Adam ElNaggar: My name is Adam ElNaggar. I'm a medical director of oncology at Natera and my focus is on Signatera, our molecular residual disease test. And so my disclosures, I work at Natera. And then our objectives are just to present Signatera, to talk about some of the prognostic and predictive value of ctDNA and then one of the last was to talk about some of the pharmaceutical industrial partnerships that are used in clinical trial design and so I'm going to talk about how ctDNA can be used for that purpose.

And so what is Signatera? It's a personalized tumor-informed approach. And so what we mean by that is that it's based on that patient's tumor. So there's not a bladder panel or a colon panel or a lung panel. It's truly that patient's. And so to do that, we need the patient's tissue and we do whole exome sequencing of that to identify those single nucleotide variants that we feel are going to be the clonal or passenger mutations that are going to persist throughout the life of the tumor and not susceptible to treatment pressure. And so we identify 16 variants there. We also run the patient's germline and do whole exome sequencing of that as well, because what we're trying to do is make sure that any SNPs that may be present in that patient's germline are not being included in the panel. And then once we have that completed, we then run that against the patient's plasma identifying their circulating tumor DNA.

And so there's tumor-naive and there's tumor-informed assays. We feel that the tumor-informed assay's very important because you're tracking multiple variants specific to that patient, giving you more shots on goal to identify if there is residual disease. You're also ruling out any germline or CHIP mutations that may be present in the patient. You're also able to quantify response to therapy because you're going to identify how much ctDNA is still present while after surgery or undergoing treatment. And we're also tracking those clonal mutations, like I mentioned, ones that we don't think are going to go away under treatment or has it recurs multiple times.

And so how has Signatera performed in muscle-invasive bladder cancer? And so you looked at Signatera at multiple points along a patient's journey within muscle-invasive bladder cancers, so treatment response monitoring, pre-, post-surgery, looking at efficacy of adjuvant treatment and then the idea of treatment response monitoring, as well as in surveillance, so being able to identify it before there's clinical signs of progression. And so one of the first studies that we used to begin really looking at that role of ctDNA was by Christiansen et al. And so it was 68 patients with muscle-invasive bladder cancer with serial time points collected. So at the time of diagnosis, so after transurethral resection, during neoadjuvant, pre-, postoperative cystectomy and then during surveillance.

And so what we saw very early on was that there was a prognostic benefit of ctDNA. So just after diagnosis, if you were ctDNA negative, you can see that the recurrence-free survival for those patients is excellent. Only one patient in that negative cohort recurred, but that was over a year out. And then the positive group you can see there was not as favorable an outcome. If we then look at different time points, so on the left there again as that pre-neoadjuvant and then we look at testing on neoadjuvant or right up until surgery, before surgery if you cleared, and again, we're seeing a very big separation of those curves. And then if we look after cystectomy and said, "If you're negative after surgery and remain negative during surveillance," those patients have an excellent outcome, a very durable recurrence-free survival. Whereas if you go from negative to positive sometime post-cystectomy, 100% of the time those patients were going on to recur. And a similar finding was seen in the overall survival as well.

And so if we're identifying a group that's at much higher risk of recurrence, can we start using that information to identify who would benefit from adjuvant therapy? So now we're trying to look at a predictive biomarker here. So IMvigor010 was a study looking at atezolizumab versus observation in muscle-invasive urothelial cancer. So patients were randomized to atezo versus observation and an exploratory endpoint was to look at biomarkers. And so you can see, 800 patient study of atezo versus obs, the intent to treat population is a negative study. And then if we looked at biomarker-evaluable patients, so you had PD-L1 or TMB, again, there was no separation of those curves with the addition of treatment.

So then if we look at that observation arm and say, "I want to look at that prognostic value of ctDNA again," you can see DFS and OS, the ctDNA negative group had a far superior outcome then the ctDNA positive group. So then let's look at this by treatment there, by those two different cohorts. And so you can see in the faded green and purple there, those are the ctDNA negative group, treatment and no treatment, and you see those lines, there's no significant difference in those curves, whereas if you looked at the ctDNA positive, you had a significantly improved disease-free survival and overall survival with the addition of atezolizumab in the ctDNA positive arm.

And so the updated overall survival was just presented at EAU this year. And so you can see here that the benefit of the addition of treatment to the ctDNA positive cohort persisted over several years. If you look at the ctDNA negative group, there starts to be this interesting trend towards decreased survival in the patients that took atezolizumab relative to those that underwent observation. And as we've done more immunotherapy, more targeted therapy trials in an all-comer population, this idea of a possible detriment in survival to patients that did not benefit from treatment is starting to be seen. And so how can we start to use ctDNA within clinical trial design?

And so to that regard, the FDA recently issued some draft guidance on how to use ctDNA within clinical trials. And one of them was just the fact that it does identify a patient that has a poor prognosis and selects a higher risk patient population. And so if we look at some of the big trials that have come out recently, so PALLAS was a hormone positive, HER2 negative study. APHINITY was one, looking at HER2 positive patients. But those studies required 6,000 and 5,000 patients respectively to show efficacy of the drug. Similarly, IMvigor, like we've mentioned, was about an 800 patients study. If we enrich that for ctDNA positive alone, we can see that the two breast trials would've drastically reduced to just under 500 patients if you enrich for ctDNA positivity. The same thing in IMvigor. Suddenly that drops down to about a 150 patient study.

And so you're really identifying a way to enrich the study to find that efficacy in less time and far cheaper. And so biomarkers have certainly been something that has become more and more present in the field of oncology. So there's the diagnostic markers, the prognostic markers, like we identified earlier, looking at who's got a better outcome, and then the predictive biomarkers. Now predictive biomarkers are certainly being used in clinical trials already. Specific examples would be tumor mutational burden, PD-L1 expression and microsatellite instability. One of the issues is they all lack specificity. So we all see this favorable cohort, these markers that should identify patients that would benefit, but they are not, as well as a lack of standardization, both between measurements, companies that are measuring TMB or PD-L1, as well as what studies use as the cutoff.

And so when I look at IMvigor010, what we see here is if you looked at the ctDNA positive patients and looked at the TMB high ones, there definitely was a benefit to the addition of atezo. If we look at the TMB low, still looking at ctDNA positivity, we see that there was not a significant benefit to the addition of atezolizumab. Similarly, PD-L1, if you were an expressor and received treatment in that ctDNA positive cohort, there's benefit whereas it was not seen in that PD-L1 lower negative cohort. And so we're starting to see the synergism here where using biomarkers to identify who would most benefit from treatment and enriching it for the ones that really need the treatment can start really shaping clinical trial design.

So those are some of the use cases. And then also treatment response monitoring can play an important role. That's particularly seen in the immunotherapy or immuno-checkpoint inhibitors where we know there's this phenomenon of pseudoprogression where radiologic imaging may not be able to distinguish who's actually progressing or who's having the idea of pseudoprogression happening in patients. And so if we look at IMvigor010 and say, "If you're ctDNA positive at the start of treatment and just six weeks in, so it's cycle three, cleared, how did that correlate with disease-free survival and overall survival?" And we can see that if you go from positive, to negative even very early on at six weeks, we're identifying a patient cohort that's really going to benefit and hopefully have a durable response to therapy.

One of the big studies that was used to look at this was the paper by Bratman et al. And so that was the INSPIRE trial. And what that was was a phase two trial looking at pembrolizumab across solid tumors. And it was enriched for different cohorts, but the idea was these recurrent solid tumors went on to study and receive pembrolizumab. And we can see that in the blue line, those are patients that were ctDNA positive and had a decrease in the amount of ctDNA relative to the green line, which is those that had an increase. If you had an increase in ctDNA, those patients had a progression-free survival just over two months and very few of them went on to have a response. Certainly if you have progression in the tumor, as well as a rise in ctDNA, none of those patients went on to achieve an overall response rate.

And I think it's important to point out that there was a 98% ctDNA detection within this patient population really highlighting the need for that tumor-informed assay. We can then separate some of these responders out a little bit further by saying, "If you're ctDNA positive and you go on to clear ctDNA, how does that correlate with survival?" And you can see it was 100% overall survival for those patients that cleared ctDNA. Then there's also those that decrease but don't go on to clear, as well as those that increase. And you can see that the survival decreases with each of those groups. And so again, very early on, as early as six weeks, both in IMvigor and this INSPIRE trial, we're identifying who's responding to treatment.

And so when starting to design clinical trials, we're seeing that ctDNA correlates with the overall response rates. And so there is Medicare approval for treatment both in the pan tumor immunotherapy monitoring, as well as in muscle-invasive bladder cancer. And so this ctDNA can be used as a early endpoint in clinical trial design. So not fully validated yet, but one that is certainly being discussed in the FDA's draft guidelines is that this may be helpful in early signal finding of drug activity, so very early trials being able to identify are patients clearing, are they responding. And so then ctDNA can start to be used as a surrogate endpoint. The idea being, for example, if 70% clear with drug A and 30% clear with drug B, before waiting for the full event rates that may take a significant amount of time and money to identify, you're identifying very early on who may be benefiting most from treatment and whether the treatment is efficacious.

And then there's also the recurrence monitoring, that there is a significant lead time to the detection of recurrent disease and that's been seen across multiple solid tumors. You can see here that the specificity approaches almost 100%, such that if you became ctDNA positive, those patients with enough follow-up time become positive and the lead time varies across solid tumors. And so there already are trials being developed that are using Signatera or molecular residual disease monitoring as part of either the inclusion criteria, has an integral or also has an integrated biomarker within studies. One of the larger studies to highlight this was the GALAXY study. And so it was a three arm trial. GALAXY was the arm that was just looking at observation. The other two arms that compose the entire study was the VEGA arm where if you were ctDNA negative, you went to a deescalation arm and if you were ctDNA positive, you went to an escalation arm.

Recently at ASCO GI, the GALAXY study, so the purely physician's choice prospective, just observation of what does the ctDNA do, we identified that prognostic value. So just looking at a single time for four weeks after surgery, if you're ctDNA negative versus positive, we're really separating out the good prognostic patients from there. And then if we look across different disease stages, the higher risk stage two, three or four, again, we're identifying a cohort within that ctDNA positive group that's benefiting from the addition of adjuvant therapy. So these patients are all ctDNA positive and the green line's with treatment and the blue line's without. Then if we say, "Let's look at that ctDNA negative cohort across those same populations," and you see they have an excellent prognosis at baseline and that the benefit to the adjuvant therapy then wasn't there. And so I hope I've been able to highlight some of the uses of ctDNA within clinical trials as both a prognostic and a predictive biomarker and some of the use cases that exist currently for Signatera. Thank you.