Familial aggregation of early‐onset cancers in early‐onset breast cancer families

The risk of early‐onset (EO) breast cancer is known to be increased in relatives of EO breast cancer patients, but less is known about the familial risk of other EO cancers. We assessed familial risks of EO cancers (aged ≤40 years) other than breast cancer in 54 753 relatives of 5562 women with EO breast cancer (probands) by using a population‐based cohort from Finland. Standardized incidence ratios (SIRs) and 95% confidence intervals (CI) were estimated by using gender‐, age‐ and period‐specific cancer incidences of the general population as reference. The risk of any cancer excluding breast cancer in first‐degree relatives was comparable to population cancer risk (SIR 0.99, 95% CI: 0.84‐1.16). Siblings' children of women with EO breast cancer were at an elevated risk of EO testicular and ovarian cancer (SIR = 1.74, 95% CI: 1.07‐2.69 and 2.69, 95% CI: 1.08‐5.53, respectively). The risk of EO pancreatic cancer was elevated in siblings of the probands (7.61, 95% CI: 1.57‐22.23) and an increased risk of any other cancer than breast cancer was observed in children of the probands (1.27, 95% CI: 1.03‐1.55). In conclusion, relatives of women with EO breast cancer are at higher familial risk of certain discordant EO cancers, with the risk extending beyond first‐degree relatives.


| BACKGROUND
It is well known that the risk of early-onset (EO) breast cancer is increased in women with a first-degree relative diagnosed with breast cancer. [1][2][3] The risk of developing EO breast cancer increases when the number of first-degree female relatives with breast cancer increases. 4 With only one affected first-degree relative, the risk is 2-fold compared to women with unaffected first-degree relatives. 4 The risk further increases to almost 3-fold with a second first-degree relative affected. 4 Furthermore, first-degree relatives of a breast cancer patient appear to be at increased risk of developing any type of cancer 5 with ovarian 6,7 and prostate 1,6,8  and other cancers in first-degree relatives. 1,5 One of the most wellestablished relations is between breast and ovarian cancer 7,8 as both of these cancers are associated with germline mutations in BRCA1 9 and BRCA2. 10 Apart from ovarian cancer, previous research has suggested an increased risk of cancers of the prostate, 1,5,6,11 colon, 1,3,5 thyroid gland 1,5 and cervix 3,11 in first-degree relatives of women with EO breast cancer. These associations are not limited to EO cancers of the relatives and many of these discoveries are far from consistent and limited by small sample sizes. Concerning shared genetic factors, germline mutations in BRCA1 and BRCA2 genes do not confer increased risks only on the cancers of the breast and ovary; both BRCA1 9,12 and BRCA2 6,10 mutation carriers have been observed to have a higher than normal prevalence of prostate and pancreatic cancers.
Familial aggregation of cancer may be due to genetic susceptibility or acquired mutations caused by shared environmental factors.
There are several known germline mutations predisposing to breast cancer with a prevalence of 10% in all breast cancers. 13 At least 10% of EO breast cancer patients have germline mutations in BRCA1 and BRCA2. 14 Despite their role as the most remarkable individual genetic factors in the etiology of breast cancer, data suggest that only a minor part of the familial excess risk of breast cancer is attributable to BRCA1/2 mutations. 14 This suggests a considerable role of factors other than genetic predisposition or yet unidentified genetic factors impacting the familial aggregation of breast cancer.
Lifestyle and environmental factors associated with the risk of breast cancer, such as alcohol consumption, 15 low physical activity, 16 body weight 17 and several endogenous 18 and exogenous 19 hormonal factors are of great interest as exposure to many of them is avoidable unlike exposure to genetic risk factors. Concerning hormonal factors, for example, early age at menarche 20 and late age at menopause are associated with an increased risk of breast cancer, 17,21 presumably due to longer lifetime exposure to estrogen. 21,22 The role of body weight is not as straightforward as being obese or overweight throughout adulthood before menopause seems to act as a protective factor against premenopausal breast cancer 23 but predisposes to postmenopausal breast cancer. 17 Independent of adulthood obesity, some studies have shown a protective effect of higher body fatness during childhood and young adulthood on both premenopausal 20,24 and postmenopausal breast cancer. 20,24 Independent risk factors of premenopausal breast cancer are mostly unidentified, but it may be assumed that they at least partly overlap with those of postmenopausal breast cancer.
Familial clustering of cancer is associated with earlier age of onset 9 and the inherited component seems to contribute more to EO cancers. 1,25 In addition, it is known that the familial risk of breast cancer is the highest in the offspring when their mother has been diagnosed with breast cancer at a young age. 3,7 As an earlier age of onset is generally suggestive of a hereditary component in the etiology of the disease, it therefore makes investigating the associations between discordant EO cancers a useful tool for evaluating the etiology of familial aggregation of discordant cancers. We estimate the relative risks of other EO cancers than breast cancer (discordant cancer) in family members of EO breast cancer probands by utilizing a prospective population-based familial cohort in Finland. Follow-up periods were modified to avoid immortal time bias which is a period of time when cancer could not be diagnosed due to study design. Therefore, family members of the proband were not considered to be at risk of cancer between 1 January 1970, and the date of diagnosis of the proband (immortal period). Immortal periods were left out when estimating the SIRs to avoid bias when evaluating familial aggregation. 27 In families with at least one breast cancer diagnosed in addition to the proband's, the original clinical and pathology notifications from the FCR data were examined and any notes indicating the presence of BRCA1/2 gene mutation were extracted. No BRCA1/2 gene mutation carriers were reported in families with multiple breast cancers but it should be noted that the pathological reports were available only until the year 2014. 27 Familial aggregation was estimated by using standardized incidence ratios (SIR) to compare sex-, age-and period-specific cancer incidence among family members to that in the population of

| RESULTS
The numbers of families, relatives, and discordant (other than breast cancer) EO cancers of relatives of EO breast cancer probands for each cancer site are shown in Table 1. Only 5.5% of the probands' families had another family member with any discordant EO cancer. The most common discordant familial EO cancers were cancers of the testicle (0.6% of the families) and the thyroid gland (0.6%).
Standardized incidence ratios (SIR) for discordant EO cancers among relatives of probands are presented in Table 2 (full table available in Table A2). The SIR for any discordant cancer in first-degree relatives was 0.99 (95% CI: 0.84-1. 16 There were no other major differences in familial risk of discordant EO cancers by histological subtypes of breast cancer. We also studied familial aggregation of testicular cancer by histological subtypes (seminoma or non-seminoma) in subtypes (ductal and lobular) of the proband's breast cancer (Tables A3 and A4). The SIR for EO non-seminoma was elevated in siblings' sons of the probands both overall (SIR 2.17, 95% CI: 1. 22-4.29) and when the proband had ductal carcinoma (SIR 2.00, 95% CI: 1.00-3.58). The risk of non-seminoma testicular cancer was not increased in relatives of lobular breast cancer probands. Familial risks of seminoma type of testicular cancer were not increased.
As a sensitivity analysis, we estimated SIRs for discordant cancers in relatives of probands when the relatives were ≥41 years at the time of diagnosis (late-onset, Table A5). Most importantly, statistically significant elevations in SIRs were seen for ovarian cancer in first-degree

| DISCUSSION
We found increased familial risks of EO testicular and ovarian cancers in siblings' children and pancreatic cancer in siblings of women with EO breast cancer. In addition, an increased risk of any other EO cancer than breast cancer was observed in offspring of females with EO breast cancer. In our data, most of the EO cancers were nonfamilial (95%) and it was rare (5%) to have at least one other family member affected with a discordant cancer in addition to proband's breast cancer. The most common familial cancers combined with proband's breast cancer were testicular and thyroid gland cancers occurring in 33 (0.6%) and in 33 (0.6%) families, respectively.
We found a moderately increased risk of any other EO cancer than breast cancer in offspring of the probands. Increased overall cancer risk in offspring of mothers with breast cancer diagnosed at a young age has been reported earlier by Anderson et al. 7 The results of our study are not fully comparable to those of Anderson and colleagues as their data included also breast cancers of the offspring, In contrast, the risk of any other EO cancer than breast cancer was decreased in mothers and fathers of the probands. One reasonable explanation behind this observation is that both female and male EO cancer survivors have a significantly lower probability of parenthood 29 compared to healthy young adults. This cannot be taken into account when calculating SIRs and therefore, might be reflected as reduced risk of EO cancer in parents of the probands. Additionally, information on either one or both parents was missing for 33.3% of the probands. Also, the observed decrease in cancer risk among spouses may at least partially be due to lesser, or later, marrying of cancer survivors. 30 Spouses may also have adopted a somewhat healthier lifestyle after the proband has been diagnosed with cancer.
EO cancer of the testicles showed an elevated SIR of 1.74 (95% CI: 1.07-2.69) among siblings' sons of the probands. The association between testicular and breast cancers has not been reported earlier in a manner where a significant increase was observed only in seconddegree relatives but not in first-degree relatives. Furthermore, there is no previous evidence of a relationship between EO breast cancer and T A B L E 1 Number of families, relatives and familial early-onset cancers (≤40 years) for the most common discordant familial cancers.  31 Estrogen exposure could serve as a common etiological factor since premenopausal breast cancer is thought to be associated with higher estrogen exposure. 21,22 Finnish men are known to have a relatively low incidence of testicular cancer compared to other Nordic countries. 41 However, a recent study discovered that Finland has had the highest relative increase in testicular cancer incidence among Nordic countries during the past years. 41 By observing the speed of the increase, it is unlikely that genetic factors would explain such a rapid increase in incidence. Ekbom et al studied the incidence of testicular cancer in Finnish immigrants in Sweden and found that the reduced risk of testicular cancer was retained independent of age at immigration or duration of stay which greatly supports the theory of early exposure to environmental as a strong predictor of testicular cancer incidence. 42 Supporting the major role of environmental factors are observations that the descendants of the Finnish immigrants do not retain the same reduced risk as their fathers 43 and that a smaller age difference between brothers contributes to a greater risk of testicular cancer. 44 For now, it remains unexplained what could have caused sons of the probands' siblings to be exposed to these shared etiological factors that the probands' offspring seem to avoid. Perhaps, the explanatory mechanism behind this connection is not a single shared etiological component but an interplay between different common predisposition factors as this connection does not seem to follow any known inheritance pattern nor offer any obvious cues of common etiology. Our findings are interesting but raise some questions about unknown genetical and environmental mechanisms that need to be further studied.
We further stratified testicular cancers by their histological subtypes to see if the clustering with breast cancer was associated specifically either with seminoma or non-seminoma as they are thought to have partially different etiological backgrounds. 45 Non-seminoma showed an increased SIR in probands' siblings' sons while the risk of seminoma was at the population level in all relatives of probands. A similar pattern of inheritance was detected in a study conducted by Seikkula et al using the same data as in our study. 33 Nonseminomatous testicular cancer tends to occur at lower ages than seminomatous testicle cancer 46 which could explain why we did not see an increased risk of seminoma as our cohort was restricted to patients aged 40 years or less.
Ovarian cancer is the most commonly reported discordant cancer in association with EO breast cancer 3,11 as the risk of ovarian cancer is substantially increased in carriers of highly penetrant BRCA1 and BRCA2 gene variants. 9 There was a notable difference in SIRs for late-onset ovarian cancer when breast cancer was separated into subtypes. The increased SIR was seen in first-degree relatives overall, siblings, and mothers of breast cancer patients when the breast cancer was ductal but was absent when the cancer was lobular. Part of this association might be explained by etiological differences between lobular and ductal subtypes as the ovarian cancer susceptibility gene BRCA1 is clearly more prevalent in ductal than in lobular breast cancer. 50 Additionally, as ductal carcinoma is more common than lobular carcinoma the statistical power is higher for ductal breast cancer than that for lobular.
In our study, siblings of probands were at an increased risk (SIR

| CONCLUSIONS
The offspring of female EO breast cancer patients were at an elevated risk of any EO cancer other than breast cancer. We also found an

ACKNOWLEDGEMENTS
The Finnish Cancer Registry and the Digital and Population Data Service Agency are acknowledged for providing data to us.

FUNDING INFORMATION
The study was funded by the Cancer Foundation Finland and Academy of Finland (Center of Excellence, grant number SA 312044) (JP).

CONFLICT OF INTEREST STATEMENT
The authors declare no conflict of interest.

DATA AVAILABILITY STATEMENT
The data that support the findings of our study are available from the corresponding author (Jasmiina N. J. Rantala) upon reasonable request and with permission of Finnish Cancer Registry.

ETHICS STATEMENT
The study was approved by the Ethical Committee of the Finnish Institute of Health and Welfare (Permit no THL/1588/5.05.00/2019).
All methods were performed in accordance with the relevant guidelines and regulations (Declaration of Helsinki). Informed consent from study participants was unnecessary as the study utilized registry data