Urologic Oncology

SLIDES & TRANSCRIPTS
Friday, December 13, 2002

Kidney Cancer

Victor Reuter, M.D.

Slide 1:

Thank you very much, Maria.

I suppose I could be more entertaining if I gave this lecture in Spanish, but I won't.

I have had the opportunity to speak in many venues, but I must tell you I have never had the distinct honor and privilege to follow two academic scholars and role models like the two that we have just heard speak, Dr. von Eschenbach and Dr. Grayhack, who serve not only as leaders for the future, but also as inspirations for us that live in the present. And for that, I thank them both for the words that we heard today of not only a recall, but also of looking forward of how we are going to treat these diseases.

And just like Dr. von Eschenbach said, we start with the patient. You start with the patient, and looking eye-to-eye. We start with the patient, looking at a tissue-to-eye. And quite frankly, our observations will go to the laboratory, and the laboratory, and then come back to us, and to the clinic.

And basically, I will give you a very quick story of what has happened in kidney cancer over the last 20 years, and why we are so excited to see that there is more money being put into this field.

When we started here in the eighties it was basically one disease, renal cancer. And in renal cancer, the higher the stage, the higher the grade, the worse the prognosis, and that is basically the way it was. Molecular biologists started working the disease somewhere in the eighties. The pathologists started describing new entities, et cetera.

Slide 2:

But it really wasn't until this gentleman that you see right here, Dr. Julakovich from Heidelberg, Germany, brought us all together, and you see some recognizable names here in the field of urology, the field of molecular biology, including Barton Zbar over here, Hogar Mach from Basel, Germany, Mohammed Ahktar from Egypt, and Bruce Beckwith in pediatric kidney cancer.

And the reason why Dr. Jula Kovach brought us together was because he said, well, you pathologists, clinicians, and molecular biologists aren't working together. You are publishing separately, and you are not talking the same language. So how about it? Why don't we start working in the same language.

And for three days we sort of bashed it out in Heidelberg, Germany. That was a real tough gig. So basically, we left there with a little bit of a different understanding of diseases, and we agreed to sort of compartmentalize them. We didn't really try to put it as a classification. That's what it has turned into, but sort of understanding the disease as biologic groups.

Slide 3:

Some were benign; some were malignant. And that with the benign and malignant diseases there were actually four or five different diseases that had a totally different biology. And lo and behold, that leads to different molecular biology. And lo and behold, they were also different morphologically. So why don't we start using common terminology, develop the programs, look at the disease differently, and see if in fact that biologically they are different diseases?

Slide 4:

So let's start with kidney cancer. The one that you feel most comfortable with is clear cell carcinoma. And those are the ones that are associated with the germ line mutation of von Hippel-Lindau gene, as you know, some of the somatic mutations. Some of these sporadic tumors will also -- most of them will also have the mutation in 3p, but basically, these are the common tumors that you recognize.

Slide 5:

One of the reasons why we don't like the term "clear cell carcinoma" is seen in this particular picture. And that is because in your left you will see your traditional acinar pattern in clear cells. But in fact, the higher the grade, the less likely it is that these tumors are going to have clear cells.

And here you see cells with larger nuclei, and glandular cytoplasm. And this is what traditional pathologists have called granular cell renal cell carcinoma.
And here you see another high-grade clear cell carcinoma that actually has some cell drop off, and has eosinophilic cytoplasm, and yet this is also molecularly a clear cell carcinoma.

So the first thing we did was redefine the disease. And rule number one is the concept of glandular renal cell carcinoma did not exist. It is just basically the end result of grade, its effect on the cytology of some particular tumors, be it clear cell or not.

Slide 6:

If you look at this, 155 consecutive tumors at Memorial hospital, you can see if you look at cytoplasm on the one side, and growth pattern on the other, that the characteristic acinar or pattern is only seen in about 15 percent of cases. And the characteristic clear cell features alone are only seen in about the same, 13 percent of tumors. Ergo, there is a relationship.

So our basic contents of what we were classifying only as clear cell, were only the lower grade tumors, but that in fact some of them granular, some are pseudo-papillary, et cetera, et cetera, and still molecular clear cell carcinomas. And what our role in pathology was, was to recognize these variants and classify them accordingly.

Slide 7:

When you look at these 150 patients, basically, you just pay attention to the top part, and I won't digest the data for you, but basically, the five-year disease-free survival, the median followed was 60 months, was approximately 62 percent, in this particular group of clear cell carcinomas.

Slide 8:

And if you looked at grade versus survival, be it disease-specific survival or not, or if I could have done the same thing for stage, you would have seen that in fact there was a good correlation between grade and stage, a good correlation between grade and survival, good survival between stage and survival. So everything was well.

Slide 9:

In the 1980s, Julakovachs published a paper, which was really a karyotypic, a traditional karyotypic paper, and said, you know, clear cell carcinomas might have abnormalities of 3p, but there is an entity called papillary carcinoma that has no abnormalities of 3p, that only has polysomies of chromosome 7, or polysomies of chromosome 17.

These are papillary carcinomas. So he classified the tumors as papillary and non-papillary. He went on to say that papillary tumors that only had the genetic abnormality of trisomy 7 were benign, a hypothesis that has never been tested.

Slide 10:

After that, much of the work done here by Barton Zbar, Maria Merino, and others, Marston also, showed that there was a hereditary part of papillary carcinoma; that these had mutations of the MET proto-oncogene, and now we understand and divide the tumors in types ones and type twos that aren't totally important to this particular talk.

Slide 11:

But again, if you look at a large cohort of papillary carcinoma, what you will see that, sure, the traditional papillary carcinomas exist. But if you apply this, and you think that this is the only way that papillary carcinomas will look, you will miss many of them. For example, look at the bottom left corner. And this is actually a solid tumor that has these tubules with cells sort of proliferating towards the lumen. We call it the glomeruloid variant. Pathologists love to give things names, or the solid variant.

And a very elegant study by Andy Renshaw from Harvard, when he was there, showed that in fact these patients, using in situ hybridization, these patients also are trisomic 7, trisomic 17, and have a normal 3. So this is a solid variant of papillary carcinoma, papillary carcinoma nonetheless.

Slide 12:

And again, if you look at a large cohort of patients, and these are 83 tumors in 79 patients, you can see that the pure variants of papillary carcinoma are really the minority, at least in this sporadic group, which are the tumors that I deal with. So you can have some that are papillary, more papillary than others.

Is there a defined percentage that a tumor has to be, to be called papillary? The answer is no. For example, the one that I showed you before, that could be solid entirely, and still is a papillary cancer.

Slide 13:

The same 60 month mean follow-up. You can see the disease-specific survival here was about 89 percent, comparing that with 62 percent that I showed you for clear cell carcinomas.

Slide 14:

Does grade work in this particular group? There is a nice study published by Malholamin from Emory University showing that in univariate analysis, in fact grade does seem to work.

Slide 15:

But as soon as you put it into multivariate analysis, firm and grading, just like you would use in clear cell, appears to fall out. And this is in fact, the experience of many people.

Slide 16:

And that is what caused Brad Delahunt and John Ebly to in fact suggest a different type of grading scheme for papillary cancers. Unfortunately, rather than saying low grade-high grade, they also chose type 1, type 2, which confused a lot of things.

Slide 17:

But the bottom line is that they chose two different levels, and then went on to show some data suggesting the type was actually more powerful, as you can see with the p value, than traditional firm and grade. The problem here of course is small number statistics, as far as the number of events of patients that died.

Slide 18:

We looked at it on 103 patients in our data. And you can see that in fact, neither using Fuhrman classification or this particular type of classification helps us in grading tumors. And that's why I'm not so certain that I am able to grade tumors. But with papillary tumors I am certain that using the Fuhrman classification does not work for me.

Slide 19:

Now, a third type of tumor is this one that you see here, that has mostly a solid pattern of growth. And you can see that there seems to be a lot of granularity eosinophyllia in the cytoplasm. What do you think these tumors were called in 1982, before Dr. Stephan Strokel and Wolfgang Tunis from Germany described them? They were called granular renal cell carcinomas.

And you will also notice -- and actually, let me go back here -- that if you look at the cytology of these cells, some of these cells look pretty angry. They are rather large, angulated, et cetera.

Slide 20:

And yet, what we know is that these tumors in fact despite the aggressive appearance of these nuclei, we have learned to ignore that, because these tumors have a much better prognosis, as I will show you.

Molecular biology has taught us that these tumors are not characterized by polysomy 7s or 17s. They are not characterized by losses of 3P, but they are losses in fact of losses across the entire genome. So these tumors are hypodiploid. They are characterized specifically losses across the entire genome.

And this is a nice little review article using LOH microsatellite analysis done by Peter Bugert and Juda Kovach, published in Laboratory Investigation several years ago, in which they show, and basically you can see that there is actually LOH across multiple chromosomes in these particular six samples of chromophobe carcinoma that I have shown you.

Slide 21:

If you look at the follow-up for these patients, this is again median follow-up of 5 years from Memorial, 60 cases. And what you will see is that in fact that the disease-specific survival here was about 92 percent.

Slide 22:

If you look at the Mayo Clinic data, you will see that their data is 94 percent. What is interesting here is that while I don't show you the sizes here, but the mean size in our particular series was the largest of them all. It was 9 centimeters, and it was about the same for the Mayo Clinic data, and here you can see that it's 8.5 centimeters.

Slide 23:

So despite the fact that many of these tumors appears to be extracapsular, about 25 percent, despite the fact that these tumors as a group were extremely large, they seemed to have much better prognosis than clear cells, and much better prognosis also than papillary carcinomas.

And finally, oncocytomas, and you can see this hypovascular stroma in the background, the nest of the eosinophilic cells. They look rather mahogany brown. Why? because they contain abundant amounts of mitochondria. Unfortunately, chromophobe carcinomas can do the same.

Slide 24:

Dr. Faro and Dr. Leiber and Dr. Rainwater from Mayo Clinic published a couple of a little bit concerning papers, despite the fact that we thought that oncocytomas were benign tumors in the middle eighties, because they showed -- you know the interest at Mayo Clinic in ploidy -- they showed that 39 percent of their tumors that they had called oncocytomas, were hyperdiploid, and 11 percent, aneuploid.

So the ground started to shake a little bit underneath us, but we understand now the fact that if you were to do ploidy analysis for example, in seminal vesicle epithelium or in mature teratomas, you are going to get a large number of hypodiploid and aneuploid tumors. So taken out of context, these particularly large pleomorphic cells that you are seeing, are actually degenerative in nature.

So that if you were to do proliferation studies, and we have done them in cells like this, you would say that these cells are in fact non-proliferative. So they are degenerative in nature, and should be ignored, be they present in chromophobe tumors, or be they present in oncocytomas. Obviously, oncocytomas you would not grade, because they are, by definition, a benign tumor.

Slide 25:

I'm not going to go into in detail, but we published a paper showing an entity in which patients had multiple oncocytomas, and I'm talking about a myriad of oncocytomas. Some of those lesions were hybrid between oncoctyomas and hybrid tumors and chromophobe tumors, and we come up with a hypothesis that there might be hybrid diseases, and we have used the term oncocytosis.

When we published that paper, Maria Merino, Barton Zabar, Marston Linehan invited us quickly over to the NHI and said "could you please look at some cases for us?" And I had no idea what they were talking about. And recently, they have published the morphologic manifestations of the renal tumors in the Birt-Hogg-Dube syndrome. It is a very, very important paper.

More important than the paper is actually the genetic information that will come out of these patients, because if you look at the morphology of these patients, in fact there are some 9 percent that also have clear cell morphology. Could there be a missing link, or could this family teach us what are the really initiating events of renal carcinogenesis, and what are the following steps to take it one pathway or the other? It is a very exciting area.

So basically, they have built on what we said with oncocytosis, and taken the syndrome of Birt-Hogg-Dube to what I believe to be a different level, a very exciting level.

Slide 26:

So in summary, I have not discussed all renal cancers, but I have discussed most of the renal cancers that matter to you in clinical practice. Yes, we believe they arise from different cells in the nephron. Does that terribly matter to you? Probably not.

What does matter is that these tumors are genetic entities; and that these genetic entities in fact, have a morphologic correlate. It's a morphologic correlate that is different from the one that you learned in medical school. And in fact, it's still being decided.

Slide 27:

Clear cell carcinoma is the most common. You see the term "conventional" here, because we wanted to get away from the word clear -- are characterized by losses of 3P. There are a couple of others that you will see that can have other losses and gains, depending on the grade.

Papillary tumors, polysome 7, 17s, C-Met proto-oncogene mutations: oncocytomas, losses of Y, losses of one: chromophobe, the same, plus losses across the entire genome. I am not necessarily going to go into the others.

The importance to the pathologist and the importance to you as a treating physician is that now there are biologic correlates to these names. And hopefully, they are much tighter than they were before. When a pathologist sees a tumor with pink cytoplasm, it is not just a granular renal cell carcinoma. There is a differential diagnosis, because glandular cell can be seen in many types.

If you see spindle cells, sarcomatoid carcinoma, there is a differential diagnosis. And if you say well, it really doesn't matter, because those are high grade disease, mais non, because as you can see here, this particular term, lipoma, is basically a low grade collecting duct carcinoma. Now there are about 35 or 40 of them in the literature. It's rare. I know of only one case with regional node metastasis, and no case that has died of disease. So in fact, even in the spindle cell category now, we can subdivide by type and genetics into higher grade and more indolent tumors.

So this is in summary, the cases from Memorial that I have shown you before. Clear cell carcinomas are the ones that are more likely to have disease outside the organ, but it's not exclusive, as you can see here.

Slide 28:

As you can see, the five year disease-free survival was 62 percent. Papillary cancers are the most likely to be multifocal, better five year survival, and chromophobe renal cell carcinomas, better prognosis despite the size.
And despite that many of them actually have extended outside the organ, and oncocytomas, basically, a benign tumor that I know of from me, or from talking to my dedicated urologic pathology colleagues, I do not know of a single case of oncocytoma that has died of the disease.

Slide 29:

What is really interesting is that this is actually a very good article written by Amin and his group using the Emory information, the Henry Ford patients, et cetera. And if you look at the breakdown of histologies, it's almost identical to what I had said before. Conventional cancers do not comprise 85 percent of tumors, as you might hear that in the old literature. It's about 60 percent, because now we have divided the groups a little bit better. And if you look at the breakdown and the size, et cetera, it's very similar to what I said before.

Slide 30:

Are there morphologic correlates to behavior in each one of the types? Yes, and it differs in each one of the types, and this is the way the disease should be looked at now, not as renal cancer, but as papillary cancer. What are the markers there, et cetera?

Now, there is one issue, if you can see here, the category that Amin published in which he called renal cell unclassified, that I have not talked about. And if you look at the five-year outcome, you will see that in fact many of these patients have died of disease, so the disease specific survival at five years was only 24 percent.

And you say, well, you didn't talk about that one. That's very important. I think that is a problem, because the concept of us not knowing how to classify some tumors is real.

Slide 31:

And as you can see here in my hands, I cannot classify using traditional morphology electromicroscopy approximately 7 percent of tumors.

But what is most important about this is, and this is a series of 1,100 consecutive kidney cancers resected at my joint.

Slide 32:

But, what is interesting is that when you look at okay, why weren't you able to classify them?, It is not as if they are all high grade tumors. So look at my first category, which actually is the largest one I have. It's 42 percent. It's whether it is an oncocytoma or eosinophilic variant of chromophobe renal cell carcinoma. Those are either benign, or extremely low-grade malignant tumors.
It is here, where I think from a diagnostic point of view, we can avail ourselves of modern molecular techniques to be able to make us better diagnosticians, and make you better predictors of how this patient might behave.

Slide 33:

Now, this is a very nice slide that was lent to me by Fred Wolman. This is work that he did with Dan Pinkle showing CGH array. And this was totally unsupervised, looking at renal cancer. And what you can see very nicely is that of these 42 tumors, the computer was able to cluster correctly 40 out of 42 tumors.

And you can see on the top line, and you can see here all the losses. Reds are losses; greens are gains. This entire group with clear cell carcinomas, conventional, you can see gains and gains, typical for polysome 7 and 17s. These were all papillary; losses across the entire genome. These were chromophobe carcinomas and oncocytomas down here. So in fact, an unsupervised CGH was able to do this.

Slide 34:

This has been duplicated now using the Affymetrix chip. And this is a paper written by Amin's group, Young last year, and can see in fact again how the chromophobes and oncocytoma very interestingly, are appearing to cluster together, and the conventional carcinomas can be separated.

Slide 35:

Now, all of this information from a diagnostic point of view is important. We understand, and we have to understand, that it's not only how big the tumor is. And we understand that it's not only what the grade of the tumor is, and clearly, all those are type-specific. There is symptomatology that matters. There is grade. There is morphologic criteria that matters.

And UCLA and Memorial have tried to look at this, and this is the paper that was published by Dr. Kattan form our institution, trying to come to grips, and trying to allow us to better assess what the prognosis of the patient will be. It's a first step, and none of us feel that it's the end step.

One of the things that I would criticize of a paper on which I am one of the authors -- I think I'm the 27th author -- I don't have grade. And I told you that for clear cell carcinomas, I think that grade is important, so for example.

The other thing is that it's nomogram based on tumors that are alike. That's like putting oat cell carcinomas together with bronchioaveolar carcinomas with squamous cell carcinomas. But we know that. The next step and the next step we will try to develop more robust information, targeting specific diseases, rather than groups of diseases. So we understand this, and the next permutations of ours, and I'm sure of others are going to be much more robust, and give you much better information.

Slide 36:

Slide 37:

Now, there is another way of looking at this, and obviously, you have heard a lot about targeted therapy. And targeted therapy is what it's all about. And one of the things that we are doing here is trying to see comparing tumors of metastasized versus it not. And obviously, if we are going to do this in a gene-array way, we are going to do it within like.

So this is data that now we are working, and this is all preliminary, sort of first, second step, and I'm sort of keeping a lot of things away not from you. But the bottom line is if you look at these circles, basically the top two are comparing seven primary tumors to their paired, sampled metastases, and these were all visceral metastases all to lung.

On the bottom you have 25 unpaired metastatic lesions. And what you can see is in fact there are 62 amplified regions, and 13 shared amplified regions between the paired metastases and the primary. But interestingly, there are 63 unpaired amplified regions only in metastatic disease. But you can see that also two of them are shared in the primary and the metastatic disease.

So this is the way that we are going to try to approach, as well as others, trying to identify those areas that in fact might be of interest, and try to see if we can identify genes, and then develop markers to be able to see if we can assist in predicting those patients that are going to have the worst outcome, number one.

Or predicting those patients that might or might not respond to compounds that are definitely on their way as far as treatment for these patients, especially a lot of agents along the anti-angiogenic pathways, but many others that other people are working with as well.

Slide 38:

So in conclusion, we have a bunch of challenges. First of all, we have to understand that a contemporary classification which takes into account pathologic and molecular data has defined, clinically distinct tumor types that have revolutionized how we understand the concept of renal cancer.

Number two, that the clinical relevance of the traditional morphologic criteria of for example, stage and grade, really are all tumor type dependent.

Number three, we have to define better prognostic markers within tumor types, and within tumor stages. And something that I have in another slide that I should have put for you as surgeons is the fact that you have to revisit the concept of how I handle solid renal masses, because, solid renal masses are a heterogeneous group of lesions. And soon, within the next decade, you will take into account the host, the patient, in trying to assess how you are going to evaluate any given solid mass.
Within a very short time, you are going to asking the pathologist to evaluate that solid mass in situ. And you will also hopefully, with more modern, robust metabolic imaging, you will be able to be helped also by our radiology friends as well, because the disease, the actual entity will define how you might or might not want to handle a solid renal mass, for example, in an 83 year old lady with marginal renal function.

Number four, define markers of response to therapy within tumor types, targeted therapy. And clearly, as I said, new compounds or first compounds will be on the way very shortly. And we also have to define, as I have just mentioned, strategies for the classification of renal epithelial tumors in vivo to help you, and guide therapy at the front end of the disease.

And with that, I thank you very much.