Familial association of specific histologic types of ovarian malignancy with other malignancies

Authors

  • Justo Lorenzo Bermejo Ph.D.,

    Corresponding author
    1. Division of Molecular Genetic Epidemiology, German Cancer Research Center, Heidelberg, Germany
    • German Cancer Research Center, Division of Molecular Genetic Epidemiology, Im Neuenheimer Feld 580, D-69120 Heidelberg, Germany
    Search for more papers by this author
    • Fax: (011) 49 6221 421810

  • Rajesh Rawal M.Sc.,

    1. Division of Molecular Genetic Epidemiology, German Cancer Research Center, Heidelberg, Germany
    Search for more papers by this author
  • Kari Hemminki M.D., Ph.D.

    1. Division of Molecular Genetic Epidemiology, German Cancer Research Center, Heidelberg, Germany
    2. Department of Biosciences at Novum, Karolinska Institute, Huddinge, Sweden
    Search for more papers by this author

  • The Family-Cancer Database was created by linking registries maintained by Statistics Sweden and the Swedish Cancer Registry.

Abstract

BACKGROUND

Population-based data on the familial association of specific histologic types of ovarian malignancy with other malignancies are limited. Such data may help to elucidate etiologic differences among histologic types of ovarian malignancy.

METHODS

The nationwide Swedish Family-Cancer Database, which includes 10.3 million individuals and 20,974 ovarian carcinomas, was used to calculate standardized incidence ratios and 95% confidence intervals for age- and histology-specific ovarian malignancies in women whose parents or siblings were affected with malignancies at the most common disease sites.

RESULTS

Ovarian malignancy was found to be associated with ovarian, laryngeal, breast, endometrial, liver, and colon carcinoma, as well as myeloma; epithelial ovarian malignancy was found to be associated with ovarian, endometrial, and skin malignancies and with melanoma and myeloma; papillary serous cystadenocarcinoma was found to be associated with ovarian and skin malignancies and with myeloma; and endometrioid carcinoma was found to be associated with endometrial, ovarian, and prostate malignancies and with melanoma. For younger women (ages 40–45 years) whose mothers were affected with endometrial malignancies, the risk of developing endometrioid carcinoma was slightly greater than the risk of developing papillary serous cystadenocarcinoma.

CONCLUSIONS

Specific types of ovarian malignancy may be associated with specific familial disease sites, with such associations depending on age at diagnosis; the strength of the observed associations varied according to histology. Associations were found between endometrioid carcinoma and endometrial malignancy and between serous carcinoma and Hodgkin disease. Cancer 2004;100:1507–14. © 2004 American Cancer Society.

Ovarian tumors are classified according to the tissue type in which they originate. The most common ovarian tumors are epithelial ovarian malignancies, which include papillary serous cystadenocarcinoma, endometrioid carcinoma, serous carcinoma, mucinous cystadenocarcinoma, and clear cell carcinoma. Nonepithelial ovarian malignancies are less common and include granulosa cell and germinal malignancies, immature teratomas, and dysgerminomas. Data regarding familial risk for specific histologic types of ovarian malignancy are limited; such data may help to elucidate etiologic differences among the various histologic types of ovarian malignancy.

Current heavy alcohol consumption was found to be associated with mucinous ovarian tumors but not with nonmucinous tumors.1 In addition, a recent study found that reproductive factors had histology-specific effects with respect to ovarian malignancy: pregnancy and tubal ligation were strongly associated with endometrial histology, whereas clear cell tumors represented the only histologic type associated with noncontraceptive hormone use.2 It also has been reported that relative to mucinous cystadenocarcinoma, papillary serous cystadenocarcinoma is associated with a higher frequency of the Arg polymorphism in p53 codon 72,3 and the PTEN gene has been linked to the malignant transformation of endometriosis into endometrioid and clear cell carcinomas.4 In families carrying BRCA1/BRCA2 mutations, most tumors are nonmucinous,5–8 whereas families with hereditary nonpolyposis colorectal carcinoma (HNPCC) exhibit a slight tendency toward endometrioid neoplasms.9 These findings support the hypothesis of histology-specific etiology, the understanding of which may lead to improved management strategies for ovarian malignancy.

The current study focused on the familial association of specific histologic types of ovarian malignancy with other malignancies. Data were obtained from the nationwide Swedish Family-Cancer Database, which was updated in 2002 and now includes more than 10 million individuals and more than 1 million registered tumors. The Database is unique in terms of its size and its unbiased structure, as data on familial relationships and malignancies were obtained from registered sources with virtually complete coverage.

MATERIALS AND METHODS

The Swedish Family-Cancer Database was created in the mid-1990s by linking census information, death notifications, and the administrative family register at Statistics Sweden to the Swedish Cancer Registry.10, 11 The Database was updated at the end of 2002 to include more than 10.34 million individuals born in Sweden since 1932 and more than 810,000 invasive malignancies diagnosed since 1959. The Swedish Cancer Registry is based on compulsory reports of individual cases provided by clinicians/pathologists or cytologists and is considered to have almost 100% completeness.12 The incidence of cancer according to the Database has been validated previously.10, 13 Data on parity were complete, and data on socioeconomic index and region were based on population censuses from the years 1960, 1970, 1980, and 1990.

Four-digit diagnostic codes from the seventh revision of the International Statistical Classification of Diseases (ICD-7) and subsequent ICD classifications are available. In the current study, ovarian malignancy was indicated by ICD-7 code 175 (tumors of low malignant potential were not considered); upper aerodigestive tract malignancy included malignancies of the lip, mouth, tongue, and pharynx; rectal malignancy excluded malignancies of the anus; lung carcinoma was indicated by ICD-7 codes 162 and 163; liver malignancy included liver, biliary, and gallbladder tumors; colorectal malignancy included adenocarcinoma only; and ocular malignancy included melanoma only. Epithelial ovarian malignancy comprised papillary serous cystadenocarcinoma, endometrioid carcinoma, serous carcinoma, mucinous cystadenocarcinoma, clear cell carcinoma, and mixed müllerian tumor. Information on both parents for each offspring (daughter) included in the study allowed us to organize individuals in the Database into 3.27 million families. Women in the Database who became mothers (n = 1.29 million) were analyzed in 2 separate ways, first as daughters and then as mothers. The age of the parents was not limited, but the maximum age of daughters was 68 years, as all individuals in the Database were born after 1931. The study included 6674 cases of invasive ovarian malignancy from 3.54 million daughters and 20,974 cases of invasive ovarian malignancy from 3.27 million mothers.

Because histology has been recorded according to the College of American Pathologists Systemized Nomenclature of Medicine since 1993, the analysis of histology-specific risks considered diagnoses made between 1993 and 2000 and included 2881 cases of invasive ovarian malignancy from 3.23 million daughters and 2470 cases of invasive ovarian malignancy from 2.22 million mothers. Follow-up began on the date of birth; the date of immigration; or January 1, 1961 (or January 1, 1993, for the histology-specific analysis); whichever occurred last. Follow-up ended on the date of diagnosis of the first primary malignancy; the date of death; the date of emigration; December 31, 2000; or the date on which the patient's age exceeded the maximum age allowed in the current analysis, whichever occurred first.

Standardized incidence ratios (SIRs) were used to estimate relative familial risks. SIRs were calculated for cases in which only parents were affected (SIRparent), cases in which only siblings were affected (SIRsibling), and cases in which parents and siblings both were affected (SIRparent&sibling). Affected parents and siblings are referred to as probands. For prostate and testicular malignancies, probands included males only; for breast, cervical, endometrial, ovarian, and other female genital malignancies, probands included females only. For the remaining disease sites, probands included both males and females. Each SIR was calculated as the ratio of the number of observed cases to the number of expected cases. Expected numbers of cases were calculated based on incidence rates that were standardized according to age (5-year bands), diagnosis decade (10-year bands), socioeconomic status (6 groups), residential area (4 groups), parity (6 groups, ranging from no births to more than 5 births), and age at first parturition (5 groups, ranging from age < 20 years to age ≥ 35 years). Age was the most important standardization variable, followed by diagnosis decade and parity. The crude incidence ratio when mothers and daughters were affected with ovarian malignancy was greater than the age-standardized incidence ratio by 25.8%. Successive standardizations for diagnosis decade and parity resulted in changes of 2.4% and 0.8%, respectively. Standardizations for age at first parturition, socioeconomic index, and geographic region altered the SIRs by less than 0.3%. Because it was possible that certain variables would have larger effects on specific disease sites, incidence ratios were standardized according to all six variables.

Confidence intervals (CIs) were calculated under the assumption of a Poisson distribution.14 The issue of multiple comparisons was addressed by calculating 99% CIs. Significant associations were explored in further detail by analyzing results according to age at diagnosis; second malignancies and other malignancies occurring in affected families also were examined. The Cochran–Armitage test was used to analyze trends.

RESULTS

Numbers of cases and mean ages at diagnosis according to histology of ovarian carcinoma diagnosed before age 69 years are shown in Table 1. The mean ages at diagnosis for patients with mucinous cystadenocarcinoma and patients with nonepithelial ovarian carcinoma were significantly lower than the mean ages at diagnosis for patients with other histologic disease types.

Table 1. Mean Age at Diagnosis of Ovarian Malignancy, by Histologic Typea
Histologic typeNo. of casesMean age at diagnosis (yrs)
  • a

    Among women ages 0–68 years.

Epithelial ovarian197552.3
 Papillary serous cystadenocarcinoma83053.6
 Endometrioid carcinoma40352.0
 Serous carcinoma27252.9
 Mucinous cystadenocarcinoma28547.7
 Clear cell carcinoma14952.6
Nonepithelial ovarian16838.9
 Granulosa cell (adult)8046.1

Data on familial risk for the development of ovarian carcinoma at the most common sites are shown in Table 2. Significant values of SIRparent (i.e., those for which the 95% CI did not include 1.00) were found for ovarian, laryngeal, endometrial, and breast malignancies, as well as for all malignancies considered together; the SIRparent values for these categories ranged from 1.09 to 2.48. Significant values of SIRsibling were found for liver, ovarian, endometrial, thyroid gland, and breast malignancies, and also for all malignancies considered together; the SIRsibling values for these categories ranged from 1.17 to 2.98.

Table 2. Familial Risk of Ovarian Malignancy in Daughters, by Site of Familial Malignancya
Site/type of familial malignancyFamilial malignancy in parent onlyFamilial malignancy in sibling only
OSIRparent (95% CI)OSIRsibling (95% CI)
  • O: number of observed cases; SIRparent: standardized incidence ratio for cases in which parents (and only parents) were diagnosed with a given malignancy; SIRsibling: standardized incidence ratio for cases in which siblings (and only siblings) were diagnosed with a given malignancy; CI: confidence interval.

  • a

    Age at diagnosis for daughters and siblings was restricted to ages 0–68 years. There were no restrictions on age at diagnosis for parents.

  • b

    Ninety-five percent confidence interval does not include 1.00.

Upper aerodigestive tract320.80 (0.55–1.13)60.80 (0.29–1.76)
Esophagus201.08 (0.66–1.67)10.43 (0.00–2.49)
Stomach1100.89 (0.73–1.07)81.10 (0.47–2.17)
Small intestine141.45 (0.79–2.44)42.16 (0.56–5.59)
Colon1711.06 (0.91–1.24)221.22 (0.76–1.85)
Rectum870.92 (0.74–1.14)100.85 (0.41–1.57)
Liver761.02 (0.80–1.28)182.98b (1.76–4.73)
Pancreas690.93 (0.73–1.18)40.66 (0.17–1.71)
Larynx271.93b (1.27–2.81)20.91 (0.09–3.34)
Lung1621.01 (0.86–1.18)251.07 (0.69–1.58)
Breast2771.17b (1.04–1.32)1161.25b (1.03–1.50)
Cervix531.10 (0.82–1.44)110.75 (0.37–1.35)
Endometrium881.48b (1.19–1.83)272.01b (1.32–2.93)
Ovary1332.48b (2.08–2.94)422.81b (2.03–3.81)
Prostate3461.08 (0.97–1.20)210.93 (0.57–1.42)
Kidney710.93 (0.73–1.18)131.19 (0.63–2.04)
Urinary bladder1030.95 (0.78–1.16)110.75 (0.37–1.35)
Melanoma511.11 (0.83–1.46)270.95 (0.63–1.39)
Skin (squamous cell)720.89 (0.70–1.12)111.31 (0.65–2.35)
Nervous system631.07 (0.82–1.37)281.18 (0.78–1.71)
Thyroid gland90.53 (0.24–1.02)151.83 (1.02–3.02)
Endocrine glands371.04 (0.73–1.43)80.62 (0.27–1.24)
Non-Hodgkin lymphoma731.20 (0.94–1.51)221.36 (0.85–2.06)
Hodgkin disease161.48 (0.84–2.40)61.31 (0.47–2.87)
Myeloma471.28 (0.94–1.71)61.46 (0.53–3.20)
Leukemia681.05 (0.82–1.34)131.18 (0.62–2.02)
All sites/types21451.09b (1.04–1.14)4631.17b (1.06–1.28)

Familial risk data for epithelial ovarian malignancy and for specific histologic types of epithelial ovarian malignancy are presented in Table 3. For individuals whose mothers were diagnosed with ovarian malignancy, the SIRmother for epithelial ovarian malignancy was 2.80 (95% CI, 2.10–3.67). The histology-specific SIRs for these women could be ordered as follows: papillary serous cystadenocarcinoma (3.42) > serous carcinoma (3.18) > endometrioid carcinoma (2.73) > clear cell carcinoma (2.65 [not significant]); the differences among these SIRs were not significant. Regarding epithelial ovarian malignancies, SIRparent was significant for cases in which the parent had an ovarian or endometrial malignancy, myeloma, or melanoma. SIRparent for papillary serous cystadenocarcinoma, the most common epithelial ovarian malignancy, was significant for cases in which the parent had an ovarian malignancy or myeloma. In 2 families in which a parent was diagnosed with myeloma, 2 daughters were diagnosed with ovarian malignancy; in 1 of these 2 families, a third daughter was diagnosed with breast carcinoma at age 46 years. With respect to endometrioid carcinoma, SIRparent was significant for cases in which the parent had melanoma or malignancy of the endometrium, ovary, or prostate. In the 15 families in which a daughter had an endometrioid ovarian malignancy and her mother had an endometrial malignancy, an unusually high incidence of malignant disease was observed among fathers: 4 had prostate carcinoma, 2 had lung carcinoma, 1 had colon carcinoma, and 1 had Hodgkin disease. With respect to serous carcinoma, SIRparent was significant for cases in which the parent had Hodgkin disease or an ovarian malignancy. None of the seven siblings of women with serous carcinoma and a paternal history of Hodgkin disease had malignant disease. Among women with sisters who had been diagnosed with ovarian malignancies, SIRsister for the development of an epithelial ovarian malignancy was 2.88 (95% CI, 1.61–4.76); the SIRsister for the development of papillary serous cystadenocarcinoma (3.66) was greater than the SIRsister for the development of endometrioid carcinoma (2.73 [not significant]). With respect to epithelial ovarian malignancy, SIRsibling was also significant for cases in which a sibling had a skin (3.11) malignancy. The risk of developing nonepithelial ovarian malignancy also was analyzed; although the number of cases was limited, we found a significant SIRsibling (11.34 [95% CI, 1.07–41.69]) for women with siblings who had thyroid malignancies.

Table 3. Familial Risk for Specific Histologic Types of Ovarian Malignancy, by Site of Parental Malignancy
Site/type of parental malignancyEpithelial ovarian malignancyPapillary serous cystadenocarcinomaEndometrioid carcinomaSerous carcinomaMucinous cystadenocarcinomaClear cell carcinoma
OSIRparent (95% CI)OSIRparent (95% CI)OSIRparent (95% CI)OSIRparent (95% CI)OSIRparent (95% CI)OSIRparent (95% CI)
  • O: number of observed cases; SIRparent: standardized incidence ratio for cases in which parents (and only parents) were diagnosed with a given malignancy; SIRsibling: standardized incidence ratio for cases in which siblings (and only siblings) were diagnosed with a given malignancy; CI: confidence interval.

  • a

    Ninety-five percent confidence interval does not include 1.00.

  • For epithelial ovarian malignancy, SIRsibling was 2.88 (95% confidence interval, 1.61–4.76), and for papillary serous cystadenocarcinoma, SIRsibling was 3.66 (95% confidence interval, 1.56–7.25).

  • b

    For epithelial ovarian malignancy, SIRsibling was 3.11 (95% confidence interval, 1.41–5.92), and for papillary serous cystadenocarcinoma, SIRsibling was 3.86 (95% confidence interval, 1.22–9.08).

Colon641.14 (0.88–1.45)251.04 (0.67–1.54)191.60 (0.96–2.50)30.40 (0.08–1.20)91.20 (0.55–2.30)71.55 (0.61–3.20)
Rectum300.90 (0.60–1.28)130.91 (0.48–1.57)71.73 (0.68–3.58)51.12 (0.35–2.64)30.66 (0.13–1.97)20.74 (0.07–2.70)
Liver230.88 (0.55–1.32)121.08 (0.55–1.89)20.36 (0.03–1.32)41.12 (0.29–2.89)51.44 (0.45–3.39)  
Larynx91.81 (0.82–3.45)52.39 (0.76–5.63)21.87 (0.18–6.87)23.10 (0.29–11.39)    
Lung581.02 (0.77–1.32)331.38 (0.95–1.94)100.83 (0.39–1.53)30.40 (0.08–1.20)101.26 (0.60–2.33)20.43 (0.04–1.57)
Breast971.20 (0.97–1.46)361.07 (0.75–1.48)191.11 (0.67–1.74)141.32 (0.72–2.22)141.20 (0.65–2.02)101.54 (0.73–2.84)
Cervix191.11 (0.67–1.73)91.27 (0.57–2.42)20.55 (0.05–2.03)31.32 (0.25–3.90)31.20 (0.23–3.54)21.44 (0.14–5.28)
Endometrium391.89a (1.34–2.58)141.63 (0.89–2.74)153.41a (1.90–5.64)31.10 (0.21–3.26)  52.96 (0.93–6.96)
Ovaryb532.80a (2.10–3.67)273.42a (2.25–4.98)112.73a (1.36–4.91)83.18 (1.36–6.30)  42.65 (0.69–6.86)
Other female genital site41.12 (0.29–2.89)21.31 (0.12–4.80)        
Prostate1301.15 (0.96–1.36)581.21 (0.92–1.56)361.51 (1.05–2.09)150.99 (0.55–1.64)120.79 (0.41–1.39)90.99 (0.45–1.88)
Melanoma261.60 (1.04–2.35)60.90 (0.32–1.97)92.62 (1.19–4.99)41.96 (0.51–5.07)41.61 (0.42–4.17)21.51 (0.14–5.54)
Skin (squamous cell)c311.08 (0.74–1.54)171.37 (0.80–2.20)40.66 (0.17–1.71)51.34 (0.42–3.16)  20.86 (0.08–3.15)
Thyroid gland50.86 (0.27–2.01)20.82 (0.08–3.02)  22.57 (0.24–9.44)    
Non-Hodgkin lymphoma221.02 (0.64–1.55)60.66 (0.24–1.45)81.76 (0.75–3.48)  30.98 (0.19–2.91)42.31 (0.60–5.98)
Hodgkin disease82.16 (0.92–4.28)21.30 (0.12–4.79)  47.77a (2.02–20.10)    
Myeloma231.79 (1.13–2.69)132.38 (1.26–4.08)20.74 (0.07–2.74)31.70 (0.32–5.04)21.17 (0.11–4.29)32.92 (0.55–8.65)
Leukemia190.85 (0.51–1.33)50.53 (0.17–1.24)40.81 (0.21–2.08)31.01 (0.19–2.98)20.66 (0.06–2.44)21.12 (0.11–4.12)

Table 4 shows familial risk data for specific histologic types of ovarian malignancy when a parent was diagnosed with malignant disease before age 50 years. Regarding epithelial ovarian malignancies, SIRparent was significant for cases in which the parent had an ovarian, colonic, or endometrial malignancy. Seven families had a parental history of colon carcinoma. In one of these families, a daughter was diagnosed with an endometrial malignancy and endometrioid carcinoma synchronously, and her sister was diagnosed with an endometrial malignancy. In another of these seven families, the father had colon carcinoma, the mother had an endometrial malignancy, and their only progeny had endometrioid carcinoma. With respect to papillary serous cystadenocarcinoma, SIRparent was significant for cases in which the parent had been diagnosed with an ovarian or colonic malignancy before age 50 years. SIRparent for the development of endometrioid carcinoma was significant for cases in which the mother was diagnosed with an endometrial or ovarian malignancy. Thyroid gland malignancies in parents who were diagnosed before age 50 years were found to be associated with serous carcinoma in daughters, and an association also was found between early-onset ovarian malignancy in mothers and mucinous cystadenocarcinoma in daughters.

Table 4. Familial Risk for Specific Histologic Types of Ovarian Malignancy among Women Whose Parents Were Diagnosed before Age 50 Years, by Site of Parental Malignancy
Site/type of parental malignancyEpithelial ovarian malignancyPapillary serous cystadenocarcinomaEndometrioid carcinomaSerous carcinomaMucinous cystadenocarcinomaClear cell carcinoma
OSIRparent (95% CI)OSIRparent (95% CI)OSIRparent (95% CI)OSIRparent (95% CI)OSIRparent (95% CI)OSIRparent (95% CI)
  • O: number of observed cases; SIRparent: standardized incidence ratio for cases in which parents (and only parents) were diagnosed with a given malignancy; CI: confidence interval.

  • a

    Ninety-five percent confidence interval does not include 1.00.

Colon73.97a (1.58–8.23)45.83 (1.52–15.0)25.20 (0.49–19.1)  13.11 (0.00–17.8)  
Rectum10.97 (0.00–5.54)12.47 (0.00–14.1)        
Liver11.80 (0.00–10.3)      19.26 (0.00–53.1)  
Larynx14.59 (0.00–26.2)111.5 (0.00–65.9)        
Lung10.61 (0.00–3.48)11.59 (0.00–9.12)        
Breast151.51 (0.84–2.49)51.29 (0.41–3.04)20.94 (0.09–3.45)21.60 (0.15–5.89)52.62 (0.83–6.15)11.27 (0.00–7.29)
Cervix91.32 (0.60–2.51)31.08 (0.20–3.20)32.03 (0.38–6.02)11.13 (0.00–6.47)10.88 (0.00–5.06)11.75 (0.00–10.0)
Endometrium63.76 (1.35–8.25)23.07 (0.29–11.3)411.7a (3.06–30.4)      
Ovary125.23a (2.69–9.16)66.55a (2.36–14.3)36.09 (1.15–18.0)26.98 (0.66–25.6)  15.48 (0.00–31.4)
Melanoma42.06 (0.54–5.34)      36.78 (1.28–20.0)16.93 (0.00–39.7)
Thyroid gland22.24 (0.21–8.24)    217.5 (1.66–64.7)    
Non-Hodgkin lymphoma11.07 (0.00–6.13)12.86 (0.00–16.4)        
Hodgkin disease23.36 (0.32–12.4)14.47 (0.00–25.6)  113.3 (0.01–76.3)    
Leukemia10.81 (0.00–4.67)    16.49 (0.00–37.1)    

Recent studies have confirmed the large excess incidence of endometrial malignancy after the diagnosis of an initial ovarian endometrioid carcinoma.15 The strong association between malignancies arising at these two sites (Tables 3, 4) prompted us to analyze age-specific values of SIRmother by histologic type of ovarian malignancy for cases in which the mother had been diagnosed with an endometrial malignancy. Figure 1 shows that the values of SIRmother for papillary serous cystadenocarcinoma and clear cell carcinoma remained nearly constant; in contrast, SIRmother for endometrioid carcinoma decreased with increasing age, but the Cochran–Armitage test indicated that this trend was not significant.

Figure 1.

Age-specific standardized incidence ratios (SIRmother) by histologic type of ovarian malignancy for women whose mothers (but not sisters) were affected with endometrial malignancies. The number of cases at each data point is provided in parentheses.

DISCUSSION

To our knowledge, the current investigation represents the largest published study on familial ovarian malignancy to date, with a number of concordant cases in mother-daughter pairs that is as large as the number of concordant cases in four previous cohort studies combined.16–19 Another advantage of the current study is that it was based on registered data, whereas most family studies rely on interviews of case and control individuals. In addition, we have reported risks using mutually exclusive proband categories; doing so has both clinical and scientific advantages. Significant associations were explored in further detail by examining other malignancies in affected families. The most notable limitations of the current study involved the recording of histologic type, for which data were available only from 1993 onward, and the maximum age of progeny (68 years), which prevented direct comparisons between SIRparent and SIRsibling.

To address the issue of multiple comparisons and the limited number of observed cases, 99% CIs were calculated. The use of these intervals was found to be conservative, as certain previously demonstrated associations, such as the one between serous carcinoma in daughters and ovarian malignancy in mothers (Table 3), were not significant at the 99% confidence level. Significant findings at the 95% confidence level are evaluated in terms of biologic plausibility and consistency with previous findings in the discussion that follows. The current study examined the risk of developing specific histologic types of ovarian malignancy in cases in which relatives had been diagnosed with other malignancies. The age structure of the Database (maximum age of progeny, 68 years) did not allow validation of the results via analysis of risk ratios for other types of malignant disease in relatives of women with ovarian malignancies. For example, there were 346 daughters with ovarian malignancy whose fathers were affected with prostate carcinoma, whereas there were only 104 sons with prostate carcinoma whose mothers were affected with ovarian malignancy.

The distinction between serous cystadenocarcinoma and papillary serous cystadenocarcinoma is based on histologic verification of whether papillary features are noted in the diagnostic description. Because some studies provide specific data on papillary serous carcinoma and because our results for these two histologic types differed slightly, we analyzed serous cystadenocarcinoma and papillary serous cystadenocarcinoma separately.

The observed familial risks for developing invasive ovarian malignancy (SIRmother, 2.48; SIRsister, 2.81) agree with other reported risk data.18–24 The association between ovarian malignancies and breast malignancies is consistent with the findings of previous studies and may be attributable to BRCA1/2 mutations and to a shared response of ovary and breast to pregnancy hormones.25–28 The association between ovarian malignancies and laryngeal malignancies probably is attributable to both genetic and shared environmental effects. Parental laryngeal malignancies tended to be diagnosed at an early age: 17 of the 27 parents with laryngeal malignancies were diagnosed before age 50 years, which suggests that the association between ovarian and laryngeal malignancies possesses a genetic component. Furthermore, of the 17 women with ovarian malignancy and a parental history of early-onset laryngeal malignancy, 1 had a sibling with urinary bladder carcinoma (diagnosed at age 50 years), and 2 had siblings with lung carcinoma (diagnosed at age 50 years and age 64 years, respectively). These findings regarding early-onset tobacco-related malignancies may indicate an inherited vulnerability to genetic damage via external sources.

Since HNPCC shows colonic malignancies with carcinomas of the endometrium, biliary tract, larynx, and breast,29 the familial aggregation of ovarian malignancy with those malignancies may be attributed in some degree to HNPCC-related genes. Values of SIRsibling were significant for ovarian, liver, endometrial, and thyroid malignancies. Hemminki and Li30 reported that malignant disease in extrahepatic bile ducts was associated with ovarian malignancy, a finding that may be related to HNPCC.31 This finding is in agreement with our observation of a significant SIRsibling for women with siblings who had hepatic malignancies (Table 2). We also observed significant SIRs associated with familial endometrial malignancies. Because endometrial malignancy and ovarian malignancy are the first and second most common extracolonic manifestations of HNPCC, respectively, it is likely that the observed associations are attributable in part to HNPCC. Nonetheless, analysis of the association between endometrial malignancy and ovarian malignancy in families that are not affected by HNPCC (results not shown) suggests that additional, non-HNPCC-related effects also are involved in this association.

Specific histologic types of ovarian malignancy exhibit noteworthy differences in terms of tumor response. A recent case–control study found that oral contraceptives had a protective effect on patients with serous tumors but had the opposite effect on patients with mucinous tumors.32 In addition, genetic background may vary with varying histology; for example, most tumors observed in families carrying BRCA1/2 mutations are nonmucinous.5–8 We have been unable to locate any literature on histology-specific associations of ovarian malignancy with malignant disease occurring at nonovarian sites; such data may help to clarify etiologic differences among histologic subtypes. Analysis of families in which a parent had myeloma and a daughter had papillary serous cystadenocarcinoma revealed a strong association between these malignancies and ovarian malignancy; of the 13 parents with myeloma, 2 had more than 1 daughter with early-onset ovarian malignancy, and early-onset breast carcinoma and prostate carcinoma also were diagnosed in the 13 families. The association between myeloma and ovarian malignancy has been reported previously;33, 34 however, the etiology of multiple myeloma is unclear, and the genetic mechanisms responsible for the association between myeloma and ovarian malignancy are unknown.24, 33, 34 The association of melanoma with ovarian malignancy may be attributable to BRCA2 mutations24, 33–36 or to HNPCC, in which both melanoma and ovarian malignancy can occur. With respect to endometrioid carcinoma, SIRfather was 1.51 for women whose fathers had prostate carcinoma. The association of prostate carcinoma with breast and/or ovarian malignancy has been reported elsewhere, and it also has been found that prostate carcinoma has an increased incidence in families carrying BRCA2 mutations;35, 37 however, because the SIR of breast malignancy for women with a familial history of endometrioid carcinoma was not significant, HNPCC syndrome represents a more likely explanation for this association. Among families in which the mother had an endometrial malignancy and a daughter had endometrial carcinoma, there were four fathers who were diagnosed with prostate carcinoma; the reasons for this association remain unclear.

Although the median age at onset of sporadic epithelial ovarian malignancy ranges from 60 to 65 years, hereditary ovarian malignancies related to BRCA1/2 mutations typically are diagnosed at ages ∼55–60 years.38 In particular, approximately 83% of hereditary ovarian malignancies diagnosed before age 50 years have been attributed to BRCA1 mutations.7 The mean age at diagnosis of ovarian malignancy in women with HNPCC is ∼45 years, approximately 20 years less than the mean age at diagnosis of sporadic ovarian malignancy.9 The significant association between epithelial ovarian malignancy and colonic malignancy diagnosed before age 50 years (Table 4) supports the idea that HNPCC contributes to familial clustering of malignant disease. Analysis of other malignancies in affected families suggests that HNPCC is associated even more strongly with endometrioid histology. Although the number of histology-specific cases of ovarian malignancy was limited and although significant differences among histologic types were not found when age at diagnosis was considered, the data from the current study also may be indicative of a stronger association between endometrioid histology and endometrial malignancy than between other histologic types and endometrial malignancy.

In conclusion, the risk of developing ovarian malignancy was high for women with a maternal history of breast, endometrial, ovarian, or laryngeal malignancy and for women with a sororal history of liver, breast, endometrial, ovarian, or thyroid malignancy. The proportion of familial clustering that can be ascribed to environmental and genetic factors (i.e., those involving known genes or characterized syndromes) should be analyzed in future studies. Epithelial ovarian malignancy was associated with ovarian, endometrial, and skin malignancies and with melanoma and myeloma; papillary serous cystadenocarcinoma was associated with ovarian and skin malignancies and with myeloma; endometrioid carcinoma was associated with endometrial, ovarian, and prostate malignancies and with melanoma; endometrioid carcinoma was associated with prostate carcinoma; and serous carcinoma was associated with Hodgkin disease. In the future, histology-specific familial associations of ovarian malignancy with specific tumor sites and the differences in the strengths of these associations may guide the search for etiologic differences among histologic disease types.

Ancillary