Hereditary Gastrointestinal Cancer Registry, Huntsman Cancer Institute, University of Utah, Salt Lake City, Utah
Division of Gastroenterology, Department of Medicine, University of Utah, Salt Lake City, Utah
Corresponding author: N. Jewel Samadder, MD, MSc, FRCPC, Hereditary Gastrointestinal Cancer Registry, Huntsman Cancer Institute, 2000 Circle of Hope, Salt Lake City, UT 84112; Fax: (801) 581-7476; firstname.lastname@example.org
Guidelines recommend that individuals with a first-degree relative (FDR) diagnosed with colorectal cancer (CRC) or advanced adenoma before age 60 years should undergo colonoscopy starting at age 40 years. The authors quantified the risk of adenomas and CRC in FDRs, second-degree relatives (SDRs), and third-degree relatives (TDRs) of patients diagnosed with adenomas and advanced adenomas.
A population-based, retrospective, case-control study was performed of residents of the state of Utah aged 50 years to 80 years who underwent colonoscopy between 1995 and 2009 at Intermountain Healthcare or the University of Utah. Controls were selected from the population of colonoscopy patients who were free of adenomas or CRC and matched to each case based on sex and birth year. Colonoscopy results were linked with cancer and pedigree information from the Utah Population Database to investigate the familial aggregation of adenomas and CRC using Cox regression analysis. The unit of analysis was the relatives of cases and controls.
Of 126,936 patients who underwent colonoscopy, 43,189 had adenomas and 5563 had advanced adenomas and defined the case population. An elevated risk of CRC was found in FDRs (relative risk [RR], 1.35; 95% confidence interval [95% CI], 1.25-1.46), SDRs (RR, 1.15; 95% CI, 1.07-1.23) of adenoma cases, and in FDRs of advanced adenoma cases (RR, 1.68; 95% CI, 1.29-2.18) compared with controls. Approximately 10% of CRCs diagnosed in relatives would have been missed if the current screening guidelines were strictly adhered to.
Colorectal cancer (CRC) is the third most common cancer in the United States and the second most common cause of cancer-related mortality. The discovery of several high-penetrance susceptibility genes (APC, MLH1, MSH2, MSH6, and PMS2) has lead to a greater awareness of the role of hereditary risk factors for CRC. Familial clustering of CRC is common even when the cases are not part of a defined genetic syndrome. Adenomatous polyps are accepted as the precursor lesion for most cases of CRC. Screening for CRC is widely believed to provide an opportunity to detect patients at risk for developing CRC, remove precursor lesions, or detect patients at an earlier stage of cancer. Adenomas have a higher potential for malignancy both at the time of diagnosis and on follow-up when they measure > 10 mm in diameter and/or present with severe dysplasia and/or a villous component, at which point they are then defined as advanced adenomas (AAs).
Screening interventions such as colonoscopy are offered earlier to individuals with such a family history. Specifically, current guidelines recommend that patients with a first-degree relative (FDR) diagnosed with CRC or AA before age 60 years should undergo screening colonoscopy starting at age 40 years or 10 years before the index patient and undergo repeat surveillance every 5 years.[3-6] The current recommendations are primarily based on a prospective study by Fuchs et al that found that FDRs of patients with CRC had a risk of CRC at age 40 years that was similar to the risk of CRC in average-risk patients at the age of 50 years (relative risk [RR], 1.72; 95% confidence interval [95% CI], 1.34-2.19).
Because the lifetime risk of CRC is approximately 6%, millions of Americans will be counseled to undergo colonoscopy at age 40 years, based on a positive family history of CRC or AA. Therefore, it is important to validate the elevated risk of CRC in the relatives of patients with adenomas or AAs in a population-based study, and to examine that risk in immediate and more distant relatives, as well as by age groups.
In this retrospective, population-based, case-control study, we quantified the risk of CRC and adenomas in the FDRs, second-degree relatives (SDRs), or third-degree relatives (TDRs; first cousins) of individuals with an adenoma or advanced adenoma, stratified by age at diagnosis. To the best of our knowledge, the current study is the first population-based assessment of the risk of colorectal neoplasia in distant relatives of individuals with an adenomatous polyp and also examines this risk based on the age of the patient at the time of the adenoma diagnosis. Comprehensive family histories were available through extensive Utah genealogies linked to a statewide cancer registry and medical records that do not rely on self-report. The design of the current study is feasible because of these unique linked resources.
MATERIALS AND METHODS
This study was approved by the Institutional Review Boards of the University of Utah and Intermountain Healthcare (IHC) and by the Utah Resource for Genetic and Epidemiologic Research (www.research.utah.edu/rge/), an administrative oversight board charged by the University of Utah to govern access to the Utah Population Database (UPDB).
We performed a population-based, retrospective, case-control study of residents in the state of Utah aged 50 years to 80 years who underwent colonoscopy from February 15, 1995 to January 31, 2009 at IHC and the University of Utah Health Sciences (UUHSC) clinical facilities. Deidentified medical information regarding patients undergoing colonoscopy at IHC and UUHSC was merged with family structure in the UPDB genealogies, which include cancer histories from the Utah Cancer Registry (UCR), a Surveillance, Epidemiology, and End Results (SEER) registry, to investigate the familial aggregation of CRC and colon adenomas.
Description of Databases
The current investigation takes advantage of unique Utah databases. The study required patient-level data integration between IHC, the UUHSC, and the UPDB. The UPDB combines genealogies often dating back to the early 1800s with data from statewide resources, including the UCR, statewide inpatient discharge and ambulatory surgery records, and driver license data, as well as birth and death certificates. This resource also has been linked to the demographic records from the UUHSC and IHC. In combination, the UUHSC and IHC together provide cancer-related care to > 80% of the Utah population. Previous demographic and genetic analyses have shown that the population recorded in the UPDB is genetically representative of US white and northern European populations, with a low level of inbreeding. Of particular interest for the current study is the inclusion of the UCR records as part of the UPDB. The UCR is a statewide cancer registry that was established in 1966, and since 1973 it has been part of the SEER network of the National Cancer Institute registries. Given an ongoing and accurate assessment of a family history of cancer that does not depend on self-report, the UPDB provides a valuable resource for a thorough analysis of the familial nature of colorectal neoplasia.
Linkage of Electronic Medical Record Data to the UPDB
These linked resources have been used to assess colonoscopy screening rates in high-risk individuals as well as recent studies on preeclampsia, spontaneous preterm delivery, cancer in twins, heritability of inflammatory bowel disease, and effects of family conditions on later-life mortality.
Colonoscopy was extracted from the institutional records using Current Procedural Terminology codes 45378, 45391, or 45392. Case subjects were defined as having adenomas detected during their colonoscopy as outlined in Figure 1. Adenomas were identified through pathology reports. For the purposes of the current study, AAs were defined as those that had any component of villous histology, also identified through pathology reports. We could not identify the other criteria associated with a definition of AA (measuring > 10 mm or high-grade dysplasia). The analysis evaluated the RR of CRC and colorectal neoplasia diagnosed in the FDRs, SDRs, and TDRs of cases compared with relatives of controls.
For each case patient, 1 randomly selected, age- and sex-matched control was used. These controls were sampled without replacement and were selected from the population of colonoscopy patients who did not have adenomatous polyps removed at the time of any colonoscopy performed between 1995 and 2009, but they could have had other findings. Controls had to have follow-up in the UPDB that was at least as long as their respective matched case. Follow-up was based on the most recent date an individual had an event recorded in Utah from vital records (deaths, births, adoptions, marriages, and divorces; Utah driver's license registrations and renewals; voter registrations; and statewide inpatient and ambulatory surgeries). Records are linked at least annually.
Using software developed specifically for UPDB kinship analysis, the magnitude of familial risk was estimated by Cox regression analysis to assess the RR of CRC and adenoma for FDRs, SDRs, and TDRs of colonoscopy patients who themselves were diagnosed with adenomas. Because observations within families are nonindependent, a robust variance estimator for clustered data similar to the generalized estimating equation approach was incorporated. This analysis corrects for the issue of families being analyzed multiple times because of multiple cases of CRC. The percentage of FDRs diagnosed with CRC based on age groups was also computed.
The current study analyzed the RRs of colorectal neoplasia from 27,646 colonoscopy patients at UUHSC and 103,680 colonoscopy patients from IHC. When combined and accounting for patients who had procedures performed in both hospital systems, there were 126,936 unique individuals who had received a colonoscopy. A total of 43,189 patients had an adenomatous polyp removed at their index or first colonoscopy and defined the case population. For cases of AA, we used 5563 patients who had an adenoma with villous features removed at the index or first colonoscopy.
Table 1 shows the descriptive features of cases (colonoscopy patients with a finding of an adenomatous polyp) and controls in the current study. The mean age at the time the procedure was performed for both cases and controls was similar at 61 years and 60 years, respectively. For cases and controls, the mean number of relatives with colonoscopies was similar, thereby reducing the likelihood of detection bias. Colonoscopy use among FDRs was approximately 30%, which is similar to statewide use of CRC screening in Utah.
Table 1. Descriptive Characteristics of Colonoscopy Patients With Adenomas (Cases) and Matched Controls
Cases diagnosed with adenomas at time of colonoscopy between 1995 and 2009.
Controls selected from colonoscopy cohort, without findings of adenomas or colorectal cancer. Matched by sex and birth year and had follow-up at least as long as the date of the case diagnosis.
Mean age at procedure, y
Mean no. of FDRs with colonoscopy
Mean no. of SDRs with colonoscopy
Mean no. of TDRs with colonoscopy
Tables 2 and 3 show the distribution of CRC among relatives of cases with adenoma and AA and controls based on the degree of relation. An elevated risk of CRC was found in FDRs of cases with adenomas (RR, 1.35; 95% CI, 1.25-1.46) and advanced adenomas (RR, 1.68; 95% CI, 1.29-2.18) compared with relatives of controls. Although the SDRs of cases who had an adenoma removed had an elevated risk of CRC (RR, 1.15; 95% CI, 1.07-1.23), the SDRs of cases with AAs did not. The TDRs of cases with adenomas were found to have a slightly elevated risk of CRC (RR, 1.06; 95% CI, 1.00-1.12).
Table 2. Relative Risk of CRC in Relatives of Cases With Adenomas
CRC in Relatives of Probands Diagnosed With Adenomas
Using the adenoma and advanced adenoma cases and controls cited above, we also examined their relatives' risk of developing similar polyps (Tables 4 and 5). The RR of an adenoma detected in a FDR of a patient affected by a similar polyp was 1.33 (95% CI, 1.26-1.40). This association demonstrated a familial trend, dependent on the degree of relationship, and decreasing as the level of the relationship became more distant. The RRs for an adenoma in a SDR was 1.15 (95% CI, 1.09-1.20), and was 1.09 (95% CI, 1.06-1.12) in a TDR. Similarly, the RR of an advanced adenoma in a FDR affected by a similar advanced polyp was 1.65 (95% CI, 1.28-2.14), and was 1.06 (95% CI, 0.83-1.35) in an SDR and 1.96 (95% CI, 1.3-2.95) in a TDR.
Table 4. Relative Risk of Adenomas in Relatives of Cases With Adenomas
Adenomas in Relatives of Probands Diagnosed With Adenomas
When stratified by age at the time of the diagnosis of colorectal neoplasia in cases, the FDRs and SDRs of those diagnosed with an adenoma at age < 60 years had a greater risk of CRC compared with relatives of controls (RR, 1.41; 95% CI, 1.27-1.56). However, even for cases diagnosed with an adenoma at age ≥ 60 years, FDRs had a moderately elevated risk of CRC (RR, 1.23; 95% CI, 1.07-1.42) (Table 6). A similar elevated risk of CRC was observed in the FDRs of cases with advanced adenoma regardless of their age at the time of polyp diagnosis (Table 7).
Table 6. Relative Risk of CRC in Relatives of Cases With Adenomas, Stratified by Age at Diagnosis
Tables 8 and 9 show the breakdown by age group of FDRs of cases diagnosed with adenoma and advanced adenoma stratified by age and demonstrate that 10% of CRCs would have been missed using the current guidelines, which do not advise screening at a younger age for FDRs of individuals diagnosed with an adenoma at age < 60 years.
Table 8. Percentage of FDRs Diagnosed With CRC Based on Age Groups, Stratified by Age of Case With Adenoma
In the current study, we report that FDRs, SDRs, and TDRs of patients with adenomas and advanced adenomas have an increased risk of developing adenomas or CRC themselves. Whereas FDRs of patients with adenomas were found to have only a moderately increased risk of colorectal neoplasia of 35%, the FDRs of patients with advanced adenomas had a nearly 70% elevated risk of developing colorectal neoplasia. This supports current guidelines for increased CRC surveillance in the relatives of patients with advanced adenomas. Although this observation in FDRs could be related in part to a shared environment, this is unlikely to be true for SDRs and TDRs.
The results of the current study add substantially to the available literature in view of its number of inherent strengths. First, it defines precise risk levels for both CRC and adenomatous polyps not only in FDRs but also in the SDRs and TDRs of patients with adenoma and AAs. Second, it is population-based, which avoids many of the recruitment biases inherent to referral-based and registry-based studies. Third, patient recall biases, which occur frequently in many family studies, are avoided by using primary medical records that are linked to cancer registry and genealogical records. Fourth, cases are defined as those who have undergone a colonoscopy, which reflects those patients seen by gastroenterologists in clinical practice. Finally, the numbers of cases and controls are sufficiently large to make precise cancer and adenoma risk estimates in various relatives by age group.
Two case-control design studies from France found an elevated risk of developing adenomas or CRC in the FDRs of patients with large adenomas.[20, 21] This was also supported by a smaller US study that demonstrated only a modest and nonsignificant increase in the risk of CRC in the relatives of patients with small tubular adenomas (odds ratio [OR], 1.26; 95% CI, 0.99-1.61) and a significant increase in the relatives of patients with AAs (those measuring > 10 mm in size and/or with a villous component or high-grade dysplasia), with an OR of 1.62 (95% CI, 1.16-2.26). This is similar to the current study findings of a moderate 35% increased risk of CRC in the FDRs of patients with adenoma (OR, 1.35; 95% CI, 1.25-1.46) and a nearly 70% elevated risk of CRC in the FDRs of patients with villous adenomas (OR, 1.68; 95% CI, 1.29-2.18). Similarly, in a 2001 systematic review, 8 of 9 studies found an increased risk of CRC associated with adenomas in a FDR, with a pooled OR of 1.99 (95% CI, 1.55-2.55). Despite variations in the study designs and populations examined, these studies support the findings of the current study that FDRs of patients with adenomas have a moderately increased risk of developing colorectal neoplasia whereas relatives of those with AAs (villous histology or large size) are at a higher risk of developing colorectal neoplasia. The results of the current study add considerable support to these other studies by examining risk in a population data set, but also extend the findings robustly to older relatives and more distant relatives.
The most recent of these systematic reviews of familial CRC studies concluded that studies published to date have had several design problems that affect their validity and generalizability. Of 12 identified studies, 10 did not answer the primary question of whether having a FDR with an adenoma increased a patient's risk of developing CRC. The remaining 2 studies suffered from a lack of generalizability due to racial homogeneity and a lack of validation of family history and colonoscopy records. The authors advocated for a cohort or cross-sectional design study that could identify all relatives and have access to medical records to establish a diagnosis of CRC, thereby reducing the effect of patient recall bias.
The design and data sets of the current study mitigate some of the weakness identified in this systematic review. To the best of our knowledge, the current study is the largest to date of familial colorectal neoplasia, thus giving it power to examine risk in SDRs and TDRs as well as FDRs. Recall bias, as mentioned above, is avoided by having primary clinical records as well as validated genealogical records for familial relationships. Detection bias, in which relatives of cases may be more likely to undergo colonoscopy and have adenomas found than controls, is also a common issue. Although detection bias may be present in the current study, we were able to demonstrate that a similar number of relatives of cases and controls had undergone colonoscopy examinations.
The current study is population-based, reflecting standard-of-care clinical practice in both a large academic medical center and a managed care organization that together provide care to approximately 85% of the population of the state of Utah. This size and inclusiveness also reduces referral bias. The population of Utah is also representative of US/European white populations, with a low level of inbreeding. Therefore, the results are applicable to similar populations in the United States.
To the best of our knowledge, only a few studies to date have examined the risk of CRC in relatives of patients with adenoma stratified by the age of the index case. The results of the current study are similar to the findings of studies from France and New York State,[25, 26] in which the risk of CRC among family members was found to increase with decreasing age at the time of diagnosis of adenoma in the index case. Familial risk for older-age probands is an important point, because increased familial risk when the index case (proband) is aged > 60 years is often not appreciated.
In the current study, we were also able to examine patient age when CRC occurred in FDRs, both when index cases were age < 60 years and ≥ 60 years at the time of diagnosis of adenoma or advanced adenoma (Table 6). With this analysis, we found approximately 10% and 1.5%, respectively, of FDRs of cases of adenoma and villous adenoma developed CRC at ages younger than those currently recommended to start colonoscopy screening (based on familial risk), and therefore would have been missed.
There are certain limitations of the current study that also deserve mention. Patients with known hereditary cancer syndromes, in particular familial adenomatous polyposis and Lynch syndrome, could not specifically be excluded. However, because both of these conditions are rare, and account for less than approximately 3% of all colon cancers, they are unlikely to modify the statistical associations demonstrated herein. Another issue is that the extraction of data from electronic medical records has limitations with regard to the information that can be gathered. The indication for colonoscopy was not specifically ascertained and therefore might introduce bias. Many screening procedures are also coded for a diagnosis found at colonoscopy or other unrelated symptoms and thus it would be nearly impossible to establish a study population of only screening colonoscopies. However, the ascertainment of the current study reflects gastroenterologists' usual clinical practice of colonoscopy, with patients referred for screening as well as other indications. We also did not have information available regarding the documented completeness of the colonoscopic examination or the quality of the bowel preparation, although these issues are likely to be similarly reflected in both the case and control groups.
We conducted a retrospective, population-based, case-control study in Utah, demonstrating that FDRs, SDRs, and TDRs of patients with adenomas and AAs who are undergoing colonoscopy are at an elevated risk of developing colorectal neoplasia (CRC, adenomas, and AAs). An increased risk was observed in relatives regardless of the age of the index patient at the time of adenoma diagnosis, although the risk was greater for index cases at a younger age. The results of the current study confirm and expand the evidence for a common familial risk for colorectal neoplasia and should become part of the data considered when screening strategies are synthesized. The data from the current study support the current screening guidelines for individuals with a family history of advanced adenomas and raise the issue of whether more aggressive screening should be considered for the FDRs of patients with adenomas diagnosed at age ≤ 60 years. Future studies to examine the risk of CRC in the relatives of patients diagnosed with AAs based on other components of the definition (large size or high-grade dysplasia) are required to validate the current CRC screening guidelines. The results of the current study further support inherited factors as a part of the etiology of colorectal neoplasia. Clinician awareness of the familial risks demonstrated in the current study will be of considerable assistance when working with individuals and families with colorectal neoplasia.
Supported by National Cancer Institute grants P01-CA073992and R01-CA040641 (both to Dr. Burt). Partial support for the Utah Population Database is provided by the Huntsman Cancer Institute (HCI) and the HCI Cancer Center support grant P30CA042014, from the National Cancer Institute. The Utah Cancer Registry is funded by contract HHSN 261201000026C from the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) program, with additional support from the Utah State Department of Health and the University of Utah.
CONFLICT OF INTEREST DISCLOSURES
Dr. Burt is a consultant for and has received payment for lectures including as a member of the Speakers Bureau from Myriad Genetics. Dr. Samadder has received speaking fees from Cook Medical Inc.