Roche Holding Ltd (OTCQX:RHHBY) Virtual Pipeline Event 2018 World Conference on Lung Cancer Conference Call September 25, 2018 8:30 AM ET
Karl Mahler - Head of Investor Relations
Alan Sandler - Interim Head of Global Development Oncology, Pharma Clinical Development Organization
Sushil Patel - Franchise Head for Lung and Rare Cancers, Head of Global Products Strategy
Matthew Weston - Credit Suisse
Sam Fazeli - Bloomberg
Tim Anderson - Wolfe Research
Richard Vosser - JP Morgan
Steve Scala - Cowen and Company
So good afternoon and good morning, ladies and gentlemen. My name is Karl Mahler. I’m the Head of Investor Relations at Roche. And I would like to extend a warm welcome to everybody joining us on this Roche Virtual Pipeline Event to review some of the key data we have presented at the 2018 World Lung Cancer Conference. I trust that you had a chance by now to download the presentation from our Web site. And today, we will not only review the key Roche data presented at this congress but also give you an overview of the Roche lung cancer strategy.
This call is a part of our recently launched event series to highlight key pipeline developments and business updates. Following on from our recent late-stage pipeline events two weeks ago this is the first oncology specific event for the second half of this year. We will host two other calls at ESMO and at ASH, and at the end of October and December respectively. So, we are looking forward to more news flow from Tecentriq, for example, in triple-negative breast cancer over the abraxane backbone from our oncology portfolio.
Looking ahead to ASH, in December, we are also excited about our new T-cell bispecific platform and are looking forward to sharing data in hematology on our two CD20/CD3s from GREx and PREx. And still, aside from oncology, we have data on SMA for risdiplam at the World Muscle Society for type 1, 2 and 3 patients in the beginning of October.
Today, we will discuss new data in our lung cancer franchise with us are Alan Sandler. He is our Interim Head of Global Development Oncology and our Pharma Clinical Development Organization and Alan will lead us through the key data that just presented during the last few days in Toronto. And Sushil Patel, our Franchise Head for Lung and Rare Cancers is the Head of Global Products Strategy. He will give us an overview of the current status of the lung cancer franchise and the outlook. And finally, we have the Q&A session. And as always, in our system here, you can ask questions via the telephone, but you can also use the web. So we have basically two opportunities to ask questions. And I wanted to already thank Gerard Tobin from our team to take the initiative for this Roche pipeline call and helping to prepare the presentation.
So I wanted to begin with slides outlining our current strategy on replacing our existing business. For example, in area such as hematology with current launch of MabThera, in our development pipeline with Venclexta, polatuzumab and our entire CD20/CD3 bispecifics. On the right hand side of the slide, you can see the busy second half of the year. There is still busy half year ahead of us and we are forward looking to share this data with you.
The next slide summarizes where we are regarding the data readout in our lung cancer franchise. We have a lot of last studies readouts in 2018, especially from our Tecentriq program, but now with a better view on how Tecentriq fits in the lung cancer landscape with strong data in important segments like EGFR, in ALK, in liver mass patients in non-small cell lung cancer. Today, we will see clinically very meaningful data in small-cell lung cancer and actually the data has been very much endorsed by the discussion today in Toronto.
At this point, I would now like to hand over to Alan. Alan, please.
Okay, thanks Karl and hello everybody. I would like to do is, if I may, set the table and look at small-cell lung cancer, which is not a new area of disease but now as you will see, a new area of some positive data. Small-cell lung cancer accounts for about 15% of all lung cancers, and the vast majority of those or two thirds or 10% represents extensive stage disease. And that’s what the study for our discussion of IMpower133 will be. Just some key highlights entirely associated with smoking, it’s very aggressive tumors that although, initially responsive to chemotherapy, becomes resistant to it quickly and survival is generally measured approximately nine or 10 months.
Moving to the next slide. So small-cell lung cancer treatment like treatment for other diseases is determined by staging of disease. However, the staging system used for small cell is a bit different than that used for the others. We don’t use the TNM classification, if you will. It’s based on an old mechanism that was designed in the 1940s and '50s, based on radiation therapy. So any patient with disease is outside the confines of one, it's a lateral aspect of the thorax, it’s considered extensive stage disease in which case the treatment is platinum-based chemotherapy. And again this represents extensive stage disease.
As I’ve mentioned, there has been no progress in this in the frontline therapy for the 30 years, so actually probably approaches 40 years. There has been some evidence that immunotherapy may have clinical activity in refractory or metastatic disease, both from preclinical data and also as you’ve seen nivolumab, has been approved in the third line treatment of metastatic or extensive -- well, a recurrent small cell lung cancer, now that's a single agent.
So if we move to the next slide, I’d like to discuss with you this chemo for IMpower133. This was a global phase 1/3 randomized placebo controlled trial that evaluated atezolizumab in combination with carboplatin and etoposide versus carboplatin and etoposide in the frontline extensive stage setting. As you see here, patients were generally good performance status patients with no prior therapy, and patients with treated but asymptomatic brain metastases were eligible. Stratification factors are also shown based on facts, performance, status and the presence or absence of brain metastases.
I would also call out that the co-primary endpoints are overall survival and also investigator-assessed PFS. Lastly, we’ll go back to the scheme on the other side. I just want to point out that the treatment consisted of induction therapy with atezolizumab with chemotherapy for four cycles powered by maintenance of atezolizumab and treated until progressive disease or loss of clinical benefit as we’ve done in all of our frontline lung cancer studies, as also I will point out that it's placebo controlled, placebo also given during chemotherapy adds maintenance.
Moving to next slide and here is our Kaplan-Meier curve for overall survival. And you see that the curve separate after a few months and then remains separated throughout the median survival and increases from 10.3 to 12.3 months. Also importantly, the hazard ratio is 0.7, representing 30% reduction in the risk of death to this disease. Median follow-up is approximately 14 months. Next slide, now we’ll go to the investigator-assessed progression free survival. This is one of two primary endpoints as I mentioned along with overall survival. Again, there’s a separation in progression free survival with median improvement in the median progression free survival from 4.3 to 5.2 month, a hazard ratio of 0.77.
The next slide looks at overall survival in several key subgroups. And I think that what I would to point out is that for virtually all of these subgroups, there is improvement in survival seen regardless of the various subgroups that is regardless of male or female and also performing status, liver metastases and we’ll also talk about the tumor mutational burden. I will call out brain metastases does seem to cross the line of unity. However, there are only 35 patients with the presence of brain metastases at the time of study. You will also see that there is no difference in outcome regardless of the cut off, pre-specified cut offs used for tumor mutational blood based tumor mutational burden.
We'll go to next slide. These next slides looks at the most frequently observed adverse events. I will first call out the fact that there was generally the majority of the side effects were related to that of chemotherapy, which you see in the top half of the table and with minimal differences between the two arms. So that atezolizumab did not enhance the toxicity of chemotherapy. Below that, you'll look at immune-related AEs. And although, there are differences of course when giving the atezolizumab, the toxicity profile is consistent with that seen in our other randomized studies with chemotherapy. And again, it appears to be tolerated rather well.
We'll go to the next and summary slide. So IMpower133 is the first study in over 20 years in small-cell overall. But the first study is nearly 40 years in the front line setting to show a clinically meaningful improvement in overall survival over the current standard-of-care in the setting of frontline extensive-stage small cell lung cancer. The addition of atezolizumab to carboplatin and etoposide provided that significant improvement in OS and PFS when compared to chemotherapy alone with the median shown here. And again, I will call out the hazard ratio of 0.7 for overall survival.
Safety profile of atezolizumab with chemotherapy was expected with no new finding, and the rates -- hematologic side effects are similar between the treatment groups. These data suggest that atezolizumab plus carboplatin and etoposide chemotherapy is a new standard of care for the first-line treatment of extensive-stage small-cell lung cancer. This was also echoed by Natasha Leighl, who was our discussant at the Presidential Symposium. The full data was published today in the New England Journal of Medicine.
So next we'll go to discuss the IMpower132. I have the trial design shown here for you; this is the study in non-small cell -- non-squamous non-small cell lung cancer; and it looks at the combination of pemetrexed with either carboplatin or cisplatin alone, or in combination with atezolizumab given comparing with chemotherapy and then followed by maintenance pemetrexed with or without atezolizumab; and again, the maintenance for atezolizumab was continued beyond progression if the patients driving benefit as defined in the protocol. Co-primary end-points for the data is investigator-assessed PFS and overall survival.
So the next slide, this shows the final investigator-assessed progression free survival. We also have overall response rate and duration of response on this slide as well. You see that survival curve for PFS separates early and remains separated throughout translating to an improvement in the median progression free survival from 5.2 to 7.6 months with the hazard ratio of 0.6. And I will point out that the minimum follow-up is nearly 12 months and the median follow-up is approximately 15 months. The value for the hazard ratio is highly statistically significant. Response rates were improved with the addition of atezolizumab from 32% to 47% and the median duration of response also improved from 7.2 to 10.1 months.
Next slide, we’ll show you PFS in key patient subgroups. In general, you’ll see that all key subgroups derived benefit with the addition of atezolizumab to the chemotherapy. I will call out liver metastases, you're familiar with the data that we have from IMpower150 where we showed its advantage with the addition of bevacizumab to paclitaxel, carboplatin and atezolizumab in patients with liver metastases. I am going to show you that data in just a moment. But what I do want to point out is that although patients did derive some benefit with the presence of liver metastases, it was -- those patients without liver metastases in this study did better again potentially arguing for the advantage and the addition of bevacizumab in this setting.
The next slide that you see here is in fact from IMpower150 showing that the addition of Avastin to Tecentriq and chemotherapy for long survival patients with liver metastases, we have hazard ratios here and also Kaplan-Meier curve I think are rather quite convincing with hazard ratio of 0.54 with when you look at Arm B versus C that the Avastin containing arm as compared to Arm A versus Arm C.
This slide now is the interim overall survival analysis. You have some relatively early separation between the curves that remain separated throughout. Minimum follow-up again is only about a year with median of about 15 months. And the improvement in survival, however, is 4.5 months, 13.6 months versus 18.1 months. And I will point out the excellent survival data that was actually seen in control arm based on -- compared to historical control. The hazard ratio is 0.81 that at this time is not statistically significant but follow-up will be forthcoming next year.
Safety summary is here, it is consistent with all of our prior frontline studies in combination with chemotherapy, showing modest differences with the addition of atezolizumab, in particular that associated with the adverse events of special interest those that were felt substantially related to atezolizumab. Key points there, however, are the very small numbers of patients all less than 3% or less seen in with grades 3 or 4 AE. I will also point that the PRO data does support the positive benefit risk profile demonstrated by the clinical data.
We’ll go to the next slide, with summary slide. And I'd state that IMpower132 did meet its co-primary end point of investigator-assessed PFS in the ITT population and the addition of atezolizumab chemotherapy with [platin] and pemetrexed-improved the medium PFS in the ITT population by nearly 2.5 months in 5.2 to 7.6 months, and a hazard ratio of 0.6, maintained across key clinical subgroups outside of liver metastases. Atezolizumab plus pemetrexed and carboplatin or cisplatin has a manageable safety profile and is consistent as mentioned with the known safety risk profile of the individual therapy and that’s seen with our other chemotherapy combination. Although, overall survival data did reveal a numerical improvement of 4.5 month, at this interim analysis, they did not reach statistical significance. The final analysis is anticipated sometime in the first half of next year.
So, I will close now with few slides on the efficacy and safety of the entrectinib in our ROS1 fusion-positive non-small cell lung cancer population. So, the slide that you see here is an MoA slide and entrectinib is an oral, potent and selective ROS1, NTRK and ALK tyrosine kinase inhibitor that also, one key element is that it is active in the CNS. Appears more important that the crizotinib against ROS in preclinical study, it's a potential pan-TRK inhibitor in clinical development and has demonstrated clinical activity in multiple tumor histologies that will not go today, and was designed to cross the blood-brain barrier and remain within the CNS. And has demonstrated clinical activity in primary brain tumors and secondary CNS metastases the data of which we'll go over shortly.
As you’re familiar, there are limitations for current treatment options for ROS1-positive non-small cell lung cancer. This represents about 2% of patients with non-small cell lung cancer that predominantly female, predominantly never-smokers and young compared to the traditional lung cancer patient with a median of approximately 50 years age. So currently, crizotinib is the only approved agent in this setting. So crizotinib was recently added as an option by NCCN guideline. I'll also just call out I think an important point is that that the CNS is often the first and only side of progression in nearly one half of patients with this disease, so a critical component in this setting.
So the next slide shows how the data was derived, where the patient came from, from the data I’m about to show you. So it's an integrated analysis of three studies, looking at ROS1-positive non-small cell lung cancer. These three studies include the STARTRK-2, well STARTRK-1, STARTRK-2, and ALKA-372-001; 37 patients from STARTRK-2, seven from STARTRK-1 and nine from the ALKA-372-001; so representing 53 ROS1-positive, ROS1-inhibitor-naïve, non-small cell lung cancer patients.
So let’s take a look at the data. And we have the objective response rate and this is by a blinded independent assessment. And you have a couple of ways looking at the data. We have a waterfall chart with color coded for whether the presence or absence of CNS disease. It's rather apparent that activity exists in both those settings. And then we have a table on the right showing response rate of around 75% to 80% in those patients whether they have CNS disease present at baseline or not.
And since I butchered those numbers a little bit, I am going to go over one more time for you. So for the overall patient population of 53 patients responses are seen in 41 patients, representing a response rate of 77.4%. Those 23 patients with CNS disease at baseline response rate is 17 patients for about 74% response rate. And then those patients with no-CNS disease at baseline 30 patients, 24 responders with response rate of 80%. So again, activity seen out independent of CNS and there were actually a few complete responses noted as well.
Let's go to the next slide, so this is another key slide. I believe looking at duration of response again using that blinded review assessment. And if you look at the table, the median DOR represents just over two years and 24.6 months. And for that total 53 patients, it’s 12.6 months for the patients with CNS disease at baseline and 24.6 months with no CNS disease at baseline. The follow-up is about 16 months. Progression free survival also using the blinded assessment, the median PFS is 19 months with a similar median follow-up. And so again, this represents a quite durable effect from entrectinib in this setting.
Let's specifically look at the impact of intracranial responses looking at ORR and DOR in patients who has CNS disease at baseline, and here’s the table that evaluates this for you. So there were 11 responders with the 55% response rate just looking at intracranial response rates shown here, including four patients 20%, that actually had a complete response; 12.9 months was the intracranial median DOR that is also shown there as well. So very key piece of information regarding CNS activity.
We’ll go the next slide, and safety. So we now have 355 patients have been treated with entrectinib across three clinical studies. The majority of these AEs are grade 1 to 2 and also reversible. And for treatment related AEs, there was only 3.9% of these treatment related AEs that actually led to discontinuation from the study treatment. Meaning that nearly just over 96% of patients maintained and stayed on treatment and tolerated -- were tolerating treatment well. About one fourth of the patients had either dose reduction or dose interruption. A list of the toxicities are shown here with AEs, again, very, very few grade 3 or higher with the majority falling into the bucket of dysgeusia, fatigue, dizziness and constipation.
So, to conclude, for entrectinib, in ROS1-positive non-small cell lung cancer and patients treated with entrectinib, clinically meaningful deep and durable systemic responses were observed in patients with and without CNS diseases, response rate was 77% median DOR of over two years and a median PFS of over two years with 26.3 months in patients without CNS metastasis and 13.6 months with CNS metastasis. We also described clinically meaningful and durable intracranial activity in these patients with baseline CNS disease with intracranial response rate of 55%, and an intracranial median duration of response of 12.9 months, that refers to the disease in the brain -- within the brain itself. And lastly, entrectinib was quite tolerable with management safety profile, most of the AEs were managed with either dose interruption or reduction, although -- and with a very, very low discontinuation rate of under 4%.
So I'll close again with some key takeaways. IMpower133 is the first study in over 30 years, nearly 40, to show a clinically meaningful improvement in overall survival over the current standard-of-care chemotherapy in frontline extensive stage small-cell lung cancer. The addition of atezolizumab with carboplatin and etoposide provided significant improvement in OS and PFS when compared with chemotherapy alone in this setting with improvement in the median overall survival of 10.3 to 12.3 month and a hazard ratio of 0.7 with a P value of 0.0069.
For IMpower132, the addition of atezolizumab to carboplatin, or cisplatin and pemetrexed, improved the medium PFS in ITT population by investigator-assessment from 5.2 to 7.6 months and a hazard ratio of 0.6, also across several key clinical subgroups. Our interim OS data shows numerical improvement as the next OS analysis is anticipated in the first half of 2019. And lastly, entrectinib represents an active agent in ROS1 fusion-positive locally advanced or metastatic non-small lung cancer, both systematically and in the CNS and overall response rate is 77.4% and median DOR of approximately two years, median PFS of 26.3 months without the CNS metastases and 13.6 month for those with CNS metastases.
So, I’m going to close now and turnover the presentation to my partner, Sushil Patel.
Thank you, Alan. Good day, everyone. So I wanted to just cover where we are with lung cancer today. And so, we have a broad coverage across a number of the key segments in the lung space. And I think it's important to note that these are important differentiated growth opportunities for us. I wanted to just start with the driver mutations segment. This is an area where we have very strong position. Alecensa has already established itself as a market leader as a standard of care, particularly in the U.S already in a number of other markets.
As Alan shared with you, we believe the entrectinib is also going to rapidly establish itself in the standard of care and set an important unmet need in ROS1, particularly in the CNS patients. We have another opportunity in the EGFR. I expect that you heard before with our IMpower150 data, particularly for those patients who have failed TKI, and this is an important area of unmet need and we've seen the 150 data has significant activity there. So that addresses I think an important segment of the market that has an unmet need today.
We’re very excited about our small-cell lung cancer data. We’re going to be the first, not only report out the data as you saw today or as you heard today, but we're also looking to be the first to file this data. And in fact, already filed with the FDA and are working very rapidly to file globally. We believe that this regimen is going to establish itself as standard of care rapidly given the unmet need. If you turn to non-small cell lung cancer, we had now three positive Phase 3 studies, two of those have overall survival, including the abraxane combination and the Avastin combinations with IMpower150. And here, we've seen strong efficacy, as Alan has mentioned, in the liver metastases, which represents about 20% of patients.
We’re also looking extensively at early stage settings and we’ve got pivotal studies ongoing in neoadjuvant and adjuvant disease. And I’d also say that we’ve also rapidly established ourselves in the second line space with Tecentriq and had approvals in over 80 countries, and are seeing rapid uptake in many markets. As high as the bar has been established in the setting of non-small cell lung cancer with a number of chemotherapy combination readouts, we’ve seen the significant unmet need, 40% of patients still don’t respond. And we also know the CIT experienced segment is a big opportunity, and we had a number of combinations we’re looking at to explore and bring options to patients in that space.
So, we expect the lung cancer market to continue to grow rapidly and we believe that we’re going to play an important role in that, particularly in terms of new areas where there has really been nothing, as Alan mentioned, such as small-cell lung cancer for over 30 years.
Next slide please. The two areas I just wanted to focus on a little bit is we now believe we have two best-in-disease molecules that actually address an important unmet need in terms of CNS activity. With Alecensa, we have the data from the ALK study. We saw an unprecedented 34.8 months benefit, almost a tripling versus the control arm with crizotinib, the hazard ratio of 0.43. We believe that Alecensa will remain a standard-of-care in the frontline setting despite the brigatinib data just recently. We have strong activity in CNS with 27.7 months versus the control. In fact the hazard ratio is even better than the overall with 0.35. And as I mentioned, Alecensa has already established itself as a standard-of-care, not only based on the efficacy but is very tolerable safety profile.
We’re also excited about what entrectinib is going to bring to ROS1 fusion-positive patients. And we shared the data with you, the 26.4 months in the non-CNS and even strong activity in CNS. The response rates were almost 80% and durable over two years. Again, we believe that this drug is going to provide a strong benefit to patients with ROS1 disease.
Next slide please. So Karl already shared this slide with you and yes, we had a number of positive readouts across many of the segments in lung cancer. The only thing I want to maybe point out in addition here is we have four drugs today. And as you saw the entrectinib data, we’re working very rapidly given the fact that we had breakthrough designation in the U.S., we had prime designation in the EU and sakigake in Japan to try and bring this molecule to patients globally as fast as possible.
Next slide please. So finally, I just want to end up with why we believe we’re so strongly positioned in the lung cancer space. Yes, we have the molecules and that’s important, but we know that lung cancer is becoming increasingly fragmented and the complicated treatment algorithm. And we believe we’re well positioned with a range of services beyond that molecule to really help customers navigate through that complexity and to maximize outcomes to patients. In terms of diagnostic capabilities, we have Roche Molecular but also partnership with Foundation Medicine, and we’re developing novel blood-based biomarkers for patients, which is an unmet need.
In terms of access, we have a number of innovative access solutions. We’re also leveraging our real world data capabilities with Flatiron to support reimbursement and filings for a number of our drugs. And then finally, we know how much data there’s in the marketplace and one part of helping to navigate that complexity is really interrogating that data and being able to use new tools and techniques to help customers figure out how best to treat their patients. And we have solutions, such as Navify which is a decision analysis tool, which we think will be important in the marketplace moving forward.
So that was my final slide. I think, Joe, we have one more slide?
Yes, I wanted to open the Q&A actually. And, operator, if you could open the line for any questions, please?
The first question from the phone is from Matthew Weston, Credit Suisse. Please go ahead.
Three questions, if I can. The first on IMpower133, the two months improvement in overall survival, no one questioning at all that’s it's not a new standard of care. But I'd be interested your experience, particularly with Avastin. How much success have you had in the past ex-U.S in gaining reimbursement for an expensive drug, which has a modest improvement, but is clearly itself novel? On IMpower132, I guess one thing that we continue to see or keep seeing is the PD1s and the PDL1s have similar benefits on PFS. But the PD1s seem to have at least faster, if not improved, benefits on OS. And I just wondered whether now having seen that signal a number of times, there's a reason you think that could attribute to that, whether it's something to do mechanistically or whether it's to do with trial design? And then finally, Sushil, you mentioned neoadjuvant and adjuvant lung on your slide in terms of new opportunities. Looking at the speed with which the different competitors are referring to when they will have data in that setting, there seems to be a huge discrepancy in terms of the number of years those trials will take. I would be very interested in Roche's view as to when you will have first data in the neoadjuvant and adjuvant lung setting?
IMpower133 OS and ex-U.S, I mean, to understand that correctly the investigator, the discussion today, we're actually quite impressed by the data. But Alan maybe you can comment on this please?
Well, I think actually the question Karl related more to the payers in Europe, and I'll leave this to Sushil.
Yes, I mean, I think payers look at a number of different things. One is the unmet need and there really hasn't been anything else for a long time. We have a clear benefit there as you mentioned. The most important end point is overall survival and we've demonstrated that clearly. I think the other point I would just make is that we have successfully got reimbursement for a number of drugs. The cost of this regimen is on top of a very established and relatively cheap chemotherapy regimen. So you’re really only covering the cost of the Tecentriq, which they already approved in Europe right now, the monotheraphy. So I don’t really see this as adding a lot of additional cost. It's not like we're combining a number of expensive regimens. So overall, we feel confident that we’re going to get a strong imbursement position for this regimen.
And 132, PD1 versus PDL1, Alan?
Yes, thanks Karl. So the question there, you highlighted a couple of points that I don’t know that we can actually answer them to-date. But I certainly can attempt to address them. I think in the head-to-head -- there’re no obvious head-to-head comparisons. But in the purest setting that we’ve had looking at monotherapy in the second line plus setting, we see that results are quite similar across the board regardless of class of agents and individual agents. And combining with chemotherapy, we have seen some -- there have been some differences with respect to overall survival. Although, you do point out that the PFS looks quite similar. So what could impact that going forward? And I won’t comment specifically on the mechanism of action, because I think that’s a bit harder to discuss in a specific way. But the trial design I think is worth commenting upon.
Particularly if we look at 132 versus 189 there’s a couple of things that do stand out that I’d like to point out to you. So look at the data for 189 was earlier where I think it was a median of about 11 months versus ours more of around 13 to 14 months or so in terms of looking at the time of data analysis. The earlier you look in a setting like that the less impact crossover therapy should have. And I think this is of particular importance in 189 in patients the design included crossover for pembro in the control arm. And you see that the potential impact with the numbers of patients that crossover less than 40% and also I would argue that the actual impact of that crossover was probably smaller than that seen in a more mature study. So it’ll be interesting to see with follow-up what 189 looks like.
The other aspect that I’ll point out is, and it’s been pointed out before is how the control arm has done in that setting for 132 and 189 with control arm in our study did a bit better. I will note that the dose intensity for our study or actually the dose, specifically for carboplatin is higher. We had AUC of 6 compared to AUC of 5 in the Merck study. And whether that or not may have contributed is unknown but again, just something I’ll point out with the trail design. So I think the time of statistical analysis may have impacted, dosing may have impacted and those are just the two things I’d like to comment upon.
Matthew has a third question on adjuvant, neoadjuvant trial results timing?
Right, I think that’s going to be for Sushil.
Sure, I can have a crack at that. So, in terms of the adjuvant, neoadjuvant, from an adjuvant perspective, actually, we believe we caught up a quite substantially with the competition. A number of the competitor trials are using cooperative studies, and we’re running the company sponsored study. In terms of readouts for those studies, it’s really hard to estimate when these adjuvant studies are going to readout. They will be using slightly different endpoints, PFS, DFS, et cetera. So I think it’s a bit hard for me to speculate. But again, we’re working actively to figure out ways to look at the different interim analyses, et cetera to bring in that timeline.
But as I mentioned I believe, we've actually caught up substantially in the adjuvant setting. And in neoadjuvant, we’re approaching that as well. We started a study and again bit hard to speculate and how quickly we can see the readout. But one aspect of neoadjuvant is you look at the NPR and have an early readout of the data. So I don’t know Alan if you have any other comments on those studies.
No, I would just echo there. I think it’s uncertain. Obviously, we have no more data on when our competitors are going to readout. And so we can really comment on ours, and Sushil has already answered that.
Thanks a lot. Matthew did this address all your questions?
It did indeed. Many, many thanks indeed.
So there’s a question over the web from Luisa Hector on IMpower 133. She asked why no maintenance space in the chemo? The PFS curve seem to separate after chemo stops. Why shouldn’t you consider treating Tecentriq on progression, any evidence to continue the treatment into second line of treatment? So that is her question.
Yes. So I will comment on the design of IMpower133. So at the time of design of the study, the concept was that the addition of chemotherapy, or atezolizumab to chemotherapy, would enhance the effect of both potentially and certainly the chemotherapy might enhance the effect of atezolizumab through its effect by cell killing and release of additional antigen. And then of course there is the maintenance effect as well. So the study was designed to look at atezolizumab in the context of both induction and maintenance, and this study has answered that question definitely. The question as to whether or not the maintenance alone would provide the same answer can only be answered with another randomized study. And so I really can’t speculate as to whether or not we would have achieved the same results in this particular setting. But I am sure that potentially academicians or other individuals may have that and finishing, and maybe something that we look into in the future.
Okay, thanks a lot. Next question please over the phone please.
Next question is from Sam Fazeli from Bloomberg. Please go ahead.
Sam Fazeli from Bloomberg Intelligence. I have so many questions that I'll just have to focus them…
We just have 15 to 20 more minutes.
Yes, I understand that. Let me just focus on 132 quickly here. First of all, the choice of AUC 6 versus AUC 5, what drove that? And then there is some incongruence in the PDL1 data in terms of the responses versus -- or survival versus different cut-offs, if you want to just comment on that? And third, there are some elements I don’t understand. In the subgroups, four cycles seem to be better than six cycles. And you're almost also suggesting that AUC 5 is better than AUC 6. Is there any way that I could interpret this as less chemo is better in terms of maybe effects on key effector cells or whatever? And why do you think the independent assessment basically didn’t show statistical significance versus the investigator PFS? Although, I understand that was the primary outcome and that was pre-specified, just some thoughts on that.
I believe I'll take all of those. This is Alan. So the AUC 6 versus 5, both of those regimens are accepted regimens with 5 versus 6. When we designed the study, we went with AUC 6 because we wanted to ensure that we were giving the control arm adequate therapy, and that we wouldn’t be critiqued when the study was completed that we use an insufficient regimen. So that was the reason for 6 versus 5. Regarding the incongruence for PDL1, I think you’re referring to the PFS data, the hazard ratios showing significant benefit in those patients with the very highest expression of PDL1 but also showing very good effect in the PDL1 negative. And benefit, although and that’s perhaps not as great in -- not as high I should say as highest and lowest so that U-shaped curve effect, if you will. That is something that we’re looking into. We admittedly do not have an answer for that at this particular time.
I will say what it does -- what it is encouraging for us is the effect in the highest expressers and also see very, very good treatment effect in those patients that by at least SP142 very good effect in those patients who are negative, so again showing that all comer benefit. Your third question related to chemotherapy effect and whether unless chemotherapy is better than more chemotherapy in this combination. And that’s certainly a very interesting question and may well be related to what you had talked about that perhaps that less effect on the two lymphocytes, allowing a better effect later on. So we're looking into that we think that’s a fascinating aspect and that is something that was actually commented upon during the discussion session as well. And I think that we’re looking at four cycles in that particular study and I think we’ll take that into account and further studies as well, as well as the preclinical data as well.
Maybe I’ll just add on in terms of the PDL1 status, we only had about 60% of samples. So again, there could be some rationale as to why we saw some unusual things there. Yes. So I just wanted to mention that.
Yes, that’s a very good point and I should have mentioned it myself, and that’s why that both of us are here. So I think the aspect for that is it was not mandated. We do not mandate tissue requirement in 132 as we did in our other studies. And then Sushil correctly points out, only 60% of patients provided tissue. So in data set that is not complete, you can have the variations that occur by a random occurrence as opposed to something that’s biologic. And so thanks Sushil for pointed that out.
Okay, I think there was one question on the endpoint?
Yes, the investigator versus independent PFS data?
So the primary endpoint is investigator assessed. And the rationale for that is that we believe that that actually is closest to real life data analysis as decisions are made in real time by the investigators themselves much as they would in the real world outside of clinical trials.
Okay, thanks. Thank you guys.
Yes, thanks a lot. Next question please.
Next question is from Tim Anderson, Wolfe Research. Please go ahead.
A couple of questions, so on 133, the survival data by TMB status, there really isn’t much of a difference where they’re routinely higher and low. And I am wondering if you’re surprised by that? And then in 132, I am wondering if you did look at TMB as well. So OS, interim OS by TMB, if you saw any difference there? And then last question on entrectinib, I guess at ESMO, you’ll be having data in the NTRK segment. I am guessing that data must be in-house at this point. I am wondering if you can preview your enthusiasm for that data relative to ROS1 segment.
So I will start with 133, and the question with TMB. And you correctly point out, there was no difference in survival between the two cut-off based on that blood based analysis. I will say that given the evidence before that there’s high levels of TMB in small cell based on the fact that a tobacco related malignancy. We did think that there might have been an effect to some degree. I will say that not necessary surprised by the all comer effect, but we would have thought that there might have been some evidence of those with the higher TMB might have performed a bit better. So that’s the answer to that one. For 132, yes, we will be looking at TMB in 132 as well that’s an exploratory analysis that we don’t have the data for that at this particular time.
And in terms of the NTRK, so we know that NTRK also has very a small prevalence in lung cancer. We did not share that at this meeting. However, we’re going to show our collective NTRK data at ESMO, and we’re more excited about that as well. And that is across a number of different tumor types.
If I could just go back to 133, so do you have any guesses as to why there was no difference? Could it be the fact that it was a blood based test versus tissue based?
No, we don’t have the data to compare with the tissue base. We don’t have the samples in this particular study, a limited number of samples were available. So, I won't be able to comment specifically as to how it did in the tissue, and we don’t necessary believe that it was a reflection of the blood based biomarker itself. Again maybe just -- yes, I mean there is an outcome that happened and that’s something that we see throughout the particular studies in terms of sometimes we see where the biomarkers do make a particular difference and sometimes we seen it in various studies that the biomarkers don’t specifically have as much of an impact as we might have expected.
Did that address your question?
It does, yes.
Okay, perfect very good. So next question please.
Next question is from Richard Vosser, JP Morgan. Please go ahead.
Hi, Richard Vosser, JP Morgan, thanks for taking my question. Two on entrectinib and two 133, please, the discussion on 133 suggested that 80% of the patients had tumor samples. So do you think -- when do you think we'll see the PDL1 divisions of the data, and do you think there will be any differences there? Second question, because on -- the survival curve on 133, it's a little bit like PDL1 negative survival curve from non-small cell lung cancer, and we see there is a couple of trials that shows -- going to benefit in the PDL1 negatives than other agents. So do you think this is an effect that could see differences in small-cell lung cancer? Or do you think this is difference in the diagnostics or something else? And then questions on entrectinib. Do you plan to do that head-to-head study I think again the discussions in real world around was suggesting a head-to-head study versus crizotinib would solidify the benefit. I know you don’t need that for approval, but would you do it? And then finally, what do you think the emerging data from larotrectinib, which is also looking to treat brain met in ROS1 patients compared to entrectinib?
Well, I take the two questions related to 133, and they both center on PDL1 level of discussion. A little bit of background for PDL1 expression in our ends and others. The level of expression in PDL1 is much lower in small cell than that seen in non-small cell lung cancer. And so, we’ve chosen the study because of that fact, as well as the fact that it’s difficult to get tissue in patients with small cell lung cancer since the majority of patients are treated with skinny needle FNAs and because of the pace of the disease are not up to -- both patients and physicians are not up to repeat biopsy. So the percent of tumor samples that we have in this study are rather low and the decision has not yet been made to evaluate the impact of PDL1 expression for those reasons.
And again, what you might see with a sample size that is so small that whatever data you obtain from that may potentially be misleading. So we’ll go the next aspect, which was the question of overall survival curve and does it in fact look like a diagnostically negative population. And again that raises an interesting point and perhaps there is a bit of that given the fact that the level of PDL1 expression in small cell is lower than that seen in non-small cell lung cancer. So it is possible that that may potentially have contributed to some degree. Although, I will say that a hazard ratio of 0.7 in that all tumor population is quite significant, both -- again statistically and clinically. The last one question, we’ll go to Sushil.
Yes, in terms of the entrectinib, I think the question is on our head-to-head study. So we actually believe the data stands on its own. Crizotinib does not have CNS activity, and we actually had a differentiated profile. Whether we need to do a head-to-head is an ongoing question and the team is certainly discussing. We know these are small prevalent populations. It could take a long time. But one of the nice things that we have at our disposal with Roche is our real-world data with Flatiron. So we may be actually using synthetic controls rather than actually having to do a study. Doing a study would actually take a long time and so that maybe another option for us.
And then I think you asked a question about larotrectinib. We haven’t seen a lot but it’s a smaller population, a higher level of grade 3 and more toxicity. So we think that might be a second line option and we believe entrectinib is going to be a good first line option for patients.
There’s another question over the web from Jessica from [indiscernible]. She asked do you believe if you are able to file on surrogate endpoints in the neoadjuvant lung cancer study.
So that is a question that something we’re working on. There’re number of surrogate endpoints that are being looked at in the adjuvant and neoadjuvant settings by us and others. We’ve held discussions with -- at various FDA meetings, one with Academic. And whether or not that this is something that can be utilized whether it'd be biomarker or other early markers of disease, it's something that we are clearly in active discussions with, we think it’s very important to be able to identify agents that are active in a potential curative setting and get these agents to patients in an expedient -- as expedient of manner as possible to enable cures in this particular setting.
The next question is from Steve Scala from Cowen and Co. Please go ahead.
If neither PDL1 or TMB can select IO sensitive small cell lung cancer patients. How can Roche follow-up on the trial to raise the bar further and/or differentiate Tecentriq in this setting? Secondly, on entrectinib, the CNS ORR is better than prior crizotinib data sets but the PFS isn’t significantly better. If PFS isn't better then what is the rationale for the accelerated FDA approval? And then lastly, just to be clear, IMpower132 won't be filed until we see final OS. Is that a true statement? Thank you.
So I will start with on the small talk question then we'll go to Sush. So with the PDL1 and TMB, again the TMB was clearly tested and did not show ability to distinguish outcome in small cell lung cancer. And PDL1, as I mentioned, the readout really have enough samples to evaluate, although, again the percent is low. Roche because of our firm belief in personalized healthcare will be making continued efforts to identify new biomarkers, if at all possible, to identify those patients that benefit best from therapeutic agents.
I'll also add that even in the other studies that we have had that we have done to-date and with other check point inhibitors, what we've been able to do with biomarkers is establish those groups of patients that appear to do better. That is we can enrich a population, but what hasn’t been able to do -- be completed and what remains even more challenging is can you identify that patient population that derive no benefit from treatment. And that one is harder and arguably maybe even more important, because it's obviously very difficult to tell a patient that I am not going to treat you, because you will not benefit from this particular therapy. So we continue to look for that, moving forward. But thank you for the question.
And then the question on the CNS, the crizotinib PFS versus the entrectinib. The crizotinib PFS number, the 90 month that was not broken out by the CNS population. So we really can’t compare but when we actually do look break it out in our study, we actually think we have an advantage there. So that’s what I would say on that.
And filings for the two IMpower132, you said you’re in discussion with FDA. But is there anything else you would like to add to that, Sush or Alan?
No, I think at this particular point that we are in discussion with the FDA currently and discussing the data that we have available.
Does that answer your questions?
Yes, thank you.
Yes, thanks a lot. Operator, we are coming to an end of the presentation. We have reserved one hour. Are there any further questions in the line?
There are no more questions from the phone.
Okay, thanks a lot for your interest in Roche. I wanted to thank all our participants today. I wanted to thank you for your interest in Roche, and I wish everybody a nice day and a nice evening. Have a good time bye, bye. All the best.