1. For personal use. Only reproduce with permission from The Lancet Publishing Group.
THE LANCET Oncology Vol 2 October 2001642
Interview
Alfred Knudson and his two-hit
hypothesis
Interviewed by Ezzie Hutchinson
TLO: What first stimulated your interest in hereditary
cancers?
Alfred Knudson: When I was an undergraduate student at
California Institute of Technology (Cal Tech) I was really
interested in maths and physics, until I met two well-
known geneticists, Alfred Sturtevant and Thomas Hunt
Morgan. My life in biology began with genetics. I went to
medical school and became a paediatrician, and then took
care of children with cancer at the City of Hope Medical
Center.
I became rather fascinated with the idea that cancer was
due to somatic mutations. It was hard to study these
mutations then, before the era of molecular genetics. Since
my work was mostly in paediatrics, I was especially
interested in genetic events that occurred at an early age.
When I read about the idea that multiple mutations were
necessary for cancer, and that the estimates were that six or
seven events were needed to cause cancer in adults, I
thought that this couldn’t be true of children, because
some of them were born with cancer. It did not seem very
likely to me that cancer could ever be the result of just one
mutation, because otherwise everyone would get cancer
really early in life. Then I began thinking about the
possibilities of hereditary cancer. There were already data
published showing that retinoblastoma in children was
caused by a mutation in a gene that could be transferred
dominantly. However, there were a couple of reports of
cases (or an affected person) having a non-case child, and
that person having a case child. This could possibly be a
coincidence, but as the incidence of retinoblastoma is 1 in
20 000 that didn’t seem likely. These reports indicated that
inheriting this mutation was not sufficient for cancer. So
then I asked myself whether two mutations would be
enough.
We already knew that there was a hereditary and a non-
hereditary form of retinoblastoma, but the picture was a
little confused because there were patients who had both
eyes affected, but who didn’t have any family history of the
disease. In fact, it is usually a minority of the patients with
hereditary cancer who have a family history, and we now
know that this is due to a new mutation. There was a paper
from Holland in the late 1960s, in which the authors
pointed out that of the people who had bilateral disease,
50% of their children were affected, so these were
obviously mostly due to new mutations. So nearly all
bilateral cases are hereditable, but there are some
hereditary onesthat are unilateral. It takes two events in
both hereditary and non-hereditary cases; the only
difference is the timing of the first event – before or after
conception. My hypothesis explained why a child born
with the first hit inherited in all of the cells (including both
retinas) would be more likely to get cancer in both eyes,
and at an early age. I suggested that a child born without
the first hit would develop cancers at a later age, and only
in one eye, as the likelihood of having two events occurring
Dr Alfred Knudson trained as a physician at Columbia University and
gained his PhD in biochemistry and genetics from California Institute of
Technology. In 1971 he published his two-hit hypothesis of cancer
causation, which explained the relation between hereditary and non-
hereditary cancers, proposed a mechanism of penetrance in hereditary
cancer, and predicted the existence of tumour-suppressor genes. This
hypothesis is recognised as an important contribution to our
understanding of oncogenesis.
Knudson has worked at the Fox Chase Cancer Center, Philadelphia,
USA, since 1976. He was previously Director of the Fox Chase Institute of
Cancer Research and was the leader of their molecular oncology
programme during the 1990s. He is also Senior Advisor to the President
of the Center. He has received many awards recognising his contribution
to science, including the Albert Lasker Award for Clinical Medical
Research in 1998.
His main research interest is in the genetic alterations involved in the
origins of cancers. In particular, he has focused on genes involved in
retinoblastoma, other childhood cancers, and the phakomatoses. He and
his colleagues are also investigating chemoprevention strategies.
CourtesyofFoxChaseCancerCenter
2. For personal use. Only reproduce with permission from The Lancet Publishing Group.
THE LANCET Oncology Vol 2 October 2001 643
Interview
independently in both eyes during
the short period of time when retinal
cells are dividing during development
was very small. I published this two-
hit hypothesis in the Proceedings of
the National Academy of Sciences USA
in 1971. I made an estimate of the
distribution of tumours in the cases I
was looking at and calculated that the
mean number of tumours was about
three and that they followed a
Poisson distribution. This was nice
because it suggested that the second
event was due to random chance. The
distribution predicted that about
15% of germline carriers would not
develop any tumours, and that
indeed was compatible with
published data. I had a chance to
present my hypothesis at a small
meeting in Virginia before
publication and to discuss it with
some prominent geneticists, one of
whom, Jim Neel, communicated it to the journal.
I could not decide whether retinoblastoma required
one mutation in each of two genes or two mutations in one
gene, there was no way we could decide at that point. Later
on I thought that the answer was more likely to be
mutations in two copies of the same gene, and I mentioned
this in a review I published in 1973. David Comings came
to a similar conclusion at about the same time.
The idea that a tumour could occur as a result of a
small number of events made sense to many of us,
especially when one considers that kidneys, neural tissue of
adrenal glands, and the retinas of children, are growing
tissues in fetuses and very young children. For example, the
retina develops from about ten retinoblast cells, growing
up to about 108
cells in the eye of a human being. So there
are huge numbers of cell divisions taking place, and there is
plenty of time for a mutation to occur at the normal
spontaneous mutation rate. You can imagine a cell with a
mutation proliferating into a clone, and then one of those
cells having a second event in the non-hereditary cases.
When you think about it, probably the majority of us have
one-hit clones in our eyes, but the cells differentiated
before the second mutation could occur.
TLO: Did you realise then that your hypothesis was much
more widely relevant – to other childhood cancers and to
adult cancers?
Alfred Knudson: I realised that the work was important, in
that it explained retinoblastoma development, but I don’t
think that I had any idea of the role the RB gene would
have in other tumours. RB1 was the first tumour-
suppressor gene to be cloned (in 1986), and it is one of the
most important pathways we know about.
In 1971–72 Louise Strong and I wrote a couple of
papers on neuroblastoma and Wilms tumour, suggesting
that they probably had the same mechanism of
development. We now know a lot about retinoblastoma,
that both copies of a gene are mutated, but it has not been
that easy in the case of Wilms tumour and neuroblastoma.
It is quite amusing and frustrating to look at how little
more we know 30 years on. We have the WT1 gene for
Wilms tumour, but the problem is that it only accounts for
about 15% of Wilms tumours. In neuroblastoma, we don’t
even have a WT1-like gene – no mutated gene has been
cloned for its hereditary predisposition. These were things
we were thinking about in 1972 and 30 years later we still
don’t have the answers.
I also thought it would be helpful to try and understand
adult tumours, especially the carcinomas, because a
carcinoma occurs in a different kind of tissue – a renewal
tissue – and because they were thought to arise from
multiple mutations. In colon cancer, for example, a two-hit
benign lesion, an adenoma, is transformed into a
carcinoma through a sequence of events, made famous by
Bert Vogelstein and his colleagues.
TLO: Why is it that the patterns of cancer seen in people
with hereditary cancers are different from those of
spontaneous cancers?
Alfred Knudson: The pattern of cancers in hereditary
syndromes has been fascinating to me. There is an overlap
between hereditary retinoblastoma and Li-Fraumeni
syndrome, as they both have considerably increased
incidence of osteosarcoma and soft-tissue sarcomas. We
now know that the long-term survivors of hereditary
retinoblastoma get some carcinomas, while in the Li-
Fraumeni syndrome nearly all female carriers of the P53
mutation develop breast cancer by the age of 50. One
would like to know why the Li-Fraumeni syndrome is so
important for breast cancer, since only about 35% of all
breast cancers have mutant P53, but that figure is 90% in
colon cancer, a tumour that is not featured in the
2nd
mutation
1st
mutation
(a) (b) 2nd
mutation
1st
mutation
Knudson’s two-hit hypothesis for retinoblastoma. (a) hereditary cancer (b) non-hereditary cancer.
3. For personal use. Only reproduce with permission from The Lancet Publishing Group.
THE LANCET Oncology Vol 2 October 2001644
Interview
syndrome. Why is there such a difference? My own
personal feeling is that it has to do with the fact that the
breast is a vulnerable target in syndrome patients because
its cells are proliferating during adolescence, as also
happens for the target cells of the sarcomas.
By the age of about 50, about 50% of people who had
hereditary retinoblastoma as children have second cancers.
In the past, these cancers have been mostly induced by
radiation in the treatment of retinoblastoma, when doses
given were much higher than those given today. So the
incidence will be lower with current treatments, but some
people think it still might be as much as 25% at age 50
years.
TLO: Your model anticipated the future discovery of
tumour-suppressor genes. What do you think about the
possibilities of of using these as therapeutic targets? Is
there any work being done on RB as a target?
Alfred Knudson: There is a difference between developing
an anti-oncogene product and an anti-suppressor gene
product. With an oncogene there is an overexpressed gene
and there is some hope of inhibition, which has been very
successful in some cases, such as chronic myelogenous
leukaemia. P53 gene therapy is being developed, but I’m
not sure whether anyone is working on RB specifically. One
might speculate that a tumour-suppresser gene is
regulating an oncogene and that the loss of this
suppression permits the oncogene to be overexpressed, as
proposed by Comings in 1973. One could then try and
suppress that oncogene, rather than trying to replace the
tumour-suppressor gene. It is more likely that tumour-
suppressor genes will be useful as targets for prevention
research.
For some reason, childhood tumours do not seem to
lose apoptotic mechanisms in the way that adult tumours
do. This means that, if you give chemotherapy, apoptosis is
readily induced. This is also true for testicular cancer. In all
of these cancers, the incidence of P53 mutations is very
low. In addition, all of the target cells grow rapidly and the
presenting tumours are much more homogenous
genetically than the typical adult carcinomas.
TLO: What is the current focus of
your research, and what are you
most excited about at the moment?
Alfred Knudson: Well, I am close to
retirement and not maintaining a lab
myself anymore, but I am the
Principal Investigator on a project
that concerns the effects of putative
chemopreventive agents on early
somatic events occurring in people
who are genetically predisposed to
cancer. With a tumour such as
retinoblastoma, if the two events are
caused by spontaneous mutations the
prospects for treatment are good, but
the possibility of prevention is not
very promising. At what point are you
going to introduce prevention strategies? How are you
going to prevent retinoblastoma in a newborn? In adults,
however, where tumours are caused by multiple events and
there is considerable heterogeneity at the time of the
presentation, we haven’t really had much success yet with
chemotherapeutic agents. On the other hand, the larger the
number of events, the greater the opportunity for
preventive measures, especially if we can pinpoint those
events and concentrate on the earliest of them. We think
that knowledge of hereditary cancer in adults could be very
helpful to us, because we can identify people who are at
high risk of cancer, who have inherited one event.
Furthermore, the knowledge gained may be useful in the
prevention of non-hereditary carcinomas because the
pathways are often similar to those of the hereditary forms.
That is what we are focusing on now – it is one thing to try
a preventive agent on the general population for a cancer
that is going to affect only 1 in 50 people and might not
give us a result for 60 years, but it is a different matter to
use it on somebody who has a mutation that confers an
almost 100% risk of getting a tumour by the age of 40.
Some tiny risk that might not be acceptable for the general
population could be acceptable for this group. There is
considerable interest in assessing putative preventive
agents in people with a genetic predisposition to cancer,
and we are particularly interested in finding one that would
act at the first event in these cases.
In particular, we are looking at two forms of hereditary
colon cancer: familial adenomatous polyposis (APC
mutation) and hereditary nonpolyposis colon cancer
(MLH1 mutation). We know that the polyps are two-hit
lesions, but we also know that many changes are occurring
in the polyps. There is some concern that the genetic
changes you could measure are secondary and not
necessarily relevant; they might be incidental mutations.
We hope to find some difference in the target tissue before
the polyp forms, as a consequence of the first somatic
event. There is evidence of agents such as sulindac causing
regression in polyps, both in mouse models and in human
beings, so obviously there is interest in knowing how they
work. Similarly, tamoxifen is being investigated as a
preventive agent in breast cancer.
Fox Chase Cancer Center, Philadelphia, USA.
CourtesyofFoxChaseCancerCenter
4. For personal use. Only reproduce with permission from The Lancet Publishing Group.
THE LANCET Oncology Vol 2 October 2001 645
Interview
TLO: A couple of months ago, TLO published an editorial
on genetic testing, calling for caution when offering tests
if their predictive value is low and there is no treatment
available (Lancet Oncol 2001; 2: 325). What is your view?
Alfred Knudson: We have some faith that we might be able
to develop a rational approach on the basis of what we
know about molecular genetics and cancer, but we don’t
want to put people at unnecessary risk. If we cannot do
anything at this time, we should certainly not pressure
people into undergoing genetic testing against their will. If
we could offer something in the way of a possible
preventive agent, for example, then people might wish to
be tested. Children, of course, should not be tested unless
parents specifically want it done.
If we found that an agent was very effective in reducing
the prospect of carcinoma in polyposis patients, and the
necessity for coloectomy, and we know that polyps begin to
appear in large numbers at about the age of 10, it might
mean that the best time to introduce such an agent would
be the first decade of life. For example, we learned long ago
that we could ameliorate the course of disease in
phenylketonuria if we tested for it very early in life, so
we’ve already done that in medicine.
Over the past 10 years, we have had intensive prostate-
specific antigen testing of men and yet the mortality from
prostate cancer in the USA hasn’t changed that much. The
case fatality rate has gone down, but the case incidence has
gone up drastically. So we are now diagnosing more cases,
but to a large extent we are diagnosing the ones that do not
succomb in the interval of time we have been studying.
Researchers are now trying to identify a subgroup with a
very rapid course on which to focus their attention.
TLO: In 1995 you were appointed as special advisor to
Richard Klausner at the National Cancer Institute. What
did this role involve?
Alfred Knudson: I was fortunate in being able to spend a
very interesting 4 years at the National Cancer Institute
(NCI). When Richard Klausner took over as Director of the
NCI in August 1995, he decided that his executive
committee should include some outside part-time
advisors, and he persuaded some of us to join him,
including Ed Harlow, David Livingston, and George
VanderWoude. Klausner is very bright and he has a very
broad knowledge base – full of energy and a terrific leader.
We knew that he was going to initiate exciting programmes
to advance the translation of all the new information
coming from molecular biology and genomics research
into something really beneficial, and it was a time of great
excitement. So we had these regular meetings, pursued
leads, called advisory boards together, and so on. Many
people in the oncology community took part in Klausner’s
initiative and it was a remarkable experience for those of us
who were involved.
It was very productive. For example, a national Cancer
Genetics Network, the Genome Anatomy Project, and the
programme of Mouse Models of Human Cancers were
established. Even more importantly, people now have more
confidence that we can actually do something practical
with all the advances of the past decade.
At the same time, I was also actively working within the
Division of Epidemiology and Genetics, led by Dr Joseph
Fraumeni. During that time we expanded the programme
in genetics, especially in the area of interactions between
genes and environmental agents.
TLO: You have worked at the Fox Chase Cancer Center
for over 20 years. Can you tell us a bit about the set up
there, and what makes it such a good place to work?
Alfred Knudson: I will have been here for 25 years this
December! Fox Chase is a free-standing Cancer Center, and
it has a historic base in basic science. It was set up by an
unusual pathologist, Stanley P Reimann in 1927, who had
the idea that we should try and learn more about growth
and development generally, including cancer; he really
believed in this and actually started a new journal called
Growth. Looking back, I think he must have been one of a
very small number of people who saw this connection and
believed in it. Since then, the Institute has grown
enormously and has three divisions: basic science, clinical
science, and population science. This is a very nice balance,
and in fact, this is now how Klausner has organised the
NCI. I think it is a natural pattern of organisation. All the
divisions work very closely together; for example, we have a
human genetics group here, and its members are from all
three divisions.
TLO: I believe that your wife is also a paediatric
oncologist – do you collaborate on projects?
Alfred Knudson: Yes, Anna Meadows and I have been
married for 25 years. She was a paediatric oncologist when
I met her. So we have considerable areas of overlapping
interest and we have written a few papers together.
Her principal interests have been the aetiology of
childhood cancer and the late effects of cancer and its
treatment in children. When I was at the NCI, she was
appointed by Klausner to develop a new Office of Cancer
Survivorship. Klausner felt that while much had been done
to study childhood-cancer survivors, the outcome of
surivors of adult cancers was not so well known. The Office
was created to promote research and education, dealing
with both adult and childhood-cancer survivors. The
success of this Office led to it becoming a permanent part
of the NCI, and more research resources to study survivors
were allocated. She returned to the Children’s Hospital of
Philadelphia, where she had been the chief of one of the
worlds largest childhood cancer programmes and where
she continues to direct the Follow-up Program. She is also
active in a national study of more than 14 000 childhood-
cancer survivors and has developed a programme for
survivors of breast and testicular cancer at the University of
Pennsylvania Cancer Center.