Research Article

Causes of Congenital Malformations

M. Hassan Toufaily 1,2, Marie-Noel Westgate1,2,
Angela E. Lin1,2,3, and Lewis B. Holmes *1,2,3

Background: Many different causes of malformations have been established. polydactyly, type B; chromosome abnormalities; vascular disruption;
The surveillance of a consecutive population of births, including stillbirths and complications of monozygous twinning; and environmental factors. The
elective terminations of pregnancy because of fetal anomalies, can identify malformations of unknown etiology were a much larger group. Conclusion:
each infant with malformations and determine the frequency of the apparent While several causes of malformations have been identified, many remain
etiologies. This report is a sequel to the first such analysis in the first 10 years unexplained. Combining the ascertainment in a future surveillance programs
of this Active Malformations Surveillance Program (Nelson and Holmes, with genome sequencing and chromosome microarray analysis will increase
1989). Methods: The presence of malformations was determined among significantly the number of malformations attributed to genetic mechanisms.
289,365 births over 41 years (1972–2012) at the Brigham and Women’s
Hospital in Boston. The abnormalities were identified from the review of the Birth Defects Research 110:87–91, 2018.
examination findings of the pediatricians and consultants and diagnostic C 2018 Wiley Periodicals, Inc.
V
testing for the live-born infants and the autopsies of the fetuses in elective
terminations and stillbirths. Results: A total of 7020 (2.4%) infants and
Key words: Mendelian inheritance; vascular disruption; chromosome abnor-
fetuses with one or more malformations were identified with these apparent malities; twinning; environmental factors
etiologies in 26.6%: Mendelian disorders, including infants with postaxial

Introduction Pauli and Reiser (1994) found that 34% had single malfor-
Population-based malformations surveillance programs mations and recognizable malformation syndromes and
have shown that approximately 2% of newborn infants 25% had chromosome abnormalities.
have a structural abnormality or malformation (Correa This report is a follow-up to the compilation of the
et al., 2007; Feldkamp et al., 2017). The frequency causes of congenital malformations identified in the sur-
increases in the first year of life as “silent” abnormalities veillance of 69,227 infants, including stillbirths and elec-
not detected at birth, such as heart defects, anomalies of tive terminations, in the first 10 years of the Active
the urinary tract, and bowel malrotation, are identified Malformations Surveillance Program, conducted at Brig-
from diagnostic studies prompted by signs and symptoms ham and Women’s Hospital (BWH) in Boston (Nelson and
in the affected infant (Thomas et al., 2018). Holmes, 1989; Holmes, 2012). Several potential causes
The frequency of malformations is much higher in the have been established for malformations during this pro-
fetuses in spontaneous abortions and in stillbirths than ject extended to a 41-year period (1972–2012): chromo-
the 2% in live-born infants. Shepard and his associates some abnormalities, mutations with autosomal or X-linked
(1989) found that 19% of spontaneously aborted fetuses inheritance; vascular disruption; multifactorial inheri-
had a localized defect or an identifiable syndrome. In addi- tance/familial; environmental factors, and complications of
tion, there is a higher frequency of associated chromosome the twinning process. An additional etiology, vascular disrup-
abnormalities in approximately half of spontaneous abor- tion, which develops after normal development has occurred,
tions (Zhang et al., 2009). In an analysis of 789 stillbirths, has been delineated as a separate group in the intervening
years.
1
The category multifactorial inheritance was a term
Department of Pediatric Newborn Medicine, Brigham and Women’s Hospital,
used by clinical investigators, such as Fraser (1986) and
Boston
2
Medical Genetics Unit, MassGeneral Hospital for Children, Boston
Carter (1976), for malformations that showed a distinctive
3
Department of Pediatrics, Harvard Medical School, Boston, Massachusetts pattern of occurrence and recurrence, such as cleft lip and
palate (Fraser, 1970) and myelomeningocele (Carter and
Supported by the Birth Defects Registry of the Massachusetts Department of Evans, 1973). The other common malformations that
Public Health, which is part of the National Birth Defects Prevention Study, a
project of the Birth Defects and Developmental Disabilities Center of the Cen- showed these patterns of inheritance included anence-
ters for Disease Control, Atlanta. phaly, cleft palate alone, hypospadias, and several common
*Correspondence to: Lewis B. Holmes, Medical Genetics Unit, MassGeneral heart defects. Other malformations, such as bilateral renal
Hospital for Children, 175 Cambridge Street, 5th Floor, Boston, MA 02114. agenesis (Carter et al., 1979) and esophageal atresia
E-mail: holmes.lewis@mgh.harvard.edu
(McMullen et al., 1996), have shown an increased likeli-
Published online in Wiley Online Library (wileyonlinelibrary.com). Doi: 10. hood of occurrence in the siblings of index cases, and have
1002/bdr2.1105 been classified as “familial” in this report.

C 2018 Wiley Periodicals, Inc.

V
88 CAUSES OF CONGENITAL MALFORMATIONS

Studies establishing the many different causes of a

specific malformation highlight the heterogeneity in the TABLE 1. Recognized Causes in Affected Infantsa
etiologies, illustrating the fact that some have a genetic
1972–74 1972–74
component and others reflect environmental factors. A
1979–1985b 1979–2012c
similar analysis of the causes of malformations was pub- (n 5 69,277) (n 5 289,365)
lished recently by Feldkamp and her associates (2017).
They analyzed 5504 infants with birth defects in a Single mutant genes 48 (3.1%) 110 (1.6%)
d
population-based study of 270,878 births in Utah (2005– (Postaxial polydactyly, type B) 545 (7.8%)
2009). They assigned a definite cause for 20.2% of the Chromosome abnormalities 157 (10.1%) 825 (11.8%)
infants with birth defects: 19.1% were chromosomal or Multifactorial inheritance/familial 581 (37.5%) 1461 (20.8%)
genetic conditions and the remainder were due to terato-
Malformation syndromes N/A 87 (1.2%)
gens, primarily poorly controlled maternal pregestational
Vascular disruption 39 (2.5%) 105 (1.5%)
diabetes.
We present here the apparent etiologies for the 7020 Environmental factors 49 (3.2%) 238 (3.4%)
infants with malformations identified by a hospital-based Twinning 6 (0.4%) 36 (0.5%)
malformations surveillance program among 289,365 births, Unknown cause 669 (43.2%) 3,613 (51.5%)
including live-born and stillborn infants and fetuses from
Total 1549 7020
elective terminations because of anomalies.
a
Limited to infants born to mothers who had always planned to
deliver at BWH.
b
Materials and Methods Published (Nelson and Holmes, 1989).
c
The malformed infants and fetuses were identified 6 days Findings through 2000, published (Holmes, 2012).
d
a week and on holidays from reading the findings Postaxial polydactyly, type B, was considered “familial” in the report
from the first 10 years (Nelson and Holmes, 1989) and has been
recorded by the examining pediatricians and the consul-
designated as due to an autosomal dominant gene mutation in the
tants in each newborn infant’s medical record. The find- current analysis. Only infants with isolated postaxial polydactyly,
ings in stillborn infants and affected fetuses in elective type B are listed.
terminations of pregnancy were determined from the
review of autopsy findings in the Department of Pathology. Malformation syndromes, such as VACTERL association
The final diagnoses were determined from the review of and hemifacial microsomia, were compiled as separate
this information by the study clinicians (L.B.H. and A.E.L.). from the common isolated malformations.
The methodology and the demographic characteristics of The designation of vascular disruption has been used
the population surveyed by the Active Malformations Sur- for abnormalities that are produced after the initial normal
veillance Program at the Boston Lying-In Hospital, later structural development has occurred. In the process of
part of the BWH in Boston, have been summarized in vascular disruption, the hypothesis is that blood flow to a
another article in this special issue (Holmes et al., 2018). region is decreased, followed by hypoxia, endothelial cell
The reports of the findings in chromosome analysis by injury, hemorrhage, tissue loss, and repair. This group of
the hospital’s cytogenetics laboratory were part of the affected infants has been presented in detail in a separate
records obtained on each infant tested. Some infants with article (Holmes et al., 2018).
a diagnosis of a chromosome abnormality from prenatal Many different environmental factors were identified,
testing at an outside laboratory had official reports or such as anticonvulsant medication taken to prevent seiz-
occasionally only verbal reports. ures, and maternal conditions, specifically, insulin-
The common malformations with distinctive patterns dependent diabetes mellitus. However, the ability of a
of clustering in families, such as cleft palate alone, cleft lip malformations surveillance program to identify these
and palate, anencephaly, myelomeningocele, hypospadias, causes of abnormalities in the exposed fetus depends on
many common heart defects, congenital hip dysplasia, etc., the accuracy of the information recorded in the medical
were assigned to the category of multifactorial inheritance. records of the infants identified as malformed and of their
They were combined with other common malformations mothers. This is particularly relevant to the identification
which have shown an increased rate of occurrence in sib- of an infant affected by exposure to an excessive amount
lings, such as bilateral renal agenesis, esophageal atresia, of alcohol or to an intrauterine infection with cytomegalo-
and other malformation with an observed increased fre- virus. These infants would be more likely to be identified
quency among the sibs or offspring of affected individuals if the hospital has a screening program with well-designed
in comparison to the general population. For this analysis, interviews about alcohol use or routine viral cultures dur-
these malformations have been combined under the head- ing pregnancy. Because there were no such systematic pro-
ing “multifactorial inheritance/familial.” grams in this setting, the affected infants were much less
BIRTH DEFECTS RESEARCH 110:87–91 (2018) 89

likely to be identified at birth. However, several rare tera- The malformations designated multifactorial inheritance/
togenic exposures were identified. familial are the largest group (20.8%), but does not represent
Identical twins have been shown to have a higher rate of a single “cause” of malformations. We would expect the ab-
malformations than nonidentical twins or singletons (Schinzel normalities included in this group to be re-designated in
et al., 1979). The like-sex twins considered to be identical were future studies, once new categories have been developed
based on the findings in the placental membranes, not genetic from the on-going research.
markers. The abnormalities identified were from pregnancies The group of malformations designated as due to envi-
complicated by twin–twin transfusion and acardia. ronmental factors was primarily malformed infants born to
mothers with pregestational diabetes mellitus, as was the
Results case for the analysis of the causes of malformations in the
The findings in the 41-year sample are very similar to those population in Utah by Feldkamp et al. (2017) (Table 2).
in the first 10 years (Table 1). The malformations attributed With better prenatal screening and more focused physical
to Mendelian inheritance have been consistent with the examinations, we would have expected to have identified
findings in Mendelian Inheritance in Man (Amberger et al., many more malformed infants with the fetal alcohol syn-
2015). One major difference is that we have tabulated sepa- drome. The listing of 29 infants with malformations whose
rately postaxial polydactyly, type B, in this analysis and high- mother had taken one or more anticonvulsant drugs is lim-
light that group of infants, whereas they were included in ited to medications that have been shown to be teratogenic.
the “familial” subset in the first 10 years. Because this mild
type of polydactyly occurs in 1% of Black infants and only 1 Discussion
in 3000 White infants, its frequency will reflect the more There were several limitations to this survey of 289,365
racially diverse nature of the population surveyed in Boston, births. First, by limiting the information about diagnosis to
in contrast to the analysis by Feldkamp et al. (2017) in the the first 5 days of life, findings in follow-up studies, such as
less diverse population in Utah. In the Active Malformations echocardiogram, a consultation with a urologist, or mutation
Surveillance Program at BWH, we used a four-level severity analysis, were not incorporated. Second, the significant pro-
score: lethal (anencephaly); severe, handicapping (Down gress made in diagnostic studies between 1972 and 2012,
syndrome); moderate, fixable (cleft lip and palate); and mild specifically mutation analysis and chromosome microarray,
(postaxial polydactyly, type B). Other malformations surveil- was used in only a few affected infants. More use of these
lance programs may exclude this mild form of polydactyly. diagnostic methods would have improved the quality and
specificity of the diagnostic findings. Whole genome
sequencing could have identified even more causative muta-
TABLE 2. Environmental Causes of Malformations tions. Third, by the 1980s approximately 20% of the mal-
formed infants were in pregnancies that the parents had
1. Infants of pregestational diabetic mothers 183a chosen to terminate (Peller et al., 2004). These fetuses were
2. Intrauterine infection likely to have more severe abnormalities (Thomas et al.,
a) Cytomegalovirus 6 2016). The termination procedure makes a diagnostic
b) Toxoplasmosis 2
assessment of that infant’s phenotype more limited. Fourth,
there were no research study-based examinations of the
3. Exposures to teratogens
malformed infants. A study-based exam could have identi-
a) Fetal alcohol syndrome 12
fied important details not identified in routine examinations
a
b) Anticonvulsant drugs 29b and led to definitive diagnostic studies. Fifth, it was not pos-
c) Misoprostol 2c sible to preserve DNA obtained from the affected fetuses,
d) Chorionic villus sampling procedure 1c stillbirths and liveborn infants.
The evaluations of the many infants with common mal-
e) Dilation and curettage procedure 2c
formations made it possible to establish the etiologic hetero-
f) Anticoagulant warfarin 1d
geneity for several, including myelomeningocele, cleft lip
Total 238 and palate, hypospadias, cleft palate alone, esophageal atre-
a
Detailed information presented in separate article on the malforma- sia, and congenital diaphragmatic hernia (CDH) (Holmes,
tions in infants of diabetic mothers in this series of articles: Nasri 2012). This experience established the fact that clinicians
et al., 2018.
b
should expect to identify several different etiologies for
Infants with malformations identified by Surveillance Program; expo-
common malformations. These diverse causes have practical
sures to either monotherapy or polytherapy with valproate, carbama-
zepine, phenytoin, or phenobarbital. value. For example, the analysis of limb reduction defects
c
Infants described in a separate article on vascular disruption in this identified a subset which attributed to the process of vascu-
series of articles: Holmes et al., 2018. lar disruption. This established the baseline prevalence rate
d
An infant reported in a case series by Hall et al., 1980. which was used to assess whether or not the fetus exposed
90 CAUSES OF CONGENITAL MALFORMATIONS

to the prenatal diagnosis procedure chorionic villus sam- anticonvulsant-exposed infants with serious malformations
pling had a significant increase in the frequency of vascular remain a “likely” causal relationship.
disruption-type of limb defects (Golden et al., 2003). The instructive new findings being developed by chro-
Another example of the importance of identifying the differ- mosome microarray and whole genome or exomic sequenc-
ent causes of a common malformation would be in evaluat- ing will add new insights into the recognized causes in
ing infants with neural tube defects. To identify those who infants being born with malformations. Two examples are
could be affected by a preconception supplement with mul- the progress in evaluating infants with heart defects and
tivitamins and folate, those infants whose abnormality was infants born with CDHs, abnormalities attributed previously
associated with maternal diabetes mellitus, prenatal expo- to multifactorial inheritance. In an analysis of 58 infants
sure to valproate, chromosome abnormalities, and syn- born with congenital heart defects in 2011 to 2012, 6.7%
dromes with Mendelian inheritance should be excluded had abnormalities identified by chromosome analysis. An
(Holmes et al., 1976). additional 22.2% had abnormalities identified by chromo-
Another benefit of an active malformations surveillance some microarray (Bachman et al., 2015). In another study,
program is that the daily searching for each affected infant exomic sequencing of 1213 infants with congenital heart
in both live-born and deceased infants, including stillbirths disease (CHD) and their parents identified protein-
and elective terminations, can identify uncommon, un- damaging de novo mutations in 20% of the children with
planned, but significant, observations and provide a timely CHD and neurodevelopmental disabilities and in 2% of the
opportunity for documentation. Some examples of these children with isolated CHD (Homsy et al., 2015). The sec-
chance observations were: (1) infants in two informative ond example is from the analysis of DNA from infants with
families who died soon after birth and who subsequently congenital diaphragmatic hernia (Pober et al., 2005) by
had a similarly affected sibling (Holmes et al., 1995, 1997). whole exome sequencing. The recent study identified an
The collaboration with the pathologists made it possible to excess of damaging de novo gene disrupting and deleteri-
determine the phenotypic features and to describe these ous missense variants in 21% of the infants with multiple
two “new” hereditary phenotypes; (2) the evaluation of a malformations including CDH and in 12% of infants with
fetus with severe limb deficiencies made it possible for the isolated congenital diaphragmatic hernias (Longoni et al.,
pathologist to examine the histologic features of the 2017). The de novo variants were more common in genes
“nubbins” in a fetus with terminal transverse limb defects expressed in the diaphragm and heart of the developing
with nubbins, a rare opportunity (Drapkin et al., 2003); (3) mouse. Similar progress is being made in research studies
identifying the rare pregnancy that continued after a dila- on a variety of congenital malformations. This progress
tion and curettage procedure made it possible to evaluate could revise the classification of the etiologies identified as
the pattern of abnormalities produced in this seldom- “causes” of congenital malformations.
reported clinical setting (Holmes, 1995).
CONCLUSIONS
Before the findings reported by Feldkamp and her associ-
Malformations surveillance programs can delineate the eti-
ates (2017) and this analysis, the frequency of malformations
ologic heterogeneity of the common malformations. This
attributed to environmental factors had been estimates by
survey of the causes of malformations among 289,365
Kalter and Warkany (1983) and Brent (2001). Brent esti-
births over 41 years (1972–2012) identified 7020 (2.4%)
mated that 10% of malformations observed in the first year
infants with one or more malformations. The frequencies of
of life were due to environmental factors, including maternal
several causes were confirmed: malformations due to muta-
conditions, intrauterine infections, and prescription drugs,
tions with Mendelian inheritance, chromosome abnormal-
high dose radiation, and hyperthermia. The frequencies of
ities, complications of identical twinning, environmental
environmental factors identified by Feldkamp et al. (2017)
factors, and the process of vascular disruption. The new
was 4.1% and in this study, 3.4%. However, it is likely that
technologies, such as chromosome microarray and genome
many newborns with fetal alcohol syndrome were not identi-
sequencing, are beginning to identify the genetic abnormal-
fied. Some would have had associated malformations and
ities present in infants with malformations referred to previ-
would have been considered to have been caused by environ-
ously as exhibiting multifactorial inheritance.
mental factors, that is, prenatal exposure to an excessive
amount of alcohol.
Prenatal exposure to anticonvulsant medication is References
another common exposure that is difficult to prove as hav- Amberger JS, Bocchini CA, Schiettecatle F, et al. 2015. OMIM.org: online
ing “caused” malformations. Study examinations of infants Mendelian I inheritance in Man (OMIM), an outline catalog of human
genes and genetic disorders. Nucleic Acids Res 43:D789–D798.
exposed to valproate, phenytoin, carbamazepine, or pheno-
barbital could have identified the associated dysmorphic Bachman KK, DeWard SJ, Chrysostomou C, et al. 2015. Array
facial features and effects on the digits that would make a CGH as a first-tier test for neonates with congenital heart dis-
causal relationship more likely. Lacking that, the 29 ease. Cardiol Young 25:115–122.
BIRTH DEFECTS RESEARCH 110:87–91 (2018) 91

Brent RL. 2001. The cause and prevention of human birth defects: Holmes LB, Westgate M-N, Toufaily MH. 2018. Malformations
what have we learned in the past 50 years? Congenit Anom 41:3–21. attributed to the process of vascular disruption. Birth Defects
Res 110:98–107.
Carter CO, Evans K. 1973. Spina bifida and anencephalus in
greater London. J Med Genet 10:209–234. Homsy J, Zaidi S, Shen Y, et al. 2015. De novo mutations in con-
genital heart disease with neurodevelopmental and other con-
Carter CO. 1976. Genetics of common single malformations. Br genital anomalies. Science 350:1262–1266.
Med Bull 32:21–26.
Kalter H, Warkany J. 1983. Congenital malformations. N Engl J
Carter CO, Evans K, Pescia G. 1979. A family study of renal agen- Med 308:424–431, 491–497.
esis. J Med Genet 16:176–188.
Longoni M, High FA, Qi H, et al. 2017. Genome-wide enrichment
Correa A, Cragan JD, Kucik JE, et al. 2007. Metropolitan Atlanta of damaging de novo variants in patients with isolated and com-
Congenital Defects Program (MACDP). 40th Anniversary Edition plex congenital diaphragmatic hernia. Hum Genet 136:679–691.
Surveillance Report. Reporting birth defects surveillance data
1968-2003. Birth Defects Res A Clin Mol Teratol 79:65–93. McMullen KP, Karnes PS, Moir CR, Michels VV. 1996. Familial
recurrence of tracheoesophageal fistula and associated malfor-
Drapkin RI, Genest DR, Holmes LB, et al. 2003. Unilateral trans- mations. Am J Med Genet 63:525–528.
verse arm defect with subterminal digit nubbins. Pediatr Dev
Pathol 6:345–354. Nasri H, Ng KH, Westgate M-N, Holmes LB. 2018. Malformations
among infants of mothers with insulin-dependent diabetes: is
Feldkamp ML, Carey JC, Byrne JLB, et al. 2017. Etiology and clini- there a recognizable pattern of anomalies? Birth Defects Res
cal presentation of birth defects: population based study. BMJ 110:108–113.
357:j2249.
Nelson K, Holmes LB. 1989. Malformations due to presumed spon-
Fraser FC. 1970. The genetics of cleft lip and cleft palate. Am J taneous mutations in newborn infants. N Engl J Med 320:19–23.
Hum Genet 20:336–352.
Pauli RM, Reiser CA. 1994. Wisconsin Stillbirth Service Programs:
Fraser FC. 1986. Evolution of a palatable multifactorial threshold II. Analysis of diagnoses and diagnostic categories in the first
model. Am J Hum Genet 32:796–813. 1,000 referrals. Am J Med Genet 50:135–153.

Golden CK, Ryan LM, Holmes LB. 2003. Chorionic villus sam- Peller AJ, Westgate M-N, Holmes LB. 2004. Trends in congenital
pling: a distinctive teratogenic effect on fingers? Birth Defects malformations, 1974-1999: effect of prenatal diagnosis and elec-
Res A Clin Mol Teratol 67:557–562. tive termination. Obstet Gynecol 104:957–964.

Hall JG, Pauli RM, Wilson KM. 1980. Maternal and fetal sequelae Pober BR, Lin A, Russell MR, et al. 2005. Infants with Bochdalek
of anticoagulation during pregnancy. Am J Med 68:122–140. diaphragmatic hernia: sibling precurrence and monozygote twin
discordance in a hospital-based malformation surveillance pro-
Holmes LB, Driscoll SD, Atkins L. 1976. Etiologic heterogeneity gram. Am J Med Genet 138A:81–88.
of neural tube defects. N Engl J Med 294:365–369.
Schinzel AA, Smith DW, Miller JR. 1979. Monozygotic twinning
Holmes LB. 1995. Possible fetal effects of cervical dilation and and structural defects. J Pediatr 95:921–930.
uterine curettage during the first trimester of pregnancy.
J Pediatr 126:131–134. Shepard TH, Fantel AG, Fitzsimmons J. 1989. Congenital defect
rates among spontaneous abortuses: twenty years of monitoring.
Holmes LB, Redline RW, Brown DL, et al. 1995. Absence/hypo- Teratology 39:325–331.
plasia of tibia, polydactyly, retrocerebellar arachnoid cyst, and
other anomalies: an autosomal recessive disorder. J Med Genet Thomas EG, Higgins C, Westgate M-N, et al. 2018. Malformations
32:896–900. surveillance: comparison between findings at birth and age one
year. Birth Defects Res 110:142–147.
Holmes LB, Schoene WC, Benacerraf BR. 1997. New syndrome:
brain malformation, growth retardation, hypokinesia and polyhy- Thomas EG, Toufaily MH, Westgate M-N, et al. 2016. Impact of elec-
dramnios in two brothers. Clin Dysmorphol 6:13–19. tive termination on the occurrence of severe birth defects identified
in a hospital-based Active Malformations Surveillance Program
Holmes LB. 2012. Common Malformations. New York: Oxford (1999–2002). Birth Def Res (Part A): Clin Mol Teratol 106:659–
University Press. 666.

Holmes LB, Nasri H, Westgate M-N, et al. 2018. The Active Mal- Zhang Y-X, Zhang Y-P, Gu Y, et al. 2009. Genetic analysis of first-
formations Surveillance Program, Boston in 1972-2012: Method- trimester miscarriages with a combination of cytogenetic karyo-
ology and Demographic Characteristics. Birth Def Res 110: typing, microsatellite genotyping and array CGH. Clin Genet 75:
148–156. 133–140.