ARTICLES

Articles

Recurrence risks in offspring of adults with major heart defects:

results from first cohort of British collaborative study
J Burn, P Brennan, J Little, S Holloway, R Coffey, J Somerville, N R Dennis, L Allan, R Arnold, J E Deanfield,
M Godman, A Houston, B Keeton, C Oakley, O Scott, E Silove, J Wilkinson, M Pembrey, A S Hunter

Summary
Background Congenital heart defects are generally
assumed to have a multifactorial aetiology. We have
tested this hypothesis by studying adults with heart
defects and their families.
Methods We identified 1094 patients who survived surgery
for major cardiac defects before 1970. We chose
individuals with disturbance of situs or segmental
connection, with atrioventricular septal defect or with
tetralogy of Fallot. After exclusion and non-participation,
727 individuals were traced. Each was visited by an
investigator and completed a detailed questionnaire. If
possible, all normal offspring were examined by a
paediatric cardiologist.
Findings The 727 individuals had 393 live offspring. There
were 71 miscarriages and five terminated pregnancies.
Overall, we found recurrent heart defects in 16 liveborn
offspringa recurrence risk of 41%. This result differed
significantly from sibling risk (21%; p=0021). More
congenital heart defects occurred in the offspring of
affected women than in those of affected men (p=0047);
when all malformations (cardiac and non-cardiac) in the

Department of Human Genetics, University of Newcastle upon Tyne

(Prof J Burn FRCP, P Brennan MRCP); Department of Medicine and
Therapeutics, University of Aberdeen (Prof J Little PhD); Clinical
Genetics Unit, Western General Hospital, Edinburgh
(S Holloway PhD); Mothercare Unit of Clinical Genetics, Institute of
Child Health, London (R Coffey BSocSci, Prof M Pembrey FRCP);
GUCH Unit, Royal Brompton Hospital, London (J Somerville FRCP);
Wessex Clinical Genetics Service, Princess Anne Hospital,
Southampton (N R Dennis FRCP); Guys Hospital, London
(L Allan FRCP); Cardiology Department Royal Liverpool Childrens
Hospital, Liverpool (R Arnold FRCP); Cardiology Department,
Great Ormond Street Hospital for Sick Children, London
(Prof J E Deanfield FRCP); Cardiology Department, Royal Hospital for
Sick Children, Edinburgh (M Godman FRCP); Department of
Paediatric Cardiology, Royal Hospital for Sick Children, Glasgow
(A Houston FRCP); Department of Paediatric Cardiology,
Southampton General Hospital, Southampton (B Keeton FRCP);
Department of Clinical Cardiology, Hammersmith Hospital, London
(Prof C Oakley FRCP); Killingbeck Hospital, Leeds (O Scott FRCP);
Department of Paediatric Cardiology, The Birmingham Childrens
Hospital, Birmingham (E Silove FRCP); Department of Cardiology,
Royal Childrens Hospital, Victoria, Australia (J Wilkinson FRCP); and
Department of Paediatric Cardiology, Freeman Hospital, Newcastle
upon Tyne, UK (A S Hunter FRCP)
Correspondence to: Prof John Burn, Department of Human
Genetics,University of Newcastle upon Tyne, Newcastle upon Tyne
NE2 4AH, UK

THE LANCET Vol 351 January 31, 1998

offspring were taken into account the excess was more

significant (p=0032). We found an excess of miscarriages
in the offspring of affected women (p=0001). In tetralogy
of Fallot, heart defects occured in seven (31%) of 223
offspring, 12 (22%) of 539 siblings, five (03%) of 1575
second-degree relatives, and eight (03%) of 2728 thirddegree relatives.
Interpretation Our findings do not support a polygenic basis
for all heart defects. Atrioventricular septal defect seems
to be a single-gene defect and tetralogy of Fallot a
polygenic disorder with a small number of interacting
genes. Our data suggest that isolated transposition of the
great arteries is a sporadic defect.

Lancet 1998; 351: 31116

Introduction
An increasing number of patients with major congenital
cardiac defects survive to adulthood and parenthood.
Data on the frequency of recurrent congenital
malformationboth cardiac and non-cardiacin the
offspring of such individuals can therefore provide
valuable clues about the aetiology of congenital heart
diseases, as well as empirical data for counselling
purposes. Many investigators have produced recurrence
data in an attempt to clarify this aetiology.112 There are
several sources of bias in the published research, relating
in particular to methods and recruitment, which
undermine the validity of the data. In many instances the
data are conflicting and of limited application.
In this study, we attempted to provide unbiased
recurrence data on major cardiac malformations by
drawing together the resources of the UK paediatric
cardiology services. The study is prospective in that the
cohort was identified from paediatric surgical records and
traced through the National Health Service; personal
contact ensured maximum recruitment, and clinical
examination of offspring made possible the accurate
assessment of clinical status.

Patients and methods

The study began in 1985. Inclusion and exclusion criteria are
shown in the panel. A major defect was defined as an
abnormality of situs (eg, right isomerism sequence),
atrioventricular
connection
(eg,
tricuspid
atresia),
ventriculoarterial connection (eg, transposition of the great
arteries),
anomalous
pulmonary
venous
drainage,
atrioventricular septal defect, or tetralogy of Fallot. Patients who
met the entry criteria, and who underwent palliative or corrective
paediatric cardiac surgery in the UK before 1970, were identified
from surgical records in participating centres. Most congenital

311

ARTICLES

Inclusion criteria
Born before 1970
Resident in UK
Major heart defect (abnormal situs or atrioventricular or
ventriculoarterial connection, atrioventricular septal defect
or tetralogy of Fallot)
Palliative or corrective surgery, including atrial septectomy
Exclusion criteria
Death during childhood
Significant learning disability
cardiac surgery in the UK was carried out at the institutions
named in this study. Data from Glasgow became available late in
the study and are included in this cohort; however, records from
Belfast and Birmingham were not available for inclusion.
We identified hospital records with approval from local ethics
committees. Individuals no longer registered with a particular
family practitioner were located via their local family practitioner
committee by means of the National Health Service Central
Register. Consent was obtained from family practioners for us to
write to all individuals asking their permission for the research
team to visit them at home for completion of a detailed
questionnaire covering diagnosis, reproductive history, and
medical status, and for the drawing of an accurate family tree.
Families were visited by a paediatric cardiology nurse (one based
in Newcastle upon Tyne, one in Southampton), an experienced
family visitor from the Institute of Child Health, London, or one
of the investigators (JB). All offspring identified as normal
were referred, with their parents permission, to the nearest
paediatric cardiologist for clinical assessment. This assessment
consisted of clinical examination, chest radiography,
electrocardiogram, and, when appropriate, echocardiography. All
dysmorphic children were karyotyped as was usual practice at
the time of recruitment. Chromosomal microdeletions may have
been overlooked, since fluorescent in-situ hybridisation
techniques were not in routine use.
Cardiac defects were grouped into the following categories for
analysis: anomalous pulmonary venous connection, abnormal
connection, atrioventricular septal defect, and Fallots tetralogy.
The abnormal connection category consisted of several
malformations,
including
abnormalities
of
situs,
of
atrioventricular concordance, and of the architecture of the
atrioventricular and ventriculoarterial valves. There was a large
enough number of patients with transposition of the great
arteries for the creation of a separate category for this defect.
Hearts with more than one malformation were categorised
Study number

Sex

according to the perceived major or underlying defectfor

example, a heart with disturbed situs and tetralogy of Fallot was
classified as abnormal connection. We analysed data for
significance using contingency tests.
For analysis of recurrence-risk ratios, we used the computermodelling method described by Farrall and Holder13 in their
investigation of cleft lip and palate, based on the work of Risch14
(further details available on request from authors). A maximumlikelihood method is a comparison of the fit of various genetic
models to the recurrence data; this comparison generates a series
of goodness of fit statistics, presented here as graphs of log10
likelihood ratio against number of loci. A perfect fit will therefore
give a log10 likelihood ratio of zero. We used recurrence-data sets
for siblings, offspring, all second-degree relatives, and all thirddegree relatives for each model. Data were selected to include
true isolated heart malformations and to exclude multiple heart
malformations. Individuals who had known associations with
environmental causation, or known or suspected genetic
syndromes (such as chromosomal abnormality, skeletal
abnormality, or mental retardation) were excluded.
A combination of recessive and dominant genes within a
population may give rise to the same phenotype (a phenomenon
known as genetic heterogeneity) and may result in the kind of
recurrence data presented here. In the polygenic model, which is
overall more likely, more than one deleterious gene (locus) is
needed in an individual to produce an abnormality. Such loci
may interact by addition, or, when their combined effect is
greater than the sum of their individual effects, by multiplication.
The computer-modelling method we used can distinguish
between a single-locus model and one containing loci that
interact by multiplication. However, the method cannot
distinguish between a single-locus model and loci that have an
effect by addition or by genetic heterogeneity. If the single-locus
model fits best, therefore, the other two possibilities must be
considered.

Results
Cohort characteristics
1094 individuals with surgically modified major structural
heart defects were identified from paediatric records in
participating centres. This cohort was reduced to 943
patients by exclusion of 108 who died in childhood, 31
who had emigrated, and 12 who were mentally retarded.
62 (66%) of the 943 cases could not be traced by
available methods. Of the 881 traceable probands, 29

Proband

Offspring

Diagnosis

Sex

Diagnosis
Extreme left-axis deviation
VSD
Critical mitral stenosis
PDA
Pathological systolic murmur
Small VSD
ASD with leaky mitral valve
ASD
TGA
Aortopulmonary window
Fallots tetralogy
VSD, mild RVOT obstruction
TAPVD, right-sided aortic arch, cat-eye syndrome
Coarctation of aorta, Turners syndrome
Perimembranous VSD, small PDA
Perimembranous VSD
ASD, Holt-Oram syndrome
Duplex kidney, double ureter
Imperforate anus, no vaginal opening, vesicoureteric reflux
Aborted fetus: spina bifida, marker chromosome

Heart malformations
90030
90078
90097
90106
90121
90158
90557
90610
90688
90722

M
F
F
F
F
F
F
F
F
F

AVSD
Fallots tetralogy
Fallots tetralogy
AVSD
Fallots tetralogy
Fallots tetralogy
AVSD
PAPVD
AVSD
DOLV

90743
90758
90849
90904
90979
90625

M
F
F
M
M
F

Fallots tetralogy
Fallots tetralogy, cat-eye syndrome
Fallots tetralogy
Fallots tetratology
Situs inversus, VSD, and PS
AVSD, Holt-Oram syndrome

M
F
F
M
M
F
F
F
M
M
F
M
M
F
F
M
M

Other malformations
90035
90327
90758

F
F
F

PAPVD, ASD
AVSD
Fallots tetralogy

F
F
F

PAPVD=Partial anomalous pulmonary venous drainage. ASD=Atrial septal defect. VSD=Ventricular septal defect. AVSD=Atrioventricular septal defect. TGA=Transposition of the
great arteries. DOLV=Double outlet left ventricle. RVOT=Right ventricular outflow tract.

Table 1: Details of probands with their affected offspring

312

THE LANCET Vol 351 January 31, 1998

ARTICLES

Defect in proband

Number of
patients

Fallots tetralogy
TGA
Abnormal connection
AVSD
APVC
All heart defects
All malformations

395
104
103
88
37
727
..

Number of miscarriages*
To fathers

To mothers

14 (101%)
1 (67%)
3 (120%)
0
0
18 (90%)
..

31 (190%)
2 (250%)
8 (320%)
9 (191%)
3 (150%)
53 (202%)
..

Total rate

149%
130%
220%
150%
100%
153%
..

Affected births/live births among offspring

Of fathers

Of mothers

2/124 (16%)
0/14
1/22 (45%)
1/13 (77%)
0/10
4/183 (22%)
4/183 (22%)

6/132 (45%)
0/6
2/17 (59%)
3/38 (79%)
1/17 (59%)
12/210 (57%)
14/210 (66%)

Total risk

31%
..
51%
78%
37%
41%
46%

TGA=Transposition of the great arteries; AVSD=Atrioventricular septal defect; APVC=Anomalous pulmonary venous connection. *Excludes five terminated pregnancies. One
female in this category had two affected offspring (case 90722). p=0077. p=0032. p=0001. If case 90625 (Holt-Oram) is included in offspring data, p=0047 and p=0022;
if terminated fetus case 90758 is included, p=0015 .

Table 2: Recurrence data by defect category

(33%) declined to participate; in 125 (142%) cases,

either the family practitioner or the proband did not
respond to repeated invitations. One male proband died
after being traced; data obtained from his wife are
included in this study. Overall, 727 (77%) probands of
943 potentially eligible individuals met all inclusion
criteria, were traced, and participated in the study.

Pregnancy outcome
The probands had a total of 469 pregnancies, which
includes four terminations of pregnancy for social
reasons, one therapeutic termination, and 71 miscarriages
(including two late-gestation stillbirths), with an overall
miscarriage rate of 153%. There were 393 liveborn
infants. The mean age of probands at delivery of all
liveborn offspring, affected or unaffected, was 255 (range
1443) years. More than half the cohort members were
aged 2030 years at interview, and the number of
offspring will have increased since the study.
All affected mothers remained healthy during
pregnancy. No proband who had an affected child was on
medication during pregnancy, and only one mother of an
affected child reported increased cyanosis during
pregnancy. There was no relation between maternal ill
health (assessed by standard American Heart Association
criteria) and miscarriage on regression analysis (data not
shown).
One stillborn infant was excluded from the main
analysis owing to a recognised genetic syndrome; a
mother with Holt-Oram syndrome with no left thumb, a
small right thumb, a shortened left forearm, and an
atrioventricular septal defect spontaneously delivered a
male fetus at 6 months gestation; the fetus had more
severe limb anomalies and a secundum atrial septal
defect. The other stillborn infant (included) was born at
25 weeks gestation to a mother with double-outlet left
ventricle and Fallots tetralogy. This mother also had a
miscarriage and then a liveborn male infant with an
aortopulmonary window, who died from neonatal sepsis.
In addition, the analysis includes a fetus aborted at 20
weeks gestation because a marker chromosome and
spina bifida were present.
Details on cardiac morphology were available in only
two of the 71 cases of miscarriage; both were reported as
normal. In 18 of the miscarriages, the father was the
affected proband (a rate of 9%), whereas in 53 it was the
mother (20%); this difference was significant (p=0001).

Sibling
Offspring*

Male probands

Female probands

Total

16/786 (20%)
4/183 (22%)

16/774 (21%)
12/210 (57%)

32/1560 (21%)
16/393 (41%)

*Livebirth data, excluding case 90625. p=0021.

Table 3: Risks to siblings and offspring

THE LANCET Vol 351 January 31, 1998

Offspring characteristics
Normality was confirmed on examination by a
paediatrician or paediatric cardiologist in 318 (81%) of all
393 offspring. In the remaining 75 (19%), further
examination was not possible, though all were examined
at birth by a paediatrician and, in most cases, by a family
practitioner afterwards. These 75 children were assumed
to have normal hearts on the basis of parental report.
Overall, 16 live offspring from a total of 393 liveborn
infants were born with cardiac malformations (table 1),
representing a total recurrence risk of 41%. If the
stillborn infant with Holt-Oram syndrome is included
with other recognised stillbirths and miscarriages, the
total recurrence risk becomes at least 36% of all
pregnancies.
Five cases of particular interest
Study number 90030a male infant with extreme left
axis deviation on electrocardiography, whose father had
atrioventricular septal defect. This infant has been
included because isolated left axis deviation is regarded as
a forme fruste of atrioventricular septal defect.15
Study number 90121a male infant with a pathological
systolic murmur whose mother had Fallots tetralogy has
been included, since the infants disorder had not been
characterised at the time of the study. The paediatrician
identified the disorder as a ventricular septal defect,
which subsequently closed spontaneously.
Study number 90158a female infant with small
ventricular septal defect whose mother had Fallots
tetralogy; the defect closed spontaneously, but is included
here as a recurrence.
Study number 90758a woman with Fallots tetralogy, a
right-sided aortic arch, and bilateral iris colobomata who
had a male infant with total anomalous pulmonary
venous drainage and a right-sided aortic arch. Both child
and proband had a marker chromosome derived from
chromosome 22 and cat-eye syndrome. The proband
later requested termination of a fetus with spina bifida
(this case is detailed in the footnote to table 2).
Study number 90849a woman with Fallots tetralogy
whose child had coarctation of the aorta and Turners
syndrome is included as representative of the background
population risk of having a child with a heart defect.
Sibling recurrences
The overall incidence of congenital heart malformation in
the siblings of probands was 21%; we found no
significant difference between male and female probands.
There was a significantly greater incidence of heart
malformations in offspring than in siblings (p=0021;
table 3).
313

ARTICLES

Tetralogy of Fallot
0
Log10 likelihood ratio

Log10 likelihood ratio

0
05
10
15
20

Atrioventricular
septal defect

05
10
15
20
25
30
35
40

25
1 2 3 4 5 6 7 8 910
Number of loci

10 100 1000
Number of loci

Log10 likelihood ratios for different numbers of loci

Computer modelling: recurrence-risk ratio analysis

After exclusions, data from 290 of 393 offspring of
probands, 861 siblings of probands, 2364 second-degree
relatives, and 4230 third-degree relatives were used for
computer modelling. Cases 90625 (Holt-Oram
syndrome) and 90758 (cat-eye syndrome) were excluded
from calculations because they represent known
Mendelian disorders.

Tetralogy of Fallot
There were seven (31%) cases of recurrent congenital
heart disease among 223 offspring and 12 (22%) among
539 siblings. Five (032%) of 1575 second-degree relative
and eight (029%) of 2728 third-degree relatives were
affected. The best-fitting model is that in which three loci
act by multiplication (figure). Such a model fits 25 times
better than a single-locus model in which the loci act by
addition, or a genetic-heterogeneity model, and eight15
times better than models with 20500 loci.

Atrioventricular septal defect

There were four (10%) cases of recurrent heart disease
among 40 offspring, and one among 112 siblings. No
second-degree relatives were affected, and three of 497
third-degree relatives were affected. Our analysis shows
that the model that acts by a single locus, by addition, or
by genetic heterogeneity fits the data 21 times better than
a model with five loci, and 165 times better than one
with 100 loci (figure).

Transposition of the great arteries

We found no cases of recurrent congenital heart disease
in the offspring of probands with transposition of the
great arteries, and only one case among 103 siblings. No
malformations were found in 767 second-degree relatives,
and computer analysis was therefore not possible.

All congenital heart disease

We carried out recurrence-risk ratio analysis for all
probands and for male and female probands separately,
including and excluding offspring data. In general, the fit
of data to any of the models tested was poor. For each
data set, the best-fitting model was one with 1000 loci,
though the likelihood ratio did not discriminate effectively
between any number of loci between ten and 1000
(results not shown).
314

Discussion
Nora16 introduced the multifactorial model for the
aetiology of isolated congenital heart defects (excluding
malformations associated with chromosomal defects,
syndromes, and known teratogenic exposure as well as
Mendelian traits) in which several genetic loci interact
together, with or without environmental factors.
Persistent patency of the arterial duct in human beings
supports this model: well-recognised environmental
insults are known to cause persistent patency of the
arterial duct (prenatal rubella infection and preterm
delivery), and Zetterqvist3 has shown that the disorder
conforms to the rules of the polygenic threshold model.
Persistent patency of the arterial duct is the exception
rather than the rule, and many studies have produced
data that do not seem to fit the polygenic or the
multifactorial model.17 For example, Whittemore and
colleagues18 found 60 children with congenital cardiac
malformations among 372 offspring born to affected
mothersan overall risk of 16%. Emanuel and
colleagues6 found the recurrence of major structural
defects in offspring of individuals with atrioventricular
septal defect to be 10%. Rose and colleagues7 examined
probands with defined malformations and found
offspring recurrence risks of 10% for coarctation of the
aorta, 11% for aortic stenosis, and 11% for atrial septal
defect. In 1994, Whittemore and colleagues12 calculated
recurrence risks for several types of defect, notably 156%
for ventricular septal defect, 150% for aortic stenosis,
and 141% for coarctation of the aorta. None of these
recurrence data support the polygenic or multifactorial
model, but the studies quoted could be criticised for
selection bias and other weaknesses in methods.
Whittemore and colleagues study18 dealt only with
affected women who agreed to be followed up during and
after pregnancy; no estimate of the participation rate was
given. In the 1994 study by the same investigators,12 data
from affected mothers were collected prospectively after
they were asked to contact one of the investigators in
early pregnancy; only 55% responded, and 2% were
subsequently lost to follow-up. Data from affected fathers
were collected retrospectively from clinic records.
Although 421 men were identified, only 267 (634%) had
fathered children, and only 191 (715%) of the children
were available for study. In addition, there were
significant differences in the distribution of types of
malformation between participants and non-participants
for some lesionsnotably maternal conotruncal
malformation, and paternal patent ductus arteriosus and
mitral-valve anomalies.
Emanuel and colleagues6 derived participants for their
study from two different sources: 60 probands from a
total of 123 (a participation rate of just under 50%),
together with an additional 30 probands from an
unspecified cohort. Rose and colleagues study7 also had
a low participation rate (43%) owing partly to limited
patient contact, and partly to the limiting of assessment
to the US state in which the studys base hospital was
located. These factors may have caused an excess of
families with affected children who were more likely to
remain close to the specialist centre.
Two other studies show the potential effects of varying
methods. Zellers and colleagues9 analysed 395 patients
with Fallots tetralogy. They identified three affected
female children from 253 offspring, two of whom had

THE LANCET Vol 351 January 31, 1998

ARTICLES

affected fathers. The childrens recurrence risk of 12% is

well below that cited by other investigators and, although
participant numbers were small, this result contrasts with
results of a growing number of studies that show a greater
recurrence risk in the offspring of affected mothers for
many defects. Zellers and colleagues acknowledge
important weaknesses in their methods: they relied on
postal questionnairesonly 58% of which were
returnedand were unable to examine any children, so
some defects might have been overlooked. Driscoll and
colleagues1-10 also used a postal questionnaire to analyse
1682 (74%) of 2262 probands with one of aortic stenosis,
pulmonary stenosis, or ventricular septal defect.
Ambiguous responses were followed with another
detailed questionnaire and, in some cases, a telephone
call. Several possible recurrences remained, which were
not characterised further, and which complicated
calculations. Since no offspring were examined, and
identification of affected offspring depended on parents,
less severe defects might have been overlooked.
By collecting information by personal interview and
arranging clinical examination of most offspring, we
acquired more reliable data. Information on more distant
relatives was less certain, though in most cases we could
verify heart defects through National Health Service
records. The weakness in the use of an analysis of all
relatives as a basis for a three-gene model for Fallots
tetralogy is that all heart malformations were included in
the modelling. 22q11 deletion is unlikely to have
accounted for the familial heart defects associated with
tetralogy of Fallot.19,20 A recent review of 558 cases of
22q11 deletion has shown that subtle but consistent
dysmorphic features were apparent in proven cases. All
probands in our study were photographed, and the
pictures were reviewed to exclude probable deletion
cases.
The suggestion that atrioventricular septal defect is
monogenic accords with published research. This specific
malformation is common in trisomy 21 and 8p deletion,
an autosomal dominant form has been mapped to
chromosome 1,22 and a further large pedigree with many
affected individuals has been excluded by linkage analysis
from all the above sites.23,24 Genetic heterogeneity has thus
been established.
The excess of malformations in the offspring of affected
women is consistent with published research. A metaanalysis of several reports212if various flaws in methods
are discountedgives a ratio for affected offspring born
to affected mothers versus those born to affected fathers
of 2.23. This ratio is almost identical to the ratio we
obtained in our study. Maternal illness does not provide
an explanation, and the mitochondrial effect suggested by
Nora and Nora8 is unlikely.
The recognition that imprinted genes influence
development suggests that some of the key genes in
cardiac development might be imprinted, and that
normal heart formation might depend on the maternal
copy of the gene. If so, deleterious alleles of maternal
origin would have a greater effect than those of paternal
origin. This phenomenon may be more general; for many
years, neural-tube defects have shown a non-specific
tendency towards matrilineal transmission.25,26 In the
statistics of our study, the significance of the maternal
excess depended on inclusion of a case of Holt-Oram
syndrome. This case was excluded because the syndrome
is a recognised single-gene defect. A 1996 family study of

THE LANCET Vol 351 January 31, 1998

Holt-Oram syndrome, however, reported greater severity

in cases of maternal origin.27 Malformation in the mother
may have a non-specific effect on a predisposition to
malformation in her offspring; one possible mechanism
for such a general phenomenon would be the
transgenerational effect of intrauterine disadvantage
described by Barker.28 Whatever the explanation, those
caring for mothers with heart malformation should be
aware of a greater risk of malformation in the offspring,
and
should
target
counselling
and
resources
appropriately.
Contributors
J Burn was responsible for study design, data collection, and overall
supervision. P Brennan carried out data analysis and presentation. J Little
was responsible for statistical analysis. Computer modelling was done by
S Holloway. All other investigators carried out proband selection and
clinical assessment.

Acknowledgments
We thank the British Heart Foundation, the Medical Research Council,
the Borwick Trust, and Glaxo-Wellcome for financial support. We are
especially grateful to our family visitors Paddy Walsh, Jenny Cole, Becky
Coffrey, and Jean Le Gassicke for their role in data collection, and to
Linda Burn for secretarial support. We also thank Anthea Stevenson for
statistical and database assistance, and Martin Farrall (Head of
Applications, Statistical Genetics Group, Wellcome Trust Centre for
Human Genetics, Oxford, UK) for use of his computer-modelling
program.

References
1

2
3
4
5

10

11
12

13
14
15
16
17

18

Pierpont MEM, Gobel JW, Moller JH, Edwards JE. Cardiac

malformations in relatives of children with truncus arteriosus or
interruption of the aortic arch. Am J Cardiol 1988; 61: 42327.
Williamson EM. A family study of atrial septal defect. J Med Genet
1969; 6: 25561.
Zetterqvist P. A clinical and genetic study of congenital heart defects.
Sweden: University of Uppsala, 1972.
Dennis NR, Warren J. Risks to the offspring of patients with some
common congenital heart defects. J Med Genet 1981; 18: 816.
Cziezel A, Poroni A, Peterffy E, Tarcal E. Studies of children of
parents operated on for congenital cardiovascular malformations.
Br Heart J 1982; 47: 29093.
Emanuel R, Somerville J, Inns A, Withers R. Evidence of congenital
heart disease in the offspring of parents with atrioventricular defects.
Br Heart J 1983; 49: 14447.
Rose V, Gold RJM, Lindsay G, Allen M. A possible increase in the
incidence of congenital heart defects among the offspring of affected
parents. JACC 1985; 6: 37682.
Nora JJ, Nora AH. Maternal transmission of congenital heart diseases:
new recurrence risk figures and the questions of cytoplasmic
inheritance and vulnerability to teratogens. Am J Cardiol 1987; 59:
45963.
Zellers TM, Driscoll DJ, Michels VV. Prevalence of significant
congenital heart defects in children of parents with Fallots tetralogy.
Am J Cardiol 1990; 65: 52326.
Driscoll DJ, Michels W, Gersony WM, et al. Occurrence risk for
congenital heart defects in relatives of patients with aortic stenosis,
pulmonary stenosis, or ventricular septal defect. Circulation 1993; 87:
1-1141-120.
Connolly HM, Warnes CA. Ebsteins anomaly: outcome of
pregnancy. JACC 1994; 23: 119498.
Whittemore R, Wells JA, Castellsague X. A second-generation study
of 427 probands with congenital heart defects and their 837 children.
JACC 1994; 23: 145967.
Farrall M, Holder S. Familial recurrence-pattern analysis of cleft lip
with or without cleft palate. Am J Hum Genet 1992; 50: 27077.
Risch N. Linkage strategies for genetically complex traits, I: multilocus
models. Am J Hum Genet 1990; 46: 22228.
Sumner RG, Phillips JH, Jacoby WJJ, Trucker DH. Forme fruste of
endocardial cushion defect. Am J Med Sci 1967; 254: 26683.
Nora JJ. Multifactorial inheritance hypothesis for the etiology of
congenital heart disease. Circulation 1968; 38: 60417.
Burn J. The aetiology of congenital disease. In: Anderson RH,
Macartney FI, Shinebourne EA, Tynan M, eds. Paediatric cardiology.
Edinburgh: Churchill Livingstone, 1987: 1563.
Whittemore R, Hobbins EJ, Engle MA. Pregnancy and its outcome in

315

ARTICLES

19

20

21

22

women with and without surgical treatment of congenital heart

disease. Am J Cardiol 1982; 50: 36170.
Wilson DI, Goodship JA, Burn J, Cross IE, Scambler PJ. Deletions
within chromosome 22q11 in familial congenital heart disease. Lancet
1992; 340: 57375.
Trainer AH, Morrison N, Dunlop A, Wilson N, Tolmie J.
Chromosome 22q11 microdeletions in tetralogy of Fallot. Arch Dis
Child 1996; 74: 6263.
Ryan AK, Goodship JA, Wilson DI, et al. Spectrum of clinical features
associated with interstitial chromosome 22q11 deletions: a European
collaborative study. J Med Genet 1997; 34: 798804.
Sheffield VC, Pierpont ME, Nishimura D, et al. Identification of a
complex congenital heart defect susceptibility locus by using DNA
pooling and shared segment analysis. Hum Mol Genet 1997; 6:
11721.

23 Wilson L, Curtis A, Korenburg JR, et al. A large dominant pedigree of

atrioventricular septal defect (AVSD): exclusion from the Downs
syndrome critical region on chromosome 21. Am J Hum Genet 1993;
53: 126268.
24 Wilson L. The aetiology of atrioventricular septal defect. University of
Newcastle upon Tyne, UK, 1996. PhD thesis.
25 Nance WE. Anencephaly and spina bifida: an etiologic hypothesis
(original article series). Birth Defects 1971; 7: 97102.
26 Burn J, Gibbens D. Hypothesis: may spina bifida result from an Xlinked defect in a selective abortion mechanism? J Med Genet 1979;
16: 21014.
27 Newbury-Ecob RA, Leanage R, Raeburn JA, Young ID. Holt-Oram
syndrome: a clinical genetic study. J Med Genet 1996; 33: 30007.
28 Barker DJP. Fetal and infant origins of adult disease. London: BMJ
Publishing, 1992.

Survival and tumour characteristics of breast-cancer patients with

germline mutations of BRCA1
L C Verhoog, C T M Brekelmans, C Seynaeve, L M C van den Bosch, G Dahmen, A N van Geel,
M M A Tilanus-Linthorst, C C M Bartels, A Wagner, A van den Ouweland, P Devilee, E J Meijers-Heijboer, J G M Klijn

Summary
Background Hereditary breast cancer has been associated
with mutations in the BRCA1 and BRCA2 genes and has a
natural history different from sporadic breast cancer. We
investigated disease-free and overall survival for patients
with a proven BRCA1 alteration.
Methods We estimated disease-free and overall survival for
49 Dutch patients from 19 consecutive families with a
proven specific BRCA1 mutation and one family with
strong evidence for linkage to the BRCA1 gene. We
compared clinical outcome and data on tumour size,
histology, axillary nodal status, contralateral breast
cancer, and oestrogen-receptor and progesterone-receptor
status with those of 196 patients with sporadic breast
cancer, matched for age and year of diagnosis.
Findings Disease-free survival for BRCA1 and sporadic
patients at 5 years was 49% (95% CI 3364) and 51%
(4359), respectively (p=098). Overall survival at 5 years
was 63% (4776) and 69% (6276), respectively (p=088).
Recurrence and death rates did not differ significantly
between groups. Hazard ratios for recurrence and death
among BRCA1 patients were 100 (065155) and 104
(063171) relative to sporadic patients (p=088), and
these did not differ significantly after adjustment for
prognostic factors. Patients with BRCA1-associated breast
cancer had twice as many progesterone-receptor-negative
tumours (p<0005) and development of contralateral
Family Cancer Clinic (L C Verhoog MD, C T M Brekelmans PhD,
C Seynaeve MD, L M C van den Bosch MSC, A N van Geel MD,
M M A Tilanus-Linthorst MD, C C M Bartels MD, J G M Klijn MD),
Department of Patient Registration (G Dahmen), Rotterdam Cancer
Institute (Dr Daniel den Hoed Kliniek) and Academic Hospital
Rotterdam, Netherlands; Department of Clinical Genetics, Erasmus
University, Rotterdam (A Wagner MD, A van den Ouweland PhD,
E J Meijers-Heijboer MD); and Department of Human Genetics,
University of Leiden, Leiden (P Devilee PhD)
Correspondence to: Dr J G M Klijn, Family Cancer Clinic,
Department of Medical Oncology, Dr Daniel den Hoed Kliniek,
Groene Hilledijk 301, 3075 EA, Rotterdam, Netherlands

316

breast cancer was four to five times as frequent as in the

sporadic group (p<0001).
Interpretation We showed that disease-free and overall
survival were similar for sporadic and hereditary breast
cancer in the presence of different tumour characteristics,
which has implications for screening prophylactic therapy,
and different treatments of hereditary breast cancer.

Lancet 1998; 351: 31621

See Commentary page

Introduction
In western countries, up to 20% of women diagnosed
with breast cancer have at least one affected relative. 5%
of all breast cancers are estimated to be attributable to
highly penetrant dominant genes, and about 4050% of
so-called site-specific hereditary breast cancers and most
cases of hereditary breast-ovarian cancer syndrome are
thought to be due to mutations in the BRCA1 gene.1
Since the mapping2 at chromosome 17q21 and
subsequent cloning3 of the BRCA1 gene, several
characteristics of BRCA1-associated breast cancer have
been reported. These characteristics include younger age
of onset,2,4 frequent bilateral occurrence,1,2,5 and worse
histoprognostic features, such as more aneuploidy, higher
grade, and higher proliferation indices.613 Although these
characteristics are generally associated with a poor
outlook for unselected breast-cancer patients,14,15 Porter
and colleagues16 found a longer survival for patients from
probable BRCA1-linked families with breast cancer than
in patients with breast cancer with similar age and date of
diagnosis identified from the breast cancer registry.
However, stage of disease was not provided. Lynch and
colleagues7 and Marcus and colleagues8 found a general
trend toward lower specific relapse and death rates, but
found similar rates after adjustment for age and stage.
For familial breast cancer, however, studies that have
investigated survival are not conclusive and results are not
consistent.8,1720 Differences in the definition of familial
breast cancer and the populations investigated could
result in a diverse contribution of genetic factors, which

THE LANCET Vol 351 January 31, 1998