Counting Colorectal Tumors During Flights to Asia

David J. Kerr, CBE, MD, DSc, FRCP, FMedSci; Alan P. Venook, MD


July 22, 2016

This feature requires the newest version of Flash. You can download it here.

David J. Kerr, CBE, MD, DSc, FRCP, FMedSci: Hi. I'm David Kerr, professor of cancer medicine at the University of Oxford. Welcome to this edition of Medscape Oncology Insights, coming to you from the 2016 Annual Meeting of the American Society of Clinical Oncology (ASCO). Joining me today is a friend and colleague of many years, Dr Alan Venook, who is professor of medicine at the UCSF Helen Diller Family Comprehensive Cancer Center in San Francisco, California. Welcome, Alan.

Alan P. Venook, MD: Thanks for having me.

Dr Kerr: You and I have been involved in colorectal cancer research for decades, trying to understand its biology and trying to come up with new treatments. One of the things that's always interested us has been trying to match the biology to the clinical behavior of the tumor and how we might be able to use that for therapeutic intervention. I know you have some fantastic new data to present at this meeting. Would you tell us a bit about it?

Two Sides to Every Question

Dr Venook: We're building on a study that I presented a couple of years ago called CALGB/SWOG 80405.[1] This was a study in patients with advanced metastatic disease where the primary goal was to look at the difference between bevacizumab and cetuximab as first-line treatment in KRAS wild-type patients.

As the dust settled and as we talked more and more about personalized medicine and about individualized care, we started understanding that there are genetic differences. We realized, as a few people had even many years ago, that perhaps there was a difference with cancers that began in the right side of the colon vs the left side of the colon. We really observed this mostly in the classic BRAF-mutant colon cancer, which we've learned is a bad disease—a bad version of colon cancer, which is almost always on the right side of the colon.

With that in mind and a couple of small suggestions—publications on small subsets suggesting that that might be the case—I happened to look back at old literature a few months ago. I was giving a memorial lecture for a colleague from many years ago, Scott Wadler, who had done some of the earliest work with fluorouracil (5-FU)-based chemotherapy. I was looking to see what he had done, so when I gave this lecture in his memory, I could cite what his work was.

Sure enough, buried in one of his papers from a study he'd done back in the 1990s and published in 2001[2] was the observation that patients who had left-sided primary colon cancer with metastatic disease lived 15.8 months, versus 10.9 months if they had right-sided primary disease. I had not appreciated that. In fact, when this was published, it just essentially established infusional 5-FU as the standard in the United States—the way to use 5-FU.

Nobody had really paid attention to [sidedness]. With that as well as our observation that there may be something there, we decided to dig in and look to see if this might not be the case in our study.

Left Colon and Right Colon Tissues Differ in Origin

Dr Venook: We had a large study with 1137 patients.[1] There was no difference in outcome between the two arms [receiving bevacizumab or cetuximab], so every patient was informative to the question of whether there might there be a difference prognostically on the basis of which side of the colon the cancer was found.[3] That's where we got the germ of the idea, and since then a number of other people have done these kinds of analyses as well.

Dr Kerr: Do you think the somatic or mutational landscape is different from left versus right? Is there anything embryologically that would lead you to that specific conclusion?

Dr Venook: There is. If we remember embryology, it turns out that the left colon derives from what's called the hindgut and the right colon comes from the midgut. They're embryologically different-origin tissues that merge at the transverse colon, and that's variable. So [sidedness] is not surprising.

Dr Kerr: So this is embryologically plausible. I seem to remember Vogelstein and the gang[4] publishing a while ago in Nature and saying they could not find a difference when looking at mutations in rectal tumors and colon tumors.

Dr Venook: One of the fascinating things is, of course, how have we split up the disease "colorectal cancer" up until now? Colon and rectum. What we did in our analysis was lump rectal and left colon together. Turns out they're really not very different.

'Duh, what were we thinking?' It should have been obvious to us that there was a difference.

But if you talk to a gastroenterologist, they will describe these flat polyps on the right colon. We know that BRAF-mutant patients [have more right-sided tumors], for example. We see women with peritoneal metastatic disease that's not BRAF mutant that's on the right side. As I looked back, it's like, "Duh, what were we thinking?" It should have been obvious to us that there was a difference. That's what prompted it. The problem was getting the data because we hadn't anticipated this; we didn't collect this in the case report form.

Dr Kerr: Because these were patients with stage IV disease.

A Case-by-Case Count on Transpacific Flights

Dr Venook: Eighty percent of these had presented with stage IV disease. And nowhere in the case report forms did we collect this information. Somebody had to go in and look at every case.

Dr Kerr: That somebody must have been somebody low-level—a mug, a student, somebody of low standing.

Dr Venook: It was me, in fact. I couldn't very well ask anybody else to do it, but I did have my research coordinators help scan all of the charts in. I had a couple of long flights from San Francisco to Asia, and I took the data along with me on my computer. I dug in.

'Holy gosh, wow. Are you sure about this?' There's a huge 14-month difference, left versus right, in survival.

A thing I hadn't really appreciated was the preponderance of left-sided versus right-sided cancers in general. I'm thinking, "Oh, that was interesting." I learned a couple of things, and then about 2 months later, after I gave all of the data to our statistician, Donna Niedzwiecki, she sent me the results and said, "What do you think of this?" I said, "Holy gosh, wow. Are you sure about this?" There's a huge 14-month difference, left versus right, in survival. That just knocks your socks off.

Dr Kerr: She put that into a multivariate model and you took a count and still it came out.

Four Subtypes of Colorectal Cancer

Dr Venook: Right. As you would expect, the left-sided and right-sided diseases are different diseases, even by the demographics—by the characteristics of the patients. The distribution of sites of metastatic disease is different. In general, right-sided patients are older than left-sided patients. By many parameters they're different diseases.

One of the things we always thought was that patients with right-sided tumors may not live as long as those with left- because they are symptomatic later. The stool is liquid on the right side, it takes longer for them to get symptoms, and they have more disease. But we couldn't really tell that, and because this was retrospective we could not control for volume of metastatic disease. With metastatic sites of disease, though, the liver was much more likely to be the main site of metastasis with left primaries, and not the liver for right primaries. All of these things added up.

One of the things we had to figure out was where to put the transverse colon patients. In our analysis, we had 60-some patients with transverse colon primaries; 700-some with left and 290 or 300 with right. We took data from the transverse colon; if we put it to the right or the left, it didn't matter; it was literally right in between.

Dr Kerr: No man's land.

Dr Venook: Right. We didn't include them in the analysis. This is clear-cut: right, which is cecum to the hepatic flexure, or left, which is rectum to the splenic flexure.

More Evidence Accumulating

Dr Kerr: Gotcha. When you started, you made the observation on sidedness and you started to dig in. I guess you got RAS, extended RAS testing, and so forth.

Dr Venook: Correct. We have over 44,000 biospecimens from this study—many blood samples looking for circulating, cell-free DNA, for example. We are doing a whole host of analyses, and obviously sidedness is a surrogate for a biological difference; it's not like magical right versus left. And we are closing in on it. In fact, at the ASCO meeting in a couple of days, my presentation[3] will be followed by a presentation from MD Anderson.[5] It's not a randomized study, which would be more reliable, but a presentation of their own experience, pretty much categorizing and making clear what the issues are with right versus left.

There appear to be four different versions of colon cancer...and with each of them comes the difference in prognosis.

There is a recent paper in Nature Medicine[6] that looked at what they are calling the consensus molecular subtypes, or transcriptional subtypes—the expression arrays that are different. And there appear to be four different versions of colon cancer. What we are pretty sure we're seeing—and we have some of the data but not all of it—is that two of these subtypes tend to be on the right, two of them tend to be on the left, and with each of them comes the difference in prognosis.

Dr Kerr: I remember a thousand years ago we looked at the fidelity of mutant RAS, comparing primary hepatic metastases and pulmonary metastases.[7] We found that there was an identical incidence of KRAS in the primary and liver [metastases], but there's enrichment in the lung, suggesting that most of the hepatic metastases were because of geography, because of portal venous circulation. Are there left versus right differences in venous circulation?

Dr Venook: There are certainly differences; the vascular supplies are different. The surgeons know that very well. But the big difference is really the rectum versus the rest of the colon. I don't think that that accounts for it, although it might. The factors that account for it are probably the mutational status, which we think is different, and perhaps the methylation. We think right-sided tumors tend to be more hypermethylated.

Dr Kerr: Microsatellite instability. We know that it involves a tiny minority of the stage IV patients.

Dr Venook: We have looked at that and BRAF. Neither of them is common enough to explain the difference. I think what's most remarkable—if we look at the subsets, and you always have to be careful when you do a retrospective look at subsets—is that the patients who clearly did the worst are patients with right-sided primaries who got cetuximab in our study. Even if KRAS wild-type, they live only about 16 months in overall survival compared with left-sided patients receiving cetuximab who were KRAS wild-type, who live about 36 months. That's a huge difference. I thought it was too big and wondered if there was something specious. The investigators from FIRE-3, Stintzing and Heinemann,[8] were kind enough to provide me their data, which is now in press, looking at this question. And they also saw 20-month difference in left versus right with cetuximab.

Dr Kerr: A fantastic observation. You sitting alone on a plane unpacking the different genotypes—fascinating. What are the clinical implications for you now? Relatively simple, we're seeing patients in our clinic, we've got a rather limited toolset in terms of the drugs that we use. Any clues? Getting the actual mutations? That sort of thing?

Moving Forward, Making Sense, and Making Decisions

Dr Venook: The first clue is about the use of EGFR antibodies cetuximab and panitumumab in right-sided patients. Our study would strongly suggest—and other data I think will soon confirm—that patients with right-sided colon cancer should not get EGFR antibodies. I think we will figure out exactly what the genetic features are. This will allow us to understand RAS, because I conclude that we don't know what we're talking about in terms of how we try to medicate or target RAS. That's one thing. Clearly, as we move forward, we have to separate or stratify by the sidedness of the tumor.

I can conclude that we don't know what we're talking about in terms of how we try to medicate or target RAS.

The other issue is trying to explain other studies in the past that have not added up to see if they might make sense. For example, from the UK, the new EPOC study from John Primrose[9] found an inferior result of RAS wild-type patients who got cetuximab. This makes no sense. Might there be a preponderance of right-sided patients who happen to be included in that?

In FIRE-3[10] versus CALGB/SWOG 80405,[1] they found a 7.5-month overall survival advantage in FIRE-3. We had no difference. In their study, 77% of the patients had left-sided tumors. We had 67% left-sided tumors, and we are doing a weighted analysis to see if that might not account for the difference.

You move 5%-10% of patients from one side to the other and you change prognosis. This is both prognostic and predictive, at least in the subset of right-sided patients and the effect of cetuximab. What do you do about it? Do you not treat a patient with a drug? That's a tougher call. These are expensive drugs with side effects. And in your mix, this may very well tip the scales one way or the other.

Dr Kerr: Fascinating. Alan, thanks for that incredible walk through an empirical observation. I'm getting a really good feel for the connection of phenotype and genotype and how we can place that within our clinical practice. Thank you to everybody who's been listening, from Professor Alan Venook and myself, from this year's ASCO in Chicago. Thanks for watching, thanks for listening. For the time being, over and out.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: