Dr. Ali Raza Khaki – Loose Regulatory Standards Portend a New Era of Imprecision Oncology – Part 2/2

Jul 1, 2021 | Rising Tide Podcast

Introduction 

On The Rising Tide Podcast, scientists from the independent, nonprofit Center for Genomic Interpretation discuss with leading experts the need to further raise the bar on accuracy and quality in clinical genetics, genomics, and precision medicine. Only through improving accuracy and quality can the promise of precision medicine be realized. CGI’s ELEVATEGENETICS services are made available to health insurers and stakeholders, to help them identify the most accurate and clinically useful genetic and genomic tests, and to help them steer clear of low quality tests that have the potential to lead to patient harm and wasted expenditure.

Transcript 

GARLAPOW: In this episode of The Rising Tide podcast, we dive even deeper with Dr. Ali Khaki into the details of his 2021 peer-reviewed publication entitled, “Loose Regulatory Standards Portend a New Era of Imprecision Oncology.” Dr. Khaki is a Clinical Assistant Professor of Oncology at Stanford University School of Medicine, and is a hematologist/oncologist with board certification in oncology, hematology, and internal medicine. In our previous episode with Dr. Khaki, we discussed the promises and pitfalls of precision medicine, and what happens when what gets approved for therapy is broader than the patients assessed in clinical trials, and what happens when the approval encompasses patients with no benefit to targeted treatment. With the growing appeal of targeted therapy, the rapid acceleration of its application to oncology runs the risk of uncoupling medicine from scientific evidence. This particular episode is geared toward anyone seeking detailed, in-depth perspectives about oncology precision therapies. Following the hype, rather than the evidence, in any treatment can have dire consequences for patients and the entire healthcare ecosystem. We ask what needs to change so that our excitement for precision medicine does not lead to patient care decisions lacking sufficient scientific support. Though precision medicine is advancing cancer care, the progress is gradual rather than meteoric.

KHAKI: There’s a lot of hype and excitement, and while progress is being made, I think that the progress in precision oncology is likely much more incremental than what is perceived. There’s a study from a few years ago with Vinay Prasad and colleagues in “JAMA Oncology” where they estimate that only about eight percent of patients with cancer were eligible for a genome-targeted therapy in 2018. And while this was up from five percent in 2006, this isn’t growing at an exponential rate. It’s growing quite incrementally.

GARLAPOW: Although less than 10 percent of patients with cancer are eligible for targeted therapy, a number that sounds small, the repercussions when precision is removed from the approval of treatments are significant.

KHAKI: The role of the FDA is to approve medications that have been shown to be safe and effective. Unfortunately, with these broad approvals, we raise questions about how confident are we about how effective these medications are for those who have not been included in this study or have not been shown to have benefit in the study. In both these cases, the broad drug approvals contradicted the principles of precision medicine, where the idea is to identify the specific group that would benefit from a therapy and only treat those patients. I thought that the specific place where this really missed the mark is including, you know, cancers like breast and prostate cancer which are much more common but have not been shown to have much benefit with immune checkpoint inhibitors. Patients can be exposed, rather unnecessarily, to toxicity and to cost without little known benefit or with actually trials that have shown that they don’t benefit. To assume that you’re going to have some benefit that doesn’t even exist in a completely different cancer is the second leap of faith.

GARLAPOW: Haste makes waste, and with poorly designed clinical trials lacking stratified randomization that includes appropriate control arms, patients bear the brunt of the uncertainty and outcomes that can develop. At the very least, we can openly converse about such uncertainty. Key improvements to trial design, however, would be even better.

KHAKI: The history of medicine will tell you that most things we do are wrong and so that’s why we have to use empirical evidence to prove ourselves that they’re right. We have many examples in oncology of promising biomarkers that have gone astray. Unfortunately, I think to do things correctly, things have to be done slowly and iteratively. Tools you could use to sort of try to alleviate some of that uncertainty or make it a bit more, you can feel a bit more confident in your conclusions is using things like stratification. So making sure you have well-represented populations of the subtypes that you are most interested in. Randomization is key, and something I think that’s being lost especially in precision medicine is, without randomized clinical trials we have a lot more uncertainty of the benefit to patients. So use stratification with randomization to help make sure you get, you know, all the populations of interest.

GARLAPOW: We need to confront this new uncertainty in clinical science if we stay on this current trajectory of imprecision in oncology.

KHAKI: At the end of the day, you probably can’t include every single cancer. So there has to be some, you know I think I mentioned this last time, there has to be some decision by patients, oncologists, providers, insurance companies, the society that if we want to do tumor agnostic approvals, we are comfortable with the uncertainty that accompanies those approvals.

GARLAPOW: I’m Dr. Megan Garlapow with the Center for Genomic Interpretation, and you’re listening to The Rising Tide podcast, where we learn from experts about improving the accuracy and quality of precision medicine and clinical genetics and genomics. Please note that this podcast does not provide medical advice and is not intended to be a substitute for professional medical advice, diagnosis, or treatment. Always seek the advice of your physician or other qualified health care provider with any questions you may have regarding a medical condition. Additionally, comments of The Rising Tide’s guests are their own and do not necessarily reflect the position of the Center for Genomic Interpretation. 

Today, I am joined again by Dr. Ali Raza Khaki, a medical oncologist in genitourinary cancer at Stanford Cancer Institute and Clinical Assistant Professor at Stanford University. This is the second of a two-part series with Dr. Khaki discussing the promises and pitfalls of precision medicine, potential consequences when FDA approvals extend beyond populations of patients and clinical trials, how to improve selection of patients for therapy, and concepts that all stakeholders should consider for these complex topics. Please note that Dr. Khaki’s views are his own and do not necessarily represent the views of his employer. Additionally, some of what we discuss is speculative in nature and/or describes use of treatment that has not been approved by the FDA. In the first episode with Dr. Khaki, we discussed two cases of FDA approvals for drugs and cancer where the populations of the approval, that is the patients who could end up receiving the drugs prescribed by their clinicians, were broader than the population of patients represented in the clinical trials on which the approvals were based or were cases of the approvals “missing the mark” in the patient population. We contextualized these approvals in the broader environment of precision medicine, and discussed what is happening currently, and what opportunities exist for promising change. In this follow-up, Dr. Khaki and I are going to break open the complexity a bit more. So welcome back Dr. Khaki.

KHAKI: Thanks Megan, it’s nice to be back.

GARLAPOW: That’s great. So what should a general stakeholder in healthcare, which could be anyone from a patient to a medical director at a community oncology practice, know about what is happening in precision medicine in oncology.

KHAKI: Yeah, I’d say there’s a lot of hype and excitement, and while progress is being made, I think that the progress in precision oncology is likely much more incremental than what is perceived. I think I mentioned last time that when we spoke that there’s a study from a few years ago with Vinay Prasad and colleagues in “JAMA Oncology,” where they estimate that only about eight percent of patients with cancer were eligible for a genome-targeted therapy in 2018, and while this is up from five percent in 2006, this isn’t growing at an exponential rate. It’s growing quite incrementally. So at the same time, recently, there have been multiple drug approvals by the FDA, where the FDA indication is significantly broader than the studied population, and that this practice introduces a lot more uncertainty into the efficacy and safety of these improved medications.

GARLAPOW: Okay. So please remind our listeners of some examples of FDA approvals in precision medicine beyond trial population or where they didn’t quite “hit the mark” by providing a general overview of your January 2021 commentary, “Loose Regulatory Standards Portend a New Era of Imprecision Oncology” published in the journal “Cancer Investigation,”  and again some of the motivation you had for writing it.

KHAKI: Yeah, so in this commentary, you know, I reviewed two recent drug approvals by the FDA that I thought the drug approval indication was broader than the studied population or the population that showed a benefit. So the first was the approval of pembrolizumab for tumors with a high tumor mutational burden, the TMB, where the trial population included 10 rare tumor types, but the FDA approval was a tumor agnostic approval that included all cancer types. We discussed last time how I thought that the specific place where this really missed the mark is including, you know, cancers like breast and prostate cancer which are much more common, but have not been shown to have much benefit with immune checkpoint inhibitors like pembrolizumab. And the second was olaparib for metastatic castration-resistant prostate cancer. And here, the FDA label included patients with a genomic alteration in a broad number of genes associated with homologous recombination deficiencies, which was the population that was studied in the Phase 3 PROfound trial, but the trial only showed that the benefit was mostly limited to those with BRCA1 or BRCA2 mutations, which is what’s also been seen in other cancer types. So I felt that in both these cases, the broad drug approvals contradicted the principles of precision medicine where the idea is to identify the specific group that would benefit from a therapy and only treat those patients. So I wanted to highlight that contradiction.

GARLAPOW: So there’s this contradiction, and these patients might not benefit. Why does that matter?

KHAKI: I think this matters because it introduces a lot of uncertainty into the practice of oncology. So, you know, the role of the FDA is to approve medications that have been shown to be safe and effective. Unfortunately, with these broad approvals, we raise questions about how confident are we about how effective these medications are for those who have not been included in the study or have not been shown to have benefit in the study. Therefore, patients can be exposed, rather unnecessarily, to toxicity and to costs without little known benefit, or with actually trials that have shown that they don’t benefit, if you include those patients like in the olaparib example.

GARLAPOW: Okay. I think that that makes sense. So in your commentary, the first approval you discuss is of the immunotherapy drug pembrolizumab, allowing doctors to prescribe the drug to patients with any kind of solid tumor, as long as the tumors have at or above a certain threshold of mutations, which in this case is a tumor mutation burden, or TMB, of 10 or more per megabase of tumor genome. The trial this approval is based on is KEYNOTE-158, which assessed patients with tumors with a mutation burden at or above that threshold of 10 compared to patients with TMB below 10. The patients with high TMB responded to pembro more with an objective response rate of 29 percent, significantly better than non-TMB-high objective response rate of 6.7 percent. There is no difference between these two groups in duration of response or in overall survival. Can you please remind us of the trial population compared to the FDA drug approval population? What happened there?

KHAKI: Yeah, so the trial population included patients with advanced cancers of 10 different types of rare tumor types. I mentioned them all last time. I won’t go through that again, but, you know, those 10 rare tumor types were what was included in the analysis to identify patients with TMB-high, as you mentioned in your summary just now. You know, we talked last time about how even in the population where there’s patients who are TMB-high, the ones that seem to be represented in that TMB-high subpopulation were overrepresented by cancer; they’ve already shown a benefit to immunotherapy. And, importantly, the FDA didn’t restrict the population to just those 10 cancer types or just those that had a benefit, it made a tumor agnostic approval, meaning that it allowed all patients with solid tumors who are TMB-high to be eligible for this medication, and this just leads to a lot more uncertainty about whether this works in all populations.

GARLAPOW: Okay. So one of the things I was looking at before this episode was variability in TMB. And I just wanted to make a note that there’s also a call for harmonization in how labs measure TMB, and as of yet it seems not really well understood. The role of TMB across different tumor types, kind of adding a layer of really important complexity to that tumor agnostic approach when you’re using it with TMB to guide your treatment decisions.

KHAKI: Yeah, and one thing I wanted to add there, you know, as I was even writing the commentary trying to review the literature of TMB and see patients who benefit, you’ll see some studies that mention TMB measures mutations per exome, some measure the mutations per megabase, and you know as an oncologist here I even had to turn to some of my lab medicine colleagues to understand how many megabases are in exomes and try to sort of make sense of how to think of the different ways TMB has been classified over different studies, and so it’s very challenging. I bet even for the people who are in lab medicine or in the diagnostic space, but as a clinician, it’s even more challenging to try to, you know, digest it all.

GARLAPOW: Absolutely, absolutely. And it’s just another complicating layer that we really need to get a handle on. So if 10 mutations per megabase may be inappropriate for prostate cancer and other cancers as you described in your commentary, there was a study that showed that in patients with prostate cancer if they had a TMB of 70 then they responded, but 53 they did not respond, which is quite a bit different than 10 as a cut off. So how can this complexity be navigated to better deliver on some of the promises of precision medicine?

KHAKI: Yeah, this is a challenging question, you know. Unfortunately, I think to do things correctly things have to be done slowly and iteratively. I think that if you’ve identified a biomarker that is prognostic in a retrospective study and it may be predictive, then the follow-up work has to be a prospective study to validate its predictive capabilities. And so in the TMB example, you might identify that the ideal cut point for lung cancer is 10 mutations per megabase, but for prostate cancer it’s 70 or 50 or 5 or you know any number under the spectrum and maybe that TMB does not actually have much separation between those who are going to and not going to respond to pembrolizumab for prostate cancer. So you might need to find a new biomarker all together. So this seems slow, it seems laborious, but this is the only way that we can have confidence that the benefit that we’re seeing is actually what bears out. We have many examples in oncology of promising biomarkers that have gone astray, and so I think that for us to convince ourselves that the biomarkers are good without necessarily doing the rigorous work to prove that, is dangerous.

GARLAPOW: Absolutely. So this next question that I’m about to ask, you just covered a lot of it, but this approval of pembro is tumor agnostic, meaning that it’s for any solid tumor so long as the tumor has a mutation load above that certain threshold or cut off, what trials would need to be done to justify the tumor agnostic approval of pembro or really of any drug in this sort of a setting with this sort of complexity? What are the shortcomings of tumor agnostic approvals, and again, Ali, why should we care?

KHAKI: Yeah, so you know I mentioned sort of how I would think about, you know, going to develop a new biomarker but I think the other part of it is that a couple other things I’d emphasize is one, randomization is key and something I think that’s being lost especially in precision medicine. Without randomized clinical trials, we have a lot more uncertainty of the benefit to patients. So in a tumor agnostic approval, I think that the heterogeneous population that makes it a lot more complex and and I think that tools you could use to sort of try to alleviate some of that uncertainty or you can feel a bit more confident in your conclusions is using things like stratification, so making sure you have well-represented populations of the subtypes that you are most interested in. So use stratification with randomization to help make sure you get, you know, all the populations of interest. So, you know, the KEYNOTE study used stratification with these 10 rare tumor types, but it didn’t randomize, so we’re still left with a lot of uncertainty. At the end of the day, you probably can’t include every single cancer, so there has to be some, and I think I mentioned this last time, there has to be some decision by patients, oncologists, providers, insurance companies, the society, that if we want to do tumor agnostic approvals, we are comfortable with the uncertainty that accompanies those approvals. Cause at the end of the day, if your cancer was not well represented in the trial that showed efficacy, then you are taking a leap of faith to trust that this therapy will work for you. You know, everything that makes sense in biology doesn’t actually bear out. In fact, most things don’t bear out. So, I’ll have patients tell me, “Oh! Like that makes sense so I don’t know why you have to do this over again,” because the history of medicine will tell you that most things we do are wrong and so that’s why we have to use empirical evidence to prove ourselves that they’re right.

GARLAPOW: Oh absolutely. There should be a journal just called, “Most Things That We Do Are Wrong.” Just be publications based on that. So as a reminder for our listeners, you also discuss olaparib, a PARP inhibitor, in metastatic castration-resistant prostate cancer based on results from the Phase 3 PROfound trial, randomizing patients with metastatic castration-resistant prostate cancer and somatic or germline mutations in genes involved in DNA repair. When mutations occur in DNA repair genes, the tumors can be particularly susceptible to PARP inhibition such as with olaparib. The most well-known mutations occur in BRCA1 and BRCA2, but there are many other genes involved in DNA repair. Patients were randomized to olaparib versus physician’s choice of enzalutamide or abiraterone. Patients who got the PARP inhibitor experienced imaging-based progression-free survival of 5.8 months versus 3.5 months in the control arm with the physician’s choice therapy, with a hazard ratio for progression to death of 0.49. A pre-specified analysis showed patients with mutations in BRCA1 and/or BRCA2 had the most benefit from olaparib compared to the control arm, whereas patients with other other mutations, genes like ATM, CDK12, CHEK2, had a null response, so appeared similar in the response to the control arm. And, one group of patients, the unlucky patients with mutations in PPP2R2A, experienced a worse outcome on olaparib versus physician’s choice. In your commentary, you described this last group of patients with PPP2R2A pathogenic variants as particularly concerning stating, “That the sub-optimal control arm biased results in favor of olaparib.” What made the control group sub-optimal and why is this group so concerning?

KHAKI: Yeah, actually thanks for bringing this up; we didn’t talk about this last time. You know, for patients to be enrolled in this trial, they had already progressed on enzalutamide or abiraterone, so enzalutamide is a non-steroidal anti-androgen or abiraterone is the CYP17 inhibitor, these are anti-testosterone level therapies for castration-resistant prostate cancer. So for patients to then again receive one of those two medications after having progressed on those medications, is not what we would do in the practice of medicine. In fact, 20 percent of patients and about everyone in all the treatment arms had already received both of those medications. So to give those patients again abiraterone or enzalutamide when they’ve progressed on both of them, that I would consider malpractice. In addition, there are prior studies that have shown that the response rates of enzalutamide after abiraterone can be decent, can be about 35 percent, but the response rates of abiraterone after enzalutamide is less than five percent. So again, for those patients who were on enzalutamide, progressed, and then were randomized to physician’s choice and received enzalutamide or abiraterone, that would be problematic. So when you consider all this and then think that a patient with a PPP2R2A mutation, despite this poor control arm, still did worse with olaparib, that’s a huge issue, right, because if patients were not on this trial they would be getting docetaxel or some chemotherapy or some other medication that would actually cause a significant benefit.

GARLAPOW: Yeah, that gives me chills. Thank you. So mutations and what other DNA repair genes might predict response to treatment with PARP inhibitors, especially in light of the results from the pre-specified analysis that showed null to inferior response in olaparib in non-BRCA1/2 DNA repair genes? 

KHAKI: Yeah, the other genes that were of interest that were studied where, you know, most of the genes that we think of were included in this trial, you know, ATM, CDK12, CHEK1, and CHEK2. So I think that in the study we didn’t see a benefit with these other mutations, so I think that there was good hypotheses to think that they would benefit or they would respond to PARP inhibitors but they didn’t in the study. I think there is some interesting new science that’s being done to think of other ways of defining BRCA-ness or HRD signatures using functional assays. There’s a lot of good science being done to develop new biomarkers, and I think that’s all great we need that science and hopefully as we identify new biomarkers we can follow that up with randomized trials, with properly-selected control arms so we can confirm that they have efficacy in the real world. 

GARLAPOW: Okay, great. So when we look at the prescribing information, the packet for olaparib, the FDA approval, metastatic castration-resistant prostate cancer is unique on the label for one thing. Even determining whether a patient with MCRPC is eligible for olaparib relies on three different companion diagnostics. In your opinion, should clinicians be concerned about three different companion diagnostics being used for olaparib in mCRPC, and what are clinician thoughts on the three different companion diagnostics also relying on three different tissues, more or less, blood and things like that, and germline versus somatic to assess the genetics and the genomics in this tumor type.

KHAKI: You know, I have to be honest with you, I think that the whole diagnostic arena is very confusing right now. In addition to having multiple different types of tissue sources (blood, plasma, tumor), you have germline and somatic, you have some platforms that don’t report the germline, the uncertainty of that’s going to be reported out to you, and then you have multiple platforms, you have Foundation, you have Tempus, you have institutional profiles like UW OncoPlex and MSK-Impact. I think it’s very challenging for even some specialized oncologists to digest and keep track of all of the changes, and then I think about those in community oncology or those who may not be around people who are experts in developing these panels and and I don’t know how anyone can keep track of all these things. I think that the science is great, it’s very exciting, but to expect an oncologist to keep track of this in addition to keeping track of the upcoming therapeutics and everything else is simply unsustainable. 

GARLAPOW: I absolutely agree, and I think that for clinicians they’re not necessarily aware that seemingly comparable tests across laboratories can vary wildly in quality. It can be really alarming how much variability there is in quality across laboratories and, you know, one lab might be incredibly well-versed in cardiomyopathy genetic/genomic tests but then they start a cancer panel and they don’t have the same level of excellence but that can kind of get lost in the messaging.

KHAKI: Yeah, and beyond that you’ll have issues. I mean, as an oncologist you’ll just see how nice a print job is, like the science on the back-end there’s issues with, and then on the front-end also like how clear is the report that you’re providing? You know, the interpretability of the report is a whole other issue that also plays into this as well. At most academic centers now, they have molecular tumor boards now because of the complexity with all this to help sort of unpack this. You need experts to help sort of digest all this information.

GARLAPOW: Absolutely, absolutely. The FDA approved use of olaparib in mCRPC is unique in that multiple genes beyond BRCA1/2 are used for the patient selection, and we’ve discussed in our prior episode and now how those non-BRCA1/2 patients had a null or inferior response,  yet all the other types of tumors on the label use only BRCA1 and/or BRCA2. What effect could this potentially have on how clinicians are using olaparib in other types of tumors?

KHAKI: Yeah, so I think that this could have the effect that other tumor types (people who are treating other cancer types) might also think that ATM or CDK12 or any number of the genes that were included in the olaparib approval could, if they are present in breast cancer or pancreas cancer, they might also try to see if those patients could benefit from olaparib. Then there’s two errors there, right? The one error is that olaparib should have never been approved for even prostate cancer for those populations because they never benefit, and then to assume that you’re going to have some benefit that doesn’t even exist in a completely different cancer is the second leap of faith. The more leaps of faith we’re taking, the more and more uncertainty we’re introducing into how we treat our patients.

GARLAPOW: So what do you make of multiple DNA repair genes guiding treatment in mCRPC? How do these, I’m just going to call them biomarkers here, how do they need to be assessed to understand the potential efficacy in other types of cancer? How could we potentially use ATM in pancreatic cancer? 

KHAKI: Yeah. I think, again, you have to prospectively validate that these work for these different cancers. Again, we haven’t really showed that olaparib works for these other mutations in prostate cancer, so if I had pancreas cancer, if I had breast cancer, I’d really want to see a study in that cancer that shows that these mutations respond to olaparib or any of the PARP inhibitors. 

GARLAPOW: So I see sometimes language saying that off-label use is permitted by a certain payer if there’s peer-reviewed literature supporting a trial. But “peer-reviewed literature,” that could be something like a case study about somebody in another country and the patient had whatever variant and then they were given this drug and then the patient responded. Is that enough for you? 

KHAKI: I mean, when I hear that I’m thinking about Vitamin C. We as oncologists need to be better than that. I think if we start settling for that then what makes us any different than a snake oil salesman, right? I think that the promise of western medicine is that we are being rigorous in our development of modern therapeutics but also the assessment of the efficacy of them, and so we really need to focus on making sure we develop high quality evidence that shows the benefit in the populations that we’re studying. I might sound like a broken record here, but I can’t say it enough. We have to show benefit in the population that’s being studied.

GARLAPOW: Absolutely, 100 percent agree. So as our classification of these mutations or the variants, it evolves you know. Sometimes you’ll get a report back that says it’s a variant of unknown significance, which means that the evidence out there in the literature isn’t sufficient to say whether it’s likely pathogenic or pathogenic, and that can be upgraded to likely pathogenic or pathogenic in a gene such as ATM. As this evolves in these genes that guide treatment, how do trials, laboratory genetics and genomics, and assessment need to evolve to deliver the right treatment to the patient? 

KHAKI: That is a great question. I’m going to mostly talk about it from the trial perspective because that’s how I think about these things. I think that what we can do as oncologists or in clinical trials is we need to find ways to make our clinical trials more agile. You know, it shouldn’t take months and months to make protocol changes. I think that there’s a fine balance between making sure you’re being cautious in your changes to a protocol to preserve the quality of the science, but to not have so much bureaucracy that conducting a clinical trial becomes a barrier to the practice of medicine. So I think that there’s work that we can do there. I think there’s work we can also do to sort of expand eligibility criteria for patients in clinical trials. I think that there’s been much that’s been written recently about representation of underrepresented minorities, you know, other populations, so I think there’s a lot of more work we can do there. You know, less than 10 percent of all patients end up on a clinical trial, and a lot of that’s because we create these unnecessary barriers, we have too stringent criteria. I know that’s not directly about these evolving mutations, but I think that being broader in how we think about clinical trials is the first step and trying to be agile if we find, for example, a new pathogenic mutation that we’re confident is pathogenic and if there’s a trial currently underway, then maybe we can amend that trial to also include those patients, trying to be a bit more agile and finding ways so we can use new tools to sort of allow for that to happen. I think the next thing I’m going to say is a bit more risky, and I say that because it’s not the highest quality evidence but I think that there is also potential to develop better real world evidence platforms to sort of help capture some of this retrospectively or in real time. I think ASCO CancerLinQ has done a great job to sort of build a great platform to help capture structured and some unstructured data from the electronic health record that we can hopefully use to sort of mine and answer these questions after the fact. There’s other companies who have been doing this work too, but I think that proceeding in that path needs to be done with much caution because retrospective evidence is not the same as prospective evidence. So while I think that there’s a lot of value in developing real world evidence platforms, in fact a lot of my research is in doing so, I think that we have to do so with caution.

GARLAPOW: Absolutely. I would like to add one additional point. I think that in addition to increasing the agility of clinical trials, we need to improve the patient selection not just in diversity but for things like somatic variants. They are really susceptible to false positives. So you end up enrolling a bunch of patients who you think have the pathogenic variant but don’t actually have it, your trial is less likely to perform in a way that’s significantly different from a control arm or something like that. 

KHAKI: Is that because of mono-allelic versus bi-allelic loss, or what’s the reason for that? 

GARLAPOW: Oh, there are a number of reasons. I could do a two-hour podcast episode on the contributing factors to false positive somatic variants in particular, even more so than in germline variants, but yeah it’s just another interesting point that selection of patients can also improve markedly there and then subsequently, treatment of patients if we can improve that then we can hopefully in the actual treatment of patients later on select the best patients for the therapy. So how much patient diversion from trials or more appropriate therapeutics, things like that, do you think is happening in oncology as a result of off-label use of targeted therapies and of on-label use when the approval is broader than the trial population? 

KHAKI: Yeah. You know, I actually don’t know if I’ve seen any good studies that have addressed this. I thought you might ask me this question, and this is something I worry a lot about but I haven’t seen good evidence of how much this is happening, but I’ll give you a couple examples of how I think or where I think this is happening. I mentioned before that less than 10 percent of patients enroll in clinical trials, and so there’s much more work that’s needed to increase access and so when you have other alternatives, that’s one more barrier to enrolling people in clinical trials. Other limitations that make it difficult is the restrictive inclusion/exclusion criteria I mentioned before, and then also the distance from patients. If you’re a patient who lives in rural America and and you have a choice between getting a drug close to home or traveling to a cancer center for a trial, many are going to pick the drug that’s close to home, just because, why, if I’m in my last months or years of life, why do I want to be traveling two, four, six hours or living in a different city just to be on a clinical trial? Not all patients will make that decision, so I don’t know how much diversity is happening but I think that the solution is to try to find ways to make trials more available close to home, and then not to just settle on off-label drugs. And then I think the other thing that’s important is that when it comes to the selection of these versus other approved therapies is that we need to be careful in how we are presenting these new therapies. We talked a lot last time about the hype in precision oncology and so if patients perceive these therapies as being targeted that they found a mutation and this is targeting that mutation, they’re going to think that this is going to work much more likely than an untargeted chemotherapy when that may not be true. There’s good evidence already that patients already can misinterpret what is told to them by the oncologist. There’s work that’s been done that shows that when oncologists are talking about palliative chemotherapy, patients will think that they’re going to be getting cured with chemotherapy, and so there’s a lot that can be lost in translation. And then when you hear all about this new targeted modern technology, I think many of us have seen the success of so much technology, I carry a computer in my pocket now, so technology has in general worked for us but unfortunately the same success of progress we’ve made with other technologies have not been as quick to be realized within biomedicine.That’s kind of a rambly answer. 

GARLAPOW: I love it. So in your commentary, you write. “Overall, the goal of precision oncology is to use advanced laboratory methods and biomarkers to better identify patients likely to benefit from a particular therapy. If FDA drug approvals deviate from the populations most likely to benefit precision oncology is likely to shepherd in a new era of imprecision.” Where are we? What is this going to do to patients? What is the long-term outlook of this if we don’t get it under control? And again, I asked this last time I’m asking it again, do doctors understand this?

KHAKI: Yeah. One thing I didn’t mention last time, I think the lab science, at least my perception of it, now I’m having doubts after our conversation today, has made tremendous strides and is much more precise than it has been historically, but oftentimes in precision oncology/precision medicine we’re oftentimes limited more by our therapeutic options than by the lab science. We understand the human genome, but we don’t actually have drugs to treat everything and we probably don’t understand the human genome as well as we think we do. But I think in general, we’re making progress in oncology. As I mentioned before, I think that progress is iterative and not necessarily at some sort of precipice of taking off, and then I think that we’re oftentimes limited by our therapeutic options. That’s a big thing, and so I think that the solution there is not to get lost in the science to convince ourselves that the new drugs are better than the old drugs, without us proving that with rigorous evidence that supports those conclusions. Even if we find mutations that we’re convinced have a role in a pathway, that doesn’t mean that if we can’t drug that with the drug that we think works that doesn’t mean that our pathway is wrong, it could be that our drug is wrong, but so either way it could be the pathways wrong, it could be the drugs wrong, whatever it is we have to produce rigorous evidence that shows that supports our conclusions, supports our hypothesis, and shows that we’re actually benefiting patients. Do doctors understand this? I think that probably the treating physicians understand it more than the lab physicians because the more you’re seeing the patient the more you see sort of what patients are experiencing. 

GARLAPOW: So I think you answered this, but I’m going to ask it again because it’s The Rising Tide podcast. How can we raise the tide?

KHAKI: Yeah, so I’m going to say the same three things I said last time we spoke which is: number one, we have to be precise in the conduct of evidence production, which I think precision oncology should start with randomization and then using other clinical trials tools to help optimize your population, things like stratification that I mentioned before. Number two, train future physicians to have critical appraisal skills of the medical literature so they’re equipped to digest the medical literature and counsel their patients appropriately. And number three, is remove that heavy hand of private enterprise that exists in biomedicine today. 

GARLAPOW: Great, Ali. This has been such a wonderful conversation. I feel like I could talk to you for like at least four more hours about this, a lot of back and forth, but this concludes the second of our two-part series with Dr. Ali Raza Khaki from Stanford University and Stanford Cancer Institute talking about precision medicine, FDA approvals, and how to raise the tide. Thank you, Ali.

KHAKI: Thanks, Meg. I’m really excited for this podcast, and I wish you guys all the best as you grow this forward. 

GARLAPOW: If you loved this episode of The Rising Tide podcast, please support us on Patreon and select us on Amazon Smile. The Center for Genomic Interpretation can help your organization identify high quality clinical genetics and genomics partners, or can help a laboratory identify areas for quality improvement. Through ELEVATEGENETICS, we engage with laboratories, encourage genetic and genomic test validity, and assess test efficacy. Find us at clarifygenetics.org and genomicinterpretation.org. 

For our listeners following along on YouTube, remember to hit subscribe and to activate alerts for when we post new episodes. You don’t want to miss any of these important conversations with leading experts. Together, we can raise the tide.

Narrated by Dr. Megan Garlapow

Produced and edited by Kathryn Mraz, Hunter Giles, and Brynlee Buhler