Gestational Hypertension, Preeclampsia, and Risk of Testicular Cancer

Testicular cancer incidence is increasing for unknown reasons (1). Pathologic studies of testicular tumor cells suggest that germ cell testicular cancer may already be developing during fetal life if precursor cells of the spermatogonia escape normal development and retain some stem cell characteristics (2). Two mechanisms have been suggested to trigger such maldifferentiation: intrauterine growth retardation and fetal exposure to altered levels of sex hormones, particularly to increased levels of hormones with estrogenic effects (3, 4).

The placenta is instrumental for fetal growth as well as for pregnancy hormone production. It is the primary producer of pregnancy hormones such as estrogens and human chorionic gonadotropin (hCG). Signs of placental malfunction, such as preeclampsia, are reflected in alterations in these pregnancy hormones and in impaired fetal growth. Preeclampsia, which is clinically defined as hypertension and proteinuria after gestational week 20, is one of the main reasons for antenatal monitoring, and has been associated with low levels of pregnancy estrogens (5) as well as high levels of hCG (6).

In this study, we have evaluated gestational hypertension and preeclampsia in relation to risk of testicular cancer. We used the unique infrastructure of antenatal and perinatal care in Sweden to obtain prospectively collected exposure information on gestational hypertension, preeclampsia, and other conditions during pregnancy among patients with testicular cancer and controls.

The Swedish Medical Birth Register. The Swedish Medical Birth Register was initiated in 1973 and includes prospectively collected information during pregnancy, delivery, and the neonatal period for 98% of all pregnancies in Sweden (7). Information is collected on standardized records used all over Sweden. At the first visit to the antenatal care unit, which occurs before the 15th week of gestation in 95% of the pregnancies (8), information on prior pregnancies, height, and smoking habits of the mother is recorded. Also, a first measurement of weight, hemoglobin, blood pressure, and a dipstick urine sample for protein and glucose is recorded. Thereafter, a pregnant woman generally visits the antenatal care unit 13 times before delivery (9). During these visits, as well as at the delivery ward, the same measurements are repeated according to specific routines. After delivery, information on birth weight, placental weight, etcetera are recorded. All records are stored at the archives of the delivery units, and part of the information is forwarded to and computerized by the Medical Birth Register.

The study design. The study design has been described previously (10). We used the Swedish Medical Birth Register to define a cohort of virtually all men born in Sweden from 1973 onwards, in which we conducted a nested case-control study through linkage to the Swedish Cancer Register (10). We manually complemented the register data with data from the paper charts at the different delivery unit archives throughout Sweden.

We identified 293 singleton cases of testicular germ cell cancer (ages ≥15 at diagnosis) diagnosed before 2002. As eligible controls, we identified the first three men born at the same hospital after a case. Only controls who were alive and without testicular cancer at the time of diagnosis of the corresponding case were included in the analysis. After excluding 18 (2%) twin controls, 293 cases and 861 controls remained for analysis.

Statistical analysis. We analyzed the following variables: maternal smoking during pregnancy, maternal age at delivery, birth order, gestational duration, birth weight, maternal height, body mass index (BMI) before pregnancy, weight gain during pregnancy, anemia during pregnancy (defined as at least one hemoglobin value of <100 g/L any time during pregnancy), mild and severe gestational hypertension (defined as at least two measurements of systolic blood pressure of 140 or more and/or diastolic blood pressure of 90 or more after gestational week 20, or at least one measurement of systolic blood pressure of 160 or more and/or diastolic blood pressure of 100 or more after gestational week 20, respectively), mild and severe preeclampsia (defined as mild or severe hypertension, respectively), with proteinuria during pregnancy (defined as at least one positive dipstick urine sample for protein after gestational week 20), glucosuria during pregnancy (defined as at least one positive dipstick urine sample for glucose any time during pregnancy), placental weight, maternal allo-immunization (diagnosis ICD-8 codes: 634.91, 634.92, or 634.93), neonatal jaundice (diagnosis ICD-8 codes: 778.93, 778.94, or 778.96), and cryptorchidism (diagnosis ICD-8 code: 752.10).

We estimated odds ratios (OR) with corresponding 95% confidence intervals (CI) through conditional logistic regression using SAS statistical software (version 9.1, SAS Institute, Inc.) procedure PHREG. All multivariable models, adjusted for age and place of birth by design, included maternal age at delivery and birth order. Gestational duration and birth weight were not retained in the final models because they exerted limited confounding. Additionally, we conducted analyses separately for the two histologic groups of testicular cancer: seminomas and nonseminomas.

We found a negative association between risk of testicular cancer and severe gestational hypertension (OR, 0.29; 95% CI, 0.12–0.74), whereas the risk was increased for mild gestational hypertension (OR, 1.62; 95% CI, 0.98–2.69). When we analyzed the joint effect of hypertension and proteinuria (i.e., preeclampsia), the risk of testicular cancer was decreased for severe preeclampsia (OR, 0.27; 95% CI, 0.06–1.17), but slightly increased for mild preeclampsia (OR, 1.31; 95% CI, 0.56–3.03; Table 1).

Low maternal BMI (<19) was associated with a decreased risk of testicular cancer (OR, 0.63; 95% CI 0.39–1.01), but there was no association with high BMI. Anemia during pregnancy was associated with an increased risk of testicular cancer (OR, 1.68; 95% CI, 0.99–2.86). We found that maternal height, weight gain during pregnancy, or glucosuria during pregnancy had no effect. Cryptorchidism was associated with an almost 3-fold increased risk of testicular cancer (Table 2).

Smoking, maternal age at delivery, birth order, gestational duration, or birth weight had no effect on the risk of testicular cancer, and the risk estimates were similar to those previously reported (ref. 10; data not shown).

Results stratified according to histologic group are presented in Table 3. Mild gestational hypertension was a risk factor for seminomas (OR, 5.55; 95% CI, 1.72–17.9), but was not associated with the risk of nonseminomas (OR, 1.17; 95% CI, 0.65–2.11). For severe gestational hypertension, the risk of both seminomas (OR, 0.24; 95% CI, 0.03–2.06) and nonseminomas (OR, 0.30; 95% CI, 0.10–0.84) was decreased.

We found a highly significant >3-fold decrease in the risk of testicular cancer among the sons of women with severe gestational hypertension, but a 60% increase in risk among the sons of women with mild gestational hypertension. Similar paradoxical patterns, although less statistically precise, appeared when we studied mild and severe preeclampsia.

Given the population-based study design with prospectively recorded exposure data, selection and information bias are unlikely. An important limitation is, however, that we defined proteinuria on the basis of the concentration of proteins detected in a dipstick urine sample, which has a high intraindividual variability and correlates poorly with the amount of protein found in 24-hour urine samples (11). This misclassification of proteinuria should be nondifferential but could be severe enough to make our proteinuria variable only weakly informative. Consequentially, our preeclampsia variable, which was created using the combination of gestational hypertension and proteinuria, may add little further to the gestational hypertension variable.

Two other limitations are that of multiple comparisons (which were not adjusted for in the analysis) and the small sample size, increasing the likelihood of spurious associations. In theory, the problem is less of a concern for associations for which we had a priori hypotheses, such as that between gestational hypertension/preeclampsia and testicular cancer. On the other hand, in the subgroup analysis presented in Table 3, the risk of false-positive findings call for more caution in the interpretation.

Fetal exposure to gestational hypertension and preeclampsia has previously been associated with a reduced risk of breast cancer in adult life (12–14), supposedly reflecting a protective effect of exposure to low levels of estrogens during fetal life. For testicular cancer, on the other hand, studies of gestational hypertension and/or preeclampsia are sparse, and results are conflicting. In accordance with our finding, Aschim and colleagues found, using registry-based exposure information, a 2-fold increase in risk of seminomas for hypertension during pregnancy, but no increase in risk for nonseminomas (15), whereas Cook and colleagues recently found an ∼50% increase in risk for both seminomas and nonseminomas (16), but no further categorization was done to distinguish severe from mild hypertension in these studies. It should be noted that when hypertension is assessed without subcategorization into mild and severe, the mildly exposed should dominate the exposed group. With respect to preeclampsia, no previous study has found a significant association with testicular cancer risk (15–18). None of the previous studies were, however, designed specifically to study these exposures, or used detailed and prospective information from repeated antenatal care visits.

It has been postulated that the mild late-onset form of preeclampsia is etiologically different from the severe early onset form (19, 20). Whereas mild preeclampsia at term may represent a hypertensive response to minimally impaired (or perhaps even normal) perfusion of the placenta in highly sensitive or predisposed women, the severe disease could be the physiologic reaction of any mother due to profoundly reduced placental perfusion. In the context of our results, this could imply that severe gestational hypertension and severe preeclampsia reflects profoundly impaired placental perfusion, whereas the mild forms do not.

Although preeclampsia is admittedly a blunt and unspecific measure of pregnancy hormones, the hormonal profile in preeclamptic women seems to be different from that in women that are normotensive during pregnancy. Women with preeclampsia, notably severe preeclampsia, have been reported to have lower estrogen levels during pregnancy (5, 21). Some studies have also reported higher levels of androgens in women with preeclampsia (22), and low levels of androgens have been implicated in the pathogenesis of testicular cancer (23).

The placenta is, in addition, a supplier of hCG to the fetus in early pregnancy (24). hCG stimulates the production of testosterone and maturation of the testes, and it is therefore biologically plausible that placental malfunctions have implications for the development of the fetal germ cells. In normal pregnancies with male fetuses, hCG levels decline considerably in the third trimester. In preeclamptic pregnancies, however, hCG remains at high levels throughout the pregnancy (6). Interestingly, these alterations seem to be present in severe, but not mild, preeclampsia (25, 26). Thus, if severe gestational hypertension and preeclampsia, as measured by us, reflect the same underlying pathology and share the same hormonal profile, our findings could indicate that exposure to low levels of estrogens and/or high levels of hCG in late pregnancy decrease the risk of testicular cancer.

Another explanation for our results may be differences in treatments between mild and severe hypertension during pregnancy. In Sweden, during the late 1970s, almost all women with severe hypertension were treated with hydralazine and diuretics or β-blockers, whereas many of those with mild hypertension were not medically treated (27). We know, however, of no biologically plausible mechanism that would accommodate a potentially protective effect of an antihypertensive substance.

Our data provide evidence against intrauterine growth retardation as a cause of testicular cancer. Low birth weight or gestational duration had no effect, and the strongly negative association with severe gestational hypertension and severe preeclamspia is diametrically opposite to what would be expected if growth restriction was indeed a causal factor for testicular cancer.

Primarily for comparability with previous studies, and without any explicit prior hypothesis, we conducted analyses stratified by histologic group. The decreased risk for severe gestational hypertension remained for both histologic groups, but several of the other results (Table 3) might indicate heterogeneity between seminomas and nonseminomas. The small number of seminomas, due to the young age of the study population, hinders the evaluation of heterogeneity as well as precluding firm conclusions, but it should be noted that the association between mild hypertension and seminomas is specifically in accordance with the study by Aschim and colleagues (15).

In conclusion, we have found a strong negative association between the risk of testicular cancer and severe gestational hypertension/severe preeclampsia. The interpretation is complicated by the finding of an increased risk associated with mild gestational hypertension. We conclude that these data provide further evidence against intrauterine growth retardation as a cause of testicular cancer. Our findings may rather reflect a potentially protective effect of the changes in the levels of pregnancy hormones such as estrogen and hCG that occur in severe gestational hypertension and preeclampsia.

No potential conflicts of interest were disclosed.

Grant support: The Swedish Cancer Society 4730-B02-01XAB, 4730-B03-02XBB, 4594-B01-01XAC, and 4594-B04-04XAB; the Compagnia San Paolo/Fondazione Internazionale in Medicina Sperimentale and the Italian Center for Research on Cancer (L. Richiardi); and the Stockholm County Council/Karolinska Institutet (O. Akre, M. Kaijser, and A. Pettersson).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Dr. Fredrik Granath and Lena Brandt for their valuable statistical input, and Ulrika Undén for her valuable contribution.