PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Gastroenterology. Author manuscript; available in PMC 2011 March 1.
Published in final edited form as:
PMCID: PMC2831153
NIHMSID: NIHMS160363

Population-based family-history-specific risks for colorectal cancer: a constellation approach

Abstract

Background & Aims

Colorectal cancer (CRC) risk estimates based on family history typically include only close relatives. We report familial relative risk in probands with various combinations, or constellations, of affected relatives, extending to third-degree.

Methods

A population-based resource that includes a computerized genealogy linked to statewide cancer records, was used to identify genetic relationships among CRC cases and their first-, second-, and third-degree relatives (FDRs, SDRs, and TDRs). Familial relative risks (FRRs) were estimated by comparing the observed number of affected individuals with a particular family history constellation to the expected number, based on cohort-specific CRC rates.

Results

A total of 2,327,327 individuals included in ≥3 generation family histories were analyzed; 10,556 had a diagnosis of CRC. The FRR for CRC in individuals with ≥1 affected FDR=2.05 (95% CI 1.96–2.14), consistent with published estimates. In the absence of a positive first-degree family history, considering both affected SDRs and TDRs, only 1 constellation had an FRR estimate that was significantly > 1.0 (0 affected FDRs, 1 affected SDR, 2 affected TDRs; FRR=1.33, 95% CI 1.13–1.55). The FRR for individuals with 1 affected FDR, 1 affected SDR, and 0 affected TDRs=1.88 (95% CI 1.59–2.20), increasing to FRR=3.28 (95% CI 2.44–4.31) for probands with 1 affected FDR, 1 affected SDR, and ≥3 affected TDRs.

Conclusions

Increased numbers of affected FDRs influences risk much more than affected SDRs or TDRs. However, when combined with a positive first-degree family history, a positive second- and third-degree family history can significantly increase risk.

INTRODUCTION

Colorectal cancer (CRC) is the third most common cancer diagnosed in the United States and the second leading cause of death among cancers. It is estimated that in 2008 alone, 148,810 people will be diagnosed with CRC and 49,960 will die from the disease.1 Screening strategies such as fecal occult blood test (FOBT), flexible sigmoidoscopy, and colonoscopy, among others, have been proven effective in reducing the incidence and mortality of the disease.2 While CRC incidence and mortality rates have been declining, a majority of US adults are still not being screened nor receiving regular screenings appropriate for their age or risk status.3,4 Knowledge of increased risk can be a motivating factor in making decisions about screening.5 If a greater proportion of adults received regular screenings appropriate for their risk it is likely that additional reductions in CRC incidence and mortality could be achieved.

Family history is a well-established risk factor for CRC.6 Many studies have estimated familial risk of CRC and consistently demonstrated that having ≥1 affected first-degree relative (FDR) doubles an individual’s risk for CRC.6,7,8,9,10 A recent random-effects analysis that pooled relative risk estimates for CRC from multiple published reports was presented by Butterworth, et al in 2006.11 While this meta-analysis comprehensively included the relevant published research to date, limitations in the source studies bring to light additional questions. The majority of studies focused on categorizing ≥1 affected FDR. Additional risk factors in other studies included: affected second-degree relatives (SDRs); age of onset in affected FDRs; sex; and relationship type (i.e., parent or child, sister or brother). Although data on FDRs is the easiest to obtain from patients and may be the most clinically relevant, it is currently not known what impact more distant affected relatives have on risk. Just as important, the risk stemming from various combinations, or ‘constellations,’ of affected relatives has not been adequately explored. Patients frequently present to physicians reporting multiple relatives affected with colorectal cancer. These often include affected relatives from first-, second-, and third-degrees, and might include positive family history from both parents (Figure 1 contains a pedigree diagram to illustrate family relationships and degrees). Physicians presently have little if any data on how to estimate risk for such patients and thereby determine appropriate screening.

Figure 1
Pedigree diagram illustrating family relationships and degrees. The proband is indicated with an arrow. FDR = first-degree relative, SDR = second-degree relative, and TDR = third-degree relative.

The objective of our study was to expand the scope of previous CRC familial risk research in order to measure and report familial relative risk estimates for a variety of specific constellations of family history. This investigation allows risk levels to be assigned for most combinations of affected relatives, thus assisting the physician in making more appropriate screening recommendations. To accomplish this we examined first-, second-, and third-degree risks in a population-based resource with a computerized genealogy linked to statewide cancer registry records. This resource, the Utah Population Database (UPDB), provides an unusual opportunity to investigate the relative risk of CRC in relatives in a large population at a more detailed level than has been previously published.

DATA

The UPDB is a population-based, computerized genealogical resource for Utah containing multiple record-linked data sources including cancer registry records, birth and death certificates, and driver’s license data, that has also been linked to inpatient and outpatient records from the University of Utah Health Sciences Center. The UPDB contains genealogies for the original Utah pioneers (members of The Church of Jesus Christ of Latter-day Saints, or Mormons) and their modern day descendants and was created in the early 1970s using data from the Utah Family History Library.12,13 The original Utah Genealogy included records for 1.6 million individuals who were part of 6–7 generation pedigrees.14 Today the UPDB includes information for about 7 million individuals, although not all have linked genealogical data. Some pedigrees are now more than 11 generations deep.

Of particular interest for this study were Utah Cancer Registry (UCR) records linked to the Utah genealogy. The UCR is a statewide cancer registry established in 1966 that includes records dating back to 1952. Since 1973 the UCR has been part of the Surveillance, Epidemiology, and End Results (SEER) network of National Cancer Institute registries. Ninety-four percent of individuals with cancer link to one or more records in UPDB and 64.2% have family information. The UCR cancer records are coded by disease site according to the International Classification of Diseases of Oncology (ICD-O) and include information on site, stage, grade, age at diagnosis, histology, and patient survival.15 The UCR is careful to report only independent primary sites of cancer in cases where an individual has multiple cancers.

Previous demographic and genetic analyses have shown that the population recorded in the UPDB is genetically representative of US Caucasian and northern European populations16,17,18,19 with a low to normal level of inbreeding.20 The majority of Utahns are members of the LDS Church, which has religious proscriptions against the use of coffee, tea, alcohol, and tobacco. Utah is among the states with the lowest rates of cancer1 and much lower smoking rates may play a role.21

Although the UPDB contains records for about 7 million people, this study utilized a subset of 2.3 million individuals who were part of ≥3 generations of Utah genealogy data and descendants of original Utah pioneers.

Access to the UPDB is governed by the Utah Resource for Genetic Epidemiology (RGE) which was created in 1982.22 RGE and University of Utah Institutional Review Board (IRB) approvals were obtained to access these data and conduct this research. Names and other identifying information were not available to the authors in order to protect the privacy of individuals in the UPDB.

METHODS

Familial relative risk (FRR) in relatives represents the ratio of the risk for a disease among relatives of probands to the risk for the disease in the general population. It is estimated as the number of observed (O) cases among relatives of probands divided by the number of expected (E) cases among the relatives (i.e., FRR = O/E). The expected number of cases among relatives is estimated using population rates. This ratio, also known as a ‘standardized morbidity ratio’, is considered a reasonable approximation of true relative risk when the prevalence of the disease and the true relative risk in the population are low.23 This method of estimating FRR in relatives has been used in previous UPDB analyses of familial risk in cancer 24,25 and other diseases. 26,27

All individuals in the UPDB who are part of ≥3 generations of Utah geneaology data (2.3 million) were assigned to 1 of 264 cohorts based on characteristics that may influence the quality and quantity of genealogical data: birth year (5-year groups), sex, amount of ancestral genealogy (>6 ancestors or not), and birthplace (Utah or not Utah). Internal, cohort-specific CRC rates were calculated by summing the number of CRC cases in each cohort and dividing by the total number of UPDB individuals in that cohort. In this study a proband was defined as an individual who has a particular constellation pattern of affected relatives (e.g., 1 affected FDR, 0 affected SDRs, and ≥3 TDRs), whether or not the proband is a CRC case him/herself. All individuals among the 2.3 million with a particular constellation pattern of affected relatives were considered probands for the corresponding FRR calculation.

After selecting a constellation pattern and determining the group of probands who fit the pattern, the number of observed CRC cases (O) in the group of probands is counted by cohort, without duplication. The expected number of cancers (E) among the defined set of probands is estimated using the formula

E = ∑PiCi/Ni (for i between 1 and 264)

where Pi, Ci, and Ni are the number probands, the number of CRC cases in the UPDB, and the number of individuals in the UPDB, respectively, in the ith cohort group. This method assumes that the morbidity and migration rates for a given cohort of probands is on average the same as that for an equivalent cohort of individuals in the UPDB. For FRR = O/E, P-values were calculated based on the null hypothesis FRR = 1.0 and the alternative hypothesis FRR > 1.0. An assumption was made that the number of observed cases followed a Poisson distribution with the mean equal to the expected number of cases. Confidence intervals for the FRRs were estimated by the method given by Agresti.28

The selection of constellation patterns for which to calculate FRRs was based on common analyses in previous studies11 and consensus of the authors. The first group of analyses performed in this study was based on systematically increasing numbers of relatives within each degree, for instance calculating FRR for those with 1 affected FDR (irrespective of affected SDRs and TDRs), then 2 affected FDRs, then 3, and so forth. Because in a clinical setting individual patients are expected to have varying degrees of family history knowledge, we calculated relative risks for the following situations: (1) only first-degree family history known, (2) only first- and second-degree family history known, and (3) first-, second-, and third-degree family history known. Additional constellation patterns and associated FRR estimates are contained in Supplement Tables A and B. Variations in age at diagnosis of CRC for the affected FDRs and SDRs (Supplement Tables C and D) were also considered. FRRs for ≥1 affected FDRs by type of FDR (e.g., mother/father, brother/sister, parent/sibling/child, offspring, and male/female) were estimated as well as the FRR for having both an affected mother and an affected father (Supplement Table E). We also estimated FRRs based on whether a proband’s family history was concentrated solely on one side of a family versus both (Supplement Table F).

RESULTS

A total of 2,327,327 individuals included in ≥3 generation family histories were included in this analysis. Among these individuals included in the study, 10,556 were identified with a primary diagnosis of CRC. Based on consensus among the authors, an FRR estimate ≥2.0 with the lower confidence interval >1.0 was considered a clinically relevant cutoff for elevated risk. All of the constellations producing FRR estimates meeting this clinically relevant criteria are presented in Table 1Table 5, along with selected others which are presented for comparison.

Table 1
Selected familial relative risks estimates for probands considering only first-degree relative family history. ‘NA’ means this category of relative was not considered.
Table 5
Selected familial relative risk estimates for probands with various affected first-degree relationship types as well as for male and female probands with ≥1 affected FDR. ‘NA’ means this category of relative was not considered. ...

First-degree risk

For those probands with no affected relatives (FDRs, SDRs, TDRs) (N=1,460,367), FRR = 0.83 (95% CI 0.81–0.86). Table 1 includes FRR estimates for probands with increasing numbers of affected FDRs, without respect to SDRs or TDRs. The most commonly published FRR is for ‘≥1 affected FDR.’ We estimated FRR = 2.05 (95% CI 1.96–2.14) for probands with ≥1 affected FDR, similar to the meta-analysis FRR = 2.07 (95% CI 1.89–2.26) that was adjusted to account for suspected publication bias among source studies.11

Second-degree risk

Table 2 presents FRR estimates for constellations with 0 or 1 affected FDRs and increasing numbers of affected SDRs. A positive second-degree family history (in the absence of a positive first-degree family history) can be associated with increased risk, but does not appear to be of the same magnitude as a positive first-degree family history. Second-degree family history does appear to affect risk when combined with first-degree family history. The FRR for 1 affected FDR with 1 affected SDR = 2.12 (95% CI 1.90–2.35), significantly higher than the FRR for 1 affected FDR and 0 affected SDRs: FRR = 1.82 (95% CI 1.72–1.93) (p = 0.007). The FRR estimate for 1 FDR and ≥3 affected SDRs is even higher; FRR= 3.37 (95% CI 2.20–4.93), and in fact exceeds the estimated FRR for 2 affected FDRs.

Table 2
Familial relative risk estimates for probands with 0 or 1 affected FDRs and increasing numbers of affected SDRs. ‘NA’ means this category of relative was not considered.

Third-degree risk

Table 3 presents selected FRR estimates for constellations with 0 or 1 affected FDRs and various combinations of affected SDRs and TDRs. A positive third-degree family history, in the absence of positive first- and second-degree family histories, does not confer a significant increased risk. For example, the FRR estimate for probands with 0 affected FDRs, 0 affected SDRs, and ≥3 affected TDRs = 1.08 (95% CI 0.97–1.20). However, in combination with positive first- and second-degree family histories, third-degree family history can make a contribution to the total risk that is significant. As an example, the FRR for probands with 1 affected FDR, 1 affected SDR, and 0 affected TDRs = 1.88 (95% CI 1.59–2.20), increasing to a significantly higher FRR = 3.28 (95% CI 2.44–4.31) for probands with 1 affected FDR, 1 affected SDR, and ≥3 affected TDRs (p = 0.004). In the absence of a positive second-degree family history, presence of affected third-degree relatives does not appear to significantly change FRR (which is significantly > 1 for both constellations); for probands with 1 affected FDR, 0 affected SDRs, and ≥3 affected TDRs FRR = 2.01 (95% CI 1.61–2.47) versus FRR = 1.76 (95% CI 1.63–1.89) for probands with 1 affected FDR, 0 affected SDRs, and 0 affected TDRs (p = 0.125).

Table 3
Selected familial relative risk estimates for probands with either 0 or 1 affected FDRs and various combinations of affected SDRs and TDRs.

Age-related risk

Table 4 contains elevated FRR estimates based on the age of diagnosis of affected FDRs. Typically disease onset <50 years is considered to be early for CRC, but we analyzed ages at diagnosis of 60 and 70 years as cutoffs as well. We estimate that for individuals with ≥1 affected FDR diagnosed <50 years of age FRR = 3.31 (95% CI 2.79–3.89); this is significantly higher than the estimate for ≥1 affected FDR when the age of diagnosis was ≥50 (FRR = 2.02 (95% CI 1.93–2.11) (p = 9.3e-9). However, when the diagnosis age (of affected FDRs) was limited to between 60 and 69 years of age, the FRR estimate of 2.22 (95% CI 2.04–2.40) was still elevated above the chosen cutoff. In fact, it was significantly higher (p = 0.045) than the FRR for probands with ≥1 affected FDR when the age of diagnosis was not considered: FRR = 2.05 (95% CI 1.96–2.14), although for counseling purposes these numbers are not dissimilar.

Table 4
Selected familial relative risks for probands with affected FDRs or SDRs diagnosed at certain ages. ‘NA’ means this category of relative was not considered.

When considering second degree relatives, age of diagnosis of the affected relative also affects risk. The FRR estimate for ≥1 affected SDR (without respect to FDRs and TDRs) was 1.27 (95% CI 1.22–1.33). The estimate for ≥1 affected SDR diagnosed <50 years of age, FRR = 1.84 (95% CI 1.61–2.09), was significantly higher (p = 6.2e-8).

Relationship type- and sex-related risks

We also estimated FRRs for specific FDR relationship types, shown in Table 5. No difference was found between FRRs for those with an affected parent versus an affected sibling. Differences between FRRs estimated for individuals with affected brothers and sisters, and individuals with affected mothers and fathers were also not significant. A statistically significant difference (of small magnitude) was observed between female and male probands with ≥1 affected FDR, with such females having FRR = 2.12 (95% CI 2.00–2.26) versus FRR = 1.96 for males (95% CI 1.84–2.09) (p = 0.04). Of particular interest, and not previously published, was the risk estimate for those with both an affected mother and an affected father, the highest FRR we observed for any group of probands: FRR = 4.97 (95% CI 2.72, 8.34).

In Table 5 we noted that the FRR when both parents are affected was 4.97 (95% CI 2.72–8.34). This is increased, although not quite significantly (p = 0.07), over the FRR for ≥2 affected FDRs when probands with both an affected mother and father are excluded (FRR = 3.21; 95% CI 2.87–3.58). This rather rare occurrence may represent the situation of more than 1 predisposing genes segregating in the offspring of the 2 affected parents. Therefore, we also investigated FRRs for a pattern of family history involving cases on both sides of the family versus cases on one side of the family only. To examine this issue, we compared FRRs for probands with ≥2 affected SDR relatives, separately for the case of ≥2 affected relatives on the same side of the family, and then for ≥1 affected relative on each side of the family. We similarly made the comparison for probands with ≥2 affected TDRs, again separately for the case of ≥2 affected relatives on the same side of the family, and then for ≥1 affected TDR on each side of the family. Results are shown in Supplement Table F. We excluded all probands with any affected FDRs. In neither analysis was there any significant difference in the FRR estimates, whether the family history was one-sided or both-sided.

DISCUSSION

The purpose of this study was to define risk estimates for CRC based on family history data from a homogeneous, well-characterized population in order to better assist physicians in determining CRC risk in their patients. We used a large genealogical and cancer registry resource, the UPDB, to calculate FRRs for various constellations of family history risk for CRC. Characteristics of the Utah population represented in the UPDB data include extended relationships, large family sizes, low outmigration, low to normal inbreeding, and a genetic composition similar to the US Caucasian population. Cancer records from the UCR, which are included in the UPDB dataset, strictly define the disease of interest. Since the UPDB includes comprehensive statewide cancer records from the UCR it is free of ascertainment and recall bias that might affect other studies that rely on interviews with probands to assess cancer in relatives. This lack of ascertainment bias is a particular strength of this resource.

Many previous studies have shown increased CRC risks for relatives of affected individuals. While the more common FRR estimates such as ‘≥1 affected FDR’ that we report here are consistent with studies included in the Butterworth meta-analysis, generally our estimates are lower with tighter confidence intervals. These differences may be due to the larger number of individuals included in our analysis and perhaps differences in the incidence of disease between our population and those in other studies. Utah has the lowest incidence of CRC in the United States for both men and women, 47.5 and 35.2 per 100,000, respectively.1

In the Butterworth meta-analysis sibling risk (RR = 2.79; 95% CI 2.36–3.29) was reported to be higher than parent-offspring risk (RR = 2.07; 95% CI 1.83–2.34) and the authors suggested that this may indicate the presence of recessive genetic factors causing a susceptibility to CRC. We found no such difference between sibling and parent-offspring risk, even though the numbers of probands with an affected parent or affected sibling included in the analysis were greater than 31,000 and 47,000, respectively. The Utah study’s population-based approach avoids the challenges of combining studies with different methods and ascertainment bias as well as accumulating higher numbers of patients for analysis thus making this study’s results more robust.

Although we were limited in our ability to explore increased risk associated with whether 1 or more CRC predisposition genes had an opportunity to segregate in a pedigree, we only observed a clearly significant effect for individuals with both mother and father affected. Having both an affected mother and affected father confers a higher degree of risk (FRR = 4.97, 95% CI 2.72–8.34) than having ≥2 affected FDRs (FRR = 3.26, 95% CI 2.92–3.63). Only a small number of probands had two affected parents (n=450), but this is an interesting result that may be worth exploring in other large population sets. The elevated FRR in individuals with both parents affected could result from gene × gene interaction, gene × environment interaction, or a combination of both. Our other FRR comparisons for second- and third-degree family history from one side of the family versus from both sides of the family did not show any significant differences.

Based on the FRR estimate for having no affected FDRs, SDRs, nor TDRs (FRR = 0.83, 95% CI 0.81–0.86) individuals with no known family history have a mild but significant protection, as might be expected (given that our overall risk estimates for all groups considered must average 1.0). Because this risk estimate relies on information on current age of all FDRs, SDRs, and TDRs (information which would not be typically available in a clinical setting) clinical recommendations for individuals with no known first-, second-, or third-degree positive family history should include standard risk estimates and screening recommendations based on their current age.

Increased numbers of affected FDRs influence risk much more than affected SDRs or TDRs. In fact, in the absence of a positive first-degree family history (i.e., 0 affected FDRs) and considering both affected SDRs and TDRs (but not age of diagnosis in affected relatives) only 1 constellation had an FRR estimate that was significantly > 1.0 in the CI (0 affected FDRs, 1 affected SDR, 2 affected TDRs; FRR=1.33, 95% CI 1.13–1.55). However, we previously noted that when combined with positive first-degree family history, the presence of positive second- and third-degree family history can significantly increase risk.

Age at diagnosis of CRC in affected relatives contributes significantly to risk estimates. While an age at diagnosis <50 years typically has been used as a cutoff for early onset, we have shown that even diagnosis between 60 and 69 years of age in affected FDRs increases risk equivalent to the level of an affected FDR without respect to age at diagnosis. Therefore, older age of onset in a FDR should not be viewed as reassuring to the patient. Additionally, even the age of onset in second-degree relatives (<50 years) can have an effect on the proband’s risk (FRR = 1.84, 95% CI 1.61–2.09 versus FRR=1.27, 95% CI 1.22–1.33 for ≥1 affected SDR without respect to age at diagnosis).

Precisely how our findings can be applied to colon cancer screening recommendations has yet to be determined, but extrapolation from current guidelines appears to be of some benefit. In the most current colon cancer screening recommendations, family histories that represent a two- to three-fold increased risk (usually any first-degree relative with colon cancer diagnosed over age 60 years) suggest that colon cancer screening, as recommended for the general population, is indicated. 29,30 Specifically this includes any one of the screening tools now used (annual fecal occult blood testing, every-five-year sigmoidoscopy, combination of the first two, every-5-year barium enema or CT colonography, or every-10-year colonoscopy). The only difference in recommendations for individuals with this level of familial risk is that they should start at age 40 years, rather than 50 years. This is because this group exhibits the same risk at age 40 years as the general population at age 50 years. Individuals with a risk of three-fold or greater compared to the general population because of family history (included are those with an FDR diagnosed < age 60 or two FDRs with colon cancer) are now recommended to have colonoscopy as the screening tool of choice, starting at age 40 (or 10 years younger than the earliest diagnosis in the family) and have repeat colonoscopy every five years thereafter. Colonoscopy findings may alter these recommendations.

In view of these widely accepted guidelines, we would suggest provisionally that constellations of family risk that result in approximately two-fold increased risk or approximately three-fold or greater risk be screened accordingly. A brief set of rules that specify constellations that meet these criteria is presented in Table 6. Those individuals with a very strong family history should always be considered for one of the inherited syndromes of colon cancer. Physicians should encourage individuals with increased, but below two-fold, risk to be screened according to guidelines for average risk.

Table 6
Family history constellations that will produce approximately 2-fold and approximately 3-fold and higher familial relative risk (FRR) estimates

This study has provided evidence that the existence of affected extended relatives increases risk for CRC in probands. However, clinicians may question whether many patients typically have valid family history information for relatives more distant than first-degree and whether the effort required to document and utilize this information in clinical practice is cost- and time-effective. With more people taking an interest in family history, and a growing number of electronic tools and standards for documenting and sharing family health histories, the collection and clinical use of data from patients on family health histories beyond the first-degree and may be reasonable in the near future.31,32

With regard to the limitations of the study, these results may not be generalizable to other populations with different racial or ethnic compositions. The Utah population has been shown to be very representative of the US Caucasian and Northern European populations. Other potential limitations include the reliance on appropriate cancer diagnosis coding and inability to capture relatives not represented in the UPDB genealogy or with cancer diagnosed outside the state or outside the UCR time period. We have not excluded individuals from our analysis with familial forms of CRC such as hereditary nonpolyposis colorectal cancer (HNPCC) since they may not be reliably identified; one may wonder if pedigrees containing individuals with these conditions have skewed the risk estimates. However, in a previous UPDB study the number of individuals meeting the Amsterdam I criteria was estimated to be small (65 out of 9458 cases or 0.7% of the cases) and none of these individuals had a histology indicating familial adenomatous polyposis syndrome.21 Finally, we observed certain constellations where the corresponding FRR estimates did not follow the anticipated trend. As an example from Table 3, FRR = 2.37 (95% CI 1.58–3.43) for probands with 1 affected FDR, 2 affected SDRs, and 0 affected TDRs. For those with 1 affected FDR, 2 affected SDRs, and 2 affected TDRs, FRR = 2.70 (95% CI 1.44–4.62). However, for probands with 1 affected FDR, 2 affected SDRs, and 1 affected TDR, FRR = 1.98 (95% CI 1.15–3.17). While there is clearly a pattern of increasing FRR for increasing numbers of affected relatives, individual estimates were not always consistent with the trend and small sample size may be a factor.

CONCLUSION

In summary, this study is unique in providing definitions of CRC risk based on first-, second-, and third-degree family history constellations that have not been reported previously. These risk estimates were based on computerized genealogy and cancer registry data for large numbers of individuals from a well-defined population. We have demonstrated that while influencing risk to a lesser extent than first-degree family history, positive second- and third-degree family histories can have a significant impact on an individual’s risk for CRC. We have also demonstrated how the age of cancer onset (estimated by age at cancer diagnosis) in relatives affects risk. We have provided a comprehensive set of supplemental tables which accommodate varying degrees of family history knowledge, which can be used to more precisely define CRC risk.

With respect to future work, producing absolute risk calculations in real-time from an individual’s family history constellation and current age based on the FRR estimates presented could be automated. A computerized CRC family history risk prediction tool could be created as part of a personal health record (PHR) application or as a decision support component in an electronic health record (EHR). While family history is an important risk factor for CRC, clinical, environmental, behavioral factors are also important but how they affect genetic susceptibility is uncertain. We are currently working to create a more comprehensive CRC risk prediction model based on a combined set of family history and clinical data for a subset of the individuals included in this current study. It is hopeful this will provide additional insight on the contributions of family history as well as other factors on total CRC risk.

Supplementary Material

ACKNOWLEDGMENTS

The authors thank Steve Backus for database and software support and Jim Farnham and Kristina Allen-Brady for statistical analysis assistance. Research was supported by the Utah Cancer Registry, which is funded by contract N01-PC-35141 from the National Cancer Institute's SEER program with additional support from the Utah State Department of Health and the University of Utah. Partial support for all datasets within the UPDB was provided by the University of Utah Huntsman Cancer Institute. Additional support was from R01, National Library of Medicine grant LM009331 (Lisa Cannon Albright), National Cancer Institute grants R01-CA40641 and PO1-CA73992 (Randall Burt), and an Intermountain Healthcare Homer Warner Center for Informatics Research Fellowship (David Taylor).

Abbreviations

BMI
body mass index
CRC
colorectal cancer
EHR
electronic health record
FDR
first-degree relative
FOBT
fecal occult blood test
FRR
familial relative risk
HNPCC
hereditary nonpolyposis colorectal cancer
ICD-O
International Classification of Diseases of Oncology
LDS
The Church of Jesus Christ of Latter-day Saints
NSAID
non-steroidal anti-inflammatory drug
PHR
personal health record
RGE
Utah Resource for Genetic Epidemiology
SDR
second-degree relative
SEER
Surveillance, Epidemiology, and End Results network
TDR
third-degree relative
UCR
Utah Cancer Registry
UPDB
Utah Population Database

Footnotes

Work performed at:

Department of Genetic Epidemiology, University of Utah, 391 Chipeta Way, Suite D, Salt Lake City, UT 84108-1266

Disclosures/Conflicts of Interest:

David Taylor: None

Randall Burt: Consultant for Myriad Genetics, but no conflict.

Marc Williams: None

Peter Haug: None

Lisa Cannon-Albright: None

References

1. Jemal A, Siegel R, Ward E, et al. Cancer statistics, 2008. CA Cancer J Clin. 2008;58:71–96. [PubMed]
2. Levin B, Lieberman DA, McFarland B, et al. Screening and surveillance for the early detection of colorectal cancer and adenomatous polyps, 2008: A joint guideline from the American Cancer Society, the US Multi-Society Task Force on Colorectal Cancer, and the American College of Radiology. CA Cancer J Clin 58:130–160, 2008, or Gastroenterology. 2008 May;134(5):1570–1595. [PubMed]
3. Smith RA, Cokkinides V, Eyre HJ. Cancer screening in the United States, 2007: a review of current guidelines, practices, and prospects. CA Cancer J Clin. 2007;57:90–104. [PubMed]
4. Meissner HI, Breen N, Klabunde CN, Vernon SW. Patterns of colorectal cancer screening uptake among men and women in the United States. Cancer Epidemiol Biomarkers Prev. 2006;15:389–394. [PubMed]
5. Palmer RC, Emmons KM, Fletcher RH, et al. Familial risk and colorectal cancer screening health beliefs and attitudes in an insured population. Prev Med. 2007 Nov;45(5):336–341. [PubMed]
6. Fuchs CS, Giovannucci EL, Colditz GA, et al. A prospective study of family history and the risk of colorectal cancer. N Engl J Med. 1994;331(25):1669–1674. [PubMed]
7. Johns LE, Houlston RS. A systematic review and meta-analysis of familial colorectal cancer risk. Am J Gastroenterol. 2001;96(10):2992–3003. [PubMed]
8. St. John DJ, McDermott FT, Hopper JL, et al. Cancer risk in relatives of patients with common colorectal cancer. Ann Intern Med. 1993;118(10):785–790. [PubMed]
9. Goldgar DE, Easton DF, Cannon-Albright LA, Skolnick MH. Systematic population-based assessment of cancer risk in first-degree relatives of cancer probands. J Natl Cancer Inst. 1994 Nov 2;86(21):1600–1608. [PubMed]
10. Hemminki K, Li X. Familial colorectal adenocarcinoma from the Swedish Family-Cancer Database. Int J Cancer. 2001;94(5):743–748. [PubMed]
11. Butterworth AS, Higgins JP, Pharoah P. Relative and absolute risk of colorectal cancer for individuals with a family history: a meta-analysis. Eur J Cancer. 2006 Jan;42(2):216–227. [PubMed]
12. Skolnick MH. The Utah genealogical data base: a resource for genetic epidemiology. In: Cairns J, Lyon JL, Skolnick MH, editors. Banbury Report No 4; Cancer. Incidence in Defined Populations. New York: Cold Spring Harbor Laboratory; 1980. pp. 285–297.
13. Skolnick MH. Prospects for population oncogenetics. In: Mulvihill JJ, Miller RW, Fraumeni JF, editors. Genetics of human cancer. New York: Raven Press; 1977. pp. 19–25.
14. Cannon Albright LA. Utah family-based analysis: past, present and future. Hum Hered. 2008;65(4):209–220. [PubMed]
15. World Health Organization. International Classification of Diseases for Oncology. 3rd. Geneva, Switzerland: World Health Organization;
16. Cannon-Albright LA, Thomas A, Goldgar DE, et al. Familiality of cancer in Utah. Cancer Res. 1994 May 1;54(9):2378–2385. [PubMed]
17. McLellan T, Jorde LB, Skolnick MH. Genetic distances between the Utah Mormons and related populations. Am J Hum Genet. 1984 Jul;36(4):836–857. [PubMed]
18. Jorde LB, Shortsleeve PA, Henry JW, et al. Genetic analysis of the Utah population: a comparison of STR and VNTR loci. Hum Biol. 2000 Dec;72(6):927–936. [PubMed]
19. Cannon-Albright LA, Farnham JM, Thomas A, Camp NJ. Identification and study of Utah pseudo-isolate populations-prospects for gene identification. Am J Med Genet A. 2005 Sep 1;137A(3):269–275. [PubMed]
20. Jorde LB. Inbreeding in the Utah Mormons: an evaluation of estimates based on pedigrees, isonymy, and migration matrices. Ann Hum Genet. 1989 Oct;53(Pt 4):339–355. [PubMed]
21. Lyon JL, Gardner JW, Klauber MR, Smart CR. Low cancer incidence and mortality in Utah. Cancer. 1977 Jun;39(6):2608–2618. [PubMed]
22. Wylie JE, Mineau GP. Biomedical databases: protecting privacy and promoting research. Trends Biotechnol. 2003;21:113–116. [PubMed]
23. Chaturvedi AK, Mbulaiteye SM, Engels EA. Underestimation of relative risks by standardized incidence ratios for AIDS-related cancers. Ann Epidemiol. 2008 Mar;18(3):230–234. [PubMed]
24. Maul JS, Warner NR, Kuwada SK, et al. Extracolonic cancers associated with hereditary nonpolyposis colorectal cancer in the Utah Population Database. Am J Gastroenterol. 2006 Jul;101(7):1591–1596. [PubMed]
25. Allen-Brady K, Camp NJ, Ward JH, Cannon-Albright LA. Lobular breast cancer: excess familiality observed in the Utah Population Database. Int J Cancer. 2005 Nov 20;117(4):655–661. [PubMed]
26. Cannon-Albright LA, Camp NJ, Farnham JM, et al. Genealogical assessment of heritable predisposition to aneurysms. J Neurosurg. 2003;99:637–643. [PubMed]
27. Horne BD, Camp NJ, Muhlestein JB, Cannon-Albright LA. Identification of excess clustering of coronary heart diseases among extended pedigrees in a genealogical population database. Am Heart J. 2006;152:305–311. [PubMed]
28. Agresti A. Categorical data analysis. New York: Wiley; 1990.
29. Winawer S, Fletcher R, Rex D, et al. Colorectal cancer screening and surveillance: clinical guidelines and rationale-Update based on new evidence. Gastroenterology. 2003 Feb;124(2):544–560. [PubMed]
30. Burt RW. Colon cancer screening. Gastroenterology. 2000 Sep;119(3):837–853. [PubMed]
31. Simon C, Acheson L, Burant C, et al. Patient interest in recording family histories of cancer via the Internet. Genet Med. 2008 Dec;10(12):895–902. [PMC free article] [PubMed]
32. Feero WG, Bigley MB, Brinner KM. Family Health History Multi-Stakeholder Workgroup of the American Health Information Community. New standards and enhanced utility for family health history information in the electronic health record: an update from the American Health Information Community's Family Health History Multi-Stakeholder Workgroup. J Am Med Inform Assoc. 2008 Nov–Dec;15(6):723–728. [PMC free article] [PubMed]