Risk of testicular cancer in men with abnormal semen characteristics: cohort studyBMJ 2000; 321 doi: http://dx.doi.org/10.1136/bmj.321.7264.789 (Published 30 September 2000) Cite this as: BMJ 2000;321:789
- Rune Jacobsen, researchera (, )
- Erik Bostofte, consultantb,
- Gerda Engholm, senior researchera,
- Johnni Hansen, researcherc,
- Jørgen H Olsen, head of departmentc,
- Niels E Skakkebæk, professord,
- Henrik Møller, professore
- a Centre for Research in Health and Social Statistics, Danish National Research Foundation, DK-2100 Copenhagen, Denmark
- b Sperm Analysis Laboratory, Health Service Physicians Organisation, DK-1112 Copenhagen, Denmark
- c Institute of Cancer Epidemiology, Danish Cancer Society, Box 839, DK-2100 Copenhagen, Denmark
- d Department of Growth and Reproduction, National University Hospital, DK-2100 Copenhagen, Denmark
- e Thames Cancer Registry, Guy's, King's, and St Thomas' School of Medicine, London SE1 3QD
- Correspondence to: Rune Jacobsen, Department of Epidemiology, Institute of Public Health, Faculty of Health Sciences, University of Copenhagen, Panum Institute, DK-2200 Copenhagen, Denmark
- Accepted 19 June 2000
Objective: To explore the associations between semen characteristics and subsequent risk of testicular cancer.
Design: Cohort study.
Participants: 32 442 men who had a semen analysis done at the Sperm Analysis Laboratory in Copenhagen during 1963-95.
Main outcome measure: Standardised incidence ratios of testicular cancer compared with total population of Danish men.
Results: Men in couples with fertility problems were more likely to develop testicular cancer than other men (89 cases, standardised incidence ratio 1.6; 95% confidence interval 1.3 to 1.9). The risk was relatively constant with increasing time between semen analysis and cancer diagnosis. Analysis according to specific semen characteristics showed that low semen concentration (standardised incidence ratio 2.3), poor motility of the spermatozoa (2.5), and high proportion of morphologically abnormal spermatozoa (3.0) were all associated with an increased risk of testicular cancer. The only other cancer group that showed increased incidence was “peritoneum and other digestive organs” (six cases; 3.7, 1.3 to 8.0). Of these, two cases were probably and two cases were possibly extragonadal germ cell tumours.
Conclusions: The results point towards the existence of common aetiological factors for low semen quality and testicular cancer. Low semen quality may also be associated with increased incidence of extragonadal germ cell tumours.
Editorial by de Kretser
Over recent decades a possible decrease in semen quality 1 2 and an increase in the incidence of testicular cancer have been reported in many populations.3-5 It is unclear whether these temporal trends are independent phenomena or somehow connected to each other.6-8 Case-control studies on subfertility and subsequent risk of testicular cancer have given conflicting results. 8 9 However, a recent Danish population based cohort study found an increased risk of testicular cancer in men with few children for their age.10 These findings supported the results of an earlier Danish case-control study.8 Both of these Danish studies used the number of children fathered at a given age as the measure of fertility. Thus some men with normal reproductive potential will inevitably have been classified as having low relative fertility because they had no or few children for reasons that were unrelated to their fertility. Subfertility can be measured more directly by analysis of semen for characteristics such as spermatocyte concentration, motility, and morphology. 11 12
Men with testicular cancer often have abnormal semen characteristics, 13 14 but the association between abnormal semen characteristics and testicular cancer has not been investigated prospectively. We studied the incidence of testicular cancer in relation to semen characteristics in 32 442 men who had semen analysis at the Sperm Analysis Laboratory in Copenhagen during 1963-95.
Participants and methods
We linked information on all men in couples with fertility problems who had a semen analysis done at the Sperm Analysis Laboratory in Copenhagen during 1963-95 (n=32 442) with data in the Danish Cancer Registry, which holds information on all cases of cancer in the Danish population from 1943 to 1995.15 Men who visited the laboratory for other reasons (such as semen analysis after vasectomy) were excluded from the analysis. The Copenhagen laboratory is one of several public semen analysis laboratories in Denmark and examines semen samples mostly from men in the area of Copenhagen. Men are referred to the clinic by general practitioners and urologists, and the investigations are paid for through the public health system. Men with cancer before the date of semen analysis were excluded. For men who had multiple semen tests only their first test was used in the analysis. Similarly, only the first cancer diagnosis in a given man was included in the analysis. The methods used for analysis of semen (sperm concentration and motility and proportion of morphologically abnormal spermatozoa) have been described previously.16 For each man we also obtained information on date of birth, dates of birth of his children, and date of death from the Central Population Register and the National Death Register.
We calculated the expected numbers of cancer cases in the cohort (by multiplying years at risk with primary cancer rates in the Danish population) and standardised incidence ratios and 95% confidence intervals using a Fortran computer program.17 The standardised incidence ratios were calculated for each type of cancer by time since first semen analysis, stratifying by semen characteristics according to standard definitions of subfertility.12 The group of azoospermic men was divided into those with and without children in order to address the possibility that some azoospermic men had not given information on sterilisation or other circumstances resulting in a sudden azoospermia. To examine the separate and joint effects of the three semen characteristics, the cohort was stratified into groups according to their combination of semen measures.
Overall, the cohort members had an increased risk of testicular cancer and of cancers of the peritoneum and other digestive organs (table 1). Risk of other types of cancer was not increased in the cohort. Eighty nine men developed testicular cancer, giving a standardised incidence ratio of 1.6 (95% confidence interval 1.3 to 1.9). Of these 89 men, 50 had seminomas (standardised incidence ratio 1.5, 1.1 to 1.9), 37 had non-seminomas (1.8, 1.2 to 2.4), and two were unspecified. For cancer of the peritoneum and other digestive organs the standardised incidence ratio was 3.7 (1.3 to 8.0) based on six observed cases. The standardised incidence ratio for cancers of all other sites combined was 1.0 (0.9 to 1.1).
Table 2 shows the standardised incidence ratios for testicular cancer stratified by time between first semen analysis and cancer diagnosis. The highest risk of testicular cancer was in the first two years after the first semen analysis (standardised incidence ratio 1.8). The risk was 1.5-1.6 for two to 11 years after the first semen analysis and 1.3 for more than 11 years since first semen analysis. The trend in the standardised incidence ratios over the four periods of follow up was not significant (P=0.46).
Table 3 shows the standardised incidence ratios of testicular cancer, stratified by measures of semen quality. In univariate analyses, low semen concentration, poor semen mobility, and a high proportion of abnormal spermatozoa were all associated with increased standardised incidence ratios, whereas the groups with normal semen characteristics had standardised incidence ratios closer to unity. The azoospermic men who had fathered children before semen analysis showed lower risk of testicular cancer than azoospermic men without children (standardised incidence ratio 2.0 v 5.3). Men who were not azoospermic but who had sperm concentrations of 20 million/ml or lower had a higher risk of testicular cancer than men with concentrations above 20 million/ml (standardised incidence ratio 2.3 v 1.1).
The univariate, separate, and joint effects of the three semen quality measures were analysed in the subgroup of 29 177 men who had some spermatozoa in the semen sample (table 4). The separate effect of low concentration on the risk of testicular cancer was roughly the same as the univariate effect (standardised incidence ratio 2.1 and 2.3, respectively). Of 10 509 men with low semen concentration, 9187 had low concentration as the only abnormal characteristic. Very few men had poor motility only or a high proportion of abnormal spermatozoa only, and no case of testicular cancer was observed in these groups. We therefore could not identify a separate effect of poor motility or of having a high proportion of abnormal spermatozoa. However, the risk of testicular cancer increased with increasing number of subfertility measures present. The standardised incidence ratio was 1.9 for one subfertility measure, 2.7 for two measures, and 9.3 for all three subfertility measures.
Table 5 gives the details of the six cases of cancer in the peritoneum and other digestive organs. Case 1 may have had a testicular cancer before his leukaemia, which probably was treatment induced. An extragonadal germ cell tumour is also possible for case 2, who had increased concentrations of tumour markers. The notifications suggest that cases 3 and 5 had extragonadal germ cell tumours. Cases 4 and 6 seemed unlikely to have had extragonadal germ cell cancers.
Our retrospective cohort study, based on more than 30 000 men in infertile couples, found a strong association between subfertility and subsequent risk of testicular cancer. All men of couples with fertility problems were 1.6 times more likely to develop testicular cancer than the Danish male population in general, and the increase was evident for both seminoma and non-seminoma. The overall analysis included some fully fertile men from couples in which only the woman was subfertile, and the observed higher risk of testicular cancer in the cohort overall would be even higher if only subfertile men were included. Men in the cohort with abnormal semen characteristics had a twofold to threefold increased risk. Our findings are consistent with the results of investigations into spermatogenesis in patients with unilateral testicular cancer18 and risk of testicular cancer in men considered subfertile on the basis of a low number of children for their age. 8 10
The observation that men with unilateral testicular cancer have impaired spermatogenesis18 does not preclude the possibility that impaired reproductive capacity is secondary to the cancer. We found that the risk of testicular cancer was relatively constant with increasing time since semen analysis. Impaired spermatogenesis may therefore have been present many years before testicular cancer was diagnosed, pointing towards a permanent state of impaired spermatogenesis.
Our use of semen characteristics to assess subfertility eliminates the misclassification problems in studies based on numbers of children, where men with normal reproductive potential who have no or few children for other reasons may bias the result towards unity. All together, the available data point towards the existence of common risk factors for impaired spermatogenesis and testicular cancer.
What is already known on this topic
The incidence of testicular cancer has increased in the past 50 years, and some evidence suggests that sperm quality has decreased in the same period
Common aetiological factors may exist for testicular cancer and male subfertility
What this study adds
This study confirms that incidence of testicular cancer is increased in men with few children for their age
The association between testicular cancer and abnormal semen characteristics is statistically robust and consistent with the hypothesis of a common aetiology
Abnormal semen characteristics may be associated with extragonadal germ cell tumours
Some evidence suggests that testicular cancer has its origin in fetal life. Incidence of testicular cancer is lower among men born during the second world war than men born before and after the war in Denmark, Norway, and Sweden.19-21 Other risk factors for testicular cancer, such as low birth weight22 and congenital malformations of the testes, 23 24 also support a fetal origin for testicular cancer. In addition, carcinoma in situ (the precursor of both seminomas and non-seminomas) has several characteristics in common with fetal germ cells.25 The specific aetiological factors in testicular cancer are unknown, but maternal oestrogens and hormonal disrupting agents have been proposed as causal factors acting on the male fetus. 26 27
We also found an increased risk of cancer of the peritoneum and other digestive organs. One explanation for this association is that some of the observed cancers in this category were misclassified testicular or extragonadal germ cell tumours. Extragonadal germ cell tumours have been associated with testicular carcinoma in situ, 28 29 suggesting a common aetiology with testicular cancer.
From a public health perspective, our study provides some reassurance to men identified with abnormal semen characteristics, despite the increased relative risks. The absolute excess of cancers is about 36 cases per 32 442 men followed for 297 750 person years. The absolute increase in risk for the individual is therefore very small.
Contributors: RJ was responsible for the study design, data collection, statistical analysis, interpretation, and reporting and is the guarantor. EB, GE, JH, JHO, NES, and HM contributed to the study design, data collection, interpretation, and reporting.
Funding Danish Research Councils.
Competing interests None declared.
This article is part of the BMJ's randomised controlled trial of open peer review. Documentation relating to the editorial decision making process is available on the BMJ's website