David Gandara: Hello, I am David Gandara from the University of California Davis Comprehensive Cancer Center where I'm a professor emeritus and co-director of the Center for Experimental Therapeutics in Cancer, senior advisor to the director, and also the chief medical officer for the International Society of Liquid Biopsy. [00:05:00] Welcome to this virtual clinical utility study. It's being done to assess use of a novel biomarker in the treatment decision-making for immunotherapy-eligible first-line advanced stage non-small cell lung cancer patients.
I'm going to go through some of the background and rationale. These are my disclosures [00:05:30] and I start with this slide, which I put together, of all the immunotherapy first-line trials and advanced stage non-small cell lung cancer. I show this slide not to dwell on the individual differences between the studies or the drugs involved, but rather to talk about the fact that they are incredibly diverse.
They're diverse in terms of the treatment regimens [00:06:00] whether it's monotherapy or immune checkpoint inhibitors or combinations with chemotherapy or even bevacizumab or combinations IO/IO together with CTLA-4. They're complex and diverse in terms of biomarkers whether there are none or PD-L1 or TMB. If so, what is the level of PD-L1? Diffuse in terms of histology, many of them are across both histologies. Some are squamous [00:06:30] only, some non-squamous. That does make a difference and also what is the primary endpoint?
The bottom line for this slide is that a treating oncologist in the United States has many different options. How do they go about sorting out who to treat with what? This is an algorithm that Solange Peters first developed a couple of years ago and I modified it to catch up with the current landscape. [00:07:00] This would be what a typical oncologist would do seeing a new stage four patient, non-small cell lung cancer, advanced stage. They would do next generation sequencing.
That's preferred in all of the guidelines with or without TMB. They would also do PD-L1 immunohistochemistry. If an oncogene driver is found, which by guidelines would require that you treat with a targeted drug upfront, then that would be done. [00:07:30] For everyone else then, they are candidates for immunotherapy whether non-squamous on the left or squamous on the right. Typically, an oncologist would look at the PD-L1 score of 50% or greater in order to make a decision.
You can see their options would include several PD-1 or PD-L1 inhibitors with or without chemotherapy. However, if the [00:08:00] PD-L1 level in that patient's cancer is lower than that, there are a variety of options as you can see below, including the addition of bevacizumab as in the IMpower150, or IO/IO combinations either by themselves as in the CheckMate 227 or in combination with chemotherapy, for example, 9LA. What are the limitations of our current biomarkers?
[00:08:30] This is a very nice slide, although it's quite old at this point, that shows that in the development of checkpoint inhibitors in all of these studies regardless of the tumor type, regardless of the PD-L1 assay that was used, those patients that have positive PD-L1 scores in their cancers do better than those [00:09:00] that do not. This is generally accepted across all cancers, but how we measure PD-L1 and what we do with the information is dramatically different.
Most of you who are listening to this know that non-small cell lung cancer is pretty much an exception where we use TPS, that is the tumor score, only. We do not use the IC or the immune infiltrating lymphocyte portion [00:09:30] of the score, which is done in most other cancer types. We use what's called TPS. Most other cancer types are using CPS. I think in some ways that is a hindrance to our use of PD-L1, but that's the way it is and all the studies have only used TPS since almost the beginning. Now here's the importance of this. As you can see on the far left, about one-third of patients [00:10:00]
David Gandara: Will have a PD-L1 that is very low. The PD-L1 greater than 50% represents a high proportion of patients. And you can see on the left that in the keynote 24, which was restricted to a TPS of 50%, that those patients who received pembrolizumab, did very well. This is a landmark study, it established single-agent immunotherapy. [00:10:30] On the right is another landmark study keynote 189. That is the combination of pembrolizumab with chemotherapy, and this is in non-squamous cell lung cancers. And you can see across the PD-L1 levels, that there is benefit to the combination over chemotherapy alone. But you can see the degree of benefit is less in the patients with a TPS at less than 1%. Nevertheless, it is there. [00:11:00] So PD-L1 has limitations, it's not a complete biomarker, and neither is TMB.
This is the five-year update on that checkmate... I'm sorry, on that Keynote 189 study. The design again is shown at the top, the combination of pembrolizumab with platinum chemotherapy versus platinum chemotherapy alone. And the point of this slide is at the five-year survival rate. [00:11:30] Even with a regimen like this is relatively low, you can see it's about 20% overall survival in those patients who received the combination, 11% with chemotherapy alone. Although there was major crossover in this study, so many of these patients, the majority did get immunotherapy afterwards. And then at the very bottom you can see really differences between PD-L1 greater than 50%, 1-49%, [00:12:00] less than 1%, but it's there for both chemotherapy and for the combination.
So are there other biomarkers for immunotherapy? And this is quite different from oncogene driven cancers where we have a mutation and it's present or absent. Most of these biomarkers that are proposed regardless of whether it's tumor neo-antigenicity or tumor immune suppression or invasion, these [00:12:30] are continuous biomarkers. And the ones that we use typically are PD-L1, as we discussed. TMB, which I won't discuss further in this presentation. But more and more we're realizing that the tumor microenvironment needs to be assessed, and it needs to be done in a way that compliments and is an addition to PD-L1 level.
And one of those assays is the PROphet assay, [00:13:00] which is a plasma proteomic signature, we'll talk about that in detail. And this just shows in a patient, actually in two patients, patient A and patient B, that they might have differentially expressed proteins, and would this would result in a high probability that they would benefit from immunotherapy versus a low probability, and we'll go through this in some detail. Well, what is the PROphet test, [00:13:30] and how would you go about using this? It is a decision support tool or assay, that's designed to predict benefit from anti-PD-1 or PD-L1 immunotherapy for advanced stage non-small cell lung cancer, with or without chemotherapy. It's a simple blood test, which could be done at the same time as PD-L1 testing and genomic sequencing, so that it would not delay [00:14:00] treatment, for example. Also, it can be done at the patient's home if that is most convenient.
This is a high-throughput proteomic assay, which includes complex bioinformatics and machine learning to refine this assay. So the blood sample is analyzed and processed and a clear registered and polar accredited lab, which is in North Carolina. The median turnaround time for the assay is seven days, [00:14:30] and a PROphet score is developed. Here you can see an example of what that would be. And this example, it's 3.6, and the interpretation is that PROphet positive is in the five to 10 level, negative as less than five. And this is then used in a decision making process together with PD-L1. And I'll show you how that is done a little later.
[00:15:00] Well, what does the test measure? That measures a variety, a multitude of plasma proteins that have previously been associated with clinical benefit to immunotherapy. These proteins are then summed to generate this score, the score I just showed you. And this score is then linearly scaled to a value, and it's converted into a probability of clinical benefit. So the test [00:15:30] informs you, PROphet positive, PROphet negative together with PD-L1, in terms of the probability of clinical benefit. Now, during the development of this assay, clinical benefit was defined as progression-free survival at 12 months. So that's in the development phase, but as I will show you, this was then translated into overall survival in real patients.
And this shows [00:16:00] how that development and validation was done. 228 treatment naive first-line patients then treated with immunotherapy were used in the development cohort, 272 treatment naive first-line patients treated with immunotherapy in a blinded validation cohort. And then a cross-validation in 500 patients. There were also 85 treatment naive first-line [00:16:30] chemotherapy-only patients that were used as a control cohort. Now here in this extensive table are the clinical characteristics, which we typically look at in terms of prognostic factors for immunotherapy. And you can see they include everything from the performance status to the PD-L1 level to histology. Also clinical benefit you can see there, [00:17:00] and this is across the development phase, the validation phase, and the chemotherapy alone cohort.
This is the independent validation set, and it shows you the association of the clinical benefit rate, as I said, defined by progression-free survival at 12 months, versus the predicted clinical benefit probability from the PROphet test. [00:17:30] And as you can see, there is a very good alignment, it's quite extraordinary. The R squared here is 0.97. So, how is this information then transferred into developing a score? So, the conjuncture here was that this clinical benefit would also translate to overall survival. You can see that the median level of five was used at the cut point to determine [00:18:00] PROphet negative, versus PROphet positive. And you can see on the right then, this applied to a clinical cohort for overall survival. And you can see positive versus negative. You can see the hazard ratio very good at 0.51, very positive P value. And you can see the median overall survivals for PROphet positive, almost 26 months, about 11 months for PROphet negative.
[00:18:30] But as I said, the test is designed to use it in conjunction with PD-L1 score. So let's look at this group for overall survival in terms of the different PD-L1 levels, and this is the analysis for the PD-L1 50% or greater. On the left, you see those patients that had a positive PROphet test, together With PD-L1 score greater than equal to 50%. And you can see here, [00:19:00] the patient is a good candidate for anti-PD-1 PD-L1 monotherapy, regardless of chemotherapy. And so, this is what many of us do in practice, unless it's a very aggressive cancer, there's some other confounding factors. But, this PROphet test then I think gives you confidence, that monotherapy can be given if it's positive, together with high PD-L1.
On the right [00:19:30] though you see, what if the PROphet test is negative, and the PD-L1 score is still very high? We'll hear the data favor immunotherapy plus chemotherapy, and you can see it's quite a widespread hazard ratio of 0.29. So the interpretation then is that the patient is not a good candidate for monotherapy immunotherapy. So consider using a combination of anti-PD-L1 with chemotherapy. As I'll come back [00:20:00]
David Gandara: ... to later, this analysis did not include IO combinations, so this was not assessed. Now we're going to that intermediate group of cancers where PD-L1 is one to 49%. And you can see on the left profit positive and PD-L1 within this range, this is a little more complex because you see ICI plus chemotherapy, ICI monotherapy, chemotherapy alone. [00:20:30] So, the clear difference here, of course, is between ICI plus chemotherapy versus chemotherapy alone. The ICI monotherapy is in between, but it is a small sample size in this analysis. So, the interpretation is the patient is a candidate for immunotherapy, preferable in combination with chemotherapy. On the other hand, if the profit score is negative and PD-L1 is between one and 49%, [00:21:00] here, the patient is expected to obtain a low benefit from immunotherapy. And a combination with chemotherapy is recommended, as you can see in the Kaplan-Meier curve split.
And lastly, I would say our most difficult group of advanced non-small cell lung cancer patients who are going to get immunotherapy. What if that PD-L1 score is less than 1%, meaning not a single cell [00:21:30] on that immunohistochemistry stained positive for PD-L1. Here, profit positive and PD-L1 less than 1%. These patients by this analysis, can actually benefit from ICI plus chemotherapy. And you can see hazard ratio here very strong at 0.41%. Now what we don't know of course in this situation [00:22:00] is about IO plus IO combinations. Profit negative PD-L1 less than 1%, the patient is not a good candidate for anti-PD-L1 therapy with or without chemotherapy. As you can see, the median survival's quite low. Even in those patients ICI plus chemotherapy, median survival, 7.5 months.
So, this population, again, I said probably the [00:22:30] most difficult. These patients may be good candidates for other sorts of immunotherapy approaches. Something together with a CTLA four inhibitor, we have these FDA approved in the United States such as nivolumab, ipilimumab, or more recently durvalumab together with tremelimumab, both in combination with platinum chemotherapy. Or they may be a candidate for a clinical trial. [00:23:00] So, this just summarizes. On the left, it says, what does an oncologist do if they don't have the profit score? And these are generally two groups, 50% or greater, or less than 50%, or I would argue a third group, less than 1%, and the options. And with profit, you can see this expands to six patient groups. I won't reiterate all of these in detail, but you can see all the categories [00:23:30] we just talked about, the profit score positive in each of the categories for PD-L1 and what the expected outcomes would be.
The report. This is an example of a negative report, meaning the profit score is negative in this particular patient. You can see the score in this example, 3.6. All the [00:24:00] information that I just described is then summarized in that report. You can see the optimal treatment recommendations for each of the groups with a profit negative score. Again, you can see on the right-hand side, negative meaning less than five, low probability of clinical benefit from immunotherapy, positive high benefit. Clinical evidence. [00:24:30] You can see these are the curves that we looked at before. So, these are put into the report with the summaries describing the studies and what the median survivals have been. These are also, these reports are linked to clinical trials, which are available in profit negative patients. And you can see how that you would access those. And then a positive report on [00:25:00] the right, you can see here, again, very similar sort of report, but here incorporating this positive score into the outcomes.
Well, what are the limitations of this analysis that was done? Well, as I already mentioned, CTLA4 combinations with immunotherapy or with anti-angiogenic therapy were not included in the development or validation [00:25:30] of the profit test. But this is being done. PD-L1 status was not determined centrally. It was the assay used. But we, for example, have reported that for our clinical benefit, meaning survival, it doesn't really matter which score you use, which scoring assay you use. PD-L1 status was not known for the chemotherapy only cohort and squamous cell patients [00:26:00] were included. But the sample size is not large enough for us to say that the results would be equal in squamous cell. But again, that work is being done. And with that, thank you for your attention and I appreciate you participating in this survey project. Thank you.
Speaker 4: [00:26:30] Okay. Thank you, Dr. Gandara.
David Gandara: Yeah, I see another major pledge when the slides were redone that I can't conceive that this was done, so I had to talk my way around it. You see here, this is supposed to be for the 30%. It's supposed to say PD-L1 [00:27:00] negative 30%. This is incorrect. In other words, you haven't countered for the group with the one ... Anyway, it's a miss cut and paste of the slide. So, what I would suggest, I went ahead, I think my language was okay, but what I would think is we might be able to [00:27:30] substitute the original i-chart here, which I believe is going to be much more clear than this and just substitute it in. If not, we can rerecord later. But I think it'll be pretty easy just to substitute another slide in there. Everything else on the slide is fine. It's just the pie chart is mistyped.
Speaker 5: Yeah. One-third was [00:28:00] supposed to be ... So, that 30% is the PD-L1 low, below 50?
David Gandara: Yeah. And that was the purpose of the slide is to talk about what happens in PD-L1 low. That's the red box there. See?
Speaker 5: Yeah, I noticed that as well. I mean, I don't think others will. So, can you guys throw a different slide in for his recording or no? Otherwise-
David Gandara: Yeah, if you go back and you take that label and you move the label up where it's right beside [00:28:30] the dark blue box-
Speaker 5: Well that PD-L1 was greater than [inaudible 00:28:34]-
David Gandara: You see what I'm saying? Move the label up here and change it to PD-L1 less than 1%. The label would go there.
Speaker 6: Okay. Yeah, we can do that. I just have have ... Oh, sorry, go ahead.
Speaker 4: I can send you the original slide, if you guys don't have that still. Or I'm sure you do, but I can resend it.
Speaker 6: Yeah.
David Gandara: Yeah, I think rather than us trying to ... If you just do that, move it up here and say PD-L1 less than 1%. The point of it is there's a lot [00:29:00] of patients with a PD-L1 less than 1%, and it's really important. And so, I couldn't say exactly that because if people concentrated on the pie, they were going to say, "What the heck is he talking about?" But anyway, I think verbally [inaudible 00:29:15]-
Speaker 4: Well-
Speaker 5: The assumption is good. The assumption's good.
Speaker 4: Actually, because I did actually go back to the original and I can share my screen, but it looks like that.
Speaker 6: Yeah, that's what I was just going to ask. If the original we have looks like that pie, so I just want to make sure we-
David Gandara: Oh, then it was a glitch. [00:29:30] Okay. In that case, it's a glitch in the original. No, it's-
Speaker 4: And it's kind of a picture, so ... Okay, so [inaudible 00:29:35]-
David Gandara: Okay. Because remember, the 30% is negative. Well, the entire rest of them aren't positive. In other words, aren't 50%. Do you see what I'm saying? You're not accounting ... This is not accounting for the ... Maybe the label. Maybe it was just the label was wrong. Again, if the label goes up here, it'll be right. But the label has to be less than 1%, of course.
Speaker 5: Yeah.
Speaker 4: [00:30:00]
Speaker 6: The
Speaker 5: Yeah.
Speaker 7: ... 30% is less than 1%.
Speaker 5: Yeah.
Speaker 8: Okay. But yeah, Kim, I just looked back at the one we have and it definitely does.
Speaker 7: Yeah, I am too, and the original has it like this, and I just have a...
David Gandara: Yeah. So maybe, again, there was a glitch in the original one and that label was supposed to be greater and equal to 1%. Do you see what I'm saying? Then that would be correct. And then the blue, dark blue would be PD-L1 negative.
Speaker 5: [00:30:30] Yep.
Speaker 8: Okay. We can add it.
Speaker 7: The 30% and PD-L1 negative. Okay. What label would be preferred? Just to put the PD-L1 negative and highlight that as the greater than 30%?
David Gandara: Yes, that would be preferred. Just move the label up there and change it to PD-L1 less than 1%. I think that's the clearest because that's what this is for, to be able to talk about the fact that there is still benefits. So PD-L1 by itself is not an adequate marker, hence the need to marry [00:31:00] it with the profit score.
Speaker 5: Yeah, your point came through loud and clear fortunately, so no need. And also, that was one take. That's pro level stuff. That's where a director's like, "Yes, we got it in one take. We saved millions of dollars." That was super, super smooth. Really well done.
David Gandara: Thank you. There are some recording artists that do everything on one take, and [00:31:30] of course, I looked at all these ahead of time so I knew what I was going to say, but actually, I thought it went pretty well.
Speaker 5: Yeah, I thought it was excellent. Excellent.
David Gandara: And I tried to personalize it a little bit in terms of speaking to the audience about what they would do in different situations.
Speaker 5: Yeah. No, that's really well-
David Gandara: But anyway, I think it'll be good.
Speaker 5: Yeah, agreed. I'm super excited about that. Okay, so we're good. So guys, if you could make the change to that slide and then send the [00:32:00] recording back out so Dr. Gandara can review it. I'm sure he'll be pleased with it, but I think we're good to go.
Speaker 8: Yeah. So the plan that we had discussed prior is we're going to just be sending the transcript and you can make your edits, which I will download that transcript as soon as we wrap up.
David Gandara: I'm not going to make any edits.
Speaker 8: Okay. Well, I'll send it to Kim and Chris if there's anything that you feel that you want to edit and we'll pop in the slide.
David Gandara: [00:32:30] If there was a stumble somewhere where I didn't say something accurately or you want me to rerecord a slide about the assay itself, let me know. But I think this is for educational purposes.
Speaker 8: I think you're fine.
David Gandara: It isn't a scientific presentation.
Speaker 5: Yeah. No, you hit every point exactly.
Speaker 7: I don't think we saw anything but we can just read through it, but I agree. I don't think there's any need to edit. If we could just edit the visual on that one slide, and...
Speaker 5: Yeah.
Speaker 7: Yeah, and Dr. Gandara, I had that ISLB final [00:33:00] webinar, and again, I think it was the final but who knows? Some of these things might have been caught but the one I had did have that, and then also, the animations weren't exactly the same, so thank you for fixing that because I probably may not even get the exact [inaudible 00:33:14] that you spoke on.
David Gandara: The reason I like the animation is it allows you to focus the audience on the point.
Speaker 7: Exactly.
David Gandara: Because you want to show them the whole enchilada here but you want them to focus on this, and I think those animated [00:33:30] boxes really help to do that.
Speaker 7: Yeah, exactly. And I love your point on TPS versus CPS.
Speaker 5: Yeah, Kim, you're going to have to educate me on that because that's a gap in my knowledge.
Speaker 7: That's a gap, that profit. Yeah.
Speaker 5: It's important.
David Gandara: Well, you remember, Chris, when you said we don't want to waste time showing the PD-L1, but it sets up the audience so that they actually are more receptive to profit. Do you see what I'm saying? And that also, they understand the complexity. We do it differently [00:34:00] in non-small cell lung cancer, really for historical reasons.
Speaker 5: Well, and that's the thing. I get that a hundred percent but I need an education on the difference between TPs and CPS. That's my gap.
Speaker 7: And I put some language of that in our dossier, so we'll talk about that and I'll make it-
Speaker 5: Okay.
Speaker 7: It got edited a little bit, but yeah, I think it's important, again, just to help to show where-
Speaker 5: Yeah, obviously. Yeah, it grabbed my attention, so-
Speaker 7: Cool. So thanks for doing that, Dr. Gandara. Yeah, awesome guys. I think we went [00:34:30] very well, so any-
Speaker 5: That was too easy.
David Gandara: Sounds good. And again, I'll give some version of the intro part at your ASCO, Friday morning.
Speaker 5: I'm going to do this. I'm going to take these slides. Once this one is fixed, I'm going to take these slides, make sure that none of the animation or anything changes, convert it to our deck, and then the only question is do you want to send me which TMB slide you want? One or two, however many you want?
David Gandara: Yeah.
Speaker 5: Why don't you just send me which two you want [00:35:00] and... Well, whatever. No, I'll send this back to you. No, no. Yeah, send me the slides, the TMB slides you want and I'll incorporate them in, and you can move them around once the background is set-
David Gandara: Yeah, it'll probably be two slides. I think that's all we'll need.
Speaker 5: Okay. And does it come after PD-L1?
David Gandara: Yes, it comes immediately after PD-L1 so it would be before this slide?
Speaker 5: Just before that slide. Perfect.
David Gandara: Or wait a minute. Actually, let's see. No, no, no. Let's see. Actually, [00:35:30] wait. Turn that down. Actually, maybe we should put it after this one, because this is concentrating just on PD-L1 so we would put it right there.
Speaker 5: Okay.
Speaker 7: Before that, yeah.
Speaker 5: Send me whichever two you want-
David Gandara: It would be right after slide seven, two slides on TMB, and then I go to this one to say, okay, we need a better test, and that shows the transition [00:36:00] to profit.
Speaker 5: And that's where L4 will take over. Yep. Perfect. Excellent. Great job, you guys. Thanks everybody.
David Gandara: Okay. Have a good day.
Speaker 5: Bye, Dr Gandara.
Speaker 7: Thank you. Bye everyone.
Speaker 8: Bye-bye.