Survival After Colorectal Cancer Diagnosis Is Associated with Colorectal Cancer Family History Free

Colorectal cancer (CRC) is the third most common cancer in the United States, accounting for 10% of all cancer diagnoses, and is the second leading cause of cancer-related death, accounting for 8% of all cancer deaths among men and 9% of all cancer deaths among women (1). A positive family history of CRC (defined as the presence of CRC in one or more first- and/or second-degree relatives) is a well-recognized risk factor for CRC. Individuals are at varying degrees of increased risk depending on age of the affected family member, number of affected relatives, and closeness of the relative. Similarly, individuals with a family history of CRC are at a higher risk to develop colorectal polyps. This is most likely due to the contribution of several genes, in addition to environmental factors. It is estimated that 10% to 30% of CRC cases show familial clustering (2, 3).

The clearest examples of familial risk for CRC are the autosomal dominantly inherited genetic cancer predisposition syndromes familial adenomatous polyposis and hereditary nonpolyposis CRC. Familial adenomatous polyposis, characterized by the presence of >100 colorectal polyps, is associated with a >90% lifetime risk to develop CRC (4), whereas individuals with hereditary nonpolyposis CRC have about an 82% cumulative risk to develop CRC (5). However, well-characterized, highly penetrant genetic syndromes account for ∼1% of CRC cases (6).

Despite convincing evidence that family history of CRC is a risk factor for colorectal polyps and cancer (7-10), conflicting reports have emerged on the association of family history with survival after CRC diagnosis. Family history of CRC has been reported to have no association with survival after CRC diagnosis in the Melbourne CRC cohort study (11) and in a study of CRC cases in the Utah Population Database (12). In a recent analysis of women enrolled in the Nurses Health Study, increased overall and cancer-specific mortality were noted for CRC cases having a family history of CRC (13). However, in a Japanese population–based study, CRC family history was associated with improved 5-year survival rates among colon cancer cases (14). Recently, an analysis of stage III colon cancer patients enrolled in an adjuvant chemotherapy clinical trial revealed that family history of CRC was associated with improved disease-free survival and recurrence-free survival but not overall survival (OS; ref. 15).

In the context of these varying reports, the role of CRC family history on survival among colon and rectal cancer cases is still unclear. Therefore, we designed the present study to determine if family history of CRC predicts improved survival among colon and rectal cancer cases using data from the University of California Irvine Gene-Environment Study of Familial Colorectal Cancer.

Incident cases of invasive CRC from the University of California Irvine Gene-Environment Study of Familial Colorectal Cancer were analyzed (6). Participants were identified through the population-based cancer registries of the Cancer Surveillance Program of Orange County/San Diego Imperial Organization for Cancer Control, using the February 2007 data file. In the parent study (6), all subjects with CRC diagnosed at all ages in Orange County, CA, from 1994 to 1996 were ascertained. All subjects diagnosed in San Diego and Imperial Counties, CA, at ages <65 y between 1994 and 1995 were also ascertained. Cases were then contacted if they were eligible for the study (alive at the time ascertained and having a contact address) and if their physicians did not deny permission to contact. At the time of study entry, subjects signed a consent form allowing for blood draws and the release of medical information, including access to pathology reports and tissue blocks. This study was approved by the University of California Irvine Institutional Review Board (#93-257).

Clinical and demographic data were obtained from the aforementioned cancer registry databases, as previously reported (16, 17). Recorded data include demographic information (e.g., age, gender, and ethnicity), site within the colon/rectum, stage at presentation, tumor histology, and grade. Stage at diagnosis was derived in adherence to tumor-node-metastasis criteria using the available Surveillance, Epidemiology and End Results Extent of Disease descriptor in CCR, as previously described (18). Tumor histology and site were coded according to criteria specified by the WHO in International Classification of Diseases for Oncology (19). Histology codes included the following: adenocarcinoma (8140, 8144, 8210, 8260-8263, 8380, and 8490), mucinous adenocarcinoma (8470, 8480, and 8481), carcinoma (8010, 8020, 8070, 8071, 8124, and 8240), and not otherwise specified (8000, 8041, 8042, 8120, 8130, 8243, 8246, 8560, 8570, and 8722). Only invasive cases of cancer were included in the analysis. Primary site code was searched using the Surveillance, Epidemiology and End Results site code for colon (21041, 21043-21048) and rectum (21051-21052). Appendiceal cancers were excluded. Therapeutic information related to the first course of treatment was abstracted from medical and laboratory records by trained tumor registrars according to Cancer Reporting in California: Vol. 1, Abstracting and Coding Procedures for Hospitals (20). This included surgical treatment rendered at the primary site (i.e., any surgical resection, including total/subtotal/partial colectomy, unknown type of colectomy, and local excision), treatment with radiation therapy, and the use of chemotherapy.

Standard therapy is defined according to the National Comprehensive Cancer Network practice guidelines as follows: resection alone for stage I disease in both colon and rectal cancer; surgical resection alone or with adjuvant chemotherapy for stage II colon cancers; and surgical resection followed by adjuvant chemotherapy for stage III colon cancers. Stage II and III rectal cancers typically require treatment with neoadjuvant (i.e., up-front) chemoradiation followed by surgery and additional chemotherapy, whereas stage IV colon and rectal cancers are treated with systemic chemotherapy (21).

Family history of cancer was ascertained by self-reporting during a telephone interview at time of study enrollment. Cases with family history of CRC in a first-degree relative (hereafter called “familial CRC cases”) were recognized as those having at least one first-degree relative (parent, sibling, or offspring) with CRC. Family history in second-degree relatives was not assessed in the present analysis. Accuracy of self-reporting was reported to be 90% in a prior validation study involving these CRC cases (22). Twenty-two cases with hereditary nonpolyposis CRC, as defined by Amsterdam criteria, in addition to one case with clinically diagnosed familial adenomatous polyposis, were excluded from analysis. The remaining sporadic and familial CRC cases (n = 1,174) were included for analysis.

Hospital registrars contacted cases regularly, and registry staff reviewed state death certificates to identify deceased registry cases. Cause of death was recorded according to the International Classification of Diseases criteria in effect at the time of death (23). Major causes of death were grouped into five categories: CRC, heart disease, lung cancer, other cause, and missing. Follow-up data through December 2006 were available at the time of analysis. The last date of follow-up was either the date of death or the last date the case was contacted.

Overall survival was defined as mortality from any cause. Thus, censoring events were created for those alive at the end of the follow-up period and cases lost to follow-up. CRC-specific survival (CRC-SS) was defined as mortality from CRC itself, such that censoring events were created for those alive at the end of the follow-up period, cases lost to follow-up, and cases that died from any cause other than CRC.

Survival curves for CRC, colon cancer, and rectal cancer cases were constructed using the Kaplan-Meier method and analyzed with the log-rank test for univariate CRC-specific and OS. Univariate stratified survival analyses were also done for colorectal, colon, and rectal cancer cases across strata for tumor-node-metastasis stage (i.e., I, II, III, and IV). Cox proportional hazards modeling was done using time since diagnosis to profile the adjusted risk of overall and CRC-specific mortality based on family history of CRC. Variables known to predict survival in CRC were included in the multivariate survival model, including age, gender, ethnicity, stage at presentation, site within the colon/rectum, tumor histology, surgical treatment rendered at the primary site, radiation therapy, and chemotherapy, with adjustment for family history.

Analyses were done for colon and rectal cancer cases, stratified by the presence or absence of family history. Comparisons of demographic, clinical, and pathologic variables among colon and rectal cancer cases were done using Pearson χ² statistic or Fisher's exact test for nominal variables and Student's t test for continuous variables. All descriptive analyses were conducted using SAS 9.1 statistical software (SAS Institute). Statistical significance was assumed for a two-tailed P value of <0.05.

We proposed a 10% or greater difference in CRC-SS based on family history of CRC for our sample size calculations. Thus, at least 684 total subjects were needed to detect a statistically significant 10% difference in 10-y CRC-SS between the two groups: 60% in sporadic CRC cases versus 70% in familial CRC cases.

A total of 1,174 CRC cases were identified from the University of California Irvine Gene-Environment Study of Familial Colorectal Cancer during the specified time period, with a median follow-up duration of 9.3 years. Seven hundred eighty one (66%) of these were colon cancer cases, 373 (32%) were rectal cancer cases, and 20 (2%) were unspecified and thus excluded from further analysis. Clinicopathologic data for the colon and rectal cancer groups are presented in Table 1.

From the 1,154 analyzed cases, 946 sporadic cases and 208 familial cases were identified (183 cases with CRC history in one first-degree family member, 19 with 2 affected family members, and 6 with >2 affected family members). Among colon cancer cases, 148 (19%) were considered familial and 633 (81%) were sporadic. Among all rectal cancer cases, 60 (16%) were considered familial and 313 cases (84%) were sporadic. Clinical comparisons between familial and sporadic cases (separately for colon, and rectum) revealed no significant differences in distribution by gender, ethnicity, site within the colon, histologic subtype, or tumor grade (Table 1). Although the median age was similar for colon cancer cases regardless of family history, familial rectal cancer cases were younger (median, 61.0 years; range, 24-99 years) than sporadic cases (median, 65 years; range, 37-93 years; P = 0.0085). Familial colon cases presented at an earlier stage than sporadic cases, with more familial cases diagnosed with stage I colon cancer, although this finding did not reach significance (P = 0.070). Individuals with familial colon cancer were also more likely to receive surgical treatment (P = 0.039), whereas radiation therapy and chemotherapy use were similar for familial and sporadic colon cancer cases (Table 1).

Because of the observed difference in the proportion of familial compared with sporadic colon cancer cases receiving surgery and near significant differences in stage at diagnosis (Table 1), the first course of treatment was further examined for familial and sporadic colon cancer cases, stratified by stage at presentation (Table 2). Among colon cancer cases, surgery, radiation therapy, and chemotherapy use were similar between familial and sporadic colon cancer cases within each stage.

Of the 1,154 colon or rectal cancer cases combined, 603 (52%) were deceased at the time of analysis. Ninety-six of these were familial cases and 507 were sporadic. Among the 603 deceased CRC cases, cause of death was available for 557 cases (92%). Cause of death was available for 91 of 96 (94.8%) familial and 466 of 507 (91.9%) sporadic cases (P = 0.33). Three hundred sixty-five (81%) cases died from CRC. Univariate survival analysis revealed no difference in 10-year CRC-SS between familial and sporadic CRC cases (69% versus 66%; P = 0.51) or 10-year OS (55% versus 49%; P = 0.075).

However, in the univariate analyses of colon cancer cases, statistically significant OS differences by family history were observed among those with nonmetastatic stage at presentation (i.e., tumor-node-metastasis stage I, II, or III): 66% 10-year OS for familial colon cancer cases compared with 56% 10-year OS for sporadic colon cancer cases (P = 0.021; Fig. 1).

For the 781 colon cancer cases, 415 (53%) were deceased at the time of analysis. Sixty-seven of these were familial cases and 348 were sporadic. Among the 415 deceased colon cancer cases, cause of death was available for 383 cases (92%). Cause of death was available for 64 of 67 (95.5%) familial and 319 of 348 (91.7%) sporadic cases (P = 0.28). Two hundred forty-one (79%) of these cases died from CRC, 55 (18%) died from cardiac disease, and there were no differences between familial and sporadic colon cancer cases in the distribution of cause-specific deaths (CRC, 81% versus 78%; heart disease,19% versus 18%; P = 0.30).

Of the 373 rectal cancer cases, 188 (50%) were deceased at the time of analysis. Twenty-nine of these were familial cases and 159 were sporadic. Among the 188 deceased rectal cancer cases, cause of death was available for 174 cases (92.5%). Cause of death was available for 27 of 29 (93.1%) familial and 147 of 159 (92.5%) sporadic cases (P = 0.90). One hundred twenty-four cases (86%) died from CRC, 19 (13%) died from cardiac disease, and there were no differences between familial and sporadic colon cancer cases in the distribution of cause-specific deaths (CRC, 83% versus 86%; heart disease, 17% versus 12%; P = 0.68).

Statistically significant differences in CRC-SS were not observed between familial compared with the reference sporadic colon cancer cases in unadjusted analysis, or in the adjusted analysis accounting for age, gender, ethnicity, stage, tumor site, histology, and treatment with surgery, radiation, and chemotherapy (Table 3). However, family history was an independent predictor of OS for colon cancer [hazard ratio (HR), 0.760; 95% confidence interval (95% CI), 0.580-0.997; P = 0.047] in the adjusted analysis, and a borderline significant difference was detected in the unadjusted analysis (HR, 0.773; 95% CI, 0.594-1.005; P = 0.054). Multivariate OS analysis for confounding of colon cancer cases by CRC family history revealed that adjusting for stage at diagnosis is the key variable that makes family history become significant. The P value for family history in the Cox model goes from a P value of 0.054 in the unadjusted model to P value of 0.053 after adjustment for age, then a P value of 0.034 after additional adjustment for stage at presentation.

In the analysis of rectal cancer cases, no differences in CRC-SS were observed between familial compared with sporadic colon cancer cases in unadjusted analysis, or in the adjusted analysis (HR, 0.920; 95% CI, 0.610-1.386; P = 0.69; Table 3). Family history was not an independent predictor of OS for rectal cancer in the adjusted or unadjusted analysis.

Subset analyses of colon cancer were done to assess the association of family history (versus no family history of CRC as a reference group) on OS stratified by gender, and separately by stage at diagnosis. In the gender-stratified analyses of colon cancer cases, CRC family history was associated with improved OS for males (HR, 0.624; 95% CI, 0.423-0.920) but not females (HR, 0.937; 95% CI, 0.631-1.392). In the multivariate analysis of OS by stage, hazards estimates for family history of CRC were as follows: stage I (HR, 0.56; 95% CI, 0.28-1.13), stage II (HR, 0.75; 95% CI 0.43-1.30), stage III (HR, 0.82; 95% CI, 0.47-1.41), stage IV (HR, 0.77, 95% CI, 0.46-1.29).

In our analysis, family history of CRC was an independent predictor of OS among colon cancer cases but was not associated with CRC-SS (Table 3). No disparities in treatment rendered by stage were apparent to explain the decreased overall mortality among familial colon cancer cases in this study.

Our results are most consistent with the report of Kirchhoff et al. (24) in their analysis cohort of 1,469 women. That study revealed a significant OS benefit among familial CRC cases when compared with sporadic and a nonsignificant trend toward improved CRC-SS among familial compared with sporadic patients (24). Our population consisted of 68% colon and 32% rectal cancer cases, compared with the Kirchhoff study, which was composed of 82% colon and 18% rectal cancer patients (24). Our conclusions are also aligned with results published by the Japanese Research Society for Cancer of the Colon and Rectum; this group detected improved survival among familial colon cancer cases when compared with sporadic cases. However, it is unclear whether overall or CRC-specific mortality was analyzed in this work (14). The recent study by Chan et al. (15) revealed significant improvements in disease-free survival and recurrence-free survival, and a nonsignificant OS benefit (HR, 0.75; 95% CI, 0.57-1.04) for stage III colon cancer cases with family history of CRC. Although the sample of colon cancer cases available was larger than ours (1,087 versus 781), follow-up time was substantially less (median follow-up of 5.6 versus 9.3 years). It is possible, then, that the improved OS difference for familial patients will become significant with additional events over the course of follow-up. Thus, our results are generally supportive of the prior research available from these three major studies.

Our results are inconsistent with those reported in the analysis of CRC and CRC family history in the Utah Population Database (12). Among colon cancer patients in that study, a positive family history of the disease did not affect colon cancer specific or OS. Moreover, when their cohort was stratified by age at diagnosis, gender, and type of relative affected, family history seemed to have a negative effect on survival, particularly among younger males who had siblings with colon cancer (12). Our analyses were adjusted for age at diagnosis and gender but were not further stratified by specific type of family member affected with CRC.

Bass et al. (13) reported that women with a family history of CRC were at increased risk both of CRC-specific mortality and death from any cause, and women with more than one affected relative were even more likely to die from CRC or from any cause. Finally, among stage IV CRC patients, significant decreases in CRC-specific and OS were observed for familial compared with sporadic cancer cases (13). Their cohort was composed of 73% colon cancer, 26% rectal cancer, and 2% unspecified CRC patients (i.e., similar to our study; ref. 13). Our conclusions may differ from those of Bass et al. (13) because our study population included both males and females. Interestingly, in our gender-stratified analyses of colon cancer cases, the association of CRC family history with improved OS was significant for males (HR, 0.624; 95% CI 0.423-0.920) but not females (HR, 0.937; 95% CI, 0.631-1.392). Consistent with these data are the stratified analyses from the Chan et al. study (15), demonstrating that family history was associated with decreased risk of disease recurrence among men but not women. Thus, it is possible that the OS effects of CRC family history are more pronounced among males than females with colon cancer; however, this gender-specific difference was not detected in CRC-SS analyses in our study (data not shown), and improved OS was detected in the aforementioned Kirchhoff study (all females; ref. 24). Perhaps the gender-stratified OS results in our study merely reflect gender-specific differences in competing risks (discussed below) for familial versus sporadic colon cancer cases.

Kune et al. (11) observed no survival difference between familial and sporadic CRC cases. However, survival rates were reported after only a 5-year follow-up period, whereas our study investigated mortality after >9 years and in a larger cohort. Additionally, the population analyzed by Kune et al. (11) consisted of a lower ratio of colon to rectal cases, with rectal cancer patients accounting for 45% of the total cohort. Thus, the large proportion of rectal cases may have negated any effect of family history on survival among colon cancer cases in their analysis.

Variability in adherence to surveillance monitoring, genetic differences, or competing risks could explain our finding that CRC family history is independently associated with improved OS among colon cancer cases. Individuals with a family history of the disease may be more likely to undergo regular CRC surveillance or pursue more aggressive treatments once diagnosed with CRC, thereby affecting survival. Differences in microsatellite instability and expression of mismatch repair proteins between familial and sporadic colon cancer cases could underlie our findings; however, two factors mitigate against this possibility. In the aforementioned study of survival among stage III colon cancer patients, the improvement in survival for familial patients was independent of microsatellite instability or mismatch repair status (15). Additionally, we restricted our analyses to nonhereditary nonpolyposis CRC cases, as defined by Amsterdam criteria, thus presumably limiting our analysis of familial cases to those without this genetically inherited syndrome, which has been shown to be highly associated with microsatellite instability (6). Thus, additional genetic factors implicated in heterogeneity of response to treatment, such as loss of chromosome 18 in the primary tumor specimen of patients receiving adjuvant chemotherapy (25, 26), must be investigated to explain the observed survival differences among individuals with familial and sporadic colon cancer (27). Differences in competing risks among familial versus sporadic CRC cases would explain differences in OS but not CRC-SS, as observed here. If familial CRC cases are less likely to suffer death from competing illnesses (i.e., cardiac disease, pulmonary disease, diabetes mellitus, etc.), they might exhibit OS differences without having differences in CRC-SS. Although we cannot entirely account for this in our study design, at least for cardiac-related deaths, observed rates were similar (11.2-17.5%) between familial and sporadic cases of the colon and rectum.

Family history seems to have no influence on CRC-specific or overall mortality among rectal cancer cases, as shown in Table 3. Although interpretation is limited by small sample size, results are substantiated by a similar finding in the Japanese Research Society for Cancer of the Colon and Rectum analysis (14). Biological differences in subsite location within the colorectum have been reported previously. Variation in CRC incidence by tumor site and its association to geography, age, and gender led to the concept of distinct CRC diseases based on proximal or distal location in the large bowel (28). Differences by subsite location within the colorectum have been reported for tumor suppressor genes, point mutations, and genetic instability (29-32). Specifically, it has been noted that left-sided colon lesions have a higher rate of chromosomal instability, whereas right-sided colon lesions more often show microsatellite instability (33-39). Azzoni et al. (40) studied distinct molecular features of sporadic CRC patients and felt that the different patterns observed according to subsite could reflect different genetic pathways of carcinogenesis. Certain environmental factors such as fatty diets and increased bile acid exposure, whether secondary to cholecystectomy or increased fat intake, together with genetic risk factors, may increase the risk of proximal CRC (28, 41-44). However, the consequences of these various derangements on survival after CRC diagnosis are unknown, and these areas remain fertile ground for further exploration.

Our study is retrospective in nature and shares limitations of other population-based studies. Performance status and comorbidity data are not available in this analysis. Differences in the observed age at diagnosis for sporadic and familial rectal cancer may be due to sampling bias in the parent study, as previously described (6). Information on disease progression and site of CRC recurrence to calculate recurrence-free survival are not available through CCR. However, CRC-SS is a reasonable approximation of relapse-free survival because the majority of CRC is incurable upon progression. Because all stage CRC cases were included, a lack of CRC-specific events (particularly among cases with early stage tumors) limits the interpretation of these results. The greater proportion of familial colon cancer cases receiving surgery may have influenced our survival analysis, despite adjustment for stage and surgery in the multivariate survival analysis. However, our analysis of treatment by stage reveals that within each stage, no differences in the proportion receiving surgery were observed between familial and sporadic colon cancer cases. Similar proportions of familial and sporadic colon cancer cases received standard therapy in our stage-specific analysis. Surgical resection rates are high in this study, due in part to the fact that we analyzed surgery as any type of surgical procedure directed at the primary tumor. Factors not so easily accounted for may also explain our findings, including differences in institutional treatment standards, physician preferences, and patient beliefs.

Our results support prior reports that CRC family history has a role in colon, but not rectal, cancer mortality. Clinical, genetic, biological, and behavioral differences to explain this epidemiologic finding are postulated, but conclusive mechanisms underlying our observations are unknown. In the current age of therapeutic advances in CRC management, and with the ever-increasing number of cancer survivors (45), understanding such mechanisms is critical, and may lead to targeted approaches to prevention of recurrent or relapsed disease among CRC survivors.

No potential conflicts of interest were disclosed.

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.