BRCA mutations and sporadic cancers: Primary Observational Data
Studies used are numbered and summarized in and . contains population studies (numbered 1-17) which include data on the incidence of cancers other than breast or ovarian cancer in BRCA mutation carriers or populations likely to be enriched in mutation carriers. reports the measured incidence of BRCA1 or BRCA2 mutations in families of patients presenting with a particular cancer. (1
Epidemiologic studies that include data on the risk of “other” cancers.
Studies of patients with individual cancers other than breast or ovarian tested for association with BRCA mutations.
lists cancers that occurred at a statistically significant greater frequency relative to controls if BRCA function is impaired in at least some subgroups of mutation carriers within individual studies (column 6): Colon cancer, colorectal cancer, gall bladder/bile duct cancer, liver cancer, stomach cancer, malignant melanoma, esophageal cancer, myeloid leukemia, lung/bronchus cancer, laryngeal cancer, Hodgkin's disease, cancer of the uterine body and cervix, testicular cancer, cancer at multiple sites including pituitary and peritoneum, all cancers except breast ovary, non melanoma skin; all cancers in females other than breast, ovarian and non-melanoma skin; any second non-breast primary cancer, and unknown primary site cancer.
plots the statistically significant relative risks extracted from the 17 studies in . The plot shows that the increased incidences of cancers is not consistent among studies and vary in studies of different population subgroups. For example, 6 of 17 studies show statistically significant elevated risks for stomach cancers in some subgroups, and 4 studies show elevated risk for pancreatic cancer. Even different groups within the same study show variation. For example, comparisons of pancreatic cancer risks among different individual groups in the study of Bermejo and Hemminki(9
) shows wide variation (), but their aggregated data includes 1 as part of the confidence interval. Increased risks for esophageal, lung, NHL, malignant melanoma, buccal cavity/pharynx, squamous skin and gallbladder/bile duct cancer are each represented in only 1 study. (Gallbladder/bile duct cancer was plotted near the results for liver cancer in )
Figure 1 A. Significant relative risks for cancers other than breast/ovarian in mutation carriers or likely mutation carriers found in various subgroups by different studies. Sites are arranged in descending frequency that statistically significant cancers were (more ...)
Eleven studies reported sufficient information to plot incidences of cancers lumped together in categories called “all cancers except breast, and ovarian” or “all cancers except breast, ovarian and non-melanoma skin cancer”. shows results for the reported incidences of cancers with statistically significant increases in relative risk of cancer at sites beyond breast or ovary. Results for different groups within each study were pooled together. shows that five of the studies do not include 1 in the confidence interval and relative risks from all of the studies are greater than 1 with p values <0.05. Combined risk was calculated as 1.52 (CI 1.21-1.93)
Most studies in and Figs and find a statistically significant increased incidence for at least one cancer other than breast or ovarian cancer in BRCA gene carriers. This increased incidence crosses different methodologies and populations. This is consistent with the idea that the cancers occur sporadically but at moderately elevated frequency.
are forest plots of risks for stomach/gastric, pancreatic, prostate and colorectal cancer, respectively. These all show elevated relative risk. Surprisingly, the largest increase in relative risk was from stomach/gastric cancer () RR=1.69 [1.21-2.38] followed closely by pancreatic cancer () RR=1.62 [1.31-2.00]
A Relative risks of Stomach/Gastric Cancer based on data in different population studies. B. Prostatic cancer relative risks C. Relative risks for pancreatic cancer. D. Relative risks for colon and for rectal cancer
In , studies 1-3 were from the Breast Cancer Linkage Consortium, a worldwide cooperative network of scientists and physicians with a major interest in inherited breast and ovarian cancer. Study 2 contains a listing of cancers as “all cancers except breast, ovary, non-melanoma skin and prostate cancer. ” The odds ratio for these other, presumably sporadic cancers is 1.90. Study 3 gives a relative risk of 1.34 [1.19-1.51] of all cancers except breast, ovary, non-melanoma skin, prostate and pancreatic cancer.
Study 2 examined data from a large number of BRCA1 carriers. Its increased statistical power detected moderate increases in numerous abdominal cancers ().
Brose and colleagues (Study 4, ) found that colon, pancreatic, and gastric cancer occurred more frequently than in the general population(4
). The cumulative age-adjusted risk of colorectal cancer in their study was a statistically significant twofold increase (P<.05). They also estimate a threefold increase in pancreatic cancer risk and a fourfold increase in gastric cancer risk. The cumulative age-adjusted lifetime risk of any cancer other than breast and ovarian diagnosis in this group of BRCA1 mutation carriers was 13.8% (95% CI 10.7% to 16.9%). Of note, the overall risk of cancers other than of the female breast and ovary was statistically significantly higher for men than women, with cumulative age-adjusted lifetime risks of 26.1% (95% CI 17.5% to 34.6%) and 10.3% (95% CI 7.2% to 13.3%), respectively. (4
A study of families related to BRCA1 probands from Sweden ( Study 5) reported nearly 6 times the risk of gastric cancer in females. In a Japanese study (not listed in ), a series of gastric cancer patients diagnosed before age 35 documented allelic losses flanking BRCA1 in 12 of 27 cases (44%), suggesting a possible mechanistic link between BRCA1 and the development of gastric cancer.(34
Johansson and colleagues ( Study 5) also found about a 6 fold increase in squamous cell skin cancer in males from BRCA1 families. Invasive cervical cancer was increased in females from BRCA2 families and prostate cancer was of borderline significance. ( Study 5). Study 5 used expanded pedigrees based on mutation testing only for probands, so Study 5 would also include mutations elsewhere in BRCA pathways. Thus effects such as amplification of EMSY genes are automatically included. EMSY overexpression may occur at least as often BRCA mutations. A disadvantage of not typing mutations is that mutation carriers are still only a minority of the population and their risk values are diluted by including individuals with normal BRCA pathways. This makes it more difficult to show increased risks for other cancers
The work by Evans and colleagues (study 6, ) also did not directly type BRCA gene mutations. Evans and colleagues used data from the Thames cancer registry to identify women under age 50 who had additional primary cancers after breast cancer. 32,799 women were diagnosed with breast cancer under the age of 50 between January 1, 1961 and December 31, 1995. Women were censored from the study at clear cut-off dates. The mean follow-up period was 7.5 years. 1448 cases had multiple additional tumors within the 9 sites monitored. 1389 had 2 tumors, 57 had 3 and 2 had 4 tumors. Based on Myriad Genetics tables, the odds that a woman related to one of these individuals would have a BRCA mutation is only 7.3%. 318 had multiple tumors under age 50 and were included in the study. The figure of about 7.3% is far above that for the general population but still omits those with mutations elsewhere in BRCA pathways.
The early age of onset of breast cancer is considered characteristic of inherited cancers. Evans and colleagues (6
) also provide data for women older than 50 and this permits useful comparisons. Comparing women with breast cancer diagnosed under age 50 with those diagnosed over age 50, Evans and colleagues noted the younger age group had an increased risk of a second cancer at site other than breast. and Figures and shows the risk for a number of second cancers is elevated and some of these rise to statistical significance.
Harvey and Brinton (, study 7) in a 1985 study listed primary cancers after a first breast cancer in Connecticut women under age 45. The authors noted a significant relationship between breast and subsequent colon cancer. Second cancers were much more numerous in this group of nearly 7000 younger women than in older women. They found a significant downward trend for risk of colon cancer (RR=1.6, 1.3, 1.1) and rectal cancer (RR=1.9,1.1,1.0) with age. This downward trend applied to ages<45, 45-54 and 55+. Similarly Teppo and colleagues ( study 8) found that colon cancer in women after breast cancer was diagnosed before age 50 occurred almost twice as often as it did in women diagnosed with breast cancer after age 70.
Bermejo and Hemminki used the Swedish Cancer Registry families to find families for which data existed for three generations. From these families, they selected individuals eligible for BRCA mutation testing. The results show an increased incidence of cancers in organs other than breast and ovary in a large population eligible for BRCA mutation testing. Cancers with increased standardized incidence ratios included prostate cancer, pancreatic cancer and pancreatic cancer before age 50. Perhaps eye cancer and stomach cancer by age 70 also achieved statistical significance ( Study 9.)
Including this study was not without complications. There was no follow-up after the first additional cancer so Bermejo and Hemminki would overlook slower growing cancers such as colorectal cancers in favor of ovarian cancer and other faster growing tumors. Ovarian and fallopian cancers were counted as an additional cancer, and the risk for these tumors varied from a few- to many-times greater than the risk for other cancers. In most groups, the odds were high that a diagnosis of ovarian cancer would have terminated follow-up, especially in the group defined by the presence of ovarian cancer in the family. Increased risk for stomach cancer did not rise to statistical significance until age 70 in one group. It is thus likely that including data of Bermejo and Hemminki leads to an underestimate of risk for other cancers, particularly slower growing tumors such as colon cancer. Although nothing was excluded, excluding even some of the data of Bermejo and Hemminki (such as the group with ovarian and breast cancer in the family) would markedly increase the risk for colon cancer.
Study 10 used a different population and methodology but still supports the increased incidence of other cancers in BRCA mutation carriers. Shih and colleagues (Study 10) found that 98 women with breast cancer who reported at least one other primary cancer in themselves or in a relative with breast cancer were twice as likely to be mutation carriers as women with breast cancer who did not report a second primary tumor. Their data support a general cancer susceptibility in those with BRCA1 and BRCA2 mutations and a heightened clinical awareness in dealing with mutation carriers (10
Easton and colleagues(11
) (, Study 11) studied 2 large breast cancer families with tumors linked to BRCA2 mutations. They found statistically significant increases in laryngeal cancer, prostate cancer and ocular cancer. In other studies, there are implications that ocular cancer is elevated, but the disease is so rare that it is difficult to achieve statistical significance.
Moslehi and colleagues(12
) (, Study 12) found relatives of Ashkenazi Jewish female BRCA2 carriers to be at greater lifetime risk (to age 75) for cancer of any type than the relatives of the BRCA1 carriers (46.3% vs. 34.9%). In men younger than 65 years, the risk was significantly higher for cancer (21.4% vs. 4.4%;) and this was attributed largely to an excess of prostate, pancreatic, and colon cancers observed in male relatives of BRCA2 carriers at that age. In first degree relatives of Jewish ovarian cancer patients, greater risks for pancreatic cancer, prostate cancer and cancers in which the primary site was unknown were statistically significant.
In a population series of 649 women with ovarian cancer, Risch and colleagues(13
) (, Study 13) examined BRCA2 mutation location relative to colorectal, stomach, pancreatic and prostate cancer in family members. For probands carrying BRCA2 mutations, colorectal cancer in family members occurred only when mutations were within the ovarian cancer cluster region of exon 11 of the BRCA2 gene (now defined as nucleotides 4075 to 6503). Statistically significant increases in risk for colorectal, stomach, pancreatic or prostate cancer (as well as for ovarian cancer) occurred for these mutations.
Aretini and colleagues(14
) (, Study 14) as part of the Italian Consortium for Hereditary Breast and Ovarian Cancer found that cancers other than breast or ovarian were significantly elevated in mutation carriers. The presence of prostate or pancreatic cancer in a family was correlated with the presence of ovarian cancer in BRCA2 mutation carriers.
Colon cancer is highly age dependent, rising from an incidence of 0.23% at age 50 to 5.6% at age 80. Because it was a peripheral issue, the publication did not list data to estimate the rate of colon cancer in a control group, so colon cancer incidence data had to be excluded. Nonetheless the authors mentioned finding an increased incidence in BRCA families of pancreatic, prostate, and gastrointestinal cancers.
Streuwing and colleagues(15
) (, Study 15) recruited over 5300 Jewish people in the Washington DC area and identified 120 carriers of one of the three BRCA founder mutations. The family histories of the carriers included increased percentages of numerous cancers other than breast or ovarian cancer. Even though the numbers of carriers were low, there were elevations in prostate, lung, multiple myeloma, and Hodgkins disease that reached sufficient statistical significance to warrant further investigation.
Goldgar and colleagues(16
) (, study 16) found a highly significant statistical association between familial occurrences of breast, colon and prostate cancers and between breast and thyroid cancers. Berman and colleagues(17
) in a small study ( study 17) noted significant cancer histories in 8 BRCA2 mutation carriers.
The occurrence of BRCA1 or BRCA2 mutations in subgroups of patients with other cancers
Mutations are prevalent in familial forms of both pancreatic and prostate cancers where multiple relatives are affected. Carriers of BRCA1 or BRCA2 mutations represent up to 4 to 7% of all unselected pancreatic cancer patients and 10% of those from Ashkenazi Jewish families. In the Swedish cancer registry study of probable BRCA mutation carriers, the incidence of early onset pancreatic cancer was elevated up to 6.54 fold(9
With considerable variability, studies of patients presenting with pancreatic cancer or prostate cancer have concluded that BRCA1/BRCA2 mutation carriers are at increased risk for these cancers (, studies 18-20). The studies in generally support those in . Because of the low percentages of mutation carriers in the general population, only low numbers of mutation carriers were involved in the studies in , ranging from a high of 64 to a low of 3 (See column 4).
in BRCA2 mutation carriers is consistent with the possibility that BRCA2 loss sometimes promotes the malignant progression of existing lesions (). Familial pancreatic cancer marked by BRCA2 mutations occurs 8-10 years sooner than sporadic disease (9
). One explanation for the higher prevalence of pancreatic carcinoma in families with >=2 affected first-degree BRCA2 mutation carriers is the early biallelic inactivation of the BRCA2 gene (35
). The risk for sporadic pancreatic cancer in BRCA2 mutation carriers rises to 3.5 (18
). Hahn and colleagues studied 64 patients (37 men and 27 women) with pancreatic cancer among 26 families. 4 families had 4 members with pancreatic cancer, 4 families had 3 affected members, and 18 families had 2 affected members. 3 families had 3 affected generations, 16 families had two affected generations, and 7 families had only one affected generation. No families were Ashkenazi Jews. (, Study 18)
Murphy and colleagues(19
) (, Study 19) found increased risk of both sporadic and familial pancreatic cancer in BRCA2 mutation carriers. Within these kindreds, both Hahn and colleagues (18
) and Murphy and colleagues (19
) found a variety of additional cancers within the families they studied but risks were not calculated.
These two reports (18
) suggest that some familial pancreatic cancers are caused by BRCA2 germ-line mutations. Of note, these pancreatic cancer families generally may not show an increased incidence of breast and ovarian cancer. Other pathways not mediated via BRCA proteins are also involved and apparently generate pancreatic cancers independent of BRCA proteins (33
Kirchhoff and colleagues(21
) (, Study 21) studied the incidence of 3 founder BRCA mutations in unselected prostate cancer patients. When results were stratified by gene, BRCA2 mutation carriers were at increased risk for developing prostate cancer (odds ratio 4.78; [1.87-12.25]. The risk of BRCA1 mutation carriers was not significantly increased. Vazina and colleagues (Study 23) found that the rate of founder BRCA1 mutations in Ashkenazi Jewish prostate cancer patients (2.3% and 3.3%) was greater than the rate of the general population (0.77% and 0.37%)(23
). Mutation carriers had double the risk of the general Ashkenazi population. Only 1 of 5 carriers had prostate cancer detected early.
In the relatively homogenous Icelandic population, one of two Icelandic population studies found significant elevations in prostate cancer risk for BRCA2 mutation carriers (, studies 25 and 26)(25
Several studies of ocular melanoma patients also found significant elevations in BRCA1/2 mutation carriers (e.g. Study 27)(27
). The incidence of ocular melanoma appears to be 2-3% in BRCA mutation carriers, but there is insufficient data published to include them in relative risk plots in Figs and .
Studies by Neill and colleagues(29
) and Kirchoff and colleagues(30
) do not exclude a small increase in risk for colon cancer. The two studies involved only a small number of mutation carriers (24 and 8 mutation carriers, respectively). Neill and colleagues found a modest elevation in colon cancer risk that did not rise to statistical significance: BRCA1 or BRCA2 mutation OR=1.47 [0.76 to 2.83)] adjusted OR 1.50 [0.77 to 2.95]. A graph of the risk of colon cancer vs. number of mutation carriers in all available studies (not shown) suggests that a much larger number of mutation carriers may be needed to demonstrate increased risk. Thiffault and colleagues(36
) noted that it is not uncommon to find families with multiple cases of breast and colon cancer. They reported on one family with 10 cases of Breast or colon cancer among 26 first-, second- or third-degree relatives. The family had an apparent dual mutation in both BRCA2 and MSH2.