An expansion or decrease in number of trinucleotide repeats in a coding gene sequence has been implicated in a growing number of human diseases.1 Some of the triplet expansion disorders seem to be a form of microsatellite instability, which has been associated with many types of human cancer,2 including testicular germ cell tumors.3, 4, 5 King et al.6 investigated CAG/CTG repeats and found expanded stretches in cell lines derived from testicular tumors and in germ line DNA isolated from a few members of testicular cancer kindreds. This finding, however, was not confirmed by 2 subsequent studies, which investigated trinucleotide repeat expansions in families with testicular cancer,7 and sporadic testicular germ cell tumors.8
One of the genes investigated in the latter study was the androgen receptor (AR). Decreased signaling via the AR can cause various disorders of the male reproductive tract, ranging from the complete sex reversal to impaired spermatogenesis.9, 10, 11 The risk of germ cell neoplasia is markedly increased in individuals with mutations in the AR gene and severe forms of androgen insensitivity.12 The vast majority of patients with testicular cancer, however, have no visible symptoms of undermasculinization and no mutations in the AR have been reported. These patients develop invasive germ cell tumors in young adulthood from a precursor lesion, carcinoma in situ (CIS, or intratubular germ cell neoplasia, unclassified).13, 14 According to our hypothesis, neoplastic transformation of germ cells into CIS occurs most probably during early development, but invasive progression into overt seminoma or nonseminomatous tumors takes place after post-pubertal onset of spermatogenesis.13, 15 Epidemiological evidence suggests that poor testicular development (testicular dysgenesis syndrome), which may be caused by an imbalance in the activity of sex hormones with a relative insufficiency of androgen action, leads to various reproductive abnormalities ranging from infertility and testicular maldescent to testicular cancer.15, 16, 17 The phenotypic severity of the clinical outcome may be dependent upon genetic susceptibility.
We hypothesized that a subtle change of the AR function caused by a polymorphic change in the gene structure might contribute to testicular dysgenesis and render some individuals more susceptible to germ cell transformation. Exon 1 of the AR gene contains a CAG repeat sequence (encoding a polyglutamine stretch) ranging in length from 8 to 37 repeats and averaging 20–22 depending on ethnic origin.18, 19 An expansion of the AR's CAG length above 38 repeats in the AR gene leads to recessive spinobulbar muscular atrophy (Kennedy's disease), which in addition to neurodegenerative symptoms, is characterized by a gradual deterioration of testicular function and spermatogenesis.20, 21 It has been demonstrated that the (CAG)n size correlates inversely with transcriptional activity of the AR,22, 23, 24 and the AR gene expression.25 Longer than average CAG repeats of the AR have been implicated as a risk factor for disorders associated with decreased action of androgens, including genital malformations26 and decreased spermatogenesis and male infertility.23, 27, 28 The issue remains debatable, however, because of conflicting results of several other studies.29, 30, 31, 32
Population-based studies of the (CAG)n polymorphism in germ-line DNA of patients with testicular germ cell tumors have not been carried out before. We report the results of our investigation of the AR's CAG repeat length in germ-line DNA in a large series of patients with unilateral or bilateral testicular germ cell neoplasia, in comparison to a matched control population of healthy men with normal reproductive parameters and proven fertility.
MATERIAL AND METHODS
Our study included 102 adult Danish patients diagnosed with testicular germ cell neoplasia. The patients were referred to our clinic for andrological assessment in the course of management of the primary disease or reproductive sequelae. None of the patients in our study presented with visible undervirilization or any other physical signs of androgen insensitivity. The clinical profile of the patients and tumor types is shown in Table II. About half of the patients had classical seminoma (48%), another half nonseminoma or mixed tumors (45%), which is a common distribution. Six patients (5.9%) were diagnosed at the preinvasive CIS stage. Most of the patients with overt testicular tumors presented with stage I (confined to the testis); disseminated disease was found only in 16.7%. Bilateral testicular neoplasia was present in 32 patients (31.4%), mainly as preinvasive CIS in the contralateral testicle. This unusually high proportion of bilateral cases is explained by the fact that our clinic serves as a regional centre for histological assessment of contralateral biopsies and for sperm banking.
Men included in the control group were recruited among partners of pregnant women in the obstetric clinic for a study of reproductive parameters of Danish men in comparison to several European populations.33 These controls were previously used for the study of the CAG repeats of the AR in the infertile men.31 After excluding a few fertile men with moderate oligozoospermia, 110 men were included in our study. In addition to being fertile and normospermic, all control subjects had normal serum concentrations of reproductive hormones (testosterone, LF, FSH, inhibin B and oestradiol). Semen analysis and hormone assays were carried out in our certified clinical laboratories according to standardized protocols.
All samples and data were coded and stored in an anonymous form. The study was approved by the local ethical committee and a written consent was obtained from all control subjects.
Analysis of the Number of CAG Repeats in the AR Gene
Genomic DNA was isolated from peripheral blood samples using reagents from Roche Molecular Biochemicals, as described by the manufacturer. The CAG-repeat-containing part of exon-1 of the AR gene was amplified by PCR using 2 sets of primers (nested PCR), in a single reaction. The primers were: P1 (80) ATCACAGCCTGTTGAACTCTTCT, P2 (710) CTCAGGATGTCTTTAAGGTCAGCGGA, P3 (93) GAACTCTTCTGAGCAAGA, P4 (698) TTAAGGTCAGCGGAGCA (numbers in brackets correspond to nucleotide numbers in the exon-1 sequence, GenBank accession number M35844). The composition of the PCR was: 10 mM Tris-HCl pH 8.3, 50 mM KCl, 1.8 mM MgCl2, 0.1% Triton X-100, 0.005% Gelatine, 250 μM dNTP, 0.0033 μM of P-1 and P-2, 0.033 μM of P-3 and P-4, 300 ng genomic DNA and 3 U AmpliTaq Polymerase (Applied Biosystems, Foster City, CA). The PCR conditions were: 96°C 5 min; 10 cycles of 96°C 30 sec, 68°C 40 sec, 72°C 105 sec; 30 cycles of 96°C 30 sec, 54°C 40 sec, 72°C 105 sec; 72°C 5 min. All PCR and cycle sequencing were carried out on Perkin-Elmer 9600 PCR machines. The DNA fragments were resolved on 2% agarose gels, extracted and directly sequenced on an ALFexpress sequencer (Amersham Pharmacia Biotech, Piscataway, NJ) using 2 CY5 labeled primers: Seq-1 (124) CY5-GGTAAGGGAAGTAGGTGGAAG and Seq-2 (448) CY5-CTTGGGGAGAACCATCCTCA), as described previously.34 All sequences were determined on both strands and the numbers of CAG repeats were counted both from the chromatographic curves and from the sequences.
The distribution of the CAG repeat numbers of the AR gene in patients and controls was compared using Levene's test and Wilcoxon's test, as well as a 2-sample t-test. Spearman's correlation was used for assessing associations. A difference was considered statistically significant if a p-value was <0.05.
Distribution of CAG Repeats in the AR Gene in Patients with Testicular Germ Cell Cancer
Figure 1 and Table I illustrate the distribution of the (CAG)n of the AR in all studied subjects in our study. Analysis of CAG repeats in normal controls with proven fertility showed a substantial polymorphism, including 19 different lengths ranging from 14 to 33. The most frequent number of repeats was 20 (17 subjects or 15.5%), the mean ± SD was 21.79 ± 3.36 and the median 21. The number of subjects with relatively long (CAG)n ≥ 26, which in 1 study was considered as a cut-off value for the significant association with decreased fertility,35 was found in 12 control subjects (10.9%). The longest CAG stretches (31–33) were found in a few men in the control group. These men had normal sperm counts (mean 65.2 ± 19.9 × 106/ml), high serum inhibin B levels (mean 260 ± 71 ng/L), low serum FSH levels (mean 1.83 ± 0.43 U/L) and normal serum testosterone concentrations (mean 21.4 ± 7.5 nmol/L) suggesting that their testicular function was normal.
Table I. Comparison of the Distribution of (CAG)n of the Androgen Receptor Between the Two Studied Groups
Mean ± SD
21.79 ± 3.36
21.69 ± 3.11
Most frequent (CAG)n
n > 25 (%)
The distribution of CAG repeat lengths of the AR among the patients with testicular cancer was very similar to that in the control group; the range was from 13–31. The most frequent number of repeats was 21 (23 cases or 22.5%), the mean value was 21.69 ± 3.11 and the median 21. The number of subjects with (CAG)n ≥ 26 in the testicular cancer group was 9 (8.8%), which was not different from the control group. Analysis of the distribution of (CAG)n among patients and controls, according to Wilcoxon's test, demonstrated that there was no difference between the 2 groups (p = 0.85).
Analysis of A Possible Association Between the Number of CAG Repeats in the AR Gene and Tumor Type and Severity of the Disease
The patients with testicular cancer were divided into groups according to the unilateral or bilateral disease, tumor type (seminoma versus nonseminoma) and Stage I disease or disseminated disease. As illustrated in Table II, no correlation between the number of CAG repeats in the AR gene and these variables was detected. The mean (CAG)n values in patients with mixed tumors and disseminated disease were 22.55 ± 2.68 and 22.32 ± 3.03, respectively, which seemed to be greater in comparison to patients with CIS only (21.17 ± 1.95) or Stage I (21.42 ± 2.94). These differences were, however, not significant. Because of the small numbers of cases, some subgroups were combined together. Mixed tumors, which comprise elements of seminoma and nonseminoma, were combined with pure nonseminomas due to similar clinical course and prognosis, whereas bilateral neoplasia was combined with disseminated disease (both are indicative of relatively severe disease). The distribution of (CAG)n did not differ between these groups and controls. Importantly, the proportions of cases with the longest CAG stretches (≥26) were not different in any of the subgroups.
Table II. The CAG Repeat Numbers in the Androgen Receptor Gene in Correlation to Various Histological Forms of Germ Cell Neoplasia, and the Extent and Clinical Aggressiveness of the Disease
Mean ± SD
n > 25 (%)
Histological type of neoplasia
21.17 ± 1.95
21.65 ± 3.38
21.63 ± 3.05
Mixed tumours (S + NS)
22.55 ± 2.68
Nonseminoma (NS) and Mixed tumours combined (S + NS)
21.85 ± 2.99
Stage I only
21.42 ± 2.94
22.32 ± 3.03
22.28 ± 2.60
Bilateral or disseminated disease
22.30 ± 2.77
21.69 ± 3.11
Testicular germ cell derived cancer of young adults and other components of testicular dysgenesis syndrome (TDS) have been increasing in recent decades. The striking geographic and ethnic differences in the prevalence of TDS can be explained by a variable degree of environmental or lifestyle impact or by genetic susceptibility, which is expected to be an important factor in the most severe form of TDS, namely testicular cancer.17 As the TDS is associated with low androgen function, a change of the AR transactivation in the individuals with longer CAG stretches could theoretically be a predisposing factor. A hypothesis behind our study was that men with the CAG repeat stretches in the AR gene close to the upper normal limit might have slightly decreased androgen sensitivity leading to testicular dysgenesis and enhanced susceptibility to germ cell transformation, especially in response to exogenous factors.
In our study, we have found no difference in the distribution of the CAG repeat of the AR in a large group of patients with testicular germ cell neoplasia in comparison to a group of fertile men with normal reproductive parameters. One could argue that some of the men in our control group, whose mean age was 31.5, could still develop testicular cancer of the young adult type, because the peak age of this disease in Denmark is approximately 30–32 years.36 It is, however, very unlikely because our controls were normospermic recent fathers, whereas men who develop testicular germ cell cancer have impaired gonadal function, spermatogenesis and fertility before development of their tumor.37, 38 Direct evidence for this was provided by recent Danish studies that documented abnormal semen characteristics and reduced fertility in men who later developed testicular cancer.39, 40
A possible correlation between the AR's CAG polymorphism and infertility was previously investigated by several groups. As mentioned in the introduction, the results of those studies were mixed and somewhat confusing. Differences in the AR's CAG repeat lengths judged to be significant in the previous studies were very small, probably because of small number of cases or the lack of information on fertility of the controls.31 The lack of association of the AR's CAG repeat polymorphism in patients with testicular cancer speaks also against an important role of this polymorphism for male fertility, although the possible association with sperm production in normal healthy men cannot be excluded.31, 41
The CAG length polymorphism of the AR (relatively shorter stretches) was linked to aggressive forms of prostatic cancer in some studies,42 therefore, we have compared the subgroups of our patients divided according to the tumor type, the extent and severity of the disease. No correlation of the CAG stretch with a clinical course of the disease was found and we did not identify any subgroup with either longer or shorter CAG stretches. The numbers of subjects in these subgroups were small, however, and the results could be a multiple testing problem, i.e., a chance finding. Therefore, our conclusion concerning the lack of association of disease aggressiveness with the CAG repeat length in the AR gene should be treated with caution. We would like to underline that caution should be exercised in general when studies of polymorphic variables are carried out. Given the broad inter-individual variation in the normal population, the numbers of subjects have to be sufficiently large, otherwise small differences observed do not have any biological meaning.
In conclusion, our study demonstrated that testicular germ cell cancer of the young adult type is not associated with a germ-line expansion of the polymorphic CAG repeat stretch in the androgen gene. Although decreased androgen function during development is undoubtedly a risk factor for germ cell neoplasia and other components of TDS, this impairment is most probably not caused by the CAG polymorphism. The mechanisms remain to be identified.
We are grateful to Drs. A.-G. Andersen, E. Carlsen and N. Jørgensen for help with collecting control data and to other members of the staff for semen and hormone analyses.