Albert Lasker
Clinical Medical Research Award

An Interview with Alfred Knudson
Interview by Dr. Richard Klausner

Dr. Richard Klausner, Director, National Cancer Institute of the National Institutes of Health, interviews Alfred Knudson.

Part 1: An Exploration of the Origins of Cancer
Dr. Knudson recounts his early interest in the origin of cancers and in particular the question of their hereditary nature. With that focus, Knudson then applied genetics to one tumor he found especially interesting, the childhood tumor retinoblastoma (RB).

Klausner: Well let me ask you a few questions. Actually I gather that this is going to be used for a whole variety of audiences who might like to hear about you and your work and how you came to do it. So I wonder if you would actually just give me your own description of the essence of the work you are being honored for.

Knudson: I started out by being interested in genetics and embryology and then went to medical school. I thought that if I went into a clinical field it would probably be pediatrics, keeping an interest in genetics. When I came across the childhood cancers, I thought, "This is a really interesting problem: these very bizarre developmental tumors." Embryonal tumors were especially interesting to me.

But in the meantime, I got interested in the origin of cancer generally, viruses and so on. I was attracted by the fact that some cancers clearly exist in hereditary form. The more I looked into it, the more I discovered that virtually every cancer has a subset that is genetically determined. One of the most striking ones for me, and even today for a lot of people, was the childhood tumor retinoblastoma, which has a very long history of being recognized as having an hereditary form.

So then the question was: "What does this mean? What is it that is being inherited? How does it work? How does it do this, or how do any of these genes work? How does the hereditary form of cancer relate to the non-hereditary form of cancer?" It was clear that some people could have the retinoblastoma gene and not get the tumor. This was known because a person might have an affected parent and an affected child but not himself or herself be affected. It was clear that it was not a sufficient condition to inherit the gene; something else had to happen. Of course we didn't know whether it was environmental or genetic or other.

We also knew that the hereditary form was most often bilateral, meaning multiple tumors were formed. We had a tumor that is occurring [at a rate of] one in 20,000, so its occurrence in both eyes made it obvious that there is something special about the person.

So when I appeared on the scene, we were faced with this fact that the bilateral cases were multiple, and, in general, hereditary cases were multiple. The non-hereditary ones were usually not. And the ages of the bilateral cases and hereditary cases in general were younger than the ages for the non-heritable form. I found that when I plotted the log of the number of the cases not yet diagnosed at a given age versus the age, I found a straight line for the bilateral ones, suggesting one event after conception, and a curved line that was compatible with two events for the non-hereditary form. That invites the idea that they both involved two events, the difference being that the first event was inherited in one case and not inherited in the other.

Then the question was: "What kind of an event is it?" Ninety-five percent of the carriers get the tumor. So, at the level of the organism, it is a very common event. Carriers get on average three or four tumors. (Our calculation at that time was three tumors.) I then thought about the retinoblast cells that are being affected by tumor. It was pretty obvious that there are millions of these at one point during development. So three divided by millions means it is rare at the cell level, and that kind of a number invites the idea that it is a somatic mutation.

I thought, "Well, these events are probably both mutations: germline and somatic in the hereditary's, and two somatic in the others ones." Then, of course, there was the problem: "What are they occurring in?" There was nothing that would tell us whether it's one mutation in each of two genes or two mutations in one gene that affect both copies of it. But for a geneticist the latter is much more interesting because it is simpler. It suggests that it might just be a one-gene disease, so it would be a dominant trait at the level of inherited susceptibility, but a recessive trait at the level of the cell and carcinogenesis.

That also led to the idea that as long as a person had one copy of such a gene he or she would be okay. So the normal copy could be viewed as what I called an anti-oncogene, now called a tumor suppressor gene; these would be quite different from oncogenes that were being recovered from tumor viruses, and the host protooncogenes uncovered by Bishop and Varmus in the middle of the seventies.

Part 2: Snagged by Genetics
Initially not a bit interested in biology, Knudson describes a change during his first years at Cal Tech when he was "snagged by genetics." He also discusses the difficulty of applying research from non-biological fields to biological research.

Klausner: It is so pleasurable to hear you say it. No matter how many times I hear it, I cannot but smile listening to it. But your telling of it raises lots of different things about the science and about you. You describe this wonderful way: "Well, I thought about this and then I thought about that," as if that sort of happens automatically. But it is so clear when listening that you brought to this a wonderfully scientific, inquiring, question-asking mind. Tell me about that. Tell me where that came from.

Knudson: I don't know. I've always wanted to know how things worked. I think this must have happened to you, too. I am not sure. When I was an undergraduate at Cal Tech, I had never had biology in high school, and was not at all interested in biology. I wanted to do either mathematics or physics. And then I got derailed by this guy named Sturtevant, in the department run by a guy named Morgan, who taught this course named genetics. And I thought: "Oh my God this biology stuff is not so bad after all."

Of course genetics was unique at that time among the biological sciences because it had a precision about it that wasn't customary in biology. I think you don't recall it personally; you are too young. But you know history well enough to know that in the pre-DNA era, that was about the way things were. Genetics was the only biological science that connected readily to what we know now or what we do now. So I was really snagged by genetics.

I think physics, the way physicists think, was put into the back of my mind too. Physicists like to analyze things to death, and you can't do that unless they are somewhat simple. And I love some of the physicists who take the idea that if an explanation is complicated, it is wrong.

Klausner: Not elegant.

Knudson: Not elegant. Right, I love their use of the word elegant. I got caught on that.

Klausner: That is wonderful. You bring up this point: your work brought both an analytic and a mathematical approach to bear on a central problem in cancer. What is your view of the role of non-biological fields in biologic research?

Knudson: I think there are not too many opportunities that come along because most biological phenomena are so complicated. We are seeing some realization of that crossover in molecular biology, genetics, and, perhaps, in physiology in certain systems. For example, I knew an aeronautical engineer at Cal Tech who moved down to UC San Diego some years ago and started working on blood-flow. You stop to think: "Oh, of course, fluid flows- fluid mechanics" So it comes slowly. Genetics is still complicated but maybe less so than some other subjects. I think everybody actually should read Mendel's paper, it is really so beautiful, and he got us off to such a good start.

Klausner: So this would be a strong recommendation of yours.

Knudson: Yes.

Klausner: And should we be looking for more Cal Tech graduates for biologists?

Knudson: Well, I think Cal Tech is not unique any more in this respect. You would probably agree that what happened in the post-World War II era is that biology became a physical science.

Klausner: Yes, and that is one of the things that is so beautiful about it.

Knudson: I think Cal Tech was special, because, at the time that I was a student there, as an undergraduate, it was already doing that to a large extent, and that was very, very uncommon at the time. It was because genetics was the discipline that drove the department; I mean, Morgan was the first head of the department.

Part 3: Surprised by the Impact of His Discoveries
Knudson sensed that the RB gene had a role in hereditary cancer, but he was surprised by the magnitude of its importance. Here, he attempts to describe his thought process in deciding which scientific paths to take.

Klausner: Has the impact of your work surprised you?

Knudson: Yes, although I had conviction that it was potentially important, and felt that the specific gene, RB, would be the best gene to attack first. Not only did we have some understanding of it, but, from our work and others', we knew that there were a few deletion cases where a piece of chromosome 13 was missing. That pointed very clearly to where to look. So as soon as somebody could dig up some markers for linkage, we knew that it would be a good gene to go for as a human cancer gene responsible for hereditary cancer. And of course it worked out that way, as you know.

Klausner: Do you recall at the time that you published that famous paper in '71 that you had any idea foreseeing what was to come as a result of this work?

Knudson: No. Although I thought it would be fascinating to explain retinoblastoma—or any one tumor. One could hardly be prepared for its importance. We now know that it is somatically mutated in all kinds of tumors that aren't found in people who have hereditary retinoblastoma. Of course, we think the explanation is that those tumors are very complicated and require mutation of other genes too. So the number of events to get those tumors in a person who has inherited the retinoblastoma mutation is too high for a lifetime, so to speak.

But I think it is just fascinating that RB not only turned out to be the first tumor suppressor to be cloned, but to be one of the two or three most important pathways we know about.

Klausner: Yes, pretty amazing it is. So it is interesting again, that when you describe the path and the process of reaching your conclusions, you describe a lot of: "Then I thought this, then I thought that and I realized…"

Knudson: Sorry about that.

Klausner: No, no, no, it is wonderful because it is the most wonderful thing about the story. But I'm curious, do you recall any particular moments of either insight or conversation or observation of a paper that fundamentally clarified things for you?

Knudson: Well, I was thinking about this question of the number of events, and I didn't like the hypotheses that said, "We are looking at age-specific incidences, and if every event occurs with equal probability and without any intermediate growth advantage, we can use the slope to decide how many events or how many events are necessary. So if we have a slope proportional to the sixth-power of age then that suggests that seven events are involved. "

So I thought: "This is crazy. There is no basis for that opinion at all. But I can't argue with them on a complicated tumor. Let's go to a real simple tumor; if we can have a tumor in a newborn baby, it has to be about as simple as one can have." So I wanted to know how few the events actually could be.

The problem was clouded because so many of the bilaterals didn't have a family history. But then a paper was published in 1969 from Holland. One investigator there had enough cases that had survived bilateral disease that he could tell from the offspring that 50 percent were affected. So these were obviously in most cases new mutations. And that cleared up the genetic part right away.

Klausner: So that paper was really important?

Knudson: Yes, that paper was important. Then there was a paper from England by a radiotherapist who was treating patients with retinoblastoma. What he did was to record how many tumors he saw in an eye.

So I was attracted by this finding and thought, "Look at the distribution of tumor number." I went through the records at MD Anderson, and there was an equally careful ophthalmologist there. Between those two I was able to assemble some 60 or so cases in which there was information on one eye at least. From that, one could calculate that the mean number of tumors was about three and they followed a Poisson distribution. So it made it inviting to suppose that the second event was a matter of chance, and would fit with spontaneous mutation.

Part 4: Past Influences and Challenges Ahead
Knudson recalls the intellectual and emotional influences that drew him into genetics. He also lists some of the important challenges ahead for cancer genetics and his hopes that the discovery of some common thread will lead to at least finding ways to delay the onset of cancer.

Klausner: Who were the major influences, either colleagues or teachers, in your work?

Knudson: Well it is hard to know those things. There are a couple of kinds of motivation. One is intellectual and one is emotional. I think my early experiences in genetics were irreversibly intellectual and emotional. I also had a really great interest in embryology, but it wasn't as precise a science.

So, an approach through genetics was almost inevitable for me. My first paper was a genetic analysis of adrenal hyperplasia, proposing that it was an inborn error in metabolism. That was in 1951. So I was interested in genetics right from the beginning.

When I went into pediatrics, I had a one-year residency at New York Hospital, and they had us rotate through Memorial Hospital because they didn't have any residents over there. So the New York Hospital residents served. Can you imagine that? Those were the days when pediatric oncology was just getting started. So I spent a month over there, and it made a deep impression on me. "How can these little kids—one-, two-, five- and six-year olds—get cancer?" I became very interested in how that could happen.

So later when I finished, I went back to Cal Tech and got a Ph.D., and my first job was the City of Hope Medical Center, where the children had cancer. I really then started to be very interested in cancer. My initial interest was in leukemia. I had an idea—well, a lot of people had an idea—that it might be due to a virus. I was working on it, but that didn't lead anywhere. So I said I am going to give up leukemia and look at the solid tumors.

Klausner: Which I imagine almost no one was working on.

Knudson: That is right. There was very little going on at that time. This was about 1960.

Klausner: So from your perspective now, what would you consider today's great challenge in cancer genetics?

Knudson: First of all, we'd like to get whatever other genes there are that are responsible for hereditary cancer. Because we've learned very well that these genes aren't important just for hereditary cancer. They are important for non-hereditary cancer, common cancers, too.

Then [we need to know] more about why is it that some people with such a gene don't get cancer, and others may get several cancers with the same gene. What is the reason? Is it environmental, or is it genetics? If we see that a cancer gene sometimes doesn't lead to cancer, then this invites the proposition that we might be able to do something to intervene, to delay or even prevent the onset of cancer even in those people that have inherited the gene. It wouldn't have to be prevention; it could be a 25-year delay in the onset. Then anything we learn from them could be useful for at least some of the non-hereditary cancers.

Klausner: Do you imagine that in those challenges, there are simplifying and unifying assumptions and realizations like the one you gave us. Or do you think that now we are sort of getting into the complexity of biology?

Knudson: We are clearly in the complexity of biology—that is for sure. I am not working in a laboratory anymore. One of the things I have been doing is thinking about all of the genes that we have in our hands—the 30 or so genes that have been cloned that underlie hereditary cancer—and thinking about them by category. Some seem clear-cut, like the DNA mismatch repair genes, which can increase the mutation rate drastically, 500- or 1000-fold, so transit through a path to cancer is going to be faster. The mutations in genes like RB and P53 can interrupt the cell cycle and apoptosis to produce an increased birth rate and decreased death rate of cells and so lead to cancer.

But there are other genes like the polyposis and neurofibromatosis genes that affect signal transduction. But that is a little too facile to say. If they interrupt signal transduction, why do we have the tumor specificity that we see? What are the differences in these genes? Do they all impinge on some final common pathway and does the tumor specificity reside in the fact that in different tissues, there are different redundancies so that a particular gene may have a redundant helper in some tissues but not in others?

I would like very much to have us compare these genes like von Hippel-Lindau and tuberous sclerosis genes, for example, and see what they have in common. Is there something we can see among these? Now whether that is going to lead to some simplifying idea, I don't know. But I think it is an interesting thing to look at.

Klausner: So, despite talking about the complexity of biology, you are thinking about trying to find some simplifying and unifying rules that may well lurk under there. I suspect everything always seems incredibly complex until you find their rules.

Knudson: That is right. You could say, "Well why do we worry about simplifying it?" I think it is disheartening to think that we have 100 kinds of cancer and that there is no carryover on our knowledge from one to another. You start thinking about how to intervene—whether it is prevention or treatment. It is a little discouraging. Whereas if we can find some thread that runs through it all, maybe we can get a theme and variations rather than various numbers of themes.

Klausner: Right, right. So you see that as one of the real challenges ahead of us.

Knudson: Yes I do, although it sounds a little pretentious.

Klausner: Not to me it doesn't. Well this has been very enjoyable for me.

Knudson: Well it certainly is an exciting time, you have to admit that.

Klausner: You know how I feel about it.

Knudson: We look back at the beginning of the 90's in the field that we are talking about, tumor suppressor genes, and as of 1990 there was only one that was documented and proven.

Klausner: Now it has exploded.

Knudson: Now it is a whole bunch and growing. It really is an exciting time.