Human Reproduction vol.14 no.2 pp.521–531, 1999
Transcervical recovery of fetal cells from the lower
uterine pole: reliability of recovery and histological/
immunocytochemical analysis of recovered cell
populations
David Miller1,4, Jackie Briggs1,
Muhammed S.Rahman1, Martin Griffith-Jones1,
Vasu Rane1, Marianne Everett1, Richard J.Lilford2,
Judith N.Bulmer3
1Centre
for Reproduction, Growth and Development, Division of
Obstetrics, University of Leeds, Level D, Clarendon Wing, Leeds
General Infirmary, Belmont Grove, Leeds, LS2 9NS, 2Department
of Health Medicine, Arthur Thomson House, 142 Hagley Road,
Birmingham, B16 9PA, and 3Department of Pathology, University
of Newcastle, Royal Victoria Infirmary, Newcastle-upon-Tyne, NE1
4LP, UK
4To
whom correspondence should be addressed
The aim of this work was to isolate, enumerate and attempt
the identification of fetal cells recovered from the lower
uterine pole. Immediately before elective termination of
pregnancy at 7–17 weeks gestation, samples were recovered
by transcervical flushing of the lower uterine pole (n J
108) or transcervical aspiration of mucus from just above
the internal os (n J 187), and their contents examined
using histological, immunohistochemical and molecular
techniques. Syncytiotrophoblasts were identified morphologically in 28 out of 89 (31%) and 50 out of 180 (28%)
flushings and aspirates respectively (mean 29%). Immunocytochemistry with monoclonal antibodies (mAbs) recognizing trophoblast or epithelial cell antigens on a smaller
number of samples (n J 69) identified putative placental
cells in 13 out of 19 (68%) and 25 out of 50 (50%) flushings
and aspirates respectively (mean 55%). These included
groups of distinctive cells with a small, round, hyperchromatic nucleus, strongly reactive with mAbs PLAP, NDOG1
and FT1.41.1. Smaller groups of larger, amorphous cells,
usually containing multiple large, pale staining nuclei,
reactive with mAb 340 and to a lesser degree with mAb
NDOG5 were also observed. Taking cellular morphology
and immunophenotype into consideration, the smaller uninucleate cells were likely to be villous mesenchymal cells,
while the larger cells were possibly degrading villous
syncytiotrophoblast. There was no significant difference in
the frequency of fetal cells obtained by the two recovery
methods. Squamous or columnar epithelial cells, labelled
strongly with antibodies to cytokeratins or human milk fat
globule protein, were observed in 97% (29 out of 30) of
aspirates. The use of cervagem in a small number of
patients prior to termination of pregnancy did not appear
to influence the subsequent recovery of placental cells. Yspecific DNA was detected by polymerase chain reaction
(PCR) in 13 out of 26 (50%) flushings and (99 out of 154)
64% aspirates analysed (mean 62%). In-situ hybridization
© European Society of Human Reproduction and Embryology
(ISH) revealed Y-specific targets in 40 out of 69 (60%) of
aspirates analysed. A comparison of PCR data obtained
from transcervical recovered samples and placental tissues
showed a concordance of 80% (76 out of 95), with 10 false
positives. Comparing the PCR data from tissues with data
derived by ISH from 41 aspirates gave a concordance of
90% with two false positives. Although syncytiotrophoblasts
were much more likely to be present in samples containing
immunoreactive placental cells, the detection rates of fetalderived DNA were similar regardless of the morphological
and/or immunological presence of placental cells. We conclude that the transcervical recovery of fetal cells, while
promising, requires considerable additional effort being
expended in further research and development, particular
in the sampling procedure.
Key words: fetal cells/immunohistology/in-situ hybridization/
polymerase chain reaction/preimplantation diagnosis-transcervical sampling
Introduction
First trimester prenatal diagnosis of fetal abnormalities is of
singular importance to obstetricians who aim to minimize the
interventional trauma caused by late therapeutic termination
of pregnancy. Amniocentesis at 14 weeks’ gestation or later,
followed by cell culture and cytogenetic analysis is accepted
as one of the most reliable and safe methods of detecting
chromosomal abnormalities, although it carries a 0.5–1.0%
risk of causing a miscarriage. Moreover, results are rarely
obtained before the 15th week of pregnancy due to the time
required for cell culture. Chorionic villus biopsy (CVB),
although capable of providing information in the first trimester,
in addition to a 2–4% procedure-related risk of miscarriage,
is thought to be associated with an increased risk of fetal
abnormality such as defective limb development, presumably
due to haemorrhage or embolism from the aspirated placental
tissues (Jahoda et al., 1993; Fortuny et al., 1995; Lunshof
et al., 1995).
The replacement of both amniocentesis and CVB is now a
serious proposition following the identification of fetal cells
in the maternal circulation. These include nucleated erythroid
cells (Liou et al., 1993; Zheng et al., 1993), granulocytes
(Wessman et al., 1992) and trophoblasts (Cacheux et al., 1992;
Johansen et al., 1995). More widespread screening for relatively
common fetal abnormalities such as Down’s syndrome (Elias
et al., 1992) may become possible. However, all these circulating fetal cells are present in exceptionally low numbers in
maternal blood and their isolation is particularly difficult. To
521
D.Miller et al.
date, enrichment methods have included expensive fluorescence activated cell sorting (Bianchi, 1995); labour intensive
density gradient centrifugation followed by magnetic cell
sorting (Ganshirt et al., 1993); and most recently, charge flow
separation (Wachtel et al., 1996).
Using quinacrine staining of the Y chromosome, a number
of early reports suggested that fetal cells, shed from the
regressing chorion laeve into the lower uterine pole, could be
recovered by simple aspiration of endocervical mucus in the
first trimester (Shettles, 1971; Warren et al., 1973; Rhine et al.,
1975). Although conflicting reports which endorsed or refuted
these findings played a significant role in the abandonment of
further research (Bobrow and Lewis, 1971; Goldstein et al.,
1973; Manuel et al., 1974), the inability to culture recovered
cells in vitro and the consequent exclusion of classical cytogenetic analysis was of crucial importance. Since the advent of
molecular techniques, particularly polymerase chain reaction
(PCR) and fluorescence in-situ hybridization (FISH) which
offer a timely alternative to classical cytogenetic analysis, a
number of workers have attempted a reassessment of trans/
endocervical fetal cell sampling using several distinct sampling
procedures. In a preliminary report, Griffith-Jones et al. (1992)
using Y-specific primers, correctly predicted fetal sex by PCR
prior to confirmation by classical cytogenetic analysis in 25
out of 26 samples swabbed from the upper cervix. Later reports
(Briggs et al., 1995; Miller and Briggs, 1996) expanded on
these findings and included transcervical aspirates and flushings
(using 1–5 ml of normal saline) from an additional 124 elective
terminations of pregnancy at 7–17 weeks gestational age.
Other reports have used PCR and FISH for sex determination
or the detection of various chromosomal aneuploidies and
single-gene defects in samples obtained by transcervical procedures (Adinolfi et al., 1993, 1995a,b, 1997; Bahado-Singh
et al., 1995; Tustschek et al., 1995; Massari et al., 1996). Yet
there has been only one report to date describing the cellular
composition of samples recovered by these procedures in any
detail (Bulmer et al., 1995) and there have been few descriptions of the recovery frequency. Such information is essential
if the material giving rise to PCR or FISH signals is to be
identified and characterized and for possible anomalies arising
from sample heterogeneity to be better understood.
The aim of the present report is to address these issues and
we report on the analysis of molecular, morphological and
immunophenotypic information, amassed over 3 years of study.
bria Biologicals (Cramlington, UK) or Gibco BRL (Paisley, UK).
Standard and biotinylated deoxynucleotides were purchased from
Promega UK (Southampton, UK). Ultra-pure bovine serum albumin
was from BDH (Poole, UK). The secondary antibodies and developer
kits used in the study were from Vector Laboratories (Burlingame,
CA, USA). Aminopropyltriethoxysilane (APES) slide coating reagent
was from Sigma. All other reagent sources are detailed below.
Materials and methods
In-situ hybridization
The procedure has been described in detail elsewhere (Briggs et al.,
1995; Miller and Briggs et al., 1996) where representative illustrations
are shown. Briefly, samples were centrifuged at 500 g for 5 min, and
cytospun onto slides precoated with APES. After air drying, samples
were fixed in PBS containing 4% (w/v) paraformaldehyde for 30
min, washed in PBS for 5 min and transferred into PBS containing
0.25% (v/v) Triton X-100 and 0.25% (v/v) NP-40 for 10 min. After
washing twice in PBS, endogenous alkaline phosphatase activity was
blocked by 15 s immersion in 20% (v/v) acetic acid at 4°C followed
by a 5 min wash in PBS and 30 min incubation in 20% (v/v) glycerol
at room temperature. Slides were then rinsed twice in 23 sodium
chloride/sodium citrate (SSC) buffer, drained and incubated with 75
Ethical approval was obtained from Leeds Health Authorities (East
and West).
Subjects
A total of 295 patients were recruited into this study. All were
informed of and gave consent to transcervical sampling under general
anaesthesia, prior to the elective termination of pregnancy.
Reagents
All laboratory reagents were of ultrapure quality (molecular biological
grade, Sigma Chemical Co, Poole, UK) unless stated otherwise.
Proteinase K and Taq DNA polymerase was purchased from Northum-
522
Sample recovery
Two sampling methods were used, the first of which has been
described previously (Griffith-Jones et al., 1992). Briefly, a hollow
flexible embryo transfer trocar (Rocket Ltd, London, UK) was inserted
transcervically under ultrasound guidance into the lower uterine pole.
Sterile physiological saline (~5 ml) containing 1.0% w/v bovine
serum albumin (BSA) was introduced into the trocar and the fluid
allowed to enter the internal os prior to recovery by gentle suction.
These samples (n 5 108) will be referred to as flushings. For the
second method, the trocar was carefully positioned just above the
internal os, without the aid of ultrasound, and samples (n 5 187)
were directly aspirated into phosphate-buffered saline (PBS) containing 1.0% bovine serum albumin (PBS-A). These samples will be
referred to as aspirates.
All recovered samples were centrifuged at 500 g for 5 min and
resuspended in PBS-A. Aliquots from 26 flushings and 156 aspirates
were immediately stored at –20°C for subsequent analysis by PCR.
Cytospin preparations from all samples were routinely examined by
microscopy following staining with haematoxylin and eosin. Samples
of placental chorionic villi obtained from 41 patients following
termination of pregnancy (from whom aspirates were obtained) were
frozen and stored at –20°C, for separate PCR analysis. Additional
small samples of first-trimester placental villi were snap-frozen in
liquid nitrogen-cooled isopentane (BDH) for sectioning (see below).
Sample preparation for polymerase chain reaction
Samples were routinely checked for the presence of DNA by sequencespecific gene amplification of proteinase K-digested material, in the
presence of primers for β-actin as described previously (Briggs et al.,
1995). Actin and Y-specific DNA were amplified using the following
PCR programme: denaturation at 95°C for 5 min followed by 30
cycles of annealing at 56°C for 1 min, extension at 72°C for 1 min
and denaturation at 95°C for 1 min. A final 5 min extension at 72°C
was also included. Products were resolved on 1.8% (w/v) agarose
gels in Tris-borate EDTA buffer (25 mM Tris, 0.1 M sodium borate,
1 mM EDTA, pH 8.0) and visualized following staining with ethidium
bromide and ultraviolet illumination. Results were compared with
those obtained following the implementation of identical procedures
on fetal tissues from terminated pregnancies and the in-situ hybridization (ISH) of Y-specific DNA probes to cytocentrifuged samples
(see below).
Transcervical recovery of fetal cells
Table I. Specificities and reactivities of antibodies used in this study
Antibody
Specificity
Reactivity
Optimal Dilution
Source
PLAP
syncytiotrophoblast;
villous cytotrophoblast
syncytiotrophoblast; chorionic
villous mesenchyme
extravillous cytotrophoblast
placental alkaline
phosphatase
hyaluronic
acid
unknown
1:200
FT1.41.1
syncytiotrophoblast;
villous cytotrophoblast
syncytiotrophoblast
oncofetal
antigen
unknown
1:5
LP34
epithelial cells
cytokeratins 5, 6, 18
1:200
HMFG1
5D3
epithelial cells
epithelial cells
human milk fat globule
cytokeratin 8, 18
1:100
1:200
Novocastra Ltd,
Newcastle upon Tyne, UK
Kind gift of Dr Ian Sargent, John Radcliffe Hospital,
University of Oxford, UK
Kind gift of Dr Ian Sargent, John Radcliffe Hospital,
University of Oxford, UK
Kind gift of Dr Lindy Durrant,
University of Nottingham, UK
Kind gift of Professor Peter Johnson,
University of Liverpool, UK
Novocastra Ltd,
Newcastle upon Tyne, UK
Unipath Ltd, Bedford, UK
Novocastra Ltd, Newcastle upon Tyne, UK
NDOG1
NDOG5
340
1:10
1:10
1:2000
µl of a biotinylated Y-probe as described previously (Lewis and
Wells, 1992) in hybridization buffer [23 SSC, 5% (w/v) dextran
sulphate, 0.2% (w/v) dried milk powder containing formamide 50%
(v/v)]. The procedure was as follows: after heating to 95°C for 10
min followed by an overnight hybridization at 37°C, the slides were
rinsed twice in 23 SSC at room temperature followed by 20 min in
23 SSC at 60°C, 20 min in 0.23 SSC at 42°C and 5 min 0.13 SSC
at room temperature. Slides were then transferred into buffer A
[0.1 M Tris–Cl pH 7.5, 0.1 M NaCl, 2 mM MgCl2, 0.05% (v/v)
Triton X-100] containing 5% (w/v) BSA for 20 min prior to draining
and transfer into a humidified box. Streptavidin (10 µg/ml in buffer
A) was added to the slides prior to incubation at room temperature
for 10 min. Slides were washed in buffer A (235 min) prior to
incubation in buffer A containing biotinylated alkaline phosphatase
(10 µg/ml) for 10 min as above. Slides were washed twice washed
in buffer A for 5 min and then equilibrated for 30 min in buffer B
(0.1 M Tris–Cl, 0.1 M NaCl, 50 mM MgCl2, pH 9.5) containing
bichloroindolylphosphate (BCIP; 0.003% v/v) and nitrotetrazolium
blue (0.004% v/v); Gibco–BRL. Development was in the dark at
room temperature for 5–10 min. Nuclei were counterstained with 2%
(w/v) methyl green (3 min) and mounted in glycerol jelly (Merck, UK).
Histology and immunocytochemistry (ICC)
Where present, contaminating erythrocytes were lysed in 0.16%
(w/v) NH4Cl (Vickers Labs Ltd; Pudsey, UK) in distilled water at
room temperature. Samples were concentrated by centrifugation at
500 g for 5 min and resuspended in 0.2–3.0 ml PBS-A, depending
on the original cell density. Samples were then cytocentrifuged onto
APES-coated slides in 50–100 µl aliquots at 250 g for 10 min.
Following air drying, cytospin preparations were fixed for 5 min in
acetone at room temperature and either processed immediately or
stored wrapped in foil at –20°C. Sample cytospins were routinely
stained with Harris’s haematoxylin (BDH). For immunocytochemistry,
non-specific binding sites were blocked with normal horse serum
[Vector Laboratories; 20% (v/v) in Tris-buffered saline (TBS; 0.15
M NaCl, 0.05 M Tris–Cl pH 7.6)] and samples were incubated for
60 min with monoclonal antibodies (mAbs) directed against various
trophoblastic and maternal cell antigens, dissolved in PBS containing
0.1% (w/v) BSA; specificities, sources and dilutions are given in
Table I. After 60 min incubation in a humidified chamber, samples
were washed in three changes of TBS and incubated for 60 min with
biotinylated anti-mouse immunoglobulin (Vectastain Elite, Vector
Laboratories). Following further washing in TBS, an avidin–biotin
peroxidase complex (Vectastain Elite) was applied to the slides for 30
min. The reaction was developed with 0.025% (w/v) diaminobenzidine
Figure 1. Frequency of flushings (open bars) and aspirates (shaded
bars) obtained at 7–17 weeks gestation. Absolute numbers giving
rise to these frequencies are given above each bar.
(Sigma) and 0.025% (v/v) H2O2 (BDH) in TBS, counterstained with
haematoxylin, cleared in xylene (BDH) and mounted in DPX synthetic
resin (BDH). For comparison with recovered transcervical samples
and to monitor mAb specificities, cryostat sections (10 µm) of frozen
placental tissue were air-dried overnight at room temperature and
fixed for 5 min in acetone.
Results
Recovery efficiencies of flushings and aspirates
As most women are referred for termination of pregnancy at
a similar stage in their pregnancy, bias in the gestational ages
of recovered samples is unavoidable; most flushings and
aspirates were obtained at 10 weeks gestation (Figure 1). In
order to investigate how this bias affected the overall efficiency
of fetal cell recovery, the percentage of samples shown
to contain syncytiotrophoblast (the most easily recognized
placental cell) for each week of gestation was determined for
523
D.Miller et al.
both flushings and aspirates (Figure 2). Despite the smaller
numbers of flushed samples and the absence of any flushed
samples at .14 weeks gestation, it is clear that the relative
proportions of samples containing syncytia were similar. Furthermore, similar frequencies of syncytiotrophoblast recovery
were obtained in flushings (28 out of 89; 31%) and aspirates
(50 out of 180; 28%) despite the application of a cervagem
prostaglandin pessary prior to termination of pregnancy in
more flushings (10 out of 28) than aspirates (two out of 50)
containing syncytia (Table II). Although these data suggested
that syncytiotrophoblast recovery was not unduly affected by
either the sampling technique or prior prostaglandin application, the flushing procedure was quickly abandoned in favour
Figure 2. Frequency of samples containing syncytiotrophoblasts in
flushings (open bars) and aspirates (shaded bars) at 7–17 weeks’
gestation. Absolute numbers giving rise to these frequencies are
given above each bar.
of aspiration in order to avoid the introduction of fluid into
the uterine cavity.
Immunocytochemistry
As suggested previously from molecular data (Briggs et al.,
1995), unexpectedly low values obtained for syncytiotrophoblast recovery in both flushings and aspirates may not accurately reflect the presence of placental cells in these samples.
Hence, both flushings and aspirates were probed with a panel
of mAbs capable of detecting placental antigens. As the
same samples were used for several different experimental
procedures, it was not possible to test every sample with all
antibodies. Thus placental cells were detected in 13 out of 19
(68%) flushings and 25 out of 50 (50%) aspirates respectively
(Table II). Displaying the percentage of immunoreactive
samples following exposure to particular mAbs (Figure 3), it
is clear that the identification of placental cells was partly
dependent on the choice of mAb used. In our hands, mAb 340
consistently performed better than any of the other reagents,
with mAb NCL–PLAP lying close behind. A smaller group
of aspirates (n 5 30) was challenged with MAbs 5D3, LP34
and HMFG1 and immunoreactive cells were detected in 29
(98%). Cells immunoreactive with the common leukocyte
antigen, CD45, were found in half (five out of 10) of the
samples probed with the corresponding mAb. As many samples
were bloody on recovery, the significance of this observation
most likely relates to procedural trauma.
Placental sections
Patterns of placental antigen expression are shown in Figure
4. Of the mAbs recognizing placental antigens, PLAP and 340
provided the strongest labelling of both the syncytiotrophoblast
and villous cytotrophoblastic layers (Figure 4A and B). In
contrast, mAb NDOG1 only weakly labelled these layers
(Figure 4C) and mAb NDOG5 labelling was patchy within
the deeper layers of the tissue (arrowheads in Figure 4D).
mAb FT41.1 appeared to be specific for the syncytiotrophoblast
layer and the occasional deeper layer of villous cytotrophoblast
Table II. Summary of immunocytochemical data. Total values are given throughout
Containing syncytia (%)
Obtained following cervagem application (%)
Containing cells reactive with mAbs to placental antigens (%)
FT1.41.1
NCL-PLAP
NDOG-1
NDOG-5
340
Total placental Ag-positive (%)
Placental Ag-positive following cervagem (%)
Containing cells reactive with mAbs to non-placental antigens (%)
5D3
LP34
HMFG 1
CD45
Total non-placental Ag-positive (%)
Total non-placental Ag-positive following cervagem (%)
Ag 5 antigen; mAb 5 monoclonal antibody.
524
Aspirates
(n 5 187)
Flushings
(n 5 108)
50/180 (28)
2/50 (4)
28/89 (31)
10/28 (36)
10/29 (34)
11/19 (58)
6/18 (33)
6/24 (25)
5/8 (63)
25/50 (50)
3/25 (12)
2/6 (33)
3/5 (50)
0/4 (0)
1/3 (33)
11/14 (79)
13/19 (68)
5/13 (38)
20/20 (100)
6/6 (100)
15/16 (94)
5/10 (50)
25/26 (96)
0/0 (0)
3/3 (100)
3/3 (100)
4/4 (100)
0/3 (0)
Transcervical recovery of fetal cells
(lower left hand corner of Figure 4E) including the trophoblast
basement membrane (arrowheads). As expected, there was
strong labelling of the syncytiotrophoblast and villous cytotrophoblast with 5D3 (Figure 4F). LP34 also labelled these
layers but with less intensity (Figure 4G). HMFG1 faintly
labelled these and the deeper layers of tissue (Figure 4H).
Transcervically recovered samples
Representative examples of samples containing syncytiotrophoblast are shown in Figure 5, labelled with PLAP (Figure
5A); 340 (Figure 5B); NDOG1 (Figure 5C); NDOG5 (Figure
5D); and FT1.41.1. (Figure 5E). A sample stained with
haematoxylin alone is shown in Figure 5F for comparison. All
these mAbs recognized syncytiotrophoblasts, although the
intensity of labelling differed between cases. Figure 5A–C and
Figure 5D–F are from samples obtained by flushing and
aspiration respectively.
Representative examples of aspirates challenged with trophoblast-specific mAbs are shown in Figure 6 for mAbs PLAP
(Figure 6A,B); 340 (Figure 6C,D); NDOG1 (Figure 6E,F);
NDOG5 (Figure 6G,H) and FT1.41.1 (Figure 6I,J). These
samples were studied to look for trophoblast cells other than
syncytiotrophoblast and various immunoreactive cells can be
seen. Large groupings of well separated cells with uniformly
counterstained round nuclei and strongly labelled with mAbs
PLAP and FT1.41.1 (Figure 6B,J) were frequently observed.
Small clumps of more tightly associating cells, strongly labelled
with mAbs PLAP (Figure 6A) and NDOG1 (Figure 6F) were
also frequent. Larger cells with more diffuse, multiple, weakly
counterstaining nuclei and weakly labelled with mAbs 340
(Figure 6C,D) and NDOG5 (Figure 6H) were less common.
The solitary mAb NDOG1-labelled cell visible in Figure 6E,
(arrowhead) resembles those present in Figure 6B and J and
may be the same cell type. The intensely labelled cells present
in Figure 6G were also found in most samples containing
leukocytes and were CD45-positive (not shown). They are
included to illustrate the presence of peroxidase-containing
cells, most likely eosinophils.
The presence of maternal cells was assessed by morphological and immunocytochemical criteria and representative
examples are shown in Figure 7 for mAb 5D3 (Figure 7A,B);
mAb LP34 (Figure 7C,D) and mAb HMFG1 (Figure 7E,F).
These mAbs were particularly useful for identifying squamous
(Figure 7A,C) and columnar (Figure 7B) epithelial cells which
could only be of maternal origin. More ambiguously, the cell
clumps present in Figure 7A resemble those labelled by
trophoblast-reactive mAbs and, as shown in placental sections
(Figure 5F,G), mAbs 5D3 and LP34 react with all trophoblast
cells. However, as HMFG is known to recognize epithelial
cells (Burchell et al., 1983), the cells shown in Figure 7E and
F are probably of cervical origin.
Comparison between molecular and cellular data
Cellular data suggested that ~50% of samples (flushings and
aspirates combined) did not contain placental cells. However,
independent estimates of placental cell presence were available,
based on determining fetal sex by PCR and ISH. The accuracy
of these data was assessed by comparison with corresponding
samples of placental tissue obtained following the termination
of pregnancy. Y-specific DNA was detected by PCR in 13 out
of 26 (50%) and 99 out of 154 (64%) samples recovered by
flushing and aspiration respectively (mean 62%). The difference in these values was not statistically significant and
compared favourably with the frequency of aspirates (60%)
containing Y chromosome targets as determined by ISH (40
out of 69). More importantly, data from placental tissues were
80% (76 out of 95) and 90% (37 out of 41) concordant with
the corresponding data derived from aspirates by PCR or ISH
respectively. Data for the 41 samples in which a three-way
comparison based on PCR and ISH on aspirates with PCR on
CV samples was carried out are shown in Table III. The 80%
concordance was due to eight discordant aspirates consisting
of five false (male) positives and three false (male) negatives.
Only eight out of 41 (20%) aspirates in this grouping contained
syncytia. Taking (placental) antigen-positive (Ag1) and negative (Ag–) aspirates into consideration where complementary
molecular data were available directly, 11 out of 20 Ag1
(55%) and eight out of 14 Ag– (57%) samples gave rise to YPCR products. Taking data from aspirates shown to contain
syncytiotrophoblast by histological analysis and subsequently
processed for immunocytochemistry, 12 out of 25 Ag1 (48%)
and one out of 29 Ag– (3%) aspirates contained syncytiotrophoblast, a statistically significant difference (χ: P , 0.05).
Discussion
Figure 3. Frequency of samples containing labelled cells from
flushings (open bars) and aspirates (shaded bars) following
exposure to a range of monoclonal antibodies. Absolute numbers
giving rise to these frequencies are printed above each bar.
Our main objectives were to examine transcervically recovered
samples for the presence of fetal material, describe the consituent cells and determine whether these cells could be separated
into fetal and maternal compartments, a prerequiste for the
detection of single gene defects. While we were unable to
meet the third objective, our findings have a bearing on the
525
D.Miller et al.
Figure 4. Sections of first-trimester placenta were probed with trophoblast-reactive monoclonal antibodies (mAbs) (A) NCL-PLAP; (B) 340;
(C) NDOG1; (D) NDOG5 and (E) FT1.41.1. Reactivities to mAbs-5D3 and LP34, which recognize epithelial cell cytokeratins, are shown in
(F) and (G) respectively. A section labelled with HMFG1 is shown in (H). Arrowheads in (D) and (E) indicate pockets of cellular and
basement membrane labelling, respectively. Original magnification: A, B, C, D, G, H 3160; E, F 3250.
potential use of transcervical sampling for prenatal diagnosis
in a clinical setting.
Most of the samples recovered during the study contained
intact cells of maternal origin; ectocervical squamous and
columnar epithelial cells, often in large numbers, were present
in 98% aspirates. In contrast, 55% of all samples examined
contained cells which were immunoreactive with one or more
of the mAbs recognising placental antigens. These findings
should be viewed in the context of the considerable variation
in reactivities of the panel of mAbs used in this study. mAb
526
340, for example, recognized cells in 16 out of 22 (73%)
samples compared with the next ‘best’ reagent, NCL–PLAP
at 14 out of 24 (58%). These data suggest that estimates for
placental cell presence may have been underestimated by the
poor sensitivity of reagents used for their detection. Detection
rates for syncytiotrophoblasts are probably more accurate but
reveal a relatively low level of abundance as illustrated by
combining the data obtained by their direct visual observation
with immunocytochemical data for all putative placental cells
(Figure 8).
Transcervical recovery of fetal cells
Figure 5. The presence of syncytiotrophoblasts in transcervically
recovered samples (A–C and F 5 flushings; D, E 5 aspirates)
were assessed following (F) haematoxylin staining alone or
following immunocytochemistry with monoclonal antibodies (A)
PLAP; (B) 340; (C) NDOG1; (D) NDOG5, and (E) FT1.41.1.
Original magnification: all 3250.
Morphologically distinct immunoreactive cell types were
observed, including discrete multinucleate syncytiotrophoblasts, larger cells containing more than one nucleus, groupings
of loosely or tightly associating cells and the occasional solitary
cell. With the exception of syncytiotrophoblast, definitive
identification of placental cells was not achieved due to
limitations in the reagents used and inter-sample variability in
their reactivity with similar cell types (see below). However,
some tentative suggestions could be made, based on cellular
morphology, the known reactivity of the mAbs and the pattern
of mAb reactivities for the ‘same’ cell type. In general, groups
of tightly associating cells were mAb PLAP/NDOG1 positive,
while the more loosely associating cell clumps were mAb
PLAP/FT1.41.1-positive. Hence, the tightly packed PLAP/
NDOG1-positive cells could have been derived from the
chorionic villous mesenchyme where both alkaline phosphatase
and hyaluronic acid are present. The loosely associating PLAP/
FT1.41.1-positive cells are more difficult to identify because
placental alkaline phosphatase is ubiquitously expressed in
the placenta. Since mAb FT1.41.1 recognized the villous
cytotrophoblast layer in placental sections, albeit weakly, these
isolated cells may be derived from this layer but could
equally derive from the villous mesenchyme. In view of these
immunophenotypes, it is unlikely that any of the PLAPreactive cells are cervical in origin, despite the reported low
level presence of a placental-like alkaline phosphatase in
cervical tissues (McLaughlin et al., 1984). The larger cells
containing more than one nucleus were usually mAb 340/
NDOG5-positive. While we were unable to confirm that these
cells were not villous cytotrophoblast, mAb 340 which is
known to recognize this cell type did not recognize equivalent
groups of cells in any of the samples challenged.
Based on the reactivity of mAb NDOG5 for extravillous
cytotrophoblast, the large 340/NDOG5-positive cells could be
of this cell type; however, in view of the occasional positive
labelling of isolated syncytiotrophoblasts with this mAb and
the syncytial appearance of these cells, it is also conceivable
that they are fragmented and degrading syncytiotrophoblast.
The latter explanation is supported by our earlier work (Briggs
et al., 1995) showing the presence of naked nuclei in recovered
samples. Although NDOG5 is not noted for its recognition of
syncytiotrophoblast in sections (as supported by this study)
there is evidence for the expression of the corresponding
epitope in syncytiotrophoblasts from term placentas
(M.Johansen, personal communication) and it is conceivable
that these giant cells could alter their pattern of antigen
expression following isolation from the main placental mass.
Artefacts may be an inevitable consequence of sampling; gain
or loss of syncytiotrophoblast antigen expression may occur
by the time the material is recovered. Alternatively, fixation
conditions can vary between whole sections and isolated
cells, masking or revealing epitopes accordingly. NDOG5,
for example, recognized syncytia in the two out of six
immunoreactive samples containing these cells (both samples
obtained at 10 weeks gestation) but not in any of the first
trimester control sections (8–10 weeks gestation).
There have been few reports describing the cellular composition of transcervically recovered samples in any detail. Rhine
et al. (1975) provided fluorescent micrographs showing clusters
of fetal cells with a ragged appearance which, in relation to
the much larger squamous cells also present and their similar
morphology to cells observed in the present study, were
probably from the villous mesenchyme. While agreeing with
our findings in relation to squamous and endocervical cells,
most samples recovered by aspiration and flushing by Bulmer
et al. (1995) contained syncytial fragments and cytotrophoblasts as demonstrated by morphology, histological staining
and immunocytochemistry. Rodeck et al. (1995) detected
syncytiotrophoblasts in 18% of samples obtained by endocervical cytobrushing and intrauterine lavage and cytotrophoblasts
in all samples. The higher levels of fetal cells reported in both
studies may have been due to their use of different mAbs or
to differences in respective sampling procedures. Rodeck et al.
(1995) included two mAbs (IO3 and H315) which may have
been more sensitive in this respect. Ultrasound guidance, which
was not used to aid the recovery of aspirates in this study,
527
D.Miller et al.
Figure 6. The presence of other trophoblast cells in aspirated samples was assessed following immunocytochemistry with monoclonal
antibodies (A, B) PLAP; (C, D) 340; (E, F) NDOG1; (G, H) NDOG5 and (I, J) FT1.41.1. Arrowheads in E, G and H indicate cells of
interest. Original magnification: A, F, H, I 3250; B, C, D, E, G, J 3400.
may also have been a contributory factor, although the similar
frequency of syncytiotrophoblast recovery in flushings
(obtained under ultrasound guidance) and aspirates suggests
otherwise. In contrast, Bahado-Singh et al. (1995) identified
trophoblasts in only five out of 10 samples obtained by flushing
using a mAb recognising human chorionic gonadotrophin.
Alternative explanations for the discrepant data regarding
cytotrophoblast recovery involve the choice of sampling procedure and its related mechanical effect on cellularity. While
intra-uterine flushing produced the best data with regard to
528
placental cell recovery and cellularity, it is unlikely that
invasive introduction of fluid in pregnancies destined for
continuation would be acceptable and this led us to abandon
this procedure early in the study. However, rather than the
perceived compromise to uterine sterility, it now appears that
mechanical trauma induced by changes in intra-uterine pressure
is a more compelling reason not to flush, following the recent
report of severe limb reduction defects in a male neonate after
an uneventful pregnancy in which first trimester flushing was
carried out (Chou et al., 1997). The aspiration data of Rodeck
Transcervical recovery of fetal cells
Figure 7. The presence of maternal cells in aspirated samples was assessed following immunocytochemistry with monoclonal antibodies (A,
B) 5D3; (C, D) LP34 and (E, F) HMFG1. Original magnification: A, B 3400; C, D, E, F 3250.
et al. (1995) were derived from the endocervix and external
cervical os using different mAbs and therefore are not directly
comparable with our aspiration data. In our earlier report
(Briggs et al., 1995; Miller and Briggs, 1996), the presence
of hybridising acellular targets to Y-probes, often in close
proximity to syncytiotrophoblasts, suggested that mechanical
disruption or other physico-biochemical effects were major
causes of cell loss during transcervical aspiration.
High levels of maternal cell contamination identified as
cervical squamous and columnar epithelial cells, as well as
leukocytes presumed to be of maternal origin, were a common
feature in both flushed and aspirated samples and was in
agreement with Rodeck et al. (1995). Leukocyte contamination
is inevitable under conditions which promote bleeding, including trauma and prior application of prostaglandins. While
reductions in their presence may have been brought about by
the use of alternative sampling tools and techniques, intersubject differences in upper genital tract physiology may also
have been an important and unavoidable factor in maternal
cell load.
While it is possible that changes in the handling of samples
might minimize cellular damage, poor cellularity due to
unavoidable natural reductions in placental cell content, as
indicated at least for syncytiotrophoblast recovery, would
restrict the use of recovered material to interphase aneuploid
detection. Certainly no cell culture of such samples would be
possible (Bahado-Singh et al., 1995; Ishai et al., 1995),
effectively excluding any classical cytogenetic analysis. The
fate of placental cells in recovered samples is of some concern
since the molecular data strongly indicated the presence of
fetally-derived DNA in most cases. In this respect, comparison
of the molecular and cellular data is revealing and supports a
number of conclusions. Firstly, the slightly higher than expected
number of designated male aspirates derived by PCR/ISH is
probably an artefact of occasional contamination with male
DNA, since the number of Y-PCR-positive samples was
similar for both antigen positive and negative aspirates (56%)
compared with all aspirates tested (64%). Secondly, the detection of fetally-derived DNA was not dependent on the presence
of intact placental cells as indicated by the large number of
aspirates containing hybridizable targets by ISH, which did
not contain immunoreactive placental cells (Briggs et al.,
1995; Miller and Briggs, 1996). This conclusion is supported
by the data showing that placental antigen positive samples
were more likely to contain syncytiotrophoblasts than antigen
negative samples, while the presence or absence of placental
cells made little difference to the detection of Y-DNA.
The concordance of our molecular data was 80% comparing
529
D.Miller et al.
Table III. Concordance of molecular data using polymerase chain reaction
(PCR) and in-situ hybridization (ISH). Numerical values for the two- and
three-way concordance are given as footnotes
Sample
Gestation
(Week)
Syncytia
(S)
Y-PCR
(T)
Y-PCR
(S)
ISH
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
Total
8
8
8
8
8
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
12
12
12
12
12
12
12
12
12
13
14
14
14
14
14
14
14
16
16
41
–
–
1
–
–
–
–
–
–
–
–
–
–
–
1
–
–
1
–
1
–
–
1
1
–
–
–
–
–
1
–
1
–
–
–
–
–
–
–
–
–
8 (20%)
1
–
1
1
1
–
–
1
1
1
1
1
–
–
1
–
1
1
1
1
1
–
–
1
1
1
–
1
1
1
1
1
1
1
1
1
–
1
–
1
–
29 (71%)
1
1
1
1
–
–
–
1
1
1
1
–
–
–
1
–
–
1
1
–
1
1
–
1
1
1
–
1
1
1
1
1
1
1
1
1
–
1
–
1
–
27 (66%)
1
1
1
1
–
–
–
1
1
1
1
–
–
–
1
–
–
1
1
–
1
–
–
1
1
1
–
1
1
1
1
1
1
–
1
1
1
1
–
1
1
27
(66%)
S 5 transcervically retrieved samples; T 5 placental tissue.
Two-way analysis: Y-PCR(T) versus Y-PCR(S): 35/41 (85%) concordance;
four false positives/two false negatives; Y-PCR(T) versus ISH(S): 37/41
(90%) concordance; two false positives/two false negatives.
Three-way analysis: Y-PCR(T) versus Y-PCR(S) versus ISH(S): 33/41
(80%) concordance.
data obtained from recovered placental tissues by PCR and
aspirates by both PCR and ISH and 90% when considering
ISH alone. These data compare with 96% accuracy in our
original PCR-based investigation of 26 samples recovered
by endocervical swabbing (Griffith-Jones et al., 1992). The
difference most likely reflects the larger sample size since our
original report and the greater likelihood of false-positives
when using PCR. Using a similar three-way technique, Adinolfi
et al. (1995b) reported 86% concordance between fetal tissues
obtained following termination of pregnancy and samples
recovered by lavage or cytobrush, using PCR and FISH;
530
Figure 8. Frequency of samples containing any trophoblasts or
other placental cells, displayed as a stack column chart. ST 5
syncytiotrophoblasts.
12 out of 22 (54%) samples contained Y-specific DNA as
determined by PCR. Three known male samples (one lavage
and two cytobrush) were non-informative because of a failure
to detect Y-targets by FISH. Interestingly, in the same study,
paternally-derived alleles were detected in only four out of 16
(25%) lavages and four out of 29 (14%) aspirates following
amplification of chromsome 21-specific short tandem repeats.
This is a clear demonstration of the likely requirement for
some form of fetal cell enrichment to counteract noise generated
by the maternal genome.
In conclusion, the latest data confirm that fetal material,
more often in the form of DNA/nuclei than intact cells, is
deported into the lower uterine pole during the first trimester
of pregnancy and can be recovered by transcervical flushing
or aspiration. However, the clinical relevance of this procedure
has not yet been proven. All studies have reported considerable
variation in both the composition and quality of recovered
material, due to sampling errors and operator discontinuities.
It is also by no means certain that the procedure is any safer
than CVS (Chou et al., 1997), despite the small trial carried
out by Rodeck et al. (1995), or indeed as reliable, given the
heterogeneity of the samples.
A much larger study is now required to establish the
application of transcervical sampling as an alternative route to
prenatal diagnosis. This would require close collaboration
between groups who are currently working in this area, in a
number of different settings. At stake is a clearer resolution
of the precise niche in which the procedure is likely to be
placed, standing as it now does somewhere between impending
non-invasive maternal blood sampling on the one hand and
existing invasive but highly accurate prenatal diagnostic procedures on the other.
Transcervical recovery of fetal cells
Acknowledgements
This work was supported by the Medical Research Council of the
UK. We are grateful to all the patients who participated in this study.
References
Adinolfi, M., Davies, A., Sharif, S. et al. (1993) Detection of trisomy-18 and
Y-derived sequences in fetal nucleated cells obtained by transcervical
flushing. Lancet, 342, 403–404.
Adinolfi, M., Sherlock, J., Soothill, P. et al. (1995a) Molecular evidence of
fetal-derived chromosome-21 markers (STRS) in transcervical samples.
Prenat. Diag., 15, 35–39.
Adinolfi, M., Sherlock, J., Tutschek, B. et al. (1995b) Detection of fetal
cells in transcervical samples and prenatal diagnosis of chromosomal
abnormalities. Prenat. Diag., 15, 943–949.
Adinolfi, M., El-Hashemite, N., Sherlock, J. et al. (1997) Prenatal detection
of Hb mutations using transcervical cells. Prenat. Diag., 17, 539–543.
Bahado-Singh, R.O., Kliman, H., Feng, T.Y. et al. (1995) First-trimester
endocervical irrigation – feasibility of obtaining trophoblast cells for
prenatal-diagnosis. Obstet. Gynecol., 85, 461–464.
Bianchi, D.W. (1995) Prenatal-diagnosis by analysis of fetal cells in maternal
blood. J. Pediatr., 127, 847–856.
Bobrow, M. and Lewis, B.V. (1971) Unreliability of fetal sexing using cervical
material. Lancet, ii, 486.
Briggs, J., Miller, D., Bulmer, J.N. et al. (1995) Non-syncytial sources of
fetal DNA in transcervically recovered cell populations. Mol. Hum. Reprod.,
10, 749–794.
Bulmer, J.N., Rodeck, C. and Adinolfi, M. (1995) Immunohistochemical
characterization of cells retrieved by transcervical sampling in earlypregnancy. Prenat. Diag., 15, 1143–1153.
Burchell, J., Durbin, H. and Taylor-Papadimitrious, J (1983) Complexity of
expression of antigenic determinants recognised by monoclonal antibodies
HMFG1 and HMFG2 in normal and malignant mammary epithelial cells.
J. Immunol., 131, 508–513.
Cacheux, V., Milesifleut, C., Tachdjian, G. et al. (1992) Detection of 47, XYY
trophoblast fetal cells in maternal blood by fluorescence in situ hybridization
after using immunomagnetic lymphocyte depletion and flow-cytometric
sorting. Fet. Diag. Ther., 7, 190–194.
Chou, M.M., Lin, S.K. and Ho, E.S. (1997) Severe limb reduction defects
after uterine lavage at 7–8 weeks’ gestation, Prenat. Diag., 17, 77–80.
Elias, S., Price, J., Dockter, M. et al. (1992) 1st trimester prenatal-diagnosis
of trisomy-21 in fetal cells from maternal blood. Lancet, 340, 1033–1033.
Fortuny, A., Borrell, A., Soler, E. et al. (1995) Chorionic villus sampling by
biopsy forceps. Results of 1580 procedures from a single centre. Prenat.
Diag., 15, 541–550.
Ganshirt, A.D., Borjesson, S.R., Burschyk, M. et al. (1993) Detection of fetal
trisomies 21 and 18 from maternal blood using triple gradient and magnetic
cell sorting. Am. J. Reprod. Immunol., 30, 194–201.
Goldstein, A.I., Lukesh, R.C. and Ketchum, M. (1973) Prenatal sex
determination by fluorescent staining of the cervical smear for the presence
of a Y chromosome: An evaluation. Am. J. Obstet. Gynecol., 115, 866.
Griffith-Jones, M.D., Miller, D., Lilford, R.J. et al. (1992) Detection of fetal
DNA in transcervical swabs from 1st trimester pregnancies by gene
amplification – a new route to prenatal-diagnosis? Br. J. Obstet. Gynaecol.,
99, 508–511.
Ishai, D., Amiel, A., Diukman, R. et al. (1995) Uterine cavity lavage: adding
FISH to conventional cytogenetics for embryonic sexing and diagnosing
common chromosomal aberrations. Prenat. Diag., 15, 961–965.
Jahoda, M.G.F.J., Brandenburg, H., Cohen-Overbeek, T. et al. (1993) Terminal
transverse limb defects and early chorionic villus sampling: evaluation of
4300 cases with completed follow-up. Am. J. Med. Genet., 46, 483–485.
Johansen, M., Knight, M., Maher, E.J. et al. (1995) An investigation of
methods for enriching trophoblast from maternal blood. Prenat. Diag., 15,
921–931.
Lewis, F.N. and Wells, M. (1992) Detection of virus in infected human tissue
by in situ hybridisation. In Hames, B.D. and Rickwood, D. (eds), In situ
Hybridisation, A Practical Approach. IRL Press, Oxford, UK, pp. 121–136.
Liou, J.D., Pao, C.C., Hor, J.J. et al. (1993) Fetal cells in the maternal
circulation during the 1st trimester in pregnancies. Hum. Genet., 92, 309–311.
Lunshof, S., Boer, K., Leschott, N.J. et al. (1995) Pregnancy outcome after
transcervical CVS: Evaluation of risk factors. Prenat. Diag., 15, 809–816.
McLaughlin, P,J., Travers, P.J., McDicken, I.W. et al. (1984) Demonstration
of placental and placental-like alkaline phosphatase in non-malignant human
tissue extracts using monoclonal antibodies in an enzyme immunoassay.
Clin. Chim. Acta, 137, 341–348.
Massari, A., Novelli, G., Colosimo, A. et al. (1996) Non-invasive early
prenatal molecular diagnosis using retrieved transcervical trophoblast cells.
Hum. Genet., 97, 150–155.
Manuel, M., Park, M.D. and Jones, H.W. (1974) Prenatal sex determintaion
by fluorescent staining of cells for the presence of Y chromatin. Am. J.
Obstet. Gynecol., 119, 853–854
Miller, D. and Briggs, J. (1996) Reliability of transcervical recovery of
placental cells from the lower uterine pole using a minimally invasive
procedure. Evidence based on fetal sexing and analysis of recovered cell
populations. Early Hum. Dev., 47, S99–S102.
Rodeck, C., Tutschek, B., Sherlock, J. et al. (1995) Methods for the
transcervical collection of fetal cells during the first trimester of pregnancy.
Prenat. Diag., 15, 933–942.
Rhine, S.A., Cain, J.L., Cleary, R.E. et al. (1975) Prenatal sex detection with
endocervical smears: successful results utilizing Y-body fluorescence. Am.
J. Obstet. Gynecol., 122, 155–158.
Shettles, L.B. (1971) Use of the Y-chromosome in prenatal sex determination.
Nature, 230, 52.
Tutschek, B., Sherlock, J., Halder, A. et al. (1995) Isolation of fetal cells
from transcervical samples by micromanipulation – molecular confirmation
of their fetal origin and diagnosis of fetal aneuploidy. Prenat. Diag., 15,
951–960.
Wachtel, S.S., Sammons, D., Manley, M. et al. (1996) Fetal cells in maternal
blood: recovery by charge flow separation. Hum. Genet., 98, 162–166.
Warren, R., Sanchez, L., Hammond, D. et al. (1972) Prenatal sex determination
from exfoliated cells found in endocervical mucus. Am. J. Hum. Genet.,
24, 22.
Wessman, M., Linen, K. and Knuutila, S., (1992) Fetal granulocytes in
maternal venous blood detected by in situ hybridization. Prenat. Diag., 12,
993-1000.
Zheng, Y.L., Carter, N.P., Price, C.M. et al. (1993) Prenatal-diagnosis from
maternal blood – simultaneous immunophenotyping and FISH of fetal
nucleated erythrocytes isolated by negative magnetic cell sorting. J. Med.
Genet., 30, 1051–1056.
Received on May 13, 1998; accepted on October 28, 1998
531