Cervical cancer is responsible for 190 000 deaths annually, about 78% of which occur in developing countries, where it is the leading cause of cancer mortality in women (Pisani et al., 1999). Squamous cell carcinomas (SCC) are the main histological type of cervical cancer while adenocarcinomas account for 10–15%. Human papillomavirus (HPV) is the main causal factor for both histological forms of cervical cancer, type 16 dominating in SCC and type 18 in adenocarcinoma (IARC, 1995;Pisani et al., 1997;Vizcaino et al., 1998). The relative risks in studies where HPV DNA has been demonstrated are >20 and the attributable risk has been estimated at 89% worldwide (IARC, 1995;Pisani et al., 1997;Vizcaino et al., 1998). The incidence of cervical SCC has declined during the past decades throughout the developed countries, largely because of Pap screening (Gustafsson et al., 1997). By contrast, the incidence of cervical adenocarcinoma has been increasing in most developed countries, particularly among younger women (Gustafsson et al., 1997;Vizcaino et al., 1998). HPV has also been implicated in other anogenital cancers including those of the anus, vulva and penis, and cancer of the head and neck, and probably also the oesophagus (IARC, 1995;Bjorge et al., 1997;Gillison et al., 2000). Host susceptibility may be a contributing factor in cervical cancer because family history and certain human leukocyte antigen (HLA) haplotypes are risk factors (Sanjeevi et al., 1996;Hemminki et al., 1999;Magnusson et al., 1999). Host immune surveillance keeps control of viral infections and a severe immunosuppression increases the risk of cervical cancer and other malignancies, such as SCC of skin and lip, and of Hodgkin's and non-Hodgkin's lymphoma (Birkeland et al., 1995;IARC, 1996).
Here, the nationwide Swedish Family-Cancer Database was used to analyse familial relationships in cervical cancers by histological type and extend earlier studies on all cervical cancer on a previous version of the Database (Hemminki et al., 1999). Data for invasive and in situ cancers were obtained from the Swedish Cancer Registry for the years 1958–1996. For in situ cancers reporting increased in the late 1960s when population screening for cervical cancer was instituted in Sweden (Hemminki and Vaittinen, 1998). Sexual behaviour is a major attribute of HPV infections, and sexual behaviour is related to many lifestyle and socio-economic factors, which family members have in common. Thus a family study of cervical cancer should consider possible confounding by a number of environmental and lifestyle factors. We include some of these possible confounders. Familial risks are calculated for daughters and mothers with in situ and invasive cervical cancer when their relatives have similar or different cancers.
Subjects and methods
The Swedish Family-Cancer Database includes all persons born in Sweden after 1934 with their biological parents, totalling over 9 million individuals; an updating has been carried out since our previous study (Hemminki et al., 1999). Cancers, including in situ cervical cancers, were retrieved from the nationwide Swedish Cancer Registry for the years 1958–1996. A four-digit diagnostic code according to the 7th revision of the International Classification of Diseases (ICD-7) was used. The following ICD-7 codes were pooled: ‘oral’ cancer codes 161 (larynx) and 140–148 (lip, mouth, pharynx), except for code 142 (salivary glands); ‘lymphoma’ codes 200 (non-Hodgkin's lymphoma), 201 (Hodgkin's disease) and 202 (reticulosis); and ‘leukaemia’ codes 204–207 (leukaemias), 208 (polycythaemia vera) and 209 (myelofibrosis). Rectal cancer, ICD-7 code 154, was separated for anus (squamous cell carcinoma, 154.1) and mucosal rectum (154.0). Basal cell carcinoma of the skin is not recorded in the Cancer Registry.
Standardized incidence ratios (SIR) were calculated for women diagnosed for invasive or in situ cervical cancer by dividing the observed numbers of second events by the expected ones, calculated as person-years at risk from the age- (5-year age groups), period- (5-year periods) and sex-specific incidence rates of all women (Esteve et al., 1994). Confidence intervals (95% CI) were calculated assuming a Poisson distribution (Esteve et al., 1994).
Trends in the incidence of cervical cancer were evaluated by means of Poisson regression to quantify the effects of time while adjustment was made for age and period of diagnosis (Breslow and Day, 1987Clayton and Schifflers, 1987a, b). This method transforms the underlying incidence rate to a regression function that describes the relationship between predictor variables, such as age, period of diagnosis year, birth cohort, and other parameters that are estimated directly by maximum likelihood techniques. Other predictor variables included in this statistical analysis were socio-economic status (four groups: farmers, professionals, blue-collar workers and others), and areas of living (two groups: big city-Stockholm, Malmö, Göteborg, and other areas). The records for cells with no person-years of observation were omitted. One last level was chosen as a baseline category. Statistical significance of each effect was examined by using the global goodness-of-fit statistic (G 2 ), which has a chi-square distribution under the null hypothesis. Relative risks (RR) were calculated from exponentiated regression coefficients.
The Swedish Family-Cancer Database included two generations, 3 017 541 mothers and 3 130 399 daughters. Mothers had 17 287 invasive cervical cancers, of which 14 885 were SCC and 1817 adenocarcinomas; daughters had 4440 invasive cervical cancers, of which 3518 were SCC and 760 adenocarcinomas. Mothers had 106 791 in situ cancers, of which 95 434 were SCC and 367 adenocarcinomas; daughters had 84 290 in situ cancers, of which 74 507 were SCC and 331 adenocarcinomas. We have carried out all analyses in two ways: in daughters by mothers’ cervical cancers, and in mothers by daughters’ cervical cancers. However, all the data presented below are on daughters by mothers (except for Table 2). The number of adenocarcinoma cases was too small for a separate analysis of this histology.
Risks for daughters’ invasive and in situ cervical SCCs are shown in Table 1
by mothers’ cervical cancer. SIRs for invasive cancers were close to 2.00 by both mother's invasive and in situ SCC cancer. Adenocarcinoma in mothers caused no significant increase in risk. Daughters’ SIRs for in situ SCC were lower, 1.43–1.92, and even maternal adenocarcinoma was a significant risk factor.
SIRs for cervical SCC in mothers are shown in Table 2
by cervical cancer in daughters. SIR for invasive SCC was highest (2.62) when two daughters had in situ cancer; SIR for in situ SCC was equally high (2.61) when two daughters were affected. SIR for maternal invasive cervical SCC was somewhat higher when a daughter had invasive SCC compared with in situ SCC (2.24 versus 1.64). None of the SIRs were significant when a daughter had only adenocarcinoma.
Age-specific relative risks of invasive cervical SCCs in daughters by invasive maternal cancers are shown in Table 3
. Overall, the RR was 1.97, and the highest RR, 4.23, was among daughters diagnosed before age 30 when mothers’ diagnostic age was below 40. However, there was another high RR of 2.42 at diagnostic ages ≥40 for daughters and 50–59 for mothers.
Age-specific relative risks of in situ cervical SCCs in daughters by in situ cancers in mothers are shown in Table 4
. The overall RR, 1.92, equalled that in the previous Table. The RRs were highest in the youngest diagnostic groups of daughters and mothers.
SIR of SCC in daughters is shown in Table 5
by maternal and paternal cancers. Sites were included if some associations to cervix were statistically significant. For invasive SCC cervical cancers in daughters, associations were found to maternal liver (SIR 1.46) and skin SCC (SIR 1.65), and to paternal lung cancer (SIR 1.26). In situ cervical SCC associated with maternal cervical SCC and adenocarcinoma, in addition to maternal and paternal oral and lung cancers. A similar analysis was carried out for maternal cervical cancers by cancer in daughters and sons (data not shown). No cancer in sons was associated with maternal cervical cancer. SIRs for invasive maternal cervical SCC were 1.82 (95% CI 1.14–2.29) and 2.08 (95% CI 0.99–3.57) for cervical SCC and adenocarcinoma in daughters. Increased SIRs for invasive maternal cervical SCC were also found for daughters’ anal (SIR 3.82, 95% CI 1.21–7.91), lung (SIR 1.79, 95% CI 1.09–2.66) and breast cancers (SIR 1.18, 95% CI 1.02–2.29). The SIR for skin cancer, 2.00, was of borderline significance (n = 11, 95% CI 0.99–3.36).
We analysed association of cervical SCCs by specific lung cancer histology (data not shown). Neither daughters’ nor mothers’ invasive cervical SCC was associated with a lung histological subgroup. By contrast, associations were found with in situ cervical cancers, due to their large number. However, neither cervical SCC nor adenocarcinoma were linked to the same histologies in lung. The highest SIR for in situ cervical SCC was to lung adenocarcinoma (1.90, 95% CI 1.03–3.02).
A Poisson regression analysis was carried out on invasive SCC and adenocarcinoma of all women (Table 6
). Both SCC and adenocarcinoma showed maximal risks in late-onset diagnostic groups but SCC also had an early age maximum at 40–44 years. Early diagnostic period and the youngest birth cohorts showed the highest RRs for both histological types. Familial RRs were lower for adenocarcinoma than SCC. However, adenocarcinomas were significantly increased from in situ cervical cancer in daughters or sisters. Both models showed good fit as deviance/degrees of freedom were below unity.
Few studies have previously addressed familial relationships in cervical cancer, and none by specific histology. Here we show that the familial risks were about 2.0, for invasive and somewhat less for in situ cervical SCC; the risks were somewhat higher at young age groups. The results were consistent irrespective of whether the risks were calculated for daughters by maternal cancer, or for mothers by daughters’ cancer. Too few cases of adenocarcinomas were diagnosed for detailed analysis. However, association of cervical adenocarcinomas in one generation with cervical SCC in another showed rather similar SIRs for SCC–SCC associations, although the significance levels were lower due to small numbers of pairs. The similarity in familial risks between SCC and adenocarcinomas, if confirmed in a larger series, would be another piece of evidence indicating a common aetiology and clinical course for these histological subtypes (Gustafsson et al., 1997;Vizcaino et al., 1998). Because of the paucity of data on adenocarcinoma we limit the discussion below to SCC only.
HPV is the predominant cause of cervical cancer and infection is related to sexual behaviour, such as number of partners (IARC, 1995). Sexual behaviour correlates with other lifestyle factors such as smoking and alcohol consumption, and this is the most likely explanation for the observed associations between cervical cancer in daughters and lung, liver and oral cancers in parents (Table 5). Lung and oral cancers are also common second malignancies after invasive and in situ cervical cancer, and they are in excess in husbands of cervical cancer patients (Hemminki and Dong, 2000;Hemminki et al., 2000). Socio-economic factors associate with the incidence of cervical cancer and in the present study farmers and professionals had lower incidence than blue-collar workers, and residents in large cities had a higher incidence than people living in small cities or the countryside (IARC, 1997;Pukkala and Weiderpass, 1999). However, these adjustments did not change the familial effects.
Familial risk estimates cannot distinguish between environmental and genetic causes of cancer. Even if we were able to control for some environmental variables there are likely to be remaining factors explaining behavioural similarity in families that we could not take into consideration. In our previous article on familial cervical cancer we discussed the findings that supported genetic and environmental interpretations of the findings (Hemminki et al., 1999). These points are still relevant to the present work. The findings that support the presence of inherited susceptibility include the relatively high familial risk, an increased risk in families where three or more individuals were affected, increased RRs at early onset and unchanged RR after adjustment for potential confounders such urban residence and socio-economic status. However, the counter argument on the sharing of promiscuous behaviour among family members cannot be rejected. The only piece of data that is not compatible with the current knowledge of target tissues for HPV is the association in skin SCC and cervical SCC between mothers and daughters. This was unlikely to be a chance finding because it was observed in daughters’ cervical cancer by mothers’ skin cancer and vice versa. This could be a marker of mild immunosuppression that could be a heritable factor (Hemminki et al., 1999).
A previous study used Swedish family data to compare mainly cervical in situ cancer risks between siblings and half-siblings and concluded that the results support genetic interpretation because the risk in half-siblings was about one-half of that of full siblings (Magnusson et al., 1999). However, half-siblings are due to remarriage of one or both parents and these families may differ in many respects from families of full siblings. In a twin study from Sweden, in situ cervical cancer showed a heritability estimate of 39–46%; no data were available for invasive cervical cancer because of lack of cases (Ahlbom et al., 1997). However, in a recent combined Nordic twin study the heritability estimate for invasive cervical cancer was 0% but with wide confidence intervals (Lichtenstein et al., 2000). In conclusion, the genetic epidemiology data on heritable effects in cervical cancer remain suggestive only and require molecular markers for confirmation.
The study was supported by the Cancer Fund.
1. Ahlbom A , Lichtenstein P , Malmstrom H et al. ( 1997 ). Cancer in twins: genetic and nongenetic familial risk factors . J Natl Cancer Inst 89 : 287 – 293 .
2. Birkeland SA , Storm HH , Lamm LU et al. ( 1995 ). Cancer risk after renal transplantation in the Nordic countries, 1964–1986 . Int J Cancer 60 : 183 – 189 .
3. Bjorge T , Hakulinen T , Engeland A et al. ( 1997 ). A prospective, seroepidemiological study of the role of human papillomavirus in esophageal cancer in Norway . Cancer Res 57 : 3989 – 3992 .
4. Breslow NE, Day NE (1987). Statistical Methods in Cancer Research
, Vol II:The design and analysis of cohort studies
. IARC Scientific Publications No. 82. IARC, Lyon, pp. 1–406.
5. Clayton D , Schifflers E ( 1987a ). Models for temporal variation in cancer rates. I. Age-period and age-cohort models . Stat Med 6 : 449 – 467 .
6. Clayton D , Schifflers E ( 1987b ). Models for temporal variation in cancer rates. II. Age-period-cohort models . Stat Med 6 : 469 – 481 .
7. Esteve J, Benhamou E, Raymond L (1994). Statistical Methods in Cancer Research
, Vol IV:Descriptive epidemiology
. IARC Scientific Publications No. 128. IARC, Lyon, pp. 1–302.
8. Gillison ML , Koch WM , Capone RB et al. ( 2000 ). Evidence for a causal association between human papillomavirus and a subset of head and neck cancers . J Natl Cancer Inst 92 : 709 – 720 .
9. Gustafsson L , Ponten J , Bergstrom R , Adami HO ( 1997 ). International incidence rates of invasive cervical cancer before cytological screening . Int J Cancer 71 : 159 – 165 .
10. Hemminki K , Dong C ( 2000 ). Cancer in husbands of cervical cancer patients . Epidemiology 11 : 347 – 349 .
11. Hemminki K , Vaittinen P ( 1998 ). Familial risks in in situ
cancers from the Family-Cancer Database . Cancer Epidemiol Biomarkers Prev 7 : 865 – 868 .
12. Hemminki K , Dong C , Vaittinen P ( 1999 ). Familial risks in cervical cancer: is there a hereditary component? . Int J Cancer 82 : 775 – 781 .
13. Hemminki K , Dong C , Vaittinen P ( 2000 ). Second primary cancer after in situ
and invasive cervical cancer . Epidemiology 11 : 457 – 461 .
14. IARC (International Agency for Research on Cancer) (1995). IARC Monograph on the Evaluation of Carcinogenic Risks to Humans: Human Papillomaviruses
. IARC, Lyon.
15. IARC (1996). IARC Monograph on the Evaluation of Carcinogenic Risks to Humans
, Vol 67:Human Immunodeficiency Viruses and Human T-Cell Lymphotropic Viruses
. IARC, Lyon, France, pp. 1–424.
16. IARC (1997) Social Inequalities and Cancer
. IARC, Lyon.
17. Lichtenstein P , Holm NV , Verkasalo PK et al. ( 2000 ). Environmental and heritable factors in the causation of cancer – analyses of cohorts of twins from Sweden, Denmark, and Finland . N Engl J Med 343 : 78 – 85 .
18. Magnusson PK , Sparen P , Gyllensten UB ( 1999 ). Genetic link to cervical tumours [letter] . Nature 400 : 29 – 30 .
19. Pisani P , Parkin DM , Munoz N , Ferlay J ( 1997 ). Cancer and infection: estimates of the attributable fraction in 1990 . Cancer Epidemiol Biomarkers Prev 6 : 387 – 400 .
20. Pisani P, Parkin DM, Bray F, Ferlay J (1999). Estimates of the worldwide mortality from 25 cancers in 1990 [published erratum appears in Int J Cancer
1999 Dec 10, 83(6): 870–3]. Int J Cancer83:
21. Pukkala E , Weiderpass E ( 1999 ). Time trends in socio-economic differences in incidence rates of cancers of the breast and female genital organs (Finland, 1971–1995) . Int J Cancer 81 : 56 – 61 .
22. Sanjeevi CB , Hjelmstrom P , Hallmans G et al. ( 1996 ). Different HLA-DR-DQ haplotypes are associated with cervical intraepithelial neoplasia among human papillomavirus type-16 seropositive and seronegative Swedish women . Int J Cancer 68 : 409 – 414 .
23. Vizcaino AP , Moreno V , Bosch FX et al. ( 1998 ). International trends in the incidence of cervical cancer: I. Adenocarcinoma
and adenosquamous cell carcinomas . Int J Cancer 75 : 536 – 545 .