Welcome, everyone, to our today's session of the Virtual Molecular Tumor Board. It's my great pleasure that we have Drs. Ian King, Nicole Park, and Eduardo González with us, and they will present today on... I cannot... Wait a second. And they will present today their project on molecular biomarkers for response to immunotherapy in endometrial cancer. But before I introduce our guests today, I would like to point out some organizational and technical points for this session. So first of all, this Tumor Board is for educational purposes only. It's a combined project of the Variant Interpretation for Cancer Consortium WIC and the Somatic Cancer Clinical Domain Working Group, and it's featured by the ACMG Genetics Academy, who is also responsible for the accreditation for your clinical credits. The planning committee consists of Beth Pittel, who is the co-lead of the Virtual Molecular Tumor Board at our Variant Interpretation Consortium. I recently joined the group, and I will lead today's session. I'm Manuela Benary, a bioinformatician from Berlin, and it's very great that we are actually sponsored by ACMG and have a great support by Jane Radford, Gordana Raka, and Maximilian Münke. We are also supported by the Variant Interpretation Consortium by Alex Wagner and Shruti Rao, who has been a co-lead in the first sessions and who actually set up the whole series in the first place and was a great help in that in the beginning. So as I said, you can get credit points for this session, so one hour is one credit point, and please use the evaluation in the end to actually get the certificate for your credit points in the end of this presentation. A recording of this webinar will be available in two weeks at ACMG Genetics Academy, and you are able to actually re-watch and join this video. However, if you want to have your credit points, please allow some time to view and print the certificate. We will come back to you on that. During the session, please use the Q&A feature for your questions to the speakers and use the chat only if you have any technical issues. If you are interested in a particular question, you can upvote them and give it a thumbs up, and we will then discuss your questions at the end of the talk. With that, it's my great pleasure to introduce to you today's speakers. We have Dr. Ian King, who is a clinical molecular geneticist from University of Health Network in Toronto, and also an assistant professor and member of our WIC consortium. It's also my great pleasure that we have Eduardo Gonzalez-Ochoa here, who is a clinical fellow and just came in right from the patient treatment, and I'm very happy that he took the time to join us today. Nicole Park is an annotation specialist, also from Toronto, and did most of the analysis presented here. We do have some disclaimers on financial issues for this presentation, so please be aware of that. And with that, I would like to hand over to Nicole, and I'm very happy to hear your presentation. Okay. Nicole, do you have control now? Yes, I see that. Okay, great. Hi, everyone. My name is Ian King, as Manuela said. We're coming to you from Princess Margaret Cancer Centre in Toronto, Ontario. Princess Margaret is a large cancer centre in Toronto, one of the fifth largest in the world and the largest cancer centre in Canada. We're part of the University Health Network, which is a provider in the Toronto area. There are four hospitals that make up UHN, Princess Margaret, Toronto General Hospital, Toronto Western, and Toronto Rehab. So it's a large healthcare provider in the Toronto area. As a large provider and a major cancer centre, Princess Margaret is involved with a lot of clinical trials and drug development and translational research. We have a lot of clinical trials going on and many links to what's going on in the drug development world. So what we're going to be talking about today is the interplay between molecular diagnostics and the therapies that are ongoing in a trial for sequencing of endometrial cancer at Princess Margaret. So next slide, please. The ongoing trials at Princess Margaret include a lot of genomics. UHN also has a large cancer genomics lab, the largest in Canada as well. We process over 20,000 samples per year between two labs actually, one at Toronto General, which handles standard of care testing, mostly funded by the Ministry of Health in Ontario and Cancer Care Ontario, and also a translational genomics lab called the Advanced Molecular Diagnostics Lab, which is housed at Princess Margaret. So the trials that we'll be talking about today are run through that AMDL, Advanced Molecular Diagnostics Lab at PMH. We've been doing large-scale genomics for some time at Princess Margaret. The first study that was launched was back in 2012. This was the IMPACT study, which was a profiling study across many tumor types that used, at the time, a fairly advanced panel, but what today would seem very small, a very targeted sort of hotspot panel. And we've grown and expanded that testing and now currently we have a number of studies underway. We have a pan-cancer study called Octane, which uses the TSO500 panel, which you might be familiar with, the Illumina panel, and we have other site-specific studies such as COMPASS, which is a study in pancreatic cancer doing called genome sequencing. But what I'm going to talk to you about today is specifically a pair of studies done for gynecological cancer called BIODIVA and VENUS. BIODIVA is a study in which we're doing large-scale molecular profiling for high-grade serious ovarian cancer, and VENUS is a companion study in which all other gynecological cancers are profiled. We are doing both sequencing on a large panel and also some longitudinal profiling with ctDNA to start looking at how these molecular mutations in the tumor evolve over time. Next slide, please. So the panel that we're using for the results that we're going to be talking about today is a custom panel that was developed in 2018, I believe, or maybe even earlier, called the UHN Hi5 panel. So this is a hybridization capture panel that sequences 555 genes, a gene list that was made in 2018 to be inclusive of more or less every gene that was known to be actionable in cancer and a bunch of other research-oriented targets. Because it's a large panel, we can do a number of things with it, and among these we can call copy number using the NGS data, and we can look at tumor mutation burden, which will be important for this talk. It's been shown in a number of studies that you need a fairly large footprint of at least a megabase for tumor mutation burden off of a small panel, or a large panel, I should say, to correlate well with whole exome sequencing, which is sort of the gold standard for tumor mutation burden measurement. So this is a large panel, 555 genes. We do have enough footprint to calculate TMB. However, this is a tumor-only workflow. I'll also emphasize a lot of large studies have used the tumor normal workflow in which normal DNA from blood is sequenced alongside tumor DNA, which allows the study to eliminate a number of mutations, which are known germline mutations, and to identify germline mutations for potential profiling later. We're not using that. This is mainly for cost reasons. To enable to profile more patients, we went with a tumor-only workflow, and also I'll mention that in practice, tumor-only workflow is still our standard for the sequencing we do for standard care because of cost reasons, and we'd like to develop our genomics tools in order to accommodate that tumor-only workflow, but it's a technical limitation that is important to talk about. Next slide, please. Okay, so at this point, I'm going to hand over to Eduardo, and he's going to discuss the clinical aspects, and he and Nicole will discuss the cases that we're talking about today. Sorry, Eduardo, you're muted. Okay, thank you very much. Okay, to put some clinical context, I was briefly about the endometrial cancer. As you know, it's the most common gynecological malignancy and the second leading cause of death just after ovarian cancer regarding gynecological tumors. It is estimated that this year in 2021, there will be 8,000 new cases and 1,400 deaths just in Canada. And this tumor, it's rising both in incidence and mortality. And from the clinical standpoint, this tumor is heterogeneous and the prognosis of this tumor depends on different factors such as histology, including the tumor grade and the histological subtype. This clinical stage of diagnosis and also some molecular features have proven to be of prognostic significance. As you see in this chart, despite most of the endometrial cancers are diagnosed at the earliest stage, nearly 30% of all the endometrial cancer presents at more advanced stage and the survival remaining poor in advanced stage with rates around 20% at five years in the stage of four clinical. Okay, can I get the next one? Please. Okay. So in 2013, the Cancer Genome Atlas published a comprehensive genomic characterization of 373 advanced endometrial carcinomas. Most of them were in the endometriosis histology. And it further classificate endometrial cancer in four big subgroups. So the first group, which is the ultra mutated subgroup, it's characterized by whole E mutation. As you can see, high burden of mutations per megabase. And it's associated most of the time with high grade endometrial tumors. And as you see in the chart C, it's the subgroup associated with the best prognosis. Unfortunately, this ultra mutated phenotype, it's the least common of the all of the four. The other, the next subgroup is the hyper mutated. It's associated with MSI, high number of mutations per megabase. And it's also associated with high grade endometrial tumor. And as you can see in table C, it's associated clinically with a intermediate prognosis. And the biggest group is the group with copy number low. This group is usually associated with low somatic copy number alterations, P10 mutations, and clinically, they're associated with endometrial tumor, endometrial tumors, but low grade endometrial tumors. And as you can see, in chart C, they're intermediate prognosis. And finally, the last, the last subgroup, referring as copy number high, it's associated mostly with P53 mutations. And also it's associated with high grade serous histology, and is the subgroup harboring the worst prognosis from the clinically standpoint. To sum up, as I already told you in the in the previous slide, the four subgroups have some prognostic significance. Usually, the ultra mutated subgroup has a gene expression expression signature of of immune regulated or immunoreactive. And also the hyper mutated subgroup with with microsatellite instability. Both subgroups accounts for 40% of the of all the tumors and are related with the signature of immunoreactive. The low copy numbers usually harbors specific genomic alterations in in the P10, PK3CA pathway. They're usually the genomic signature is hormone responsive, it's the largest group, as I told you before, and is associated with the low with low grade endometriosis histologies. And finally, the high copy numbers, around 25% of these are grade three endometriosis, but most of them are associated with the serous histology. And the specific genomic variants in these type of tumors relies especially in TP53 alteration as seen in basal-like breast cancer and also in ovarian cancer. As I told you before, this is the subgroup harboring the worst, the worst prognosis. So this, despite there's a high frequency of genomic variants in each subgroup, matching patients to the appropriate target, the therapy, it's not uniform and remains elusive. And that could be explained because not all of these genetic alterations have a good predictive value, and also our incomplete understanding of the biology of these tumors. Regarding treatment, for the first-line systemic treatment, platinum-based chemotherapy is the main strain for systemic treatment in advancing endometrial cancer. Usually, in the first-line setting, this chemotherapy is quite effective and produces response rates over 50%. But as you can see, the progression-free survival is around one year in this type, in this, in all of types of endometrial calcinoma, which means that most patients who are diagnosed in stage three or stage four disease will ultimately have a recurrence disease within one year and will need further systemic treatment even after an initial good response with platinum-based chemotherapy. And for the second-line chemotherapy, usually the standard platinum-based chemotherapy has much lower chances to response, which drops from 50 to 20%. And usually in the second-line stage, the chemotherapy by itself controls the disease just about three to four months and the expected overall survival with the second-line chemotherapy is expected overall survival with this type of treatment is less than a year. So recently, we have a new standard of care for those tumors who are mismatched repair proficient, which is the combination of anti-PD1 therapy, pembrolizumab, plus praslembatinib, which is a tyrosine kinase inhibitor, and recently proved survival benefit comparing with chemotherapy of about seven months. And these patients now have a life expectancy over two years, even if they are mismatched repair proficient. We have seen on the other side that up to 30% of the recurrent endometrial cancer are mismatched repair deficient. And that would predict some kind of response to immunotherapy. So when we treat these patients who are mismatched repair deficient with a single agent immunotherapy, it could lead to overall response rates from 40 to 50%. And what is interesting is that from all the patients who initially respond to chemotherapy, most of them will have an ongoing response to treatment. So the life expectancy is much longer in patients with mismatched repair deficient tumors. But we need to keep in mind that the other 50% of mismatched repair deficient tumor will not respond to immunotherapy. So it's not a uniform, it's not a 100% accurate predictive biomarker. So we currently don't have such predictive biomarkers to tell us who which patients are going to respond to or which patients would not respond with immunotherapy or with combination strategies. And the future for clinical genomics in oncology that's here when comes our studies like translational studies like VENUS can facilitate longitudinal and real-time molecular somatic profiling. And this is complementary to the clinical pathologic characteristic of the of our patients. And these translational studies like VENUS can provide insights into genomic vulnerabilities and mechanism of resistance and response to these very heterogeneous tumors. Okay, so to put everything into context, we will present three clinical cases of patients treated here at our clinic at Princess Margaret and correlate the clinical characteristics with the genomic findings to see, highlight the outcomes of and the likelihood of response of these kind of treatments. So for the first case, it's the first patient is a 49 year old female with recurrent grade two endometrial endometrial carcinoma with mismatched repair deficient tumor. So this patient, this patient presented in 2019 with abnormal uterine bleeding. She had DNC deletion in puritage, which show a high grade adenocarcinoma of the endometrium. And for their studies, these can show no evidence of this disease, but so she had a primary surgery. During the surgery, they found that the patient was a stage four, because of extrauterine serosal involvement and mesenteric nodularities in the in the small bowel, and she was the bulk to microscopic residual disease. The pathology report showed grade three, I'm sorry, I said grade two, it's it's grade three endometric endometrial carcinoma, which was a P53 wild type. ER and estrogen and progesterone receptors were negative, and she was mismatched repair intact on the immuno histochemistry. That pathology sample was further submitted for the venous study, which will be discussed further. So after surgery, the patient had three cycles of carbotaxel to paclitaxel, we usually give six cycles, but the interval scans after three cycles, showed progressive disease with a bulky peritoneal deposit, which poor response to chemotherapy. And after the initial progressive disease, she was enrolled in in March 2020, in the NEC, which is a phase two trial, evaluating the Starling map, and TPD1 treatment plus Mirapariv and IPAR. She was enrolled in March 2020. And she has been with the same treatment since then. So by now she has had 28 cycles and her latest CT scans from October this year showed no evidence of of progressing disease. So this is the first the first example of a patient with grade three endometrial endometrial tumor mismatch repair intact that had a very good response on the Starling map plus Mirapariv. This clinical trial called NEC, it's a, as I told you, it's a phase two trial with recurrent endometrial cancer previously treated with platinum based chemotherapy like this patient. So this patient had a clinically very good response to the to the treatment. Okay, we will discuss now the molecular, the genomic findings. Yeah. Hi, so what I'll do right now is review her molecular results, and then go through some brief background on the clinical utility of tumor mutation burden and certain genes like poly and how we're handling these findings in our study. So in this case, there were more than 400 variants detected across 230 genes. So here I'm showing an analysis of that gene list. So here I'm showing an analysis of that gene list using a tool from reactome, which displays the level of enrichment of these genes and pathways here indicated in shades of yellow. So we saw an over representation of genes involved in the DNA repair pathway as we might expect in this case. And one of the genes in this pathway. Oh, sorry. I think we just stopped sharing for some reason. Oh, that's not what I wanted. Oh, sorry, guys. One second. Current slide. Okay. So yeah, one of the pathways that we saw was the DNA repair pathway that was over represented. And one of the genes in that pathway is the pulley gene. So for brief background, pulley encodes for the DNA for a catalytic subunit of the DNA polymerase epsilon complex, and it's involved in proofreading activity, largely mediated by this exonuclease domain. And what happens is when the exonuclease domain is altered, the proofreading function is impaired and leads to defects in DNA repair replication, and ultimately results in an increase in genomic mutations and leads to overall high tumor mutation burden. Briefly, somatic variants in pulley have been reported and well documented in colorectal and endometrial cancers. And then in the germline context, pulley variants have also been reported in affected family members, and individuals affected with colorectal, endometrial, ovarian and brain cancers as well. So in this case, we detected one pulley missense variant, the L424I missense variant, which occurs in the exonuclease domain. Published studies also suggest that this particular variant is associated with the hyper mutated phenotype. So, you know, in functional studies using an excision rate based assay demonstrated that this variant does indeed result in reduced exonuclease function, and that the mutation load actually varies depending on the amino acid change occurring at this particular codon. And so for this particular change, this has been associated with the hyper mutated phenotype. Interestingly, another variant located at this particular codon position, the valine change has been reported in the germline context in colorectal cancers and even segregating with disease in a few of the families. So all of this is to say that the we detected a pulley variant and that was associated with a hyper mutation phenotype, which indeed fits with our case. The other thing we wanted to do is determine a way to measure tumor mutation burden. And briefly, tumor mutation burden or TMB has been mainly assessed for Foundation Medicine MSK impact. And in both cases, what's been generally found is that the treatment with immune checkpoint inhibitors or ICI are associated with better response rates among the patients with a high tumor mutation burden. So in the SAMHSA study here below, they looked at all tumor types and selected the group with the highest mutation load. So the top 20% of cases that had a higher TMB and show that this group had overall better survival for those treated with ICI, which was not observed in the top 20% cohort that were never treated with ICI. And then furthermore, in that same study, they found that when you stratify with each tumor type, so the relative TMB is defined by the top 10% or the top 20%, there is better overall survival when compared to the the bottom 80% in that group. So we applied the same principles to our study and using the BioDivaVenous cases, which includes both ovarian and uterine cancers, we stratified based on these tumor types, and then plotted the distribution of mutations per case. And what we found in this sort of preliminary analysis is that compared to ovarian cancers, endometrial cases, you know, within this top 20% cohort accounts for cases with relatively higher mutations per megabase, and includes these exceptional ultra hyper mutated cases. And our particular case that we're discussing here is within this 98th percentile. So another thing to note, you know, in this analysis is that this might illustrate why ovarian cancers generally don't respond to immunotherapy, as these cases also tend to have relatively lower TMB. So the take home messages for our first case is that this patient with endometrial cancer belongs firmly to that ultra hyper mutated subtype, and also demonstrates remarkable clinical response to immunotherapy, which is in line with what's been published in the last few years. And to highlight another interesting point, I think is that these ultra hyper mutated cases tend not to be MMR deficient, nor the MSI positive, suggesting that potentially, you know, there may be different disease mechanisms at play here, or that perhaps too much mutational load or burden would be absolutely lethal to the tumor cell and ultimately not advantageous for survival so that we don't tend to see TMB high plus MSI high and MMR deficient. So I think we're going to move on to the second case now. Okay, the second case is about a 66-year-old patient with advanced endometrial cancer, which was mixed histology, clear cell, and high-grade endometrium, and mismatch repair deficient. This patient presented back in 2017 with postmenopausal bleeding and abdominal discomfort. She had initially endometrial biopsy on October. This reported a high-grade adenocarcinoma with clear cell features. Investigations and further CT scans show locally advanced disease with involvement of the uterine body, low uterine segment, parametria, and islet lymph nodes. This patient underwent neoadjuvant chemotherapy with six cycles of carboplatin plus Taxol, followed by external beam radiotherapy, 45 rates in 25 fractions, completed in May 2018. And then she had a good response and proceeded with surgery in June with a hysterectomy, bilateral salpingo-operectomy, and a mentectomy, followed by vaginal brachytherapy, 11 rates in two fractions. So, the pathology report in this case was mixed clear cell and high-grade endometrial carcinoma with estrogen receptor positive, progesterone receptor positive, P53 overexpressed, HNF1 positive, which confirmed the clear cell feature, and loss of MLH1 and PMSA2. Scans after she completed treatment show no evidence of residual disease, and the patient underwent surveillance until mid-2020. On September 2020, she had some CT scans, and they show progressive disease with new peritoneal nodules. They were biopsy, and that's the biopsy that we use for the venous study with the recurrent peritoneal nodules. And the biopsy was consistent with the previous diagnosis with high-grade endometrial carcinoma, and we used the chemistry showing consistency with the previous diagnosis. She was then, in December 2020, she was enrolled in the LIP Phase III trial. This clinical trial, it's a Phase III randomized trial for patients with Stage III, Stage IV, or recurrent endometrial cancer, and in the first-line therapy, meaning that they shouldn't have received any chemotherapy previously for advanced disease. In this case, the patient had chemotherapy in the neoadjuvant setting, so she was eligible to this trial in the first line. So, this patient was randomized to pembrolizumab plus lembatinib in December 2020, and she's been in the same treatment for almost a year. She has received 15 cycles of treatment, and her latest CT scan showed complete response by resist criteria. Okay, so in this case, we detected a high-ish level of mutation. 70 variants were reported, including a KRAS G12C and a BRCA frameshift variant, which has also been previously reported as a pathogenic variant in the germline context. The TMB profile for this case is moderately higher. It's within the 65th percentile, but one of the more interesting findings here was this Pol D1 frameshift variant, which was detected and occurs outside of that exonuclease domain. Briefly, Pol D1 is part of the DNA polymerase delta complex, which, like Pol E, mediates DNA replication and repair. Germline mutations of Pol D1 have been associated with susceptibility to colorectal and endometrial cancers and an increased risk of breast cancer as well. Three things I wanted to make note about this particular variant is, firstly, this variant is a deletion of one guanine within a track of guanines that continues actually into the splice site junction near the C-terminal end of the gene, particularly within the exon 24 area. So, even though by nomenclature, this variant is predicted to result in a frameshift, it's actually unclear how this might also or may in fact alter splicing instead. But ultimately, we aren't able to predict the functional consequence of this variant and leads me to the second point, which is this variant has not been functionally characterized in the literature. However, the third point I wanted to make was, though, even though this variant occurs outside of that exonuclease domain and occurs within the zinc finger domain, it turns out that unlike Pol E, the hypermutation phenotype is associated with variants distributed across the entire Pol D1 gene, and it's not just restricted to the exonuclease domain like it is in the case of Pol E. And so, it's also worth noting that another missense variant very near to our interest has been associated with a hypermutant phenotype. So, data mining studies also suggest that just by having Pol E or Pol D1 mutation of any type is associated with immunotherapy response, and that this better response rate is also observed even in the context of an MSI-stable scenario. So, to conclude this case, I wanted to link it with another interesting case. We wondered if we saw this particular Pol D1 variant in other cases in our study, and indeed, in another endometrial case that was metastatic and MMR deficient, her NGS results detected the same Pol D1 variant and also had a somewhat higher TMD profile. It was within the 75th percentile range. Interestingly, this patient also responded to a clinical trial of an immune checkpoint inhibitor back in 2015, and she had a complete response and has been on surveillance ever since, suggesting that perhaps this Pol D1 may in fact be functionally and clinically relevant. Next slide. Okay. The last case, the last clinical case, it's a 49-year-old patient with recurrent grade 2 endometriic carcinoma, who is, her tumor is mismatched repair deficient, and this case highlights a different response to immune therapy. So, this patient presented in 2017 with abnormal bleeding and back pain. She had a CT scan which showed a mass in the left psoas with paravertebral involvement of L2 to L5. She had a biopsy of this mass which showed endometriic adenocarcinoma, which was PAX8 positive, ER positive, P53 wild type, and with loss of MLH1 and PMS2. She was therefore diagnosed with stage 4 disease and got first-line chemotherapy with six cycles of carboplatin plus paclitaxel. She had significant response after the chemotherapy, and she underwent a hysterectomy and bilateral salpingo-ophorectomy, and she had almost a complete response to chemotherapy with just one residual foci of endometrial adenocarcinoma, which was grade 2 bifigo, and that sample was submitted from the VENUS study, which Nicole will tell us about the results. Yeah, it was overall unremarkable. She was part of the lower TMB cohort. It was within the 25th percentile, and there were just nine reported variants, all of which were of uncertain significance. This was a mismatch repair deficient tumor. She then got radiotherapy, 45 rays to the psoas muscle, and underwent surveillance for a short period of time. She developed progressive disease in mid-2018 with enlargement of the psoas mass and involvement of the L3 and L4 vertebrae. She started treatment with Magestrol since her tumor was hormone receptor positive, but unfortunately she had a progressive disease after the first interval CT scan with enlarging of the psoas mass. She was re-challenged with carbotaxel for another six cycles, and after she completed chemo, immediately after, she had further progressive disease with enlargement of the paravertebral mass, and also this mass was invading the spinal canal, and they showed new liver metastases. She underwent palliative radiotherapy again to the paravertebral mass because of the involvement of the spinal canal. After the radiotherapy, she was enrolled in the NEC clinical trial, the one that I presented for the first patient, but this time it showed a different response. She was enrolled in this clinical trial in October 2019, and in the first interval CT scans, three months after the beginning of the treatment with the Starlimab plus niraparib, the patient developed significant toxicity from the treatment, and also her performance status declined. The CT scans showed initially a mixed response in the liver with some of the lesions enlarging and some of the lesions were getting smaller, and also a slight enlargement of the paravertebral mass. At that time, the patient was considered to have no benefit from treatment, and it was decided that she was going to get off the trial. We proposed a new second-line chemotherapy, but before the start of the second-line chemotherapy, she developed gastric obstruction. It was nonmalignant, it was related to strictures, and she couldn't get the second-line chemotherapy at that time, so we decided to follow up with new CT scans. All of the following CT scans show overall stability of the disease, and to this point, she has not gotten any new treatment, which in retrospect suggests some benefit of the treatment with the Starlimab plus niraparib. This different response might be related to the progression or even delay response that is sometimes seen in patients treated with immune therapy, and this highlights the different patterns of response to immunotherapy in patients with mismatch repair deficient tumors in endometrial cancer, and also it's a good example of how clinical decision-making can be challenging in this case, given that initially she was told to have progressive disease and no benefit from the treatment, but when we see backgrounds, it is suggested that she may have had some benefit from the immune therapy treatment because she's been overall stable, the CT scans, and she hasn't been in any active treatment since then. Ian? Yeah, I will take over for the last few slides here, sorry, I was finding my cursor, which was lost amongst my screen. So, just to end on sort of a future directions note, the last case especially highlights that we have a lot to learn about biomarkers for immunotherapy. We have a case there where we do have one positive biomarker, MMR deficiency, but a sort of an intermediate response and very little insight from the sequencing that we did. It's our hope that in the future, we'll have some more information from some of these emerging biomarkers. And one of the questions in the Q and A was asking if prediction of new antigens was something that we could think about. And that's definitely an active area of research. It is possible to predict the consequences of the mutations that you see in these tumors. And we can say with some degree of confidence whether the sort of random mutations that you're getting from these DNA repair pathway defects are actually increasing neoantigen load in a tumor cell. This is an area of computational biology actually, and it's hoped that artificial intelligence will help us sort of refine those algorithms. And there's already some published studies about improvements you can make to mutation burden as a biomarker by taking into account what the neoantigens that are produced are. And then taking it a step further, people have published on whether neoantigen genes being methylated is a mechanism by which a tumor evades immune detection and so forth. So it's definitely an active area that we'll probably see more of in the future. Just also to mention that there's a lot of research on tumor and immune microenvironment. So it's possible to discern the level of immunogenicity of a tumor by looking at gene expression. So if you have a tumor sample that has a lot of lymphocytes, the gene expression signature starts to look like a lymphocyte. And you can also tease out things within a gene expression signature that tell you either in a directed way or an unbiased way, you can tease out things that tell you what the response to immunotherapy might be. Tumor infiltrating lymphocytes, TILs, are also important. People have counted the number and type of lymphocytes and gotten information about response to immunotherapy from that as well. And finally, it's also important to mention that circulating tumor DNA is going to be important in this field in the future. We're not gonna just want to know upfront whether a patient has a high probability of response. We're going to want to monitor that response as the patient is treated. Immunotherapy, of course, has a lot of side effects and a lot of sort of costs to it. So if a therapy is not working, we want to know very quickly. And circulating tumor DNA is a way to know that to gauge the response to the tumor. Also, people have just looked at marks of response in the immune system. Neutrophil to lymphocyte ratio is one thing that people have looked at. And these other sort of biomarkers can be useful in determining whether a tumor is responding to immunotherapy. So plenty of things on the horizon and in addition to mutation burden and the biomarkers that we've been looking at. So if you could advance next slide. So just to summarize what we've talked about, there's good data to divide endometrial cancer into a number of molecular classes. These classes have importance for prognosis and response to immunotherapy. And we've shown you some of the biomarkers that can be used to predict response to immunotherapy that relate to these molecular classes of endometrial cancer. In particular, we've talked about mutation burden and poly Pald1 variants and how they can complement the current sort of standard care biomarker MMR deficiency to predict response to immunotherapy. So next slide, please. This is just one more slide to show you that it's probably a complex picture even in the biomarker part of it. And we shouldn't think that one biomarker is necessarily going to give us all the information that we need. This is a study from the same group that you saw with the data mining study of poly and Pald1 variants. This is showing that in large clinical data sets, this is three separate clinical cohorts, patients that had one mutation in multiple DNA repair pathway. So a mutation in the mishatch repair pathway plus a mutation in the basic excision repair pathway in say poly. If a patient had more than one pathway mutated, DNA repair pathway mutated, they have a higher probability of responding to immunotherapy. So it's just a sort of a taste of what can be, maybe teased out by looking at multiple biomarkers and not just concentrating on one. And we've shown that in our clinical cases, the first case in which the patient was MMR proficient, but hyper mutated is an example of the interplay between those biomarkers. So last slide, please. So the last thing I'd like to leave you with is also that it's important from just a practical lab perspective as well. Poly and Pald1 mutations are something you can get on a small panel. Tumor mutation burden requires a large panel of at least a megabase. In our neck of the woods, Cancer Care Ontario recently approved funding for biomarker testing for sequencing of Poly1 and Pald1 and endometrial cancer. So this is now standard care where, or at least this is paid for where we are. Tumor mutation burden is not something you have to be a part of a study to get a large sequencing panel for or to self pay. So this is a, at least in the short term, also a practical consideration and working on just one biomarker like tumor mutation burden is probably not a practical lab strategy either. So I think we'll conclude there and open to questions. Well, thank you very much for this very interesting talk and the different types of cases then in the end with respect to immunotherapy. Also, thank you that you already answered so many questions and especially the question of how tumor mutational burden then really relates to the effects of immunotherapy. And there's an additional comment now in the Q&A that it seems that like only about 10% of the predicted neoantigens actually have confirmed immune recognition. How would you take that information into account when you think about approving then immunotherapy for a patient? Yeah, there was some good questions in the chat about that. Yeah, neoantigen prediction is like I said, an emerging field. I don't, I mean, we certainly haven't delved into using that clinically and I don't, I'm not aware of anybody who's kind of actively using that clinically, but it's definitely, you know, it's another piece of information. Yeah, it's, even with high tumor mutation burden, there are plenty of non-responders. So there's plenty of other things that are going on. What seems to be important is sort of the breadth of neoantigens too. If you're presenting with a single neoantigen that's highly immunogenic, that can predict a good response. So there, but there's a lot of subtlety in that and that's sort of yet to be worked out, but it's definitely a tool that will enhance the use of mutation burden as a biomarker in the future unquestionably. All right, thank you. Also, maybe it's in the same line. Then the question is, you showed the data on the different endometrial and ovarian cancer and what the spectrum of TMP is. How, what do you think, how many cases would you need to actually identify what kind of spectrum do you have? Yeah, using the strategy identified by MSK in that Sandstein paper, where they look within a tumor type. There's two considerations. There's the histology. So just bundling things into histology to treat them as the same tumor type. For instance, in endometrial cancer, we have serous endometrial cancer, which is a different pathologic entity and molecular entity than endometrioid adenocarcinoma and things like that. It's fair to group some of those things. It's not fair to group other things. Like for instance, in our analysis, we kept serous ovarian cancer or endometrial cancer out, but we lumped all other endometrial cancers into one bucket to do that analysis. And if that's just a judgment call that you make. Numerically, you just have to have enough cases that the distribution isn't changed by adding a case. So, you don't want the number that you get for the percentile rank to change by 10% every time you add a case or something like that. So it just takes a little, you have to have high enough numbers for the numbers to stabilize. I would say empirically 50, 100 cases and you get pretty stable numbers. Okay, yeah, that sounds reasonable. When you were pointing out that you do have, of course, a different type of classification than just the molecular classification, which you showed here, what type of cancer classification do you actually use in your lab? ICD, Oncotree, or what are you using? I think that question means to group the... In clinical reporting, we don't actually do any other classification other than just endometrial cancer, basically. I mean, just practically speaking, these come through with histology reported by the pathologist. Sometimes that histology is fairly exact. Sometimes it's not. Sometimes it's just the differential on which endometrial cancer is one. And it's actually, clinically, it's just very hard to get into that kind of detail. So we don't really delve into that level of classification for a clinical case. I think Beth had a response to that. Yeah, and do you use a certain classification then for your study in addition to the molecular one? Oh, I see what you're saying. No, we haven't done data analysis using any sort of classifications yet, but it's definitely something that could be important when we start to analyze things. Right. Yeah, and there's another question. So you've been focusing now mostly on immune therapy by itself. And then there was the question at the beginning, how many KRAS mutants you actually have? And now then the question would be how valid is a combination therapy? Do you actually have any patients treated with a combination therapy against KRAS and ICI? Yeah, we have... Eduardo might know better than I. I don't know of any patients treated with sodaracid or anything like that, but Eduardo, do you know if that's a thing? No, we don't have that kind of treatment for our patient with endometrial cancer. We don't have any patient being treated with sodaracid. So it's proven effective in lung cancer with the KRAS mutation, but my understanding is it hasn't been proven yet in the endometrial cancer patients. Yeah. Yeah, I'm not aware of any trials at least going on at Princess Margaret. And the question asked previously was what the frequency of G12C was in endometrial cancer. KRASs are, and just to answer that orally, KRASs are reasonably frequent, but not overwhelmingly frequent. And G12C is not any more frequent than any other activating KRAS mutation. It's not biased for or against. So, yeah, there's some, there's definitely some cases that would have that and might potentially be targetable. Yeah, okay. We have one more functional question. I hope we can have the time for that. The question is regarding the Pol D1 variant. This is a variant of unspecific significant. Why do you think it might be important? Says it would be seen based on the underlying sequence that the observation in two different tumors is mostly due to the nature of the repeat sequence, especially since loss of function variants are not generally considered pathogenic in Pol D1. Yeah, that's a great question. Yeah, it's a great question. First of all, it is in a repeat sequence. So it's a deletion of a G and about a poly G round of about six, I think. We, if it's an artifact or if it's something, if it's just due to the repeat difficulty of sequencing it, we'd expect to see it in more cases. We actually only see it in the two out of, we've profiled about, I wanna say 800 cases right now for Biodiva venus, and we only see it in two. So it's not a common artifact. We're not actually sure that it's a loss of function variant. It, like I said, it affects a poly G run that covers a splice site. So it's possible that if splicing is preserved, it results in a frame shift. It's possible that it affects splicing or it's possible that it has basically no effect at all. We just don't know. And it's actually really hard to parse that out from anything but a functional study. So I would just keep an open mind. I mean, it's not, and also to emphasize, Pol D1, the mechanisms of pathogenesis are much less understood than poly E. Poly E, we have this pretty good idea that we have hotspot mutations in the exonuclease domain that are, that comprise a lot or most of what's going on in poly. With poly, as the results that Nicole showed from that large study, it's less clear with Pol D1 what's going on. So I would just, again, keep an open mind as far as Pol D1. Pol D1 mutations are much less frequent than poly. They're about 10 fold less frequent. So it's, we just don't have very much data on Pol D1 right now. Yeah. Okay. Thank you very much. It's been a very interesting discussion. And I think we would have probably a lot more questions coming up, but we're at the end of our session. It would be really cool if you could just pass one more slide for everyone. There's a call for abstracts for the annual ACMG meeting next year in March 22. And the deadline is November 19. So it's just eight days from now. So please consider sending in your abstract. And also the video from the last meeting in October is already available online. And this video will also be provided for you for your convenience within about two or three weeks. So thanks everyone for joining. Thanks for this great presentation and discussion. And it was really interesting and hope to see you next time again. Thank you.

molecular biomarkers

response to immunotherapy

endometrial cancer

molecular subgroups

prognosis

tumor mutation burden

Poly D1 variants

tumor microenvironment

circulating tumor DNA

biomarker testing