Supplementary file2 (MP4 272492 KB)

Additional Information: A video abstract, and the podcast and transcript can be viewed in the online version of the manuscript. Alternatively, the podcast and the video abstract can be downloaded here: https://doi.org/10.6084/m9.figshare.23975517.

1 Xiuning Le: [00:00]

Hello. Thank you for joining us for this podcast entitled EGFR Tyrosine Kinase Inhibitors for the Treatment of Metastatic Non-Small Cell Lung Cancer Harboring Uncommon EGFR Mutations. My name is Xiuning Le. I am an Assistant Professor from UT MD Anderson Cancer Center. Today, we have three other experts in the field for this discussion. Dr Daniel Costa, can you please introduce yourself?

2 Daniel Costa: [00:27]

Hello. I am Daniel Costa. I am a Thoracic Medical Oncologist at Beth Israel Deaconess Medical Center. I am the group leader for our Thoracic Oncology Program and an Associate Professor of Medicine at Harvard Medical School.

3 Xiuning Le: [00:41]

Dr Eric Nadler, can you please introduce yourself?

4 Eric Nadler: [00:44]

My name is Eric Nadler. I am a Head and Neck and Thoracic Oncologist at Baylor University Medical Center in Dallas, and I run health informatics for thoracic and head and neck cancers in the US oncology networks. I have a huge database of information from which to cull.

5 Xiuning Le: [01:04]

All right, Dr John Heymach, can you please introduce yourself?

6 John Heymach: [01:07]

Hi, I am John Heymach. I am a Thoracic Medical Oncologist and I am Chair of the Department of Thoracic and Head and Neck Medical Oncology at MD Anderson Cancer Center.

7 Xiuning Le: [01:15]

All right. For today, first of all, I want to ask Dan to provide some background information on what we really mean by saying uncommon EGFR mutations, and what are uncommon EGFR mutations’ role in lung cancer?

Infographic

figure a

8 Daniel Costa: [01:34]

Thank you, Xiuning, for that question. I think that it is very important to understand the prevalence of different EGFR mutations. As we know now, EGFR mutations are present in around 15% of all lung adenocarcinomas diagnosed in the United States [1,2,3]. The prevalence changes depending on worldwide characteristics, with higher frequency seen in East Asian countries [1, 3]. Amongst the various types of EGFR mutation, the predominant changes occur in the kinase domain [1, 3]. The most frequent mutations are within the exon 19—small deletions or small insertion deletions (indels) [1,2,3]. They account for around 45% or so of all EGFR mutations, followed by a common exon 21-point mutation called L858R that accounts for another 35% of all EGFR mutations [1,2,3]. So, these two classes of mutants are the most common EGFR mutations and are usually called classical EGFR mutations [1,2,3]. The third most common type of EGFR mutation are small insertions within exon 20 of EGFR [1, 3]. They account for around 5–10% of all EGFR mutations [1, 3].

These three main classes are where most clinical trial development and drug approvals have occurred. For the classical EGFR exon 19 deletions and L858R mutations, there are many classes of first-, second-, and third-generation EGFR inhibitors approved, all of them with high-level, randomized clinical trial data to support their use in clinical practice. These EGFR inhibitors have supplanted the use of traditional cytotoxic chemotherapy for the first-line care of patients with metastatic disease, and some of them have entered our use in more early-stage disease, including the adjuvant setting. The current preferred EGFR inhibitor for mutations worldwide is the third-generation EGFR inhibitor osimertinib, which has the best adverse event profile, the highest central nervous system activity, and best clinical outcomes, as it relates to both progression-free and overall survival, both in the metastatic and in the adjuvant setting [4].

For the EGFR exon 20 insertion mutations, the two approved drugs are different to the ones approved for the classical mutations. For EGFR exon 20 insertion mutations, the FDA, over the last 2 years, approved the EGFR inhibitor mobocertinib and the monoclonal antibody amivantamab, due to the activity seen in groups of patients who harbor these mutations that, in preclinical models, are not as sensitive to first-, second-, and third-generation EGFR inhibitors [5, 6].

Now, after these three classes of mutations, we come up with another group of around 10% or more of kinase domain mutations that are grouped together and called less common, uncommon, or atypical EGFR mutations [1,2,3]. Amongst this group, the most frequent is the EGFR G719X series of mutations. The X is usually either G719A, G719C, or G719S. They account for around 3–6% of all kinase domain mutations [7]. The next most common amongst these uncommon mutations is the exon 21 L861Q mutation, accounting for 2–3% of all kinase domain mutations, and then the exon 20 S768I point mutation that accounts for approximately 1% of all kinase domain mutations [7]. All these mutations, as we will discuss in this podcast, have their particular preclinical response rates to EGFR inhibitors, and some clinical approvals that match that pattern, as we will be discussing, including both for approved drugs and guideline-approved therapies.

9 Xiuning Le: [06:23]

Thank you, Dan. So, like you pointed out, I think that currently there is a lack of a robust clinical dataset for uncommon mutations. Now, the question is, can we use preclinical models and their findings to help support a treatment decision? John, can you please comment on that?

10 John Heymach: [06:45]

Thanks, Xiuning. You know, in preclinical studies, we have looked at more than 100 different mutations and combinations, and the data clearly tells us that sensitivity to different tyrosine kinase inhibitors (TKIs) varies between different EGFR mutation subtypes [8]. So, given all these different mutations, we can not do separate studies for each mutation. So, the question is, how to best classify or group them. [EGFR mutations have] been grouped or classified by a couple of different methods in the past. One is based on the frequency, just common or uncommon [8]. A second is to classify mutations based on the location, like exon 18, exon 19, exon 20 mutations [8]. But neither one of these [approaches] is a great predictor of the sensitivity. The key issue is how a mutation impacts the three-dimensional structure of the EGFR protein, and how it interacts with the TKI. So, it is now possible to categorize the mutations based on the structural impact on the EGFR protein, and it turns out that this is a much better predictor of their sensitivity to TKIs.

Recently, we defined four different categories [1]. The first one is classical-like mutations, and these have sensitivities like the classical mutations [1]. One classical-like mutant is L861Q, and this tends to be broadly sensitive to all different EGFR TKIs [1]. The second group of mutations occur on the interior surface of the ATP-binding pocket [of EGFR], and they are predicted to squeeze together that binding pocket and make it a bit smaller [1]. We call these PACC mutations [1], which stands for P-loop alpha-C helix-compressing [1]. This group includes mutations in different exons [1]. Common ones include the G719X mutations and S768I [1], and it turns out that these mutations are most sensitive to second-generation TKIs, such as afatinib, but are less sensitive to first- or third-generation inhibitors [1]. The third class are insertions in the loop at the C-terminal end of the alpha-C-helix of exon 20 [1], so we call these exon 20 loop insertions [1]. Actually, we can divide the loop insertions into two groups—far-loop and near-loop insertions [1]. The far-loop insertions are generally resistant to TKIs, but they may be sensitive to a new class of EGFR exon 20 inhibitors that are more mutant-selective [1], whereas the near-loop mutations may be more sensitive to second-generation TKIs and exon 20 insertion-specific TKIs, such as mobocertinib, as well as amivantamab, but they are not sensitive to first-generation TKIs, such as erlotinib and gefitinib, or to third-generation drugs such as osimertinib [1]. The final category is T790M-like mutations [1]. These occur in the hydrophobic core of the EGFR protein, and we have two different subgroups of these, according to their sensitivity to the third-generation EGFR TKIs; T790M-like third-generation sensitive, and T790M-like third-generation resistant [1]. So, to put this all together, and I know I just covered a lot of different information there, but the key take-home [message] is that there is an increasing body of preclinical data that lets us better select which TKI for an uncommon mutation based on the structural subgroup of EGFR; but this does require more robust clinical data to confirm these preclinical observations.

11 Xiuning Le: [11:03]

Thank you, John. In thinking about clinical data, in 2021, National Comprehensive Cancer Network (NCCN) guidelines had an update that included afatinib and osimertinib as preferred first-line treatment options for patients with metastatic lung cancer, specifically harboring G719X, L861Q or S768I mutations [4]. Dan, can you comment on the clinical evidence behind this NCCN guideline update?

12 Daniel Costa: [11:38]

Sure. I think that it is important to note, as John just highlighted, the incredible translation of the findings from the laboratory in preclinical models to what we see in the clinic. So, I think, as John highlighted, some EGFR mutation subtypes, specifically the G719X-type mutations L861Q and S768I are EGFR TKI-sensitive. They do have a pattern of response that is slightly different, depending on the mutation, than the common mutations, but there is evolving clinical data to support the use of either FDA-approved inhibitors for that mutation type, or FDA-approved inhibitors for other mutations. So, as an example, for these three classes of EGFR uncommon mutations, there was a 2018 approval for the second-generation EGFR inhibitor afatinib at its 40 mg daily dose [7, 9], and this was based on a post-hoc analysis of the LUX-Lung 2, 3, and 6 registration trials of afatinib [7, 10] (Table 1). In that particular analysis, 38 patients with G719X, L861Q or S7687I were analyzed and demonstrated an overall response rate of 71%, progression-free survival of 10.7 months, and overall survival of 19.4 months. These data were quite similar to that observed with the more common mutations [10] and led to the clinical approval of afatinib for these three less common EGFR mutations. But it has been pretty clear as we have evolved data, not only from the FDA approval but other larger cohorts, including large retrospective databases of use of afatinib, that there is some heterogeneity by which type of tumors with each mutation respond best, or not, to afatinib [1, 11, 12] (Table 1). As John mentioned eloquently, based on his preclinical work, PACC mutations such as G719X and S768I are the ones that have the most sensitivity preclinically to second-generation inhibitors such as afatinib, but in the clinic, those tumors are the ones also that have the highest response rates and the longest durations of response, and both the post hoc analysis of the prospective LUX trials and these large retrospective databases, plus other large multicenter studies, have confirmed this. So, I think for the data for afatinib, it is pretty clear that it is a pan-EGFR mutation active inhibitor, but for the G719X and the S768I mutations, it looks particularly clinically active, and it is where we see the highest response rates and the longest durations of response.

Table 1 Summary of clinical data in patients with NSCLC with uncommon EGFR mutations
Full size table

13 Xiuning Le: [15:05]

So, thank you Dan for summarizing the data, but really to me, I hope you agree as well, those clinical data, although they are not from a large, randomized study, are really consistent with our preclinical work from Dr John Heymach's lab, showing that the PACC mutations, especially the two that you just talked about, respond best to the second-generation EGFR TKI. Would you agree?

14 Daniel Costa: [15:30]

Yes, I agree. I think that we all know that second-generation EGFR inhibitors have a distinct pattern of adverse events, with high levels of both skin and gastrointestinal toxicities, that are related to inhibition of wild-type EGFR [12]. But in the two mutations that you mentioned, and that correlate with the preclinical work, the EGFR G719X and the EGFR S768I, the clinical responses, even when dose reductions are necessary, seem to be the most robust, really highlighting the preclinical work showing that these mutants, because of their structure–function properties, are most sensitive to this type of EGFR TKI inhibition [1, 11, 12] (Table 1).

15 Xiuning Le: [16:16]

Dan, can you go back to the retrospective analyses you mentioned? Some of them are actually pretty large-scale and also included first-generation and third-generation TKIs, as well as second-generation EGFR TKIs. Can you comment on what this research has shown, and is it consistent with preclinical data? Can you give us some insight on first-generation and third-generation EGFR TKIs?

16 Daniel Costa: [16:48]

Definitely. I think I can respond to that specifically by mutation type. So again, the other high-level correlation with preclinical data is for the L861Q-mutated tumors. So, as John mentioned, L861Q, in this novel structure–TKI response model, is within the classical-like mutations, more akin to exon 19 deletions and L858R. And I think that the large retrospective database and the few prospective trials of the third-generation inhibitor osimertinib, for these less common mutations, have shown this: that for L861Q, you see the highest response rates with drugs such as osimertinib and the longest durations of response. So, as an example, there was a phase II clinical study of osimertinib for these less common mutations, run out of Korea, and in the 36 patients that they identified for this particular study, they observed an overall response rate of around 50% for all these less common mutations [13]. But when you look at the data for the L861Q, they had response rates that were much higher, more in concert with what you see with the exon 19 deletions in L858R, while the G719X-mutant group, which was the second most prevalent in this particular study, had response rates that were slightly lower and the durations of response were also [shorter].

So, I think for the third-generation inhibitors, this small prospective study, combined with some ongoing larger datasets that have come out, including one called Unicorn that has evaluated, in large datasets around the world, the effects of osimertinib in less common EGFR mutants [14,15,16] (Table 1). It has confirmed this observation that L861Q-mutated, EGFR-mutated tumors tend to respond quite well to osimertinib, matching the preclinical data that this is a classical-like mutation. While these cohorts have consistently shown that for drugs such as osimertinib, the third-generation inhibitor, although there are responses with the G719X and the S768I, they tend to be with overall response rates that are either at 50% or below, and the durations of response seem somewhat shorter.

And now, Xiuning, the last part of your question about the use of first-generation EGFR inhibitors such as gefitinib and erlotinib; although there is slight preclinical activity in these three most common uncommon mutations (S768I, L861Q, and G719X) the clinical data have shown that the degree and duration of response is modest [17], hence why many companion guidelines, including NCCN, have not included routine use of these drugs in their recommendations. Although, one must say that for a particular healthcare system that does not have access to a drug such as afatinib or osimertinib, there is some clinical activity for these three mutants with the first-generation EGFR inhibitors.

17 Xiuning Le: [20:26]

Thank you. Now we have talked about the current NCCN guidelines for treating those uncommon EGFR mutations and we talked about the evidence behind the NCCN guideline. Eric, can you please comment on, in day-to-day, real-life practice, from the oncologist’s perspective, do you think those guidelines are followed? And then, do you think patients really derive benefit from those guideline recommendations?

18 Eric Nadler: [20:55]

It is a complicated question. Our testing patterns are in flux. If we went back 3 or 4 years, to 2019 and 2020, the testing patterns of the United States [18], particularly in the community, roughly 70–80% of people would test for one mutation [19,20,21]. But multiplex testing, or more than one mutation, was tested [for] only 45–50% of the time [19,20,21]. And at that time, next-generation sequencing was being utilized in the teens, 15–16%, even as we crossed over through 2020. If you look over the year or two after that, overlapping with the emergence of COVID-19, those numbers changed markedly, but they are still not probably where we want them [22].

First and foremost, single genetic testing panels in the metastatic setting really do not exist anymore. So, most patients are being tested for more than one thing at a time, and that has been true for at least 3 years now [22]. And if you look at next-generation sequencing adoption throughout the nation, it went from about 30% in mid-2020 to now—more than 90% of all testing is next-generation sequencing. Now, [there are] lots of challenges with next-generation sequencing because that could be both plasma and tissue, and sometimes both. And of course, in our national community of practicing oncologists, there is a lot of heterogeneity in how people test, and what people test, and which tests they use. Whether or not it is both RNA- and DNA-containing, or whether it's just tissue or plasma [23, 24]. That being said, it is certainly moving in the right direction [25, 26].

But a couple of big points here that deserve mention. The first is, are we testing all the non-squamous, or are squamous actually being tested? And we know we can see EGFR and KRAS and other mutations in squamous histologies and certainly that is happening more and more. The second thing is, turnaround time has pushed a lot of people toward testing plasma-based approaches initially, sometimes in lieu of tissue-based testing, just because of the ease of it, and the turnaround time of some of the plasma-based approaches, and whether that be Guardant® or cobas®; they tend to be quicker than our tissue-based counterparts [27,28,29]. Lastly, there is a lot of geographic heterogeneity in our testing. I work in Texas in a large community practice and almost all of our patients are being tested in non-small cell lung cancer in the advanced setting. But you go to the state to the left or the right of me, and those rates could fall to 70% even today. So, it is a challenging, complicated time, and things like this podcast I hope will help inform and educate [to perform] at least tissue-based testing, and when that is difficult for turnaround time and tissue, you could add plasma to that easily; [it] should be done in all patients with non-small cell lung cancer [30]. And like I said, over the past 3 years, our numbers have more than doubled but they are still not where any of us would consider optimal.

19 Xiuning Le: [24:37]

Eric, may I ask a follow-up question? So, for physicians, when they see a report with an EGFR G719S, for example, how often can the clinical oncologist really pick the recommended afatinib upfront? Or do you think there is still a need to educate even our peers so that they can use the current guidelines to match the appropriate genotype to the right drug?

20 Eric Nadler: [25:13]

Basically, if you look at the testing patterns of the last 18 months in the United States, we are certainly moving more toward an optimal direction in which the majority of our patients, now nationally, are being tested with next-generation sequencing [18]. Furthermore, we are moving more and more toward a more comprehensive testing pattern of thoracic malignancies in which not just the non-small cell non-squamous patients are being tested but also an emerging and larger and larger percentage of the squamous patients are being tested. And we are seeing that in our community databases as well, both of which I think is a giant step forward.

21 Xiuning Le: [25:55]

Thank you, Eric. Although afatinib is now approved for non-resistant mutations, other uncommon mutations are still lacking an approved TKI. So, there is a clear unmet need for the mutations we actually know that exist in the PACC, for example, and the other classes, but not the three that we mentioned. We do not have a real approved TKI. John, can you please comment on how you envision the field of uncommon EGFR mutation will evolve going forward?

22 John Heymach: [26:33]

It is straightforward preclinically to match different mutations with the best drug because we can do those studies in the laboratory, and they are relatively straightforward. The greater challenge is how do we test this in the clinic, given the enormous number of mutations that are out there. You know, as I mentioned, we have seen more than 100 different mutations or combination mutations, [or] compound mutations that may have two or three mutations combined together [1]. So, how do we implement this clinically and verify that what we find in the lab matches what happens in the clinic? Well, this is where the classification system I described before can be helpful, because we can not do separate clinical trials for every mutation, but we can do clinical trials that encompass a whole group of mutations, like PACC mutations or classical-like mutations, and test drugs for that specific subgroup [1]. So, I think that is really going to be an important step forward now, which will let us start testing mutations much more broadly than just classical mutations or the three next most common mutations—the G719X, the S768I, and the L861Q. I think having a classification in place will really let us start testing for any mutation that comes along, as long as we know which group it belongs to. And so, you know, I think this is really the greatest unmet need, these rare, atypical mutations.

So, moving beyond these three mutations, the G719X, the S768I, and the L861Q, these other uncommon mutations are much more heterogeneous, and this is where the classification can be helpful for enabling patients with any of these mutations to go on studies. Developing a larger clinical dataset for these rare, atypical mutations, then, is really the most important next step, and the larger this database grows and the more practical it is to do these clinical studies, the more we will learn about how to treat these rare, atypical mutations. Now, we do have some data with some other rare mutations. For example, L747X and E709X have demonstrated some clinical sensitivity to afatinib based on clinical databases and case studies that we have [11, 31, 32]. And both of those would be classified as PACC mutations. So, that is consistent with the data I mentioned before; that PACC mutants seem to be more sensitive to afatinib. In case reports, osimertinib, a third-generation drug, has demonstrated variable activity against L747P, with some studies saying that [this] mutation is resistant to osimertinib, others saying that there is some sensitivity [33,34,35]. So, I think we need more data really to see how that settles out.

Now, one other challenge that is worth mentioning is tumors sometimes have more than one mutation occurring in the same gene. We call these compound mutations. As an example, and this was actually early work from Dan Costa that I think really highlighted some of this: tumors will often have a mutation in, for example, the G719 and a 709 mutation or a 758 mutation at the same time. So, there are pairs of mutations that commonly occur together, and there is not an easy way to predict what the sensitivity is going to be other than to test those directly. So, there I think we are going to need to just have a large retrospective database to get clues, use the preclinical modeling, and then test them prospectively in the same type of clinical trials that we mentioned before. You know, in some recent meta-analyses, outcomes in patients who had EGFR inhibitors were generally better among patients with compound mutations if they had a classical mutation like a deletion 19, or an L858R with the second mutation, as compared with when there were two rare independent compound mutations [22]. So, if you have got a classical mutation as part of the compound mutation, it is more likely to be sensitive, is what the data are suggesting. And there are data supporting that those compound mutations, if they have a classical mutation as part of it, that are often sensitive to erlotinib, gefitinib, so first-generation drugs, or afatinib [36]. But if they have got two uncommon mutations, afatinib seems to perform better [36]. There are also some data saying that osimertinib is effective against certain compound mutations [22]. In one study, and it was just with four patients, there was a response rate of 75%, among those four patients, to osimertinib if they had a compound mutation that did not include deletion 19 or L858R [13] (Table 1).

In the recent Unicorn study, there was an encouraging response rate and progression-free survival in patients with some uncommon mutations [15]; so, in patients who had an uncommon mutation plus an L858R, or exon 19 deletion, or T790M, the response rate was 61% and the median progression-free survival was 30 months [15]. And in a much larger retrospective database with over 8000 patients with next-generation sequencing data, among patients who received osimertinib, there were eight patients who had a T854A mutation plus an L858R classical mutation [37]. And there the response rate was 80% and median progression-free survival was 10 months [37]. So, in summary, the real-world data can help support treatment decisions for some mutations, but we just do not have enough data from enough different mutations to make firm decisions for every mutation that comes along. A major unmet need is to develop more data with more different drugs so that we can best start matching drugs with mutations.

23 Xiuning Le: [33:03]

From today's conversation, we have heard [about] many different EGFR mutations, some at the mutational level, sometimes we use different words to refer to as uncommon, atypical. So, Dan, let me ask you, is it a problem that we have so many different names and different nomenclatures around the mutation? Is it a barrier for physicians to know how to offer the best treatment for a specific EGFR mutation to a patient?

24 Daniel Costa: [33:32]

Yes, I think that this is a major, major issue that needs to be addressed. I think that the lack of standardization on how to both classify the diversity of EGFR mutations and how to ask our partners in the diagnostic setting to use any classification that we use creates a lot of potential places where there could be both gaps in knowledge and also gaps in implementation of the evolving knowledge that we have been discussing here from preclinical models to prospective, real-world, retrospective datasets. The information is not always easily accessible to a practicing provider. I think, as Eric pointed out, although there had been an uptick in the use of more comprehensive genetic profiling, including next-generation sequencing that can capture the breadth and diversity of both single or compound EGFR mutations, the explanation of an identifiable mutation is not always easy for a provider to understand and to link to both FDA approval of particular EGFR inhibitors and also NCCN or other large guidelines that exist. And, obviously, I think for physicians that are specialized in EGFR-mutated lung cancer, similar to myself, John, and a few academic oncologists that do this on a day-to-day basis, like yourself Xiuning, obviously we can keep up with the evolving use of different terminologies and we can adapt much easier with the ideas of using a mutation-rich, structure-based classification where we are almost just focusing on a particular mutation, and then understanding all the work that has gone on from preclinical models to all clinical studies. But I think that this does not translate well for a community practicing oncologist. Obviously, it will be important to get Eric's perspective, what he hears from colleagues and others. But I do believe that one clear unmet need is to have better reporting where all this knowledge that we've gathered since the discovery of the EGFR mutations in 2000 and through 2004 to now is put together so that the clinician can read [and] understand in simple terms what that particular mutation means from sensitivity to inhibitors; give them a list of the FDA-approved inhibitors, and why they were approved, and also other options such as NCCN guidelines where they can use off-label. And then match with the clinical scenario, like if they see a patient with brain metastasis and then they have the ability to select between drug X or Y, based on both efficacy, clinical parameters, toxicity, expectations, and others. So, I think that this is a major unmet need to address, Xiuning.

25 Xiuning Le: [36:42]

Eric, can you also comment? I think you probably work with more diagnostic platforms as well. Can you also comment on what is your perspective of this challenge with so many mutations and so many different nomenclatures?

26 Eric Nadler: [36:57]

Well, I completely agree with Dan, and he stated it beautifully. But the challenge is when you're working outside of a large academic institution where they have a strong preferential internal testing platform. In the community or in smaller academic centers, you are talking about potentially two or three different platforms that are being used by every practitioner, sometimes more [38]. In my institution, the hospital uses one, and I routinely use another one. And then when I see patients in second and third opinions, I am seeing different tests. And the challenge is, each one of these testing companies reports this information in their own unique and distinct way with no homogenization of how they classify the mutation itself; whether it be typical or atypical, how they describe the mutation, whether it be the full mutation out to seven or eight digits, or kind of lump and group it into a more simplified form. There are few, if any, recommendations in terms of treatment in any of these reports that really direct you in terms of recommended treatments. And equally importantly, and I'll just mention, it does not direct you toward a clinical trial if one is available in your community or neck of the woods, which I think would be very helpful as well. So, one would hope that over the next 12–24 months that the testing communities themselves both simplify the reports, make it more streamlined and homogenized, so that we have a repeatable and understandable form in which this information is communicated to us. And lastly, that this is coupled with both NCCN or IASLC or other types of guidelines that are given to the practitioner, as well as clinical studies that they could look for that are within 100 or 200 miles of them. And I think all of these things would be important; and they are all underway. So, I am not saying anything that is not currently being discussed and happening, but I think that is very important, that this really gets codified in a rigid way that you can look at a report, know what you are looking at, and know how to take that report and make actionable decisions based on it.

27 Xiuning Le: [39:24]

To summarize today's discussion: an update to NCCN Guidelines® [4] regarding G719X, L861Q, and S768I is supported by a growing body of clinical evidence. A new characterization system allowing prediction of sensitivity to TKI for infrequent, often we call them uncommon, EGFR mutations exists now [and] we can classify them into four different classes based on the structural functions, evidence supported preclinically [1]. Real-world data and the mutation database are improving with clinical evidence for EGFR TKI selection against different uncommon EGFR mutations. However, clinical trial or prospective data are still lacking [39]. Therefore, that is an unmet need for us to conduct clinical trials to improve what we currently have. Last, broader uptake of next-generation sequencing for tumor molecular profiling, and the availability of better guidance for physicians, are key obstacles to address going forward so that we can offer precision oncology to our lung cancer patients. I thank you for your attention and I also want to thank Dr Daniel Costa, Dr Eric Nadler, and Dr John Heymach for joining us today.