466 --> 00:16:49,366
saying—parietal, visceral, this positive, that positive, touches the parietal basement membrane, sits on the capillary—in a way you’ve

105
00:16:49,366 --> 00:16:59,999
got to try these definitions and the word “podocyte” is very handy, in a way, but then it’s starting to become a generalization kind of an

106
00:17:00,000 --> 00:17:08,100
issue. So, I guess you’ve got to come up with your definitions and see if they kind of…you can put cells into different categories and stick to it or

107
00:17:08,100 --> 00:17:18,533
change it after, etc.; it’s not straightforward. In terms of if you see three or four or five nuclear profiles and ask the question, “Is that four cells,

108
00:17:18,533 --> 00:17:28,333
three cells, five cells?” How I would try to tackle that, and I don’t know if it’s feasible with your tissue and your archive and what you’re getting,

109
00:17:28,333 --> 00:17:39,733
etc., is I would try to tackle that with confocal optical sections and marking them where you might have six, seven, eight 1-micron sections

110
00:17:39,733 --> 00:17:49,066
through that nucleus and then see if they connect up. But if you don’t have access to that kind of tissue, well that’s impossible, but that’s my first

111
00:17:49,066 --> 00:17:54,499
thought on how I might try to tackle that.
JENNIFER WEIL: I shall send my patients to

112
00:17:54,500 --> 00:17:59,933
Australia.
JOHN BERTRAM: Well, you could send blocks to

113
00:17:59,933 --> 00:18:05,399
Australia.
JENNIFER WEIL: Thank you.

114
00:18:05,400 --> 00:18:21,500
FEMALE: To throw something else out, in other fields and in fields that are far away, often one needs confirmation of something in terms of how

115
00:18:21,500 --> 00:18:34,466
many people judge something. Because the work that’s been described over the last day is so labor-intensive, I suspect it’s rare for two people

116
00:18:34,466 --> 00:18:43,232
to do the counting, but when so much rides on it I thought at least one should bring up the question.

117
00:18:43,233 --> 00:18:53,466
JENNIFER WEIL: Right. So, I will say that we have 120 subjects that have been biopsied and we have 3 glomerular tufts, sometimes more, from

118
00:18:53,466 --> 00:19:03,132
each one. I’ve been at it now for six years and I am the sole counter of cell numbers for that reason, and even so, I have experienced some

119
00:19:03,133 --> 00:19:10,599
drift in my own counting as I’ve gotten smarter and I’ve tried to understand the issues better. So, whether I’m even internally consistent is

120
00:19:10,600 --> 00:19:18,800
worrisome, but you raise very good points. When I’m really in a conundrum I e-mail the picture to Behzad and I try to get some second opinion on

121
00:19:18,800 --> 00:19:26,466
this, but there aren’t enough…I guess my plea…Susanne gave an excellent plea for “where do we go from here?” My plea is that I would like to

122
00:19:26,466 --> 00:19:33,866
know how other people are answering these questions or dealing with them or at least to handle the problems in a consistent way so that

123
00:19:33,866 --> 00:19:38,266
we can pool our data and start to make some sense of this.

124
00:19:38,266 --> 00:19:44,699
FEMALE: I mean, it seems to me akin to other fields when there is some sort of consensus about how to do it.

125
00:19:44,700 --> 00:19:48,766
JENNIFER WEIL: We need consensus. FEMALE: Maybe the data can be posted or the

126
00:19:48,766 --> 00:19:52,866
images can be posted so there can be confirmation.

127
00:19:52,866 --> 00:19:55,866
JENNIFER WEIL: Yes. FEMALE: Because what none of us wants is to

128
00:19:55,866 --> 00:20:00,699
have unintentional individual bias.
JENNIFER WEIL: Absolutely, and I work in

129
00:20:00,700 --> 00:20:12,233
isolation. I work in a lab full of epidemiologists, so they’re out writing computer programs to model the course of nephropathy or of diabetes in these

130
00:20:12,233 --> 00:20:19,133
people and I’m the sole person looking at these images, which is why I have an e-mail relationship with Behzad, but I’d be happy to have

131
00:20:19,133 --> 00:20:25,499
an e-mail relationship with anybody else who wants to look at the pictures or post them in a public place, but I think that’s exactly right. Thank

132
00:20:25,500 --> 00:20:29,766
you so much.
MALE: I would like to make a comment that more

133
00:20:29,766 --> 00:20:34,899
often than not I have no clear answer to Jennifer’s question.

134
00:20:34,900 --> 00:20:50,766
MALE: I’ve got a comment. How often are these…I mean, is this one per glom or two per glom that these are the problems? I guess I don’t need an

135
00:20:50,766 --> 00:21:01,866
answer, but I’m more concerned about: as pathology progresses that then it becomes more difficult to identify the podocyte.

136
00:21:01,866 --> 00:21:10,699
JENNIFER WEIL: Absolutely. Well, if the tuft gets bigger and starts to rub up against the Bowman’s capsule, it definitely becomes more difficult. You

137
00:21:10,700 --> 00:21:19,000
can see from the overview of this particular tuft that there’s a lot of mesangial matrix, a lot of incipient nodularity here and it definitely gets

138
00:21:19,000 --> 00:21:29,666
worse as the disease progresses. So if the question is, “Is podocyte loss a marker of early diabetic glomerular injury?” one thing we could do

139
00:21:29,666 --> 00:21:36,932
is exclude these people entirely, and I think that this problem is probably less bad in the people with less disease, but in terms of understanding

140
00:21:36,933 --> 00:21:46,266
the way that the disease progresses, you lose something by losing this particular individual. But I have problems like this on virtually every image. I

141
00:21:46,266 --> 00:21:53,966
should have prepared better for today’s meeting by accumulating the right images. Over the course of the last couple of months I knew I

142
00:21:53,966 --> 00:22:00,532
would be coming, but in fact right before I left, I thought, “Oh, let me pull just a couple of pictures to show,” and I just picked a recent glom and I

143
00:22:00,533 --> 00:22:10,466
could pick any glom and it was amazing that I could just pick anything and show you the problems; it’s very common.

144
00:22:10,466 --> 00:22:16,232
KEVIN LEMLEY: Perhaps, Wendy, if you could add one more comment and then we could go on.

145
00:22:16,233 --> 00:22:25,033
WENDY HOY: Might I suggest when we are working out how to tackle these tricky problems that we think more collaboratively and that for

146
00:22:25,033 --> 00:22:35,433
people like yourself, Jennifer, who’s been working on these issues for six years, that instead of suggesting that a multiple lab set-up,

147
00:22:35,433 --> 00:22:45,966
sophisticated mechanisms and techniques and equipment for the sort of stereology, for example, that John Bertram does, that indeed we consider

148
00:22:45,966 --> 00:22:57,099
collaborative arrangements of sending tissue blocks to labs such as that to answer these questions—is this one cell or five cells and so

149
00:22:57,100 --> 00:23:05,166
forth—at least to clarify that sort of an issue. Then maybe if one lab has markers that are working well for visceral versus parietal cells or

150
00:23:05,166 --> 00:23:14,566
whatever, that instead of trying to reproduce that in many locations, that we have collaborative arrangements to send them to a few places

151
00:23:14,566 --> 00:23:19,766
where we know does this in a standardized way.

152
00:23:19,766 --> 00:23:25,266
JENNIFER WEIL: Thank you, Wendy.
MALE: Nicely done, Wendy. First, a comment

153
00:23:25,266 --> 00:23:35,199
from my past, working on ischemic injury. It’s always difficult to show a good picture of something that looks inherently awful and there’s

154
00:23:35,200 --> 00:23:45,100
a big problem with selecting what is a picture that shows all the biology. I think your random choice worked just as marvelously well as anything else

155
00:23:45,100 --> 00:23:54,333
you would have done in a directed search. But I guess the question I would then now direct to Roger, which is: based on your proteomics and

156
00:23:54,333 --> 00:24:05,233
gene chip, when the podocytes are sloughing off, as it were, are there markers that stay constant and are there some that are lost, and is

157
00:24:05,233 --> 00:24:21,199
that the route we need to go when we do histochemical analysis for identification of the podocytes and these sort of disease processes?

158
00:24:21,200 --> 00:24:32,966
ROGER WIGGINS: I’m afraid I can’t really answer that but I think by looking at transcription factor profiling and other podocyte-specific proteins we

159
00:24:32,966 --> 00:24:44,666
will be able to define phenotypes that carry useful information, clinical information, and I think the challenge…is Matthias still…no, I think

160
00:24:44,666 --> 00:24:56,266
Matthias went back to Ann Arbor, but Matthias Kretzler is looking at that kind of question using systems biology kinds of approaches. He and I

161
00:24:56,266 --> 00:25:04,032
work somewhat together to try to address some of those questions but I can’t answer your question right now.

162
00:25:04,033 --> 00:25:11,299
JENNIFER WEIL: The problem is that Matthias is also a collaborator of ours and so Matthias is getting my data to look at the transcription to see

163
00:25:11,300 --> 00:25:17,433
what’s going along with this. If I make a wrong call then I’m telling the wrong thing to Matthias, so it becomes a feedback loop that doesn’t actually

164
00:25:17,433 --> 00:25:21,233
advance this field at all.
ROGER WIGGINS: Well, from what I understand,

165
00:25:21,233 --> 00:25:35,466
you guys are making great progress.
JENNIFER WEIL: Thank you.

166
00:25:35,466 --> 00:25:45,199
JOHN BASGEN: Well, we were all told on the panel that we had three slides to present and I didn’t know which three to present, so I made

167
00:25:45,200 --> 00:26:02,033
five groups of three slides to sample from and Kevin presented the first two slides of my first set, so I’m going to skip this where we compared,

168
00:26:02,033 --> 00:26:12,199
and he presented our data this morning where we didn’t differentiate between podocyte, mesangial cell, endothelial; we grouped those

169
00:26:12,200 --> 00:26:22,333
three and compared the Weibel-Gomez method and the fractionator-disector method to this exhaustive cut method that nobody wants to

170
00:26:22,333 --> 00:26:38,999
repeat, I don’t think. So I’m going to skip over the description of that and go right to our results. In the left-hand column are the nine glomeruli from

171
00:26:39,000 --> 00:26:52,166
one normal mouse and the color purple represents a glomerulus measured using the three different techniques, and you’ll see that the

172
00:26:52,166 --> 00:27:07,766
Weibel-Gomez method always overestimates the number of podocytes in that glomerulus when you use the exhaustive cut method. The

173
00:27:07,766 --> 00:27:20,799
fractionator-disector method, when we did the statistical test, there was no significant difference between those two methods, so

174
00:27:20,800 --> 00:27:31,433
we’re saying that convinced us that it was an unbiased method. There was a statistical difference in the estimate of podocyte number

175
00:27:31,433 --> 00:27:42,666
using Weibel-Gomez, so we’re saying that that’s a biased method. And then this whole question about “how much bias can you have in your

176
00:27:42,666 --> 00:27:55,232
study and still get the information out of the study?” is a question we all have to keep discussing, I think. So now I’m going to…that only

177
00:27:55,233 --> 00:28:08,533
counts as one slide and I’m going to jump to my second group which was to try and answer this question that was proposed for this panel: is it

178
00:28:08,533 --> 00:28:16,999
possible to accurately measure podocyte number in clinical biopsy material? And I think “yes” is the answer, but you need to have an unbiased

179
00:28:17,000 --> 00:28:27,866
sample of the complete glomeruli in the biopsy; you need an unbiased sample of those complete podocytes that were in the tissue that you were

180
00:28:27,866 --> 00:28:45,632
looking at; and like John reminded us, you need an unambiguous marker for the podocytes; and then you need an unbiased counting method. I

181
00:28:45,633 --> 00:28:59,799
think if you can do those four things—and Agnes, maybe it’s not possible in clinical biopsies—then do we have to do it in human research biopsies?

182
00:28:59,800 --> 00:29:15,466
Anyway, if you can meet these four criteria I think you can get an unbiased estimate which is an accurate method of podocyte number, and I

183
00:29:15,466 --> 00:29:31,399
haven’t done this in a human biopsy. I’ve done lots of this counting in mice and rats and I tried to make an estimate of the time I thought it would

184
00:29:31,400 --> 00:29:43,600
take to use that disector-fractionator method in human biopsies. So, you need the plastic-embedded tissue that was embedded for EM,

185
00:29:43,600 --> 00:30:01,533
then you need, which we say are 1mm cube blocks or 1,000 cubic millimeter blocks, and we would then cut 1-micron-thick sections, which

186
00:30:01,533 --> 00:30:13,033
would be 1,000 sections through a block. We would save pairs of sections every 20 microns and every 20 microns would be the fractionator

187
00:30:13,033 --> 00:30:26,399
part, so we’re saving 1/20th of the sections, and then for the disector we needed to save a second section so we could use the disector

188
00:30:26,400 --> 00:30:36,333
technique. In estimating that it would take two hours to do that sectioning, you then would stain these sections in toluidine blue which I think

189
00:30:36,333 --> 00:30:48,166
would be about 30 minutes to do those sections. Then you need to map the glomeruli to know which of these glomerular profiles that you see in

190
00:30:48,166 --> 00:30:59,066
these sections come from complete glomeruli because if you have incomplete glomeruli you don’t know how many podocytes are not there.

191
00:30:59,066 --> 00:31:15,766
That mapping takes, I estimate, about 30 minutes per block. So, now we’ve got three hours that we’ve spent knowing what are the complete

192
00:31:15,766 --> 00:31:29,832
glomeruli in this block of tissue and now we can start to image the complete glomeruli, which I estimate would take 15 minutes. I think there’s

193
00:31:29,833 --> 00:31:47,199
going to be about seven pairs of sections through glomeruli that are four million cubic microns. So, you’d end up imaging those seven layers through

194
00:31:47,200 --> 00:31:59,366
a glomerulus and you’d have the image from the two sections, so you’d have 14 images from a glomerulus. Then you go through and you mark

195
00:31:59,366 --> 00:32:09,632
what are the podocyte nuclei; that’s what I’d do. I mean, you could put up the two images together and I don’t think it’s as hard to orientate these

196
00:32:09,633 --> 00:32:24,666
images as John presented yesterday. Photoshop as the software that I use to do this and you’ve taken these two images and the sections on your

197
00:32:24,666 --> 00:32:33,899
slide have been twisted, but in Photoshop I’m good at saying, well, this has to be rotated clockwise 37 degrees and they’re lined up well

198
00:32:33,900 --> 00:32:49,333
enough to know that the profile from a podocyte in this section is the same as in this section. Then after all that work, it only takes about five minutes

199
00:32:49,333 --> 00:33:00,999
to see which of the podocytes have a profile in this section but it’s disappeared in this section and that’s what you count. So, you see that

200
00:33:01,000 --> 00:33:13,066
those steps five, six, and seven is per glomerulus so it’s maybe 30 minutes per glomerulus. I don’t know how many glomeruli were in the original

201
00:33:13,066 --> 00:33:25,632
block of tissue and I don’t know how many glomeruli we need to get the precision that we want for our particular study. So, this is my rough

202
00:33:25,633 --> 00:33:41,299
estimate of three-and-a-half hours to get the first unbiased estimate of the podocytes in one glomerulus and now add a half-hour to each

203
00:33:41,300 --> 00:33:54,866
additional glomerulus that is present and the concern is that the variability in the number of podocytes in the glom within a patient, the

204
00:33:54,866 --> 00:34:09,832
statistician would say you need lots of glomeruli to get the power to determine or to reach the statistical difference. So, that’s my presentation

205
00:34:09,833 --> 00:34:22,299
for today. Any questions?
JOHN BERTRAM: Something we’ve perhaps not

206
00:34:22,300 --> 00:34:33,400
explicitly stated during this meeting—if it was, I might have missed it—but what John is talking about there, as I understand it, is trying to get a

207
00:34:33,400 --> 00:34:45,966
count for the number of podocytes in individual glomeruli in a biopsy and that is much more time and work than trying to get an estimate of the

208
00:34:45,966 --> 00:34:54,866
number of podocytes in an average glomerulus where you wouldn’t have to do the serial sectioning, etc., etc. So that comes down, I

209
00:34:54,866 --> 00:35:05,599
guess, to the clinicians in the room dealing with human clinical material and I’m not a clinician but it gets down to that question: do you want your

210
00:35:05,600 --> 00:35:16,833
data on a per glomerulus basis or do you want your data on an average glomerulus basis for your biopsy for IgA or FSGS or whatever?

211
00:35:16,833 --> 00:35:26,266
They’re different pieces of information and they require different sampling and they require different amounts of time. So if you were just

212
00:35:26,266 --> 00:35:35,132
trying to say in this biopsy there are 12 glomeruli and the average number of podocytes per glomerulus is 610 or whatever, that’s a lot quicker

213
00:35:35,133 --> 00:35:41,499
than trying to do what you’re doing, but of course you’re getting a lot more information doing it on a per glom basis.

214
00:35:41,500 --> 00:35:54,933
JOHN BASGEN: Well, I didn’t mention that you get almost free the glomerular volume from these images. You can do Cavalieri that is very quick. I

215
00:35:54,933 --> 00:36:10,633
believe, to get accurate numbers, you have to use these unbiased methods and I think if this is done again we need a stereological statistician to

216
00:36:10,633 --> 00:36:27,266
help us decide how much bias is allowed into our studies and then go from there, that yes, we can live with this much error if we’re trying to detect

217
00:36:27,266 --> 00:36:36,532
this much difference in our different groups.
AGNES FOGO: This approach and measuring

218
00:36:36,533 --> 00:36:46,766
and counting is so very, very different when you’re talking about a fairly homogeneous population in terms of destructive scarring altered

219
00:36:46,766 --> 00:36:55,532
lesions versus normal, and Jennifer’s already talked about the abnormalities of the glomeruli when they’re big and abutting Bowman’s space

220
00:36:55,533 --> 00:37:05,066
of just seeing it and that’s not a sclerosed glomerulus or a glomerulus that has a lesion of proliferation like in IgA nephropathy. So again,

221
00:37:05,066 --> 00:37:14,366
yesterday we talked a little bit…I brought up the question and asked John: how do these methods apply? What is the sampling size needed? What

222
00:37:14,366 --> 00:37:22,099
is the variability? What is your power to detect differences with an intervention if you have repeat biopsies or different time points in a lesion

223
00:37:22,100 --> 00:37:33,100
with a treatment? And when you take into account the complexity of the lesions and the tremendous difference in glomeruli that we see at

224
00:37:33,100 --> 00:37:40,800
a light microscopic level, it becomes really challenging. So, John and Wendy shared with us that they don’t count the globally sclerotic

225
00:37:40,800 --> 00:37:47,766
glomeruli in their specimens. There’s no technical reason why you couldn’t; you can recognize a globally sclerosed glomerulus. It was tactical

226
00:37:47,766 --> 00:37:57,966
decision and that’s information about what’s going on, but we can’t count podocytes in a glomerulus that is globally sclerosed. So, how do

227
00:37:57,966 --> 00:38:08,132
we deal with the complexities of disease and how would that add to the time of being able to map glomeruli, or if you had a bigger piece of

228
00:38:08,133 --> 00:38:18,666
tissue—not the little 1 mm cube pieces—for me at least, to map the glomeruli. I did a serial section study on LM, mapping glomeruli to find complete

229
00:38:18,666 --> 00:38:29,566
profiles, and it took a really, really long time to map in a long biopsy core the same glomerulus through the 80-100 sections that I had and make

230
00:38:29,566 --> 00:38:38,032
sure that that glomerulus was cut completely. So if you have a small 1 mm cube and you don’t have that many players and glomeruli in there, it’s easy.

231
00:38:38,033 --> 00:38:46,266
I’d really like to know your take on the variability on the sampling and the size and the mapping because I sure can’t map in a bigger piece of

232
00:38:46,266 --> 00:38:53,599
tissue three dimensionally how many glomeruli are followed from core to core in 30 minutes.

233
00:38:53,600 --> 00:39:16,733
JOHN BASGEN: Well first of all…where did I go around here? I want to back up one slide. If I hit the down button…this is optical sectioning on the

234
00:39:16,733 --> 00:39:21,366
what?
FEMALE: [inaudible comment]

235
00:39:21,366 --> 00:39:31,432
JOHN BASGEN: Oh, yes, yes. Well, the point I’m trying to make is number three: that we have to know what we’re counting. So as the pathology

236
00:39:31,433 --> 00:39:41,933
gets to the point where we’re not sure if it’s a podocyte or not, we’re going to be inaccurate in counting those podocytes, so that’s one thing.

237
00:39:41,933 --> 00:39:58,666
The other question that Agnes was asking, I think, is not accuracy but precision: how much do we have to count? The good news is, by doing more

238
00:39:58,666 --> 00:40:10,199
work you can make things precise enough. So if the means of your two groups that you’re trying to detect that normal podocytes numbers here

239
00:40:10,200 --> 00:40:19,400
and there’s hardly any podocytes left here, you don’t have to do a lot of work to detect that difference, but as the means of your groups get

240
00:40:19,400 --> 00:40:31,500
closer together, then you need this more precision and I think in the human biopsies where there aren’t a lot of glomeruli, you’ve got to spend

241
00:40:31,500 --> 00:40:46,133
time doing more precision. Now, I suggested this 20 micron be the spacing and I’m guessing that you’d get about 7 disector pairs going at that 20

242
00:40:46,133 --> 00:41:01,133
micron space. That might be too precise. Maybe going 50…well, if you went 100 microns then you might only get 3 pairs and that’s not feeling good

243
00:41:01,133 --> 00:41:12,599
enough to me, but I don’t know. But these are questions for doing it this way and I think for doing it other ways. You’ve got to get this feeling

244
00:41:12,600 --> 00:41:22,600
of how much variation there is at the different levels within the glomerulus among the glomeruli within the biopsy, and that’s where we don’t

245
00:41:22,600 --> 00:41:32,533
have a whole lot of information yet.
BEHZAD NAJAFIAN: So, I was faced with the

246
00:41:32,533 --> 00:41:44,033
same question and I admit that I cheated and have more than three slides but I will try to speak fast in taking advantage of being the last person to

247
00:41:44,033 --> 00:41:54,366
talk on the panel and don’t be repetitive. So, is it possible to accurately measure podocyte numbers in clinical biopsies? I think that in order to

248
00:41:54,366 --> 00:42:06,699
answer that question we have to acknowledge questions and limitations. Do we know how to accurately count podocytes? The talks that we

249
00:42:06,700 --> 00:42:18,233
had, they mainly tried to deal with that question, although I’m sure that we reached a consensus or we are not very hopeful, that especially when

250
00:42:18,233 --> 00:42:30,366
we have a limited sample, it will be easy to reach to a consensus. We also need to consider challenges of podocyte counting in limited

251
00:42:30,366 --> 00:42:44,599
samples, that our biopsies in practical issues of studying clinical specimens should be considered. So, the first thing would be to define

252
00:42:44,600 --> 00:42:58,500
what is our definition for podocytes and that pretty much, I think, goes back to the question of our research. So, well, a cell is sitting on the

253
00:42:58,500 --> 00:43:10,733
glomerular tuft, by morphology having foot processes: do we call it a podocyte or not? The question that Jennifer raised, if we have a

254
00:43:10,733 --> 00:43:19,799
bridging cell between the capillary loop and Bowman’s capsule basement membrane, is that a podocyte or not? If a cell that shows markers of

255
00:43:19,800 --> 00:43:29,166
parietal epithelial cells is sitting on the capillary tuft, is it really functioning as a podocyte? Should we count it as a podocyte or, as Dr. Wiggins

256
00:43:29,166 --> 00:43:38,399
mentioned, we should define by cell markers that, okay, we define that we are only interested in cells that are expressing these types of markers

257
00:43:38,400 --> 00:43:52,533
and I think that that goes back pretty much to the main questions of our research and those should be determined ahead of time. Then, if we are

258
00:43:52,533 --> 00:43:59,699
calling podocytes based on the morphology saying that, okay, podocytes cover the exterior surface of the glomerular basement membrane,

259
00:43:59,700 --> 00:44:10,233
we should be careful about the sample that we’re studying as well as the pathology in the glomeruli, as people talked about. So for example, I found it

260
00:44:10,233 --> 00:44:25,066
difficult, especially when we start to have pathology conditions in the glomeruli to tell apart podocytes from, let’s say, other cell types,

261
00:44:25,066 --> 00:44:33,432
especially parietal epithelial cells in paraffin sections or in frozen sections, why it’s easier on plastic-embedded if you look at the transmission

262
00:44:33,433 --> 00:44:47,666
electromicroscopy and then you need, probably, going to higher magnification. So again, if you are planning for doing a study you’ll want to consider

263
00:44:47,666 --> 00:45:00,332
which would be the priority for specifying that, okay, I’m going to specify more samples for electromicroscopy or embed in plastic or other

264
00:45:00,333 --> 00:45:13,133
conditions to do immunohistochemistry for cell markers I alluded to. And then: what to measure? What is the parameter of our interest? Is it the

265
00:45:13,133 --> 00:45:22,066
average number or is it the number density per glomerulus or the number density per glomerular basement membrane surface? These are

266
00:45:22,066 --> 00:45:29,666
different questions. So if we are interested, for example, in the number of podocytes per glomerulus basal membrane surface, I don’t think

267
00:45:29,666 --> 00:45:40,566
that the paraffin-embedded tissue will be an appropriate sample to answer that kind of question. So, going back to the same thing, that

268
00:45:40,566 --> 00:45:50,732
we have to define what are the questions and based on that, decide what would be the appropriate sample to use. Then, how are we

269
00:45:50,733 --> 00:46:02,533
going to measure podocyte number? How should be our sampling strategy and are we allowed to consider assumptions, to involve assumptions? In

270
00:46:02,533 --> 00:46:16,266
order to answer that question I would like to go back to this very basic principle in sampling, not only about histology but that can be applied to all

271
00:46:16,266 --> 00:46:31,332
kinds of, I think, different sampling strategies, that we often measure observed variants—this is what we are able to measure by our

272
00:46:31,333 --> 00:46:39,533
methodology that we choose—and actually that observed variance is composed of two different components, and what would be the biological

273
00:46:39,533 --> 00:46:52,699
variants that we want to measure? It’s the variance of expectation of the estimate plus the expectation of the sampling variance, which is

274
00:46:52,700 --> 00:47:03,266
variance coming from sampling and this is something that we want to minimize in order to actually or what we observe to be as close as

275
00:47:03,266 --> 00:47:13,899
possible to biological variants that we are interested to detect. In order to reach to this goal and minimize this component, we want to have

276
00:47:13,900 --> 00:47:25,200
an unbiased sampling and choose a methodology which is efficient combining these two together; I think that doing a systematic uniform random

277
00:47:25,200 --> 00:47:39,866
sampling would be the most efficient way as the authorities in this field such as Gundersen has showed that clearly. Kevin and other people went

278
00:47:39,866 --> 00:47:51,932
beautifully over the different methods to measure podocyte number. The disector-fractionator’s unbias is shape and size independent; it often

279
00:47:51,933 --> 00:48:01,399
requires plastic-embedded, although I found the data presented by John Bertram very interesting and I look forward to see what comes out of it for

280
00:48:01,400 --> 00:48:12,200
using optical disector in the paraffin-embedded tissue as well; it’s technically demanding; it’s time consuming. However, we need to decide

281
00:48:12,200 --> 00:48:24,366
whether the question that we are asking is important or how important it is and if it is worth to try this method or not. Weibel-Gomez method is

282
00:48:24,366 --> 00:48:39,266
model-based; it involves shape and size assumption; it’s volume-weighted; and it requires thin sections, as Kevin showed. You’ll want to try to

283
00:48:39,266 --> 00:48:49,899
do this, for example, in paraffin sections. You will come up with quite different numbers or

284
00:48:49,900 --> 00:48:56,966
estimations, so we want to do this on thin sections and it can very well be applied to transmission electromicroscopy. A question that is

285
00:48:56,966 --> 00:49:07,932
raised with Weibel-Gomez is that podocyte nuclei may not have the ideal shape for this method unless we include some assumptions. Another

286
00:49:07,933 --> 00:49:18,366
method that was not discussed here—and I’m not ready to go into details but after this session if anybody’s interested I can go into that—is a point

287
00:49:18,366 --> 00:49:29,999
sample intercept method which is an unbiased method. It is shape-independent completely but the sampling strategy is volume-weighted but at

288
00:49:30,000 --> 00:49:34,366
least it’s completely unbiased, and based on the measurement of the nuclear

289
00:49:34,366 --> 00:49:43,799
volume and volume fraction, it can give us an unbiased estimation of the number density of podocytes as well; and thin and thick section

290
00:49:43,800 --> 00:49:57,400
method, that was also discussed. So, here’s an example of things that need to be considered. I’m just quoting the original paper by Weibel and

291
00:49:57,400 --> 00:50:10,200
Gomez for estimating the particle number and these are those things that Kevin mentioned in his talk, that this equation—the basic equation for

292
00:50:10,200 --> 00:50:20,933
Weibel-Gomez method—the authors say is applicable if the following is satisfied. One is that the profile density must be fairly constant at any

293
00:50:20,933 --> 00:50:30,899
level, any direction; the particles must be comparable to finite bodies and any section through any particle may yield only one

294
00:50:30,900 --> 00:50:42,900
transection. Well, based on what Jennifer showed and also often we see that the shape of the nuclei in humans, especially, of podocytes is

295
00:50:42,900 --> 00:50:54,933
more kind of a kidney shape or a dumbbell shape and it’s not very rare, although it’s not very common either, to see two profiles that could be

296
00:50:54,933 --> 00:51:07,866
coming from one nucleus. Of course, unless we have look-up sections or we use optical means, we can’t understand whether they’re coming

297
00:51:07,866 --> 00:51:16,232
from different nuclei or from the same nucleus. The sections must be very thin compared to the smallest diameter of the particles, it must be

298
00:51:16,233 --> 00:51:33,866
randomly oriented so, as I said, this method should be limited to very thin sections; TEM would be ideal. The way that people calculate the shape

299
00:51:33,866 --> 00:51:46,766
coefficient is based on actually this assumption that the shape of the particle that we are estimating can be simplified into revolution

300
00:51:46,766 --> 00:52:00,732
through an axis, so it’s basically a symmetrical shape. So, I’m sure how much deviation from this assumption would affect our methodology. I also

301
00:52:00,733 --> 00:52:20,699
wanted to quote from a review paper by Mayhew and Gundersen that’s basically for disector method. I do not want to disrespect

302
00:52:20,700 --> 00:52:33,566
people that are not using unbiased methodology at all. However, I think that we all should be careful about the biased methodologies and also

303
00:52:33,566 --> 00:52:39,832
model-based methodology because any model-based methodology involves assumptions that may work in special conditions, but not

304
00:52:39,833 --> 00:52:55,099
necessarily in others, so we should be very careful and a priori, at least test, whether those assumptions hold true or not. Another thing that in

305
00:52:55,100 --> 00:53:04,866
clinical samples especially is very important is be careful about the reference trap when we come up with numerical density—how we’re going to

306
00:53:04,866 --> 00:53:17,832
calculate the total number by measuring the total glomerular volume if we don’t have an adequate sample to do that. So, all of these methods, even

307
00:53:17,833 --> 00:53:31,699
the disector-fractionator, these require you to do enough counting and we have a general guideline that, okay, for complex structures if we count

308
00:53:31,700 --> 00:53:46,500
about 150 different events we should come up with a robust estimation. We should test these in pilot studies and determine how many profiles—if

309
00:53:46,500 --> 00:53:54,066
we were going to do Weibel-Gomez—how many profiles do you need to measure to come up with a robust number? If you’re going to do disector-

310
00:53:54,066 --> 00:54:05,799
fractionator in a limited sample, again, how many events, how many cue bars do you need to count in your disector to come up with a robust

311
00:54:05,800 --> 00:54:20,100
number? Otherwise, the numbers could be quite off. Questions that need to be answered, basically, as John said, depends on the variability

312
00:54:20,100 --> 00:54:32,000
that we have in different conditions, so I think that we won’t be able to give a fixed guideline, that okay, you need to look at these many glomeruli,

313
00:54:32,000 --> 00:54:40,800
count these many podocytes, you will come up with the right answer. For each study, for each condition and each disease, you would need to

314
00:54:40,800 --> 00:54:51,100
do a pilot study before that to answer these questions to determine your sampling strategy. At the end I think that it’s also very important, if we

315
00:54:51,100 --> 00:54:59,800
want to make any sense of our measurements, we want to do a structural functional relationship. So one side is to use appropriate methodology to

316
00:54:59,800 --> 00:55:10,166
come up with the right morphometric answer, but on the other side, it’s also very important to do appropriate functional studies, functional kidney

317
00:55:10,166 --> 00:55:24,166
studies. If you don’t have good estimations of glomerular filtration rate, it’s very unlikely that…you may miss a structural-functional relationship

318
00:55:24,166 --> 00:55:35,599
even if you can detect the structure changes; same thing for albumin excretion rate. So, with that I conclude my talk.

319
00:55:35,600 --> 00:55:45,933
KEVIN LEMLEY: So maybe we can open up to general discussion. Jeffrey, I don’t know how much time we have. Five minutes more. I’ll take

320
00:55:45,933 --> 00:55:54,833
advantage of the first point just to say something I said before, that to make theoretical arguments against the model-based methods, to me doesn’t

321
00:55:54,833 --> 00:56:04,933
hold much water and if one has stuff like Kath’s data where it does agree, I’d be very interested in seeing John’s raw data because before when I

322
00:56:04,933 --> 00:56:13,933
asked you for that to see if the cross-section correction weighting…I guess you found it again…to see if that would eliminate that small

323
00:56:13,933 --> 00:56:26,866
bias, but fundamentally I’d like to see data would be the answer and unfortunately, with very rare exceptions, we don’t have data.

324
00:56:26,866 --> 00:56:34,199
KATHRYN WHITE: I’d just like to talk about how many levels that we looked at in the disector when we estimated podocyte number in our Type

325
00:56:34,200 --> 00:56:41,366
II and Type II diabetic patients, and we looked at five glomeruli per patient and we started by looking at about eight pairs. We also compared

326
00:56:41,366 --> 00:56:45,699
that to…we did four pairs…
KEVIN LEMLEY: Eight per…

327
00:56:45,700 --> 00:56:54,100
KATHRYN WHITE: …eight per glomerulus and worked out that, by half the amount of work, we got very similar answers, so we ended up doing

328
00:56:54,100 --> 00:57:04,433
sort of four disector pairs per glomerulus, and as for mapping the time to take to map and that type of thing, I think Agnes is right: in a small Epon

329
00:57:04,433 --> 00:57:13,433
block it’s quite easy to map because you’re lucky if you’ve got more than one glom at a time. You get a piece of tissue in a biopsy or I also do it in

330
00:57:13,433 --> 00:57:21,299
mouse and rat and it can take a lot longer. Gloms are appearing and going all the time and to make sure that one hasn’t disappeared and another

331
00:57:21,300 --> 00:57:30,133
one’s come in at the exact place is a lot more difficult in large pieces of tissue. The other point I’d like to make is John also said that you get

332
00:57:30,133 --> 00:57:36,566
glomerular volume for free when you do the disector, which is true. You also can get the percentage of sclerosed glomeruli while you’re

333
00:57:36,566 --> 00:57:47,999
going through there and you can also use these sections for looking at atubular glomeruli. So, you can, if you’ve got 1 micron sort of serial section…

334
00:57:48,000 --> 00:57:54,833
MALE: I’m not sure you can atubular glomeruli because we threw away 20 microns where the tubular…

335
00:57:54,833 --> 00:58:02,966
KATHRYN WHITE: Well, we’ve got them all. We’ve got the sections, so that’s how I got atubular glomeruli was from taking all these 1

336
00:58:02,966 --> 00:58:06,332
micron sections. JOHN BASGEN: So if you spend more time

337
00:58:06,333 --> 00:58:08,933
saving all the sections, then you can get more information.

338
00:58:08,933 --> 00:58:09,899
KATHRYN WHITE: When you start these studies,

339
00:58:09,900 --> 00:58:15,100
sometimes you don’t know what you’re going to be looking for in five years’ time, and if you’ve kept all the tissue,

340
00:58:15,100 --> 00:58:20,266
it can be a very valuable resource.
MALE: Forgive me, this is not really

341
00:58:20,266 --> 00:58:21,999
my field, but I was just wondering…

342
00:58:22,000 --> 00:58:28,600
it seems like there’s a lot of arguments about whether different models are right and some of them it seems like there’s really

343
00:58:28,600 --> 00:58:37,166
no ground truth to compare against, and so that’s why there’s a lot of these arguments, and I was wondering if anybody had ever tried doing flow

344
00:58:37,166 --> 00:58:44,466
cytometry on large sections or whole kidneys broken up or single glomeruli and trying to see exactly how many cells there are per glomerulus

345
00:58:44,466 --> 00:58:48,632
and then comparing that against the techniques.
KEVIN LEMLEY: I can tell you no one has done

346
00:58:48,633 --> 00:59:01,299
that. So you’re saying if you wanted to ground this, if you took a sample and used whatever technique and said “this is how many podocytes I

347
00:59:01,300 --> 00:59:12,133
get per glom” and then you somehow isolated gloms, enumerated them, digested them, stained them, I would encourage anybody to try that. It

348
00:59:12,133 --> 00:59:25,266
sounds like there would possibly be a little bit of noise in the technique but the advantage would be in a completely independent way. So, go for it.

349
00:59:25,266 --> 00:59:33,132
SUSANNE NICHOLAS: Can I make one comment on that? I mean, some of the issues you may have to take into consideration for a study like

350
00:59:33,133 --> 00:59:40,833
that is antibody specificity and also in terms of the antigen in health may be different in disease, and so you may actually lose a lot of information

351
00:59:40,833 --> 00:59:46,099
in that regard.
MALE: No, I agree, and I was just thinking of in

352
00:59:46,100 --> 00:59:56,533
healthy patients or animals, just testing techniques that way.

353
00:59:56,533 --> 01:00:05,733
MALE: I just wanted to bring up a point as a clinician. In the practice of nephrology we use a lot of biomarkers that are inaccurate, imprecise.

354
01:00:05,733 --> 01:00:15,566
We don’t do 24-hour urine collections, we use albumin-to-creatinine ratios, we generally don’t do [---] clearances, we use serum creatinine or

355
01:00:15,566 --> 01:00:23,832
equations-based on it and there’s a whole literature on the bias and accuracy of all of these markers. So, I would just bring up the point that if

356
01:00:23,833 --> 01:00:33,066
you have a method that can be done in clinical practice that takes 5 minutes and not 12 hours and it’s biased and imprecise but it does better

357
01:00:33,066 --> 01:00:41,666
than what we currently have to risk stratify patients, that might be of value, and I just wonder what your thoughts would be on that.

358
01:00:41,666 --> 01:00:51,799
KEVIN LEMLEY: I would call attention to the people to the diskette on the paper by [---] from Japan where they looked at glomerular density in

359
01:00:51,800 --> 01:01:01,500
core biopsies from IgA; a very non-stereologic approach. I mean, it was number of profiles per area, so we know that that can’t possibly be

360
01:01:01,500 --> 01:01:10,200
right. The only problem is, it was a stronger predictor of progression than proteinuria. I’m trying to remember what the other…I guess

361
01:01:10,200 --> 01:01:19,933
crescents, the binomial presence of crescents or not were the only two in a multivariate that predicted outcome. So, that kind of speaks to that

362
01:01:19,933 --> 01:01:29,533
this would be considered a very primitive morphometric or pseudo-morphometric predictor, but in predicting clinical outcome, it beat a lot of

363
01:01:29,533 --> 01:01:35,766
things that you might think would be stronger.
SUSANNE NICHOLAS: And just one addition to

364
01:01:35,766 --> 01:01:45,299
that, too. As we move forward in terms of identifying and using biomarkers in the diagnosis of disease even before we can see some of the

365
01:01:45,300 --> 01:01:53,833
typical clinical parameters that we used to measure, like albuminuria, for example, I think studies such as the morphometric analyses really

366
01:01:53,833 --> 01:02:05,099
become much more important where we can relate structure and function and have a better predictor in terms of disease progression and

367
01:02:05,100 --> 01:02:34,233
disease outcome down the road.




Date Last Updated: 10/3/2012

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