Mar 21, 2011
Jacquelyn K. Beals, PhD
March 21, 2011 — Editor’s note: Genomic sequencing is moving from bench to bedside, being used to design and direct treatment strategies for colon tumors, predicting response to therapies, and revealing startling genomic heterogeneity within a single tumor. Although current applications are mainly in the field of cancer genomics, conditions as diverse as Charcot-Marie-Tooth disease and healthy aging are beginning to benefit from sequencing technology.
In an interview with Medscape Medical News, Samuel Levy, Ph.D., director of genomic sciences at Scripps Translational Science Institute, Scripps Health, and The Scripps Research Institute, in La Jolla, California, discusses his own presentation and other highlights of the sessions he comoderated at the The Future of Genomic Medicine (FGM) IV conference, held March 3 and 4 in La Jolla, and sponsored by his institution.
Medscape: Your presentation at the meetings focused on new insights from cancer sequencing. Can we talk about what you presented?
Dr. Levy: I covered 2 topics. One was a project in which we’d been applying sequencing to help direct treatment for a colon tumor. A fairly young individual, 27 years old, had presented with a stage IV tumor and didn’t respond well to standard therapies (cetuximab therapy, which works for 75% of individuals with colon cancer) because he had a mutation that had only been described in a KRAS gene. The idea was to determine whether we could apply sequencing to understand more about the other mutations in his colon tumor to help direct therapies that could help mitigate circumstances.
We sequenced his colon tumor, and also his blood DNA, in an attempt to understand the full set of mutations in his genome. We used 2 separate approaches — with Illumina as a collaboration to sequence the genome, and we ourselves sequenced the exome (protein-coding regions of his genome) at a higher depth to ensure that we got good accuracy. It’s hard to be sure you have comprehensive coverage of all the mutations and that you have characterized those in a comprehensive manner, and also that you’re being very accurate. As a first attempt to work with this kind of methodology to help guide the patient’s therapy, we wanted to be sure that things were correct.
The sequencing revealed the mutations described for him already and a whole host more suggesting 2 forms of therapy that might help. One would be to block genes further downstream in the RAS/RAF pathway; a gene called PI3KCA also was mutated, and therapies are available for that, too — the combined effect of those 2 therapies might help. In addition to the standard and targeted therapies, there were chemotherapies that could be applied. He was put on a standard chemotherapy of FOLFOX, a 3-drug cocktail [folinic acid, 5-fluorouracil, oxaliplatin]. We predicted from sequencing his germline DNA that he was likely to develop a mild neuropathy, which indeed he did develop.
Medscape: Had you predicted that in advance?
Dr. Levy: Because of the timing of this, we found out after the fact that he had the neuropathy, but the actual sequencing revealed that. In doing this, we were not in a position to actually guide the therapy. The idea was to determine whether we could learn something more from the sequencing. Clearly, going forward, the idea is that this would be done hand in hand with a therapy plan. One of the other therapies used in a combined fashion for colon cancers that don’t respond to cetuximab is the VEGF [vascular endothelial growth factor] inhibitor bevacizumab (Avastin). Because VEGF was upregulated, on sequencing we found suggestions that that therapy would be useful for him since the gene was amplified to 3 or 4 copies in his genome.
At least for this patient in this so-called “n of 1” trial, we found utility in this approach because it gave us more solid information on the course of therapy, which is very useful in communicating with a patient for their success. Going forward, we hope this kind of approach will be useful — not just in revealing something that we wouldn’t have seen before. There are mutations for which existing gene tests won’t be tried, and indeed in a different patient, a different drug-therapy strategy might work because of a different set of mutations. But in this patient’s case, we provided good support for why the existing therapy plan might actually be beneficial, as well as suggesting other therapies that might be turned to if the existing therapies didn’t work. From that perspective, we were very encouraged by it.
The second thing I talked about was a study that tried to determine the amount of mutational heterogeneity within a tumor. Again, we turned to another patient with a colon cancer, and asked: “If you want to diagnose somebody’s tumor by sequencing — using the methodology I described in the first part of the talk — does it matter where you sample that person’s tumor?” We looked at 5 different loci within a colon cancer patient’s tumor, and applied the same strategy of sequencing — in this case, just the exome — and asked whether the same mutations are found equally in all parts of the tumor.
In fact, we found that that is not the case. It did matter where you sampled the tumor from. What we’re now trying to understand, for this particular patient, is whether that would have potentially changed the strategy in terms of the therapies proposed for this patient. We’re trying to see how we can look at the spectrum of heterogeneity. How different are these loci in multiple patients? We really wanted to do the experiment to try to understand the best strategy to apply going forward if genome sequencing is going to be applied as a diagnostic test.
At the conference, not just in cancer but also in other disease areas, the notion of how you could apply sequencing was very much in the forefront of everybody’s minds and in many of the talks. The 2 days actually reflected some of the successes people have had using sequencing in different disease areas.
Medscape: Is there any way to tell from sampling a tumor in several spots which mutations arose earliest and which developed later in the tumor’s history?
Dr. Levy: That’s a very good question. We’re actually trying to do that now. We’re taking the mutational data from the 5 different areas and trying to use analysis methodologies to see if you can provide a lineage, an evolution of how the tumor might have come about. There’s a fair amount of analysis still to be done, but some of the initial findings suggest that regions on the edges of the tumor, the transition regions where you go from the tumor edge into normal tissue, provide the most mutational variation — which is something we didn’t expect to see. We have to corroborate that in other samples, but if that turns out to be the case, it is a very interesting observation.
We’ve yet to establish how the flow of mutational changes exists. The other question is: Beyond understanding how cells evolve and how mutations change across the tissue, how does that change how you diagnose? If you can establish in multiple tumor types that there’s a pattern you can almost imagine is not random, you could come up with a strategy — presenting a pathologist with a plan of where to excise tissue from if you want to get a consistent diagnosis each time.
We’re still working toward that, but we hope these kinds of studies will lead to that kind of approach. Within the hospital system where I work, that’s a very important aspect of trying to apply genomics in a patient-care-focused manner.
Medscape: Do you foresee this type of approach being widely used clinically, maybe in 10 or 20 years? Or will it only for be used for unusual tumors or those that are resistant to standard treatment?
Dr. Levy: In terms of timeline, I think it’s here already. A group in British Columbia published something a year or so ago on a rare adenocarcinoma of the tongue. In doing the sequencing, they could characterize which pathways were mutated, and they came up with a treatment strategy that prolonged the life of the individual by about 4 months.
I think that certainly begs the question of where the best way to apply this is. Is this going to be something that will become standard to every patient who has a cancer? I think ultimately it will. One thing most people do, when they’re trying to think of the best way to apply these methodologies, is to see where the most benefit would be had first. That was behind our choice of this individual with an early-in-life colon tumor that was quite advanced — with the hope of trying to understand, with these early attempts, whether we can get successes that show the benefit of the approach. I think that between now and 2 years from now, you’re going to be seeing it routinely applied in this mode, which is a research mode.
Medscape: What kinds of questions or reactions did you get in the panel discussion?
Dr. Levy: How soon, or when, or under what conditions this will be applied going forward was certainly among the questions, Others were: How important is sequencing someone’s DNA? Is the epigenetics of the tumor as important? (I think the response was ‘yes’). Certainly this is an important aspects too, and we’re at the beginning stages of understanding that.