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

Breakout Session Reports
  • Glomerular Area and Volume
    John Bertram, Monash University, Australia and Wendy Hoy, The University of Queensland, Australia
  • Interstitial Capillaries
    Michael Mauer, University of Minnesota and Agnes Fogo, Vanderbilt University
  • Interstitial Fibrosis
    Kevin Lemley, University of Southern California and Cindy Nast, Cedars – Sinai
  • Digital Pathology
    Stephen Hewitt, National Cancer Institute and Laura Barisoni, New York University Langone Medical Center

Video Transcript

1
00:00:00,000 --> 00:00:07,100
JEFFREY KOPP: Okay. Let’s go ahead and get started with our reports in the breakout session. I know planes are departing, I know people are

2
00:00:07,100 --> 00:00:15,466
departing, so we’ll go ahead and do reports from the four sessions and then Kevin has a brief wrap-up, and we’ll try to get all of this done in the

3
00:00:15,466 --> 00:00:30,166
next 25 minutes or so. John is going to start.
JOHN BERTRAM: So, Wendy Hoy and I kind of

4
00:00:30,166 --> 00:00:39,166
chaired this little session looking at glomerular area and volume. I don’t know that we did everything we were asked to but this is where

5
00:00:39,166 --> 00:00:52,166
we got to. I think there was agreement that the reality is that not all of the quantitative morphology QM techniques that we’ve heard about here in

6
00:00:52,166 --> 00:01:01,732
the last day or so may be applicable to the renal biopsy. Where we’re dealing with a limited amount of tissue, the priority has to be the correct

7
00:01:01,733 --> 00:01:13,133
pathological diagnosis, but perhaps we can think about how best to utilize this precious resource, and maybe by talking to the pathologists more in

8
00:01:13,133 --> 00:01:27,599
terms of how sections are collected and stained and shared and so forth, we may be able to get more quantitative data out of those biopsies. I

9
00:01:27,600 --> 00:01:40,400
think the issue of how these techniques we’ve heard about, again, that the workshops hold up in pathology is an important point. We heard from

10
00:01:40,400 --> 00:01:52,000
Jennifer Weil today and her talk on the Pima project, the issue of cell identification and how that’s tough and as the pathology gets more and

11
00:01:52,000 --> 00:02:01,400
more advanced, this issue becomes more and more difficult. So, maybe some of these methods we’ve been talking about here some of them may

12
00:02:01,400 --> 00:02:12,600
do very well and be very useful there and others may start to fall down. The issue of specific markers and so forth for cells of glomeruli or

13
00:02:12,600 --> 00:02:24,600
whatever, again, may well change with disease. Our group was very interested in the emerging opportunities that we’ve heard about today to

14
00:02:24,600 --> 00:02:34,566
combine the structural and functional data with these new methods. The obvious one we heard today about doing the MR and getting the nephron

15
00:02:34,566 --> 00:02:44,966
count, getting the volume of every glomerulus; being able to combine that with estimates, non-invasive estimates, of GFR and even longitudinal

16
00:02:44,966 --> 00:02:57,699
studies is a very powerful new approach to be able to understand the functioning of the kidney. The point was also made that if glomerular

17
00:02:57,700 --> 00:03:07,866
hypertrophy is so important—I guess most of us assume it probably is because we spend a lot of time thinking about how to size glomeruli—then if

18
00:03:07,866 --> 00:03:17,466
we were able to identify the molecular markers that drive that hypertrophy—the growth factors, the cytokines, etc.—then to be able to measure

19
00:03:17,466 --> 00:03:25,632
those factors in blood or urine would be a big step forward. None of us assume that it’s going to be the same factors and the same mechanisms

20
00:03:25,633 --> 00:03:36,766
in every setting of glomerular hypertrophy; biology and medicine is never that simple. But as these drivers of hypertrophy emerge, we may

21
00:03:36,766 --> 00:03:49,499
well be able to measure those factors as well. There was a call for step-by-step protocols to be made available for these techniques to enable

22
00:03:49,500 --> 00:03:59,700
beginners or new users to get started, and rather then have to do a Ph.D. in stereology, if they got some easy step-by-step protocols they could get

23
00:03:59,700 --> 00:04:09,966
off to a quick start and hopefully avoid some of the pitfalls that come into these techniques and they come into all techniques, so why don’t we

24
00:04:09,966 --> 00:04:21,966
try to make it easier for new people? Also, there was a call for standardization of approaches and protocols. In terms of new approaches for

25
00:04:21,966 --> 00:04:34,732
measuring glomerular area and volume, that was our brief, in just a short amount of time six methods were mentioned, or five or six new

26
00:04:34,733 --> 00:04:43,099
methods were mentioned which made me think we could have probably kept going at this meeting for a little longer or do another one at

27
00:04:43,100 --> 00:04:52,300
some point in another way. Bob Bacallao told us about a method he’s developing and has had some experience with involving confocal

28
00:04:52,300 --> 00:05:01,666
microscopy where he’s taking kidney slices, he’s not embedding them in formalin or plastic or whatever, he’s simply immersing them in saline or

29
00:05:01,666 --> 00:05:12,799
a physiological solution, fluorescently labeling these slices with some of these markers here and then optical imaging through to a depth of

30
00:05:12,800 --> 00:05:20,666
about 150 micrometers to be able to measure some of the parameters we’ve been talking about. O course there’s a lot of interest in the MR

31
00:05:20,666 --> 00:05:30,232
work that Kevin Bennett and Norbert Gretz talked about. Kevin was in our group. He thought the biggest issue going forward will be the correcting

32
00:05:30,233 --> 00:05:40,233
for the RF artifacts, the radio frequency artifacts; the development of better coils, better surface coils. Perhaps he’s lucky enough to have a 19

33
00:05:40,233 --> 00:05:52,233
Tesla machine you saw was about the size of a small building but perhaps for many of our applications 3T or 7T will be adequate. Julie

34
00:05:52,233 --> 00:06:04,099
Ingelfinger reminded us of methods developed in the 70s or so using dextrans of different sizes and charges and so forth and sieving

35
00:06:04,100 --> 00:06:16,766
approaches to be able to estimate glomerular function, and I presume…I’m not an expert in this but perhaps the functional surface area or

36
00:06:16,766 --> 00:06:24,066
capacity of the filtration barrier and perhaps that should be something as we go forward with these technologies is maybe an approach that

37
00:06:24,066 --> 00:06:33,832
might be worth thinking about resurrecting in one form or another. Yu Chen told us about his work with optical coherence tomography; I hope this is

38
00:06:33,833 --> 00:06:44,833
the right words, I apologize for any mistakes. He’s using this approach in rats and in people as a clinical study in progress. He’s getting

39
00:06:44,833 --> 00:06:57,366
resolution of the 3-10 micrometer level, which is quite impressive, so I’m sure we’re going to hear a little bit more about that over the coming months

40
00:06:57,366 --> 00:07:08,032
and years. Ashraf El-Meanawy is using ultrasound to measure glomerular diameter in human kidneys and is about to launch a study or

41
00:07:08,033 --> 00:07:19,299
is involved in a study looking at glomerular size in control subjects versus subjects with glomerulomegaly. So there are a lot of new

42
00:07:19,300 --> 00:07:30,633
methods kind of in the pipeline, if you will. Some of them are already being applied to patients and the like, which hold great promise and who

43
00:07:30,633 --> 00:07:38,599
knows what we might be talking about in two or three years’ time? The point was made that, you know, with the techniques we’ve heard about in

44
00:07:38,600 --> 00:07:47,433
the last day but also with these new techniques coming along, that gives us even more meaning but also more opportunity to try to cross-validate

45
00:07:47,433 --> 00:08:01,933
these approaches and see how reproducible and similar they are in the kinds of data that, at the end of the day, they provide. I think the final task

46
00:08:01,933 --> 00:08:14,533
we were asked to do was look at opportunities and how using quantitative morphology might help us improve our understanding of disease. David

47
00:08:14,533 --> 00:08:20,466
Archer pointed out, and we’ve heard it plenty of times here at the workshop, we need something at least in the clinical setting that’s quick and

48
00:08:20,466 --> 00:08:32,132
effective. Time efficiency is paramount. We need a morphological marker, ideally, of progression or non-progression so we can start to do

49
00:08:32,133 --> 00:08:42,433
longitudinal studies and be able to measure things like glomerular drop-out; so, the safe non-invasive approach either in laboratory animals or

50
00:08:42,433 --> 00:08:50,133
in patients and so forth. The point was made “don’t forget the tubules and the interstitium,” which apparently are some part of the kidney as

51
00:08:50,133 --> 00:08:59,966
well. I guess we haven’t talked about them a lot here, but I was reminded of a paper I think it was in JASN last year by Inoue and colleagues who

52
00:08:59,966 --> 00:09:14,532
used MR to measure interstitial fibrosis in the kidney. I think that’s, again, something of great interest, so I recommend you have a look for that

53
00:09:14,533 --> 00:09:23,266
paper. Finally of course, there was a lot of interest in non-invasive quantification, quantitation of glomerular number and size in patients. This

54
00:09:23,266 --> 00:09:32,399
would assist us with issues of organ allocation, with risk stratification, being able to monitor disease progression, assess the effects of

55
00:09:32,400 --> 00:09:41,600
various treatments, assess glomerular endowment in premature kids and low birthweight babies and so forth and really give us

56
00:09:41,600 --> 00:09:55,633
a lot of new power and new opportunities in our research. So, that’s as far as we got. I hope that reflects what we talked out.

57
00:09:55,633 --> 00:09:58,833
KEVIN LEMLEY: I didn’t know I was the appointed speaker, so Laura’s getting…

58
00:09:58,833 --> 00:10:02,599
LAURA BARISONI: We didn’t decide.
KEVIN LEMLEY: Yeah, we didn’t decide. So the

59
00:10:02,600 --> 00:10:12,733
group, which was about eight of us, seven or eight of us, talked and discussed…primarily the focus that our group discussed was the

60
00:10:12,733 --> 00:10:20,066
challenges of the implementation of morphometric analysis within the context of clinical trial settings, and we spent actually far more time

61
00:10:20,066 --> 00:10:27,632
talking about the complexity of doing the clinical trials and collecting the pathology. This is something that has been an ongoing developing

62
00:10:27,633 --> 00:10:39,633
subject within NEPTUNE and is really the first large multi-center trial in nephrology to use virtual pathology or whole-slide imaging in this space.

63
00:10:39,633 --> 00:10:46,766
So, most of the discussion was the complexity and how one would implement that and some of the advantages of moving to an independent

64
00:10:46,766 --> 00:10:55,532
rather than consensus microscope review process. We also discussed the complexities of developing appropriate scoring systems that are

65
00:10:55,533 --> 00:11:03,266
piloted because, you know, you come in with a system and you need to test it a priori before you do this because it may not fit the data set that

66
00:11:03,266 --> 00:11:09,432
you’re collecting, either because the natural history of the disease is changing or the questions that you’re trying to evaluate are

67
00:11:09,433 --> 00:11:17,099
different. We also discussed slightly some of the technical challenges, especially multi-center environment, in collecting the imaging to make it

68
00:11:17,100 --> 00:11:25,233
adequate for morphometric analysis and other utilities, and these are ongoing developments.

69
00:11:25,233 --> 00:11:52,166
FEMALE: [inaudible question]
LAURA BARISONI: So, do we have it? So, we

70
00:11:52,166 --> 00:12:09,532
talked about the difficulties in the regulations, that’s more Steve’s job, and my part is actually to say why we use it and which are the

71
00:12:09,533 --> 00:12:17,966
advantages, and if you think about classic clinical trials as we normally do it, we have to ship the slides, we have to meet to have consensus

72
00:12:17,966 --> 00:12:34,732
meetings, pathologists look at the same slide, not necessarily the same structure, etc., and so I just want to show a couple of things that we have

73
00:12:34,733 --> 00:12:50,766
done. Okay. So basically, this is the diagram if we use the conventional light microscopy. The slide is stained with…the stain of the slide’s fate with

74
00:12:50,766 --> 00:13:05,599
time. We have description errors if we use [---] sheet. We have to look at multiple slides, multiple structures, etc. So, the possibility of error is high

75
00:13:05,600 --> 00:13:17,166
and these are our consensus meeting, but if we use the virtual microscopy methodology we can cut down on some of these issues and we have

76
00:13:17,166 --> 00:13:29,266
can two processes. We can have digital imaging created in a data center, so slides are indeed mailed but once and don’t’ go around the world

77
00:13:29,266 --> 00:13:41,366
between the different pathologists; or we can have local scanning. Of course, this brings up a lot of QC issues and regulations and protocols

78
00:13:41,366 --> 00:13:52,099
that need to be defined up front so we have a system that is consistent among biocenters if we use that. The other thing that virtual microscopy

79
00:13:52,100 --> 00:14:04,333
can allow you, because now we’re talking about imaging, we can select the structures that we want to score, for example, measure, whatever

80
00:14:04,333 --> 00:14:13,599
we want to do. So for example, NEPTUNE introduced the annotator step, so the annotator selects, for example, the glomeruli that need to be

81
00:14:13,600 --> 00:14:21,966
scored by the other pathologists and that decrees tremendously the intra- and inter-reader viability so there is a chance that there is more

82
00:14:21,966 --> 00:14:32,832
reproducibility of whatever we want to do. I want to go fast. These are the digital slides as they can appear to you and this is what we do when we

83
00:14:32,833 --> 00:14:43,366
do the annotation. Each glomerulus, for example, for NEPTUNE receives a number and the same number is maintained throughout all the sections,

84
00:14:43,366 --> 00:14:51,366
so glomerulus number one would be glomerulus number one throughout all the sections of the biopsy. That allows us to have a multi-level

85
00:14:51,366 --> 00:14:59,732
reconstruction of the glomerulus and have a better sense of what happens to that given glomerulus. Of course, we are working with

86
00:14:59,733 --> 00:15:11,233
renal biopsies, so we don’t have a complete 3D reconstruction, but as in this example, we can see how we can follow what happens to any

87
00:15:11,233 --> 00:15:24,066
given annotated glomerulus. Multiple pathologists or multiple investigators can view the digital images at the same time on the phone and

88
00:15:24,066 --> 00:15:38,732
discuss each of these features in a consensus meeting or whatever it is. I don’t want to go into this core. I have a disclosure. I am consulting with

89
00:15:38,733 --> 00:15:53,733
Amicus and Protalix for Fabry disease and digital pathology, but this is the summary of the application of the system with advantages. So,

90
00:15:53,733 --> 00:16:04,233
multiple users; you can avoid mailing system; you can create a library of the images that can be maintained permanently with the data linked to it

91
00:16:04,233 --> 00:16:15,433
permanently; full transparency for regulatory agencies to think about a clinical trial for drug development; same slides can be used for

92
00:16:15,433 --> 00:16:26,766
multiple studies simultaneously and for future studies; the cost becomes low in the long time, the more you use it, of course; you can annotate

93
00:16:26,766 --> 00:16:35,299
the slides; you can have multi-level construction of renal biopsies; you can do morphometric analysis of certain structures, at least; and you

94
00:16:35,300 --> 00:16:42,966
can measure the density of certain structures in the renal parameter such as the case of peritubular capilliaries, and this is where my

95
00:16:42,966 --> 00:16:56,699
disclosure comes. For example, here we counted 300 capillaries per biopsy and you can do that in a precise way only if you use digital pathology

96
00:16:56,700 --> 00:17:04,766
because if you’re trying to do that using light microscopy you end up counting the same capillary twice, you get interrupted, you have to

97
00:17:04,766 --> 00:17:15,566
start again, etc. So, we mark each capillary and we can count the number of GL3 inclusion per endothelial cells in this specific example and you

98
00:17:15,566 --> 00:17:27,332
can see that appears as a number next to the arrow that identifies the capillary. This is a specific study for Fabry disease, but if you think

99
00:17:27,333 --> 00:17:38,733
about reproducibility of these data, of course now we’re looking at the same exact capillaries. So, two pathologists can disagree on a certain

100
00:17:38,733 --> 00:17:50,466
number of those but at least they are the same, and you can do a reconciliation process just pulling out those capillaries you disagree upon

101
00:17:50,466 --> 00:18:01,066
and you don’t have to review the entire biopsy, recount 300 capillaries as you would do by light microscopy, and we have shown that there is an

102
00:18:01,066 --> 00:18:16,666
increased reproducibility if you use virtual microscopy versus light microscopy using this scoring system. So, the applications are multiple

103
00:18:16,666 --> 00:18:27,499
and we are just at the beginning. I don’t know of any other studies on renal biopsy beside the Fabry and the NEPTUNE, but certainly there are

104
00:18:27,500 --> 00:18:38,533
advantages and the key, as Steve pointed out, is just very rigorous protocols, a lot of training, that is critical, and training of multiple personnel, from

105
00:18:38,533 --> 00:18:52,299
the study coordinator to the pathologist to all the elements that are involved in this process. I don’t know, Steve, if you want to add anything.

106
00:18:52,300 --> 00:19:06,466
That’s it.
BEHZAD NAJAFIAN: So, I just summarized the

107
00:19:06,466 --> 00:19:16,232
highlights of our breakout session on peritubular capillary estimations on behalf of Agnes Fogo and Mike Mauer, who left earlier, so I apologize. I don’t

108
00:19:16,233 --> 00:19:24,999
have any slides to show; that was not my plan to summarize this session. So, basically we discussed different parameters that can be

109
00:19:25,000 --> 00:19:38,600
quantitated in peritubular capillaries, including length, density, volume density, surface density, per volume or surface density per surface of

110
00:19:38,600 --> 00:19:47,966
tubules, and number densities. These are different parameters that can be estimated. The importance of appropriate sampling was

111
00:19:47,966 --> 00:19:59,566
discussed in that we need to do a systematic uniform random sampling in order to reduce the variability, as it was discussed in other sessions

112
00:19:59,566 --> 00:20:11,199
during today and yesterday. Different information we can obtain by looking at the light microscopy and electron microscopy samples. By electron

113
00:20:11,200 --> 00:20:20,933
microscopy, Mike Mauer discussed the breakdown of different components of interstitium, including the volume fraction of the

114
00:20:20,933 --> 00:20:31,733
capillaries, the surface of those, and how much of the interstitium is composed of collagen fibers and a space. In light microscopy, Agnes Fogo

115
00:20:31,733 --> 00:20:46,566
discussed the huge amount of variability that we can get in relation to interstitial fibrosis and also peritubular capillary densities. Therefore, it will be

116
00:20:46,566 --> 00:20:57,866
very important to define which areas we are interested in for quantitation of peritubular capillaries in interstitial fibrosis. Do we want to

117
00:20:57,866 --> 00:21:05,199
include areas with scarring or do we want to avoid areas with scarring where we are interested to obtain an average of those areas?

118
00:21:05,200 --> 00:21:16,400
Combination of electron microscopy and light microscopy findings can be useful, however, the pitfall would be that the areas that we choose for

119
00:21:16,400 --> 00:21:27,066
electron microscopy studies should correspond to the same areas in light microscopy if we want to correlate those findings together. Adding of

120
00:21:27,066 --> 00:21:40,932
immunohistochemical studies in order to visualize peritubular capillaries for quantitation, such as CD31 or elected markers was discussed that

121
00:21:40,933 --> 00:21:53,933
would facilitate quantitation, as well as using cell markers to study other cells that increase in the interstitium—inflammatory cells, fibroblasts, or

122
00:21:53,933 --> 00:22:05,033
pericytes—will be also informative. We do not have much information about importance of peritubular capillaries in relation to progression of

123
00:22:05,033 --> 00:22:17,066
the disease. Performing longitudinal studies would be especially helpful to see if a reduction in peritubular capillary density comes first or just

124
00:22:17,066 --> 00:22:31,732
accompanies fibrosis. One of the important challenges, of course, because of the variability, would be that basically we don’t know what is

125
00:22:31,733 --> 00:22:40,699
going on by looking at the biopsy; we may not necessarily come to a conclusion about what is going on in the whole kidney. So, a reference trap

126
00:22:40,700 --> 00:22:52,733
will be a very important issue in studying the peritubular capillaries and therefore non-invasive methods to basically study the vasculature of the

127
00:22:52,733 --> 00:23:06,699
kidney hypoxia would be very important. Jeff Kopp mentioned about the usefulness, potential usefulness, of markers of hypoxia in the biopsies

128
00:23:06,700 --> 00:23:16,633
that together, with the quantitation of peritubular capillaries, could provide important information on which one comes first in areas that there is not

129
00:23:16,633 --> 00:23:38,266
significant scarring. The importance to do additional studies in clinical trials, doing clinical studies in areas that basically contribute a lot in

130
00:23:38,266 --> 00:23:47,566
end-stage kidney disease, diabetic nephropathy, especially Type II, hypertensive nephrosclerosis and focal and segmental glomerular sclerosis, the

131
00:23:47,566 --> 00:23:59,366
importance of APO1, those were discussed and the limitations that we have so far for the availability of the biopsies for those clinical trials

132
00:23:59,366 --> 00:24:09,732
and also the limitations that we’re facing for availability of control tissue. So, the importance of doing collaborative studies, sharing the biopsy

133
00:24:09,733 --> 00:24:25,166
materials, and the control material was discussed in our session. Also, the differences between the autopsy material and biopsies obtained from living

134
00:24:25,166 --> 00:24:37,866
donors or living controls was discussed. Some of the parameters, perhaps, can be studied, can be learned, from looking at the autopsy material such

135
00:24:37,866 --> 00:24:48,499
as glomerular scarring, interstitial fibrosis, maybe those things, while some other parameters could be significant or different. So depending on the

136
00:24:48,500 --> 00:25:00,966
study, those materials become handy or there might be more limitations. That was basically the wrap-up of what we discussed. Any questions

137
00:25:00,966 --> 00:25:07,332
or comments?




Date Last Updated: 10/5/2012

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