evaluarea riscului de purtator brca
TRANSCRIPT
8/7/2019 evaluarea riscului de purtator BRCA
http://slidepdf.com/reader/full/evaluarea-riscului-de-purtator-brca 1/16
Breast Disease 27 (2006,2007) 5–20 5IOS Press
Assessing Breast Cancer Risk and BRCA1/2
Carrier Probability
Julie Culver∗, Katrina Lowstuter and Lauren BowlingDepartment of Clinical Cancer Genetics, City of Hope Comprehensive Cancer Center, Duarte, CA, USA
Abstract. By identifying individuals with an increased risk of breast cancer, health professionals can offer prevention strategies
tailored to individual risk levels. Such strategies may include early initiationof cancer screening, more frequent screening, targeted
therapeutic or behavioral interventions, or prophylactic surgery. In order to achieve clinical benefits with this approach, however,
risk assessment strategies and effective prevention measures must be available. In this article we review current knowledge about
cancer risk assessment for unaffected women and probability models for identifying individuals who are carriers of a mutation
in BRCA1 or BRCA2, the two genes most commonly implicated in hereditary breast cancer. We review BRCA1 and BRCA2
mutations in various ethnic populations and how this information factors into risk assessment. Additionally, we summarize the
current guidelines for when to make a referral to genetic services for risk assessment and evaluation.
Keywords: Breast carcinoma, BRCA1, BRCA2, cancer genetics, cancer risk assessment, probability models, hereditary breast
cancer
INTRODUCTION
Cancer risk assessment is practiced in clinics spe-
cializing in genetics as well as other health care set-
tings. We will describe many of the tools used by ge-
netic counselors and others to evaluate breast cancer
risk and determine the likelihood of hereditary breast
cancer caused by BRCA1 or BRCA2.
PEDIGREE-BASED RISK ASESSMENT
Breast cancer has long been known to “run in fam-
ilies.” Epidemiological studies have established fami-
ly history as a major risk factor for breast and ovarian
cancer. Relative risks associated with an affected first-
degree relative range from 2 to 4 (Table 1). Relative risk
increases with increasingnumbers of affected relatives,
∗Corresponding author: Julie Culver, MS, CGC, Department of Clinical Cancer Genetics, City of Hope Comprehensive Cancer Cen-ter, 1500 E. Duarte Rd. Mod 173, Duarte, CA 91010, USA. Tel.: +1626 256 8662; Fax: +1 626 930 5495; E-mail: [email protected].
greater biological closeness of affected relative(s), and
earlier age of onset of cancer.
Epidemiological studies provide evidence for two
general categories of risk based on familyhistory alone:
(1) moderate risk , typically associated with a family
history of breast cancer in a close relative (2) high risk ,
typically associated with a family history pattern of
breast cancer in two or more relatives, indicating the
inheritance of a highly penetrant breast cancer gene
mutation. Example pedigrees are shown in Fig. 1.
Features of moderate risk families include later ages
of onset of breast cancer (50 years and above), few rel-
atives affected, lack of ovarian cancer, and no evidence
of autosomal dominant transmission. It is likely that
a large proportion of this type of familial clustering of
breast cancer is due to the presence of genetic traits
that only modestly contribute to cancer risk. Similar
environmental exposures as well as gene-environment
interactions are also likely to account for some of the
risk associated with moderate family history.
High risk families are characterized by early age
of onset of breast cancer (less than 50 years), bilater-
al breast cancer, multiple affected individuals, ovarian
cancer at any age, male breast cancer at any age, and a
0888-6008/06,07/$17.00 2006,2007 – IOS Press and the authors. All rights reserved
8/7/2019 evaluarea riscului de purtator BRCA
http://slidepdf.com/reader/full/evaluarea-riscului-de-purtator-brca 2/16
6 J. Culver et al. / Assessing Breast Cancer Risk and BRCA1/2 Carrier Probability
Table 1Empiric Risk of Cancer Based on Family History
Cancer Family History Relative Risk
Breast First degree relative with breast cancerAll ages 2.1 (95% CI 2.0–2.2)Affected< 50 years 2.3 (95% CI 2.2–2.5)Affected50 years 1.8 (95% CI 1.6–2.0) [50]
Breast Sister with breast cancer at:20–29 years 4.68 (95% CI 0.92–11.36)30–39 years 3.28(95% CI 1.91–4.65)40–49 years 2.56 (95% CI 1.89–3.24)50–59 years 2.68 (95% CI 1.98–3.38)60–69 years 1.71 (95% CI 0.98–2.44) [64,66]
Breast Second degree relative with breast cancer 1.5 (95% CI 1.4–1.6) [50]
Ovarian Breast cancer in mother or sister before age 40 SIR 1.7 (95% CI 1.3–2.1) [46]
Ovarian Parent or sibling with breast cancer OR 1.6 (95% CI 1.3–2.0) [51]
(cancer mortality)
Ovarian First degree relative with ovarian cancer 3.1 (95% CI 2.6–3.7Daughter with ovarian cancer 1.1 (95% CI 0.8–1.6)Sister with ovarian cancer 3.8 (95% CI 2.0–5.1)Mother with ovarian cancer 6.0 (95% CI 3.0–11.9)More than one affected relative (first or second degree) 11.7 (95% CI 5.3–25.9) [59]
Ovarian Second degree relative with ovarian cancer 2.5 (95% CI 1.5–4.3) [59]
pattern of autosomal dominant transmission often ap-
pearing to “skip” males in the family. Families with
these characteristics are more likely to have a muta-
tion in a highly penetrant autosomal dominant cancer
susceptibility gene such as BRCA1 or BRCA2, com-
pared to moderate risk families. Ductal carcinoma insitu (DCIS) may be part of the phenotype of hereditary
breast cancer [32,35]. A population based study found
the rate of BRCA mutation detection is similar in DCIS
and invasive breast cancer cases [16].Of note, when breast cancer is seen in families with
additionaltypes of cancers, other hereditarysyndromes
(reviewed by Nusbaum et al. in this issue) may be the
cause of breast cancer in the family and should be con-
sidered. For example, breast, endometrial, and thyroid
cancers occurring in an autosomal dominant pattern
may be the result of an inherited PTEN mutation.
When evaluating small families, it is particularly im-portant to consider family structure and the number of
women surviving through later ages. A single case of a
womanwith early onset breast cancer whohas a limited
family history due to lack of female relatives or early
ageat death of femalerelatives mayindeed have a high-
er probability of carrying a BRCA1 or BRCA2 mutation
than a similarly affected woman from a large families
with many unaffected women surviving through later
ages. Among 204 single breast cancer cases before
age 55 who underwent genetic testing of BRCA1 and
BRCA2, family structure was a strong predictor of mu-
tation status (P = 0.009), with BRCA mutations iden-
tified in 17.3% of women with limited vs. 5.7% with
adequate family structure [71]. Some of the BRCA1/2
probabilitymodels, such as BRCAPRO, do take into ac-
count family structure, (see section below on this topic)
but others do not.
The rate of de novo (non-inherited) BRCA mutationsis thought to be negligible, however there are rare re-
ports of such mutations in the literature [61,69].
EMPIRIC MODELS OF BREAST CANCER
RISK ASSESSMENT
Family history data can be utilized in various models
for predicting individual breast cancer risk. The four
risk models to be reviewed in this section include the
Claus, Gail, Tyrer-Cuzick, and BRCAPRO models. A
comparison of risk estimates produced by these modelsis shown in Table 2; because these four risk models in-
corporate different risk factors, theysometimes provide
strikingly different risk estimates and may be utilized
to provide a range of risks in the clinical setting.
Claus (or CASH ) Model
The Claus model estimates the probability of an un-
affected woman developing breast cancer based on her
family history of breast cancer [17]. This model was
derived from empiric observations in the Cancer and
Steroid Hormone Study (CASH) [73]. Genetic mod-
8/7/2019 evaluarea riscului de purtator BRCA
http://slidepdf.com/reader/full/evaluarea-riscului-de-purtator-brca 3/16
J. Culver et al. / Assessing Breast Cancer Risk and BRCA1/2 Carrier Probability 7
Table 2Lifetime breast cancer risks for hypothetical patients, based on four risk models
Family History Claus1 [17] Tyrer- BRCAPRO3 Gail4 Notes on model limitationsCuzick 2 [67] [13,47] [29]
Case 1 40 year old woman– mother BC 35– maternal aunt BC 41
34% 24% 18% 19% Gail does not incorporatematernal aunt (SDR)
Case 2 40 year old woman– paternal aunt BC 28– paternal grandmother BC 39
23% 21% 18% 11% Gail does not incorporatepaternal (SDR) relatives
Case 3 40 year old woman– mother OC 55– maternal aunt BC 45– maternal grandmother BC 49
19% 25% 23% 11% Gail does not incorporateOC or SDRs; Claus doesnot incorporate OC
Case 4 40 year old woman of Ashkenazi Jewish
ancestry
– mother OC 55
– maternal aunt BC 45– maternal grandmother BC 49
19% 31% 30% 11% Same as Case 3; addition-ally, Gail and Claus do notincorporate AJ ancestry
BC= Breast cancer, OC = ovarian cancer, SDR = second-degree relative, AJ = Ashkenazi Jewish.1Claus model calculates breast cancer risk to age 79 years.2Tyrer-Cuzick model calculates lifetime breast cancer risk to an unspecified age. Other personal characteristics included in the model for eachcase were: age at menarche = 12, age at first birth = 28, height = 1.37 meters (5 feet, 4 inches), weight = 61 kg (134 lbs), woman has neverused HRT, no atypical hyperplasia or LCIS.3BRCAPRO calculates breast cancer risk to age 85 years.4Gail model calculates breast cancer risk to age 90. Other personal characteristics included in the Gail risk model for each case were: age atmenarche= 12, age at first birth = 28, breast biopsies = 0, race =White.
els were developed to fit the age-specific incidence of
breast cancer among first- and second- degree relatives
of 4730 Caucasian breast cancer cases and 4688 Cau-
casian controls, aged 20–54 years. Although the Clausmodel is based on an assumption that risk associat-
ed with a family history can be exclusively attributed
to rare autosomal dominant mutations with high pene-
trance, which is almost certainly incorrect, the results
of this model agree with observations concerning the
association of family history andbreast cancerrisk. For
any given unaffected female patient, the model incor-
porates up to two relatives affected with breast cancer
(first- or second- degree) and the decade of onset of
breast cancer for each relative. The model provides
risk estimates for each decade of the patient’s life up to
age 79. However, the Claus model does not incorpo-
rate family size, ethnic background, or other risk fac-
tors. Therefore, it may not be not suitable for women
with more than two affected relatives, as it may un-
derestimate risk. Furthermore, the Claus model does
not include incorporate relatives with ovarian cancer.
A separate Claus paper allows for the calculation of
breast cancer risk for women with a first-degree family
history of ovarian cancer [18].
Claus risk estimates are easily calculated using the
published tables in the original paper [17]. However,
one adjustment must be made using a formula on page
645 of the paper, which accounts for the patient’s cur-
rent age, to appropriately reduce her risk due to having
passed some of her years of breast cancer risk. For ex-
ample, in Table 2, Case 1, the 40 year old woman with a
34% lifetime risk of breast cancer should have her risk re-calculated if she ages and does not develop breast
cancer; at age 50 the Claus model would predict a 28%
lifetime risk. Another approach to making the adjust-
ment for current age is to use the software available to
calculate Claus risks on a Palm pilot [1], which makes
calculating these risks very simple. Also, the same ad-
justment to the Claus model is made by CancerGene
(BRCAPRO) program, discussed below (downloadable
from link in Table 3).
Tyrer-Cuzick Model
The Tyrer-Cuzick model [67] incorporates the prob-
ability of a BRCA1 or BRCA2 mutation, the likelihood
of a low penetrance gene mutation, and personal risk
factors. For an individual unaffected woman, family
history is used in conjunction with Bayes’ theorem to
produce the likelihood of a BRCA mutation. The asso-
ciated breastcancer risk is then calculated andmodified
to reflect the relative risk associated with the woman’s
personal risk factors. Personal risk factors included
are: current age, age at menarche, parity, age at first
livebirth, age at menopause, history of atypical hyper-
plasia or lobular carcinoma in situ, height, and body
8/7/2019 evaluarea riscului de purtator BRCA
http://slidepdf.com/reader/full/evaluarea-riscului-de-purtator-brca 4/16
8/7/2019 evaluarea riscului de purtator BRCA
http://slidepdf.com/reader/full/evaluarea-riscului-de-purtator-brca 5/16
J. Culver et al. / Assessing Breast Cancer Risk and BRCA1/2 Carrier Probability 9
Table 4BRCA1/BRCA2 carrier probability estimates for hypothetical patients, based on five models
Case Family History Modified Myriad BRCA- Tyrer- Manchester Notes on models
Couch1 PRO Cusick 2
1 40 year old women with BC3
– 3 older unaffected sisters4N/A 7% 2% N/A BRCA1 score= 3
BRCA2 score= 3Does not meet 10%threshold
– Couch is not designed to calcu-late probability for a single affectedindividual– BRCAPRO considers the unaf-fected sisters
2 50 year old women with BC at40 and OC at 50– Mother deceased at 40
accident– Mat aunt BC 35
88% 55% 88% N/A BRCA1 score= 15BRCA2 score= 12 if BRCA1 negativeMeets 10% threshold forboth genes
– BRCAPRO considers the moth-er’s early death– Risk estimates are high becauseof BC andOC in a single individual
3 51 year old women with BC3
– Mother BC 52– Mat aunt BC at 56
4% 3% 8% N/A BRCA1 score= 6BRCA2 score= 6Does not meet 10%
threshold
– BRCAPRO mayoverestimaterisk as unaffected relatives (if any)werenot entered
4 −35 year oldunaffected woman5
– Sister BC 40– Pat aunt OC 55
21% 12% 10% 4% – BRCA1 score= 15BRCA2 score= 12 if BRCA1 -negativeMeets 10% threshold forboth genes
– Risks are approximately doublefor the patient’s sister
5 −35 year old unaffectedwoman5 of AshkenaziJewish ancestry– Sister BC 40– Pat aunt OC 55
44% 27% 36% 25% Model does not apply toAJ population
– In comparison to Case 4, eachmodel shows a higher probabilitywith AJ ancestry.– Manchester does not apply to AJpopulation
BC= Breast cancer, OC = ovarian cancer, AJ = Ashkenazi Jewish.1Probabilities are derived from modifying Couch to include BRCA1 and BRCA2 probability of a mutation (see description of modification in thetext); Couch model is not used for Case 1 because the model does not apply to single cases.2Tyrer-Cuzick model cannot be used for Cases 1–3 because it is only applicable to unaffected patients. See Table 2, footnote 2 for personal
characteristics of Cases 4 and 5 included in Tyrer-Cuzick model.3Age of women equals age of diagnosis with cancer.4Ages of sisters entered into to BRCAPRO was 50, 55, and 60 years of age.5Risk is for the unaffected patient. For the Couch model, this represents 50% of the family risk calculated, modified to include BRCA2 asdiscussed in the text.
mass index (BMI). This is a statistical model basedon combining relative risks, and not an actual sampleof women. The model has been incorporated into acomputer program, which produces very user-friendlyoutputs of both the likelihood of the patient developingbreast cancer as well as the probability of the patientcarrying a BRCA mutation (Table 3).
BRCAPRO Model
The BRCAPRO model [13,47] provides risk esti-mates for breast and ovarian cancer based on the likeli-hood that a person carries a BRCA1 or 2 mutation. Us-ing a patient’s current age, cancer history, and familyhistory of breastand ovarian cancerin first- andsecond-degree relatives, the program uses Bayesian analysisto calculate the probability of a BRCA mutation, andfrom that probability, the risk of breast and ovarian can-cer. See detailed discussion of the BRCAPRO modelin the section on BRCA probability and related models,below.
Gail Model
The Gail model [29] estimates the probability of an
unaffected woman developing breast cancer over spec-
ified time intervals based on her age and personal risk
factors. It was developedusing data from a nested case-
control subset of the 284,780 women participating in
the Breast Cancer Detectionand Demonstration Project
(BCDDP) [11]. These were predominately Caucasian
women 35 to 79 years of age, receiving annual mam-
mography screening. The model includes risk factors
that were important predictors of risk in the BCDDP
and was derived from an unconditional logistic regres-
sion analysis. Risk factors (and their associated codes)
include: age [<50; 50], age at menarche [14; 12–
13; <12], age at first live birth [<20; 20–24; 25–29;
30; or nulliparous;], number of previous breast biop-
sies [0; 1; 2] and whether any biopsy found atypical
hyperplasia (yes, no), and number of first-degree rela-
tives (mother or sisters) with breast cancer [0; 1; 2].
8/7/2019 evaluarea riscului de purtator BRCA
http://slidepdf.com/reader/full/evaluarea-riscului-de-purtator-brca 6/16
10 J. Culver et al. / Assessing Breast Cancer Risk and BRCA1/2 Carrier Probability
(a)
Br, 42
d. 43
Br, 50; Br, 68
Ov, 53
d. 56
Br, 40 Br, 35
(b)
Br, 52
Br, 68
Fig.1. Example pedigrees illustrating ahigh and moderate risk breastcancer families.
The Gail model determines an odds ratio for a given
womanto developbreast cancer andcombines this with
baseline age-specific hazard rates and competing mor-
tality risks, resulting in an absolute risk of breast can-
cer over specified time intervals. Later, the National
Surgical Adjuvant Breast and Bowel Project’s Breast
Cancer Prevention Trial modified the Gail model to
incorporate race and include daughters as first-degree
relatives [25].
When using the Gail model for breast cancer risk
assessment, it is importantto consider the limitations of
this model. TheGail model is inadequate for evaluating
family history because it does not incorporate second-
degree relatives (including aunts, grandmothers, or any
paternal relatives) or the age of onset of breast cancer
in any relative. For example, Cases 2, 3, and 4 in
Table 2, theGail model does notcalculate increasedrisk
of breast cancer attributable to family history because
second-degreeand paternal relatives are not included in
the risk calculation. Therefore, we do not recommendits use for evaluating patients with a significant familyhistory of breast cancer. However, the Gail model is aneffective clinical tool in determining whether a patientmeets a minimum risk thresholdto be offered tamoxifenfor chemoprevention. This risk threshold was the entrycriteria of the BCDDP trial and was equal or greaterthan the risk of a an average 60-year old woman, whichis equivalent to a 5 year predictedrisk of breastcancer of at least 1.66% [25] For breast cancer risk estimation inclinical practice, the Gail model is most appropriate forwomen with affected first-degree relatives or womenwith a history of biopsies. Both Palm Pilot [3] and web
versions [2] are available, and the Gail model is alsoavailable in CancerGene (BRCAPRO), with web link shown in Table 3.
Validation studies have been performed on the origi-nal Gail model, which demonstrated that in some casesit failed to accurately predict cancer risk. Two stud-ies found the Gail model overpredicted the absoluterisk of breast cancer in women less than age 60 whodid not undergo annual mammography screening [14,57]. Additionally, the model tended to overpredict risk for women less than age 60 and underpredict risk forwomen over age 60 [14].
It is important to note that risk estimates for the same
woman using either the Claus and Gail models may notbe identical, in part based upon the differentparametersof the models. Indeed, when looking at large numbersof women with a family history, Gail estimates tend tobe higher than Claus estimates [39,40].
Although no risk assessment model is appropriatefor every patient, clinicians often choose one modelover another for different types of patients. Patientswith a significant family history of breast cancer insecond degree relativesshould notbe evaluated with theGail model (Table 2). However, patients with a biopsyhistory, especially a biopsy with atypical hyperplasia,may best be evaluated with the Gail model. The Gail
andClausmodels shouldbe used with caution if ovariancancer is present in the family. The Tyrer-Cuzick andBRCAPRO models can incorporate Ashkenazi Jewishancestry, while Gail and Claus cannot. Additionally,BRCAPRO can account for the size and structure of thepatient’s family and current age of family members. Inthe clinical setting, a patient can be provided with arange of risk estimates from the models that are deemedappropriate for her circumstance. Providing the rangewill also enable the patient to see that risk estimationis an imprecise science. Based on an assessment of the risk numbers provided, screening and preventionprograms can then be tailored individually.
8/7/2019 evaluarea riscului de purtator BRCA
http://slidepdf.com/reader/full/evaluarea-riscului-de-purtator-brca 7/16
J. Culver et al. / Assessing Breast Cancer Risk and BRCA1/2 Carrier Probability 11
BRCA 1/2 Probability and Related Models
There are numerous models now available to esti-
mate the probability of an individual having a mutation
in BRCA1 and BRCA2 genes. A few of these models
also predict individual cancer risks. (Table 3) Some of
these models were first developed around 1997, soon
after BRCA clinical genetic testing became available.
However, the field of mutation and cancer risk proba-
bility modeling continuesto evolve as evidenced by the
recent revision of the Couch model and publications of
new models such as BOADICEA.
Couch Model
The Couch model, published by Couch et al. in 1997,
is a widely used logistic regression model that predicts
the probability of a BRCA1 mutation in a given fami-
ly [19]. The purpose of Couch et al.’s original publica-
tion was to define the incidence of BRCA1 mutations in
women with breast cancer who were referred for breast
cancer risk assessment. Couch et al. gathered personal
cancer history, family history of cancer, and blood from
263 women with breast cancer seen for cancer risk as-
sessment between 1993 and 1995. The Couch model is
presented in a table format within the publication [19].
Themodelutilizes personal and familyhistory of breastand ovarian cancer in first and second-degree relatives
to estimate the mutation probability. The model also
considers Ashkenazi Jewish ancestry. An average age
of onset of breast cancer in the family is used to gen-
erate the mutation probability, (age of onset of ovarian
cancer is not included in the average age calculation).
Therisk provided is the family’s probability of a BRCA1
mutation and applies to all affected (diagnosed with
breast or ovarian cancer) family members. The proba-
bility of a mutation in unaffected first-degree relatives
of breast/ovarian cancer patients is half of the family’s
probability for carrying a mutation. For example, if a
Couch mutation probability is 10% for a family, then
the daughter of an affected individual in the family has
a 5% chance of having a BRCA1 mutation, as she has
a 50% chance of inheriting the mutated allele from her
mother. The model can be applied by using the table in
the original paper [19] and in the CancerGene program
(Table 3).
The limitations of the Couch model should be con-
sidered when being applied to a clinical setting. The
model does not account for other types of cancer as-
sociated with the BRCA genes aside from breast and
ovarian cancer and does not consider male breast can-
cer or bilateral female breast cancer. The model is
further limited as it predicts for BRCA1 mutation sta-tus only and the study population consisted mainly of
Caucasian women. In clinical practice, the Couch may
be modified to include BRCA2 mutation probability by
multiplying the estimated BRCA1 mutation probability
by a factor of 1.33. The modification by a factor of 1.33
accounts for the contribution of BRCA2 to the over-
all load of hereditary breast cancer due to BRCA1 and
BRCA2 mutations and is based on the published data
from the combined analysis for the original cohort [56].
Thesame research team from theUniversity of Penn-
sylvania Abramson Cancer Center that developed the
original Couch model has recently revised and updat-
ed the model, entitled “Penn II,” to predict for both
BRCA1 and BRCA2 mutations and to consider other
personal and family cancer history (Table 3). The mod-
el takes into account three generations of breast and
ovarian cancer (e.g. including cousins) as well as oth-
er BRCA-associated cancers including pancreatic and
male specific cancers (prostate and male breast cancer).
The development and validation paper for the Penn II
model is currently under review [58].
Myriad
The mutation prevalence tables published by Myri-ad Genetic Laboratories provide easily accessible risk
estimates for detecting a BRCA mutation. These tables
are based on methods published by Frank et al. [27].
The risk estimates presented were originally derived
from BRCA clinical test results over a three year period
from 10,000 individuals with a personal and/or family
history of breast and/or ovarian cancer. Of these in-
dividuals, 7,461 had full gene sequencing of BRCA1
and BRCA2 and 2,539 were screened only for the three
Ashkenazi Jewish founder mutations. Approximately
90% of individuals tested were women and ∼45% had
a personal history of breast cancer only. The recently
updated mutation prevalence tables, released on the in-
ternet by Myriad in March of 2006, (Table 3) are based
on clinical test results from ∼49,000 individuals who
had full gene sequencing and∼15,000 individuals who
were screened for the three Ashkenazi Jewish founder
mutations.
The benefits and limitations of the Myriad mutation
prevalence tables should be considered when provid-
ing mutation risk estimates in the clinical setting. The
greatest advantage of using these tables is that the risk
estimates are based on a large clinical sample and cat-
egorized by Ashkenazi Jewish versus non-Ashkenazi
8/7/2019 evaluarea riscului de purtator BRCA
http://slidepdf.com/reader/full/evaluarea-riscului-de-purtator-brca 8/16
12 J. Culver et al. / Assessing Breast Cancer Risk and BRCA1/2 Carrier Probability
Jewish ancestry. In addition, these tables are easy to
use and updated frequently. The tables are available fordownload on palm pilot (Table 3) and the risk numbers
are produced by the CancerGene program (Table 3).
However, the risk estimates presented in these tables
do not take into account: the specific age of onset of
breast cancer, the number of affected relatives, bilater-
al breast cancer, unaffected relatives, or other BRCA-
associated cancers. There is also no distinction be-
tween first and second degree relatives or maternal ver-
sus paternal affected relatives. The Ashkenazi Jewish
tables also include some women who had testing for
the three founder mutations because of a known muta-
tion in the family, and the numbers in the table could
be an overestimate. Furthermore, these risk estimates
are entirely dependent upon the personal and family
history information provided on test requisition forms
completed by the ordering clinician, which is subject
to errors and omissions by health care providers. In
summary, these tables are widely used in the clinical
setting to provide risk estimates prior to BRCA testing,
but the tables may under- or over-estimate the risk of
detecting a BRCA mutation in some families andshould
be interpreted with caution.
BRCAPRO
BRCAPRO is a mathematical model that predicts the
probability of a BRCA mutation [13,47]. The founda-
tion of this model uses Mendelian genetics and Bayes’
theorem to evaluate a family history of cancer for mu-
tation probability. Specifically this model predicts
the probability having a mutation in either gene, both
genes, or neither gene [12]. The model also estimates
breast and ovarian cancer risk, as described above. The
model incorporates all family members (up to second-
degree relatives), their history of breast and ovarian
cancer, bilateral breast cancer, male breast cancer, and
whether the family has Ashkenazi ancestry. Mutation
probabilities can be calculated for both affected and
unaffected individuals; however, cancer risk estimates
only apply to unaffected individuals. The model also
takes into account mutation status in the family (i.e., if
a family member has tested negative for BRCA muta-
tion).
The model can be downloaded for free as part of
CancerGene (Table 3) and is also available as part of
the Progeny pedigree software package [5]. Cancer-
Gene Version 4.3 accounts for whether a woman had an
oophorectomy. The output is easy to interpret; howev-
er, in order to obtain the most accurate mutation prob-
ability, the pedigree must be entered, which can be
time-consuming. Another limitation of this model isthe penetrance data used to derive the model were tak-
en mainly from Caucasian families therefore its use in
non-Caucasian families may be limited.
Validation studies were conducted in 2002 by com-
paring the estimated probability of carrying a BRCA
mutation as computed by BRCAPRO to actual genetic
test results. These studies found that BRCAPRO gives
an accurate measurement of the probability of a muta-
tion and therefore is a useful instrument in the counsel-
ing process [12].
Tyrer-Cuzick Model
The Tyrer-Cuzick breast cancer risk assessment
model (discussed in detail above) [67] also calculates
BRCA mutation probabilities. This model incorporates
first and second- degree gamily members with breast
and ovarian cancer and their ages of onset. However, a
disadvantage is that the model calculates the mutation
probability only for an unaffected individual, which
is usually not the ideal candidate for initiating testing
within a family. Software is available to calculate risks
(Table 3) and a user-friendly printout is produced.
Manchester Model
The Manchester Model is a scoring system that will
determine whether a family has 10% probability of
a mutation in either BRCA1 or BRCA2 [22] or a 20%
combined risk [23]. This model was developed in
Manchester, England using a population of 422 non-
Ashkenazi British individuals. Using this population
the authors designed a scoring system to determine
whether a family may have a deleterious BRCA1 or
BRCA2 mutation. The cancers included in the scoring
system are female breast cancer, male breast cancer,
ovarian cancer, pancreatic cancer, and prostate cancer.
Between 1 and 8 points are given for each cancer di-
agnosis, depending on type of cancer and age of onset.
Higher scores are given for earlier ages of onset, and
the decade of diagnosis is included in calculating the
score for breast cancer cases. For ovarian, male breast,
and pancreatic cancer cases, a distinction of diagnosis
before or after age 60 is made. Separate scores are cal-
culated for BRCA1 and BRCA2 and a total score of 10
for one lineage in a family is equivalent to a 10% prob-
ability of a mutation in that gene. For example, female
breast cancer diagnosed <30 years is given a score of
6 for BRCA1 and 5 for BRCA2; ovarian cancer <60
8/7/2019 evaluarea riscului de purtator BRCA
http://slidepdf.com/reader/full/evaluarea-riscului-de-purtator-brca 9/16
J. Culver et al. / Assessing Breast Cancer Risk and BRCA1/2 Carrier Probability 13
is given a score of 8 for BRCA1 and 5 for BRCA2 (if
BRCA1 has already been tested). Therefore, a familywith one case of breast cancer before age 30 and one
case of ovarian cancer before age 60 is given a total
BRCA1 score of 14anda BRCA2 score of 10 (if BRCA1
testing is negative). Thus testing is justified for this
family, with BRCA1 testing to be done prior to BRCA2
testing. Using this model, the gene with the higher
score (over 10) may be tested first; if no mutation is
found in that gene, scores may be subsequently adjust-
ed for the other gene. This model does not suggest that
mutation analysis is justified in an isolated breast or
ovarian cancer case at any age.
An important advantage of the Manchester model is
that the scoring system can be easily used in the clini-
cal setting and does not require the use of software or
the input of an entire pedigree into a computer. Ad-
ditionally, validation studies by the authors compared
the Manchester model against BRCAPRO, Couch, and
Myriad and found the Manchester model to outperform
the other models in discriminating families with a 10%
likelihood of a mutation [22].
An important limitation of this model is that it does
not calculate the exact probability of a mutation, but
rather distinguishes whether a family meets the 10%
or 20% probability cutoff or not. This model may not
be as useful in a clinic that uses a different probabilitycutoff or does not useany specific numerical probability
cutoff for offering BRCA testing. Also, the model is
not designed for use in Ashkenazi Jewish individuals.
BOADICEA
The BOADICEA model was developedby Antoniou
et al. in 2002 in order to predict the probability of a
BRCA1 or BRCA2 mutation and provide breast and
ovarian cancer risks [6–8]. The model was devel-
oped using complex segregation analysis of the occur-
rence of breast and ovarian cancer in two data sets
(population based series of 1484 breast cancer cases
and 156 multiple case families). Both data sets were
ascertained from probands with breast cancer, main-
ly from the United Kingdom. The model takes in-
to account that familial breast cancer is explained by
both BRCA1 and BRCA2 mutations as well as a poly-
genic component (reflecting the joint multiplicative ef-
fect of multiple genes of small effect on breast cancer
risk). Furthermore the model accounts for the possi-
bility of genetic modifiers, which may affect the pene-
trance of BRCA1 and BRCA2 mutations. The remain-
ing clusters of cancer in families not accounted for by
Table 5Selected examples of recurrent and founder mutations in the BRCA
genes
Population BRCA1 BRCA2
Ashkenazi Jewish 185delAG5382 ins C
6174delT
Icelandic 999del15British 6-kb dup exon 13
4184 del4Dutch 2804delAA
del exon 13del exon 22
Chinese 1081delGAfrican American 943ins10
1832del55296del4
Hispanic 185delAG
del exon 9–12French Canadian 4446C>T
2953del3 + C3768insA
8765delAG2816insA6085G>T6503delTT
the BRCA1 and BRCA2 genes are assigned to poly-
genic factors. In 2005, the model was found to ac-
curately predict the carrier probability in individuals
of French Canadian ancestry [6]. The researchers are
developing a web-based software interface, which en-
ables clinicians to enter pedigree information to deter-
mine probabilityinformation. Updates on this software
can be obtained by checking the BOADICEA web site(Table 3). One can also utilize the published tables [8]
to assess risk.
In clinical practice, using multiple BRCA probability
models is time consuming and therefore it may be best
to choose a model or two that best suit the patient. We
have indicated some of the benefits and limitations of
each model in Tables 3 and 4. Additionally, Table 4
shows the probabilityestimates producedby eachavail-
able model for various pedigrees. Of note BOADICIA
was not included in this table as the software needed to
use the model is not available at this time.
Here are some considerations of when to use the
published models discussed above. These comments
are based on our clinical practice and others may have
differing viewpoints. The Couch model should not be
used when there is only a single case of breast or ovar-
ian cancer in a family since the probability table in the
original article is based on families with multiple af-
fected individuals. In families with multiple affected
relatives, the Couch model estimates should be modi-
fied to include BRCA2 probabilities, as explained in the
section on the Couch model above. The Myriad model
is useful for both single cases and families; this mod-
el is extremely quick to calculate probability estimates
8/7/2019 evaluarea riscului de purtator BRCA
http://slidepdf.com/reader/full/evaluarea-riscului-de-purtator-brca 10/16
14 J. Culver et al. / Assessing Breast Cancer Risk and BRCA1/2 Carrier Probability
Table 6Published guidelines for referral to cancer genetics services
Non-Ashkenazi Jewish Ashkenazi Jewish
Affected National Comprehensive Cancer Network (United States)*BC 50y#*Two primary HBOC cancers in an individual (bilateral breast cancer andbreast and ovarian cancer)*Two breast primaries or breast and ovarian cancers in close relatives) on thesame side of the family*Clustering of breast cancer with male breast cancer, thyroid cancer, sarco-ma, adrenocortical carcinoma, endometrial cancer, pancreatic cancer, braintumors, dermatologic manifestations or leukemia/lymphoma on the same sidethe family*Member of a family with a known mutation in breast cancer susceptibilitygene*Male breast cancer*Clustering of ovarian cancer
National Comprehensive Cancer Network (United States)*Less stringent than for non-AshkenaziJewish
Unaffected United States Preventive Services Task Force*Two FDR with BC, with one diagnosed <50y*Combination of 3 FDRs or SDRs diagnosed with BC regardless of age*Combination of both BC and OC among FDR and SDR*FDR with bilateral BC*Combination of 2 or more FDR or SDR with OC*Single FDR or SDR having both BC and OC*Male BC in any relative
United States Preventive Services Task Force*Any FDR with BC or OC*Two SDR on the same side of the familywith BC or OC
National Institute for Health And ClinicalExcellence (United Kingdom)*Seek advice fromtertiary care contact aboutlevels of risk and appropriateness of referral
National Institute for Health And Clinical Excellence (United Kingdom)*Two FDR with BC, diagnosed before average age 50y*Three FDRs or SDRs, diagnosed before average age 60y*Four relatives diagnosed at any age (including at least 1 FDR)*Combination of one OC in any relative and:
– FDR or SDR with BC < 50
– Another OC– Two FDR or SDR with BC, diagnosed before average age 60y*Bilateral BC in a FDR diagnosed before average age 50y*Bilateral BC in a FDR or SDR and one FDR or SDR with BC diagnosed<60y*Male BC in any relative and:
– FDR or SDR with BC diagnosed <50y– Two FDRs or SDRs diagnosed with BC before average age 60y
Key BC= Breast cancer, OC = ovarian cancer, FDR = first-degree relative, SDR = second-degree relative, y= years.# Includes both invasive and ductal carcinoma in situ.
and we find it very useful in the clinic. BRCAPRO
model is often useful if a family is particularly large
or small because its Bayesian analysis considers family
size. Additionally, the model is useful if genetic testing
has been performed in the family and is negative, but
one wishes to calculate the probability of a mutation in
other family members. Finally, BRCAPRO is the only
published model currently available that incorporates
bilateral breast cancer.
The Tyrer-Cuzick model is only applicable to unaf-
fected women. A nice feature of this model is that it
provides both a breast cancer risk estimate and a BRCA
mutation probabilityestimate; however,when possible,
it is almost always most informative to test an affect-
ed family member first, so a probability estimate may
be needed for that individual, which the Tyrer-Cuzick
model does not calculate. The Manchester model is
useful in a setting of limited resources when a muta-
tion probability of 10% is used as a threshold for offer-
ing genetic testing. Finally, it is important to remem-
ber that none of the risk models was developed from
non-Caucasian populations.
When sharing these probability estimates with pa-
tients, providing a range of numbers may be helpful
to illustrate that the likelihood of a mutation may vary,
depending on the factors considered in their family. In
many cases, the probability of a mutation is not criti-
cal to the decision to undergo testing, because of how
critical the results of testing can be. Results will often
influence medical management, such as whether a pa-
tient should undergo an oophorectomy or more inten-
sive breast cancer surveillance.
8/7/2019 evaluarea riscului de purtator BRCA
http://slidepdf.com/reader/full/evaluarea-riscului-de-purtator-brca 11/16
J. Culver et al. / Assessing Breast Cancer Risk and BRCA1/2 Carrier Probability 15
Giving a patient probability estimates often sets an
expectation for the likelihood the test will be positive.A patient who has a very low probability of a mutation
who is insistent upon having genetic testing may be
convinced that testing is not worthwhile if the proba-
bility of finding a mutation is very low (additionally,
her insurance company may not cover the cost of test-
ing and seeing her probability estimate may help her
understand why that is the case.)
RISK ASSESSMENT IN SPECIFIC
POPULATIONS
In the overall population the incidence of BRCA
mutations varies in the literature from approximately
1/500–1/1100 [4]. Higher allele frequency of BRCA
mutations in specific populations is due to founder ef-
fect or recurrent mutations.1 The most well-known and
prevalent founder effect in BRCA genes occurs with-
in the Ashkenazi Jewish population. Approximately
95% of all hereditary breast cancer within the Ashke-
nazi Jewish population is attributed to three founder
mutations: 185delAG and 5385insC in BRCA1 and
6407delT in BRCA2. Approximately 1 in 40 (2.5%)
individuals of Ashkenazi Jewish ancestry is a carrier
of one of these three BRCA mutations [53]. This isdramatically higher than the frequency in the general
Caucasian population.
Other populations also have founder effects or recur-
rent mutations. For example within the Icelandic pop-
ulation a single BRCA2 mutation (999del5) accounts
for a high proportion of familial breast cancer [33,65].
Some founder mutations, such as the exon 13 duplica-
tion, have been identified across geographically diverse
populations who originate from a common background.
In 1999, a BRCA1 exon 13 duplication was identified
and found in one Portuguese family and three appar-
ently unrelated families of European ancestry from the
United States; via haplotype analysis, these families
appeared to be from a common ancestor. In 2000,
the Exon 13 duplication Group set out to estimate the
1Founder effect is defined as a high frequency of a mutated allelein a population, which was founded by a small group where memberof the group was a carrier of the mutated allele [43]. Individualswho carry the same founder mutation also share common markerswithin the gene or adjacent to the gene (same haplotype). Recurrentmutations are commonly seen mutations that do not segregate withthe same markers (i.e. have different haplotypes) among differentcarriers. These mutations are likely due to areas within the geneprone to mutation or ‘hot spots.’
geographic diversity and frequency of this mutation.
This group concluded that the Exon 13 duplication islikely a founder mutation in countries that have a his-
torical link to Great Britain. Another high frequency
mutation identified within the British population is the
BRCA1 4184del4 mutation. Because different haplo-
types have been identified with the 4184del4 mutation
it is unlikely to be a true founder mutation; most likely,
it is a highly recurrent mutation found in the United
Kingdom [24]. In 1997, Peelen et al. characterized the
2804delAA BRCA1 Dutch founder mutation as origi-
nating 32 generations ago in the Dutch population [48].
Within the Dutch population, large deletions of exon
13 and 22 in BRCA1 have also been characterized as
founder mutations [49]. Although there have been few
studies of BRCA mutation prevalence in Asian popula-
tions, a number of founder mutations have been iden-
tified to date. One of these mutations, the 1081delG
mutation in BRCA1, was described as a likely founder
mutation in China by Khoo et al. in 2002 [37].
Even within a single country, individuals of differ-
ent ethnic backgrounds have founder mutations. In
the United Sates, approximately 12% of the popula-
tion is of Hispanic ancestry. In a clinic based cohort
study of primarily Mexican Hispanic individuals seen
for breast cancer risk assessment, 4 out of 110 (3.6%)
had the BRCA1 185delAGmutation [72]. Interestingly,the 185delAG mutation found in these individuals had
the same haplotype as the Ashkenazi Jewish founder
haplotype. None of the four unrelated Hispanic pa-
tients with the 185delAG mutation had any knowledge
of Jewish ancestry. This may be the result of the Jewish
population in Spain during the Spanish Inquisition be-
ing forced to convert to Christianity or they would have
been expelled from the country. Many of the conver-
sos (Jews converted to Christianity) and crypto-Jews
(conversos who secretly practiced Judaism) may have
migrated to the United States carrying with them the
185delAG BRCA1 mutation. Also in 2005, Weitzel et
al. identified an apparent founder rearrangement mu-
tation within the Mexican Hispanic population involv-
ing a deletion of exons 9–12. This was identified in
three apparently unrelated families [70]. There are also
documentedfounder mutations identified in families of
African ancestry in America. For example the BRCA1
943ins10 was described in 1999 as being associated
with a single haplotype in five families of African an-
cestry from different geographic locations (three from
the United States, one from the Ivory coast and one
from the Bahamas) [41]. This mutation has been de-
scribed as an ancient founder mutation of West African
8/7/2019 evaluarea riscului de purtator BRCA
http://slidepdf.com/reader/full/evaluarea-riscului-de-purtator-brca 12/16
16 J. Culver et al. / Assessing Breast Cancer Risk and BRCA1/2 Carrier Probability
origin of a similar age to the 185delAG founder mu-
tation in the Ashkenazi Jewish population [45]. Therehave also been two other BRCA1 recurrent mutations
described in the literature in African American fami-
lies: BRCA1 5296del4 and BRCA1 1832del5 [30]. One
challenge in interrupting BRCA genetic test results in
individuals of African ancestry is the high rate of vari-
ants/polymorphism identified within African Ameri-
cans compared to Caucasian Americans [45]. Interpre-
tation of these variants of uncertain significant is cov-
ered in more detail in the Brown et al. article in this
issue.
In other parts of North America there are other pop-
ulations with a high rate of founder mutations. For
example there are seven founder mutations within the
BRCA genes which account for a significant proportion
of the hereditary breast and ovarian cancer in individ-
uals of French ancestry in Canada [62]. Of these sev-
en mutations two occur more frequently than the oth-
ers, BRCA1 4446C>T and BRCA2 8765delAG [15].
A complete list of the seven founder mutations can be
found in Table 5. These mutations are so common that
often individuals of French Canadian ancestry initiate
testing of the BRCA genes with a panel of the founder
mutations and if negative then consider reflex to com-
plete evaluation of the genes. This testing strategy is
comparable to that used for individuals of AshkenaziJewish ancestry.
Knowledge of the numerous recurrent and founder
mutations in BRCA1 and BRCA2 genes have important
implications for risk assessment and clinical care. In
countries such as Iceland where a single BRCA2 muta-
tion accounts for a large proportion of BRCA mutations,
it may be worthwhile to initiate testing with the com-
mon BRCA2 999del5 mutation first; if negative, testing
for the other BRCA mutations may be considered. In
individuals of British and Dutch ancestry, it is critical
to include rearrangement studies to evaluate for the re-
current exon 13 duplication in the British population
and the large deletions of exon 13 and 22 in the Dutch
population; these rearrangements cannot be detected
by full gene sequence analysis alone. For individuals
presenting with breast cancer who are of Ashkenazi
Jewish ancestry it is recommended to start genetic test-
ing with evaluation for the three founder mutations. If
the family is of high risk (i.e., highly suspicious for
a BRCA mutation such as a history of multiple cases
of breast and/or ovarian cancer), complete analysis of
BRCA genes may be warranted. For such families it
is worthwhile to continue with further genetic testing
to increase the negative predictive value of the test, as
approximately 5% of hereditary breast cancer in the
Ashkenazi Jewish population is caused by non-foundermutations [36,55]. When a founder mutation has been
identified in an Ashkenazi Jewish family, testing at-
risk family members should include evaluation for the
other two-founder mutation, as there have been numer-
ous reports of families segregating with more than one
founder mutation [28,38].
Furthermore, knowledge of these mutations helps to
estimate the likelihood of a BRCA mutation. For exam-
ple, in the Ashkenazi Jewish population, the likelihood
of being a carrier of a BRCA mutation is higher due to
the relatively high frequencyof the three founder muta-
tions and thereforethe thresholdof when to offer genet-
ic testing is lowerthan in the general Caucasian popula-
tion. Many of the mutation probability models detailed
previously consider Ashkenazi Jewish ancestry for
BRCA mutation probability calculations. Knowledge
of founder and recurrent mutations leads to more ac-
curate risk estimation, cost effective mutation-targeted
testing, and guides appropriate testing for when to offer
rearrangement analysis or other testing methodologies.
LOW PENETRANCE GENES
Germline mutations in high-penetrance breast cancersusceptibility genes (BRCA1, BRCA2, p53, and PTEN )
are rare in the general population and only account
for approximately 5–10% of all breast cancers [21,26,
44]. On the other hand, variants in low-penetrance
breast cancer susceptibility genes, which are relatively
common in the general population, are expected to ac-
count for the majority of breast cancers [34,44,52,60].
Variants in these genes may confer a modest increase
in breast cancer risk and are thought to interact with
both exogenous (diet and pollution) as well as endoge-
nous (hormonal) risk factors [52,54]. Numerous low-
penetrance genes have been identified including en-
zymes involved in DNA repair, cell signaling process-
es, detoxification of reactive oxygen species, as well as
metabolism of estrogen, carcinogens, and alcohol [21].
These genes have almost exclusively been studied in
Caucasianpopulations and limited informationis avail-
able on whether these low-penetrance genes contribute
to increased cancer risks in other populations. Al-
though some laboratories have recently started offering
clinical testing for some of these low-penetrancegenes,
clinical management based on test results of these genes
remains controversial. In addition, many of these tests
were developed for other genetic syndromes and were
8/7/2019 evaluarea riscului de purtator BRCA
http://slidepdf.com/reader/full/evaluarea-riscului-de-purtator-brca 13/16
J. Culver et al. / Assessing Breast Cancer Risk and BRCA1/2 Carrier Probability 17
not initially intended to screen for breast cancer risk
(i.e. ATM gene testing was originally developed for di-agnostic testing for ataxia-telangiectasia or CDKN2A
(p16 ) gene testing was originally developed for heredi-
tary melanoma). In addition, some commercial labora-
tories (for example, a lab called Oncovue) have begun
offering breast cancer risk genetic testing on various
low-penetrance genes as well as variants in single nu-
cleotide polymorphisms (SNPs), however, such testing
is generally considered premature and is not typically
practiced in cancer genetics clinics. A detailed review
of low penetrance genes is presented by Nusbaum et
al. in this issue.
CRITERIA FOR REFERRAL FOR GENETIC
COUNSELING AND TESTING
It is critical that individualswho areat risk forheredi-
tary cancer syndromes be identified as they may benefit
from increased surveillance to detect cancer at earlier
more treatable stages as well as preventative interven-
tion strategies. Dueto the myriadof differentpublished
diagnostic and referral criteria, it is often challenging
for healthcare providersto determinewho should be re-
ferred for genetic cancer risk assessment. Furthermore,
there are differences in guidelines for who shouldbe re-ferred for risk assessment versus who is appropriate for
genetic testing. Most referral guidelines are less strin-
gent than diagnostic/genetic testing guidelines, which
helps to ensure that all high-risk individuals are identi-
fied.
Below is a review of some of the published clinical
practice guidelines for referral to cancer genetics ser-
vices. Table 6 also provides a summary of some of the
criteria below. Additionally, in the United States, crite-
ria for genetic testing for BRCA1 and BRCA2 have been
established by third party payers (such as Kaiser, Medi-
care, Aetna, Blue Cross, etc.). These criteria change
frequently and can be accessed directly from these in-
stitutions.
NCCN The National Comprehensive Cancer Net-
work (NCCN)is a group of 20 UnitedStates cancercen-
ters designated by the United States National Cancer
Institute as comprehensive cancer centers. The NCCN
Genetic/Familial High Risk Assessment panel, which
consists of experts within the field from the NCCN
member organizations,has published guidelinesfor ge-
netic/familial risk assessment of breast and ovariancan-
cer [20], shown in Table 6. The guidelines include con-
sensus statements from the panel of experts of accepted
referral patterns. In general, the guidelines recommend
referral for cancer genetics services for early onsetbreast cancer or diagnosis of any age of ovarian cancer
and/or a family history concerning for hereditary beast
ovarian cancer syndrome. The guidelines also take into
account non-BRCA genes, which can cause hereditary
breast cancer such as PTEN and p53, by including fam-
ily history of thyroid cancer, sarcoma, adrenocortical
carcinoma, endometrial cancer, brain tumors, dermato-
logic manifestations or leukemia/lymphoma [20].
USPSTF The United States Preventive Services Task
Force published a recommendation statement for the
referral of unaffected women for genetic counseling
and evaluation for BRCA testing [68]. The recommen-
dation, published with supporting scientific evidence,
recommended whichwomen without a personal history
of breast or ovarian cancer should be referred. Guide-
lines are shown in Table 6.
NICE The National Institute for Health and Clinical
Excellence (NICE) of the United Kingdom published
guidelines to classify women at risk of familial breast
cancer. NICE is an independent organization respon-
sible for providing national guidance on the promo-
tion of good health and the prevention and treatment
of ill health. The NICE guidelines include primary,
secondary, and tertiary management for women with
a family history of breast cancer. Recommendationsfor referral to tertiary management (genetic counsel-
ing) are shown in Table 6, and are more stringent than
the USPSTF guidelines. The guidelines also include
recommendations for cancer surveillance in high risk
individuals [42].Hampel Hampel et al. published risk assessment
criteria designed to guide health care professionals in
identifying individuals who are appropriate for referral
to cancer genetics services [31]. Guidelines for referral
were created from review of published diagnostic cri-
teria for hereditary cancer syndromes. When the pub-
lished guidelines differed from each other, the authors
used expert opinion to develop their guidelines. The
goal of the criteria was to assist health care providersin
recognition of individuals who would benefit from risk
assessment and create uniformityin referral guidelines.
ASCO The American Society of Clinical Oncology
published a consensus statement in 1996 and an updat-
ed consensus statement in 2003 guiding when cancer
susceptibly genetic testing should be used. ASCO rec-
ommends that genetic cancer risk assessment and ge-
netic testing be offered to individuals when 1) there is a
personal or family history suggestive of a genetic can-
cer susceptibility syndrome 2) the genetic test can be
8/7/2019 evaluarea riscului de purtator BRCA
http://slidepdf.com/reader/full/evaluarea-riscului-de-purtator-brca 14/16
18 J. Culver et al. / Assessing Breast Cancer Risk and BRCA1/2 Carrier Probability
adequately interpreted 3) the results of genetic testing
will influence medical management. Initial guidelinesin 1996 suggested that a person with a 10% or greater
probability of having a mutation for hereditary cancer
syndrome should be offered genetic testing [10]. The
2003 guidelines were revised to reflect that for many of
the syndromes it is difficult to accurately predict muta-
tion probability or there is variance between the muta-
tion probability models; therefore, ASCO does not rec-
ommend a numerical threshold of when genetic testingshould be offered but instead states that expert clinical
judgment is more appropriate [9].NSGC The National Society for Genetic Counselors
recommends that a referral for cancer genetic risk
assessment and counseling should be considered forclients with personal or family history features sugges-
tive of familial or hereditary cancer and should not be
limited to just those individuals who are potential can-
didates for genetic testing. Individuals from high-risk
families may benefit from a detailed discussion about
hereditability of cancer in their families, appropriatecancer risk management strategies, and the option of
genetic testing [63].
In conclusion, the tools in this article will enable
health care providers to identify individuals with in-
creased breast cancer risk, estimate their risk, as well
as determine the probability that they may carry aBRCA1 or BRCA2 mutation. By finding individuals
with an increased risk of breast cancer, health profes-
sionals can offer prevention strategies to reducerisk as-
sociated with breast and ovarian cancer. Additionally,
the published guidelines summarized above will helpprovidersmake appropriatereferrals to genetic services
for further evaluation and genetic testing.
ACKNOWLEDGEMENTS
The authors thank Wylie Burke, MD, PhD for her
assistance in compiling the material in Tables 1 and 2,as well as the discussion of pedigree-based risk assess-
ment and the Claus model.
REFERENCES
[1] BreastCa for Palm 2001 http://www.palmgear.com/index.cfm?fuseaction=software.showsoftware&prodID=29820.
[2] Gail Model Breast Cancer Risk Assessment Tool http://www.cancer.gov/bcrisktool/.
[3] NSABP Model 2, BeastCa for Palm 2001 http://www.palmgear.com/index.cfm?fuseaction=software.showsoftware&prodID=29820.
[4] Prevalence and penetrance of BRCA1 and BRCA2 mutationsin a population-based series of breast cancer cases. AnglianBreast Cancer Study Group, Br J Cancer 83 (2000), 1301–1308.
[5] Progeny Software, LLC. Southbend, Indiana; Copyright 2006.http://www.progenysoftware.com/.
[6] A.C. Antoniou, F. Durocher, P. Smith et al., BRCA1 andBRCA2 mutation predictions using the BOADICEA andBRCAPRO models and penetrance estimation in high-risk French-Canadian families, Breast Cancer Res 8 (2006), R3.
[7] A.C. Antoniou, P.D. Pharoah, G. McMullan et al., A compre-hensive model for familial breast cancer incorporating BRCA1,BRCA2 and other genes, Br J Cancer 86 (2002), 76–83.
[8] A.C. Antoniou, P.P. Pharoah, P. Smith et al., The BOADICEAmodel of genetic susceptibility to breast and ovarian cancer,Br J Cancer 91 (2004), 1580–1590.
[9] ASCO, American Society of Clinical Oncology policy state-
ment update: Genetic testing for cancer susceptibility, J ClinOncol 21 (2003), 2397–2406.
[10] ASCO, Statement of the American Society of Clinical On-cology: genetic testing for cancer susceptibility, Adopted onFebruary 20, 1996, J Clin Oncol 14 (1996), 1730–1736.
[11] L.H. Baker, Breast Cancer Detection Demonstration Project:five-year summary report, CA Cancer J Clin 32 (1982), 194–225.
[12] D.A. Berry, E.S. Iversen, Jr., D.F. Gudbjartsson et al.,BRCAPRO validation, sensitivity of genetic testing of BRCA1/
BRCA2, and prevalence of other breast cancer susceptibilitygenes, J Clin Oncol 20 (2002), 2701–2712.
[13] D.A.Berry, G. Parmigiani, J. Sanchez et al., Probability of car-rying a mutation of breast-ovarian cancer gene BRCA1 basedon family history, J Natl Cancer Inst 89 (1997), 227–238.
[14] M.L. Bondy, E.D. Lustbader, S. Halabi et al., Validation of a
breast cancer risk assessment model in women with a postivefamily history, J Natl Cancer Inst 86 (1994), 620–625.
[15] P.O. Chappuis, N. Hamel, A.J. Paradis et al., Prevalence of founder BRCA1 and BRCA2 mutations in unselected FrenchCanadian women with breast cancer, Clin Genet 59 (2001),418–423.
[16] E.B. Claus, S. Petruzella, E. Matloff et al., Prevalence of BRCA1 and BRCA2 mutations in women diagnosed with duc-tal carcinoma in situ, Jama 293 (2005), 964–931.
[17] E.B. Claus, N. Risch and W.D. Thompson, Autosomal domi-nant inheritance of early-onset breast cancer: Implications forrisk prediction, Cancer 73 (1994), 643–651.
[18] E.B. Claus, N. Risch and W.D. Thompson, The calculationof breast cancer risk for women with a first degree familyhistory of ovarian cancer, Breast Cancer Res Treat 28 (1993),115–120.
[19] F. Couch, M.L. DeShano, A. Blackwood et al., BRCA1 muta-tions in women attending clinics that evaluate the risk of breastcancer, N Engl J Med 336 (1997), 1409–1415.
[20] NCCN Practice Guidelines in Oncology: Genetic/familialhigh-risk assessment: breast and ovarian 2006 http://www.nccn.org/professionals/physician gls/PDF/genetics screening.pdf.
[21] R.G. Dumitrescu and I. Cotarla, Understanding breast cancerrisk – where do we stand in 2005?, J Cell Mol Med 9 (2005),208–221.
[22] D.G. Evans, D.M. Eccles, N. Rahman et al., A new scoringsystem for the chances of identifying a BRCA1/2 mutation out-performs existing models including BRCAPRO, J Med Genet
41 (2004), 474–480.
8/7/2019 evaluarea riscului de purtator BRCA
http://slidepdf.com/reader/full/evaluarea-riscului-de-purtator-brca 15/16
8/7/2019 evaluarea riscului de purtator BRCA
http://slidepdf.com/reader/full/evaluarea-riscului-de-purtator-brca 16/16
20 J. Culver et al. / Assessing Breast Cancer Risk and BRCA1/2 Carrier Probability
[61] A. Tesoriero, C. Andersen, M. Southey et al., De NovoBRCA1
mutation in a patient with breast cancer and in inheritedBRCA2 mutation, Am J Hum Genet 65 (1999), 567–569.
[62] P.N. Tonin, C. Perret, J.A. Lambert et al., Founder BRCA1
and BRCA2 mutations in early-onset French Canadian breastcancer cases unselected for family history, Int J Cancer 95
(2001), 189–193.[63] A. Trepanier, M. Ahrens, W. McKinnon et al., Genetic Cancer
Risk Assessment and Counseling: Recommendations of theNational Society of Genetic Counselors, J Genet Couns 13
(2004), 83–114.[64] H. Tulinius, V. Egilsson, G.H. Olafsdottir et al., Risk of
prostate, ovarian, and endometrial cancer among relatives of women with breast cancer, Bmj 305 (1992), 855–857.
[65] H. Tulinius, G.H. Olafsdottir, H. Sigvaldason et al., The effectofa single BRCA2 mutation on cancer in Iceland, J Med Genet
39 (2002), 457–462.
[66] H. Tulinius, G.H. Olafsdottir, H. Sigvaldason et al., Neoplasticdiseases in families of breast cancer patients, J Med Genet 31
(1994), 618–621.[67] J. Tyrer, S.W. Duffy and J. Cuzick, A breast cancer prediction
model incorporating familial and personal risk factors, Stat
Med 23 (2004), 1111–1130.[68] USPSTF, Genetic risk assessment and BRCA mutation testing
for breast and ovarian cancer susceptibility: recommendationstatement, Ann Intern Med 143 (2005), 355–361.
[69] R.B. van der luijt, H.A. can Zon, R.P.M. Jansen et al., Denovo recurrent germline mutation of the BRCA2 gene in apatient with early onset breast cancer, J Med Genet 38 (2001),102–105.
[70] J. Weitzel, B. Hendrickson, V. Lagos et al., Identification of a Novel Frequent Large Genomic Rearrangement of BRCA1
in High-risk Hispanic Families. In: The American Society of
Human Genetics: 2005; Salt Lake City, Utah; 2005: A:362.[71] J.N. Weitzel, C.A. Cullinane, V.I. Lagos et al., Family Struc-
ture Predicts BRCA Mutations in Single Cases of Breast Can-cer, Submitted.
[72] J.N. Weitzel, V. Lagos, K.R. Blazer et al., Prevalence of BRCA
mutations and founder effect in high-risk Hispanic families,
Cancer Epidemiol Biomarkers Prev 14 (2005), 1666–1671.[73] P.A. Wingo, H.W. Ory, P.M. Layde et al., The evaluation of the
data collection process for a multicenter, population-based,case-control design, Am J Epidemiol 128 (1988), 206–217.