Read the Full Video Transcript

Sam Chang: Hello everyone. This is Sam Chang. I am here in Nashville and looking forward to actually having an excellent presentation today from Dr. Bishoy Faltas, who is the Assistant Professor of Medicine, Cell, and Developmental Biology, at Cornell University at The Weill Medical School there.

Bishoy, thanks so much for spending some time with us, and as always, appreciate your insight in examining the germline and somatic interactions in the bladder cancer kind of continuum. Welcome.

Bishoy Faltas: Thank you, Sam. I really like any opportunity to speak with you and discuss all the new aspects of bladder cancer research for our patients. And I'm grateful for the opportunity to talk to you today about this story, and some of the recent work, and the recent data that is emerging in the role of germline mutations in bladder cancer patients.

So these are my disclosures.

So let me start out by talking about, giving everyone a brief introduction about germline variation as a source of missing heredity in urothelial bladder cancer. So throughout the rest of my talk, I'll be using the term urothelial cancer because as you know, that is the most common histology of bladder cancer. I will discuss our approach for identifying deleterious germline variance from whole-exome sequencing data. And I'll go over some of the results of our analysis of the functional impact of these germline variants on protein structure and function.

Germline variants are variants that we are born with. They can be transmitted to progeny, and because they are inherited, they are present in all cells of the body with some exceptions. For example, postzygotic events that result in a mosaic pattern, or some degree of heterogeneity.

This last point is actually very important because when we do target whole-exome or whole-genome sequencing of a tumor, the germline mutations are computationally subtracted, and therefore unless we asked specifically for germline testing, the germline data is largely discounted from the clinical management of bladder cancer patients. So it's not really wrong to use the iceberg metaphor here, in the sense that germline variants require an intentional and active search process to look for them, especially in patients with cancers that we do not necessarily think of as having a strong hereditary component, bladder cancer being one of them.

And there are several challenges for integrating germline variants into the routine care of bladder cancer patients that may have contributed to this being under-studied. These include having the proper regulatory frameworks for studying them clinically, having mechanisms in place that will protect patient anonymity, bioinformatic pipelines to identify truly deleterious variants, and finally, in my opinion, the lack of ground truth for a lot of the functional annotation data for the majority of germline variants that are specific for individual cancer types is a major issue that limits our interpretation and actionability of this information.

A very well-known example, probably the most well-known example, of the involvement of germline mutations in urothelial cancer, is the association between upper tract urothelial carcinoma, or urothelial carcinoma in general, and mismatch repair deficiency.

This is a very well-known association. And as I mentioned, Lynch syndrome, which is caused by a loss of function mutations and economical mismatch repair genes in urothelial carcinoma in general, upper tract urothelial carcinoma in particular. We know that Lynch syndrome patients have an increased risk, up to 22 fold, of developing upper tract urothelial carcinoma. Over the general population, we know that these patients tend to be younger. It is important to recognize patients with Lynch syndrome, because they tend to have a higher tumor mutational burden, also a higher burden of microsatellite instability, and these have therapeutic applications for treatment with immune checkpoint inhibitors. However, it is also important to note that overall, the majority of upper tract urothelial carcinomas are actually sporadic.

Interestingly, even in these sporadic upper tract urothelial carcinomas, If you look at the mRNA and the protein levels of the canonical mismatch repair proteins, such as MLH-1, PMS-2, MSH-2, MSH-6, and we and others have shown this before, we see that they are actually lower in the tumors, compared to patients with sporadic bladder cancers.

However, I'd like to emphasize that this downregulation, most of the time does not actually translate into an increase in microsatellite instability, as you can see here. In fact, we saw that the tumor mutational burden in some cases was actually lower in our sporadic UTUC cohorts, compared to bladder cancer. This is likely because one needs complete loss of these MMR proteins, which is usually the case with germline loss of function variance, to result in downstream microsatellite instability.

This has been confirmed in the MSJ cohort. As you can see here, the tumor mutational burden of the sporadic upper tract urothelial carcinomas, which are shown here in yellow, is quite low. And there are other studies that are confirming these observations.

Now I'd like to expand a little bit beyond Lynch syndrome and talk about germline mutations in DDR genes. This is from a review that was recently published by us, by our group, working with Dr. Sonpavde at the Dana Farber Cancer Institute. Looking at three recent publications, one from Dana Farber, one from Memorial Sloan Kettering, and one from our group, showing a high prevalence of germline variants in about 20% of urothelial cancer patients. This is something that we are starting to realize now. It depends on the definition of what are germline mutations and which ones are deleterious, but we are certainly starting to appreciate that this is a lot more common than we have known before.

What led us to this, at least our group, is that we know from epidemiological studies that were done in the past, and this is one of them, that there is about a 30% hereditary predisposition to bladder cancer, which is almost the same as what one would see with other cancers, such as breast cancer, which we traditionally think of as having a much more prominent role for germline mutations.

Sam Chang: Bishoy, I'm going to stop you there. So from a urologist standpoint, and someone who sees different types of urologic cancers, bladder cancer, we've never really sought a family history, as opposed to our prostate cancer patients. And so should we be asking our patients who... It makes sense for someone who doesn't have a lot of risk factors, but should we be screening all our bladder cancer patients with basically a family history profile?

Bishoy Faltas: Thank you, Sam. This is a great question. I would say the answer is, absolutely yes. I do this now routinely with all my patients, where I start taking a much more careful family history. And I think this is probably an example of when it would identify more cases of familial cancers, and it may or may not be bladder cancer, with a more careful family history. And essentially, the more we look, the more that we're likely to identify that.

But to give you an example, in our published cohort, which I will show you in the next few slides, we have a fairly high prevalence when looking at the medical records of these patients, either having a second primary cancer, and it's hard to know whether that is genetic, or because of other common environmental exposures, such as smoking and lung cancer, for example.

But we also have a significant number of patients who have a family history of cancer at a fairly young age. I think this is something that we urologists, medical oncologists, and everyone taking care of these patients, should pay more attention to.

Sam Chang: Excellent.

Bishoy Faltas: So we started looking at this question, and trying to understand really what is the source of this missing heredity in bladder cancer? And that is where our recent publication comes in. So to answer that question, we used whole-exome sequencing to identify putative deleterious germline variants, or what I refer to you as pDGVs, in urothelial cancer patients. And this is an overview of the study.

We performed whole-exome sequencing of germline samples in 157 tumors from 80 patients with advanced urothelial cancer at Weill Cornell. We developed a new computational tool, or workflow, called DGVar, to identify pDGVs using a series of different filtering steps. We adopted a strict functional definition, by only including variants that are either known to be pathogenic on the ClinVar database or the truncate known tumor suppressor genes. And then we compared the alia frequencies of pDGVs in our urothelial cancer cohort, to the TCGA urothelial cancer cohort, and with other patients with other cancer types, and over 13,000 exomes from non-cancer subjects as controls. And finally, we performed several additional analyses to understand how these germline variants are impacting protein structure, protein function, and how they are interacting with these somatic genomic alterations in the tumors.

And as you can see on this CoMut plot, most patients in our cohort were males, smokers, and the majority had metastatic disease. 90% were of European ancestry, and 49% had a history of a second non-urothelial primary, and 50% had a family history of cancer in a first-degree relative, as we were just discussing. We identified 61 pDGVs, in 45 out of 80 patients, or 56%, in our cohort, which is actually a bit high. But I would remind you that the majority of these are actually [inaudible 00:11:46] stopped gain or frameshift mutations in tumor suppressor genes. And when we performed a similar analysis in using the raw data from the TCGA bladder cancer cohort, looking at the same exact genes, we found that we can identify these pDGVs in 48% of the TCGA cohort.

The pathway analysis of these genes showed enrichment in the DNA repair pathways. The genes that were involved here were palQ, palK, FANCA, Xba, and BRCA2.

When we looked at these pDGVs in the gene set of 158 genes that we have looked at, we saw that these were significantly enriched in the Weill Cornell Medicine and the TCGA urothelial cancer patients, compared to ethnicity matched non-cancer subjects.

And then when we looked at the TCGA bladder cancer cohort, and also, actually all the TCGA Pan-cancer cohorts, we find that urothelial cancer was one of the top cancers, near ovarian cancer, that is associated with significant enrichment of pDGVs, compared to non-cancerous subjects.

So we then tried to gain some insight into the functional impact of these pDGVs, and for that, we performed the following analyses.

We found that they had a higher CADD score, which is used to predict functional impact, compared to randomly selected background variants, suggesting a high predicted effect, or downstream impact, on these proteins.

Because these genomic variants, if they were conferring, actually, a fitness advantage in the tumor cells, we hypothesized that they would tend to aggregate in functional protein domains, and we tested that hypothesis. And as you can see here, we found that they are actually forming distinct topological clusters with known somatic cancer mutations within three-dimensional protein structures, as you can see here on this bubble plot. We found that 27 out of the 28 pDGVs that had available structural information for the protein, that the germline mutations we identified clustered with known somatic variants.

Just to give you an example of how these clusters look, these are lollipop plots. I would like to focus on one particular patient where we identified a pDGV in xeroderma pigmentosum complementation group A, or XPA, in a urothelial cancer patient in our cohort. And this resulted in a stop codon that eliminated the DNA binding domain near the C-terminus of the protein, which is highlighted here in gray. What we did next is we confirmed the presence of that mutation using Sanger sequencing of the patient's germline DNA.

We then tried to understand what this alteration is doing, and XP is actually a component of a multi-subunit protein complex, called the XPA transcription factor 2H complex. And when we superimposed the truncating germline variant on the structure, we saw that it eliminated the DNA binding domain, as you can see here, which is that this entire alpha helix would be eliminated, but based on this mutation, suggesting that it would significantly disrupt DNA binding. And we also saw that this mutation clustered with known germline mutations in patients with Xeroderma Pigmentosa that have been described before.

We then turned our attention to understanding the nature and the biological consequences of these pDGVs in tumors. And the conceptual framework that is underpinning these analyses is the Knudson two-hit model, which suggests that tumor suppressor genes require activation of both alleles to cause phenotypic change.

And we indeed found evidence supporting that this is happening in the tumors. We found that 53% of these pDGVs that we identified were showing evidence of somatic loss of heterozygosity.

And we observed that there was actually deepening, or progressive loss, of heterozygosity in the metastatic tumors, compared to the primary tumors, in 23 out of 29 pairs. This suggests that there is an evolutionary pressure on these pDGVs that drives this progressive loss of heterozygosity in metastatic urothelial cancer, and suggests that these genes may be playing a role in tumor progression.

And then finally, we found that the presence of germline variants compounds the somatic genomic heterogeneity because they result in activation or alteration of oncogenic pathways differently in different tumors from the same patients. So private germline-somatic interactions are happening, mostly because of the extensive somatic heterogeneity that we've identified before.

Sam Chang: So Bishoy, I'll stop you there. So that means within the same patient, the different stages, in terms of their tumor progression, results in differences that you just outlined. Is that correct?

Bishoy Faltas: That is correct. As we've described, and we and other groups have published extensively before, advanced metastatic and urothelial cancers have extensive genomic heterogeneity. So on average, about a third of the genomic alterations are only shared between any two tumors that we sequence, on average. And this is obviously, an oversimplification.

Sam Chang: Sure. And then, as that patient's tumor progresses, its signal also seems to change as well. Is that correct?

Bishoy Faltas: That's correct. So these different pairings, if you will, between the germline alteration and the somatic alteration, because typically, when we do the sequencing, we discount all the germlines. So we subtract it out. But it's still there in the somatic cells. So if you remember the first slide in my talk, these germline mutations are present in all cells of the body, including the cancer cells. So the somatic mutations are subsequent to them and compound heterogeneity. And that is something that I think is very important to point out, and it is something that we need to pay more attention to. And that is what we are currently working on, trying to understand the functional consequences of these germline-somatic interactions.

Sam Chang: Great.

Bishoy Faltas So I wanted to end on sort of a more clinically useful framework. So this is again from the review that we published last year, about a proposed framework for germline genetic testing in urothelial carcinoma.

So we asked, which patients should be offered germline testing? And obviously, 1) all patients that meet the Amsterdam or the Bethesda criteria, or hereditary breast and ovarian cancer guidelines, or have a close relative who has met these guidelines, 2) patients with MSI high or MMR deficient tumors, with all the caveats that I've mentioned initially, about upper tract urothelial cancers and the role of MMR deficiency there. And then 3) patients with the early-onset diagnoses, so less than, or at 45 years of age, which is outside the 5% confidence interval of the median age of onset for urothelial cancer, which we know to be mostly a disease of the elderly patients.

So I'd like to summarize; we've identified a high prevalence of putative deleterious germline variants that truncated tumor suppressor genes in urothelial cancer patients, we show that there is a progressive loss of heterozygosity in serial tumor samples, suggesting a critical role for these germline variants in tumor progression, we identified significant intra-patient heterogeneity in germline-somatic variant interactions, resulting in divergent biological processes within primary and metastatic tumors from the same patient, and we want to point or raise awareness, of the rule of germline variation and its potential downstream effect, as another layer of complexity to clinical genomics that needs to be considered for precision medicine strategies.

Again, thank you very much for giving me this opportunity to share our work with you, and I would love to talk to you more about this.

Sam Chang: Bishoy, this is just fantastic. Obviously, the work you and your group have done at Cornell is fantastic. So I guess always, when you present work like this, the next step is, okay, where are you guys going to go next? What's the next step, in terms of either screening, evaluation, prediction, versus screening, evaluation, prediction, and then the response to therapy? Where are you guys going to go next? I don't want to divulge all the upcoming things that are going to be presented, but if you could give us some insight, in terms of where the research is going to next explore.

Bishoy Faltas: Thank you for that question. One thing that we realized very quickly is that our study did not represent all ethnicities. And when we are talking about germline variants, it's obvious that this relates to the place of origin and ethnicity. We are fortunate to have funded work in collaboration with the New York Genome Center, as part of a great initiative called the Polyethnic-1000 research initiative in New York City, which is, and my group is leading the bladder cancer work, and in collaboration with Olivier Elemento, and we are planning to perform whole-genome sequencing on a large number of patients from diverse ethnic backgrounds here in New York City. But this actually extends to many other cancers, breast cancer, pancreatic cancer, I believe cervical cancer, there are many other, prostate cancer. I think this is a very important knowledge gap because we are starting to realize how much we do not know about the germline differences, or the similarities in patients from different ethnicities, or different places of origin.

Sam Chang: It makes me think about your, back to your pie chart, where you showed the majority didn't have a DDR change, but then there was a wheel that showed all the different noted mutations that have been identified. And then it makes my head spin, thinking about what that may look actually very, very different from a different non-European ancestry line that comes down. The majority of work, obviously, as you outlined has European ancestry. So the work in terms of all places, New York is obviously a fantastic place that to start that evaluation of the different ancestries and geographic locations in terms of where these changes may actually evolve, in terms of race and location. And that type of thing, I think is essential. 

Bishoy, thank you so much for taking the time. And it's clear, from a clinical standpoint, that just as you ended your talk about how it's becoming clearer and clearer, how little we really know. And so we appreciate all your efforts. It really is starting to expand all of our knowledge. So thank you again for your time, and obviously, thank you so much for all your efforts.

Bishoy Faltas: Thank you so much, and always a pleasure to talk to you.