Human Reproduction Vol.22, No.1 pp. 167–173, 2007
doi:10.1093/humrep/del320
Advance Access publication August 18, 2006.
Towards a non-invasive method for early detection
of testicular neoplasia in semen samples by identification
of fetal germ cell-specific markers
C.E.Hoei-Hansen1, E.Carlsen, N.Jorgensen, H.Leffers, N.E.Skakkebaek and
E.Rajpert-De Meyts
University Department of Growth and Reproduction, Rigshospitalet, Copenhagen, Denmark
1
To whom correspondence should be addressed at: University Department of Growth and Reproduction (GR-5064), Rigshospitalet,
Blegdamsvej 9, DK-2100 Copenhagen, Denmark. E-mail: [email protected]
BACKGROUND: Testicular germ cell tumours (TGCTs) originate from a common precursor, carcinoma in situ
(CIS). Diagnosis at the CIS stage is desirable as it minimizes the necessary treatment. A detailed clinical evaluation of
an approach to detect CIS cells in the ejaculate using primordial germ cell/gonocyte markers is presented. METHODS:
Immunocytological staining for AP-2g [and in some cases, OCT-3/4, NANOG or placental alkaline phosphatase
(PLAP)] was performed in semen samples from 294 infertile patients and 209 patients with TGCTs or other diseases.
RESULTS: Presence of AP-2g-stained cells was detected in 50% of participants with CIS and in 33.9% of TGCT
patients before treatment (non-seminomas: 56.6%, seminomas: 17.4%). OCT-3/4 results were similar to those of AP-2g,
whereas NANOG and PLAP stainings were unsuitable. Sensitivity was 54.5% for participants harbouring pre-invasive
CIS but reduced in participants with overt TGCTs, perhaps because of obstruction. Assay specificity was 93.6%, positive predictive value (PPV) 83.3% and negative predictive value (NPV) 60.3%. CONCLUSIONS: Immunocytological semen analysis based on expression of fetal germ cell markers in exfoliated cells has auxiliary diagnostic value, as
it detects some patients with CIS/incipient tumour, but a negative result does not exclude TGCT. Further effort is
needed to improve this assay, for example, by employing a more sensitive biochemical method of detection.
Key words: AP-2γ/carcinoma in situ/intratubular germ cell neoplasia/OCT-3/4/testicular germ cell tumour
Introduction
The majority of patients with testicular germ cell tumours
(TGCTs) survive with the current treatment regime, but regardless of recent improvements in the outcome of this disease, it is
potentially lethal, especially in poor-prognosis patients with
disseminated non-seminomas and in patients with relapsed testis cancer, which is often refractory to chemotherapy (Schmoll
et al., 2004). The subset of patients that require chemotherapy
and/or radiotherapy may experience side effects, severe psychological stress and reduction of their reproductive potential
(Brydoy et al., 2005; Huddart et al., 2005). TGCTs of young
adults originate from a common precursor, the carcinoma in
situ (CIS) cell (Skakkebaek, 1972), also known as intratubular
germ cell neoplasia. Because CIS can be cured by low-dose
irradiation or unilateral orchidectomy alone, there is a great
incentive for non-invasive or minimally invasive methods for
detection of CIS. A surgical biopsy is at present the only reliable method for diagnosis of CIS, and unilateral or bilateral
biopsies are performed in selected patients at risk of CIS, for
example, those with a history of cryptorchidism or where clinical examination has revealed atrophic testes or ultrasonic
microlithiasis (irregular echo pattern). Hence, very few
patients are diagnosed at the CIS stage, particularly because
most patients with CIS have no symptoms (Hoei-Hansen et al.,
2005a). An additional incentive for obtaining diagnosis at the
pre-invasive stage is the concern for the patients’ fertility.
While the presence of an overt tumour requires surgery within
a few days, a patient with CIS can be given some weeks, or
even months, to fulfil his wish of fatherhood or to cryopreserve
semen.
Seminiferous tubules with CIS are nearly always present in
the vicinity of the tumour (Skakkebaek, 1975; Jacobsen et al.,
1981; Oosterhuis et al., 2003). Abnormal cells in semen samples from patients with TGCTs were first described based on
their cytological features (Czaplicki et al., 1987; Howard et
al., 1989; Yu et al., 1990). Our group demonstrated cells with
CIS characteristics in semen of TGCT patients, based on
immunohistochemical detection of the oncofetal marker M2A
(Giwercman et al., 1988a; Giwercman et al., 1988b) or based
on aneuploidy of CIS cells (Giwercman et al., 1988c). The
possible presence of cells with isochromosome 12p was
explored in another study (Meng et al., 1998). However, owing
© The Author 2006. Published by Oxford University Press on behalf of the European Society of Human Reproduction and Embryology. All rights reserved.
For Permissions, please email: [email protected]
167
C.E.Hoei-Hansen et al.
to a large overlap with control specimens, hyperdiploidy turned
out not to be a parameter predictive of testis tumour (Salanova
et al., 1999). An approach of flow cytometry, immunochemistry and enzymochemistry for alkaline phosphatase was reported,
but it failed to discriminate CIS because of the high level of
cross-reacting germ cell alkaline phosphatase (Brackenbury
et al., 1993). RT-PCR-based detection has been attempted by
our group (results not published) but showed a very low specificity. Thus, none of the described methods based on examination of semen samples have proved sufficiently reliable to be
used for diagnostic purposes in the clinical setting, mainly due
to frequent false-negative or false-positive results.
Nevertheless, we decided to re-visit the immunocytological
approach after the discovery of novel transcription factors
specific for neoplastic germ cells in the adult testis (Almstrup
et al., 2004), and localized to the cell nucleus, which is usually
well preserved in semen. We recently published an encouraging
case report, where we diagnosed a young man with CIS testis,
based on an immunocytological detection of the marker AP-2γ
(TFAP2C) in CIS-like cells in his ejaculate (Hoei-Hansen
et al., 2005b). AP-2γ is expressed in fetal gonocytes, but not in
adult germ cells or structures in the ejaculatory pathway, and
has been established as a marker of CIS and many types of
overt TGCTs, including the most common classical seminoma
and the embryonal carcinoma component of non-seminomas
(Almstrup et al., 2004; Hoei-Hansen et al., 2004; Hong et al.,
2005; Pauls et al., 2005). Since then, we have tested this
approach to detect CIS on more than 500 patients and controls
in our andrology clinic. Here, we present our detailed experience with the putative method, which has also been expanded
to analyse other stem cell and gonocyte marker proteins,
including OCT-3/4, a transcription factor that is also not
expressed in the adult testis and genital tract and is a marker of
CIS, seminoma and embryonal carcinoma (Looijenga et al.,
2003; Rajpert-De Meyts et al., 2004).
Materials and methods
Characteristics of participants and semen samples
A total of 503 men were included in the study. A trained doctor
approached participants for inclusion in the project in May 2004–
January 2006. Participants attended our department either for semen
cryopreservation or for semen analysis as part of an andrological evaluation and were enrolled in this study if at least 200 μl of the sample
was available after standard analysis. Inclusion was after written
informed consent, and the local medical research ethics committee
approved the study. Approximately 92% of the participants were Danish
(based on their surnames). Data regarding referral and history were
attained from medical charts. Period of abstinence and concentration
of spermatozoa was recorded according to the World Health Organization 1992 guidelines (World Health Organization, 1992), with small
modifications (Jorgensen et al., 1997).
The participants were divided into groups as shown in Table I: (i)
TGCT before orchidectomy [histology was seminoma (n = 35), nonseminoma (n = 23) and mixed GCT (n = 7), (ii) TGCT after orchidectomy [histology was seminoma (n = 19), non-seminoma (n = 19) and
mixed GCT (n = 4)], (iii) CIS-only (three had unilateral CIS-only and
three had CIS in the testis contralateral to a testis removed because of
TGCT), (iv) other testicular tumours/lesions [including Leydig cell
168
tumour (n = 1), spermatocytic seminoma (n = 1), leiomyosarcoma
(n = 1), adenomatoid tumour (n = 2), spermatocele (n = 1) and fibrous
changes (n = 2)], (v) patients referred to semen analysis because of
infertility, (vi) a group of other control participants with various diseases including participants with gonadal-related diseases/conditions
[(n = 13), including testis torsio, inflammation, sterilization and
inguinal hernia], non-testicular malignant diseases [(n = 19), including
extragonadal GCT (this patient had no CIS in bilateral testicular biopsies), prostate cancer, lymphoma and leukaemia], other diseases [(n = 21),
including prostate hypertrophia, rheumatoid diseases, epilepsia and
sclerosis] and apparently healthy young semen donors or participants
in other research projects (n = 35). Abstinence period (number of days
since last ejaculation) was for the whole group 3.6 days, with small
variations among the groups. Mean age of participants was 32.9 years
(range 15.9–59.3), which was very similar in all groups except the
group of patients with CIS-only where mean age was 25.5 years.
Number of analyses per participant was in the range of 1–10 analyses,
where more analyses were performed for participants with borderlinepositive results. Evaluation of immunocytological staining as a
putative diagnostic test was determined based on the included participants where bilateral histology was known (n = 151). This group
included andrological patients who had testicular biopsies performed
and TGCT patients where histology of the tumour was known and
who all had a contralateral biopsy performed as is standard practice in
Denmark.
Immunocytochemistry
Semen cytospin preparations were obtained for all included participants.
Staining with AP-2γ was performed in all cases (n = 503), staining
with OCT-3/4 in random samples where additional cytospins were
available (n = 84 cases) and staining with NANOG and placental alkaline phosphatase (PLAP) in a few test samples. A small aliquot of a
fresh semen sample from each participant was diluted according to the
following protocol: 0–25 × 106 spermatozoa/ml non-diluted, 25–45 ×
106 diluted 1:2, 45–55 × 106 diluted 1:3, 55–65 × 106 diluted 1:4, 65–
75 × 106 diluted 1:5, 75–85 × 106 diluted 1:6, 85–95 × 106 diluted 1:7,
95–130 × 106 diluted 1:8, 130–180 × 106 diluted 1:9, 180–230 × 106
diluted 1:10, 230 × 106 and above diluted 1:15. For each cytospin, 100
μl was used, and the amount of cells present in each cytospin was
approximately between 0 and 2.5 × 106 per cytospin. One hundred
microlitres of the diluted portions was spun down on a microscope
slide with 400 μl of PBS buffer, dried, fixed 10 min in formalin,
washed, dried again and stored up to 2 weeks at room temperature.
The following antibodies were used: monoclonal mouse anti-AP-2γ
antibody (6E4/4:sc-12762, Santa Cruz Biotechnology Inc., CA,
USA), monoclonal mouse anti-OCT-3/4 (C-10, sc-5279, Santa Cruz
Biotechnology), polyclonal goat anti-human NANOG (AF1997, R&D
Systems, MN, USA) and monoclonal mouse anti-PLAP, (BioGenex,
San Ramon, CA, USA). A standard indirect peroxidase method was
used for staining, as previously described (Hoei-Hansen et al., 2004;
Rajpert-De Meyts et al., 2004; Hoei-Hansen et al., 2005c), except for
small modifications especially regarding concentration of the antibodies, which were adjusted after analysis of control samples of testicular
tissue containing AP-2γ-positive CIS or semen samples spiked with
AP-2γ-positive seminoma cells (Figure 1E,F). Briefly, to unmask the
antigen, the fixed cytospins were heated in a microwave oven in a
buffer (either TEG buffer = TRIS 1.21 g/l, EGTA 0.19 g/l, pH 9.0; 5%
urea buffer, pH 8.5 or citrate buffer = 10 mmol/l, pH 6.0). Subsequently,
the samples were incubated with 1.5% H2O2 to inhibit the endogenous
Non-invasive detection of testicular neoplasia
Table I. Characteristics of the participants
Participant category
n
Age (years)
Sperm concentration
(×106/ml)
Abstinence
(days)
Number of analyses
per participant
Testicular germ cell tumour before orchidectomya
Testicular germ cell tumour after orchidectomy
Carcinoma in situ-only
Other testicular tumours/lesions
Infertilityb
Other
Total
65
42
6
8
294
88
503
31.4 (15.9–43.2)
32.6 (15.9–51.1)
25.5 (20.3–30.1)
33.1 (18.8–41.6)
33.7 (18.8–53.6)
32.1 (18.5–59.3)
32.9 (15.9–59.3)
23.8 (0–153)
14.4 (0–119)
2.9 (0–8.2)
71.6 (0–331)
23.0 (0–244)
53.3 (0–266)
28.1 (0–331)
3.1 (1–22)
4.9 (1–42)
1.8 (1–3)
3.9 (2–7)
3.7 (1–30)
3.1 (1–11)
3.6 (1–42)
2.1 (1–6)
1.5 (1–9)
2.3 (1–4)
2.1 (1–4)
1.4 (1–9)
1.3 (1–10)
1.5 (1–10)
For each group, the number of participants, mean age, sperm concentration, abstinence time and number of analyses per participant are listed (95% confidence
intervals are given in brackets).
a
Three of these patients had had the other testis removed because of a testicular germ cell tumour. Three other participants had contralateral carcinoma in situ, all of
whom had a borderline score.
b
Including the participant diagnosed with carcinoma in situ in the course of the study (SC-008) and the infertile participant who had a false-negative AP-2γ staining
(SC-371).
Figure 1. Examples of staining results according to the categories used for evaluation. (A) AP-2γ-stained cells. A1-4 classified as positive, A5-8
classified as borderline, A9-12 classified as negative. A1 from a participant with testicular CIS before irradiation, after orchidectomy of contralateral teratoma; A2 from participant SC-228, a control participant where biopsies have not been performed; A3 from a participant with a seminoma
before orchidectomy, no contralateral CIS; A4 from a participant with unilateral CIS diagnosed in the course of the study; A5 from an infertile
participant; A6 from a patient with an embryonal carcinoma and teratoma before orchidectomy, no contralateral CIS; A7 from an infertile participant, where similar, unspecific staining was seen in the negative control cytospin; A8 from an infertile participant, where bilateral biopsies were
without CIS; A9-12 all from infertile participants. (B) Examples of staining with OCT-3/4; B1 and B3 from patients with a known testicular cancer; B2 and B4 from infertile participants. (C) Examples of staining with PLAP in a patient with CIS. (D) Examples of non-specific staining with
NANOG in a control participant. (E) AP-2γ-stained semen sample with spiked seminoma cells. (F) AP-2γ-stained testicular tissue with a tubule
containing CIS. CIS, carcinoma in situ; PLAP, placental alkaline phosphatase. Magnification is the same for all images; scale bar = 25 μm.
169
C.E.Hoei-Hansen et al.
peroxidase, followed by diluted non-immune goat or human serum to
block unspecific binding sites. Incubation with diluted primary antibodies
(AP-2γ 1:30–1:50, OCT-3/4 1:200–1:250, NANOG 1:50 and PLAP
1:400) was carried out overnight at 4°C, followed by a secondary link
antibody and the horse-radish peroxidase–streptavidin complex.
Between all steps, the sections were thoroughly washed. The bound
antibody was visualized using aminoethyl carbazole substrate (all reagents from Zymed, S. San Francisco, CA, USA). Sections were lightly
counterstained with Mayer’s haematoxylin to mark unstained nuclei.
For negative control, for each participant, another cytospin was
treated as above, except that it was only incubated with the dilution
buffer. The sections were examined under a light microscope (Zeiss,
Oberkochen, D) and scored systematically and independently by two
investigators (C.H.H. and E.R.M) with no prior knowledge of the
diagnosis. The time course from staining to analysis was in the range
of 1–14 days. Stained cells were scored using an arbitrary scale based
on the intensity and morphological resemblance of stained elements to
CIS or tumour cells: ‘negative’ = no staining or only some unspecific particles stained, including trace reaction in small fragments of nucleusresembling structures; ‘borderline’ = clear staining in a part of a
nucleus or weak staining in a cell with correct morphology; and ‘positive’ = clear staining in one or several whole nuclei with correct, CISlike morphology with a large nucleus and prominent nucleoles (see
examples of the staining categories in Figure 1A). A final score was
determined as the maximum score obtained in one of the analysed cytospins. The scoring was slightly modified from the categorization
used in the previous publication (Hoei-Hansen et al., 2005b), as we in
the present study applied a more strict definition of positive scoring
and added the category ‘borderline‘. Discrepancy in positive/negative
scoring results (especially in the ‘borderline’ category) was detected
in 7.2% of analyses, in which cases slides were re-evaluated by the
observers together and a consensus score was reached. A tendency
towards fading of stained elements with time was noted, perhaps especially in elements stained unspecifically. In two cases, positivestained elements were detected in both AP-2γ-stained and negative
control cytospins and were thus regarded as non-specific.
marker PLAP were much less consistent with a substantial
amount of unspecifically stained elements. The PLAP example
shown is from a participant harbouring approximately 150 AP2γ-stained CIS cells in another cytospin (Figure 1C). NANOG
is a nuclear CIS and GCT marker, but the example of NANOG
staining is from an infertile participant where repeated AP-2γstained cytospins were negative, whereas the applied polyclonal NANOG antibody had scattered non-specifically stained
elements (Figure 1D). The NANOG and PLAP antibodies were
judged unsuitable for further use in this project and were abandoned after a few experiments.
The combined results of AP-2γ and OCT-3/4 stainings are
shown in Figure 2. In participants with CIS, we detected positive
or borderline-positive AP-2γ-stained nuclei in 50% of cases,
and in patients with an overt TGCT before treatment, this figure
was 33.9%. For six patients with TGCTs and for one patient
with CIS-only, samples were available before and after treatment (orchidectomy/chemotherapy/radiotherapy). These cases
were used for evaluation of whether positive elements disappeared from the semen with adequate treatment. Among these,
four patients had positive or borderline-positive elements in
cytospins before treatment, but after treatment all cytospins
were judged as negative. For some TGCT patients, we only
had a semen sample available after treatment for TGCT, and
among a few of these cases, AP-2γ-stained elements were
noted up to a few weeks after orchidectomy.
Characteristics of the putative diagnostic test
For determination of sensitivity, specificity and predictive
values of the putative detection assay, we analysed the positive
and borderline test results together in the subgroup of participants where bilateral testicular biopsies were available (n = 151).
The sensitivity of the AP-2γ assay as a marker of neoplastic
cells in semen analysis was 34.2% [95% confidence interval
Statistics
Statistical analysis of the data was performed with the SPSS version
14.0 software. A significance level below 0.05 was considered significant. Correlations of score with various parameters were performed
with non-parametric Kruskal–Wallis analyses. Confidence intervals for
the sensitivity and specificity were calculated using the Confidence
Interval Analysis (CIA) software, version 2.0 (Altman et al., 2000).
Results
Scoring results of AP-2g and other markers
Examples of elements in semen samples stained with AP-2γ,
OCT-3/4, PLAP and NANOG are presented in Figure 1. These
markers were chosen because they were established markers of
CIS, and AP-2γ was our focus because the antibody already in
initial experiments showed a consistent staining of the majority
of CIS cells in tissue samples (Hoei-Hansen et al., 2004). For
AP-2γ, examples of elements categorized as ‘positive’, ‘borderline’ or ‘negative’ are shown. OCT-3/4 is a nuclear marker
and the antibody is very specific; hence, stainings with OCT-3/4
were similar to the AP-2γ results. Stainings with the cytoplasmic
170
Figure 2. Summary and evaluation of the results of (A) AP-2γ and
(B) OCT-3/4 staining of semen samples from the various participant
categories. Depicted is the percentage of participants in each group
that had semen cytospins assessed according to an arbitrary score of
negative, borderline or positive. The evaluation was performed systematically and independently by two investigators, who had no prior
knowledge of the diagnosis.
Non-invasive detection of testicular neoplasia
(95% CI) = 24–45%] when assessing all patients with a positive/borderline score and 54.5% (95% CI = 28–79%) when
assessing only testicular GCT patients with contralateral CIS,
infertile and CIS-only patients. The specificity of the test was
93.6% (95% CI = 86–97%). The positive predictive value
(PPV) was 83.3% (95% CI = 66–93%), and the negative predictive value (NPV) was 60.3% (95% CI = 51–69%).
There was no correlation between presence of AP-2γ- or
OCT-3/4-stained elements and semen concentration or abstinence time (P = 0.9, P = 0.64, respectively). The AP-2γ score
was positively correlated with the number of analyses performed (P < 0.001); the OCT-3/4 score showed similar tendency, but without reaching statistical significance. We also
found an increased discrepancy between the observers with an
increased number of analyses (P < 0.001).
In 26 cases, infertile patients with negative AP-2γ staining
had bilateral biopsies performed as a part of the andrological
evaluation for reasons unrelated to the present project (mainly
due to very severe oligozoospermia with or without ultrasonic
microlithiasis). In one case, the biopsies revealed bilateral CIS
(participant ID SC-371). Despite performing a total of 10 separate
cytospin analyses in this patient, we only detected ‘borderline’
AP-2γ-stained elements and no OCT-3/4-stained elements. In
the remaining infertile participants, bilateral biopsies did not
detect CIS, but varying abnormalities, including Sertoli cellonly, hypospermatogenesis and signs of atrophy. For the eight
patients with other testicular tumours/lesions (see Table I) and
a contralateral biopsy without CIS, no AP-2γ-positive cells
were detected in semen samples.
We performed a sub-analysis of immunocytological score
depending on the histology of the overt tumour in patients with
TGCTs. Scoring was positive or borderline positive in 13/23
(56.6%) of non-seminomas but only 6/35 (17.4%) of seminomas,
which was a significant difference (P = 0.015). A similar tendency
was seen for OCT-3/4.
Follow up of participants with a suspicion of CIS due
to AP-2g-positive cells in semen
In the infertile and control groups, positive scoring results were
unexpectedly detected in a few participants (<5%), and their
follow up is described in Table II. In most cases, a second
semen sample was requested from the participant. Because this
putative test has not yet been approved as a diagnostic procedure, we have not yet implemented a consensus procedure. We
chose to individualize the follow up for each patient after a discussion between the researchers (C.H.H. and E.R.M.) and
attending andrologists (E.C., N.J. and N.E.S.), who informed
the patients and obtained consent. As the only definite confirmation of the presence of testicular CIS, bilateral biopsies were
performed in four patients with positive cells in semen, also
because of the presence of other risk factors for CIS. In only
one participant (SC-008), the biopsy revealed CIS in one testis,
and so, he had an orchidectomy and required no further treatment [(this case story was described in Hoei-Hansen et al.
(2005b)]. In three cases, biopsies failed to detect CIS, despite
AP-2γ-positive or borderline elements present in semen. In
four cases, biopsies were not performed, either because of control
semen analysis showing no AP-2γ-positive cells or because of
participant’s refusal. In one case, the patient had had bilateral
biopsies showing no CIS performed a few years before the AP2γ tests, and we decided not to repeat biopsies. Finally, one
participant is still awaiting biopsies to be performed.
Discussion
In the present study, we analysed semen samples from a total
of 503 participants and demonstrate that the immunocytological AP-2γ assay is able to detect CIS cells in semen. Particularly informative was a group of eight participants who had
CIS without an overt tumour being present. Previous studies
have only sporadically included patients at the stage of CIS;
the majority were patients with a GCT and CIS in the vicinity.
Tumour cells are usually not exfoliated into semen, because
the tumour has no connection to the seminal ducts, unless it is
present in the intratubular form. In one previous study, CIS
cells were detected in 4/8 patients with CIS before tumour
development (Giwercman et al., 1988c), and in a study applying immunohistochemistry to detect M2A-positive cells, 4/4
CIS patients (three in the vicinity of a TGCT) were positive,
but 2/10 controls showed non-specific staining (Giwercman
et al., 1988b). Our present study indicates that CIS/tumour
cells perhaps are not always detectable in semen, at least not by
a relatively insensitive immunocytological method. There are
several possible reasons for this. Firstly, there may be very few
CIS cells in semen, especially in testes with an overt tumour,
where compression of tubules by the tumour may hinder the
release of CIS cells into the ejaculate, which could account for
the reduced sensitivity of our test in patients with an overt
TGCT. Secondly, there may exist inherent properties of the
CIS cells determining whether they detach from the basement
tubular membrane and may be shed into the semen. Our observation of the higher proportion of AP-2γ-positive cells in
semen samples of patients with non-seminomas supports these
explanations. Firstly, seminomas do not metastasize as fast as
non-seminomas and tend to grow larger within the testis, exerting mechanical pressure and sometimes destroying the entire
remaining parenchyma. Secondly, testicular parenchyma
adjacent to seminomas contains CIS less frequently than that
adjacent to non-seminomas (85 versus 97%, respectively), and
the number of CIS tubules is also significantly smaller (26 versus 32%), probably because of the immunological host
response, which in some cases may eradicate CIS (Oosterhuis
et al., 2003). On the contrary, intratubular tumour growth is
found more frequently in seminomas. Thirdly, animal studies
have suggested that seminiferous tubule fluid may be
decreased in cryptorchid testes, testes containing Sertoli cellonly tubules, or in cases with reduced androgen action (Setchell,
1970). As this fluid is the vehicle via which exfoliated CIS
cells would be transported into seminal plasma, it is possible
that men with atrophic testis and/or history of cryptorchidism
might thus shed fewer CIS cells. We do not have data in the
present study regarding testicular size, but it is known that testes harbouring CIS have a reduced volume, and the presence of
Sertoli cell-only tubules and partially undifferentiated Sertoli
cells is also quite common, and therefore, a reduced flow of
171
C.E.Hoei-Hansen et al.
Table II. Summary of the follow up in the participants without overt testicular tumours who had AP-2γ- or OCT-3/4-positive stained elements in cytospin
Participant ID>
Diagnosis category
Number of analyses
positive/total
Follow up
SC-008
SC-228
AP-2γ: 1/3
AP-2γ: 1/10 positive
and 2/10 borderline
AP-2γ: 2/6
Biopsies with unilateral CIS
Awaiting participant’s decision, will perhaps have biopsies performed
SC-231
Infertile
Control from epilepsia
research project
Infertile
SC-237
SC-238
Infertile
Infertile
SC-341
SC-353
Infertile
Infertile
AP-2γ: 1/2
AP-2γ: 1/4 positive and
1/4 borderline
AP-2γ: 1/3
AP-2γ: 1/4
SC-365
Cryopreservation before
inguinal hernia operation
Extragonadal GCT, no CIS
bilaterally (2002)
Infertile
SC-506
SC-605
AP-2γ: 2/9OCT-3/4: 0/3
Biopsies without CIS (right testis: normal; left testis: spermatogenic
arrest and degenerated cells)
Participant does not want testicular biopsies
Biopsies without CIS (right and left testis: reduced spermatogenesis
and lymphocytic infiltration)
No biopsies, control semen analysis negative
Biopsies without CIS (right testis: reduced spermatogenesis; left testis:
spermatogenic arrest)
No biopsies, control semen analysis negative
AP-2γ: 1/2OCT-3/4: 0/1
No follow up
AP-2γ: 0/6OCT-3/4: 1/2
Control semen analysis negative, biopsies are planned for other
reasons
CIS, carcinoma in situ; GCT, germ cell tumour.
seminiferous tubule fluid is likely (Hoei-Hansen et al.,
2003; Holm et al., 2003). Another reason for a low sensitivity of the AP-2γ-based test for detection of CIS cells in semen
may be the small volume of the semen sample used for the
test (50–1000 μl of total normal volume 3–5 ml, depending on
sperm concentration and number of samples). If possible,
either repeated analysis or larger volume should be analysed to
improve the assay. In the present analysis, we applied several
CIS markers, of which AP-2γ was the most suitable. Staining
for NANOG and PLAP revealed substantial non-specific reaction, whereas OCT-3/4 was nearly as specific as AP-2γ and
detected positive cells in the same range as AP-2γ for patients
with an overt GCT but did not show stained cells in patients
with CIS-only. For both AP-2γ and OCT-3/4, we were in most
cases only able to detect stained cells in a subset of analysed
cytospin samples, as seen in Table II.
As the lifetime risk of TGCT (and assumed prevalence of
CIS) is about 1% in the male population in Denmark, and probably even higher among subfertile men, we anticipated at least
one case of testicular CIS per 100 normal participants or subfertile participants analysed. In our study, 292 participants
were included, because they were in an andrological evaluation
for subfertility, and among these, two participants were identified with CIS. We do not have biopsies on all participants and
therefore cannot rule out that other patients may harbour CIS.
In addition, we cannot rule out that the participants we have
stated as false positive (because of positive elements in semen,
but no CIS in testicular biopsies) may later develop a TGCT, as
false-negative testicular biopsies have been reported (Dieckmann
and Loy, 2003). The outcome of our assay was, however, completely opposite in the two identified CIS cases. The first participant (SC-008) was a very encouraging example of early
diagnosis, thanks to our AP-2γ assay (Hoei-Hansen et al.,
2005b). The other infertile participant with testicular CIS
(SC-371) was an example of low sensitivity of our assay. No
AP-2γ-positive cells were detected in his semen sample when
he was first examined, but the patient had bilateral biopsies
172
performed due to other clinical findings raising the CIS suspicion (unilateral cryptorchidism, hypospadias, atrophic testes,
severe oligozoospermia and ultrasonic microlithiasis). At histological evaluation, CIS was present in large areas of biopsies
from both testes. Subsequently, we repeated the AP-2γanalysis
several times but only detected borderline-positive elements
without being able to identify unequivocal CIS cells in semen.
As these two examples and the detailed evaluation of the
assay presented here demonstrate, the present analysis does not
yet meet criteria for a valid and sensitive diagnostic analysis
that can be recommended for screening in routine clinical practice. However, the method does have some clinical value as it
detects about half of patients harbouring CIS or incipient
tumour in a non-invasive and simple way. Thus, this method
can be used as an auxiliary diagnostic procedure in specialized
andrology centres, where there is a possibility of repeated analysis and careful follow up of individual patients. The most
important point for the attending clinician and the patient is to
be aware that a negative result does not exclude the presence of
germ cell neoplasia in the testis. There is some room for
improvement in the immunocytological assay, and we are currently working on methods of enrichment of the fetal markerpositive cells in semen samples. However, in a subset of
patients, especially where there are very few CIS cells or they
are not exfoliated into ejaculate, another approach with much
higher sensitivity has to be developed, for example, based on
biochemical detection of a low concentration of CIS-specific
markers.
Acknowledgements
The authors thank all the participants for their cooperation and H. Kistrup,
K. Lambrethsen, L. Andersen, M. Simonsen, P. Timmerby, S.
Soultanova and M. Bekker for excellent technical assistance. The
Danish Cancer Society, Vissing Foundation, Carla Thiel Kraghs
Foundation, Kirsten & Freddy Johansen’s Foundation, Copenhagen
University Hospital and the Danish Medical Research Council supported this study.
Non-invasive detection of testicular neoplasia
Conflict of interest statement
None of the authors have conflicts of interest, but a patent covering the use of newly identified markers for CIS, including
AP-2γ, for diagnosis of germ cell neoplasia is pending.
References
Almstrup K, Hoei-Hansen CE, Wirkner U, Blake J, Schwager C, Ansorge W,
Nielsen JE, Skakkebaek NE, Rajpert-De Meyts E and Leffers H (2004)
Embryonic stem cell-like features of testicular carcinoma in situ revealed by
genome-wide gene expression profiling. Cancer Res 64,4736–4743.
Altman DG, Machin D, Bryant TN and Gardner MJ (2000)Statistics with Confidence. Confidence Intervals and Statistical Guidelines, 2nd edn. BMJ
Books, Bristol, UK, pp. 1–240.
Brackenbury ET, Hargreave TB, Howard GC and McIntyre MA (1993)
Seminal fluid analysis and fine-needle aspiration cytology in the diagnosis
of carcinoma in situ of the testis. Eur Urol 23,123–128.
Brydoy M, Fossa SD, Klepp O, Bremnes RM, Wist EA, Wentzel-Larsen T and
Dahl O (2005) Paternity following treatment for testicular cancer. J Natl
Cancer Inst 97,1580–1588.
Czaplicki M, Rojewska J, Pykalo R and Szymanska K (1987) Detection of
testicular neoplasms by cytological examination of seminal fluid. J Urol
138,787–788.
Dieckmann KP and Loy V (2003) False-negative biopsies for the diagnosis of
testicular intraepithelial neoplasia (TIN)—an update. Eur Urol 43,516–521.
Giwercman A, Marks A and Skakkebaek NE (1988a) Carcinoma-in-situ
germ-cells exfoliated from seminiferous epithelium into seminal fluid.
Lancet 1,530.
Giwercman A, Marks A, Bailey D, Baumal R and Skakkebaek NE (1988b) A
monoclonal antibody as a marker for carcinoma in situ germ cells of the
human adult testis. APMIS 96,667–670.
Giwercman A, Clausen OP and Skakkebaek NE (1988c) Carcinoma in situ of
the testis: aneuploid cells in semen. Br Med J 296,1762–1764.
Hoei-Hansen CE, Holm M, Rajpert-De Meyts E and Skakkebaek NE (2003)
Histological evidence of testicular dysgenesis in contralateral biopsies from
218 patients with testicular germ cell cancer. J Pathol 200,370–374.
Hoei-Hansen CE, Nielsen JE, Almstrup K, Sonne SB, Graem N, Skakkebaek
NE, Leffers H and Rajpert-De Meyts E (2004) Transcription factor AP2gamma is a developmentally regulated marker of testicular carcinoma in
situ and germ cell tumors. Clin Cancer Res 10,8521–8530.
Hoei-Hansen CE, Rajpert-De Meyts E, Daugaard G and Skakkebaek NE
(2005a) Carcinoma in situ testis, the progenitor of testicular germ cell
tumours: a clinical review. Ann Oncol 16,863–868.
Hoei-Hansen CE, Rajpert-De Meyts E, Carlsen E, Almstrup K, Leffers H and
Skakkebaek NE (2005b) A subfertile patient diagnosed with testicular carcinoma in situ by immunocytological staining for AP-2gamma in semen samples: case report. Hum Reprod 20,579–582.
Hoei-Hansen CE, Almstrup K, Nielsen JE, Brask SS, Graem N, Skakkebaek
NE, Leffers H and Rajpert-De Meyts E (2005c) Stem cell pluripotency factor NANOG is expressed in human fetal gonocytes, testicular carcinoma in
situ and germ cell tumours. Histopathology 47,48–56.
Holm M, Hoei-Hansen CE, Rajpert-De Meyts E and Skakkebaek NE (2003)
Increased risk of carcinoma in situ in patients with testicular germ cell cancer with ultrasonic microlithiasis in the contralateral testicle. J Urol
170,1163–1167.
Hong SM, Frierson HF Jr and Moskaluk CA (2005) AP-2gamma protein
expression in intratubular germ cell neoplasia of testis. Am J Clin Pathol
124,873–877.
Howard GC, Hargreave TB and McIntyre MA (1989) Carcinoma in-situ of the
testis diagnosed on semen cytology. Clin Radiol 40,323–324.
Huddart RA, Norman A, Moynihan C, Horwich A, Parker C, Nicholls E and
Dearnaley DP (2005) Fertility, gonadal and sexual function in survivors of
testicular cancer. Br J Cancer 93,200–207.
Jacobsen GK, Henriksen OB and von der Maase H (1981) Carcinoma in situ of
testicular tissue adjacent to malignant germ-cell tumors: a study of 105
cases. Cancer 47,2660–2662.
Jorgensen N, Auger J, Giwercman A, Irvine DS, Jensen TK, Jouannet P,
Keiding N, Le Bon C, MacDonald E, Pekuri AM et al. (1997) Semen analysis performed by different laboratory teams: an intervariation study. Int J
Androl 20,201–208.
Looijenga LH, Stoop H, de Leeuw HP, de Gouveia Brazao CA, Gillis AJ, van
Roozendaal KE, van Zoelen EJ, Weber RF, Wolffenbuttel KP, van Dekken H
et al. (2003) POU5F1 (OCT3/4) identifies cells with pluripotent potential in
human germ cell tumors. Cancer Res 63,2244–2250.
Meng FJ, Zhou Y, Giwercman A, Skakkebaek NE, Geurts van Kessel AD and
Suijkerbuijk RF (1998) Fluorescence in situ hybridization analysis of chromosome 12 anomalies in semen cells from patients with carcinoma in situ of
the testis. J Pathol 186,235–239.
Oosterhuis JW, Kersemaekers AM, Jacobsen GK, Timmer A, Steyerberg EW,
Molier M, Van Weeren PC, Stoop H and Looijenga LH (2003) Morphology
of testicular parenchyma adjacent to germ cell tumours. An interim report.
APMIS 111,32–40.
Pauls K, Jager R, Weber S, Wardelmann E, Koch A, Buttner R and Schorle H
(2005) Transcription factor AP-2gamma, a novel marker of gonocytes and
seminomatous germ cell tumors. Int J Cancer 115,470–477.
Rajpert-De Meyts E, Hanstein R, Jorgensen N, Graem N, Vogt PH and
Skakkebaek NE (2004) Developmental expression of POU5F1 (OCT-3/4) in
normal and dysgenetic human gonads. Hum Reprod 19,1338–1344.
Salanova M, Gandini L, Lenzi A, Chiarenza C, Filippini A, Dondero F, Di
Silverio F and Stefanini M (1999) Is hyperdiploidy of immature ejaculated
germ cells predictive of testis malignancy? A comparative study in healthy
normozoospermic, infertile, and testis tumor suffering subjects. Lab Invest
79,1127–1135.
Schmoll HJ, Souchon R, Krege S, Albers P, Beyer J, Kollmannsberger C, Fossa
SD, Skakkebaek NE, De Wit R, Fizazi K et al. (2004) European consensus
on diagnosis and treatment of germ cell cancer: a report of the European
Germ Cell Cancer Consensus Group (EGCCCG). Ann Oncol 15,1377–1399.
Setchell BP (1970) The secretion of fluid by the testes of rats, rams and goats
with some observations on the effect of age, cryptorchidism and hypophysectomy. J Reprod Fertil 23,79–85.
Skakkebaek NE (1972) Possible carcinoma-in-situ of the testis. Lancet 2,516–517.
Skakkebaek NE (1975) Atypical germ cells in the adjacent “normal” tissue of
testicular tumours. Acta Pathol Microbiol Scand 83,127–130.
Yu DS, Wang J, Chang SY and Ma CP (1990) Differential diagnosis of testicular mass by cytological examination of seminal fluid collected by ejaculation or prostatic massage. Eur Urol 18,193–196.
World Health Organization (1992) WHO Laboratory Manual for the Examination of Human Semen and Sperm–Cervical Mucus Interaction, 3rd edn.
Cambridge University Press, Cambridge, UK.
Submitted on May 23, 2006; resubmitted on June 30, 2006; accepted on July 11,
2006
173