Transcript
MIC CAVZAZZINI: Welcome to Pomegranate Health 2025. This is a podcast about the culture of medicine and I’m Mic Cavazzini for the Royal Australasian College of Physicians.
Over the next two episodes you’re going to hear from Professor David Thomas, Director of the Centre for Molecular Oncology at the University of New South Wales and founder of a not-for-profit company called OMICO. I met with him in Sydney, to learn about Omico’s program to connect patients with overlooked cancers to new molecularly-targeted therapies.
Professor Thomas wanted to emphasise just how dramatically genomics has changed the precision of cancer medicine and the pace of development. He’s not the only commentator to have noted that the last paradigm shift in the understanding of cancer came with the light microscope. New concepts in cell pathology transformed what had, since ancient Egyptian times, been largely an anatomical and surgical model. But this was far from an overnight revolution.
The microscope made its debut in the public consciousness in 1665 with the publication of Robert Hooke’s ‘Micrographia’, containing impeccably detailed engravings of a fly’s eyes among other wonders. While this was a best-seller, it would be another two hundred years before microscopy was taken seriously by medical investigators.
In 1830, London surgeon Everard Home was the first person to publish drawings of tumours viewed under a microscope. Though of poor quality, he recognised that the lymph nodes near the primary tumour could harbor cancerous cells and that metastasis might occur through the blood vessels.
Berlin anatomist Johannes Müller followed in 1838 with a book containing a hundred careful drawings from his observations suggesting that cancers were derived from normal cells and could be classified by tissue of origin. This was supported by the work of his student, Theodor Schwann who, a year later, argued that the cell theory of proliferation was true for animal tissues not just plants.
Over the next few decades other German and French microscopists collated evidence of the subcellular differences between benign and malignant tumours. And, in 1882, Walther Flemming published his historic drawings of dividing cells, their chromosomes being guided to new nuclei along nanometre-scale filaments. Flemming’s images had a level of contrast not available to earlier investigators. Chemical engineers had found a way to extract the orange-red eosin compound from coal tar, and this dye was actually used by Vincent van Gogh to create some of his vibrant paints.
But in 1877, a Russian chemist called Nicolaus Wissozky described a protocol for staining blood cells with eosin on top of the commonly used plant dye hematoxylin. These dramatic improvements in imaging tissues catalysed the need for a new kind of medical expert, the pathologist. Cancer by cancer, tissue by tissue, they made countless advances over the next 150 years that would improve patient management. Think of all those histopathology-derived risk stratification tools, such as the Breslow depth and Clark Level for staging of melanoma. And consider that the H&E protocol is still used for 80 percent of the staining work performed by path labs today.
But that pool of foundational knowledge has been dwarfed the last three decades by the output from a genome-based examination of cancer. And as Professor David Thomas will explain, this has been reflected in the improvement in patient outcomes resulting from rationally-designed therapies targeted at known molecules.
DAVID THOMAS: So, when I graduated as a doctor in 1988 the paradigm that we were using in oncology was based around understanding of cancer as an anatomic disease. That is a breast tumour or a lung tumour, based on anatomy and also, based around the pathologic patterns we observed under the light microscope. So, we would have a non small cell lung cancer, we would have a small cell lung cancer, literally the appearance of cells under the microscope.
But the drugs we were using for the most part in that time were developed empirically. And what I mean by that is drugs had certain effects on dividing cells in culture, it was reasoned that cancers were comprised of dividing cells, and therefore these drugs might have an effect on cancer cells. Now the trick, of course, is lots of normal cells divide and, as a consequence, it took trials to demonstrate whether a drug had what's called a therapeutic window—a gap between its effect upon cancerous cells whilst having a lesser effect on normal cells.
That was variably successful with chaemotherapies. It's not to say that there weren't enormous advances from chaemotherapy. Cancer survival in New South Wales is now 70 percent five year survival, whereas it used to be in the 50s, only two decades ago, and it used to be in the 20s, back in the 1960s and 70s. All of that is because of incremental models of clinical research, where we've empirically made enormous advances in a whole range of areas, including prevention, which we shouldn't forget. But there wasn’t a concept about the cancer at that time that led you to choose a drug. And it wasn't possible to design a drug to say, here's a target, let's design a molecule that binds to and inactivates that target.
MIC CAVAZZINI: What was the process of fishing for oncogenes like before the human genome was fully mapped in 2003?
DAVID THOMAS: Well, that's right. So, molecular biology and cellular biology over the preceding 20 or 30 years, had identified oncogenes. We started to map the first tumour-suppressors around the beginning of the 1980s we had the first proto-oncogene in HRAS, identified and characterized in the early 1980s and we started to identify more of these proteins that drove growth of cancers. And it was painstaking, you know, it might take, you know, an entire PhD to characterize a novel oncogene. It would take 10 or 15 years to understand this biology. And we had techniques, genomic techniques like PCR and positional cloning, and techniques which were very low throughput at that point. So that explains the relatively, low number of targets that we were introducing into clinical practice.
We started to use antibodies to do immunohistochemical stains for particular proteins in the 1990s. We were certainly using oestrogen receptor and progesterone receptor staining to determine whether we were dealing with a receptor-positive breast cancer, for example. We also had drugs like tamoxifen that came along. That's effectively molecular targeted treatment, but not designed in the way that we're talking about now.
MIC CAVAZZINI: In my little history, I came across the 1908 Nobel laureate in Medicine, Paul Erlich. This German physician is known as the “father of immunology,” but his work had repercussions across medicine. In experimenting with the organic dye methylene blue to stain bacteria, he also, noticed it’s cytotoxic effect. This led him to conceive of chaemotherapy, and in playing around with arsenic, Erlich actually discovered the world’s first antibiotic. But it had some nasty side-effects, and Erlich fantasised about a therapeutic agent which would only kill pathogens or cancer cells but leave healthy tissue untouched. He termed this a Zauberkugel, or magic bullet. We’d call it a silver bullet today. But he recognised the limitations of existing technology. So, I’ll ask you to read the prescient words he gave in his Nobel Prize lecture. I’ve edited this down a little.
DAVID THOMAS: You want me just to read this out?
MIC CAVAZZINI: Yeah, have you ever seen this before?
DAVID THOMAS: No, no.
MIC CAVAZZINI: I’m after your impressions.
DAVID THOMAS: Okay, so “I am inclined to think that the limit of what the microscope could do… for us is now approaching. And that for a further penetration into the important, all-governing problem of cell life, even the most refined optical aids will be of no use to us. Now, at this moment, the time has come to penetrate into the most subtle chemism of cell life, and to break down the concept of the cell as a unit into that of a great number of individual specific partial functions.
But since…the configuration of chemical structures lies beyond the limits of the eye’s perception, we shall have to find other methods of investigation for this. This approach is not only of great importance for a real understanding of the life processes, but also the basis for a truly rational use of medicinal substances.” I mean, this is perfect example of logic and science asking questions for which the tools had not yet developed to answer it.
MIC CAVAZZINI: This was 1908. Well before Watson and Crick.
DAVID THOMAS: Well, he doesn’t talk about DNA, of course, and he doesn't talk it doesn't refer to Mendel and any concept of genetics here. The idea of somatic genetics hadn't been developed. So, it is remarkable how much we've advanced. What you can say about this quote, the idea that one replaces completely something so that it’s no longer of any use, I think that’s certainly not true. I think we're never going to not consider anatomy relevant when we're curing cancer, until surgery is no longer important.
The morphology under the microscope doesn't tell us what makes the cancer tick, but it makes it recognizable as a certain class of cancer which has properties like prognosis, curability, response to chemotherapy and so forth. So, it's very useful, and still remains useful today. So, I think what he’s describing here is that by 1908 it was obvious to scientists like Ehrlich that we had to have a molecular understanding of a disease if we were to hope to be able to rationally tackle them. But as you say, that didn't really come to fruition until now.
MIC CAVAZZINI: Until imatinib. It might have been the first “Zauberkugel.”
DAVID THOMAS: Yeah. The first really successful designer drug was the drug imatinib methylate. And that was based around a target that had been identified in 1963 by Peter Nowell and David Hungerford. It was the Philadelphia chromosome, which you could observe as a mutation under the light microscope because of cytogenetics. They knew it was a mutation seen in chronic myeloid leukemia. So, the folk at Novartis started to develop this drug imatinib, which would inhibit the ABL kinase, and that was the first designer drug. That drug, which is still widely used today for chronic myeloid leukemia, was so effective that there are still patients who are in remission having been treated with imatinib 10 years, 15,20, years later. That simply wasn't imaginable in the 1990s.
MIC CAVAZZINI: Where the lead compound for imatinib was a small pyrimidine molecule that had been sitting on a shelf at Novartis, the first precision oncotherapy designed from scratch was trastuzumab. This is a monoclonal antibody raised against an extracellular domain of the protein HER2. That’s a human epidermal growth factor receptor which we know to be associated with one in five breast cancers.
As Professor Thomas has already stated, conducting such research before the advent of high throughput genomics was a painstaking process. The first sequences for growth factor receptors were cloned in the early 1980s one gene at a time. Expressed in plated cell lines they were found to be tyrosine kinases that would convert an extracellular hormone signal into a nuclear response.
It was then observed that some of the human sequences had homology with a gene carried by a virus that caused fibrosarcoma in birds. In fact, four members of this gene family were identified, but number 2 came to the fore by studying chemically-induced neuroblastomas in rats. It was only after this that HER2 overexpression was observed in the wild, in a human mammary carcinoma. Dozens of copies of the gene lead to a 40–100 fold increase in expression of the HER2 protein.
By 1987, enough screening of breast cancer patients had been done to establish that HER2 amplification was a significant predictor of early relapse and short survival. The native ligand for HER2 still isn’t known, but the kinase activity and cell proliferation is switched on by dimerization with other HER subtypes. With this understanding, so began the process to cultivate monoclonal antibodies that would block this dimerization, and this resulted in the drug known as trastuzumab.
Trastuzumab went into trials in 1992 and gained full approval from the US Food and Drugs Administration six years later for patients with advanced disease. Large randomized controlled trials showed that trastuzumab on top of adjuvant chaemotherapy resulted in a 5 year survival rate of 85 percent, compared to 75 percent for standard care alone. Such outcomes have been supported by real world observations since and equally dramatic outcomes for patients with early stage breast cancer as well.
Since 2010, two more monoclonal antibodies have been approved against HER2. More recently still, trastuzumab has been used not just for its own therapeutic merit, but as a vehicle to deliver cytotoxic payloads in a very targeted way. As one oncologist from Washington told the Financial Times, “Originally it was simply bad news if you had HER2 expression…Now the pharmaceutical industry has developed such effective treatments that you want to be HER2-positive.” I asked Professor Thomas to reflect on how trastuzumab has paved the way for different applications in precision oncotherapy.
DAVID THOMAS: Well, trastuzumab is probably the second, really, rationally-designed therapeutic, and probably the first blockbuster. It's 30 years since trastuzumab was developed and brought into the clinic, and yet, we're still modifying with these antibody-drug conjugates where trastuzumab is now bound to a cytotoxic. So, the targeting actually brings the cytotoxic specifically to the cancer cell—as opposed to chaemotherapy, which you pour into the body and you hope it reaches the cells of interest at sufficient concentration. This really allows us to deliver those payloads, those chaemotherapies, right to where the cancer cell was, probably at concentrations are unachievable if you were delivering it systemically. So, one of the elements here, that's, I think, important, is it's not as if you find a drug for a target and that's the end of it.
MIC CAVAZZINI: An example of this is in 2013, a conjugate of the anti-HER2 antibody and a cytotoxic drug called emtansine or DM-1 was described as the first targeted chaemotherapy. The conjugate undergoes receptor-mediated internalization into cells where DM1 blocks microtubule formation and therefore cell division. I believe it’s indicated only disease that’s recurred after trastuzumab or taxane. Is that because it’s too expensive or too toxic or…?
DAVID THOMAS: Well, because to this point in time, the clinical trials have only established that it has superior efficacy against treatments required after that point, not before that. Typically, when we do clinical trials, we start off with the most advanced cancer patients who perhaps lack any options that would be regarded as standard of care. But as the drug proves effective, typically what happens is the drug will be moved earlier and earlier in the cancer journey.
So that's another lesson of trastuzumab. It got used in the advanced incurable setting in the 1990s but it wasn't long before it started to be introduced into the adjuvant and neoadjuvant treatment of HER2 positive breast cancer. And in that setting, it can not only prolong life, but it could actually increase cure rates, and that's where it's used today as much as it is in the advanced setting.
MIC CAVAZZINI: Okay, so 2013 is still very recent in terms of where the trials will be.
DAVID THOMAS: We've become extraordinarily demanding when we say, you know, 150 years ago, we used a fundamental technology called the light microscope, 70 years ago we had empiric therapies, and something that is 11 years old is somehow sort of taking too long to get to.
MIC CAVAZZINI: Then in 2022 the FDA listed a conjugate of trastuzumab and deruxtecan for “HER2-Low” breast cancer, as well as broader indications we’ll hear about later. So, cancers that don’t have a ton of HER2 may not respond so well to the monoclonal antibody alone. But deruxtecan inhibits enzymes involved in DNA coiling and ligation so the drug basically leads to a junkpile of DNA strands that triggers cell death.
DAVID THOMAS: Yes. This antibody-drug conjugate era, which is probably only three or four years old in a serious way—we're starting to observe some very odd effects from that. The first is, it's working on breast cancers that don't have HER2 overexpression. It's even working on cancers where we can't detect HER2 by immunohistochemistry, and we think that's because immunohistochemistry as a screening tool is not sensitive enough to pick up the presence of HER2 on the surface of breast cancer cells. But there is enough of HER2 on the breast cancer cell for an ADC, an antibody drug conjugate, to bind to and still kill the cells.
MIC CAVAZZINI: And as you said, the dose is enough, even with that lower expression of HER2.
DAVID THOMAS: That’s right, because the mechanism of action isn't simply binding to, and effectively hindering the functions of HER2, it's actually acting as a targeting agent for cytotoxics. But think about the extraordinary implications of that. The first implication of it is, immunohistochemistry may not be sensitive enough for the use of modern, future agents. We may have to use things like mass spectroscopy to be able to identify levels of protein targets which may be below the detection of immunohistochemistry.
Secondly, it expands the market. So instead of just dealing with 15% of the breast cancer population, now we could potentially be targeting 20 or 30 or 40% with an advance that was made in the 1990s. The breakthroughs we're making at the moment, they're not finished, that's merely the beginning.
It could well be—here’s an idea that's disturbing—it could well be that some of the drugs that we're developing now we've gone into clinical trials based around a dosing, a schedule, which may be sub-optimal. We may need to go back to old-fashioned pharmacology to work out how to better use those drugs to eke out the maximum benefit from them. And it’s not just trastuzumab-deruxtecan and emtansine, it's also radioligand therapies. You know, think about radiotherapy as a non-targeted anti-cancer therapy but it's so widely used and so critical for the management of patients. Now we're beginning to use molecules deliver the payload of radiotherapy to where the cancer cells are. And that's a very interesting second example of the way in which we will keep going back and taking good ideas from the past and rejuvenating them in that way, which is very exciting.
MIC CAVAZZINI: Yeah, and those unknowns are still there. You know, this, this particular combination might work well, it might have some side effects. There's still a lot of empirical work to do.
DAVID THOMAS: Oh yeah. To me, that’s a very important point you just made. I think we can be led into the perception that everything is now advancing on a rational basis and there's no question that is generally true. But there's still space for the serendipitous observation and we shouldn't be so arrogant as to think, “Well, we know this is only going to work in this space, so we’ll only test in this space”. And I think that combination of what we've learned about being open minded in the use of chaemotherapies and using statistics to show benefit—because the effects are clearly real, even if we don't understand them—as well as rationally-designed therapies, those two things, probably there needs to be a portfolio approach where maybe 80% is rationally designed but we should also be testing these agents use outside of where we might expect them.
MIC CAVAZZINI: That’s a great analogy the risk portfolio. Is it worth just mentioning, not discussing at depth, resistance and whether targeted therapies have any advantages in resistance, or should we skip that?
DAVID THOMAS: No, I think it's an important point, because almost all of our rationally-designed and molecularly-targeted therapies, and indeed, chemotherapies—with rare exceptions, ultimately, resistance does emerge in most cancers, and resistance could be argued, therefore, to be one of the great problems of the age. Now, second generation, third generation agents are often designed to prevent resistance or to target the mutations which are resistant to the first drug. So, if the predominant mechanism of resistance to first generation EGFR inhibitors in lung cancer is the emergence of resistant mutations in EGFR that aren't hit by that first round of drugs, then drugs which target T790 M, like osimertinib become effective. Now, if you used osimertinib in the upfront setting, obviously T7090M is not an exit strategy for the cancer.
MIC CAVAZZINI: So, the equation hasn't changed, except in the sense that the response rate is higher and there are fewer clones that are going to survive the first round.
DAVID THOMAS: Yes, and the science of resistance is really poorly understood, so we don't know—it looks like at least some resistance does emerge from pre-existing clones. But it's also possible that there are de novo mutations that weren't there when you introduced the first drug, that are generated through increased plasticity of cancer cells under stress, thereby increasing the probability of resistance—and that process itself, could be druggable.
I guess we could consider PARP inhibitors in that light, right? So, PARP inhibitors directly target a particular mode of genomic instability, a mechanism that generates genetic diversity. So, maybe we should be thinking about combining agents that target mutability itself with our tyrosine kinase inhibitors. We hit the target, but we also somehow block the emergence of resistance mutations that render the first drug less effective. And that's an example of an idea which I think needs to be tested going forward. I really want to reinforce for the audience that there we're just at the beginning of a new era. We're having a huge number of new drugs coming along, which we will have to take the same painstaking approach.
MIC CAVAZZINI: We’ve heard some examples of precision therapies against breast cancer and leukemia. Treatment options for lung cancer have also multiplied in the last thirty years, from a few chaemotherapies with short-lived benefit, to eleven different targeted treatments that evoke deep responses in the majority of patients. Indeed, for the most common cancers, Professor Thomas has estimated that there will be forty rationally-designed and officially registered molecules by the end of the decade.
But it’s a different story for rare cancers, which are defined by having an incidence rate below 6 cases per 100,000 population. The financing model for drug development means that such diseases are often overlooked by pharma companies, given the small pool of potential buyers relative to the massive costs and risks of investment. Even at the level of basic research, academics whose grant funding depends on citation counts in high-impact journals can be dissuaded from investigating rare cancers.
In the next episode we’ll evaluate the cost-risk paradigm for drug development, but first, let’s hear about the three-pronged strategy that Omico is offering patients with rare cancers right now. For starters, any patient referred to the program with advanced cancer will have their cancer genome profiled.
Second, in the ideal outcome, positive hits are found that will match the patient to so-called pan-cancer therapies. Finally, for patients with less certain targets identified, Professor Thomas says that clinical trials should be the standard of care. All this is orchestrated though an entity called PrOSPeCT, which stands for Precision Oncology Screening Platform Enabling Clinical Trials.
DAVID THOMAS: So OMICO is the organization we established in 2018 to be the focal point for federal government investment in a national precision oncology program. Precision oncology is a high-tech area of medicine. High-tech areas of medicine are deployed initially, at least, through hospitals. Hospitals are funded by state and territory governments. So, there were two ways we could deploy a national precision oncology program. One of them was to go through each state and territory government, and the other was to create a company which links together, in perpetuity, the cancer centres within those states and territories, so that all states and territories were represented in our program. That was a necessary precondition for us receiving the federal government MRFF money to start off, what we called then, the Molecular Screening and Therapeutic study, or MOST study, that was in 2018.
So, 8300 patients later, we set up a second program called PrOSPeCT. And PrOSPeCT isn't a particular study. It's taking the view that we are dealing with a complex ecosystem involving multiple stakeholders; patients, of course, clinicians, cancer centres, governments state and federal, and also, necessarily industry, who produce all the drugs that make a difference. So, PrOSPeCT is the ecosystem, but the protocol by which we enrol patients into our program give patients access to the molecular profiling, link them to treatments through trials, that's called CaSP for Cancer Screening Protocol.
So, we've now got to the 15,000 odd patient mark. All those patients have incurable cancer. For the most part, in our early programs, at the end stage of incurable disease. Now we're taking patients who have newly-diagnosed, incurable disease, in the hope that we might give them more options available through their cancer journey. And we, by the end of 2025 we’ll have got to 30,000 patients with incurable cancer, and hopefully put a good chunk of those patients onto these emerging targeted therapies.
MIC CAVAZZINI: So, patients with rare cancers might not have therapies tailored for them, but they may be candidates for ‘pan-cancer’ therapies, also, known as ‘tumour agnostic’ or ‘tissue-agnostic’ therapies. And this brings us back to trastuzumab. The HER2 receptor isn’t just amplified in aggressive breast cancers. You also find overexpression in cancers of the bladder, ovaries and stomach. Even pancreatic cancer, which while being the 8th most common cancer in Australia is still treated by a cocktail of very dated chaemo agents. Five-year survival for those with regional cancer is just 16 percent and investment is disproportionately low compared to the burden of disease. This year trastuzumab and deruxtecan combination therapy was listed in the US for any solid cancer with HER2 augmentation. Explain the significance of this tumour-agnostic listing and is that a watershed moment?
DAVID THOMAS: Oh yeah. Tumour-agnostic therapies are a key potential solution for a group of cancer patients that have not been previously druggable, and they are patients with two problems. The first is when we don't know where the cancer started, cancers of unknown primary. Cancers of unknown primary mean that the cancers spread, we can't infer where it started by looking under the microscope, we've got no clues. We don't even need know what therapies, standard therapies, chaemotherapies, to use because we don't know where it started. And for that group of cancer patients, having a molecular target and a tumour-agnostic therapy could be a game changer, because we don't care where it started anymore. We just care that it has that particular target.
And the second is the group of patients with rare cancers, and the reason that they are disadvantaged is that the cancer types, histologically, are so individually rare that we simply cannot do the sort of randomized studies that will prove a benefit. You know, there might be 10 people in the country with that target. Even if every single one of them has a tumour disappear with a treatment, we can't do a randomized study because there's only 10 people in the country. So, those populations of rare cancer patients have always been behind in the model the previous era of empiric drug development.
The idea that it doesn't matter what the label is, no matter how rare the disease, you can aggregate across cancer types based around a molecular target and then treat them with the same drug, is obviously a game-changer there. And the reason why we should care about that is because one in five cancer deaths are due to those groups, and they're markedly behind the rest of the population in terms of the number of standard-of-care treatment options. In our program, for example, there are an average of 2.7 lines of standard therapy that patients have received if you've got a common cancer, but it's down to 1.7 on average for those who've got a rare cancer diagnosis, simply because they've been outside the mainstream of rational drug of empiric drug development over the past 50 to 70 years.
So, tumour-agnostic therapies represent potentially a breakthrough for those groups of patients. We just have to then make a mainstream out of genomic profiling so that we can identify whether they do or do not carry the targets. And as of my last looking at the list, there are now nine tumour-agnostic therapies that have been approved. Trastuzumab-deruxtecan was approved in April and on HER2, 3+ that is the highest level of the immunohistochemical stain it has an average response rate of 61 percent. That’s incredible. Clinicians now have options for patients outside of breast cancer who also have HER2 overexpression so it has the potential to really transform clinical outcomes.
MIC CAVAZZINI: And, to be clear, when you talk about Comprehensive Genomic Profiling for these patients, you’re not talking about deep sequencing or looking for germline polymorphisms that convey cancer risk. You’re using tumour tissue extracted at biopsy or resection, and screening for nine targets that are registered with the FDA, another 50 potential ones, or a hundred oncogene risk factors that we know about. Are they all in the screen as well?
DAVID THOMAS: Okay, so, you raise an important distinction that many people, even surprisingly sophisticated people, will not understand clearly. When we talk about comprehensive genomic profiling, tumour profiling, we're talking about somatic mutations that were not present in the genes we inherited from our parents, but which have arisen during the course of the tumour growth.
And there are many different technologies for genomics, and they range from doing a single looking for an individual mutation, not even an individual gene, but an individual mutation, all the way through to stuff where we call whole-genome sequencing, where we're literally sequencing every single one of the 3 billion sites where there's a nucleotide in our in our genomes. And what we're using within the CaSP program is called Comprehensive Genomic Profiling, which is, sort of, in the middle. It's about 380 to 400 genes that we sequence. In some cases, we sequence the RNA component for various technical reasons, and we sequence that from the tumour to look for therapeutic targets, primarily. But we also get improved accuracy of diagnosis, because sometimes we see mutations which mean that the cancer diagnosis was wrong. And those are the primary benefits that flow to a patient.
However, one of the less well-appreciated aspects of tumour profiling is that we see also what's in the germline. And so, we can see when somebody has inherited a mutation which caused them to get the cancer in the first place, and it completely changes the way in which we think about heritable cancer risk. Because at the moment we look at breast cancer or bowel cancer or ovarian cancer, we focus on those diseases, and then we do a test to determine whether there's a heritable component. Now, we're just screening patients out in the wild, so to speak, and we're finding that one in 12 patients have these heritable risk mutations, and that’s not a rare event. And that has implications for just the volume of work that's going to go through to the familial cancer centres. I think we're going to increase the volume of work to be done by perhaps by perhaps, two, three, five-fold.
MIC CAVAZZINI: Let’s go now from the most recent of the listed pan-cancer therapies to the first. We already talked about pembrolizumab back in episode 79 as one of a number of immune checkpoint inhibitor therapies developed against melanoma. ICIs are not targeted to an oncogene, as such, but rather to native proteins that control activity of the patient’s immune system.
You see, the immune system recognises mutant proteins expressed extracellularly by cancer cells, just as it does invading pathogens. To avoid this scrutiny, cancers present on their surface a ligand to the programmed cell-death receptor expressed by cytotoxic T cells. It’s a bit like a Jedi mind trick, telling the T cell; “This is not the cancer you’re looking for. It’s just native tissue.” But when you administer pembrolizumab, it interferes with this fakery and exposes cancer cells once again to T-cell attack.
Given this role, pembrolizumab is most effective against cancers like melanoma that express a high density of mutant epitopes to arouse the patient’s immune response. Prior to the launch of immune checkpoint inhibitors around 2011, 5 year survival for patients with metastatic disease was just 17% according to a large real world data set aggregated by the US National Cancer Institute. Standard care at the time involved a decades-old agent called dacarbazine and cytokine therapies had not shown much promise. Since then, pembrolizumab treatment in patients with unresectable melanoma has resulted in 5 year survival rates of around 40 percent, with complete remission in most of those responders.
Pembrolizumab and other ICIs have in the last decade been adopted as first line treatment not just in metastatic melanoma but also as an adjuvant therapy in earlier stages of disease. In 2015, pembrolizumab was listed for use in non‐small cell lung cancer and a year later for head and neck squamous cell carcinoma.
Then in 2017, came the bombshell indication; any unresectable or metastatic solid tumour with an inherited predisposition to genetic hypermutability. It’s now been broadened to include cancers with a high tumour mutation burden from any cause. I asked Professor David Thomas to explain in more detail the DNA editing conditions known as mismatch repair deficiency and microsatellite instability.
DAVID THOMAS: So, the general idea is this, that the immune system recognizes antigens. And cancers, when they mutate, will produce proteins which have never been seen before, at least amino acid sequences, which can therefore be recognized by the immune system, in principle. The higher the tumour mutation burden, the greater the likelihood that there's an antigen in there that the immune system will be able to latch onto. And that's probably the basis for tumours with a high tumour mutation burden showing the highest response rates to immunotherapies. Think of melanoma, where you have UV exposure to the skin. It's amongst the highest tumour mutations of any cancer burdens of any cancer type, and immunotherapies work beautifully in melanoma.
There are other cancer types which don't get exposed to UV light, which have a high tumour mutation burden, which is driven by a genetic defect which results in an increased rate of mutations being generated. And mismatch repair deficiency and microsatellite instability belong to that group. Typically seen in colorectal cancers, for example, or in uterine cancer, but you can see it in many different cancer types.
What happens is that there are four or five genes which, when mutated, causes a loss of control over the generation of new mutations, and those mutations occur through an intermediate phenotype called microsatellite instability. So, the way to think about these terms is that mismatch repair deficiency will generate microsatellite instability, which will in turn generate a high tumour mutation burden. Now, mismatch repair deficiency is typically tested for by either sequencing of the mismatch repair genes and showing a mutation in them, or by immunohistochemistry to show mismatch repair protein loss of expression and therefore deficiency. [For] microsatellite instability, [there’s] is a particular type of molecular test which measures the consequence on DNA of creating instability around what are called microsatellites in the genome.
MIC CAVAZZINI: And is it the end phenotype and the high tumour mutation burden that makes them susceptible to immunotherapies, or is there anything more specific about those?
DAVID THOMAS: No, no, I think it’s generally that. I think the ultimate common route is probably high tumour mutation burden. Microsatellite instability and mismatch repair deficiency produce a more profound change in tumour mutation burden than you might see as a consequence of smoking, for example, and lung cancer so the response rates to immunotherapy tend to be higher, but there’s a range. So, the various thresholds are defined about what is an excessive tumour mutation burden. You know, one boundary might be 10 mutations per megabase, using one particular assay, others use 16 mutations per megabase. And we have seen patients with hundreds of mutations per megabase. Having said all that, there are patients in whom we see mismatch repair deficiency or microsatellite instability, but we're unable to measure tumour mutation burden for technical reasons.
MIC CAVAZZINI: And you’ve screened about 15,000 patients so far, of which about 8% had a high mutational burden, another couple had of percent had mismatch repair deficiency or microsatellite instability. That’s not a bad hit rate, but it’s actually only about half of what was reported from a European registry known as AACR GENIE. Any ideas why this difference?
DAVID THOMAS: Well, our patient populations are not precisely the same as the populations in AACR GENIE, almost certainly. And they're not the same as the distribution of cancers in the community. So, we provide access to molecular profiling for cancer patients where the system does not routinely provide that. But as you know, since November of last year, there's a small gene panel that's been developed for non-small cell lung cancer. For breast cancer as well, and for colorectal cancer, there are tests in place which are used to guide the choice of therapies. It means that the patients that come to us are biased towards those that have nothing at all. So, we have 70% of our MOST cohort was rare where that should be, something like probably 20% of the cancer population because they had nowhere to go. There were no options for those patients. There were fewer standards of care therapies. That's why we saw that distribution.
MIC CAVAZZINI: Right, okay. Another important target in this tumour-agnostic realm is the BRAF gene, in which a missense mutation at residue 600 causes disinhibition of a MAP kinase signal pathway leading to enhanced proliferation and tumour survival. This mutation was first described in melanoma but now is also associated with certain leukemias, colorectal cancers, thyroid cancers, colorectal cancer, non-small-cell lung cancer, neoplastic histiocytoses and ameloblastoma. The drugs dabrafenib and trametinib were given tumour-agnostic approvals in the USA in 2022 and listed with the TGA a year after that. I don't even have a question there…
DAVID THOMAS: Well, BRAF is an important gene to talk about because, first of all, it has a 50 percent response rate for dabrafenib and trametinib across any cancer type, it appears. The interesting thing is that in people's minds, initial BRAF inhibitors like vemurafenib worked beautifully in melanoma, but they didn't work in colorectal cancer—same target, just didn't work in colorectal. Same target, just didn’t work in colorectal. And so, this is an example, it was argued, of histotype-restricted therapy that is not tumour-agnostic even if they share the same mutational target.
However, it subsequently emerged that if you combine targeting of BRAF with another drug, that you could induce responses in BRAF-positive colorectal cancer. It's important that people don't think that, because there are examples of tissue-restricted responses that all are like that. That’s going to break down as we develop combination therapies and other ways of trying to understand why in a particular tissue that particular target doesn't work, there will be some other explanation that we may be able to exploit with time.
MIC CAVAZZINI: That’s really interesting. So you're saying that if we'd have given up just because that first generation of monoclonals didn't work, then we wouldn't have got...
DAVID THOMAS: Yeah, encorafenib and cetuximab which is currently standard of care for…
MIC CAVAZZINI: Right, and two final examples of tissue-agnostic drugs in current use are directed against the membrane receptors NTRK and RET. They’re involved in cell differentiation and cell proliferation, and it’s when they get mutated by gene fusions that the become oncogenic. There are a couple of dozen genes that can be involved in these fusions, so, given this mishmash what domain are the drugs larotrectinib and entrectinib actually targeting?
DAVID THOMAS: What domains? They typically target the NTRK—the fusion component of NTRKs 1 to 3, which is the consistent partner in the fusion. It doesn't seem to work with overexpression or with amplification or even point mutations, it seems to be specific for the fusion. Now, whether that's because tumours that have a fusion are particularly dependent on that pathway, whereas ones with point mutations or amplification are less dependent, it's not clear whether it's for that reason…
MIC CAVAZZINI: I just want to unpack some of the numbers about response rate, and hopefully the listeners will stay with us. For people with NTRK fusions the response rate to larocetinib is a whopping 75% but the frequency of that target is very low. HER2 amplification is a lot more common, and the response rate to deruxtecan is 60%. Selpercatinib for RET fusions 44%. Pembrolizumab has “only”, inverted commas, around a 30% the response rate but there about five times as many candidate patients. These response rates, how do they compare to the standard of care till now?
DAVID THOMAS: Yeah, that’s easy. I’m a sarcoma specialist so I’ll just go back to my own experience. So, our current standard of care therapy, which is used all the way across this country in cancer centres, in specialist units treating sarcoma, is the drug doxorubicin. And doxorubicin in advanced soft tissue sarcoma has an objective response rate of between 18 and 23%. So, the worst of the FDA therapies, these pan-tumour, tumour-agnostic therapies, is still better than the best standard of care we have now, according to the information we have available. And some of it is not just better. It's three times as good.
So, these drugs that are being FDA approved as having tumour-agnostic activity are potentially a game changer for diseases like sarcoma, where there are maybe 50 different histological subtypes of sarcoma. And if you were to do trials in each one of those to determine the objective response rate in leiomycosarcoma compared to synovial sarcoma, compared to liposarcoma, you'd just never get those trials done. It would take far too long to be able to do those trials in a randomized way. These tumour agnostic therapies, if we're prepared to accept the evidence that's in front of us now, could mean that those patients could get treatments that are just going to advance the field in a single jump by 40 years.
But because it's an emerging concept, those recommendations have not flowed through into public health care systems with single-taxpayer funding. They work in the US, where it now means that clinicians can prescribe the drug and that the health maintenance organizations will fund the provision of that drug, based on the FDA ruling. But in single-payer healthcare systems, like in Australia or the UK or where you have to demonstrate value for money, the complexity of using a completely different paradigm of tumour-agnostic drugs, working across any cancer type, and providing the molecular profiling is challenging for our health technology assessment process.
That's where I think there's a significant need for work to be done, because the science is out that you can show this benefit, the patients could benefit from it today, it's a question of how we assess that value for money from a taxpayer perspective and then integrate it into routine healthcare. If we could do that, would make a difference for a lot of patients in this country.
MIC CAVAZZINI: Many thanks to Professor David Thomas for this magic carpet ride over this rapidly progressing field of medicine. Keep an eye out for the next episode where he and I talk about the outcomes for the 15,000 or so patients who have been through the PrOSPeCT program to date, and some of the regulatory and funding challenges faced by tumour-agnostic therapies.
Professor Thomas has, received sponsorship or consultancy contracts from a number of different pharmaceutical companies, whether in his own capacity as a principial investigator or for Omico. Rather than list them all here I’ll put them in the show notes at racp.edu.au/podcast.
There you’ll also find a transcript loaded up with links to all the academic citations I’ve referenced. Another cool resource is the College Learning Series, which features a polished audiovisual lecture on antibody-drug conjugates in cancer therapy. That’s at elearning.racp.edu.au along with tons of great CPD tools.
Before I go, I also have to thank all the College members on the podcast editorial group who provided feedback on early drafts of this story. They’re listed by name at the website as are the composers responsible for the great soundtrack you’ve heard. Pomegranate Health is recorded and produced on the lands of the Gadigal clans of the Yura nation. I pay respect to the healers who walked this country for tens of thousands of years. I’m Mic Cavazzini, thanks for listening.