Subsequent primary malignancies after endometrial carcinoma and ovarian carcinoma

Authors

  • Kari Hemminki M.D., Ph.D.,

    Corresponding author
    1. Department of Biosciences at Novum, Karolinska Institute, Huddinge, Sweden
    2. Division of Molecular Genetic Epidemiology, German Cancer Research Center, Heidelberg, Germany
    • Department of Biosciences at Novum, Karolinska Institute, Huddinge 141 57, Sweden
    Search for more papers by this author
    • Fax: 011 (46) 86081501

  • Lauri Aaltonen M.D., Ph.D.,

    1. Department of Medical Genetics, Biomedicum Helsinki, University of Helsinki, Helsinki, Finland
    Search for more papers by this author
  • Xinjun Li M.D., M.Sci

    1. Department of Biosciences at Novum, Karolinska Institute, Huddinge, Sweden
    Search for more papers by this author

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

Abstract

BACKGROUND

Population-based data on subsequent neoplasms after women are diagnosed with endometrial and ovarian carcinomas are limited, particularly regarding specific histologic tumor types.

METHODS

The nationwide Swedish Family-Cancer Database of 10.2 million individuals, which includes 19,128 invasive endometrial carcinomas and 19,440 ovarian carcinomas, was used to calculate standardized incidence ratios (SIRs) and 95% confidence intervals (95% CIs) for second primary carcinomas. SIRs were calculated for specific follow-up periods. Data on histopathologic types also were used.

RESULTS

An excess of subsequent malignancies after women were diagnosed with endometrial carcinoma was noted at 11 sites. The highest SIRs were recorded for synchronous or metasynchronous ovarian carcinomas (SIR, 55.77; 95% CI, 48.82–63.43) and carcinomas of the small intestines (SIR, 14.71; 95% CI, 4.64–34.59). Primary ovarian carcinoma was followed by an increased risk of developing endometrial carcinoma, and the risks of developing many other malignancies also were increased after women were diagnosed with endometrial carcinoma, including intestinal malignancies, renal cell carcinoma, bladder carcinoma, squamous cell skin carcinoma, connective tissue malignancies, and leukemia. When ovarian endometrioid histology was diagnosed synchronously with primary endometrial carcinoma, the SIR was 140; when endometrial carcinoma was the subsequent neoplasm, the SIR was 87. A small familial component was found in the cooccurrence of endometrial carcinoma and ovarian carcinoma.

CONCLUSIONS

The current data show a strong clustering of endometrial carcinomas and ovarian carcinomas, particularly involving tumors of endometrioid morphology. The patterns of second neoplasms also suggest that hereditary nonpolyposis colorectal carcinoma may contribute to the association between endometrial and ovarian malignancies. Increased risks for connective tissue tumors and leukemia may signal a response to treatment, and an increased risk for squamous cell skin carcinoma may signal a depressed immune function. Cancer 2003;10:2432–9. © 2003 American Cancer Society.

DOI 10.1002/cncr.11372

Multiple primary tumors arise because of inherited or acquired cellular lesions or deficiencies, and their causes may depend on the first primary malignancy, or they may be entirely independent.1 Because of the increasing success of therapy in achieving long-term remissions for patients with many types of malignant disease, second primary tumors are an increasing disease category.2 The 5-year relative survival rates for patients with endometrial carcinoma and ovarian carcinoma were > 80%, and > 40%, respectively, in Sweden at the beginning of the 1980s.3 Many subsequent neoplasms can be expected after women are diagnosed with endometrial carcinoma because of their favorable survival rate and because endometrial carcinoma is ranked as the third most common female malignancy. Survival from ovarian carcinoma is intermediate; although ovarian carcinoma is ranked sixth among all female malignancies, its incidence is not much lower than the incidence of endometrial carcinoma.4 Ovarian carcinoma and endometrial carcinoma are linked mechanistically to reproductive hormones, and the number of ovulatory cycles is a risk factor for both malignancies.5–7 Thus, similarities in the spectrum of second neoplasms may be expected. Most ovarian tumors are adenocarcinomas derived from the surface epithelium.8 They present in many histopathologic forms, such as cystadenocarcinoma with serous or mucinous secretions or endometrioid tumors with endometrial-like, tubular gland structures. Endometrial neoplasms also include adenocarcinomas, which may retain a well-defined, glandular structure referred to as endometrioid carcinoma.8 Based on clinical experience, endometrial tumors and ovarian tumors often are diagnosed together, and tumor histopathology is frequently endometrioid carcinoma.9 The association between endometrial tumors and ovarian tumors also has been noted in population-based epidemiologic studies, but tumor histology has not been considered.10, 11

Familial predisposition to a particular malignancy may entail an increased risk for other types of malignancies, a risk that has been noted for many cancer syndromes and for epidemiologically defined familial clustering.12–18 Ovarian carcinoma is a manifestation in families with BRCA1/BRCA2 mutations and in families with hereditary nonpolyposis colorectal carcinoma (HNPCC).19–22 Endometrial carcinoma is the most common extracolonic manifestation in families with HNPCC, and mismatch-repair defects also are common in patients with sporadic tumors.23–25 The PTEN gene also is affected commonly in patients with sporadic endometrial tumors.25–27 In addition to the known cancer syndromes, both ovarian carcinoma and endometrial carcinoma show familial clustering. The reported familial risk among first-degree relatives has ranged from 2.0 to 3.0.10, 28–38 Many of those studies also analyzed the association of ovarian carcinoma and endometrial carcinoma with other malignancies in families.10, 28, 31, 32, 34, 36–38 Some studies noted an association of these malignancies with breast carcinoma, and a few other sites have emerged in individual studies. An association between ovarian carcinoma and endometrial carcinoma was observed only in three studies, all of which analyzed young populations of women.10, 31, 38

In this report, we examine second primary malignancies after women were diagnosed with endometrial carcinoma using the nationwide Swedish Family-Cancer Database, which was updated in 2000 to include > 10 million individuals, and > 1 million registered tumors retrieved from the Swedish Cancer Registry from 1961 to 1998.39 We also considered histology and familial risks in the aggregation of endometrial carcinomas and ovarian carcinomas.

MATERIALS AND METHODS

The Swedish Family-Cancer Database was created in the middle of the 1990s by linking an administrative family register of all Swedish families to the Swedish Cancer Registry from 1961 to 1998.39 The Database includes all persons who were born in Sweden after 1931 with their biologic parents, totaling over 10.2 million individuals. It covers 760,000 first primary invasive tumors and 67,200 multiple primary tumors; the incidence rates of cancer are practically identical in the Swedish Family-Cancer Database and in the Cancer Registry.39 Information on the Database also is available at the Nature Genetics website as supplementary information.40 The completeness of cancer registration in the 1970s has been estimated at > 95% and now is considered close to 100%. The percentage of cytologically or histologically verified diagnoses of ovarian and endometrial carcinoma is close to 100%.41 The Swedish Cancer Registry has no data on treatment.

The Swedish Cancer Registry is based on compulsory notification of cancer incidence.41 A four-digit diagnostic code according to the seventh revision of the International Classification of Diseases (ICD-7) has been used since 1958, together with a code for histologic type (World Health Organization/HS/CANC/24.1 histology code). From 1993 onward, the second edition of the International Classification of Diseases for Oncology (ICD-O-2)/ICD with histopathologic data according to the Systematized Nomenclature of Medicine (SNOMED) was used; we refer to this classification as SNOMED. Leukemia codes included ICD-7 204–207 (leukemias), 208 (polycythemia vera), and 209 (myelofibrosis). According to the ICD-7 classification, lymphomas are classified as lymphomas irrespective of the site at which they occur. The classification all nonmelanoma skin malignancies includes only squamous cell skin carcinomas in the Cancer Registry (ICD code 191). Data were presented for a tumor site if > 10 women were diagnosed with a tumor at that site after they were diagnosed with ovarian carcinoma.

All tumor occurrence rates were based on the data in the Swedish Family-Cancer Database. All individuals, parents, and offspring were included in the analysis of second events. Even synchronous second malignancies were included, and follow-up was divided into three periods, allowing assessment of the effect of follow-up. Follow-up was started at the time a patient was diagnosed with primary endometrial or ovarian carcinoma and was terminated when she was diagnosed with a second malignancy, died, or emigrated or on the closing date of the study (December 31, 1998). Standardized incidence ratios (SIRs) were calculated as the ratio of observed to expected numbers of diagnoses. The expected numbers were calculated from 5-year age specific, gender specific, tumor type specific, period (10 intervals) specific, and socioeconomic status (6 groups) specific standard incidence rates. Ninety-five percent confidence intervals (95% CI) were calculated assuming a Poisson distribution

RESULTS

The Swedish Family-Cancer Database covered years 1961–1998 from the Swedish Cancer Registry and included 19,128 primary endometrial carcinomas and 19,994 primary, invasive ovarian carcinomas (Table 1). The median age at the time of onset was 61 years for women with endometrial carcinoma and 58 years for women with ovarian carcinoma. Follow-up after the first primary neoplasm was diagnosed accumulated 255,000 person-years for endometrial carcinoma and 157,000 person-years for ovarian carcinoma. Altogether, 2413 second primary neoplasms were diagnosed after endometrial carcinoma, and 1275 second primary neoplasms were diagnosed after ovarian carcinoma. Few concordant diagnoses were noted, because the treatment usually involved removal of the whole organ. Concordant sites are not included in the data discussed below.

Table 1. Number of Second Primary Tumors after Endometrial and Ovarian Carcinoma
CharacteristicFirst primary tumor
EndometrialOvarian
No. of first primary tumors19,12819,440
Person-years of follow-up255,211156,877
Second primary tumors24131275
 Endometrial1296
 Ovarian27366

SIRs for second primary malignancies after women were diagnosed with endometrial carcinoma are shown in Table 2. The data are presented according to follow-up. The overall SIR was 5.59 during the first year of follow-up, 1.30 during years 1–10 of follow-up, and 1.43 after 10 years of follow-up, for an overall SIR of 1.54. The SIR for ovarian carcinoma was 55.77, and the SIR for carcinoma of the small intestines was 14.71 during the initial year of follow-up. When endometrial carcinoma was diagnosed before age 40 years, the SIR for ovarian carcinoma was 800 during the first year of follow-up. Upon closer inspection, we noted that almost all second primary ovarian carcinomas were diagnosed synchronously with endometrial carcinomas. Overall, the incidence of second primary malignancies at 11 sites increased, with ovarian carcinoma (SIR, 3.16) and carcinoma of the urinary bladder (SIR, 2.35) increasing more than malignancies at other sites. The incidence of intestinal carcinomas increased to an SIR of about 2.0. The SIRs for some sites, such as genitalia, kidney, and bladder, were high in the initial follow-up, but they also increased toward the end of follow-up. Squamous cell carcinoma of the skin and connective tissue tumors also were in excess. Among connective tissue tumors, there were no great differences between tumor sites or histologies for tumors that appeared early or late during follow-up. Myxosarcomas and undifferentiated sarcomas emerged relatively late after endometrial carcinoma, although their numbers were too few for statistical treatment.

Table 2. Standard Incidence Ratios for Second Primary Tumors After Endometrial Carcinoma
Second primary tumorFollow-up (yrs)
< 11–10> 10All
OSIR95% CIOSIR95% CIOSIR95% CIOSIR95% CI
  • O: observed; SIR: standardized incidence ratio; 95% CI: 95% confidence interval.

  • a

    95% confidence interval does not include 1.00.

Stomach0250.840.54–1.25240.990.64–1.48490.870.64–1.15
Small intestine514.714.64–34.59a81.870.80–3.7130.880.17–2.61162.001.14–3.25a
Colon182.871.70–4.55a1231.571.30–1.87a1432.071.74–2.44a2841.841.64–2.07a
Rectum92.991.36–5.70a631.681.29–2.15a712.271.77–2.87a1431.991.68–2.35a
Liver51.810.57–4.25230.650.41–0.98a451.501.10–2.01a731.070.84–1.35
Pancreas20.840.08–3.08331.080.75–1.52351.350.94–1.88701.190.93–1.51
Lung51.350.43–3.17491.060.78–1.40411.160.83–1.57951.110.90–1.36
Breast512.561.91–3.37a3321.401.25–1.56a1941.221.05–1.41a5771.391.28–1.51a
Ovary23255.7748.82–63.43a340.670.46–0.94a70.220.09–0.46a2733.162.80–3.56a
 First endometrial cancer diagnosed   before age 40 yrs8800.00341.67–1584.04a13.450.00–19.7711.370.00–7.85109.714.62–17.92a
Other female genitals34.350.82–12.87101.180.56–2.17202.701.65–4.18a331.991.37–2.80a
Kidney83.571.53–7.07a401.421.02–1.94a371.761.24–2.42a851.651.32–2.05a
Urinary bladder42.160.56–5.59441.901.38–2.55a582.882.19–3.73a1062.351.92–2.84a
Melanoma10.480.00–2.77311.250.85–1.78251.440.93–2.13571.290.98–1.67
Skin (squamous cell)20.960.09–3.52411.611.16–2.19a421.511.09–2.05a851.541.23–1.90a
Nervous system31.260.24–3.73331.140.79–1.61251.360.88–2.01611.230.94–1.58
Thyroid gland0131.440.77–2.4840.710.19–1.85171.110.64–1.78
Endocrine glands20.960.09–3.54301.150.77–1.64291.541.03–2.21a611.300.99–1.67
Connective tissue12.170.00–12.46122.151.10–3.76a81.860.79–3.68212.031.25–3.11a
Non-Hodgkin lymphoma41.820.47–4.70351.280.89–1.78220.930.58–1.41611.150.88–1.47
Myeloma21.800.17–6.63191.350.81–2.11141.150.63–1.94351.280.89–1.78
Leukemia52.690.85–6.32341.481.03–2.07a211.100.68–1.68601.371.04–1.76a
All3625.595.03–6.20a10321.301.22–1.38a8681.431.34–1.53a22621.541.48–1.61a

A similar analysis was carried out for second primary malignancies after women were diagnosed with ovarian carcinoma (Table 3). The overall SIR was 2.95 during the first year of follow-up, 1.28 during years 1–10 of follow-up, and 1.45 after 10 years of follow-up, for an overall SIR of 1.47, almost identical to the data on endometrial carcinoma for all but the initial follow-up period shown in Table 2. The SIR was 12.74 for endometrial carcinoma and 13.79 for small intestinal tumors during the initial year of follow-up. Practically all endometrial carcinomas were diagnosed synchronously with ovarian carcinoma. Their risk was independent largely on the age at which women were diagnosed with ovarian carcinoma. Overall, the incidence of second primary malignancies increased at 12 discordant sites after women were diagnosed with ovarian carcinoma, and most of those sites (n = 8 sites) were the same sites at which the incidence of second primary malignancies increased after women were diagnosed with endometrial carcinoma. The exceptions were carcinomas of the breast and genitalia, which were significant sites only after endometrial carcinoma, and carcinomas of the pancreas and thyroid, which were significant sites only after ovarian carcinoma. However, at all of these sites, the SIRs were greater than unity. Connective tissue neoplasms had the greatest overall excess SIR of 3.40. The numbers of patients (n = 18 women) were too few for statistical analysis, but myxosarcomas and undifferentiated sarcomas emerged relatively late after women were diagnosed with ovarian carcinoma.

Table 3. Standardized Incidence Ratios for Second Primary Tumors After Ovarian Carcinoma
Second primary tumorFollow-up (yrs)
< 11–10> 10All
OSIR95% CIOSIR95% CIOSIR95% CIOSIR95% CI
  • O: observed; SIR: standardized incidence ratio; 95% CI: 95% confidence interval.

  • a

    95% confidence interval does not include 1.00.

Stomach73.241.28–6.71a181.250.74–1.97181.761.04–2.79a431.601.16–2.16a
Small intestine413.793.59–35.67a62.901.04–6.35a21.310.12–4.81123.081.59–5.41a
Colon295.523.70–7.94a501.351.00–1.78a571.931.46–2.51a1361.901.59–2.24a
Rectum20.800.08–2.93341.911.32–2.67a312.261.54–3.22a671.971.53–2.51a
Liver52.090.66–4.92160.960.54–1.55151.170.65–1.94361.130.79–1.56
Pancreas52.450.77–5.77241.671.07–2.49191.721.03–2.69a481.751.29–2.32a
Lung10.330.00–1.91271.210.80–1.76241.430.91–2.12521.230.92–1.62
Breast191.090.65–1.701341.050.88–1.25991.180.96–1.432521.100.97–1.25
Endometrium5312.749.54–16.67a190.630.38–0.98a241.190.76–1.78961.761.43–2.15a
Other female genitals051.220.38–2.8772.170.86–4.50121.520.78–2.66
Kidney147.223.93–12.14a191.400.84–2.19111.180.58–2.11441.771.29–2.38a
Urinary bladder21.310.12–4.81211.931.19–2.95a192.191.32–3.43a421.991.43–2.69a
Melanoma52.690.85–6.32191.340.81–2.10111.150.57–2.07351.370.95–1.91
Skin (squamous cell)31.840.35–5.45191.640.98–2.56181.560.92–2.47401.611.15–2.20a
Nervous system62.801.01–6.14a120.780.40–1.36131.350.72–2.32311.140.77–1.62
Thyroid gland22.740.26–10.08101.910.91–3.5341.320.34–3.42161.781.02–2.90a
Endocrine glands52.820.89–6.64181.370.81–2.1670.770.31–1.60301.250.84–1.79
Connective tissue12.500.00–14.33103.471.65–6.41a73.481.38–7.22a183.402.01–5.39a
Non-Hodgkin lymphoma10.550.00–3.17161.210.69–1.97181.711.01–2.71a351.370.95–1.91
Myeloma071.060.42–2.1940.780.20–2.02110.870.43–1.56
Leukemia21.260.12–4.63343.052.11–4.26a121.430.74–2.51482.271.68–3.02a
All1662.952.52–3.44a5181.281.17–1.40a4201.451.31–1.59a11041.471.39–1.56a

To study the effect of specific histopathology between synchronous endometrial carcinoma and ovarian carcinoma, we used SNOMED codes, which have been used systematically since 1993 in the Swedish Cancer Registry. The analysis in Table 4 was based on SNOMED coding for ovarian tumors and an older histology coding for endometrial carcinoma, because SNOMED provided no further details on endometrial carcinoma. When endometrial adenocarcinoma was the first event, 58 ovarian malignancies were diagnosed synchronously or metasynchronously, for an SIR of 45.67, and 21of those tumors were endometrioid carcinomas, with an SIR of 140. In addition, papillary cystadenocarcinomas, which are the most common type of ovarian tumor, and mucinous cystadenocarcinomas were in excess. Reversing the order, with ovarian endometrioid carcinoma the first primary malignancy, yielded an SIR of 86.67 for second primary endometrial adenocarcinoma.

Table 4. Standardized Incidence Ratios for Second Primary Endometrial Carcinoma after Ovarian Carcinoma and for Second Primary Ovarian Carcinoma after Endometrial Carcinoma by Histopathology during the First Year of Follow-Up
First primary tumorSecond primary tumorOSIR95% CI
  1. O: observed; SIR: standardized incidence ratio; 95% CI: 95% confidence interval.

EndometrialOvarian (all)5845.6734.67–59.07
 Adenocarcinoma Endometrioid21140.0086.52–214.36
 Adenocarcinoma Papillary cystadenocarcinoma923.0810.46–44.00
 Adenocarcinoma Mucinous cystadenocarcinoma654.5519.63–119.51
 Adenocarcinoma Other1458.3331.78–98.13
OvarianEndometrial (all)2920.7113.86–29.78
 Endometrioid Adenocarcinoma1386.6745.96–148.63
 Seropapillary cystadenocarcioma Adenocarcinoma410.262.67–26.52
 Other Adenocarcinoma625.009.00–54.78

We probed the possible familial effects in the aggregated population with endometrial and ovarian carcinoma by searching for family members, either mothers or sisters, who were diagnosed with either of these malignancies and then determining the risk for developing second primary malignancies. The analysis was extended to cover the whole follow-up period, but only data on the first year of follow-up are shown in Table 5, because all relevant findings could be narrowed to this period. When a family member was diagnosed with endometrial carcinoma, the risk of the first-degree relative for developing a second primary ovarian carcinoma after endometrial carcinoma was 125.00. In addition, colorectal carcinoma was in excess (SIR, 6.38). The risk for developing second primary endometrial carcinoma after ovarian carcinoma was 50.00. When a family member was diagnosed with ovarian carcinoma, her daughter or sister experienced an increased risk of ovarian carcinoma after endometrial carcinoma (SIR, 40.00). No small intestinal malignancies were found as second primary neoplasms among patients who had this familial risk component.

Table 5. Standardized Incidence Ratios for Subsequent Carcinoma in Women with a Familial History of Endometrial or Ovarian Carcinoma during the First Year of Follow-Up
Family historyOSIR95% CI
  • O: observed; SIR: standardized incidence ratio; 95% CI: 95% confidence interval.

  • a

    95% confidence interval does not include 1.00.

Family member had endometrial carcinoma   
 First primary endometrial carcinoma   
  Second primary colorectal tumor36.381.20–18.89a
  Second primary breast tumor16.250.00–35.83
  Second primary ovarian tumor5125.0039.44–294.03a
 First primary ovarian carcinoma   
  Second primary colorectal tumor0
  Second primary breast tumor0
  Second primary endometrial tumor250.004.71–183.88a
Family member had ovarian carcinoma   
 First primary endometrial carcinoma   
  Second primary colorectal tumor0
  Second primary breast tumor17.140.00–40.94
  Second primary ovarian tumor240.003.77–147.10a
 First primary ovarian carcinoma   
  Second primary colorectal tumor0
  Second primary breast tumor14.350.00–24.92
  Second primary endometrial tumor0

We listed all family members and their malignancies using women with a double primary endometrial and ovarian carcinoma (or ovarian and endometrial carcinoma) as probands (data not shown). In these 369 families, there were 1202 offspring ages birth–66 years. None of the families fulfilled the Amsterdam criteria for HNPCC of including three family members with colorectal carcinoma; however, three families did fulfill the wider Bethesda criteria.42, 43 No sibling with carcinoma of the small intestinal was found among the offspring of these women.

DISCUSSION

The increased occurrence of second primary malignancies after women are diagnosed with an initial primary malignancy may result from 1) intensive medical surveillance after the first diagnosis, 2) therapy-induced exposure to X-rays and antineoplastic agents, and 3) shared environmental and hereditary factors between the first and second primary malignancies.1 In the current study, we observed examples of each of these possibilities. The high risk of developing second neoplasms during the first year of follow-up most likely is due to lead-time bias. Diagnoses of endometrial, ovarian, and most other carcinomas have been verified histologically and cytologically in almost 100% of the tumors in the Swedish Cancer Registry, suggesting that diagnostic misclassification should not be an issue in the current study. In an ad-hoc study of 209 multiple primary tumors from the Swedish Cancer Registry, a reevaluation showed that 98% of second malignancies were classified correctly.44 However, diagnostic accuracy does not exclude the presence of surveillance bias during the treatment for the first primary malignancy and the ensuing follow-up period. Treatments for patients with endometrial carcinoma and ovarian carcinoma usually involve major surgery, which leads to a careful examination of the patient and diagnosis of tumors at a variety of sites.

Radiotherapy and chemotherapy agents cause DNA damage, which may cause the development of a subsequent neoplasm. These therapeutic modalities have been used more to treat women with ovarian carcinoma than to treat women with endometrial carcinoma: The increase in connective tissue tumors after a diagnosis of ovarian carcinoma to an SIR of 3.40, compared with an SIR of 2.03 after a diagnosis endometrial carcinoma, is consistent with radiotherapy-related effects.45, 46 Connective tissue tumors presented as many histologic types, although there were too few patients for a proper statistical treatment; however, myxosarcomas and undifferentiated sarcomas were the types of second primary tumors that emerged during the late follow-up period. The incidence of leukemias also increased more after ovarian carcinoma (SIR, 2.27) than after endometrial carcinoma (SIR, 1.37), suggesting that the increases were the result of therapy for the first neoplasm. Therapy also may have contributed to the increased occurrence of renal and bladder carcinomas, although there may have been other reasons, as discussed below. It has been well established that immunosuppressed patients who undergo kidney transplantation experience a high risk for squamous cell carcinoma of the skin and non-Hodgkin lymphoma, in addition to many other neoplasms.47–49 In these patients, tumors may appear within a short period after the onset of therapy, depending on the severity of their immunosuppression.49 Non-Hodgkin lymphoma was not in excess in the current study, but squamous cell skin carcinoma was increased to an equal extent after endometrial carcinoma (SIR, 1.54) and ovarian carcinoma (SIR, 1.61), suggesting that the result was not fortuitous. Patients may have suppressed immune function due to therapy, particularly if they have undergone a large surgical procedure, as suggested in a study of patients with skin malignancies and non-Hodgkin lymphomas as second neoplasms after any first primary malignancy.50

Breast carcinoma was in excess after endometrial carcinoma, but not after ovarian carcinoma. Hormonal factors may explain these findings. These three malignancies also show familial aggregation, although the familial association between breast and ovarian carcinomas is stronger than that between breast and endometrial carcinomas.38 HNPCC may explain some of the excess risks of second intestinal, stomach, kidney, and bladder carcinomas, in agreement with previous studies.18, 51–53 HNPCC also may contribute to the association between endometrial and ovarian carcinomas, although the main reason for this clustering appears to be elsewhere. A diagnosis of endometrial carcinoma before age 40 years substantially increased the risk for second ovarian carcinoma, but the diagnostic age of patients with ovarian carcinoma did not influence the risk for subsequent endometrial carcinoma. There was a small familial component to this clustering, but few women with colorectal carcinoma were identified in these familial clusters (Table 5). None of the 369 families with double primary endometrial and ovarian carcinomas fulfilled the Amsterdam criteria for HNPCC, and only three families fulfilled the more permissive Bethesda criteria.42, 43 The caveat is that carriers of the rare MSH6 germline mutations appear to show a high risk of endometrial carcinoma but a low risk of colorectal carcinoma.24, 54 Moreover, the clustering of endometrial and ovary carcinomas was not limited to, but was vastly strengthened by, the endometrioid histology of ovarian carcinoma, which is not known as pathognomonic to HNPCC, although it may be somewhat more common in patients with HNPCC.12, 55 Ovarian and endometrial carcinomas also are manifestations in Peutz–Jeghers syndrome; however, the lack of women with carcinoma of the small intestines in the families from the current study suggests that this syndrome is not involved.56

Based on previous epidemiologic studies from the 1970s and 1980s, the risk for synchronous endometrial carcinoma and ovarian carcinoma reportedly increased 2–10 times, and it has been suggested that shared etiologies are the basis for the cooccurrence of these malignancies.57 In a recent study, the risk of ovarian carcinoma increased 17-fold during the first 4 years after women were diagnosed with endometrial carcinoma, but that risk was limited to women who were diagnosed with endometrial carcinoma before age 50 years. In the current study, the risks were greater for women with endometrial carcinoma compared with the risks reported for women with ovarian carcinoma (56-fold) and endometrial carcinoma (13-fold): When ovarian endometrioid histology was involved, the SIRs were 140 and 87, respectively. However, any ovarian histology conveyed a risk > 10-fold. Zaino et al. recently discussed the limited evidence based on analysis of polymorphic DNA markers, X chromosome inactivation, mutations, and allelic loss, suggesting that double primary endometrial and ovarian carcinomas are independent primary tumors.9 If, indeed, synchronous endometrial carcinoma and ovarian carcinoma are independent tumors, then one likely common mechanism may be the response to pregnancy hormones, which also may explain the increased risks of carcinomas of the breast and genitalia after women are diagnosed with endometrial carcinoma. Overall, the pattern of increased risk for subsequent tumors was remarkably similar for first primary endometrial carcinoma and first primary ovarian carcinoma. Of the 20 discordant sites that were analyzed, SIRs for 8 sites increased significantly after both endometrial carcinoma and ovarian carcinoma, providing further evidence for the etiologic similarity of endometrial carcinoma and ovarian carcinoma. Heritable causes may contribute, and an association between endometrial carcinoma and ovarian carcinoma has been observed in previous studies.10, 31, 38

The current data indicate a strong coupling of endometrial carcinoma with ovarian carcinoma in which a relatively rare form of ovarian carcinoma, the endometrioid type, commonly was involved. Hormonal mechanisms may underlie the shared susceptibility for double primary carcinomas. The patterns of second primary malignancies also suggest that there was a contribution by HNPCC to the association between endometrial carcinoma and ovarian carcinoma. Increased risks for connective tissue tumors and leukemia may signal a response to treatment, and an increased risk for squamous cell skin carcinoma may indicate an immunologic disturbance.

Ancillary