Webinar with Intermountain Health
Hosted by Genomenon
Clinical biomarkers can be valuable tools for assessing prognosis and treatment options for cancer patients. PIK3R1 and POLE are emerging biomarkers in endometrial cancer. In endometrial cancer patients, PIK3R1 mutations are highly prevalent and associated with poor prognosis, but their biochemical effects in the cell remain diverse and understudied. Conversely, detection of a POLE mutation is associated with good prognosis, which can guide clinicians to spare patients from the toxicity of overtreatment. Thus, detecting and understanding these biomarkers are essential for optimizing patient care.
In this webinar, special guests from Intermountain Health discuss their use of precision genomics to sequence and analyze patient tumors, and present in-depth findings on PIK3R1 and POLE in the context of endometrial cancer gleaned through the literature using the Mastermind Genomic Search Engine as part of their workflows.
You Will Learn:
- How biochemical and clinical trial research may shed light on the impacts of PIK3R1 mutations in endometrial cancers
- How POLE mutations are associated with favorable prognosis and improved progression-free survival in endometrial cancer
- How the detection of a POLE mutation can spare patients the toxicity of overtreatment
Try Mastermind for yourself. Create your free account today.
Aneesa Al-Soodani, PhD
Variant Scientist, Intermountain Health
Natalie Peer, PhD
Variant Scientist, Intermountain Health
Field Application Scientist
CANDACE: Hello, everyone! Welcome to today’s webinar. We’ll be discussing how POLE and PIK3R1 are changing how we approach endometrial cancer. My name is Candace Chapman, and I’ll be your host. Let’s get started!
The Mastermind genomic search engine is the most comprehensive source of genomic evidence, and can quickly be used to identify papers for patient diagnosis and treatment decisions. In today’s webinar, guest speakers from Intermountain Health, based in Salt Lake City, Utah, will discuss their use of precision genomics to sequence and analyze patient tumors using Mastermind as part of their workflow. You’ll learn how different mutations can be associated with either a poor or favorable prognosis, and how detecting and understanding these biomarkers are essential for optimizing patient care.
We’re really excited to hear from our speakers, but first, here are just a few quick housekeeping notes. Today’s presentation will include Professional Edition features of Mastermind, which are necessary for clinical variant interpretation and reporting. If you don’t already have a Mastermind account, you can create one today easily using the bit.ly link you see on the screen, and it will also be dropped in the chat window. This will start you with a free trial of Mastermind Pro, so definitely take advantage of that. If you’re joining us live, feel free to drop your questions into the Q&A panel, and we’ll get to these toward the end of the event today if we have time. If we run out of time, we’ll follow up with you after the event. Today’s webinar is being recorded, and will be emailed to you to review or share after we wrap.
Now, I have the privilege of introducing our speakers. We’re joined today by Natalie Peer and Aneesa Al-Soodani from Intermountain Health. Hello there! Both Natalie and Aneesa are variant scientists at Intermountain, where they specialize in classifying and curating genomic variants in both solid and liquid tumors. We’re also joined by Genomenon’s field application scientist, Denice Belandres, who will be moderating today’s discussion. Denice is going to kick us off with a quick introduction to Mastermind. Nice to see you all! Denice, take it away.
DENICE: Awesome, thanks so much, Candace, for the introduction! I’m going to stop my video here so you can get a better look at the slides that I’m going to be sharing. I know we have some folks in the audience today who may not be as familiar with Mastermind yet, so I’m going to share a few slides so that everyone can understand how the system is working, what the interface looks like, and just get primed for the content that Natalie and Aneesa will be presenting here shortly. One concept to understand about Mastermind is that it’s very much an associations engine. You can search many different types of information. We take in things like genes and variants, so SNVs, indels, and even CNVs as well. Then, we can link those across to other concepts, like phenotypes and diseases, and even other types of information, like, is this a functional study or a case report? These are the kind of connections that you can make in the system. What you’re seeing on the screen now is a schematic of how all of those play together. Really, the strength of Mastermind is in finding and connecting genomic concepts.
On this slide, I’m showing a screenshot of the user interface after having launched a gene and a variant search. We have a few different panes or sections to focus on, and I’ll dive a bit into each of these in just a moment. Basically, we want everyone here to be able to follow along when our speakers start sharing their own screenshots of Mastermind. On the screen here, there are three main sections. The very top is the variant information section, so you’ll find our ClinVar integration here, along with related variants. Then, below that, there are two hemispheres: on the left, we have information about variants, and on the right, we have information about articles.
Taking a look at the variant side here, we have the variant diagram and the variant list right below that. In this section, we display all published variants found in the gene that you’ve searched. The diagram shows you potential hotspots in the gene, where there are lots of articles at a specific point along the x-axis, which is the length of the protein. We also display protein domain information below that as well. Then, the variants table is searchable and it’s sortable, so you can filter for things like, show me all the nonsense variants in that gene, or show me all of the changes found at residue R43, or even just sort them by cDNA position.
Then, we have the literature component. We display articles returned for your search. Here, we have the articles list, and below that, the full text matches from whichever article you’re currently clicked on. The full text matches are sentence fragments pulled in from the article for every mention of your search terms. Every time the author mentions your gene or your variant that’s in your search bar, you can see that sentence fragment and get some context for what it is they’re talking about in the paper.
Now, the system is highly sensitive. You can see that we’re picking up a bunch of different nomenclatures for this variant, we pick up the protein change, the cDNA, the rsID, all of those are getting recognized as matches for the variant that we searched for. When you add things like filters or phenotypes to your search, those types of keywords will also get pulled out and displayed in that full text matches section. We’re also looking into the supplemental data, so that’s toward the bottom of your screen. We’re actually going to show you how the variant is described. That’s where it says “match text,” the number of times that it’s mentioned in that supplemental data set, and even the name of the file or data set where you can go and find it. No more digging into 16 different supplemental tables to find your variant match, because we’re going to show you exactly where to look.
The last thing I want to share before I hand it off to Natalie: We also have an API which provides programmatic access to our database on the back end. We have a file annotations endpoint where we essentially take in a VCF, search all the variants in that file in Mastermind, and then provide an annotated VCF back, with two pieces of information. One is going to be the counts, or the number of articles found in Mastermind for each of the variants. The second is the link or the URL to take you back to Mastermind to go look at those articles from the user interface. With that, I hope we’ve all been level set, and are ready to hear from our fantastic speakers! Natalie, take it away.
NATALIE: Thank you, Denice! I hope everyone can see my screen. Today, Aneesa and I will be discussing the genes POLE and PIK3R1, and how mutations within them can change how we approach endometrial cancer. We are both variant scientists at Intermountain Health, as they said, located here in Salt Lake City. Let’s kick it off!
Before we get into the variants themselves, I’d like to talk a little bit about Intermountain Precision Genomics and what we do. We have multiple genetic testing offerings for our patients. These include pharmacogenomics analysis and genetic counseling, so people can better understand their mutations and what that means for them, as well as their families. We also offer precision cancer care, which is what we’ll mainly focus on here today. To provide this precision care, we use our genetic testing platform, TheraMap, which is based on the Illumina TSO 500 pipeline, which can identify mutations in all of these genes that you see here. The text is probably so small, because there’s so many, that you can’t even make them all out. You can appreciate, here, that since there are so many genes, we encounter many, many variants across our different tests that we perform for our patients. We have a lot of variants to curate, and we use a lot of different resources to do that.
We use a smattering of resources that we have represented here, all the way from COSMIC, where we just want to see if the mutation has ever been reported, to ClinVar, to see if it’s relevant in the germline, NCCN guidelines to see how far along this is in gaining access to patient approval. But first and foremost, right big in the center of our slide, is Mastermind, and this has become an integral piece of our variant curation process. I literally don’t know how we would do it without Mastermind’s search capabilities. That’s why we’ll be highlighting that today.
Next, I just wanted to highlight endometrial cancer itself, as that’s what we’ll be focusing on today. I’m sure we all know, endometrial cancer is a very devastating disease. It’s the most common cancer type of the female genital tract in the U.S. and in Europe. In the U.S. alone, in 2022, it was estimated that there were about 66,000 new cases diagnosed. Unfortunately, 13,000 of those patients are estimated to have passed from the disease. In Europe, we actually see the same ratio, with 122,000 identified in 2018, and about 30,000 of those passing as well. This mortality rate comes out to about four to five percent of all patients being diagnosed succumbing to the disease. Really, just highlighting here that identifying different methods of helping these patients is very very critical, and genetic biomarkers that we’ll be discussing today have never been so important. With that, I will hand it over to Aneesa to discuss our first biomarker of the day, POLE.
ANEESA: Hi, everybody! My name is Aneesa Al-Soodani, also a variant scientist at Intermountain Healthcare. I will be switching to a more positive note on endometrial cancer, because these mutations usually are associated with a positive prognosis, and they help with treatment decisions further down the line if a patient does have a POLE mutation. So, what is POLE, and how does it cause genomic instability? Then, subsequent positive prognosis? The POLE gene encodes the major catalytic subunit of DNA polymerase-epsilon. This protein is a DNA polymerase that extends the leading strand during DNA replication. It also provides proofreading activities during that step, so that errors are not occurring during DNA replication. Specifically, mutations in the exonuclease domain where the proofreading occurs in POLE compromises its proofreading function. This leads to loss of replication fidelity, genomic instability, and subsequently, an ultra-mutated phenotype.
We’ve heard about tumor mutational burden and high mutation rate, and highly mutated protumors, but POLE mutations lead to an ultra-mutated phenotype, where you have even more mutations. These tumors can have upwards of 500, 600, 700 mutations per megabase. In these tumors, it’s quite high, the error that occurs when you have a POLE hot spot mutation. These poly-mutated tumors, since they do have a lot of mutations, they also have elevated tumor infiltrating lymphocytes, or TILs. TILs and PD-1 and PDL-1 expression. This leads to a good response to immune checkpoint inhibition, so immunotherapy works particularly well in these tumors, since they have that elevated TIL. They just have so many mutations in them that it makes it easier for immunotherapy to work and kick these tumors out.
How many POLE mutations are there in endometrial cancer, or how often are they seen in endometrial cancer? We can see here on the right, a POLE from cBioPortal for cancer genomics, that endometrial cancer harbors the most POLE mutations compared to other tumor types, that span left to right. Endometrial has the most, upwards of 15 percent of endometrial cancers have a POLE mutation. Luckily, these are associated with a favorable prognosis, based on retrospective studies. They have higher progression-free survival than POLE wild type. They have disease-specific survival and overall survival. This helps patients avoid overtreatment, given their excellent prognosis. Instead of going through rigorous chemotherapy regimens, they can be treated with immune checkpoint blockade with anti PD-1 antibodies, such as pembrolizumab, that we’ve all heard of, because these show a good response in these patients with a POLE mutated cancer. Instead of going through a hard chemotherapy regimen, you can get immunotherapy, and it’ll have a better progression-free survival, and better response.
So, stratifying these tumors is pretty important. Here’s a new guide, or a new stratification scheme, that’s been coming out to look at patients that are POLE mutated or POLE wild type. You can see here on the left, which this talk focuses on, is a POLE mutated endometrial cancer. That is an ultra-mutator; you have lots of mutations, but you have an excellent prognosis. This is compared to microsatellite instability in green, which are usually hypermutated as well, and can have a good response to immunotherapy, but their prognosis is intermediate. Compared to a POLE wild type or microsatellite stable tumor, that has a lot of copy numbers, either low or high, they have less mutations, but they will have higher copy number alterations, with an intermediate or poor prognosis. The stratification of the blue and green is something that’s coming out recently. This is more merging the difference between having a POLE mutated, ultra-mutated tumor, versus a microsatellite unstable tumor, because their prognosis and their response to immunotherapy may be different.
You can see this here was also a new emerging stratification for these tumors. The difference between a microsatellite unstable tumor and a tumor mutational burden high tumor. The MSS is microsatellite stable, and the green MSI is microsatellite instability. They can merge. You can have a POLE, or its friend, POLD1, that also leads to a high mutational burden. They can be associated with MSI stable or microsatellite instability, or they could just have a high tumor mutational burden. They can merge, you can have a mix and match. What we’re finding is that, with POLE mutations, they are usually associated with upwards of 100 mutations per megabase. It’s an ultra-mutator, they’re usually missense.
However, microsatellite instability or microsatellite unstable high tumors have around a range of 10 to 100 mutations per megabase. These are usually, due to the nature of microsatellite instability, frame shifts. They’re usually small indels, a deletion or insertion, so their mutation signature is different than a POLE or POLD1 mutation. On the right, where POLE or POLD1 mutate, they usually have a tumor mutational high. You’ll see an elevated immune cell in with the infiltrating tumor cells, suggesting that immunotherapy would work better for these types of mutations.
Now, there are a lot of mutations in POLE that have been reported to date. However, most of them cluster in the exonuclease domain, where most of the DNA repair and proofreading activities occur. That’s in green, it’s about amino acids 268 to 471. There are definitely some hot spot mutations that have been well-characterized in this domain, such as the first three that I’ve listed in this table: proline 286 (p.P286R), valine 411 (p.V411L), and phenylalanine 367 (p.F367S). These have been reported in hypermutated tumors, and they are located in either the DNA binding pocket and they have been shown to have increased replication error rates compared to wild type, for in vitro and in vivo studies. These are three very well-known mutations in POLE, and they have shown positive prognosis for endometrial cancer patients, as well as a good response to immunotherapy.
What happens if you have a hypermutated case, a hypermutated tumor endometrial cancer, but you don’t have these three very well-known, very well-characterized mutations, but there still happens to be a POLE mutation that is located in the exonuclease domain, such as the two that I have listed here, as serine 297 (p.S297F) and alanine 465 (p.A465V)? Those are also represented on the arrows. They’re pretty prevalent in endometrial cancer, but a quick Google, or a quick search on COSMIC or other databases, you’re not finding much. Then, you really want to know, can I give the provider anything? This is clearly a hypermutated tumor, can I give the provider something to work on? Maybe it won’t be concrete, but these are usually late stage cancer patients, and anything can help them out. I’m going to walk through how Mastermind really helps patients understand their tumor type, given that they have these lesser characterized POLE mutations.
What I first would do is go to Mastermind, and it seems like they’re having a webinar right now, so you should definitely check that out. I will type in POLE, and then type in the variant. It’s very genome aware, very easy to use. This was just by typing in POLE — it knows it’s a gene, and then typing in a variant, the p.A465V, it knows it’s a variant, gets that little tag in. Very easy to use and then search. You get back this screen, where it has found quite a few articles on this variant, about 87 full text, 102 that include the supplemental that was discussed earlier. That’s quite a few hits, quite a few papers for this lesser-characterized mutation. At this point, I’m thinking, okay, there’s something going on here. Compared to the other bar graphs, it’s pretty high, so there’s a lot more papers than outside of the exonuclease domain. I don’t have it marked here, but you can see a cluster of papers around that 200 to 400 mark, where the exonuclease domain is. It’s within that cluster. I’m already thinking, wow, I’m probably going to find something. I might not, but let’s just keep digging.
I did find this paper, where this mutation was detected in a ultra-mutated tumor. Ultra-mutated is usually over 100 mutations per megabase, and the TMB high is over 10. So ultra-mutated is over a hundred, and this is definitely over a hundred, because you can see here, the TMB is 841. That’s a pretty high tumor mutational burden signature. This variant was detected in that. You can see that the POLE signature has a C to A peak, and a C to T signature, and it interacts with the site 275 of the exonuclease catalytic region and affects the structure of the active site. This is all stuff that I found through Mastermind. It did not increase the sensitivity to radiotherapy or chemotherapeutic in mouse embryonic stem cells. This suggests that, for POLE p.S297F mutation positive patients may not benefit from overtreatment, and they may benefit from immunotherapy.
This is a pretty good indication of looking at a variant through Mastermind that was not necessarily a hot spot, but it is prevalent in endometrial cancer, and it is detected in ultra-mutated tumors, and it has that signature. These patients, if they have a tumor mutational burden that is high, they are likely to respond to immunotherapy and not respond to chemotherapeutics.
Here’s the next variant that I was looking for. It doesn’t have as high a number of papers as the last one, there’s only 16 full texts, a total of 23 with supplemental. It is VUS in ClinVar, so it’s a variant of uncertain significance in ClinVar in the germline. This one’s a little bit leaning towards, or in between, actionable and a variant of uncertain significance. Even less is known about this one, but you can see that, still, quite a few papers have been published on it compared to others, and it is located in that exonuclease domain, and it is within that cluster of highly reported variants. So, I’m still thinking, there’s got to be something here. I did get a whole bunch of papers back and started sifting through these.
What I want to highlight on this page is, on the right, you can sort by how many times the paper mentions your specific variant. If a paper only mentions your variant once, and it might be in the supplemental article, they’re probably just saying that it’s reported, but we don’t know anything about it. I’m usually more interested in a paper that mentions it five times, three times, then they probably start talking about the variant and characterizing it more deeply than just saying, “oh, it’s reported but not characterized.” This is a very interesting feature to determine, there are 16 articles, but which one am I going to give my full attention?
Also, the year — so, this article that has five hits in the paper for my variant, it was published in 2020. Then, there’s 2019 for three mentions. 2022 only mentions it once, so I will definitely be interested in the first two hits here. I might go down to the 2022, I’ll glance into 2014. This is very helpful in deciding which of these papers am I going to give my full attention, and which ones can I just skim. I did find that in the 2019 paper, it was also detected in an ultra-mutated tumor. This tumor, they don’t say what the tumor mutational burden was of this one, but the field has agreed that ultra-mutated tumors are a TMB of over 100.
Then, I was very interested in continuing on and seeing if this variant was characterized. Another paper, the 2022 paper, that top hit, did look at TCGA data for endometrial cancer with this mutation, and looked at the different types of mutations and compared it to an MSI unstable or an MSI stable tumor, then our POLE wild type. The top is the POLE mutated endometrial cancer, and then the bottom two are POLE wild type. The different colors represent the different mutation types, and this was a pretty difficult figure to get through, but it was also lucky that I did find this. It was buried deep in the literature, but it was easily found. It was one of the top hits in Mastermind, and it did show a summary on this figure: Compared to wild type POLE, MSI unstable, and MSI stable, it had an elevated C to T transition, so that’s that light blue. It had less indels, which are unlikely to be a POLE signature, and they’re more MSI unstable signature, and it was detected in an ultra-mutated tumor. This variant is likely to inform on a TMB high endometrial cancer. It is not as well-characterized as the other hot spots, or the last mutation I showed, but this is still something that you can give a provider if they have an ultra-mutated endometrial cancer.
I’d like to point out that, yeah, we don’t know much about this p.A465V. However, it has been detected in ultra-mutated tumors, and it has similar characteristics and mutational signature to a POLE mutated cancer, and your patient may have a better prognosis, and they may respond to immunotherapy. It’s nothing concrete, like the hot spot mutations that I talked about earlier, such as that p.V411L. Everybody knows that that one is an inactivating and pathogenic variant, but this lesser-known p.A465V may also have clinical potentiaI. I will hand it off to Natalie for her PIK3R1 discussion.
NATALIE: Thank you, Aneesa! I’ll be taking it more from the more established prognosis into more of the potential with PIK3R1. Unlike POLE, this is more of an up-and-comer in the field that we can use Mastermind to start to analyze. Similarly to how Aneesa showed the prevalence of POLE mutations across different cancer types, here, we can see that same representation with PIK3R1. With this plot at the left, we can appreciate that PIK3R1 mutations are detected in multiple cancer types. If we start taking a little bit of a closer look, we can see that these first four are actually all types of endometrial cancer. We can appreciate right away that PIK3R1 is more heavily reported in endometrial cancers than any other tumor type. Across these different studies and types, we can see anywhere from 20 to 40 percent of PIK3R1 mutations occurring within these cancers.
We were really interested, since we at Intermountain see a lot of endometrial cancer patients ourselves. We have queried the amount of PIK3R1 mutations we have detected in our patients, and asked how many of those were endometrial cancer cases, and we have found a pretty consistent number: about 40 percent of our PIK3R1 mutated cases are indeed endometrial cancers. It’s really interesting to see that, not only is this established in the field, but we also see it mirroring within our internal clinical cases as well.
PIK3R1 is, like we said, heavily mutated in endometrial cancer. What we can see here is a representation of a clinical study, where endometrial cancer patients at different stages were analyzed for different mutations in different genes. We can see a list of genes here, and right away, we can see that our PIK3R1 is popping up as one of the most prevalent genes mutated. Taking this into more of the scope of, what does this mean for the patients? It does have an emerging presence in prognosis for these patients. We can see this with the survival curve of PIK3R1 mutated patients in blue, that there is a significant increase in mortality rates for endometrial cancer patients harboring PIK3R1 mutations. This is one study, but it is really recent, in 2019. Just highlighting that PIK3R1 is becoming more and more relevant in this field.
Also, taking another look at the plot at the left, we see these other genes that are mutated often as well. PTEN, which is a well-known and established tumor suppressor, case as well as KRAS and PIK3CA, which are known oncogenes. Mutations in all four of these genes are really important in endometrial cancer. It actually makes sense, if we start to look at a signaling diagram for them, we can see that these four proteins all play into this very complicated web, this signaling pathway that is very often mutated in endometrial cancer. Taking a closer look here, we can see the Ras/MAPK pathway here at the left where Ras genes can signal into the Raf/MEK/ERK pathway. We can also see at the right, the PI(3) kinase/Akt pathway, which plays into the mTOR pathway as well. Then, here at the center is our p110, our PIK3CA protein, as well as p85, which is the PIK3R1 protein that we are focusing on today.
You can see already that it makes sense that PIK3R1 is going to have a broader effect in endometrial cancer, since these are common genes as a whole. They’re mutated, and this is hanging out right front and center, managing these multiple branches of signaling. So, how does p85 or PIK3R1 work? Well, it works in tandem with both PIK3CA and PTEN. At the left, we can see that PIK3CA, or p85, can physically bind to p110, or PIK3CA, and it can attenuate the PIK3CA oncogenic activity. It can make it more stable and keep it from sending off its oncogenic signals. On the right, we can see that p85 also physically binds to PTEN, and it can help stabilize it and increase its activity. PIK3R1 is a tumor suppressor in that it can regulate an oncogene and promote the activity of a tumor suppressor. Really, if you have a deleterious effect on PIK3R1, you’re hit with a double whammy. You’re going to have dysregulation on both sides of this little seesaw that it tries to balance.
How does it accomplish this? PIK3R1 has multiple domains, and we can start breaking this up into two pieces. The front half, towards the left of this diagram, is the major region that’s involved in PTEN binding. There is an n-terminal SH3 domain mixed alongside a BH/GAP domain. These two together are what can physically bind to and stabilize and activate PTEN. On the latter half, we can see multiple SH2 domains. there’s an n-terminal, a c-terminal in yellow, and then, in that spot in the green, this is the inter-SH2 domain, the domain that lies between the n and c. By and large, this latter half of the protein is what’s involved in binding to and stabilizing and attenuating the PIK3CA protein, the p110. Both of these pieces of PIK3R1 are important for its normal function within the cell.
Looking at Mastermind, we can see a diagram of the protein here, and I’ve outlined where our PTEN and PIK3CA binding regions are. With this plot, you can see that the columns on the y-axis are showing number of publications per variant at different locations along the protein. It’s just completely across the whole protein. There are publications for variants at every spot along this gene, and it really just highlights that, since the majority of the protein is required for its normal activities, mutations can occur at literally any point of this protein. We can see that, furthermore, here, in this lollipop plot at the top, again highlighting the different diagrams and regions. We can start to see that we have mutations at multiple points. We can see them clustering in that inter-SH2 domain, this PIK3CA binding region. We can see some of these lollipops hitting higher points, and again, just highlighting that these are occurring along the full length of the gene.
Additionally, depending on the location of the mutations, they can have different functional impacts on the PIK3R1 protein. We can see that below, where we have multiple mutations occurring. We have truncations, missenses, deletions, and they all can have different effects. They can change the binding to PTEN, or change the binding to PIK3CA, activation of different pathways, even the non-canonical pathways that are occurring through the Ras branch of the signaling pathway with the ERK pathways. This is just really trying to highlight here that, PIK3R1, if you mess it up, you’re going to have a whole bunch of things happening downstream.
The variants that we’ll be focusing on today are highlighted here: truncation at arginine 348, as well as a missense at asparagine 564. These are outlined here as well. I also wanted to point out two other variants that are occurring within this lollipop that are enriched, that are also hot spots, along with these two. Keeping in mind, with those four variants, we can start to look at PIK3R1 in Mastermind in the context of endometrial cancers. By putting two filters into Mastermind, our gene filter for PIK3R1, as well as a disease filter for endometrial neoplasms, we can pull out not just mutations that are prevalent in PIK3R1, but mutations that have been reported in endometrial neoplasms. We could see the list that we have down below, we have that truncation that we mentioned on the previous slide, as well as these three mutations that also show up very enriched, just in PIK3R1 as a whole. Turns out, these four mutations are also enriched specifically within endometrial cancers.
Taking this first, let’s look at our p.N564D variant. We can see here that our filters to show the endometrial neoplasms reports that there are four articles outlining this variant within endometrial neoplasms. Taking a look at those papers, we see what Aneesa was mentioning: sorting by our match count, how many times the variant has been reported in your paper. It’s only been reported, this variant, four times within the article, and then descending, three, two, one. We can click on this first paper, and Mastermind will show the full text matches. You can find where your gene and where your variant are mentioned in the paper. Right away, without even having to open the paper, I can see that there’s probably not going to be too much information regarding the functional effect of this variant. If I wanted to know its functional characterizations, this probably wouldn’t be the best way for me to look at it, because this is just showing me when it’s being reported with endometrial neoplasms.
Let’s say I was looking for functional information to find out if this was truly deleterious or not. I would change up my filters a little bit, to get just straight functional information. The first thing I would do is remove the endometrial neoplasms filter, because it’s very possible that this variant is characterized, just not within endometrial cancers that this literature research is pulling. I could remove that filter, and then, I could also add a functional data filter. Going to the filter categories option at your search bar, you can open up these very detailed and diverse filter options. Going to the functional filters within the ACMG interpretation heading, you can access either in vivo or in vitro filter options. For my purposes here, I was looking for in vitro studies to see if there were any cell assays or things like that to show me what’s going on with this variant. I can submit this with chosen filters to pull out articles that are only going to be talking about my variant with functional in vitro data. Right away, I am now getting 57 articles instead of just those four that were limited to endometrial cancer. I also am, in this case, instead of using a match count sort, I’m using the relevance sorting, because this is showing which articles are going to best show me what I am looking for. Mastermind’s algorithm can sort these for me.
Just looking at this first paper, I can select that, the full text display is showing me so much great information without me even having to open the paper further. I’ve sort of highlighted the information here that would be important to me, as a variant scientist curating this variant. I could see that they are indeed talking about my p85-alpha mutation that I’m interested in. Now, I can see that these mutants showed elevated kinase activity, promoted anchorage, independent growth, increased cell proliferation, and they had increased survival as well. So I’m just like, wow, this was easy! It gives me all the information I need. If I wanted more information, I could easily click this link out to the PubMed page directly, and this would pretty much be my one-stop shop.
Now, if I wanted to find more information, maybe not on the aspartate substitution here, but maybe I wanted some information on the codon itself, I could also edit my filter options and add a text search, a simple text search, to find any papers that use the word “residue,” because maybe they are going to be talking about the asparagine residue rather than the asparagine to aspartate alteration in the variant itself. If I just add an easy text filter here for “residue,” I end up finding a paper right away that says, this residue itself is within hydrogen bonding distance of this other base, or this other amino acid in PIK3CA, and this is starting to make sense. This residue is within the PIK3CA binding region the inter-SH2 domain, so it’s going to be bonding with PIK3CA. A mutation in this residue is probably going to change this interaction, and I can start piecing these things together just with these easy filters.
One of the other variants that we were talking about is a truncation at arginine 348. Heading into this one, even with retaining that endometrial neoplasms filter, I am finding information characterizing this variant in endometrial cancers already, which is pretty exciting. The other missense we were looking at doesn’t seem to have that foothold yet, but this truncation is already further along the discussion, even back in 2014. This paper itself is giving me some functional information as well. The truncation that we’re describing is activating AKT signaling. It also activates the non-canonical ERK and JNK through the Ras/Raf pathway. This sentence, these two, particularly, down here, indicating that a MEK inhibitor was effective in cells harboring this alteration, and the speculation that other cancers harboring this mutation may be susceptible to PI(3) kinase pathway inhibitors as a whole. Not only, with these filters, am I gaining information about functional activity, but I can start translating PIK3R1 mutations into possible therapeutic effects, perhaps even in patients. This is at the in vitro level here, but since they are now suggesting maybe cancers as a whole could be susceptible, I can start editing my filters to try to get into that space: Is PIK3R1 reported in relation to therapy sensitivities? If I wanted to know that, I could start using my filters for that reason.
Just before I head into that, I wanted to supply a more simplified figure of how all of these proteins are interacting with each other, more simplified than the one I had showed previously. Again, just highlighting the major players here: PIK3CA and our PIK3R1, how that feeds into AKT and mTOR pathways, our PTEN here as well as the Raf/MEK pathway as well. Multiple drugs exist that target this pathway, but since PIK3R1 is emerging, there are more therapies going into this, and I’m curious to see what therapies have been reported in relation to PIK3R1. I’ll start exploring that within Mastermind. Here, I have specifically left my endometrial neoplasms filter on, because I am really curious if PIK3R1 has been reported in endometrial neoplasms with regard to therapies.
Going back to the filter categories, I can go now to this clinical significance tab. At the left, I can select “therapy” and enable all of these filters to start pulling in articles for that. I can also go to the clinical trials option and enable those to see if maybe, not only there are therapy sensitivities reported for PIK3R1 mutations and endometrial cancers, but maybe some of these have made these through to clinical trials. Maybe there’s patient data. Using these two filters, I was able to pull out two papers, very recent, 2018 and 2022, that discussed PI(3) kinase/mTOR inhibitors in the context of phase one clinical trials. The information I’m pulling out here was very, very interesting, showing that patients that had mutations in PIK3R1 showed a partial response and a complete response in treatment with these drugs.
This other paper that we can see here, with another PI(3) kinase/mTOR inhibitor, also shows that patients had a durable partial response, lasting over a year and a half with a PIK3R1 mutation. What was really great about this search is, not only was I finding that patients with a PIK3R1 mutation had a response to these specific drugs, but Mastermind was also telling me specifically what variant this patient had that showed a partial response. I was just really gaining a lot of information. To be honest, when I was performing the search, I didn’t expect an actual variant to come out of this, but that was such a great piece of information, because dels within the iSH2 domain are also hot spots in endometrial cancers. This just really hit home that point, that PIK3R1 mutations are really up-and-coming in the endometrial cancer human care field.
Fitting back onto this image here, with different drugs and how they can target these pathways, the drugs that are here listed in bold are the drugs that we were just looking at in those phase one studies, and how they fit into this pathway. The drugs that are not in bold are current FDA-approved drugs that are available to patients that have mutations in respective pieces of this pathway. These three drugs are PI(3) kinase inhibitors. These are mTOR inhibitors, and then, these are inhibitors of the MEK pathway as well. There are many drugs that are available that are trying to inhibit this pathway, but the role that PIK3R1 is playing in them. Now that we’re seeing that some patients, specifically with PIK3R1 mutations, are potentially responding to some of these, it’s just really really exciting to be able to see this coming along in the field.
With that, Aneesa and I would like to thank our department, our subdivision, Intermountain Precision Genomics. Specifically, we would like to thank our variant science team lead, Laura Gonzalez, as well as our bioinformatics team lead, Dustin Miller, who helped us curate those statistics for our internal endometrial cancer cases. With that, I think we’ll head into some Q&A. Thank you!
CANDACE: Thank you so much for the excellent presentation, Natalie and Aneesa, and thank you to our audience members who have submitted questions! Let’s see how many we can get through. I’m going to welcome Natalie, Aneesa and Denice back for this section. So, the first question I’m going to ask Aneesa: do different POLE variants have different roles in cancer induction? For example, colorectal versus endometrial cancer? And there’s two questions in one — does loss of function lead to cancer for POLE?
ANEESA: Yeah, I can take that one. As you saw with my one of my earlier slides, POLE mutations are more prevalent in endometrial cancer than colorectal, but yes, they are prevalent in both. They can lead to a hypermutated phenotype in both colorectal and endometrial cancer. The data is more promising, probably, given that there’s more mutations than endometrial, and it has been specifically associated with a positive prognosis in endometrial cancer. For colorectal, I’m not as familiar, but more data is available for endometrial. Different types of POLE mutations, such as whether or not they’re in the exonuclease domain and the type of mutation signature they have, so the C to A transitions or C to T transitions, they can have a different effect on tumor progression and a hypermutated phenotype. If you have an exonuclease domain mutation that has that POLE signature and the ultra-mutated phenotype, those are the ones that are now well-known to have the positive prognosis, but we still don’t know much about the others. They are different, and we know more about the positive prognosis in endometrial cancer, but for colorectal cancer, I think, still, the response to immunotherapy is promising.
CANDACE: Great, thank you. Natalie, I’ll ask you: do you report all kinds of variants? VUSs, fusions, etc.?
NATALIE: Yeah, so, with our TheraMap test at Intermountain, we report multiple types of variants. We have both DNA and RNA sequencing data that we use. A lot of the variants that we will report, like with our PIK3R1 deletion variants that I mentioned today, we’ll report those. Then, we also use our RNA data to report fusion variants. There are a lot of canonical fusion events that occur in multiple cancer types, that are very important in patient care, for example, EML4-ALK fusions in lung cancer. Those are heavily reported in the NCCN guidelines, and have specific FDA approved therapies. So, yes, we do report both DNA and RNA data for our cases.
CANDACE: Okay, awesome. Denice, this one is, I think, for you. How do you find variants in specific domains in Mastermind?
DENICE: Absolutely. In the variants diagram where I showed that Manhattan plot at the beginning — and I think Natalie showed it, as well, at other times — right under that plot is a separate track where we display the protein domain information. That information gets pulled in from UniProt, and you can hover over the little bars to see the range of the residues. Remember that you can also sort the variants list or table by position. Once you know the range of residues, you can then go find those within the variants list as well, once you sort.
CANDACE: Okay, awesome. Then, this one is related to that, also for you, Denice. Can you search for specific clinical trials in Mastermind?
DENICE: Definitely. Natalie showed in her slides how you can use the filters to look for general clinical trial information. That’s under filter categories, and then clinical significance. Then, as some subcategory of clinical trial type keywords. You can enable them all, like she showed, or you can pick and choose more specific terms. Also, if you know the specific NCT number or identifier, you can type or paste that directly into the search bar and launch that as a text term. Natalie also showed, at one point, she used “residue” as a text term, so you can essentially type any word you want in the search bar, and launch that as a text. That also works in the full text matches, so you start to see those getting pulled out if there are matches in the articles.
CANDACE: Okay, awesome. I think we have time for one more question for Aneesa: Can POLE associated, hypermutated EC evolve into some other subtype over treatment, or can it be treated and maintained with ICI therapy once and for all?
ANEESA: Oh, wow, that’s an interesting question. I recently read a case report that a patient who had a POLE endometrial cancer was being treated with chemotherapy, not responding, the tumor was growing. Then they detected that POLE mutation, and then they switched to immunotherapy, and the tumor shrunk, and they had good prognosis overall. I mean, I’m not making any promises, that it will never come back, or it won’t change, but the data now is very promising that most patients do pretty well on immunotherapy maintenance. Of course, there’s probably some that, you might get a secondary hit, and it’ll evolve into some other subtype. That can happen, but right now, that case study I read was pretty incredible. They just stopped the harsh chemotherapy and used immunotherapy, and the patient just responded very quickly, and was doing very well. It’s really just such a promising biomarker right now.
CANDACE: Wow, great job answering that! That sounded like a very complex question, great job. That is all the time we have. I just want to say thank you, once again, to our speakers, and to all of you for attending today. Here’s one last look of the bit.ly link to create your free Mastermind account for yourself, or pass it along to colleagues. Of course, feel free to reach out to us with your questions to email@example.com, and look out for the recording of this webinar in your email inbox soon. We hope to see you all again at our next event. Thanks again, and have a wonderful rest of your day! Bye.