Common susceptibility genes for cancer: search for the end of the rainbowBMJ 2006; 332 doi: https://doi.org/10.1136/bmj.332.7550.1150 (Published 11 May 2006) Cite this as: BMJ 2006;332:1150
All rapid responses
Response to letter by Professor Luzzatto regarding “Common susceptibility genes for cancer: search for the end of the rainbow”
We thank Professor Luzzatto for the letter
1. Based on the studies we had cited, we start with the premise that
the primary cause of cancer is a defect in communication from the stroma
to the parenchyma. Under this premise we do not think it likely that an
inherited mutation in the parenchyma would have much impact on the risk of
cancer. We do agree that an inherited mutation in the stroma has the
potential to increase the risk of cancer, and for that reason
investigators should take samples from the stroma, which is not usually
the case. However, we cited one study that did not find such inherited
mutations in the stroma. This is an area where more research is needed
2. We agree that, within a population, some individuals are at
greater risk of cancer than others. For example persons exposed an
environmental carcinogen or with a family history of cancer would be at
higher risk of cancer than other persons. Perhaps the question is better
phrased as “Why do some individuals get cancer and some do not although
their environmental exposures appear the same and their family histories
of cancer appear the same?” Of course the nub of the question is whether
or not the environmental exposures and family histories are, in fact,
precisely the same, which is not possible to determine, as environmental
exposures may be overlooked or cancers in other family members may be
imprecisely determined or may be due to environmental exposures.
3. We agree that our twin study estimate of a “small to moderate
contribution” of genes to the incidence of cancer is still a contribution.
However, a key point is that our twin study also estimated a high
penetrance for these genes as would be found in the family studies
mentioned, not the low penetrance common genes that were the focus of our
4. We agree that there are methods for searching for low penetrance
cancer susceptibility genes besides genome wide high throughput studies.
This does not change our opinion that these studies are expensive, because
much of the cost is due to the very large sample sizes that are needed,
regardless of the technology. The low penetrance cancer susceptibility
genes mentioned were identified based on much smaller sample sizes than we
recommended. Until these genes are confirmed in larger studies, we are
skeptical that they are truly associated with cancer.
5. We do not understand why the existence of high penetrance rare
cancer susceptibility genes would imply the existence of low penetrance
common cancer susceptibility genes. Also in the “compelling review
regarding colon cancer”, 95% confidence intervals were reported for the
estimated association between the genes and cancer, but, as mentioned in
our article, to reduce the chance of false positives we recommend a much
higher level for the confidence intervals. Regarding the benefits of
chemoprevention, as discussed in our article, the identification of the
common cancer susceptibility gene would need to lead to the identification
of a modifiable risk factor and its evaluation in a large randomized
trial, which is a long-term and expensive undertaking.
Competing interests: No competing interests
Failure of cancer genetics; Are we ready to accept the argument for transfer of resources to observational and experimental research in cancer biology ?
Baker and Kaprio have reviewed the status of clinical benefit from
cancer genetics(1). We would like to congratulate them for demonstrating
the fallacies of this approach in a systematic manner. This is quite
obvious also at common sense level. But genetics appears to catch both
professional and lay imagination like nothing else in cancer biology. This
situation is like use of Lasers in medicine. The mere words like genes and
lasers have profound effect. The vast amount of resources are spent year
after on cancer genetics without clear long term broad objectives. It is
no surprise that a month after Baker and Caprio’s paper in BMJ there is an
article in a national newspaper perpetuating the myths of value of cancer
genetics(2)There is clear desire here to simplify a complexity as Mr
Wishart has tried to explain.
We believe that more emphasis needs to be given to the the fact that
a cancer is a dynamic process( made up of several very well co ordinated
sub processes). These processes are being dealt with mostly in a clinical
setting. There are structural changes within the host of a cancer and in
the cancer itself which are used as clinical, investigational and
laboratory signs. Cancer genetics is study of one important aspect of the
structural changes and will always be of the value for what it is worth
under such complex circumstances.
Various structural attributes of a cancer e.g. histology, and grades,
tumour stage; clinical and imaging information, serological estimation of
tumour activity only give us a cross section or snapshot information
regarding a cancer. Trillions of dollars and pounds have been wasted in
the reductionist approach in cancer research. Common susceptibilty gene
approach is an example of this misguided direction of research. This
approach grossly underestimates the environmental infulences. The
separtion of the two in a practical world will remain improbable until we
have at least few starting reliable and reproducible equations. At present
unfortunately we have none.
What we have not explored is treating each cancer as an individual
process using the biological or observational approach(3). This may
involve more respect for the cancerous phenomenon than is being given at
present. For example we may have to visualise a cancer as an organism
within an organism with at least some autonomy. For example carrying out
serial MRI spectroscopy, PSAs and serial targeted biopsies in prostate
cancer over a peroid of time will help us to assess the individual
behaviour of a cancer.. That will be one type of such assessment.
Additionally provocative in vitro assessment is possible using the
stimulus and response model. In vitro response to various investigations
and treatments may give us information which may have more value than the
static tests. The information so gained will be akin to a multimedia file
compared to the snap shots assessment which are carried out at present.
What is not being appreciated is that time is a compulsory variable which
is being omitted too often to simplify various equations Omission of time
leads to making imperfect impressions about a dynamic process which leads
to n number of probabilities and rarely concludes to a certainty. Bringing
time back into these equations will pose problem. There will be obvious
ethical questions in an assessment where time is allowed to elapse after
diagnosis without treatment. Applying thinner slices of time using ultra
sensitive measurments and using examples of slow growing cancers e.g.
prostate cancer may provide some answers to these ethical dilemmas.
Relating the audio-video file. of each patient with their genotype and and
their environment may open newer avenues of cancer research. As clinicans
we use this approach every day processing real time information on our
patients, but our measurements tools are rather crude. And we do not
record what we measure all the time. Refining these tools may lead us to
insights regarding hard facts not yet imagined by us. But for present we
should focus resources in clinical research and slow down on orthodox
1. “COMMON SUSCEPTIBILITY GENES FOR CANCER: SEARCH FOR THE END OF THE
RAINBOW” Stuart G Baker, Jaakko Kaprio BMJ 2006;332:1150-1152,
2. “A CURE FOR CANCER” lead story, Nigel Hawkes and Adam Wishart, The
Times, Times2 l No 68728, Friday June 16 2006; 4-5.
3. "POST-BIOPSY RISE IN SERUM PSA: A POTENTIAL TOOL FOR DYNAMIC
EVALUATION OF PROSTATE CANCER/PROSTATIC INTRAEPITHELIAL NEOPLASIA" (PIN)
Shiv Bhanot, R. Gopalakrishnan, R.T.D. Oliver. Cancer Biology &
Therapy 2.1,59-62,Jan/Feb 2003
Competing interests: No competing interests
Baker and Kaprio1 (BMJ of 13 May 2006) have given a serious critical analysis of the value of searching for low penetrance cancer susceptibility genes, and they conclude that the game is not worth the candle. I think this conclusion should not be allowed to go unchallenged.
In essence, all the points made in the paper can be rebutted.
1. Early changes in cancer may be in the stroma rather than in the parenchyma. Whether this is true or not (the authors themselves seem to have doubts), it is irrelevant: an inherited mutation could increase the risk of somatic mutation in either a stromal cell, or a parenchymal cell, or both.
2. Migration studies are obviously informative regarding environmental causes of cancer. They say nothing about whether, within a population, some individuals are at greater risk of cancer than others.
3. With respect to the evidence from identical twins, the argument is of a quantitative nature: a ‘small to moderate contribution’ is a contribution. Of course the inherited contribution observed in identical twins is supported by numerous other studies on the cancer risk in relatives of patients with cancer (see for instance 2 3). In standard oncology textbooks a family history is ranked as a high risk factor for prostate cancer and other cancers.
Baker and Raprio rightly express concern about the fact that finding low penetrance cancer susceptibility genes will be hard and expensive. However, this issue is of a different nature, and this is not the only case where setting priorities in research is complex, whether in the cancer field or in other areas. It is also important to note that genome wide high throughput studies are not the only approach: the discovery that TGFBR and HDM2 as low penetrance cancer susceptibility genes 4 5 are recent examples.
Finally, I must make a conceptual argument. Baker and Raprio do not question that ‘rare genes related to familial cancer’ do exist. These genes have, by definition, high penetrance: but still their penetrance is variable, ranging from about 85% (e.g. BRCA1, BRCA2) to about 30% (e.g. in ataxia-telangectasia). Considering this, it would be quite surprising if other genes with even lower penetrance (e.g., between 1 and 30%), did not exist. Of course it is impossible to predict whether and in what cases identifying such genes might have clinical implications (but see a recent compelling review regarding colon cancer6. It is certainly not impossible that, for instance, genes that cause a slightly increased rate of somatic mutation may be accessible to some form of chemo-prevention.
1. Baker SG, Kaprio J. Common susceptibility genes for cancer: search for the end of the rainbow. Bmj 2006;332(7550):1150-2.
2. Narod SA, Stiller C, Lenoir GM. An estimate of the heritable fraction of childhood cancer. Br J Cancer 1991;63(6):993-9.
3. Cannon-Albright LA, Thomas A, Goldgar DE, Gholami K, Rowe K, Jacobsen M, et al. Familiality of cancer in Utah. Cancer Res 1994;54(9):2378-85.
4. Pasche B, Kolachana P, Nafa K, Satagopan J, Chen YG, Lo RS, et al. TbetaR-I(6A) is a candidate tumor susceptibility allele. Cancer Res 1999;59(22):5678-82.
5. Bond GL, Hu W, Bond EE, Robins H, Lutzker SG, Arva NC, et al. A single nucleotide polymorphism in the MDM2 promoter attenuates the p53 tumor suppressor pathway and accelerates tumor formation in humans. Cell 2004;119(5):591-602.
6. de la Chapelle A. Genetic predisposition to colorectal cancer. Nature Reviews Cancer 2004;4(10):769-780.
Competing interests: No competing interests
Response to letter from Professor Lubinski regarding "Common susceptibility genes for cancer: search for the end of the rainbow"
We thank Professor Lubinski for the letter.
(1) We commend Professor Lubinski for collecting and maintaining a
large biobank relatively inexpensively. There is a still a cost for the
genetic tests, which may be high. We disagree with the claim the costs of
clinical trials for late-stage cancer drugs are an order of magnitude
greater than for detection and prevention studies. Early detection and
prevention studies require tens of thousands of subjects followed for many
years, while clinical trials for late-stage cancer drugs are much smaller
and shorter in duration.
(2) We commend Professor Lubinski for obtaining large sample sizes
for the studies of association between common genes and cancer. (Because
it was not clear from the letter, we want to point out that the sample
size for each particular cancer investigated needs to be large.) However
some of the genes listed in Table 1, such as BRCA1, are not the common
susceptibility genes that we had discussed. The high relative risk of 4
for one CHEK2 variant also makes us wonder if this gene runs in families.
Considering the most recent paper cited, Debniak et al, International
Journal of Cancer, 2006 118, 3180-2, the smallest p-value in their Table 1
was .0004 which is about eight times larger than the type I error we had
computed in our paper of 0.0000474 (with a prior probability of an
association of .001 and a false positive reporting probability of .05). In
any event, large confirmatory studies will be necessary to determine if
these preliminary estimates of associations between common genes and
cancer represent true or false positives.
(3) Detection of premalingnant lesions or early cancers does not
necessarily save lives, as some of these may never cause medical problems
in a person’s lifetime if undetected and untreated or simply because early
treatment may not be effective. Furthermore there are harms associated
with early detection programs such as unnecessary biopsies and sometimes
application of toxic therapy when treatment is not necessary. Based on
results from randomized trials, mammography at age 50 is generally
recommended for women regardless of the results of a genetic test.
Without data from a randomized trial, it is not possible to clearly know
if benefits outweigh harms in breast cancer screening among younger women
with a particular genetic variant.
Competing interests: No competing interests
Response to the article „Common susceptibility genes for cancer: search for the end of the rainbow” BMJ 2006; 332; 1150-1152 by SG Baker, J Kaprio.
Response to the article
�Common susceptibility genes for cancer: search for the end of the rainbow� BMJ
2006; 332; 1150-1152 by SG Baker, J Kaprio.
I read with
great interest this article, as the common susceptibility genes for cancer
constitute the primary focus of my research and clinical practice. Since 1992,
I have been managing the
Research data and clinical practice experience of the Center suggest
that the conclusions of the above article are unduly pessimistic.
- Cost of studies to detect
common cancer susceptibility genes
Authors claim that these studies are expensive because they require
large sample size. In our center during the last 15 years we have been collecting
biological samples (mainly peripheral blood) and clinical data from over
135,000 cancer patients, their relatives and appropriate controls. Our costs of
collecting and maintaining what is probably the largest bio-bank in the world
are less than 1 mln EURO annually. Our studies are part of our normal research
and clinical practice work. �. Obviously,
the costs in
considerably smaller than in Western countries. Nevertheless, we can conclude
that collection of samples and data is not a limiting factor of discussed
studies. Furthermore, costs of detection and prevention studies should be
compared to the costs of clinical trials for late-stage cancer drugs, which are
orders of magnitude more expensive.
- Common cancer susceptibility
genes certainly exist
We have identified several common cancer susceptibility genes,�� Studies of consecutive cases allowed us to
identify cancer susceptibility markers for 85% of breast cancers [3,500 cases
dgn <51 yrs; 2,000 cases dgn at any age; 4,000 controls, (table 1), 80% of
colorectal cancers, 72% of malignant melanomas, 38% of ovarian cancers and 28%
of prostate cancers.�
For many of above markers we have been able to prove distinct clinical
phenotype � different age at diagnosis, characteristic histopathology,
predisposition to cancers of various but specific sites (1-6)
For example � CHEK2 mutations are predisposing to cancers of the breast,
thyroid, colon, kidney, prostate; NOD2 (3020insC) to cancers of the breast,
colon, lung, ovary; CHEK2 � I157T is associated with lobular sub-type and NOD2
3020insC with DCIS sub-type of breast cancers.
Our experience suggests that studies on identification of common cancer
susceptibility genes should be performed initially mainly on populations
showing high level of genetic homogeneity. This was the case for the Polish
population. In heterogenous populations achievement of the same data may require
much larger size of registries .
- Clinical benefits
Obviously, large prospective studies are needed in order to conclusively
prove clinical benefits accruing from diagnosing moderate / low cancer risk
However, even at present, we can, and in the case of our Center we
already do, offer to our patients options resulting from identification of
Classical examples of options different from standard surveillance
programme which we offer in our center include � colonoscopies at age 60 yrs
for carriers of CHEK2, NOD2 or p16 changes, MRI of the breast beginning from
the age 40 yrs for carriers of CHEK2 I157T, mammographies from the age of 35
yrs for carriers of NOD2 3020insC, BRCA2 5972C/T or of p16 A148T. In such way
we detected already hundreds of premalignant lesions or early cancers.
In summary � common cancer susceptibility genes exist and by diagnosing
them we can save lives. Studies on moderate/low cancer risk genetic markers can
be performed effectively provided we use initially the well organized models
from homogenous populations.
Jan Lubi�ski, Ph.D
Cybulski C, Gorski B, Huzarski T, Masojc B, Mierzejewski
M, Debniak T, Teodorczyk U, Byrski T, Gronwald J, Matyjasik J, Zlowocka E,
Lenner M, Grabowska E, Nej K, Castaneda J, Medrek K, Szymanska A, Szymanska J,
Kurzawski G, Suchy J, Oszurek O, Witek A, Narod SA, Lubinski J.CHEK2 is a multiorgan
cancer susceptibility gene.
Am J Hum Genet. 2004
Dec;75(6):1131-5. Epub 2004 Oct 18.
Scott RJ, Huzarski T, Byrski T, Rozmiarek A, Debniak B, Gorski B, Cybulski C,
Medrek K, Mierzejewski M, Masojc B, Matyjasik J, Zlowocka E, Teodorczyk U,
Lener M, Klujszo-Grabowska E, Nej-Wolosiak K, Jaworowska E, Oszutowska D,
Szymanska A, Szymanska J, Castaneda J, van de Wetering T, Suchy J, Kurzawski G,
Oszurek O, Narod S, Lubinski J. CDKN2A common variant and
multi-organ cancer risk-a population-based study. Int J Cancer. 2006 Jan 4.
Lubinski J, Huzarski T, Kurzawski
G, Suchy J, Masojc B, Mierzejewski M, Lener M, Domagala W, Chosia M, Teodorczyk
U, Medrek K, Debniak T, Z�owocka E, Gronwald J, Byrski T, Grabowska E, Nej K,
Szymanska A, Szymanska J, Matyjasik J, Cybulski C, Jakubowska A, Gorski B,
Narod SA, The 3020insC
Allele of NOD2 Predisposes to Cancers of Multiple Organs, Her Can in
Clin Pract 2005; 3(2), 59-63.
Huzarski T, Cybulski C,
Domagala W, Gronwald J, Byrski T, Szwiec M, Woyke S, Narod SA, Lubinski J.
Pathology of breast cancer in women with constitutional CHEK2 mutations. Breast
Cancer Res Treat. 2005 Mar; 90 (2): 187-9.
Huzarski T, Lener M,
Domagala W, Gronwald J, Byrski T, Kurzawski G, Suchy J, Chosia M, Woyton J,
Ucinski M, Narod SA, Lubinski J.: The 3020insC allele of NOD2 predisposes to
early-onset breast cancer. Breast Cancer Res Treat. 2005 Jan; 89 (1): 91-3.
Breast cancer risk markers; consecutive cancers; n=2012.
|Gene||Mutation||Percentage in %||RR|
|NOD2||3020insC||9,00||2,0 (<_50 yrs="yrs" td="td"/>|
|X5||-||10,00||2,0 (>60 yrs)|
Competing interests: No competing interests