Correspondence to: Morten Frisch, Department of Epidemiology Research, Statens Serum Institut, Artillerivej 5, 206/317, DK-2300 Copenhagen S, Denmark, Fax: +45–3268-3868, E-mail: email@example.com
Few population-based studies have investigated the relation between living arrangements and risk of invasive penile squamous cell carcinoma (iP-SCC). Using long-term national cancer registry data in Denmark we examined incidence trends of iP-SCC. Furthermore, we examined the relation between marital status, cohabitation status and risk of iP-SCC using hazard ratios (HRs) with 95% confidence intervals (CIs) obtained in Cox proportional hazards regression analyses as our measure of relative risk. Overall, 1,292 cases of iP-SCC were identified during 65.6 million person-years of observation between 1978 and 2010. During this period, the WHO world age-standardized incidence remained relatively stable (p-trend = 0.41) with an average incidence of 1.05 cases per 100,000 person-years. When compared to married men, those who were unmarried (HR 1.37; 95% CI: 1.13–1.66), divorced (HR 1.49; 95% CI: 1.24–1.79) or widowed (HR 1.36; 95% CI: 1.13–1.63) were at increased risk of iP-SCC. Regarding cohabitation status, single-living men were at increased risk of iP-SCC compared to men in opposite-sex cohabitation (HR 1.43; 95% CI: 1.26–1.62). Risk increased with increasing numbers of prior opposite-sex (p-trend = 0.02) and same-sex (p-trend < 0.001) cohabitations. In conclusion, single-living Danish men and men who are not currently married are at increased risk of iP-SCC, and the risk increases with the number of prior cohabitations, perhaps reflecting less stable sexual relations in these subgroups.
Invasive penile cancers, of which up to 95% are penile squamous cell carcinomas (iP-SCC), are rare malignancies that occur mainly among elderly men at a mean age around 60 years. In industrialized countries the age-standardized incidence of penile cancer is generally around or less than 1 per 100,000 person-years.[2-4] However, the annual incidence varies considerably worldwide from around 4 per 100,000 in countries such as Uganda and Brazil,[4-6] while it is close to zero among Israeli Jews. In Denmark, the incidence declined from 1.15 to 0.82 per 100,000 person-years between 1943 and 1990.
Several population-based studies have evaluated the trends in penile cancer incidence over time. Most studies have reported a declining incidence,[5-11] while some have indicated a stable incidence over time.[12, 13] A recent study from the Netherlands even suggested a slightly increasing trend of penile cancer incidence between 1989 and 2006, but this was explained by a significant increase in penile carcinoma in situ over time, whereas the incidence of invasive penile cancer remained stable. A recent Danish study also reported an overall increase in the incidence of penile cancer in recent years. However, concerns regarding methodological problems in that study have been raised, such as the use of supplementary data sources that were incomplete before the mid-1990's, thus contributing to the impression of an increasing trend.
Risk factors for penile cancer include phimosis,[17-21] tobacco use[17, 18, 20, 21] and multiple sex partners.[17, 20] For decades childhood circumcision has been associated with reduced risk of penile cancer, presumably due to the reduction in penile tissue per se and the elimination of phimosis in circumcised boys.[18, 19] It has been estimated that it may take up to 322,000 neonatal circumcisions to prevent one case of invasive or in situ penile cancer.
Infection with cervical cancer-associated high-risk types of human papillomavirus (HPV) has been established as a risk factor for iP-SCC.[17, 18, 20] It seems that there may be two etiologically distinct forms of iP-SCC, of which around half are believed to be attributable to such high-risk HPV types.[23, 24] In a recent Danish study, 52% of 116 iP-SCCs were positive for high-risk HPV, notably HPV 16.
Since penile HPV infection is sexually acquired, differences in sexual history and marital status might plausibly impact on risk of iP-SCC. In previous case-control studies men with penile cancer reported a greater number of lifetime female sex partners and an earlier age at first intercourse,[17, 19, 20] although not all studies have replicated these findings. The number of same-sex sex partners seems unrelated to the risk of penile cancer.[17, 20] Regarding marital status, one previous register-based case-control study showed that Danish men diagnosed with penile cancer between 1943 and 1990 were more likely to never have married than patients with cancers of the colon or stomach, and penile cancer patients who had never married were younger at diagnosis. Similarly, a case-control study in western Washington reported that men diagnosed with penile cancer were more likely to never have married than healthy controls. However, another Danish case-control study with patients diagnosed between 1993 and 1998 showed no association between marital status and risk of iP-SCC.
As recent studies have shown equivocal results, we decided to assess the time trends of iP-SCC in Denmark for the 33-year period between 1978 and 2010. Furthermore, former studies have only sporadically investigated the association between living arrangements and risk of penile cancer. In Denmark, people increasingly live in informal partnerships without being married. Therefore, in addition to marital status, we used a newly developed algorithm for cohabitation status to investigate the risk of iP-SCC in Denmark according to cohabitation status.
Material and Methods
As our study cohort we identified the entire Danish male population age 18 years or older (2.4 million men in 2010) in the Civil Registration System (CRS), a continuously updated national demographic database. Since its establishment in 1968 all Danish residents have been assigned a unique 10-digit identification number, enabling identity-secure linkage of data between registers. In analyses of incidence trends in iP-SCC we used data for the period 1978–2010. For the analysis of associations of marital status and cohabitation status with risk of iP-SCC we restricted the period to 1982–2010, as potential socioeconomic confounders were available only since 1982.
Between January 1, 1978 and December 31, 2010 all incident cases of invasive penile cancer were identified in the Danish Cancer Registry, which is a high-quality national register containing information on all malignancies diagnosed in the Danish population. Specifically, penile cancers were identified under topography codes C60.0-C60.9 according to the International Classification of Diseases for Oncology, 3rd edition. Invasive penile cancers with histology codes 8050–8076, 8094 or 8120–8124 delineated cases of iP-SCC.
Marital status and cohabitation status
Each cohort member's continually updated marital status history was established on a day-to-day basis in the following categories: (i) unmarried (i.e. never-married), (ii) married, (iii) divorced or (iv) widowed between April 1, 1968, when the CRS was established, and December 31, 2010. Since October 1, 1989, an additional category of (v) current or past same-sex marriage was put into operation with the implementation of the world's first national law on registered same-sex partnerships.
Similarly, the cohort's observation time was categorized on a day-to-day basis in five cohabitation categories: periods of cohabitation with one or both parents were categorized as (i) living with parents. Periods when cohort members lived with ≥9 unrelated adults at the same address were categorized as (ii) living in a multiadult household. Periods when cohort members did not share a specific address with any current or former spouse or any unrelated adult with an age difference of up to ±15 years were categorized as (iii) living as a single person. Periods when cohort members shared address with only one other adult (either a spouse regardless of age difference or an unrelated adult with an age difference of up to ±15 years) were categorized as (iv) living in opposite-sex cohabitation or (v) living in same-sex cohabitation, as appropriate. If cohort members lived at the same address as 2–8 unrelated adults with an age difference of up to ±15 years, further criteria (which are described in more detail elsewhere) were used in an attempt to identify the possible cohabiting partner of each cohort member. When successful, such periods were categorized as living in opposite-sex cohabitation or same-sex cohabitation, as appropriate; otherwise periods of living with 2–8 other adults were categorized as living in a multiadult household.
Using the same data on marital status and cohabitation status we also kept track on a day-to-day basis of each cohort member's continually updated number of prior opposite-sex marriages and of opposite-sex cohabitations and same-sex cohabitations of at least one year's duration since April 1, 1968.
We used weights according to the new WHO World Standard Population to calculate age-standardized incidence rates of iP-SCC for each year between 1978 and 2010. These incidence rates were plotted graphically. We performed linear regression on the age-standardized incidence rates, assessed the slope of the best fitting line, and evaluated it against the null hypothesis of no change in incidence over the 33-year study period by means of the maximum likelihood test.
Next, we used cubic splines restricted to be linear in the tails to graphically depict age-specific incidence rates of iP-SCC for the entire study period 1978–2010, applying knots every 10 years between age 25 and 85 years.
In analyses of associations of marital status and cohabitation status with risk of iP-SCC, we used Cox proportional hazards regression to calculate hazard ratios (HRs) with 95% confidence intervals (CIs) as our measure of relative risk. In these analyses, which were carried out for the observation period 1982–2010, we used periods in opposite-sex marriage and opposite-sex cohabitation as reference categories, respectively. In all analyses we used attained age in 1-day intervals as the underlying time scale. All analyses were stratified according to birth year in 1-year intervals, thereby also indirectly taking possible effects of calendar period into account. In addition to controlling for age, birth year and calendar period, we further stratified for actual values of municipality (in 98 categories), and we included population density (in five categories), educational level (eight categories) and relative income two years prior to the actual year (in quartiles) as class variables in all analyses. These potential confounders, which were obtained in the CRS and in Statistics Denmark, were chosen because a recent Danish study showed that these factors were nontrivial confounders of associations between marital status and cohabitation status on one side and mortality on the other.
We also used Cox regression to calculate HRs for the possible impact on iP-SCC risk of prior opposite-sex marriages (categories 0, 1, 2, 3 or more), opposite-sex cohabitations of at least one year's duration (categories 0, 1, 2, 3, 4 or more) and same-sex cohabitations of at least one year's duration (categories 0, 1, 2, 3 or more) since April 1, 1968; corresponding trend tests were based on actual values without truncation of the upper categories. Ninety-five percent CIs excluding unity and two-sided p-values <0.05 were considered indicators of statistical significance. All analyses were performed using SAS version 9.2.
Incidence of invasive penile squamous cell carcinoma in Denmark (1978–2010)
During 65.6 million person-years between 1978 and 2010, a total of 1,428 men living at a Danish address were diagnosed with invasive penile cancer (regardless of histological type) in the Danish Cancer Registry. Of these, iP-SCC accounted for 1,292 cases (90%). Age distribution and selected tumor characteristics are shown in Table 1.
Table 1. Patient and tumor characteristics of 1292 invasive penile squamous cell carcinomas, Denmark 1978–2010
ICD-10, International Classification of Diseases, 10th edition.
Abbreviations: NOS, not otherwise specified; SCC, squamous cell carcinoma.
Plotting the annual WHO world age-standardized incidence estimates revealed no obvious change in incidence of iP-SCC between 1978 and 2010 (Fig. 1). Reducing the data to a linear description was statistically acceptable (p = 0.37). The slope of the best fitting line suggested an annual increase in age- standardized incidence of 0.003 case of iP-SCC (95% CI: −0.004–0.01) per 100,000 person-years, a change that was not statistically significant (p-trend = 0.41). In the absence of evidence of an increasing trend we assumed a constant incidence rate. Thus, we calculated a summary estimate of the age-standardized incidence for the 33-year study period of 1.05 per 100,000 person-years.
Next, we evaluated the age-specific incidence patterns of iP-SCC based on data for the entire study period (Fig. 2). Incidence rates were low before age 50 years and gradually increased with age. The median age at diagnosis for the 1,292 men with iP-SCC was 68 years (range 22–102) between 1978 and 2010 (Table 1).
Association of marital status with risk of invasive penile squamous cell carcinoma (1982–2010)
During 58.3 million person-years between 1982 and 2010, a total of 1,139 men were diagnosed with iP-SCC. When compared to the reference group of men in opposite-sex marriage, unmarried men, divorced men and widowers were at 37%, 49%, and 36% greater risk of iP-SCC, respectively, after controlling for age, birth year, calendar period, educational level, income, municipality and population density (Table 2). For same-sex married men the HR was 0.88 (95% CI: 0.12–6.31) based on only one case.
Table 2. Risk of invasive penile squamous cell carcinoma as a function of actual marital status and number of prior opposite-sex marriages, Denmark 1982–2010
Controlled for age, birth year, calendar period, educational level, relative income, municipality and population density. Number of prior opposite-sex marriages additionally adjusted for actual marital status.
After taking the actual marital status into account, the risk of iP-SCC increased with the number of prior marriages, although the estimated increase of 10% per prior marriage was not statistically significant (p-trend = 0.24) (Table 2).
Association of cohabitation status with risk of invasive penile squamous cell carcinoma (1982–2010)
Single-living men were at 43% greater risk of iP-SCC compared to men living in opposite-sex cohabitation (HR = 1.43; 95% CI: 1.26–1.62) after controlling for confounders (Table 3). Furthermore, men living with parents (HR 2.14; 95% CI: 0.85–5.41) and those living in same-sex cohabitation (HR 1.29; 95% CI: 0.66–2.50) had elevated risks of iP-SCC, but these findings were not statistically significant. Men living in multiadult households were not at increased risk of iP-SCC (HR 0.92, 95% CI: 0.61–1.38).
Table 3. Risk of invasive penile squamous cell carcinoma as a function of actual cohabitation status and number of prior opposite-sex or same-sex cohabitations, Denmark 1982–2010
Controlled for age, birth year, calendar period, educational level, relative income, municipality and population density. Number of prior opposite-sex and same-sex cohabitations additionally adjusted for actual cohabitation status.
Prior cohabitations of at least 1 year's duration.
After taking the actual cohabitation status into account, the risk of iP-SCC increased linearly with the number of prior cohabitations (13% increase per prior opposite-sex cohabitation, p-trend = 0.02; 38% increase per prior same-sex cohabitation, p-trend < 0.001) (Table 3). In a supplementary analysis where we divided currently same-sex cohabiting men into subgroups according to the duration of their same-sex cohabitation, we found that iP-SCC risk was similar among men who currently lived in a long-term (>5 years) same-sex cohabitation as compared to men who currently lived in opposite-sex cohabitation (HR = 1.02; 95% CI 0.33–3.20, n = 3 cases of iP-SCC).
In the present study, we examined incidence trends of iP-SCC in Denmark and the impact of marital status and cohabitation status on risk of this cancer. Salient findings were that the incidence of iP-SCC remained statistically stable between 1978 and 2010 and that men living alone and those who were not currently married were at increased risk.
Our documentation of a stable incidence of iP-SCC over time corroborates recent findings between 1964 and 2003 in other Nordic countries. Likewise, an Australian study showed a constant incidence in the period 1982–2005. Frisch et al. previously showed a significantly decreasing incidence of penile cancer in Denmark from 1.15 per 100,000 person-years in 1943–1947 to 0.82 per 100,000 person-years in 1988–1990. Following up on that study we here show that the incidence has remained at a stable level of 1.05 new cases per 100,000 person-years between 1978 and 2010. The slightly higher point estimate in the present study (1.05 per 100,000 person-years in 1978–2010) than in the previous Danish study's late period (0.82 per 100,000 person-years in 1988–1990) reflects the use of a new WHO world standard population in the present study, which attributes greater weight to older segments in the population. Other studies from Finland,[8, 9] Uganda[5, 6] and the US[10, 11] have shown a decreasing incidence of penile cancer over time. It has been suggested that the falling incidence in some countries might be due to improvement in hygienic standards and socioeconomic status.
A recent study from the Netherlands suggested a small overall increase in penile cancer incidence (including both in situ and invasive lesions) from 1.4 per 100,000 person-years in 1989 to 1.5 per 100,000 person-years in 2006. The authors explained this finding by an increase for penile carcinoma in situ from 0.1 per 100,000 person-years in 1989 to 0.3 per 100,000 person-years in 2006. We did not include in situ lesions in our study, as cancer registry data on such lesions are incomplete. However, while precancerous penile lesions may well be increasing in incidence, further studies are needed to shed light on this point. Possibly, the increased attention to HPV-related anogenital cancer precursors in recent years may have led to increased detection and reporting of premalignant penile lesions.
The most remarkable findings regarding living arrangements were significantly increased risks of iP-SCC among men who had never married and currently single-living men. Additionally, risk was increased among men who were divorced or widowed, while there was no relationship with actual same-sex marriage or actual same-sex cohabitation. Two previous studies also reported an increased risk of penile cancer among men who had never married. Danish men diagnosed with invasive penile cancer between 1943 and 1990 were more likely to never have married as compared to patients with colon or stomach cancer, and unmarried penile cancer patients were diagnosed at a younger age than those in other marital status categories. Similarly, Daling et al. showed that US men diagnosed with invasive or in situ penile cancer were more likely to never have married than controls. In that study there was no association between divorce or separation and the risk of penile cancer, which stands in contrast to findings in the current study. Another case-control study among Danish men did not find any relation between marital status and risk of penile cancer.
Around half of iP-SCCs are believed to be attributable to sexually transmitted high-risk types of HPV.[23, 24] A meta-analysis of 31 studies including 1,466 penile carcinomas estimated the global HPV prevalence of invasive penile carcinomas to be 47%. This estimate is comparable to that of a recent Danish study showing an HPV prevalence of 52% in 116 specimens of iP-SCCs. Since HPV is sexually transmitted, differences in sexual history and other factors related to risk of HPV acquisition would also plausibly be related to the risk of iP-SCC. Several case-control studies suggest that a greater lifetime number of female sex partners, earlier age at first intercourse as well as a history of genital warts increase the risk of penile cancer,[17, 19, 20] but a Swedish study from the 1980s found no association with sexual history.
Our finding of increased risk of iP-SCC among single-living men and men who are not currently married might reflect less stable partnerships and a greater number of sex partners and thus a greater exposure to HPV. This speculation is indirectly supported by our finding of a dose-response association with numbers of both opposite-sex and same-sex cohabitations of at least one year's duration. A greater number of prior marriages also increased the risk of iP-SCC, although that finding was not statistically significant.
We found no indication of an association between homosexuality and risk of iP-SCC. The risk among men in same-sex marriage and men in same-sex cohabitation was statistically inconspicuous, even among those same-sex cohabiting men who had lived with their same-sex cohabitant for at least five years, a subgroup believed to comprise a relatively high proportion of homosexual men. This is in accordance with other studies that have failed to show an association with male homosexuality.[17, 18, 20] The apparent lack of an increased risk of iP-SCC among homosexual men is interesting in light of the well-documented excess of HPV-related anal squamous carcinomas in this group. It seems, therefore, that the significant dose-response associations with numbers of prior opposite-sex and same-sex cohabitations and the indication of a similar trend with number of prior marriages may reflect a general pattern of unstable living conditions that place these men at higher risk of iP-SCC, presumably through unstable sexual encounters with HPV-infected female partners.
Several factors may influence the interpretation of our findings. We were unable to control for potential confounding factors such as circumcision status, phimosis or smoking, which are known to influence iP-SCC risk.[17-21] Therefore, to the extent that these unmeasured variables correlate with marital status or cohabitation status, confounding may have occurred. In a recent Danish survey, circumcised men were slightly less likely to be married than uncircumcised men, although not significantly so (54% vs 62%, p = 0.07). This suggests that circumcision status is unlikely to be a major confounder of the observed associations between living arrangements and iP-SCC in this study. We are unaware of studies linking a history of phimosis to marital status or cohabitation patterns. Consequently, we consider it unlikely that lacking information about history of phimosis in our cohort would have seriously affected our results. With respect to the possible role of tobacco smoking, previous research suggests that married persons smoke less than unmarried and divorced persons. Therefore, theoretically, our findings linking single-living men and unmarried, divorced and widowed men to an increased risk of iP-SCC might to some extent be explained by lifestyle factors and behaviors other than those associated with increased risk of HPV acquisition. Tobacco smoking could be one such contributing nonsexual factor underlying at least part of the higher risk in nonmarried and single-living men.
In many Western countries, including Denmark, it is increasingly common for couples to live in marriage-like unregistered partnerships. To qualify the interpretation of our marital status analyses we used a newly developed algorithm for cohabitation status to evaluate the risk of iP-SCC. Consistently, married men and men living in opposite-sex cohabitation constituted the groups at lowest risk, presumably due to higher levels of partner stability than what may be the case in other living arrangements. A limitation of our study was that an unknown proportion of opposite-sex and same-sex cohabitations consist of people who live at the same address without being intimate partners. Likewise, while divorced men in Denmark and Sweden are almost four times more likely to have a lifetime number of ≥10 sex partners than married or widowed men, we lacked detailed data on sexual partner numbers to directly examine the extent to which married men and cohabiting men truly have fewer sex partners than unmarried or single-living men. Consequently, marital status and cohabitation patterns may not directly reflect the number of sexual partners. In spite of this we observed an increased risk of iP-SCC both among single-living men and men who were not currently married.
The present study was conducted as a register-based cohort study using routinely recorded high-quality data. Register data was available for exposures (marital status and cohabitation status) and potential confounders (socioeconomic variables) obtained in the CRS and Statistics Denmark, and information about the study outcome (iP-SCC) was obtained in the Danish Cancer Registry. Thus, by study design we avoided the possible selection and recall biases that may influence findings in questionnaire-based case-control studies. Furthermore, with the implementation of cohabitation status we have introduced a partnership measure that may more accurately capture actual living arrangements in Denmark than those reflected by people's marital status.
In conclusion, the incidence of iP-SCC has remained stable in Denmark between 1978 and 2010. Men who are not currently married and men who live alone are at increased risk of iP-SCC, just as a greater number of prior cohabitations, presumably a surrogate measure of relationship instability, was associated with increased risk of iP-SCC.