Transcript

KRISTINE BARLOWE-STEWART: So we were testing for inherited cancer but we’ve now found that you are predisposed to Alzheimer’s disease—totally unexpected. That’s the problem, these incidental findings.

MIC CAVAZZINI: Welcome to Pomegranate, podcast of the Royal Australasian College of Physicians.
I’m Mic Cavazzini, and this is the second of two episodes on the application of genomics in clinical practice.

Last time we talked about how whole genome sequencing can be used to diagnose rare and poorly understood conditions as it allows a wider net to be cast with each new diagnostic hypothesis. Healthy people might even adopt lifestyle changes and interventions on the basis of inherited risk.
But risk is a slippery concept. Only the classic Mendelian conditions manifest predictably from the presence of a mutation, presumably because of there is little redundancy in the biochemical role these gene products play. And BRCA1 mutations, for example, are ominous because they confer a 65% per cent lifetime risk of breast cancer.

But most genetic associations we know of have a much weaker influence. What do you advise people who carry a variant of small effect or parents expecting a child with one of these known markers? Clinical geneticist Michael Gabbett, of the Royal Brisbane and Women’s Hospital:

MICHAEL GABBETT: I guess, one of the tenets of clinical genetics really is to take a position where you provide the information, help people process that information, but how they perceive the risk is something that is very individual. So certainly, I’ve had people when I’ve said, “The chance of this recurring again is 25 per cent,” they’ll say, “Oh, is that all?” I don’t know if it’s that they focus on the positive—that three in four chance of it not occurring is actually the positive that they like to focus on.

Conversely, I’ve had people where I’ve said, “Look, the chance is as close to zero as possible. I couldn’t put a number on it but it’s certainly less than one per cent,” but they are absolutely petrified of something recurring, then they’d like to take any steps possible to prevent that. And it’s good to have an understanding of your patient’s background. If you can come up with analogies that I guess are visually something that they can relate to quite easily, more so than if you use some esoteric example. And I try to put them in a comparative fashion. I guess one of the great analogies I always use in my counselling is there’s no such thing as a risk-free life. I could drive home from work today and get killed by a semi-trailer. You know—risk is inherent in life, risk is inherent in having babies. I can never take away risk.

MIC CAVAZZINI: Associate Professor Kristine Barlowe-Stewart, Director of the Masters of Genetic Counselling program at the University of Sydney:

KRISTINE BARLOWE-STEWART: Risk perception is very personal. I often say to my students, “You have a ten per cent chance of winning the lottery. Stand up if you think that’s a good chance.” They all stand up.

“You have a ten per cent chance of your child having a problem.” They’re not often the same people who think that’s a risk.

If they have experience of a particular condition their perception of that risk is going to be different to someone who’s just read about it. People are not necessarily numerically literate, you can’t use percentages. You’ve got to present that number in a whole different way perhaps using the 100-person figure. And it takes time.

MIC CAVAZZINI: These discussions about risk are going to become more nuanced as the findings from genetic epidemiology research trickle down to the clinic. One sensitive tool is the genome-wide association study, or GWAS for short. These are large scale case-control studies in which researchers fish out thousands of single nucleotide polymorphisms that are common in the human population.

These common variants, known as SNPs, are not like the pathogenic mutations we’ve talked about until now. SNPs are usually harmless silent mutations, or fall outside protein-coding regions entirely. But they sometimes associate with a slight increase in disease risk, perhaps because they sit within an important regulatory sequence, or segregate with some unknown causal gene factor nearby.

Research is only just beginning to make sense of these variants, says David Thomas, Director of the Kinghorn Cancer Centre and the Cancer Division at the Garvan Institute of Medical Research.

DAVID THOMAS: A SNP is a change, a variant that is quite common typically amongst people, perhaps more than five percent of the population might have that SNP but the effect side is quite small. So many people, you know, ask themselves, “What’s the point of a genome-wide association study in terms of deriving clinically useful conclusions from something which has a 1.2-fold risk, a 20 per cent risk?”—for example.

But the reality is these things don’t operate as singletons, they’re not lone wolves—they hunt in packs. And because they’re so common, for individuals who have, say, not one disease-associated SNP but maybe five at once, all of a sudden the cumulative total risk for that person goes from 1.2-fold to maybe eight-fold.

And we certainly see that in breast cancer. There have been studies, including Australian studies actually by Paul James in Melbourne, which have shown the polygenic risk score—that is, using the genome-wide association study results and aggregating all of those the risk—is phenotypically indistinguishable from carrying a BRCA1 mutation. For the woman concerned it makes no difference whether you’ve got BRCA1 or five of these nasty SNPs together. And that’s a really interesting thing, which I think is still to enter into clinical practice in terms of risk stratification, but the data are unambiguous.

MICHAEL GABBETT: I guess commercial testing for these normal variants has been available for many years now. You know, you can hop on the internet and these companies use what information is available in the literature to look at your risk profile for, you know, dementia or heart disease. Now, as our understanding of normal variants increases that will become more powerful, but we’re looking at sort of relative risks of, you know, 1.5 times more likely to develop disease X, and if the background incidence of the disease is low then 1.5 times a small number is still a small number. So, I guess us clinical geneticists aren’t big fans of these commercial tests. Certainly next generation sequencing will be able to find all these normal variants and as our knowledge improves then we should be able to design algorithms to give a risk figure at the end of the day. But there’s still a lot we don’t know.

DAVID THOMAS: So, here’s something that might happen over the coming years. Typically people come into a familial cancer centre because they know that people in their family have had cancer. In a way you’re primed, you know—you’ve seen your sister, your mother get breast cancer and now you’re worried about breast cancer.

But increasingly we’re going to be discovering genetic causes that are not going to be dominant and Mendelian in their transmission patterns, which means there may not be a family history, and it might be quite different in terms of clinical genetics practice to give people information which they weren’t expecting because they’ve had no kind of conditioning for that.

MIC CAVAZZINI: Whole genome sequencing allows you to screen for every genetic risk factor in the book, but the problem of casting such a wide net is that you might detect disease markers you weren't expecting. How such incidental findings should be dealt with is one of the main ethical issues with the technology. The question becomes even more fraught in the context of prenatal screening, since foetal DNA can now sampled non-invasively from the maternal blood.

KRISTINE BARLOWE-STEWART: That’s the problem, these incidental findings. The chance is not high—one to two per cent, but it can happen. There will be enormous uncertainty generated from that because of our still-limited capacity to interpret what we find. Even if we find something and we know that it can cause a problem, it may not.

MIC CAVAZZINI: And is that a risk you explain to the patient that, “OK, we’re looking for this but we might turn up other things. Do you really want to go down that path?”

KRISTINE BARLOWE-STEWART: Yes, that’s a conversation we have because the consent forms now have that on them. We ask them, “Do you want to know this other information?” So people have a choice. If that’s what the patient wants—some patients will say, “I want to know everything.”

And I think that increasingly that choice is not given. Everyone has an ultrasound, and it’s a test, it’s a prenatal test. People don’t think it’s a prenatal test because they think they’re just getting a picture of the baby, but it’s a test. And because it’s a blood test, and they have zillions of blood tests in pregnancy, more and more there is the expectation that you will do these tests. And I think you have to tell them that this is a special blood test. But not everybody then wants to go down that path.

Making a decision to terminate a pregnancy, for example, is the most difficult decision anyone can ever make. But if we’re going to be doing whole genome sequencing, we’re going to find these babies are at risk for adult-onset conditions like Alzheimer’s disease and breast cancer and bowel cancer, in a foetus. And you know we, as a society, need to have the conversation about how appropriate is this, and should there be limits on the technology?

MIC CAVAZZINI: Genetic pathologist, Professor Leslie Burnett is more confident that systems are in place to avoid incidental findings from even coming to light.

LESLIE BURNETT: Well I’m Leslie Burnett, and I’m the Chief Medical Officer at Genome One at the Garvan Institute of Medical Research.

If we are searching for a narrow diagnosis and we know what we’re looking for, we will adjust the bioinformatics filters so we never see these things. We are, in effect, going in with deliberate blinkers, consciously applied, because we know that either the requesting clinician has asked us to look no further or the patient has not consented for us to look further. But for more difficult cases where the net necessarily has to be cast wider we will find unexpected findings, we’ll have to evaluate them and decide whether they’re relevant.

Look, the issue of incidental findings is very important. Some see it as an issue to be feared and avoided, others see it as an issue to be welcomed because if they’re present it’s a risk that the patient has and they should know about it. And this was brought to a head two or three years ago by the American College of Medical Genetics and Genomics which, I think very bravely, put out a list of 56 important findings that they claimed that one is professionally obliged to search for and include whether or not it had been requested. Because these were sufficiently important, would have an impact on the patient’s life, and you could do something about them, therefore don’t you have an obligation to inform the patient and the clinician?

Now, this caused much controversy and is still debated around the world, and I think, I respect them and support them for having made the statement, but it probably didn’t quite capture the consensus views of others working in the field around the world. And the current position is that this list is widely respected, but you don’t have an obligation to include it. One should seek the patient’s consent as to whether they wish to know about it or not, and one shouldn’t go searching for it just because one can.

Now, it’s not a black and white thing. I’ve had my genome sequenced—I was one of the early adopters, because I was doing this as an experiment to see how I would react. I discussed this around the dining room table with my family and my view was, “I want to know because if there’s something I can do about it, well, I’d be foolish to close my eyes, it’s not going to go away.” And my wife looked at me and she said, “Well, if you are going to be mad enough to have it, under no circumstances tell me anything.” And then my daughter, who works in a similar field, she said, “Oh, you’d be crazy not to know. Look, if you find anything you have an obligation to tell me because I share half my genes with you!”

DAVID THOMAS: Look, so actually this is an interesting point. You cannot consent someone in an informed way to the full implications of a whole genome test.

MIC CAVAZZINI: Medical oncologist, David Thomas once again:

DAVID THOMAS: You know, to consent somebody to the full spectrum of human disorders which might be identified or for which the information might be relevant would take far longer than is clinically feasible, and knowledge is changing so rapidly that you could consent at one point in time and it would become irrelevant 12 months later.

So I think what you’re consenting people to understand is that the information is changing rapidly and that we will use our best judgment, clinical judgment, and always putting the patient’s best interests at the forefront of our decision-making. Electing to return those results which we feel we have a duty of care to return to the patient, whether or not they express a certainty that they don’t want to receive that information or not.

And the reason I say that is there’s quite a lot of literature about people making a firm decision at one point in time and changing it the next, and the last thing want you do is take a result which might change the course of a person’s life and not return it to them because of a decision which couldn’t possibly, in retrospect, be called informed maybe some years ago. And of course, if it’s not relevant to you as the subject it’s relevant to children and sisters and brothers and so forth.

Dealing with uncertainty—it’s actually not an unusual feature of medical practice, you know. When you get a chest X-ray because somebody’s got some chest pain and you see an incidental spot in it, people don’t panic and think, “We needed to have consented people that we might identify a spot.” You just deal with it because you’ve got a duty of care to the patient to do what’s in their best interest. Those things are part and parcel of the way clinical practice evolved, and I think the same thing will apply with genomic information.

MIC CAVAZZINI: The apprehension some clinicians feel towards genomics testing will change with time, says clinical geneticist Michael Gabbett:

MICHAEL GABBETT: You know, with anything new you always have to tread a little bit carefully and there are certainly some traps I’ve seen generalists fall into and do quite a bit of damage by not providing adequate pre-test counselling or giving an erroneous interpretation to the families. But that, I guess, is a function of not being familiar with the test.

A good analogy I think of is HIV. So certainly when I went through medical school for someone to have an HIV test was a very big deal. There had to be significant pre-test counselling and then to give the result it was all very planned—you know, a quiet room, one-on-one. Now HIV testing is done quite, I would say willy-nilly, simply because people are more accustomed to knowing what HIV is, but also our ability to treat it is much, much better. So, the test itself hasn’t necessarily changed but the meaning of the test has changed and our familiarity has changed. And I think genetic testing is something similar. With next-generation sequencing there are a lot of caveats to the test, they are still a little bit special, but I have seen certain genetic tests over the years becoming less special as people get more familiar with them.

MIC CAVAZZINI: Another concern about collecting a person's entire genome sequence is what then happens to the information. Who has a right to gain access to it and what kind of decisions will it influence? Kristine Barlowe-Stewart was a founder of the Genetic Discrimination Project aimed at dealing with some of these questions.

KRISTINE BARLOWE-STEWART: So, what we’re talking about is being treated differently by a third party like insurer or employer. Access to a right that you would have normally had has been changed on the basis of your genotype rather than anything that you were presenting with.

And so, we investigated that and we certainly found in life insurance that’s where there was the greatest concern. In many other countries life insurers can’t ask you about your genetic information. Here, if you know your genetic test result you have to tell them—“Have you had a genetic test?”—on every form. And we got access to all of the applications that were disclosed, where a genetic test was disclosed, and basically most of the results were negative, but that’s because people are not really disclosing. I think the fear—

MIC CAVAZZINI: —and so you’re still pushing to have that requirement removed?

KRISTINE BARLOWE-STEWART: —no, look, I think it’s a problem if we start saying a genetic test is different from every other test. What I am pushing for is that if it is disclosed, the interpretation has to be evidence-based. If the data isn’t there, they shouldn’t be using it. So my message is, use your genetic counsellor as your advocate. If you are concerned, challenge the insurance company.

MIC CAVAZZINI: Probably the most advanced area of genomics is oncology, where more than 140 genes have been identified that can be involved in tumorigenesis. Every cancer is in a sense a unique experiment in cell regulation. Some of the mutations are inherited in the germline and set down a baseline risk profile. But on top of this a perfect storm of spontaneous somatic mutations must brew before the tissue veers onto an aberrant developmental pathway.

DAVID THOMAS: It’s an important distinction, the distinction between tumour-testing, looking for genetics signatures in a tumour that reflect the changes that have occurred in that cell or the population cells that allow it to form, and testing of the stuff that we inherit from our mother and father that we call the germline—which we typically measure in blood. Which is actually a tissue in its own right.

And of course, the big problem of genomics now in this era is not mapping genetic variants but interpreting them. There’s stuff that’s bleeding obvious, there’s a whole lot of stuff that’s grey. But a lot of what’s driving genomics into clinical practice is actually this tumour-testing question because, think about the risk benefit: if you’re going to recommend something as radical as mastectomy or removal of the ovaries, or potentially drug prophylactic treatments you need to have a very high standard, a bar, in order to intervene in somebody who is well. So, you’re talking about predictive testing, making a statement about someone’s future that is so strong that they would effectively undergo surgery to mitigate that risk. But if you’ve got somebody who’s dying from cancer that risk benefit ratio about what you call significant is very different.

MIC CAVAZZINI: Testing of breast cancer biopsies on a 21-gene panel can reveal how aggressive they are likely to be, and which patients can be safely spared the discomforts of chemotherapy. And still at the cusp of research translation are findings about whether cancers will be sensitive to conventional drugs or to radical new therapies instead.

DAVID THOMAS: Most of the targeted therapies that we’ve been using over the past 20 years to treat cancer ultimately buy time, but don’t eradicate the last cancer cell. They become resistant. So David Bowtell has been one of the world’s leading researchers in ovarian cancer and using genomics to study it. And he put out a very important paper describing genetic changes that predict for sensitivities to cisplatin, resistances to cisplatin, and which became mechanisms for acquiring resistance. And I think that’s going to be an increasingly important aspect of cancer treatments. These drugs are very expensive and they don’t cure people in many cases—I should say, up until the advent of the immunotherapies where the “c-word” has become tantalisingly close on the horizon.

So the immunotherapies—which are really cool developments in cancer drug therapies because they harness our immune system to recognise the cancer cell and eradicate it—those drugs appear to work in a way that is in some sense related to their mutational burden. And some of the ideas behind this are not that you’re targeting a particular mutation in a particular gene, but you’re targeting all these new proteins that are generated by the mutations. There are enough new epitopes that the immune system can get hold of and recognise and can form the basis of an effective immune response. So, melanoma and lung cancer are the two flagship diseases where the immunotherapies have had enormous impact—tumours which are just screaming white noise. You end up with a thousand mutations per tumour cell, and many of them are passenger mutations; they’re just part of the broken machinery or the very high exposure to carcinogens.

But at the other end of the spectrum you have cancers that affect children and young people that are eerily quiet mutationally. These are things that we really understand very poorly. So how do these things interrelate? Targeted therapies, immunotherapy, genomic profiles, you know, precision medicine—all of these things are extremely complicated.

Clinical practice is increasing and becoming a really weird blend between received knowledge, the sort of canon of knowledge of medicine, and the cutting-edge of what we’re learning through large-scale clinical studies. There’s a really important paper by Razelle Kurzrock, the head of phase 1 at M.D. Anderson, probably the largest cancer centre in the world. And she did a meta-analysis of phase 1 studies where we were trying drugs for the first time. We’re not really after getting a signal-response, we’re trying to work out how to use the drugs, use the right dose and the right schedule.

So in that meta-analysis of 360 phase 1 studies she asked three questions. She said, “What is the benefit to patients of a chemotherapy used in that setting?” And it was five per cent; five per cent of patients had something like a response. And that’s what we’ve known for chemotherapy used in that setting in a phase 1 study, and that’s really quite dismal. And then she asked a second question that was kind of cool, she asked, “What about a targeted therapy designed to target a particular gene, but not paying any attention to whether that mutation is present in a tumour, just trying it?” And the result was five per cent. Which means if you develop a fancy new drug on its own, it may be fancier or newer, but unless you use it rationally it doesn’t work.

And then she asked the killer question which is, “If you connect a targeted drug to a target within a tumour, the notion of precision medicine—the right patient, the right drug?” The response rate was 43 per cent. Now, in a phase 1 where you’re using the drugs for the first time what that tells you is that human science and reason actually does work, even when you don’t have empiric historical evidence. And that’s why I think, as knowledge is accelerating, as this data comes out, there needs to be an impetus to bring science into the lives of patients who don’t have the time to wait for phase 3 studies to emerge, there needs to be increased capacity to offer people an alternative to a phase 1 cytotoxic. And you know that’s the future, I think, and that’s where genomics comes in because it represents an enormous advance in our ability to identify those biomarkers.

Now, it’s important that people understand that giving the test does not necessarily mean your tumour’s going to melt away. You have to then get a hold of a drug, those drugs are often not available, they may be available on trials, they may be available overseas. But let me put that into clinical perspective. So, when I see a patient with metastatic soft tissue sarcoma and I offer them the frontline therapy that the government pays for, the response rate to that is 18 to 23 per cent, and usually lasts two, three months and then the tumours progress. And that’s what I do as part of clinical practice. The likelihood of identifying a therapeutic target with a panel test is about the same order of magnitude—between 10 and 20 per cent depending upon the panel. And if I’d run out of my doxorubicin’s and my ifosfamides I wouldn’t think it unreasonable to get a molecular test, for a possible, a more rational pairing of a drug to a target.

I have patients who’ve got advanced cancer, for example, for whom I can quite reasonably say, “Look, we don’t have a treatment right now but the pace of knowledge is such that within 12 months there could be something turn up.” And look, 10 years ago that would not have been plausible.

MIC CAVAZZINI: That’s all for this Pomegranate series on “Genomics for the Generalist.” Please go back and listen to the previous episode of you missed it.
Thanks to Kristine Barlowe-Stewart, Leslie Burnett, Michael Gabbett and David Thomas for their contributions. The views expressed are their own, and may not represent those of the Royal Australasian College of Physicians.

For resources mentioned in the podcast, or to claim CPD credits for listening, visit the Pomegranate website at racp.edu.au/pomcast. You’ll also find our email address if you have any feedback, and please share the story around using the hashtag #RACPpod.