Dr. Darryl Irwin: There have been some fantastic advances in technology over the last couple of years. Next generation sequencing is getting much of the limelight and that’s because it’s allowed accelerated biomarker discovery in a hypothesis free environment without the need for linkage [sp] disequilibrium or association, but the high false positive and false negative rights of next generation sequencing, the variable coverage and the challenges with annotating insertions and deletions has many clinical laboratories concerned. I think we can all see the upside potential of next generation sequencing that it has the potential to discover novel mutations that you otherwise would not see, but the risk of missing a known actionable mutation is too high. It needs to be run in parallel with orthogonal confirmation techniques and technologies and proven tests for those markers that have therapeutic action
When we move over into clinical samples there are increased technical and biological challenges. In the area of oncology, in particular, we have the challenge of sample heterogeneity, meaning that we need a highly sensitive and specific technique to overcome the background of normal cells. There is also the challenge in turnaround time. Ideally, we want that to be less than a week and preferably one to two days
The next generation sequencing is fueling the content, and we now need mid-density panels on these pure proven technologies. So, today I’m going to introduce to you the oncology and translational research workflow on the sequene [sp] on the system. We have not been standing still in advances either. We’ve driven down sensitivity by orders of magnitude. We’ve improved our software so that you can design your custom panels, your content, rapidly. We’ve improved our technology for reliability and robustness
So, in the Brisbane Laboratory that I manage, we are heavily focused along with our brother and sister laboratories in other countries, but we’ve heavily focused on translating customer content into robust panels and experimental designs that meet the needs of our customer. We don’t believe one solution suites all questions
Today, I’m going to focus on the area of lung adenocarcinoma and specifically non-small cell lung cancer. We’re going to look at some proven examples of panels that we’ve delivered to our customers in Asia, including the OncoCarta and LungCarta panels, a panel for analysis of ABL kinase mutations, so slightly move outside of the lung adeno space there for a moment, but also our panel for the ultrasensitive detection of the EGFR T790M mutation and a panel for EML4-ALK translocation analysis
So, the modern paradigm of molecular profiling is to improve anticancer treatments in our patients by targeting therapies to patients that are receptive. There’s a lot of drugs available that are targeted to specific lesions in specific genes and there’s even more in clinical trials, but cancer is an invasive beast. A quiet resistance means we have a consistently moving target. And there are examples of acquired resistance for a variety of drugs. In the ABL kinase domain we’ve got the T315I mutation, which infers resistance to Imatinib and nilotinib and dasatinib. In gastrointestinal chambers, KIT mutation, T670I infers resistance to Imatinib. In EGFR T790M infers resistance to Erlotinib. For the EML4-ALK translocations the C1156Y and the L1196M mutations infer resistance to Crizotinib. So, it’s not a single test at a single point in time
So, in the lung adenocarcinoma space the testing paradigm is now well defined. You start with a KRAS mutation. You’ll find KRAS positive mutations in around about 15 to 30 percent of samples and generally there’s a poor outcome. Cytotoxic chemotherapy is basically the main option at this point in time. There are some drugs coming through clinical trials that may change that. But, then if you’re negative for KRASE you go on to analyze EGFR mutations. In the Caucasian population we’ll find them in around about 10 percent of the population. In Asia, where I spend most of my time, we find them in 40 to 50 percent, so substantially enriched in Asia. Good news for the Asians because they respond reasonably well for to EGFR tyrosine kinase therapy, but either a baseline at very low frequency or sometime during the course of therapy the majority of these will acquire resistance and the main mechanism for this is the T790M mutation. Ideally, we’d like to detect that mutation as early as possible and quantify it during the course of therapy such that we can determine clinical switchover points
So, in the next couple of months we will be releasing a focused EGFR KRASE panel, which has higher content than any of the current available panels on the market and also overcomes the challenges of next-gen sequencing in analyzing EGFR because they’re majority alleles. I’m going to talk to you today about a T790M resistance assay as well. But, if you’re negative for EGFR you then go on to analyze the EML4-ALK translations. If you have this these patients respond quite well to ALK tyrosine kinase inhibitors. And I’m not excluding RET [sp] and ROSS [sp] translocations, but I’m going to focus on EML4-ALK. We are also working in the RET and ROSS space as well, but EML4-ALK is better defined for me to present to you today
If you’re negative for EML4-ALK then, as Suzanne was talking about, you’d like to maybe do some molecular profiling for allocation to clinical trials. And that is analyzing other genes with unclear action such that you can apply these patients to clinical trials or research studies. And that’s a further panel that I’m going to talk to you about today and that is our LungCarta panel, which was released just last week
So, this is a commonly seen slide. This is the process that a biomarker takes to go from discovery through to validation. Of course in discovery, we have the challenge that there’s small sample sizes that are analyzed and large ararights [sp]. So, it requires technical confirmation to rule out the ararights, and it requires clinical validation to ensure that the marker is actually biologically active. Sequenom works in this space, technical confirmation and validation and, as you heard from the previous speaker, we’re moving further into the clinical deployment space. There are different needs at each stage of the process. No one technology can take you from start to finish. And the challenge of today’s laboratories is to determine which complimentary technologies to deploy to answer the questions you’re asking
So, therapy has two dynamics, pharmacokinetics and pharmacodynamics. Pharmacokinetics is looking at the target, what the therapy is targeting. And in this space we have a series of pharmacokinetic panels, our OncoCarta and LungCarta panels, the flagships that I’m going to talk to you about mostly today. Pharmacodynamics is about the passage of the drug through the body, the absorption, distribution, metabolism and excretion of that drug and how long and how effective that drug will be within the body. Ninety percent of drugs work in 30 to 50 percent of samples. That’s a statistic we hear all the time. To me that sounds fine for a headache. I don’t suffer much from headaches, but it sounds fine for a headache. It doesn’t sound all that great for cancer. And I think we’d all agree we’d like to improve it. Adverse drug reactions are a leading cause of death globally
So, we’d like to look at the ADME panels to look at how these drugs are being processed by the body and how they’re affecting the body on a global scale. So, we have our ADME toxicology panel. It’s a broad panel targeting those most relevant genes in the sit4 50 [sp] and other drug metabolizing genes so that you can look at both pharmacodynamics and pharmacogenetics in your clinical research studies. I won’t talk any more about ADME. There’s talks this afternoon at 1 o’clock, which will cover ADME
So, Suzanne, thank you. You did a great job in introducing OncoCarta version one. So, OncoCarta version one is a panel of 238 mutations in 19 different oncogenes. When this was developed in 2007 it contained 90 percent of the drugable [sp] targets at the time, and it’s still highly relevant today, but many groups wish to diversify the panel, increase the content, decrease the content and that’s where we’ve been focusing. To increase content, we’ve augmented this panel with a version two add-on and a version three add-on to increase the content, but we’ve also developed panels that are tissue specific. We have our melanoma panel, which is an eight reaction panel, looks at 78 [sp] genes, 72 mutations, our call-on [sp] panel seven genes, 32 mutations, our pancreatic panel 42 genes. And I have to read this. I can’t remember all these panels off my heart. Forty-two genes, 140 mutations. And our GyniCarta [sp] panel, which looks at 12 genes and 92 mutations. So, sequenom is not a one trick meets all needs company. We’re developing multiple different panels and also collaborating with our customers to make their own panels
So, our LungCarta was released just last week, and we’re very excited about it. We’re generating a lot of interest in the market. In fact in my laboratory there have been--there is 900 samples sitting--waiting for analysis or waiting for the launch last week for analysis. So, we’re very busy doing that
Lung cancer is a leading cause of death. One million deaths per annum globally, 1.2 million new cases diagnosed per year and generally it has a poor outcome because of the late stage of diagnosis and very few treatments available. In 2008 Ding [sp] and other authors published in nature a study on 600 candidate genes using next generation sequencing. They found somatic mutations in 26 genes, and they found a thousand different somatic mutations. We have used that publication and key opinion leader guidance from across the world to determine the content for our LungCarta panel. Our LungCarta panel is 26 genes, 214 mutations, and the flyer is on the memory stick. So, you don’t need to remember this. It’s received thorough in-house and field testing. We don’t release products until they are thoroughly tested. When we make a commitment of 10 percent mutation detection, we meet it or better it. It’s not an average
So, all of these detect 10 percent mutation detection, and we’ve done thorough testing to prove it. And you can start from 240 nanograms of sample from fresh tissue. With FFPE I suggest a little bit more and probably 450 nanograms is a good starting point
So, in our first field study, we did this up in Japan in the custom laboratory. We analyzed 37 FFPE samples and controls, and we found mutations in 25 samples with the LungCarta panel, a 68 percent mutation rate. That’s generally the rate that we’re seeing across various cohorts for these large types of panels, from 40 percent, as Suzanne described, up to 60 to 70 percent in some cohorts
So, we found those mutations in 10 different genes. We identified all 10 previously non-mutations. Of the 36 novel mutations, 15 were high enough in frequency for us to confirm with direct Sanger sequencing. The remainder were below the detection limit of Sanger sequencing, and we didn’t have sufficient tissue in order to do other techniques or Klinal [sp] sequencing
So, we moved on to the NCI-60 cell lines where we have plenty of tissue to--plenty of DNA to play with. We analyzed 38 of the NCI-60 cell lines with the LungCarta panel and detected 45 mutations. Thirty-six of these mutations have previously been described in posmic [sp]. Of the nine novel mutations, we confirmed these with an ofaganal [sp] technique the Illumina TruSeq cancer panel. Those that weren’t on the Illumina TruSeq cancer panel we confirmed with OncoCarta version one
I mentioned before that resistance is a challenge. Ideally, we want to detect resistance marks as early as possible so that we can guide patient therapy, but previous studies of ultrasensitive mutation detection have not correlated well to Klinal expansion. And that’s because with alleles specific PCR you’re actually introducing a mutation in order to detect the mutation. It also can detect down to one copy without quantitation and this is really important. The clinical threshold in a solid tumor will be different to a noninvasive fluid such as plasma or uron [sp]. One copy of a resistance mutation in a solid tumor may be unlikely to clonally expand. One copy detected in plasma may be likely to clonally expand. Sensitivity needs to be coupled with quantitation
The other challenge in clinical tissues is the amount of available sample. So, we’ve done titration studies, and this is a published from Beadling [sp] where they titrated out the amount of genomic DNA in the reaction to look at the sentle [sp], the technical sensitivity of the test going all the way down to .3 of the nanogram, .3 of a nanogram being 50 cells or 100 copies, and you can see we can get very robust extension even down at a 100 copies with 100 percent of the assay converted. We can go below 100 copies, but I urge severe caution that you’ll move into the area of pasoydin [sp] noise. But, also we want sensitivity. So, we can look at a titration of a mutation. Through the dynamic range you can see a clear mutation peak all the way down to 7.5 percent in this particular assay. So, the industry believes seven point--that 5 to 10 percent is a good range for solid tumors, but sometimes you want to go lower
So, if we’re thinking about plasma analysis, sequenom knows a lot about plasma analysis, because of our work, as Fritz [sp] described earlier, in cell free fetal DNA analysis for aneploite [sp]. So, we know that in a normal, healthy reproductive aged female in a milliplasma [sp] there will be 1,400 cell free copies, and with fetal DNA you’ve got generally about 4 percent, 60 molecules, so a sensitivity limit of 4 percent, plus or minus 1.3 percent is what’s required. Now, if we apply that theory across to noninvasive cancer analysis. So, in cancer if we assume the same background of 1,400 normal cell free DNA copies, if we have one copy of a mutation that would be .07 percent, but, of course, if you only have one copy then you’re going to be right in the area of stochastic noise. In the half the samples you’ll get one, in half the samples you won’t. So, if we bring in stochastic noise a detection limit of .21 percent is what’s required to detect a mutation in plasma in 95 percent of samples taking into account that sampling noise
So, over the last few years we’ve been working on a technique called SABER, single alleles based extension reaction. And what we do is we drop out the wild-type nuclear type, such that we can enrich for the detection of the mutant. The first clinical study that we performed was in BCR-ABL in chronic myeloid leukemia. The group--in collaboration with the group at South Australia pathology down in Adalata [sp], Australia, we worked with them to develop a 27 PLEX full reaction panel for the 27 common mutations in the ABL kinase domain. We thoroughly validated the detection limit of those assays, and the detection limit using SABER goes from .05 to .5 percent in those 27 assays. So, you can see that’s driving us down in order of magnitude, but in fact we have some improvements coming down the line. Our R&D group are working heavily on improving this and it looks like will be detecting reliable .1 percent mutation detection. So, here’s just some example of the data going from 100 percent, 10 percent, 1 percent clearly detectable, .5 percent clearly detectable and then a flat baseline for 0 percent
So, if we look at the resistance marks in adenocarcinoma T790M is now well described and well understood. Basically, in this study here that was done on the MassARRAY system, they were able to detect more samples with T790M mutations because of the improved sensitivity and also show that there’s a difference in the survival progression. So, we thought, “Well, that’s a very good start, but let’s see if we can apply SABER to it and if we can see T790M in plasma.
So, in conjunction with Kinki University up in Japan, we designed a SABER assay and analyzed 75 plasma samples after EGFR-TKI. Using synthetic standards, we were able to confirm the sensitivity limit of the assay was 0.3 percent mutation detection, and we observed the T790M mutation in 21 of 75 plasma samples. We were able to confirm that mutation by subcloning and sequencing in two-thirds of the samples, so that is 14 of the 21 positives were confirmed. You’ll recognize that cloning only 100 clones with a .3 percent sensitivity doesn’t go deep enough, but our pockets weren’t deep enough to clone a thousand clones. So, this the proof of principle that it’s possible to detect T790M in the plasma but needs further validation really to confirm the robustness of the test, but in this small cohort we’re actually able to see differences in the survival curbs with those with T790M mutations in the plasma versus those that don’t. So, it’s a nice proof of principle
The final area that I wanted to talk to you about was EML4-ALK translocations. And, again, working with the group up at Kinki University in Japan, they designed assays to all of the common EML4-ALK translocations, V1 through V7. These are single base extension assays amplifying over the breakpoint, over the fusion point. Controls confirm that they were able to detect all of the assay--all of the control samples with the correct translocation. And in a clinical study they were able to detect variant one in three of the FFPE biopsies and confirmed that with fish. We have now been working on an EML4-ALK panel where we’re combining iPLEX genotyping and SABER, jewel direction confirmation, and we’ve added in the resistance mutations, so the ALK resistance mutations. And just the week before last, I was in Singapore to perform a clinical study on 15 fish positive FFPE samples. You got extremely good validation. We detected more translocations than the current real-time PCR panel, because we include the alternative variance, and we also detected samples with both a translocation and a resistance mutation with similar frequencies. So, again, this is a further panel that we’re working on to develop and validate, and we’re happy to talk to anyone who’s working in this space to help us progress that through
So, what do you get with the Sequenom system? You get extremely high quality data. You get quantitation combined with sensitivity, an essential combination. You get precision, accuracy and robustness. Endpoint PCR combined with short amplicons means this works on a variety of tissues, as Suzanne explained. We focused on somatic mutations today. There are a variety of other applications. This is a toolbox for targeted nucleic analysis across a variety of different biomarker classes. We have software so that you can design your own assays, but we also have an extremely active and skilled workforce, support force throughout the globe. Hemmer [sp] we’ve heard about today. We’ve got Marian [sp] here as well today, and myself from the support field. We work actively with our customers to help you design your panels with your content. And it’s a proven technology. It’s low cost and high throughput as required for translational research and clinical deployment and a proven technology with over 1,500 publications in the area of discovery confirmation and deployment of mature panels for screening studies
So, I thank you for your attendance today. We do have another lecture at 1 o’clock, which will, again, touch on somatic mutation but also going to ADME toxicology studies. Are there any questions