A Precision Medicine Roadmap for Rx and CDx Co-development and Regulatory Approval in Japan

WEBINAR TRANSCRIPT

MARK: Hello, and welcome to today’s webinar. Thank you so much for joining us! Today, we’ll be discussing a Precision medicine roadmap for RX and CDx co-development and regulatory approval in Japan, with a distinguished guest speaker from Loxo@Lilly. My name is Mark, and I’m the chief science officer and founder of Genomenon. Before we get started with today’s presentation, let me cover a quick couple of housekeeping issues. I will be back after Susanne’s presentation to cover some questions and answers, so feel free to use the live Q&A chat feature and go to webinar. We’ll get to those at the end of Susanne’s presentation. Also note that this webinar is being recorded, and will be emailed to you once we’ve wrapped up. Now, I’d like to introduce Dr. Susanne Rhodes. Susanne, if you could turn on your camera. Hi, Susanne!

SUSANNE: Hello!

MARK: Susanne is currently the associate vice president of diagnostic development and
program management at Loxo@Lilly. She received her PhD in Pharmacology from Vanderbilt University School of Medicine, has 11 years of companion diagnostic development experience, and worked on a number of novel companion diagnostic programs. Susanne, we’re all the more grateful for you to be here today, because we know that you’re traveling. I just came off a very busy travel January, where I spent a week in a broom closet for 800 dollars a night, so I empathize, wherever you are, whatever you’re doing in your travel.

SUSANNE: Thanks! I’m in Sunny Orlando, so it’s not too bad.

MARK: It hailed in San Francisco when we were there! I didn’t think that was possible. Susanne — I and my colleagues at Genomenon had the distinguished pleasure of working with her on this effort. She will be covering a broad range of the challenges and successes she faced in seeking approval for selpercatinib in Japan. I will turn the floor over to you, Susanne, and you’ll see me again at the Q&A.

SUSANNE: All right, thanks, Mark, for that kind introduction! I just want to start out by saying that this material was shared last September at the world clinical biomarkers and companion diagnostics conference. In Japan, selpercatinib is approved for the treatment of RET-positive non-small cell lung cancer (NSCLC) and RET-positive thyroid cancers (TC), with the Oncomine Dx Target Test (ODxTT) as the companion diagnostic. This presentation will discuss these recent co-approvals and unique considerations for drug and companion diagnostic development in Japan. Some of the questions that we’ll talk about for the development timeline are choosing between co-approval or co-launch as a goal; what are some of the essential elements of the bridging study design; and what steps can be taken to facilitate the regulatory review?
Before I get started, I’d like to give a little bit of background on selpercatinib. It’s a small molecule kinase inhibitor, and it’s designed to address two primary mechanisms of RET activation, RET fusions in non-small cell lung cancer, as well as capillary and other thyroid Cancers and RET mutations in medullary thyroid cancer. Next, the Oncomine Dx Target Test, which is manufactured by Thermo Fisher Scientific. It is a next generation sequencing (NGS) based companion diagnostic, which was clinically validated and received initial approval by PMDA in 2019 for the genes listed here, these four genes for patients with non-small cell lung cancer. In addition to the biomarkers clinically validated as a companion diagnostic, 46 genes have been analytically validated. From a formalin-fixed paraffin-embedded tumor biopsy, collected as either a surgical resection, core needle biopsy, or fine needle aspirate, the ODxTT can detect genomic alterations of both DNA and RNA at quantities as small as 10 nanograms. Loxo@Lilly selected the ODxTT as the selpercatinib CDx for Japan due to the existing Ministry of Health, Labor, and Welfare (MHLW) approval, and placements of instruments and clinical laboratories in Japan.
As we began this partnership with Thermo Fisher to develop a companion diagnostic, we discussed our initial PMDA submission strategy for the ODxTT. Here are some key points for that strategy. First, the submission would be an amendment to extend the ODxTT indication for use as a companion diagnostic for selpercatinib in lung and thyroid cancer. Our goal would be to have co-approval of selpercatinib and Oncomine Dx Target test. This is important because PMDA regulations do not have flexibility to approve a targeted drug in the absence of a CDx. This is unlike the U.S., where there is some flexibility, where the drug can be approved without a corresponding CDx in certain situations. Another key point for our submission strategy was that the same companion diagnostic clinical validation study, specifically, the bridging study, would be used for the U.S. and Japan. Lastly, Thermo would seek PMDA feedback on our CDx submission package prior to submission.
First, we’ll talk about lung cancer. As we were working with Thermo on the development timeline for lung cancer, we had several milestones to consider. We knew that selpercatinib wouldn’t be approved without a CDx. Because of this PMDA regulation, Loxo asked Thermo to prioritize the Japan submission over the U.S. submission. Some key points regarding the timeline are that, first of all, selpercatinib received orphan drug designation, resulting in a nine-month PMDA review time. The ODxTT analytical validation package was submitted at an official PMDA consultation meeting prior to the official CDx submission. This pre-evaluation is allowed by PMDA, and has the potential to reduce the review timeline if PMDA finds minimal issues with the package. In selpercatinib’s case, the review time was reduced to five months. In the time between approval and launch, reimbursement with MHLW was granted. As I mentioned before, the goal here was co-approval and co-launch of selpercatinib and the ODxTT. You can see that, based on this development plan, we would meet that goal of co-approval and co-launch.
Thyroid cancer had a different timeline for both selpercatinib and ODxTT. For thyroid cancer, Thermo needed to complete the analytical validation work for RET, which led to the submission occurring separately from lung cancer. Again, selpercatinib had orphan drug designation, which resulted in expedited PMDA review time of nine months rather than twelve months. The actual selpercatinib submission for thyroid cancer occurred in May of 2021, but Lilly Japan resubmitted the package for the SNDA, or the supplemental new drug application, in October, after the lung cancer approval had been received.
Because selpercatinib was already approved in Japan for lung cancer, no additional time was needed for reimbursement and launch in thyroid cancer. You’ll see a note on the top gray bar that approval and launch occurred at the same time for selpercatinib. However, additional time was needed for ODxTT to be reimbursed and launched in thyroid cancer as a new indication. We ended up being in a situation where the drug was approved and the CDx was approved, but not commercially available. I’m happy to say that Thermo worked with MHLW to get the reimbursement and launch completed as quickly as possible so that patients in Japan could have access to both the drug and the diagnostic. However, in hindsight, we realized that the goal should have been looking at co-launch rather than co-approval, and that the submissions adjusted accordingly to make co-launch of the drug and the diagnostic.
Before we get into the submission package, I’d like to give a brief overview of the LIBRETTO-001 study design. The efficacy outcome data and samples from this study were used for both the selpercatinib and the ODxTT submissions. The LIBRETTO-001 study is a phase 1/2 study with a dose escalation phase, which is the phase one portion, followed by dose expansion phase, the phase two portion. Submissions for both the drug and ODxTT were comprised of both Phase 1 and Phase 2 patients. The study was a single arm study, meaning that no RET negative samples or patients were enrolled into the study. Patients were enrolled into the study based on a laboratory-developed test result, or an LDT result. The LDT could be one of any several technologies, including NGS, PCR, or FISH. As you can see, there were multiple tumor types enrolled into the study: lung cancer, thyroid cancer, and MTC, medullary thyroid cancer. Each of these indications was broken down into further patient populations, with patients that had received prior therapy — either prior platinum therapy for lung cancer or other treatments for thyroid cancer — or patients that were treatment-naive, and had not received any prior therapy. One final important note is that the selpercatinib submission in Japan utilized a later enrollment cutoff and different efficacy analysis sets than the U.S. submission. We’ll talk about that in a little bit more detail in a couple of slides.
Going back to the initial PMDA submission strategy, I’d like to dive a little deeper into the point that the same CDx bridging study would be utilized for the U.S. and Japan. That pretty much all changed after PMDA provided feedback on the selpercatinib and ODxTT submission package. Several requests from PMDA resulted in a change in the submission strategy. First, the drug division of PMDA requested that a later data cutoff be used than the U.S. Additionally, PMDA requested that the statistical analysis plan, or SAP, must align with the protocol cohorts for the analysis sets. This is in contrast to the U.S. SAP, which described analysis that was based on negotiation with FDA, and were based on a primary analysis set and supplementary analysis sets.
The best analogy I can use here is to think about a deck of cards. The same deck, the same trial, was being used for the U.S. as was being used for Japan — however, for the U.S., it was a slightly smaller deck. For those of you that are from Indiana or the Midwest and play Euchre, that might be a good analogy for you to consider. It’s a slightly smaller deck, and the cards, or patients, are separated by suit, so, hearts, clubs, and spades, whereas for the Japan submission, the patients were in a slightly larger deck, a later data cut off. More patients were enrolled, and the patients were separated by face card, so, kings, queens, aces, etc. It’s not a perfect analogy, but hopefully, it helps illustrate the differences between the submissions.
The second piece of feedback was that the drug division also requested that an efficacy analysis of the patients enrolled in Phase 2 only be provided. Lastly, the device division of PMDA provided feedback that the CDx package, or the ODxTT package, must match the selpercatinib NDA submission package, meaning that it had to follow the same analysis sets and data cutoffs as the selpercatinib submission. They also requested that as many samples as possible must be tested. Because the device division said that the CDx package must match the drug package, we would not be able to use the FDA ODxTT clinical bridging study results for the submission. As a result, and to accommodate the feedback, we, in partnership with Thermo Fisher, made a unique protocol statistical analysis plan, and a report was written for Japan.
Thermo Fisher had to prepare two major components of the submission for the ODxTT. The first is the analytical validation. Several studies were conducted by Thermo, one of which involved the clinical trial samples from LIBRETTO-001 to establish analytical accuracy, which is an agreement with an orthogonal method. The second is the clinical validation. There’s two components of the clinical validation. For this project, because patients were enrolled using local laboratory-developed tests, a bridging study was required, and so specimens from a subset of patients enrolled in Phase 1 and Phase 2 were used for the bridging study. The bridging study, as I said, has two pieces: the clinical concordance, which is agreement with the enrolling LDTs, as well as an efficacy analysis.
On the next slide, we’ll start getting into the fun stuff, some of the results from the submission, but first, we’ll talk about the analytical accuracy in lung cancer. The goal of this study is to establish performance of the ODxTT with reference to a validated orthogonal method. In this case, an NGS method was appropriate. The acceptance criteria were defined ahead of time, which include a positive percent agreement, or PPA, of at least 90%, and a negative percent agreement of at least 90%. You can see in the table below that these acceptance criteria were met. In the next slide, we’ll move into the bridging study for lung cancer. The first piece is the clinical concordance. The goal is to establish concordance between the enrolling local LDTs and the ODxTT results. The results from this study show that the agreement estimates of PPA and NPA were high. You may recall that in the overview of the LIBRETTO-001 study, I mentioned that many different LDTs were used, so establishing concordance between LDTs and ODxTT was important for the submission.
The last piece for the clinical validation in lung cancer was the efficacy analysis. The goal here is to determine the overall response rate based on ODxTT results from LIBRETTO-001 lung cancer specimens. Just to explain the graph a little bit, the light blue bars represent patients that were enrolled into LIBRETTO-001. These patients were all positive by the laboratory-developed test. The dark blue bars are patients that were enrolled in LIBRETTO-001, and were positive by the ODxTT. Each different set of bars is a different breakdown of the patient population. The first set are patients that had lung cancer that had been previously treated. The second set is lung cancer patients that were treatment-naive. The third set is all patients combined, and then the fourth set is patients that were enrolled into the Phase 2 portion only, as requested by PMDA. As you can see, estimates of the primary clinical endpoint overall response rate were consistent between the ODxTT-positive population and the LDT-positive, or the drug efficacy analysis sets.
Just to summarize, the data supports the claim for RET fusions in formalin-fixed paraffin-embedded lung cancer samples to identify patients eligible for treatment with selpercatinib. The first three bullet points are what we covered today, the data that was submitted. I didn’t talk about the sensitivity and other analysis that was conducted by our stats team at Loxo@ Lilly. That also supported the above conclusions for the clinical concordance in the efficacy analysis. I gave away the spoiler at the beginning, but the ODxTT was approved as a CDx in lung cancer in early September, and a few weeks later, selpercatinib was approved in Japan for its first indication in lung cancer.
Although the timing for thyroid cancer was later, a similar approach as lung cancer submission was taken, incorporating the PMDA feedback given for lung cancer. The thyroid submission was a little bit more complex, because it was really comprised of two indications, RET fusion-positive thyroid cancer and medullary thyroid cancer for detection of RET mutations. I’m not going to go through the bridging study results in thyroid cancer, just to say that they were very similar to what was observed for the lung cancer. There is one topic that I would like to discuss, and that’s a question that was raised by the drug division of PMDA during the selpercatinib review. PMDA requested a rationale for each of the mutations detected and reported by ODxTT, that we provide evidence that those mutations were activated, and the patients that had those activating mutations would respond to Retevmo (selpercatinib).
To address this question from the drug division of PMDA, Lilly worked with Genomenon to extensively evaluate and annotate each variant. This annotation was based on a large, comprehensive database of genomic evidence based on thorough review by scientific experts consistent with the ACMG and AMP guidelines. Approximately 97 variants were categorized as pathogenic, likely pathogenic, variant of unknown significance (VUS), no data, or likely benign or inactivating. As we look through these, the variants that were categorized as pathogenic or likely pathogenic were automatically included into the CDx claim. Those that were likely benign or inactivated were automatically excluded from the CDx claim. That leaves the two yellow boxes in the middle. We took those variants under further investigation and looked at the patients enrolled into LIBRETTO-001. If those variants were observed in the clinical trial, then we would look at the response data to see if we could have evidence to support that the variant was activating, and that the tumor responded to selpercatinib. As a result of this evaluation, several variants were removed from the CDx claim.
So, based on the Loxo experience for selpercatinib and Oncomine Dx Target Test, just a few points or tips to share regarding and to facilitate regulatory review. First is to discuss the submission package and forecasted timing with PMDA prior to starting the validation work. This was critical for our approach and how we handled the submission with Thermo Fisher. Then, to test all available clinical trial samples for the bridging study; to provide clear sample accountability diagrams, those are very important to illustrate how many patients were enrolled, how many patients had samples, which samples were eligible for testing, etc.; importantly, having a CDx partner that has experienced a prior consultation with PMDA is very beneficial, as they can help us navigate these kinds of issues and questions, and get resolution quickly. Lastly, be prepared to quickly answer thorough and meticulous questions regarding the submission, particularly if you have a large number of variants to be reported as a CDx claim.
To summarize the key points or take-home messages, first of all, Japan CDx regulations do not allow for drug approval of a targeted agent in the absence of a corresponding CDx. Submission timing of the drug and diagnostic needs to account for co-approval or co-launch, depending on the prior approval status of the drug. The CDx submission must match the drug submission package in requirements. Lastly, discussion with PMDA on the submission strategy is critical. With that, I’m happy to turn it back over to Mark and take any questions.

MARK: Thank you so much, Susanne, for that presentation! I think you and I can both come back on camera. We’ve got a couple of questions that have come in from the audience. I’m going to glance down to read some of them. There’s a question that I have before we go to the audience questions — in your experience, how does the complexity of this situation compare with your previous programs? For instance, RET has both loss-of-function and gain-of-function variants, it’s associated with both hereditary and somatic cancer, and a number of different diseases, in addition to the fact that it’s associated with fusion events and mutations. I feel like it’s as complicated as you can get. I don’t know, in your experience, if that’s true, or you took each of those challenges in stride, and it wasn’t overly burdensome?

SUSANNE: No, it was definitely a complicated submission. It’s really three different indications, which made it much more complicated, and then, as you said, fusions as well as mutations. That added complexity. This was really my first experience with an NGS test, and it has its own details as well that are important. You’re not only concerned about the RET gene, but there are other genes that are being looked at as well through the testing. That doesn’t necessarily impact the submission, but it maybe impacts how Thermo Fisher would work on the submission, because they need to be consistent with their prior submissions and provide the same data package as they would for other submissions.

MARK: Good. And you handled it, despite how complicated it was. I’m going to make this question from an audience member a little bit more generic, to get a higher level view. It builds on what you said about this being your first experience with an NGS test. Can you help me understand the validation process, and particularly, the sample acquisition process? You’ve talked about FFPE, which is standard. Was that required of the samples? You mentioned that there were multi-assay modalities that you would accept. Can you give me a sense for how the regulatory bodies viewed the analytic validation and the clinical validation from a sample perspective?

SUSANNE: Yeah. Thankfully, Thermo Fisher is responsible for the analytical validation piece, and they procured the samples that they needed for analytical validation, but that is not an easy task, especially for RET and lung cancer, where the prevalence is only 1-2%. Imagine you have to screen 100 samples to maybe find one or two, and you need to have enough material to run an extensive, reproducibility study, where you’re running the same sample over and over again, but under different conditions. It can be quite challenging. With the MTC, we actually had the opposite problem, where the prevalence is about 60%, and we needed to identify negative samples to support the negative percent agreement and the analytical accuracy and the clinical accuracy studies. There we had the opposite problems. MTC is not all that common compared to lung cancer, but 60% of the patients you do find have some kind of RET mutation. We were looking for the negative sample, so it made it a little bit more challenging to find those. A lot of companies — Thermo didn’t really feel this way, but some companies think, “well, the pharma company has access to all these tissues because they’re running all these clinical trials,” but we try to collect as little as possible in the clinical trial, because that’s what’s best for the patient, and so we can’t really support those huge amount of tissue requirements that are needed for the analytical validation study. We maybe had one or two that were very rare, where we had a few slides we could spare for Thermo, but not really very many. So, it is quite a challenge.

MARK: Yeah, that echoes my experience in working with other pharma companies as well. That is a pretty significant time constraint. It takes quite a long time and a lot of effort to get those samples together to prove out the validation. Another question that we had was asking if you could summarize the differences between the regulatory submission process in the U.S. versus Japan, as well as contrast that with your experience working in other countries.

SUSANNE: Thankfully, again, for the submission in Japan, Thermo has an affiliate in Japan that is well-versed in the submission process. We were able to write up the same kind of reports that we did for the U.S. submission, and then the Thermo team nuanced it to meet the PMDA requirements. In general, they’re fairly similar, but the types of questions that we receive from the regulatory agencies were a bit different across the globe. Obviously, in the EU, they have just had a change in their regulations from the IVD directive to the IVDR regulations. Thermo Fisher has achieved the CE mark under the IVDD, the old regulations, and are in the process of getting it through the IVDR process. I don’t have a lot of experience. I don’t think anyone does, because there’s only one companion diagnostic approved under IVDR, to really know how those are going to go through the system.

MARK: Since you brought that up, how incipient many of the diagnostic modalities are and how the pace of change is brisk, can you speak to any educational component from the pharma to the regulatory bodies? FDA, or in this case, PMDA? How much do you feel like you had to help guide their understanding of the samples and the clinical circumstance and the interpretation of the result?

SUSANNE: One of the things that we had on our side was that the efficacy profile for Retevmo was very good. That helps, because the regulatory agency wanted patients to have access to the drug. The questions, I think, were intended to help us get to that point, where the test could be available. With PMDA, the questions were different, in that I think PMDA assumed that the drug and the diagnostic were developed literally together at the same time, so that the mutations that were reported out, specifically for MTC, Retevmo, or selpercatinib, would only work in those mutations. However, for the ODxTT, we started development back in 2015, and as you know, we’re constantly learning new variants that could be implicated in the disease. Things change over time. Selpercatinib was started later, completely independent of the ODxTT, and didn’t really have a specific list of variants in mind that they were aiming for — just looking to block RET kinase activity. In the beginning of our questions from PMDA, they wanted to know more about the development of the tests and the diagnostic, and how they were developed in parallel, but that was one point that we had to explain, that they were different, and now, we were trying to merge them together and fit them together as best as possible.

MARK: We’re with a different group, and we’re addressing some of those challenges of looking at the data in light of new evidence, new information, and basically proving out the case that these variants will respond similarly to the previous variants. A question from the audience, I’ll read verbatim here: you said Genomenon was involved in providing the evidence for the variants in response to the PMDA question, but was Genomenon involved with the initial variants on the preliminary or initial CDx claim? If you could speak to that.

SUSANNE: No, I don’t believe so. Thermo created the amplicons, the regions of the gene that, at the time, they felt were of interest for cancer and what was driving the cancer. That’s what led to the questions from PMDA, how are those developed? Initially, we tried to address that at a high level, explaining that certain regions of the gene were targeted based on literature at the time, but that really wasn’t sufficient for PMDA. What they really wanted was the detailed report that Genomenon provided, characterizing each variant based on its pathogenicity.

MARK: Another question from an audience member: when, during the review process, did PMDA ask the question about providing evidence? At what point? Was it toward the end, and there was sort of a scramble to get that information together, or was it toward the beginning with the initial submission review?

SUSANNE: That’s a good question. Initially, they had those high level questions about the development, kind of early on, I’d say a month or two within the drug submission. We waited to see how they responded, and then they came back with more detailed questions, and that’s when we turned to Genomenon. We thought we had a time frame of a couple of months to answer those, and to do the detailed annotation, but I think about a month into it, PMDA came back and said, “actually, we need that right now.” Thankfully, your team was excellent at responding very quickly, and we were able to turn around the report. Then, they meticulously reviewed the report, and had more questions, and again, the Genomenon team was phenomenal at getting those questions. I remember it distinctly, because it was December 23rd, two days before Christmas, and we were still getting questions, and your team was still responding. I was very appreciative.

MARK: Yeah, you get a chill down your spine whenever you see a client have an email that says “URGENT” in the subject. I remember, Jeffrey Bissonnette on our side spearheaded a lot of these emails and requests for further information, and would dig in a little bit deeper for me with respect to the types of questions that the PMDA was asking, whether it was solely focused on the interpretation, were they predominantly interested in the activation state of each variant and the empirical evidence, the clinical and functional evidence, to prove out that information, or did it have to do with the classification framework or interpretation framework, the AMP and ACMG guidelines.

SUSANNE: Yeah, early on, we proposed to use a report produced by Genomenon, and gave them an example report, and they didn’t seem to question or have concerns about how we were approaching the question. Then, once the detailed report was provided, I can’t remember any of the really detailed questions off the top of my head, but they were finding things that I didn’t even know were on the spreadsheet. They were taking a very close look at it, just making sure that it was correct. I obviously didn’t see all the questions from the drug side. The other big question that I recall, and my colleague Anya handled most of those, were the LDTs that were used to select the patients, and how were the patients with specific mutations selected to be enrolled into the trial? What were the criteria that were used, not only to allow the test, but also the variant itself into the trial? Again, they wanted to see consistency between the ODxTT and the variant selection process that was used for the clinical trial. I think we were able to show that because our process within LOXO@Lilly was fairly close to what Genomenon uses. It’s just simpler, because it was a one-by-one kind of scenario, whereas, all of a sudden, we had 87 variants that we needed to look at, all at once.

MARK: I’m sure it was a relief to survive such meticulous scrutiny. It makes you feel very confident. I liked very little when a teacher would give me an A and there would be no red marks on my essays. It sounds like, as you said, there was very careful attention to the detail, and the evidence survived that process of scrutiny, so having a great deal of confidence to the end result. Last question: I was just at the Precision Medicine World Conference last week, and real-world evidence was a big topic of discussion. I wonder, in the context of somatic or hereditary cancer indications, what are your thoughts about the role or usefulness of real-world evidence and real-world data in regulatory submissions? That’s a little bit of my take on the question, but it’s a question from an audience member.

SUSANNE: I think that’s a great question. You know, I haven’t seen it used for CDx yet, but I’ve heard talk. This might be another way that we can approach it. I’m one of those people that, until I see it, it’s hard for me to really describe what I think it would look like, but I think it would be great for patients, if that’s another avenue that we can use to support a CDx submission.

MARK: Good. Excellent. Let me end there by thanking you again, Susanne, and wishing you safe travels wherever you’re going after the webinar, and also thank the audience for their questions and for their participation. I’ll remind everyone that a recording of this webinar will be shared with all of you who’ve attended, or for those of you who weren’t able to, by email. Visit our website, genomenon.com, to contact us or learn more, including a deeper discussion about some of the needs that you may have that are analogous to what Susanne talked about today. Thank you so much, everyone, and have a great day!

SUSANNE: Thanks!

MARK: Thanks, Susanne. Bye.