Analysis of the CDKN2A Gene in FAMMM Syndrome Families Reveals Early Age of Onset for Additional Syndromic Cancers

Melanoma of the skin is a significant public health issue with an estimated 87,110 new cases and 9,730 deaths in the United States during 2017 (1). Approximately 5% to 10% of cutaneous melanomas occur in families with hereditary melanoma predisposition (2), suggesting that 4,350 to 8,700 cases of cutaneous melanoma in the United States annually can be attributed to familial and/or hereditary predisposition. Factors that can increase risk 10-fold include atypical moles, more than 100 typical moles, or family history of two first-degree relatives with melanoma. In general, familial melanoma cases have an earlier age at diagnosis than sporadic melanoma cases (approximately 34 years compared with 54 years; ref. 3). In individuals with hereditary melanoma, life-time risk for melanoma is 50% to 90% (3).

Familial atypical multiple mole melanoma (FAMMM) syndrome is a cancer disorder that was first identified due to the presence of multiple atypical nevi (moles) in several first-degree relatives of a family. It was first documented by Norris (4), who described what may be the first recorded example of FAMMM syndrome, which he referred to as a “fungoid disease.” Norris suggested that the condition was hereditary as the proband's father, brothers, and children also manifested multiple moles. Over a century later, Lynch and colleagues noted what appeared to be a greater than expected occurrence of additional cancer types in one FAMMM kindred (5) followed by a study with four kindreds that showed significantly increased cancer rates for several anatomical sites when compared to population rates (6). However, due to the limited number of families and cancer data available, the conclusions of these studies were limited. In addition, another study found contradictory evidence where rates were not significantly higher than expected when compared to the then current cancer statistics (7). There were additional critiques of possible patient referral bias where a physician may selectively refer FAMMM kindreds with extraordinarily high cancer rates for research studies. It was argued that these families may not have been representative of most FAMMM kindreds and may have led to over estimation of cancer rates in these families.

In 1991, Lynch and Fusaro (8) were able to identify an association between FAMMM and pancreatic cancer, showing clear clustering of pancreatic cancer among a FAMMM kindred and strong evidence for phenotypic heterogeneity. Yet, the gene responsible for the syndrome had not yet been identified, which hindered researchers from finding a clear genetic link between FAMMM and additional cancer types beyond pancreatic cancer. Without the ability to clearly identify carriers of the mutation who did not express the nevi or melanoma phenotype (which was used to help determine carrier status before the disease genes were identified), any type of association or statistical inference would be under-powered. In addition, pedigrees would need to be expanded and family members followed over time in a longitudinal study to identify the prevalence of latent cancer types that tend to occur later in life.

Linkage analysis in families with multiple cases of cutaneous melanoma (9), followed by candidate gene analysis (10, 11) and gene characterization (12), resulted in the identification of a FAMMM syndrome risk gene now known as CDKN2A. Although a number of genes have been found to strongly increase risk for FAMMM syndrome, including CDKN2A, CDK4 (13), and MITF (14, 15), mutations in the CDKN2A gene have been identified in approximately 20% to 40% of families manifesting a predisposition to melanoma (2). In addition, other cancers, especially pancreatic cancer, are known to be associated with some of these mutations (16–18). In consideration of the central role that CDKN2A plays in cell-cycle regulation (it encodes two distinct cell cycle proteins that inhibit cell cycle progression), it is not surprising that multiple mutations in this gene (generally 1 mutation per high-risk family) have been associated with melanoma as well as various other cancer types and clinical features. These cancers include melanoma and pancreatic cancer, and suggestive associations with breast and lung cancer (17), as well as the precancerous dysplastic nevi, which are the cardinal features of FAMMM syndrome. For simplicity, we refer to the aforementioned well established FAMMM syndrome conditions (melanoma cancer, pancreatic cancer, and dysplastic nevi) as FAMMM-related features. Since the discovery of the CDKN2A gene, FAMMM kindreds that each segregate a high-penetrance variant in CDKN2A (different variants in different families) have been followed over long periods of time. Additional cancers other than the FAMMM-related cancers have been observed in both carriers and noncarriers in these families. This observation led to the question of whether any of these additional cancers should be considered to be FAMMM syndrome related. Studies examining relative risk of nonmelanomas in close relatives of individuals with melanoma have identified an increase in relative rates of nonmelanomas in these individuals (19, 20). These studies, however, did not genotype family members for any specific FAMMM-related genes. This likely resulted in a sample set with heterogeneous disease etiologies, which would decrease the power to detect specific factors that may increase risk of other cancers. Another study looked only at first-degree relatives of CDKN2A mutation carriers and found an increase in additional cancers outside of FAMMM-related cancers, but they did not assess the effects in more distant relatives that are more dissimilar in their overall genetic and environmental background (21). These individuals potentially have different genetic backgrounds, such as common variants with small additive effects on cancer risk (i.e., the polygenic component), that can interact with or modify the effects of CDKN2A, leading to a wider array of cancer outcomes. Also, more distant relatives are more likely to have larger differences in environmental exposures than do close relatives.

For this study, we sought to perform survival analysis in extended FAMMM syndrome families that have been genotyped for mutations in the CDKN2A gene to determine whether there is an elevation in the occurrence of additional cancers in these families and/or an age effect that would warrant early genetic assessment of genetically closely-related as well as distant family members. We present a comprehensive statistical analysis of well characterized extended FAMMM syndrome families that have been followed longitudinally and have been directly genotyped for risk variants in the CDKN2A gene to assess this question.

These families have been enrolled in genetic studies at Creighton University and have been followed for over 30 years at the Creighton Hereditary Cancer Center (Table 1). These family members gave written informed consent and have been approved for whole exome sequencing, whole genome sequencing, and database of genotypes and phenotypes (dbGaP) data sharing by the Institutional Review Board at Creighton University.

All families in this study were identified as having a deleterious CDKN2A mutation that segregated with FAMMM syndrome. CDKN2A genotyping for the families included in this study have been previously described in past publications (22, 23). Briefly, FAMMM syndrome families that were previously shown to have linkage for melanoma on chromosome 9p21 (the region that harbors the CDKN2A gene) were further evaluated for specific mutations within the CDKN2A gene. For one such study, exons shown to harbor disease mutations were PCR-amplified and sequenced for a set of FAMMM syndrome families (23). Additional families for whom a CDKN2A exonic mutation was not identified were tested for splice variants within the introns using PCR and reverse transcriptase (PCR; ref. 22). These families were shown to have mutations at sites that resulted in alternatively spliced mRNA that produced functionally altered proteins. The mutations identified in the aforementioned studies were shown to be disease-causing and are listed in Supplementary Table S1. Once a pathogenic CDKN2A mutation was identified in an affected family member, subsequent testing within the family was performed for the site-specific family mutation.

We developed criteria using Mendelian inheritance patterns to unambiguously determine obligate mutation carriers for some individuals who were not directly tested for CDKN2A mutations. These individuals were included in a strict analysis. We also developed less strict criteria for a second analysis (“loose analysis”) that included individuals who are likely to be carriers, but could not be unambiguously determined as such.

First, carrier status was determined based on a positive or negative CDKN2A mutation test of the family-specific mutation (see Fig. 1, for the results of this testing for one family). Then, two different sets of criteria were generated to infer mutation status for nongenotyped individuals: one to be used for a loose analysis composed of less strict criteria as well as a set of stricter criteria for a stringent analysis (Fig. 2). For both analyses, non-bloodline individuals who did not have a FAMMM-related cancer (e.g., melanoma and pancreatic cancer) were categorized as negative inferred because these mutations are rare in the general population, especially in those who do not have the aforementioned cancer types. Also, for both analyses, individuals who were descendants of someone who tested negative were categorized as having “negative inferred” carrier status. Individuals who were FAMMM bloodline ancestors of someone who tested positive for CDKN2A mutations were categorized as “positive inferred” carriers of the same mutation because they are obligate carriers. Finally, bloodline individuals who did not have enough information to unambiguously infer mutation status (descendants of a carrier who had not themselves been tested) were categorized as having unknown mutation status and excluded from both analyses.

In the “loose” analysis, untested bloodline family members (without a mutation carrier descendant or a noncarrier bloodline parent) who had the aforementioned previously implicated, relatively rare FAMMM-related cancers (melanoma or pancreatic cancer) were designated as positive inferred. These individuals are very likely to be carriers of a CDKN2A mutation as they have bloodline relative(s) in which the mutation was identified and they have a relatively rare cancer that has been previously implicated as part of the FAMMM syndrome. For the “strict” analysis, we accounted for the possibility that some of the aforementioned likely carriers were misclassified and are actually sporadic cancer cases. Therefore, for the strict analysis we only categorized untested bloodline individuals as positive inferred mutation carriers if (i) any of their descendants tested positive; or (ii) this untested individual had both a less common FAMMM-related cancer (melanoma, pancreatic) and at least one additional primary cancer (any primary cancer), because multiple primary cancers are a hallmark of an inherited cancer susceptibility allele. Although breast and lung cancer have shown a suggestive association with the syndrome, these cancers were not used in this determination because they are too common in the general population and we could not be as confident that they are the result of a CDKN2A mutation. Mutation testing of additional family members is required to answer this question definitively.

Ten FAMMM syndrome families for whom a causal mutation in the CDKN2A gene was identified were used for these analyses. FAMMM-related cancers in these families were classified as such only if the condition was previously shown to occur at increased frequency in carriers of CDKN2A gene mutations through multiple replication studies (melanoma and pancreatic cancer). Breast and lung cancer, which both have been found to have suggestive associations with FAMMM syndrome, were tested in a replication analysis to determine if our dataset shows a similar relationship between these cancers and FAMMM syndrome. All other cancers were classified as “other cancers” and tested for new associations. Individuals who had both a FAMMM-related and “other” primary cancer event were included in both the FAMMM-related and other cancer analyses as these cancer events were found to be independent based on histologic analyses. However, the age at the time of the diagnosis for the FAMMM-related or other cancer was used for the FAMMM-related or other cancer analysis, respectively. The observed FAMMM-related cancers, replication analysis cancers and other cancers in these families are listed in Supplementary Table S2A. Supplementary Table S2B contains the full names associated with the abbreviations in Supplementary Table S2A.

Data were plotted as a survival curve and a mixed effects Cox regression (24) was performed with age at cancer event or follow-up as our time variable and presence or absence of a FAMMM-related cancer, replication cancer, or other cancer (each in a separate analysis) as our censoring variable adjusting for gender and family grouping. The mixed effects Cox regression accounts for relatedness between family members by generating family-specific y-intercepts and slopes. Essentially, this removes the random effects of shared environment. This method was chosen as opposed to generating relatedness matrices that account for the effects of genetic similarity between family members because the main effect variants for this syndrome have already been identified, so a multigenic model was not tested.

The predictor variable was presence or absence of a mutation in CDKN2A with the exclusion of individuals with ambiguous carrier status as described in the methods section. All carriers and noncarriers who had other cancers, including individuals who may have also had a FAMMM-related cancer (multiple primary cancers), were included in the analysis as well as carriers and noncarriers who have never had cancer. These analyses were performed using either the loose or strict criteria-generated datasets separately. Analyses were performed using R package version 3.4.2 and an α level of 0.025 was used to assess significance in the initial analyses because we performed two analyses, on the “loose” and “strict” carrier status definitions, respectively. Sensitivity analyses were then performed to determine whether changing the conditions and assumptions of the analyses changed the results. These sensitivity analyses consisted of the following changes: (i) we included breast and lung in the “other” cancer group to see if the P value would change for the “other” cancer group analysis; (ii) we removed nonmelanoma skin cancers from the other cancer group (coding individuals with these cancers as unaffected with an “other” cancer); (iii) using the original definition of “other” cancers, we included individuals with FAMMM-related cancers in the “other” cancer analyses as unaffected instead of excluding them from the analysis (strict criteria); and (iv) using only bloodline carriers and bloodline noncarriers in the other cancer analysis (strict criteria) to remove the possible effects of biased family ascertainment from the analysis.

We constructed pedigree drawings using Cranefoot version 3.2.3 for the 10 families to look for general patterns of co-segregation between mutations in CDKN2A and the occurrence of other cancers. These pedigrees reflect the strict carrier status criteria and include all the individuals in the files, unlike some of the forthcoming summary tables, which exclude individuals with incomplete data for all variables. Supplementary Fig. S1 is a pedigree for one such family and is an example of the cancer inheritance patterns observed in the study. The colors yellow and blue mark individuals with other cancers and FAMMM-related cancers, respectively. Individuals in green had both FAMMM-related cancers and other cancer(s). The shapes with two triangular patterns represent individuals who are known carriers whereas those with slanted lines or one square are those whose carrier status is unknown or who are potential carriers, respectively. Individual 5 is a carrier who developed both a FAMMM-related cancer and an “other” cancer primary. This individual produced an offspring (individual 7) who is also a carrier but only developed an “other” cancer primary. We also observed that in carriers who exhibit multiple primary cancers, we saw a combination of FAMMM-related and “other” cancers. Another visible pattern is the occurrence of dysplastic nevi or melanoma as the first condition followed by various combinations of other cancers (Supplementary Table S3). Similar inheritance patterns were seen in other pedigrees.

Next, we generated summary statistics for the individuals who were included in the statistical analysis of the FAMMM-related cancers, replication analysis cancers, or “other” cancers versus no cancer (Table 2; Supplementary Table S4). As we included covariates (gender, family group, and carrier status) in the analyses, only individuals with information for all relevant variables (cancer status, age at onset or last follow-up, and gender) were included in our summary statistic tables. Supplementary Table S4 has the summary information for additional pertinent variables. As it relates to the “other” cancer analysis, 453 individuals were excluded: 258 individuals were excluded due to lack of information on carrier status, 103 due to missing age information, and 79 had only a FAMMM-related cancer or replication cancer event and therefore were not included in the initial “other” cancer analysis (Supplementary Table S5). Based on summary Table 2, there is a much larger proportion of carriers that developed “other” cancers (38%) when compared with those who are noncarriers (3%).

We first generated survival curves and Cox regression hazard ratios (HR) for the FAMMM-related cancer group for comparison purposes. This analysis serves as a positive control as we expect a strong and visible effect of the mutation on cancer occurrence in this group. As expected, the survival curves show significant separation between the curve for the carrier group versus that of the noncarrier group (Fig. 3). The curve for the other cancers shows a similar pattern with the median cancer-free survival probability being ∼55 years old (Fig. 4) whereas that observed for FAMMM-related cancers is ∼50 with noncarriers having little to no occurrences of cancer by this age. The replication analysis of lung and breast cancer, which only consisted of eight cancer events, shows that these cancers seem to occur earlier for carriers (Supplementary Fig. S2; Supplementary Table S6). However, there were not enough cancer events or carriers who developed breast or lung cancer to accurately determine this to be true. Nor could a median survival probability be determined with this small sample set.

The Cox regression reflected what was seen in the survival curves with all models being highly significant for the loose analysis (models adjusted for gender and family grouping; Table 3). In Table 3 it can be seen that the HR and 95% confidence interval (CI) for the FAMMM-related cancer and other cancer analyses overlap. We performed the strict analysis in which we repeated the analyses and excluded individuals who are likely carriers, but were not tested for CDKN2A nor were their descendants. There was little change in the HR and the P-value for both the FAMMM-related cancers and “other” cancer analyses and both remained extremely significant. In addition, we added breast and lung cancers to the “other” cancers (for this analysis only) to see if the P value for the point estimate of other cancers would become more significant and we found this to be true (HR 0.06; 95% CI, 0.03–0.12; P = 6.10E−14; Supplementary Table S7; Supplementary Fig. S3). Because melanoma screening in these high-risk families will identify nonmelanoma skin cancers but not the “other” cancers observed here, and because these families may seek skin cancer screening more frequently and at earlier ages than noncarriers of these mutations, we sought to determine if the results were being driven solely by differences in nonmelanoma skin cancer rates in carriers versus noncarriers or whether there was evidence that new sites should also be commonly screened. Hence, we repeated the other cancer analysis with nonmelanoma skin cancers removed to ensure that this cancer type was not driving the effects seen. The result was still highly significant (HR 0.02; 95% CI, 0.02–0.13; P = 3.8E−10). Next, we performed an analysis where the FAMMM-related cancer group was included in the control group instead of excluded from the other cancer analysis. Although this more conservative analysis scheme results in a larger number of carriers being included in the analysis as unaffected, the analysis still gave highly significant evidence of increased risk of the occurrence of other cancers for CDKN2A mutation carriers compared to non-carriers (HR 0.13; 95% CI, 0.06–0.31; P = 2.90E−06).

Ascertainment bias is a common concern in epidemiologic studies and was of particular concern for this study because many individuals in this study developed cancer before enrollment. This presents the potential for erroneous statistical associations as a medical provider may have recommended families with abnormally high rates of “other” cancers for this study. As many non-bloodline individuals were used as controls and were not referred to the study based on these same criteria, we wanted to ensure that the results we generated were not due to the aforementioned reasons. Hence, in our third sensitivity analysis, we performed the same analysis as above with only bloodline family members who either had or did not have a CDKN2A mutation (strict carrier criteria used).

We first generated summary statistics (Supplementary Table S8) with rates of “other” cancers in each group and we did not see a large difference in other cancer rates between noncarriers who were non-bloodline (4%) or were bloodline (2%) family members whereas bloodline carriers had a much higher rate (37%). The survival plot for these groupings indicate that bloodline noncarriers had similar cancer age of onset as non-bloodline noncarriers with little to no occurrence of cancer by age 55, which is the median cancer-free survival age for bloodline carriers, but the age effects in this group could not be accurately assessed post-age 70 due to high censoring after this age (Supplementary Fig. S4). We then performed the same Cox regression analysis as before, but only for bloodline family members with and without CDKN2A mutations and still found a significantly earlier age of onset for bloodline carriers compared with bloodline noncarriers (HR 0.15; 95% CI, 0.04–0.60; P = 7.10E−3) despite the reduction in sample size of noncarriers in this analysis.

Considering that mutations in different locations of CDKN2A may have different effects, we examined the locations of the various mutations in these families to see if there was any clustering. This might warrant an analysis to look for mutation-specific effects on cancer age of onset or increased risks of specific cancer types. The mutations were spread across the gene and did not seem to cluster in one location although there were three mutations located in the N-terminus of the protein (Supplementary Fig. S5). We also did not find any of the known founder mutations in the gene. We then examined which specific cancers occurred in carriers versus noncarriers to determine if any occurred at a higher rate in carriers and found that there is a higher proportion of carriers with skin (not melanoma or sebaceous), colon, nervous system, soft tissue, bone, esophageal, ovarian, and testis cancer than noncarriers (Supplementary Table S9).

FAMMM syndrome has long been suspected to be associated with additional cancers, outside of the syndromic FAMMM-related cancers, yet the field was not yet advanced enough to definitively make this conclusion. After years of gene discovery and characterization as well as the expansion of FAMMM families, we have been given the opportunity to perform a more comprehensive analysis of one of the genes that contribute to the syndrome and assess its phenotypic heterogeneity. The objective of this study was to describe a wealth of information that has been revealed through the statistical analyses of these families and has guided the suggestion to expand surveillance to include that of cancers outside of the known FAMMM-related cancers in carriers.

We first reviewed the patterns of inheritance of FAMMM-related cancers and other cancers, from which we made a few important observations. First, in addition to family members exhibiting high rates of melanoma and pancreatic cancer, we also see that some of these individuals give rise to carrier offspring who exhibit only “other” cancers. Furthermore, in carriers who exhibit multiple primary cancers, which is a hallmark of inherited cancers, we see a combination of FAMMM-related cancers and other cancers. These suggest that the familial mutation likely contributed to each independent primary at different time points in an individual's life. Another visible pattern is the occurrence of dysplastic nevi and melanoma as the first primary followed by various combinations of other cancers. It is known that there is an environmental contribution to melanoma with sun ultraviolet (UV) rays being a strong factor in cancer initiation (25). It is likely that melanoma is the initial cancer to occur in a series of cancer occurrences as the result of a CDKN2A mutation tumor-driving effect paired with almost unavoidable UV light exposure, which is a common and abundant environmental cancer risk factor that directly affects skin. The additional cancer primaries that occurred in these individuals are likely the result of other acquired somatic mutations and/or environmental exposures that have latent effects in the tissue where these other cancers occurred.

Our examination of FAMMM-related cancer incidence rates as a function of time in the form of survival plots indicates that carriers show a strong age effect in cancer occurrence, which is another hallmark of familial cancers. We indeed expected this result because it is well known that melanomas occur at a younger age in CDKN2A carriers than in non-carriers. As we expected the FAMMM-related cancer group to display a strong age effect, this plot serves as a control and models the pattern we expected to see in the other cancers group if they indeed were the result of familial mutations. Just as with the FAMMM-related group, we see a sharp drop in cancer-free survival probability in carriers after the age of 25 with the 50% survival probability occurring around age 55 for the “other” cancer group. For both groups, the noncarriers have little to no incidence of cancer by this age providing strong evidence that the age effect seen in both FAMMM-related and other cancers analyses is due to CDKN2A familial mutations. Our analysis of lung and breast cancer, which both have been shown to have suggestive associations with FAMMM syndrome, supports previous findings. However, there were only eight cancer events in the analysis so this replication analysis was not conclusive. A larger dataset is needed to definitively state that this relationship exists.

The Cox regression analysis reflected what was seen in the survival plots. Individual who are carriers are 100 times more likely to develop a FAMMM-related cancer than those who are noncarriers (P = 7.15E−20). The effect of CDKN2A mutations on other cancers was also strong, albeit less so, with carriers being 20 times more likely than noncarriers to develop these other cancers (P = 5.00E−13). Despite the difference in magnitude of the effect between these groups, the HRs and their CI greatly overlap. This difference is possibly due to differences in sample size, resulting in the other cancers group having a wider CI. An additional explanation is that the stronger age effect in the FAMMM-related cancers may be driven by the melanoma events, which tend to occur at younger ages than other cancers. Finally, it is also possible that these mutations produce a stronger increase in the risk of FAMMM-related cancers than for the other cancers. Our assessment of ascertainment bias by examining only bloodline family members bolsters our survival analysis results and suggest that they are not due to flaws in participant recruitment design. They also suggest that the significant age effect and increased rate of other cancers is specific to carriers and not simply familial. Follow-up studies with larger sample sizes of known CDKN2A mutation carriers and known noncarriers within large families should perform similar analyses stratified by cancer type to further refine this age effect. In addition, larger studies may also examine the effects of the various mutations in CDKN2A since each may have different functional consequences and effects on cancer development. Those that were identified in this study were scattered across the gene and could not be grouped in a meaningful way (Supplementary Fig. S5).

Finally, it also appears that several cancers, especially nonmelanoma, non-sebaceous skin cancer, are observed more frequently in mutation carriers than noncarriers in this dataset (Supplementary Table S9). Thus, there may be greater susceptibility for these particular cancers (nonmelanoma skin cancer, colon, nervous system, soft tissue, bone, esophageal, ovarian, and testis cancers) in carriers when compared with other cancers. There are not enough samples in this dataset to suggest this is a feature of FAMMM syndrome in general, but distinguishing between other cancers that occur more frequently in mutation carriers will be important for future studies. We showed that the significant excess of additional cancers in CDKN2A carriers was not due solely to non-melanoma skin cancers since the analysis excluding these nonmelanoma skin cancers still showed highly significant increased risk for other cancers in the carriers compared with noncarriers (HR 0.02; 95% CI, 0.02−0.13; P = 3.8E−10). This gives strong support to the clinical implication that carriers of these mutations should also be carefully screened at early ages for additional types of cancer.

These findings support our hypothesis that the high frequency of other cancers observed in these FAMMM syndrome families manifest due to familial mutations in the CDKN2A gene. These mutations cause an excess of other cancers and FAMMM-related cancers, thereby strongly arguing for much broader cancer screening recommendations in these families for not only the FAMMM-related but now the additional cancer types. Doing so may allow for earlier detection of these additional cancers and lower mortality rates due to earlier treatments.

Our findings emphasize the importance of performing carrier detection among more distant relatives of known carriers and earlier screening guidelines for both FAMMM-related and other cancers in the carriers of CDKN2A mutations. Our results also have implications for strategies of detecting germline mutations in individuals with a family history of cancer or with multiple primary cancers. It is most common for a cancer patient to have germline and tumor sequencing performed on a set of known cancer-predisposition genes based on the tumor type of the patient. Our results suggest that a broader set of cancer predisposition genes should be sequenced because a patient with one of the other cancers here might not be screened for CDKN2A at present. Support for this idea also comes from a recent study of adults with multiple primary cancers, which used whole genome sequencing (26). This study found that by sequencing all known cancer predisposing genes, rather than those targeted by cancer types in the patient, they could detect a deleterious variant in about a third of individuals and that of those with a pathogenic or likely pathogenic mutation, over 40% had a tumor type that appeared unrelated to the mutated gene.

Rapidly evolving genomic technologies are influencing cancer control through the diagnosis of hereditary cancer syndromes. A comprehensive cancer family history that includes cancers of all anatomic sites, paired with optimum cancer education and targeted therapy will enhance personalized medicine initiatives.

No potential conflicts of interest were disclosed.

Conception and design: C.D. Middlebrooks, C. Snyder, M. Rendell, J.E. Bailey-Wilson

Development of methodology: C.D. Middlebrooks, C. Snyder, M. Rendell

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C. Snyder

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C.D. Middlebrooks, Q. Li, C. Snyder, J.E. Bailey-Wilson

Writing, review, and/or revision of the manuscript: C.D. Middlebrooks, C. Snyder, T.G. Shaw, M. Rendell, P. Silberstein, M.J. Casey, J.E. Bailey-Wilson, H.T. Lynch

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M.L. Stacey, T. Richardson-Nelson

Study supervision: J.E. Bailey-Wilson

Other (Assistant to Dr Lynch in review): C. Ferguson

This work was supported in part by the Intramural Research Program, National Human Genome Research Institute of the U.S. National Institutes of Health, which supported the research led by J.E. Bailey-Wilson. This publication was also supported by revenue from Nebraska's excise tax on cigarettes (LB595 funds) awarded to Creighton University through the Nebraska Department of Health & Human Services (DHHS), which supported the clinical studies led by H.T. Lynch. Its contents represent the views of the authors and do not necessarily represent the official views of the State of Nebraska or DHHS. Funding was also received from the Liz's Legacy fund through Kicks for a Cure, which supported the clinical studies led by H.T. Lynch.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.