Association between hereditary predisposition to common cancers and congenital multimalformations

In a previous article we reported that mutations favoring cancer at adulthood seemed to improve fertility and limit miscarriages. Because spontaneous abortion may result from anomalies in embryo, we questioned if an increased frequency of congenital malformation could be evidenced among cancer‐prone families. Oncogenetics database (≈193 000 members) of the comprehensive cancer center Jean Perrin was crossed with regional registry of congenital malformations (≈10 000). Among children born between 1986 and 2011, 176 children with malformation matched in both databases. In breast/ovaries cancer‐prone families, the risk for malformations was multiplied by 2.4 [1.2‐4.5] in case of a BRCA1 mutation. Frequencies of malformation in BRCA2 and MMR mutated families were similar to families without a cancer syndrome. In comparison to malformations concerning a unique anatomical system, multimalformations were significantly more frequent in case of BRCA or MMR mutations: compared to families without cancer syndrome, the risk of multimalformations was multiplied by 4.1 [0.8‐21.7] for cancer‐prone families but with no known deleterious mutation, by 6.9 [1.2‐38.6] in families with a known mutation but an unknown parental mutational status and by 10.4 [2.3‐46.0] when one parent carried the familial mutation. No association with the type of anatomical system was found, nor with multiple births. These results suggest that BRCA and MMR genes play an important role in human embryogenesis and that if their function is lowered because of heterozygote mutations, congenital malformations are either more likely (BRCA1 mutations) and/or more susceptible to concern several anatomical systems.

Funding information FEDER (European Funding of Regional Development) and Conseil Régional Auvergne, Grant/Award Number: convention n 36760 (2013); Conseil Régional Auvergne; FEDER (European Funding of Regional Development) In a previous article we reported that mutations favoring cancer at adulthood seemed to improve fertility and limit miscarriages. Because spontaneous abortion may result from anomalies in embryo, we questioned if an increased frequency of congenital malformation could be evidenced among cancer-prone families. Oncogenetics database (≈193 000 members) of the comprehensive cancer center Jean Perrin was crossed with regional registry of congenital malformations (≈10 000). Among children born between 1986 and 2011, 176 children with malformation matched in both databases. In breast/ovaries cancer-prone families, the risk for malformations was multiplied by 2.4 [1.2-4.5] in case of a BRCA1 mutation. Frequencies of malformation in BRCA2 and MMR mutated families were similar to families without a cancer syndrome. In comparison to malformations concerning a unique anatomical system, multimalformations were significantly more frequent in case of BRCA or MMR mutations: compared to families without cancer syndrome, the risk of multimalformations was multiplied by 4.1 [0. 8-21.7] for cancer-prone families but with no known deleterious mutation, by 6.9 [1.2-38.6] in families with a known mutation but an unknown parental mutational status and by 10.4 [2.3-46.0] when one parent carried the familial mutation. No association with the type of anatomical system was found, nor with multiple births. These results suggest that BRCA and MMR genes play an important role in human embryogenesis and that if their function is lowered because of heterozygote mutations, congenital malformations are either more likely (BRCA1 mutations) and/or more susceptible to concern several anatomical systems. The persistence of such mutations in the population is demonstrated with founder mutations known to be thousands of years old. [1][2][3] In a large retrospective survey of our oncogenetic database, we found that in families predisposed to breast/ovarian cancer, women from BRCA mutated families had a 36% lower miscarriage frequency (P = 0.015) than those from families with no known deleterious mutation; among individuals of both genders tested for a BRCA mutation, childless individuals were 22% less frequent for carriers (P = 0.0022) and the interval between first and last child was 16% longer (P = 0.042). 4,5 Although the underlying biological mechanisms are not yet known, this finding could suggest two opposing hypotheses: 1. Natural miscarriage triggers could be inhibited by defective genetic pathways that predispose to cancer. Considering that about 50%-70% of miscarriages are caused by cytogenetic abnormalities 4-8 of which a majority are induced by de novo aneuploidy, 9 a higher frequency of congenital anomalies might be seen among carriers of mutations in these pathways.
2. Mutations predisposing to cancer at adulthood could reduce the risk for embryonic or fetal malformation-by an unknown genetic mechanism-and thus diminish the resulting frequency of miscarriage in the mutated population.
To the best of our knowledge, no research has been published on the incidence of congenital malformations in the offspring of parents with hereditary cancer risk. Sources of data on both conditions were available in the Auvergne region of France and were extensive enough to enable research regarding this issue. We crossed the information available in two large databases, the first used by the oncogenetic service of Centre Jean Perrin Comprehensive Cancer Center containing pedigrees of cancer-prone families and the second from the regional register of congenital malformations that includes all children born with a congenital malformation. The study period corresponded to 26 years, from January 1986 to December 2011.

| MATERIALS AND METHODS
The Auvergne region contains about 2% of the national population. 10 Because of its distance from ports and major trade routes and its traditional rural character, migration inflows have always been limited.
This makes Auvergne a good model for long-term oncogenetic studies.

| Regional registry of congenital abnormalities
The Centre d'Etude des Malformations Congénitales en Auvergne (CEMC-Auvergne) is one of seven regional registries of congenital malformations in France. Launched in 1983, it is certified by the National Committee of Registries. It concerns all mothers giving birth in either public or private clinics in the region, about 14 000 births per year, including stillbirths and therapeutic abortions. All malformed newborns are registered as they are born after a pregnancy of at least 22 weeks of amenorrhea or if pregnancy was interrupted for congenital malformation regardless of the duration of amenorrhea. For livebirths, the diagnosis of malformation must be made during the first year of life. Exhaustivity is ensured because each health institution must declare cases of malformation and more than 99% of women give birth in public or private maternities.
All types of malformations are registered, including single or asso- at the oncogenetics department but for who no cancer risk was diagnosed and therefore no genetic test was performed. A priori, this group was not exposed to a higher cancer risk than the general population, and was assumed to carry the same congenital malformation risk. Another control group was composed of individuals testing negative for a known familial mutation.

| Database matching
A temporary mixed database was constituted to evaluate the frequencies of congenital abnormalities: children born alive between 1986 and 2011, regardless of parental mutation status and type of cancer predisposition were extracted from our oncogenetic database. A computer "robot" was developed to match its records with those of the malformations register, based on similar/close children names, date of birth, mother's names, and age of parents; a visual control list permitted to validated manually each proposed match. To limit bias, that is, artificially increase the risk for malformation in particular subgroups of cancer-prone families because of a syndrome combining both cancers and malformations, four cases of microcephaly were excluded from the analysis: these cases were referred for genetic analysis on the basis of specific congenital abnormalities suggestive of Nijmegen breakage syndrome (NBS), a syndrome that includes strong predisposition to lymphoid malignancy. 16 Their corresponding families were thus excluded from the data-matching diagram ( Figure 1). Fanconi anemia was considered for exclusion, but none of our cancer-prone families presenting with this syndrome had children born after 1985.
One hundred and seventy-seven children with congenital anomalies corresponded to children in our oncogenetic database. The

| Statistical analysis
Each family where a deleterious mutation was diagnosed yields three sets of children: those born to carriers of the familial mutation, those born to parents of unknown mutation status, and children born to non-carriers. In the first set, 682 children were born to parents Because the risk for congenital malformation increases in cases of multiple births, 17-21 the frequency of these events was computed per subgroup. These frequencies were compared to frequencies of congenital anomaly both to verify that our statistics were not biased because of this, and also to control that these frequencies were in accordance with national figures.
To compare proportions of children with abnormalities between subsets, χ 2 test were used, or Fisher's exact test if needed. 95% confidence intervals were calculated assuming the number of observed cases followed the Poisson distribution of rare events. Cochran-Armitage test for trend was used to evaluate the proportion increase of syndromal, chromosomal, and multiple malformations across mutational groups. The relation between proportions of twins and congenital abnormalities were tested using standard Pearson correlation. R version 3.0 and SEM software 22 were used for statistics and datamanagement.

| RESULTS
In the oncogenetic database, 16 798 children with a known date of birth were born between 1986 and 2011, of whom 2292 belonged to families with a BRCA mutation and 390 to families with an MMR mutation. The registry contained 11 234 cases diagnosed during the same period, with 10 026 corresponding to the selection criteria. Considering about 364 000 children were born in Auvergne during the 26-year study period, this corresponds to a malformation rate of 2.8%, a rate compatible to the nation-wide estimation equal to 2.4% in 2011 to 2012. 11 Overall, 176 children with congenital anomalies corresponded to children in our oncogenetic database (the matching process is described in the material and methods section).

| Frequencies of congenital malformation
Because most families recruited at our center consult for breast/ovarian cancer syndrome, 62% of children belonged to HBOC families, that is, exposed to Hereditary Breast or Ovarian Cancer susceptibility ( Table 1)  Children of deleterious mutation carriers had a 10-fold risk of "extended" malformation in comparison to the control group. A same trend was found if we only considered multi-malformations (P = 0.0013). But no significant trend was found for syndromes alone (P = 0.13) and chromosomal anomalies (P = 0.18).
The frequencies of "extended" malformation per group increased proportionally to the probability that children might carry a deleterious mutation (ie, 100% divided by two in the group of carriers parents, 25% when the mutational status of the parent was unknown but from a mutated family, and so on): the proportion of "extended" malformations ( Table 2) varied accordingly if it was calculated either over the total number of malformations or over the number of children per group (respectively P = 0.008 and 0.003). Meanwhile in these four groups, the global frequencies of malformation whatever the type (unique or "extended") did not differ (P = 0.25).
Finally, as the risk of malformation tends to increase with father's age, due to an increased frequency of de novo mutations, 23,24 fathers' age was compared according to malformation types. Overall, no difference was found between the four groups of parents (P = 0.28) although the association between chromosomal malformations and older fathers was close to significance (P = 0.06). Mean fathers' age was respectively 31.
In our population, systems concerned by malformations did not significantly vary according to cancer risk groups (P ≈ 0.90).

| Analysis of multiple births frequency
Because associations have been described between multiple births and congenital malformation incidence, we investigated if frequencies of abnormalities and twinning were associated in our population. Two families registered triple births: they were grouped with twins. The overall frequency of multiple births was 2.47%. No association was found between frequencies of twins and frequencies of malformations (r = 0.07, P = 0.81). Seven malformations were reported among our 341 twins/triplets, all unique except one, multiple, malformations in one twin born to a BRCA2-mutated parent. Five concerned heart anomalies and one skeletal anomaly, while the only multiple case associated heart, digestive and CNS malformations. Surprisingly, no twins were observed in families with MMR mutations: P = 0.00023 vs 3.38% of twins in the control families ( Figure 5).
When all cancer syndromes together were compared to the control group, the frequencies of multiple births were similar (respectively 2.95% and 3.38%, P = 0.32). No interaction regarding congenital malformations was observed between multiple births and BRCA mutations (P = 0.89).

| DISCUSSION
In our previous study, 4 we observed fewer miscarriages in cancerprone families and suspected a possible increase of congenital malformations. In the present study, cancer predisposition was weakly associated with a higher incidence of congenital abnormalities in offspring. This association was entirely attributable to BRCA1 mutated parents who were associated with an increased risk of 2.37 [95% CI     43 Finally, the frequency of syndromal and chromosomal malformations in our sample was low when compared to our registry (2% vs 8% in the registry for syndromes, P = 0.009 and 8% vs 14%, P = 0.023 for karyotype defects): indeed, genetic anomalies corresponding to labeled syndromes rarely induce cancer and those favoring cancer (Nijmegen for example) were excluded. This is also true for chromosomal anomalies (trisomy for example) that highly reduce the risk for solid tumors while it increases the risk for leukemia. 44 Conversely multimalformations were more frequent in our sample (14% vs 10%, P = 0.05).
One source of bias may be related to the evolution in recent decades in prenatal diagnosis, where sensitivity of screening has strongly improved and malformations can be diagnosed earlier. 45  mutations did not seem to increase the risk for malformation in case of multiple births (P = 0.80), but our study was not powered to investigate this issue. We may however conclude that multiple births per category are not likely to bias our statistics. The only particularity found in our pedigrees is the absence of twins in the MMR families.
The associated probability when compared to our reference group (P = 0.00023) let us suggest this might not be an artifact and MMR mutations may indeed interfere with the mechanisms favoring multiple births.

| CONCLUSION
In our study, BRCA or MMR mutations significantly increased the risk for congenital multimalformations. This suggests that DNA repair genes play a role in embryonic development and that some congenital malformations may result from either less efficient repair, or from non-repair functions of these genes. This agrees with the recent review on this issue by Terabayashi et al. 46 This study partially confirms one of our working hypotheses: BRCA1 mutation carriers seem more likely to give birth to children with malformations. Further study is necessary to evaluate the influence of BRCA and MMR mutations on fetal deaths and/or medical abortions in case of fetal anomalies.