CME Slides Forum - H.P. Marti

NEW MARKERS OF CHRONIC ALLOGRAFT NEPHROPATHY

Hans-Peter Marti, Bern, Switzerland

Chair: Josep M. Grinyo, Barcelona, Spain

Yves Vanrenterghem, Leuven, Belgium

marti

Dr H.P. Marti
Division of Nephrology and Hypertension
University of Bern
Bern, Switzerland

Dear Chairman, Ladies and Gentlemen, thank you very much for the opportunity to present you some mostly ongoing studies devoted to the topic of the evaluation and definition of new markers of chronic allograft nephropathy. Since it’s an emerging field I’ve also included data of acute allograft rejection.

First, after a brief introduction I’d like to lead you through the field of biomarker identification and validation.

So, first what are the main problems in transplant medicine? Of course, the shortage of organ donors, then second the need to increase efficacy to control graft rejection safely, predictably and durably. Therefore, the next generation of immunosuppressive drugs should be more selective and safe. Thus, we put forward the hypothesis that new biomarkers (diagnostic and predictive) of acute and chronic allograft rejection should increase our understanding of molecular mechanisms and should help to identify and validate new drug targets.

Second, I’d like to give you some definitions of important terms. First what’s a biomarker?

A biomarker is a characteristic feature that is measured and evaluated as an indicator of normal biological processes and also pathogenic processes or response to a therapeutic intervention. A clinical endpoint means a variable or characteristic measure how a patient feels, functions and survives. In our case we work with surrogate end-points meaning that we like to identify a biomarker that can substitute for a clinical endpoint.

So far in the literature there have been a number markers described for acute rejections like intragraft cytokine expressions, granzyme, perforin, FasL and cytokine growth factors. There have also been put forward some markers of chronic rejection on the histological level, Sirius red staining with quantification of the interstitial volume affected. Then on the electronic microscopic level lamellation of capillary basement membranes and again cytokine expression.
Furthermore, there have been put forward some genetic markers of immune responses and also pharmacodynamic biomarkers.

Now, what’s the situation in reality? We have some biomarkers in clinical kidney transplantation. First organ function assessments mostly serum creatinine and creatinine clearance of course, however, they have a low specificity and if the creatinine has risen, allograft damage is already present as also shown by the previous speaker. Histology biopsies are invasive and the accuracy to compare immunosuppressive drug efficacy is also challenged and there are many confounding factors on histology like nephrosclerosis, diabetic nephropathy and so on. Thus we think novel biomarkers are clearly in clinical demand.

First I would like to show you some ongoing studies of a rat kidney allograft model, how we and our collaborators from Novartis, Doctor Raulf tried to identify novel biomarkers.

The hypothesis in these experimental studies was that early allo-response markers and early organ function parameters can be developed as new diagnostics for renal allograft rejection. The approach was to do a comprehensive unbiased profiling of a rat kidney transplant model by several approaches and we hope to identify a diagnostic pattern.

In the laboratories of the Transplant Institute of Novartis they use the acute rejection model Brown Norway-to-Lewis rat, it’s a life-supporting orthotopic kidney transplantation model. Graft survival 7 days and 3 groups of 10 allo- and 10 isografts were studied among other controls each terminated at day 3, 4 or 5.

The approach was a comprehensive profiling doing transcriptomics from blood and kidney grafts by affymetrix and PCR, doing metabonomics from urine and blood with mass spectrometry, assess the different model parameters like kidney function, histology and doing proteomics from urine and blood with several methodologies.

So the first results from affymetrix genechip expression, Pair wise analysis comparing 3 day blood from allografts versus controls also 4 days and 5 days developed several probesets which have been clearly different to 600 genes were different in acute rejection versus controls at day 3, 4 and 5. We did an overlap and 28 probesets are clinically significantly different at all 3 time points meaning day 3, 4 and 5. We selected 22 genes which were at least 2-fold differently expressed as acute rejection marker candidates.

The 22 candidates have been defined in these blood transcriptomics a various arrays of genes from caspases to chemokines and other growth factors. We did the validation method by clustering of samples.

The clustering of samples was based on the expression levels of the selected genes. We use this method and it’s an unsupervised method, it means the software doesn’t know to which group the individual sample belongs.

By clustering the results of the blood samples the 22 probesets which defined acute rejections were managed to cluster all acute rejection blood samples correctly here allografts with acute rejection on the left panel and right the healthy isografts and most of the genes are upregulated. So by using these 22 different genes, you’re able to classify all samples from acute rejection to healthy control.

The next question was, do these rat acute rejection markers also work in monkeys? There we performed a monkey renal allograft rejection model and did a cross-validation and using 17 out of these 22 probesets were able to also separate normal monkey kidneys versus kidneys of monkeys with acute rejection.

So, these results from the rat model were able to also classify the monkey renal allograft samples. So to summarise on the experimental level the acute rejection marker set correctly identifies acute rejection in blood and also in kidney graft samples in rats as early as on day 3 preceding the rise in serum creatinine of day 5 and graft loss at day 7. These marker sets of 22 genes were also able to correctly cluster monkey kidney biopsies and mouse acute rejection heart samples. These results I did not show due to lack of time. So, therefore, the candidate markers are not species specific nor are they organ dysfunction markers. So we think these 22 marker genes could potentially be used as alloresponse markers of acute rejections across species and across different transplanted organs.

Now, I’d like to move on to the topic of biomarker identification using clinical biopsies.

First I have to present you the landmark study of Sarwal et al published in the New England Journal of Medicine a couple of years ago. The authors from California they did DNA microarrays from 67 biopsy samples from 50 patients, they’re mostly paediatric patients with renal allograft mostly 1 month and 10 years post transplant.

The authors were able to differentiate these samples showing acute rejection into 3 groups 1, 2 3 based on transcript or gene expression profiling. So according to the different genes, differentially expressed they classified the acute rejection into 3 different groups and this has prognostic value because the 3 different groups had a different probability of graft survival here in the acute rejection group 3 from 5 grafts they survived and 0 incomplete recovery and the group 2, 9 grafts with 3 incomplete recovery and the group 1 with the worst prognosis. So gene expression profiling can be able to be used to classify acute rejection into several groups with prognostic value.

One of the most important findings was the identification of CD20 as a marker also of acute humoral rejection having a bad prognosis of CD20 positive samples from 9 that had 8 graft loss or incomplete recovery whereas when CD20 was negative then they only had 1/11 graft losses. As a consequence they had even therapeutic interventions with rituximab being used to treat acute rejection so using transcriptomics they were able to classify acute rejection in different groups and also define a new therapeutic intervention.

Another approach to find markers of acute rejection has been presented to you yesterday from the group of Professor Mueller from Germany. They used proteomics from urine samples and here is --- of urine profiles, here normal controls stable transplants and patients with acute clinical rejection, tubular necrosis, glomerulopathy and urinary tract infection and you can see on this gel-view that the protein pattern detected in the urine, difference in samples from acute rejection versus the other groups of patients. So they conclude that proteomic technology together with stringent definition of patients can detect urine proteins associated with acute renal allograft rejection. So this is another approach that may lead to the identification of specific markers of acute rejection together with the transcriptomic approach.

Now, I’d like to move on to chronic allograft nephropathy. I’d like to present you first results from an ongoing study which has been designed in Paris at the INSERM unit of Hopital Tenon together with Alexandre Hertig and Eric Rondeau and the microarray analyses have been performed by the statistician and members of the transplant institute at Novartis, Pierre Saint-Mezard and Friedrich Raulf.

We collected a collection of kidney biopsy samples and also blood samples from most of the patients. At least 50 nanograms of mRNA were required and we had several groups of patients. Most important patients with acute rejection, chronic rejection and 2 different controls, normal renal allograft biopsies and control kidney from nephrectomy. These biopsies have been done on clinical grounds and they’re not protocol biopsies. At the end we had 80 samples that were usable for further analysis.

We processed the samples, the renal biopsy fragments and also currently blood cells from concurrently --- blood samples, we isolated total RNA, did quality checks with --- electrophoresis excluding degradation and contamination. It was transferred to cDNA labelled, fragmented, hybridised on affymetrix microarrays and then scanned and analysed by the biostatisticians.

As far as the statistics are concerned we did a three tier analysis. First the ANOVA, second application of Shrunken Centroid or genetic algorithms to determine the features or genes that contribute mostly to discriminating patterns. Then also pathway enrichment analysis.

So we started with the affymetrix genechip arrays, analysed the data with statistical filtering normalisation then proceeded to these 3 statistical tests to allow gene signature and biomarker detection and also potential targets and we also compared the results obtained with 5 independent datasets from other studies.

So, just to explain you two of the most important methods, one is nearest Shrunken Centroid method, it’s a supervised method to identify subsets of genes that best characterise each class and they rank the genes based on their contribution to the classification.

The other was the pathway-centric analysis, it’s the gene set enrichment analysis. You start by sorting data by expression ratios between conditions, project known pathways onto the data and identify pathways significantly enriched at the top and bottom of the list. So the genes, the differently expressed were assigned to different pathways. This method has the goal to detect modest but coordinated changes in the expression of groups of functionally related genes. So for instance the group of genes expressed by T cells or matrix related genes which may just move on the --- level or modestly but as a group together still might be important if you analyse the whole pathway.

First the samples of acute rejection by the pathway analysis we observed in acute versus controls increased expression of genes related to activation of B cells, macrophages, T cells and so the genes mostly upregulate and reflect the recruitment activation of antigen presenting cells, B cells, macrophages, dendritic cells and also T cells as could be expected.

Then we did a metanalysis, clustering of different datasets, we used 36 out of the old genes that are be able to be used to classify acute rejection and analyse 3 different datasets here from Novartis non-human primates and then published study from Flechner et al 2 years ago and our ongoing studies and using the same 36 genes defined in our study we were able to clearly classify acute rejection in non-human primates in our samples and also in a sample published 2 years ago. So by defining a set of genes one can be able not only to define acute rejection in their own samples but also in other published datasets.

Then we moved on to chronic allograft nephropathy, these are the initial results. We obtained a list of bit less than 1000 genes that changed significantly between the four groups control, allograft nephropathy grade III, grade II and grade I. Part of the genes were more upregulated another part of the genes were downregulated and by using this list of these genes one is able to differently classify the different stages of chronic allograft nephropathy in humans. Here a bit more clearly you can see a heat map pattern of these 937 genes.

Here you have the control grade I you see there’s already a difference between control grade I and then even more so between grade I, grade II and grade III. So, one can classify the different histology grades also based on the expression pattern of these 937 genes.

Then among these genes we selected 33 genes prioritized for chronic allograft nephropathy and using these 33 genes we have been able to differently classify grade I, grade II and grade III. So we extracted the 33 most important genes among the 900 and also already with these 33 genes one can classify the different grades of chronic allograft nephropathy.

Then taken again these 33 genes in a gene-centric analysis we were able to correctly predict the grade of rejection, the different grade and control in our renal biopsy samples. So 33 genes can be used to predict the different grade and also to classify the different biopsies.

So if you look at the key events in chronic allograft nephropathy, some of them are common in acute rejection also and some of them maybe more common or more predominant in chronic allograft nephropathy.

Thus we did a pathway centric analysis and we obtained the result that if we compare grade III chronic allograft nephropathy versus control we also obtained an upregulation of genes related to activation of T cells, B cells, macrophages and also related to extracellular matrix production, metabolism and genes related to TGF-beta. On the contrary, downregulated pathways described in kidney loss of function are mostly devoted to transport and metabolism.

So going back to the pathogenesis of chronic allograft nephropathy we have as already shown by the previous speaker we have antigen dependent mechanisms, antigen independent mechanisms both leading to tissue injury. Two different expressions of cytokines, chemokines, growth factors and at the end alteration in extracellular matrix deposition and fibrosis. So in a --- analysis we asked the question if genes related to extracellular matrix metabolism can also be used as markers to classify chronic allograft nephropathy.

So we defined an expression profile of 80 metzincins and related genes. Metzincins are matrix metalloproteinases, ---, meprin so proteases involved in the metabolism of extracellular matrix and by using these 80 genes we have also been able to separate the samples in control grade I, grade II and grade III. Upregulated genes in brown and downregulated genes in green.

So by using metzincins genes related to matrix metabolism we can also separate the different grades of allograft nephropathy versus controls and the key genes have already been confirmed by RT-PCR. Examples are MMP-2, MMP-7, MMP-8 and TIMP-1 which have been clearly upregulated especially in grade II and grade III allograft nephropathy which may serve as marker candidates for this condition. These results are in line with a study published by us recently using a Fisher to Lewis rat kidney transplant model of chronic rejection by PCR and microarray analysis most MMPs and TIMPs have also been upregulated and also in histology upregulation of MMP-2 but here downregulation of MMP-9. Most genes are in the same way dysregulated in the human situation however, there are exceptions like MMP-9 which is downregulated in the rat condition but was upregulated in the human specimen of chronic allograft nephropathy.

So, so far the gene expression profiling of the samples they allow changes which are consistent with the pathophysiology of acute and chronic rejection importantly dynamic changes were observed in gene expression signatures from chronic allograft nephropathy biopsies from grade I to grade II, an increasing upregulation going on from grade I to grade III of the marker sets. It provides gene signatures able to classify samples of acute rejection or chronic allograft nephropathy.
Pathway analysis identified a panel of genes relevant for fibrosis progression most importantly metzincins and the related genes may represent novel markers for allograft nephropathy.

What has been published in the literature? Just to cite some of the most important studies one by Hotchkiss et al in Transplantation just recently published. The authors used a differential expression of profibrotic and growth factors in chronic allograft nephropathy. First they also have been able to differentiate between controls and chronic allograft nephropathy by using transcriptome analysis and control biopsies here, they demonstrate a different global gene expression pattern than samples of chronic allograft nephropathy as already shown by us in this ongoing study before.

In relation to matrix genes they also defined genes which have been upregulated like TGF beta induced factor, thrombospondin, PDGF-C in chronic allograft nephropathy whereas other genes were downregulated.

They performed immunohistology, TGF beta clearly upregulated on immunohistology and also on the PCR in allograft nephropathy. VEGF was downregulated on the glomerular level but not on the vascular level and other factors have clearly been upregulated in allograft nephropathy relating to the matrix metabolism like firbonectin and MMP-7 sort of going in line with our study.

A very nice and elegant study has been published by Eikmans last year. They analysed patients with acute rejection and divided them in progressors and non-progressors. 10 patients who had graft loss through chronic allograft nephropathy following acute rejection were classified as progressors and 8 patients had a stable graft function over time after acute rejection were classified as non-progressors. --- beta globulin was downregulated in the groups progressing from acute rejection to allograft nephropathy as compared to the non-progressor patients with acute rejection not progressing to chronic allograft nephropathy and also the calcium binding protein A8 was downregulated by progressors and the calcium binding protein A9 was also downregulated. However, the surfactant protein C was upregulated in the progressors. So these authors defined markers that allow in chronic rejection to predict whether patients are going on to develop chronic allograft nephropathy or not.

The preceding speaker stressed the importance of doing protocol biopsies which is very important to see also in marker studies whether some changes in markers actually occur before overt kidney damage. I’ll just cite one of the studies by Scherer et al published in 2003. They did gene expression profiling with renal protocol biopsies at 6 months post transplantation.

They had 2 groups, one normal histology at 6 months at the time of biopsy and also at 12 months and the pre-chronic rejection group also normal at 6 months but they developed chronic rejection at 12 months. They asked the question if there are some markers at the time of the protocol biopsies that determine or can be used as prognostic factors to predict whether graft function stays the same or whether patients develop chronic allograft nephropathy. There are about 10 predictor genes which have been upregulated in the group who was about to develop at a later stage chronic allograft nephropathy and some of them have been downregulated at 6 months in this group. So this is certainly the way to go to use protocol biopsies to define the marker set which can be used to predict the development of chronic allograft nephropathy.

I think that gene expression profiling and proteomics as shown by others yesterday may lead the way to the detection of diagnostic and prognostic markers for chronic allograft nephropathy. The first data from our ongoing studies from France allowed a reasonable classification in accordance with histology between normal tissue and the different stages of chronic allograft nephropathy.
However, in future studies the validation of diagnostic, as well as predictive rejection markers needs to be done and is already ongoing especially metzincins genes related to matrix metabolism maybe used as a marker set.

Also we have to move from invasive to non-invasive rejection biomarkers such as blood transcriptomics, I showed initial experimental results who defined a marker set in blood leukocytes which corresponded to the kidney samples and have also been able to be used across species.