PRENATAL DIAGNOSIS, VOL.

15: 209-214 (1995)

AN INCREASED INCIDENCE OF
HAEMANGIOMAS IN INFANTS BORN
FOLLOWING CHORIONIC VILLUS SAMPLING

BARBARA K. BURTON, CHARLENE I. SCHULZ, BRAD ANGLE AND LAURENCE I. BURD

University of Illinois College of Medicine and Center f o r Medical and Reproductive Genetics, Michael Reese
Hospital and Medical Center, Chicago, IL, U.S.A.
Received November I993
Revised April I994
Accepted April 1994

SUMMARY
The incidence of haemangiomas was ascertained by questionnaire in infants born to 578 consecutive CVS patients
and 445 consecutive mid-trimester amniocentesis patients seen at a single institution between 1 January 1989 and 31
May 1991. The incidence of 7.4 per cent reported in the amniocentesis group was comparable to previous estimates
of the incidence of haemangiomas in the general population. In contrast, a 21.1 per cent incidence, three-fold higher
than that observed in the amniocentesis group, was observed among CVS-exposed infants (P<O.OOl).This increased
incidence was largely confined to patients undergoing a transcervical procedure. No correlation was observed
between the incidence of haemangiomas and gestational age at sampling, sample size, number of sampling attempts,
or a history of bleeding following the procedure.
KEY WORDS: chorionic villus sampling; amniocentesis; haemangiomas

INTRODUCTION MATERIALS AND METHODS

Several recent studies have suggested that there
might be an increased incidence of haemangiomas
in infants born following chorionic villus sampling
(CVS). In 1990, Kaplan et aZ. examined 95 infants
born following CVS and 87 born following amnio-
centesis and reported a 12.6 per cent incidence of
haemangiomas in the CVS group as opposed to 3-4
per cent in the amniocentesis group. More
recently, Burton et al. (1993) noted that among a
group of 14 CVS-exposed infants with limb disrup-
tion defects, 50 per cent had one or more hae-
mangiomas. T h s prompted us to examine the
incidence of haemangiomas in a series of patients
undergoing CVS at our institution.
Addressee for correspondence: Barbara K. Burton, MD,
Michael Reese Hospital, 2929 S. Ellis Avenue, Chicago, IL
60616-3390, U.S.A.
CCC 01 97-385 1/95/030209-O6
0 1995 by John Wiley & Sons, Ltd.
This study focused on all patients with singleton
pregnancies undergoing CVS (n=578) or mid-
trimester amniocentesis ( ~ 2 ~ 4 4 5at) Michael Reese
Hospital and Medical Center between 1 January
1989 and 31 May 1991. Approximately 95 per cent
of all CVS procedures were performed by a single
perinatologist, whereas 5 per cent were performed
by fellows under the direct supervision of the same
perinatologist. CVS procedures were performed
transcervically, using either a Trophocan catheter
(ConcordRortex, Keene, NH) or a CVS catheter
manufactured by Cook Ob/Gyn (Spencer, IN), or
transabdominally, using an 18-gauge thin-walled
needle with stylet, under continuous ultrasound
guidance. Details of the CVS technique utilized
have been previously described (Burton et al.,
1992). CVS procedures were performed at 9-12
weeks' gestation.
Amniocentesis was performed in 90 per cent of
cases by one of several experienced perinatologists
210 B. K. BURTON ET AL.
under continuous ultrasound guidance using a
22-gauge needle at 14-20 weeks’ gestation.
Approximately 10 per cent of amniocentesis
procedures were performed by residents in obstet-
rics and gynecology under the supervision of a
perinatologist.
CVS patients were routinely contacted by mail
approximately 6 weeks after the expected date of
delivery and were asked to complete a brief ques-
tionnaire regarding the outcome of their preg-
nancy. Data on the outcome of pregnancy
obtained from this contact had been previously
published for most of the CVS subjects except for
those seen between 1 December 1990 and 31 May
1991 (Burton et al., 1992). At the time of this initial
contact, data were obtained regarding compiica-
tions of pregnancy, labour, and delivery, and birth
defects in the infant but there were no specific
questions related to birthmarks. Therefore, a se-
cond questionnaire specifically addressing birth-
marks and haemangiomas was sent to all CVS
subjects in November 1992. Questions were
worded in such a way as to differentiate between
haemangiomas (strawberry birthmarks), common
vascular stains (such as stork bites and salmon
patches), and other types of birthmarks. Patients
were asked to report any skin lesions developing
during the first year of life as well as those present
at the time of birth.
Amniocentesis subjects had not been previously
contacted for pregnancy outcome data, so were
asked to complete a brief outcome questionnaire
addressing birth weight, gestational age at deliv-
ery, and the presence or absence of birth defects, in
addition to the same questionnaire about birth-
marks and haemangiomas sent to CVS subjects.
Both groups received a cover letter explaining that
the purpose of the study was to determine the
incidence of various types of birthmarks and hae-
mangiomas (described as benign overgrowths of
blood vessels often referred to as strawberry birth-
marks) in infants born to mothers undergoing
prenatal testing. Because of the timing of the
mailing, all infants were at least 1 year of age at the
time of this contact. Some were as old as 3 years of
age.
A second mailing was sent to subjects who did
not respond within 4 weeks. If there was no
response to the second mailing, a telephone con-
tact was attempted. In the case of positive
responses indicating the presence of haemangio-
mas, paediatric records and/or photographs were
obtained for confirmation or verbal confirmation
Table I-Pregnancy outcome and participation rates in
CVS and amniocentesis subjects
CVS Amniocentesis
Total number 518 445
NO follow-up 28 65
Spontaneous abortion 29 4
Elective abortion 21 11
Known livebirth 500 365
Questionnaire completed 432 365
was obtained over the telephone from the infants’
physicians.
RESULTS
Questionnaires were completed by a total of 432
CVS subjects. This represents 86.4 per cent of CVS
subjects known to have had a livebirth and 96.4
per cent of those with current addresses who could
be contacted (see Table I). Questionnaires were
completed by 364 amniocentesis subjects, which
represents 84.7 per cent of the original group
(excluding those known to have had a spontaneous
or elective abortion) and 96-6 per cent of those
with current addresses who could be contacted.
The characteristics of the two groups of respon-
dents are shown in Table 11. CVS subjects were
slightly, but not significantly, older than amnio-
centesis subjects and a significantly larger percent-
age were white. Most subjects in both groups
underwent testing because of advanced maternal
age but a significantly larger percentage of amnio-
centesis subjects (27.2 per cent) than CVS subjects
(10.3 per cent) had other indications, most
commonly elevated or low maternal serum
alpha-fetoprotein (MSAFP).
The incidence of haemangiomas in the CVS and
amniocentesis groups is shown in Table 111.
Because a significant difference was observed
between the transcervical CVS (TC-CVS) and
transabdominal CVS (TA-CVS) groups, the data
have been presented for all CVS subjects and
separately for the TC-CVS and TA-CVS groups. A
total of 13 subjects who underwent a single
attempt at sampling using both TC and TA tech-
niques were included in the total CVS group
only. A striking three-fold increased incidence of
haemangiomas was noted in the total CVS group
with one-third of the infants with haemangiomas
 HAEMANGIOMAS IN CVS-EXPOSED INFANTS 21 1

Table 11-characteristics of the CVS and amniocentesis groups

CVS n=432 Amniocentesis n=365 P

Age (mean) 36.3 & 2.7 35.0 f 2.4 NS

Race (YOwhite) 85.0% 63.0% <0.001
Indication for procedure (% AMA) 89.8% 72.9% <0.001
Mean GA at time of procedure (weeks) 10.6 16.6 <0.001

AMA=Advanced maternal age; GA=gestational age; NS=difference not significant.

Table 111-Incidence of haemangiomas in CVS- and amniocentesis-exposed infants

All CVS TC-CVS TA-CVS Amniocentesis

n=432 n = 243 n=176 n=365 PI p2 p3
~~ ~~~ ~

Single 61 (14.1%) 48 (19.8%) 13 (7.4%) 27 (7.4%) <0.005 <0.001 NS

Multiple 30 (6.9%) 25 (10.3Yo) 5 (2.8%) 0 <0.001 <0.001 CO.01
AU haemangiomas 91 (21.10/0) 73 (30.0%) 18 (10.2%) 27 (7.4%) <0.001 <0.001 NS

TC-CVS=Transcervical CVS; TA-CVS = transabdominal CVS; P , =all CVS vs. amniocentesis; P2=TC-CVS vs. amniocentesis;
P,=TA-CVS vs. amniocentesis; P,,Pz and P, all determined by xz analysis; NS=difference not significant, D-0.05.

having multiple lesions. This is in contrast to the
amniocentesis group in which there were no
infants with multiple haemangiomas reported.
Most of the haemangiomas in the CVS-exposed
infants were observed in the group exposed to
TC-CVS, where the overall incidence was four
times higher than that observed in the
amniocentesis-exposed group. The incidence in
the TA-CVS group was slightly higher than that in
the amniocentesis group but the difference was
statistically significant only when infants with mul-
tiple haemangiomas were considered separately.
Within the TC-CVS groups, a higher percentage of
patients sampled with the Cook catheter (50/156;
32.1 per cent) than with the Trophocan catheter
(23/87; 26.4 per cent) reported haemangiomas but
the difference was not statistically significant.
Because there were a larger number of white
subjects in the CVS group than in the amnio-
centesis group, the incidence of haemangiomas
was analysed separately in white vs. non-white
subjects. This is seen in Table IV. Although the
incidence of haemangiomas was significantly
higher in non-white than in white subjects, the
incidence was s i d c a n t l y greater in CVS-exposed
infants than in amniocentesis-exposed infants in
both groups. The distribution of CVS subjects
between the TC-CVS and TA-CVS groups was
similar for whites and non-whites.
All the haemangiomas reported except one were
cutaneous. A single patient in the amniocentesis
group reported a laryngeal haemangioma. A single
patient in the CVS group reported an infant with a
large port wine stain of the face. There were no
other reports of major vascular birthmarks in
either group. Two of the CVS patients reported to
have haemangtomas were among the four infants
with limb disruption defects previously reported
from this same series (Burton et al., 1992).
A number of factors related to the CVS proce-
dure were examined in CVS subjects who reported
infants with and without haemangiomas. The
mean sample weight for TC-CVS subjects with vs.
without haemangiomas was 14.0 mg vs. 13.7 mg;
for TA-CVS subjects it was 12.8 mg vs. 11.6 mg.
The mean gestational age at the time of sampling
was 10-4weeks vs. 10.6 weeks for those with and
without haemangiomas, respectively. Ten of 73
TC-CVS subjects (13.7 per cent) with haemangio-
mas had more than one sampling attempt vs. 29 of
170 (17.1 per cent) without. For TA-CVS subjects,
the comparable numbers were 1/18 (5.6 per cent)
for those with haemangiomas vs. 13/158 (8.2 per
cent) for those without. Vaginal bleeding or spot-
ting was reported following the CVS procedure by
18.0 per cent of TC-CVS subjects with haemangto-
mas vs. 16.8 per cent of those without and by 4.2
per cent of TA-CVS subjects with vs. 3.5 per cent
212 B. K. BURTON ET AL.
Table IV-Incidence of haemangiomas in CVS and amniocentesis infants, separated
into white and non-white groups
White CVS n=367
Single 48 (13.1%)
Multiple 25 (6.8%)
All haemangiomas 73 (19.9%)
Non-white CVS n=65
Single 13 (20.0%)
Multiple 5 (7.7%)
All haemangiomas 18 (27.7%)
P determined by x2 analysis; NS=difference not statistically significant, P>0.05.
of TA-CVS subjects without haemangiomas. None
of these differences was statistically significant.
DISCUSSION
Earlier studies comparing the safety of CVS and
amniocentesis for the prenatal diagnosis of genetic
disorders identified an increased risk of fetal loss
following CVS but no increased incidence of birth
defects or other abnormalities in CVS-exposed
infants (Rhoads et al., 1989; Canadian Collabora-
tive CVS-Amniocentesis Clinical Trial Group,
1989). Kaplan et al. in 1990 reported the frequency
of major and minor malformations in a group of
infants following CVS and amniocentesis and
noted a 12-6 per cent frequency of haemangiomas
in the CVS group as opposed to 3.4 per cent in the
amniocentesis group. The number of reported
infants was small and the frequency reported in the
amniocentesis group was lower than that previ-
ously reported in the general population, making it
difficult to interpret these findings. More recently,
several reports have suggested the possibility of an
increased risk of oromandibular limb hypogenesis
and isolated limb disruption defects following CVS
(Firth et al., 1991; Burton et al., 1992; Report of
the NICHHD Workshop on Chorionic Villus
Sampling and Limb and Other Defects, 1993).
Although this association remains controversial, a
number of plausible mechanisms have been put
forward including transient fetal hypoperfusion
secondary to bleeding into the sampling site and/or
the release of vasoactive substances from the
placenta causing vasoconstriction or haemorrhage
Amniocentesis n =230 P
12 (5.2%) <0.005
0 co.00 1
12 (5.2%) <0.001
Amniocentesis n = 135 P
15 (11.1%) NS
0 <0.005
15 (11.1%) <0.025
in the fetus. Quintero et al. in 1993 reported
embryoscopically detectable haemorrhagic lesions
on the fetus following CVS performed prior to
elective abortion, particularly when the placenta
was deliberately traumatized during the CVS pro-
cedure. They suggested that there might be a
relationship between these haemorrhagic lesions
and the disruptive defects associated with CVS.
We previously reported the findings in a series of
14 CVS-exposed infants with limb disruption
defects and noted that 50 per cent of the infants
had one or more haemangiomas (Burton et al.,
1993). This finding, in conjunction with the previ-
ous report by Kaplan et al., prompted us to study
the frequency of haemangomas in our entire CVS
population.
Previous studies have suggested that the inci-
dence of haemangiomas in the general population
during the first year of life is approximately 8-12
per cent (Jacobs and Walton, 1976; Holmdahl,
1955). The incidence of 7.4 per cent noted in our
population of amniocentesis-exposed infants is
comparable to this. In contrast, a significantly
increased frequency of haemangiomas was noted
in our CVS-exposed infants. This increased inci-
dence was largely confined to patients who under-
went a TC-CVS procedure. The occurrence of
haemangiomas did not appear to be related to any
other factors examined such as sample weight,
number of sampling attempts, gestational age at
sampling, or a history of bleeding following the
procedure. None the less, a correlation with one or
more of these factors might have been observed
if larger numbers of subjects had been available
for study. Although haemangiomas were more
 HAEMANGIOMAS IN CVS-EXPOSED INFANTS
common in non-white subjects, the increased inci-
dence in the CVS group was observed in both
whte and non-white subjects.
The mechanism by which haemangiomas
develop in the general population is not under-
stood. Therefore, it is difficult to suggest a mech-
anism relating their occurrence to the TC-CVS
procedure. It is tempting to speculate that the fetus
might respond to the haemorrhagic lesions
described by Quintero et al. in a manner that
would predispose to the development of a haeman-
gioma at that site.
There are obvious pitfalls in a questionnaire
study of this nature which must be kept in mind in
the interpretation of the data. There could be
underascertainment of haemangiomas if mothers
were unaware of their presence on their infants,
had forgotten about lesions which had already
disappeared, or did not recognize the name or
description given for haemangiomas. The fact that
the frequency noted in the amniocentesis group
was comparable to that previously reported in the
general population is somewhat reassuring in that
regard. It is also possible that mothers of infants
with haemangiomas were more likely to respond to
the questionnaire than those without, leading to an
overestimation of their frequency. While this is
certainly possible, we believe that it would be
unlikely that this would sigmficantly alter the
findings since it should be true for both the CVS
and the amniocentesis groups. In the case of
patients living in the Chicago area, it is possible
that CVS mothers may have scrutinized their
infants more carefully than amniocentesis mothers
as a result of exposure to media coverage of the
possible link between CVS and limb defects. While
this could have led to overparticipation in the
study by mothers whose infants had haemangio-
mas, it could not entirely explain the findings.
Even if it is assumed that none of the CVS infants
lost to follow-up had haemangiomas, the incidence
in the CVS group would still be 18.2 per cent (91 of
500). T h s is substantially higher than any previous
estimate of the frequency of haemangiomas in the
general population.
Although there is clearly an increased incidence
of haemangiomas in our series, it is possible that
our CVS population does not reflect the entire
population of CVS-exposed infants worldwide if
there are site-specdc differences which relate to the
associations observed with both limb disruption
defects and haemangomas in this patient popula-
tion. Although no specific technical factors, aside
213
from the gestational age at the time of sampling,
have previously been shown to correlate with
CVS-associated limb disruption defects, it is pos-
sible that one or more will be identified in the
future. The observation that the fetal loss rate
among TC-CVS patients previously reported in
our series is higher than that reported in many
other TC-CVS series has been cited as evidence
that greater placental trauma may have occurred
in our patients (Blakemore et al., 1993).
The occurrence of one or more cutaneous hae-
mangiomas in an otherwise normal infant is rarely
of medical or cosmetic significance. We believe
that it is important, however, that other studies
be done in other patient populations to c o n b n
the association with CVS observed in our series.
Ideally, infants would be systematically examined
in CVS and control groups not only to determine
the frequency of haemangiomas, but also to assess
whether or not the lesions occurring in CVS-
exposed infants are different from those occurring
in the general population. Additional data may
help to shed some light on the mechanism by
which haemangiomas develop, in addition to
furthering our understanding of how the fetus is
affected by placental manipulation, including
chorionic villus sampling, in early gestation.
REFERENCES
Blakemore, K., Filkins, K., Luthy, D.A., Platt, L.D.,
Medearis, A.L., Carlson, D., Priest, J., Korotkin, J.,
Verp, M.S., Padilla, L.-M., Warsof, S., Weston, P.V.
(1993). Cook obstetrics and gynecology catheter mul-
ticenter chorionic villus sampling trial: comparison of
birth defects with expected rates, Am. J. Obstet.
Gynecol., 169, 1022-1026.
Burton, B.K., Schulz, C.J., Burd, L.I. (1 992). Limb
anomalies associated with chorionic villus sampling,
Obstet. Gynecol., 79, 726-730.
Burton, B.K., Schulz, C.J., Burd, L.I. (1993). The
spectrum of limb disruption defects associated with
chorionic villus sampling, Pediatrics, 91, 989-993.
Canadian Collaborative CVS-Amniocentesis Clinical
Trial Group (1989). Multicentre randomised clinical
trial of chorion villus sampling and amniocentesis:
first report, Lancet, 1, 1-6.
Firth, H.V., Boyd, P.A., Chamberlain, P., MacKenzie,
I.Z., Lindenbaum, R.H., Huson, S.M. (1991). Severe
limb abnormalities after chorion villus sampling at
5 6 6 6 days’ gestation, Lancet, 337, 762-763.
Holmdahl, L.K. (1955). Cutaneous hemangiomas in
premature and mature infants, Acta Paediatr., 44,
370-379.
Jacobs, A.H., Walton, R.G. (1976). The incidence of
birthmarks in the neonate, Pediatrics, 58, 2 18-222.
214 B. K. BURTON ET AL.
Kaplan, P., Normandin, J.J., Wilson, G.N., Plauchu,
H., Lippman, A,, Vekemans, M. (1990). Malforma-
tions and minor anomalies in children whose mothers
had prenatal diagnosis: comparison between CVS and
amniocentesis, Am. J. Med. Genet., 37, 366-370.
Report of the NICHHD Workshop on Chorionic Villus
Sampling and Limb and Other Defects (1993). Am. J.
Obstet. Gynecol., 169, 1-6.
Rhoads, G.G., Jackson, L.G., Schlessehan, S.E., De La
Cruz, F.F., Desnick, R.J., Golbus, M.S., Ledbetter,
D.H., Lubs, H.A., Mahoney, M.J., Pergament, E.,
Simpson, J.L., Carpenter, R.J., Elias, S., Ginsberg,
N.A., Goldberg, J.D., Hobbins, J.C., Lynch, L.,
Shiono, P.H., Wapner, R.J., Zachary, J.M. (1989).
The safety and efficacy of chorionic villus sampling
for early prenatal diagnosis of cytogenetic abnormali-
ties, N. Engl. J. Med., 320, 609-617.
Quintero, R.A., Romero, R., Mahoney, M.J.,
Abuhamad, A., Vecchio, M., Holden, J., Hobbins,
J.C. (1993). Embryoscopic demonstration of hemor-
rhagic lesions on the human embryo after placental
trauma, Am. J. Obstet. Gynecol., 168, 756-759.