Colorectal cancer (CRC) is the third most common malignancy in the US and the second leading cause of cancer death [1]. Compared to European Americans (EA), African Americans (AA) have a 20% higher CRC incidence rate and a 45% higher mortality rate [1]. At least some of the excess mortality in AA is caused by the reduced survival associated with a higher prevalence of metastatic disease at diagnosis [2-4]. However, the disparity persists at each stage and is likely to be influenced by differences in treatment utilization and/or CRC tumor biology.

CRC survival has improved significantly over the past fifteen years due to the introduction of several new chemotherapeutic agents; yet, the racial difference in survival has worsened, especially among advanced stage patients [5, 6]. Until the year 2000, there were few treatment options other than 5-Fluorouracil (5-FU) and leuvocourin for metastatic CRC; however, with the introduction of combination chemotherapy in 2000 (i.e., irinotecan or oxaliplatin with 5-FU) and the addition of biologic agents to the combination chemotherapy in 2004 (i.e., bevacizumab or cetuximab), the median survival has increased from 8-12 months in the mid-1990’s to 18-24 months in 2008 [7]. Although few studies have examined the effect of the new therapies on CRC survival at the population level, historical evidence suggests that AA may not experience the same survival benefit as EA [5, 8, 9]. To explain, AA have been reported to have lower rates of initiation of standard therapy [10-12] and refuse therapy at a higher rate than EA [13, 14]. More recently, AA have been found to have a lower response to the newer treatment regimens, which may translate to differences in survival by race [8, 15].

Differences in CRC tumor biology by race, age, and or gender may influence the difference in stage-specific survival. AA compared to EA are more likely to present at diagnosis at a younger age and with features indicative of greater tumor aggressiveness, including a higher proportion of proximal compared to distal neoplasia (11-17). Proximal colonic location is associated with worse survival [16-19], especially among advanced stage CRC patients. Earlier age of onset, irrespective of race, is associated with aggressive tumor histopathology, a higher grade and stage of CRC, and a higher proportion of hereditary CRC [20-27]. Furthermore, gender may also be associated with difference in tumor aggressiveness since women appear to have a tendency to develop tumors with lower rates of metastasis and better prognosis [28, 29]. However, few studies have considered the intersection of race, age and gender on survival.

In the present investigation, statewide, population-based data from South Carolina Central Cancer Registry (SCCCR) is used to examine if there are racial differences in survival in advanced stage CRC. This study is restricted to a single stage because the magnitude of the racial disparity in survival has been shown to differ by stage. This study focuses on metastatic disease because of similar recommended standard therapy, large racial difference in the relative survival, and a documented change in chemotherapy usage in both 2000 and 2004. We investigated the relationship between race and age on CRC survival during three calendar periods: (1) 1996-1999; (2) 2000-2003; and (3) 2004-2006, each corresponding to a change in chemotherapeutic regimen. We also explored the influence of clinicopathologic features by race in younger and older patients, and in males and females, and their impact on survival.

In this analysis of South Carolina Central Cancer Registry (SCCCR) data, key findings were that AA had worse overall survival for advanced stage colorectal cancer. This racial disparity was largely concentrated among CRC patients under 50 years of age. A disconcerting observation was that the racial disparity has worsened during the past 15 years. In the survival analysis, we observed that among the younger patients, AA had worse survival especially AA females with proximal tumors. For older patients, AA males had worse survival, but the association with race was not nearly as pronounced as it was in the younger age groups.

Despite improvements in survival among those with advanced stage CRC, our study highlights the high fatality rate of this disease, and thus emphasizing the importance of prevention and early detection of CRC. In the present study, there were a higher percentage of stage IV CRC cases under the age of 50 years in AA than EA (17% vs. 12%) and the prognosis of AA patients was worse suggesting a more aggressive phenotype. This touches on the broader issue of developing strategies for identifying high-risk patients who have not yet reached the age for recommended screening. These findings are thus in line with the recommendations of two national organizations that recommend AA begin CRC screening earlier than EA, 40 [30] or 45 [31] years of age.

Our findings of poorer CRC survival in AA compared to EA parallel those of several previous investigations. Our finding that the disparity among late stage CRC patients is largely concentrated among the younger patients is novel. In general, the racial disparity has been shown to occur at each stage for both men and women and has worsened during the past 40 years (23). Most (10, 24-34), but not all (35-37), studies have observed that the disparity persists even after adjustment for variables such as age, sex, stage, socioeconomic status, insurance status, and in some studies pathologic characteristics. After accounting for these factors, the survival differences by race range from 8-50% across the various patients populations that have been studied. The reasons for this persistent racial disparity in CRC survival remain poorly understood. However, results from the present study suggest that the effect of ‘race’ on advanced stage disease survival is not uniform across the population. Factors such as patient’s age at diagnosis, gender, type of CRC, and date of diagnosis (possibly reflecting differences in treatment utilization) all influenced the relationship between race and survival.

On average, AA are known to be diagnosed with CRC at an earlier age (11, 12, 49), but to our knowledge there are no previous reports of racial differences in survival by age. A large body of evidence documents that CRCs in younger persons are more likely to have familial hereditary CRC and, independently of family history, are likely to have more advanced and aggressive tumors (41, 42 2008, 43, 44). Younger patients also appear to have more genome complexity (assessed as copy number aberrations), and higher P53 expression or mutation (41, 42, 45). Within our data, younger AA and EA had different tumor profiles. AA had a higher prevalence of proximally located (+13%), mucinous (+4%) and lower grade (+6%) CRCs at diagnosis. Whether this represents a unique phenotype associated with poorer survival will need further study. On the other hand, despite the presence of more aggressive disease, many studies have shown that younger persons (predominantly EA) have better survival from CRC [20, 23, 24, 27]. Another possibility is that poor clinical performance status coupled with an aggressive phenotype added to the worse outcomes among AA, but we could not assess this in our dataset. Finally, the role of hereditary colon cancer in younger AA and its impact on survival is largely unexplored. Two studies have reported a higher prevalence of MSI-H tumors in AA – a common phenotype in familial CRC-- compared to historical controls and, in one, the tumors exhibited a different mutational pattern [32, 33].

Outcomes in our investigation differed by race and by sex. Similar to two recent studies [28, 29], we observed that younger EA women had better survival than younger EA men. In the present study, however, young AA women had worse survival than younger EA women did. In part, the racial difference by gender in the young may be influenced by the colonic location of the CRC. For example, among younger women, proximal colonic location was associated with poorer survival and younger AA women had a higher prevalence of proximal tumors compared to EA (44% vs. 33%). For older women, there was no difference in prevalence of proximally located tumors by race (AA 48% vs. EA 50%) and no racial difference in survival. Several studies have shown that estrogen protects against CRC [34-36], and it is possible that a subset of younger EA women have a different hormonal profile compared to younger AA woman. A similar pattern of poorer survival and aggressive phenotypes has been observed in younger AA breast cancers patients [37, 38] compared to EA. Among advanced stage CRC patients older than 50 years, a modest increase in risk among AA men compared to EA was seen with no difference among women.

Many studies, like ours, have found that clinicopathologic characteristics influence survival. Similar to previous investigations [39-41], we observed the prevalence of proximally based tumors is higher in AA. As in the present study, several previous studies have previously observed that proximal colonic location is associated with worse survival [16-19, 42]. Compared to distal tumors, proximal colonic tumors were observed to be associated with greater mortality among stage IV CRC patients (HR 1.22; 95% CI 1.16-1.28)) [16]. However, proximal cancers are not uniform. A subset of proximal tumors is associated with favorable prognosis (i.e. MSI-H cancers). CRC is increasingly recognized as a heterogeneous disease, which can evolve through at least three distinct carcinogenic pathways [43-49]: (1) the chromosomal instability pathways (CIN), (2) The CpG methylator pathway-1 (CIMP-1) and (3) the CpG methylator pathway-2 (CIMP-2). The CIMP pathways predominate in the proximal colon and demonstrate CpG methylation. The CIMP-1 cancers tend to exhibit MSI-H (Microsatellite instability-high) status, BRAF mutations, and low levels of KRAS and p53 mutations and have good prognosis; CIMP-2 CRCs have a higher proportion of Microsatellite stability (MSS), more frequent KRAS mutations and poor prognosis. The CIN pathway tumors show distal predominance, APC, P53 mutations, MSS, and lack extensive CpG methylation and have average prognosis. Furthermore, a recent study has found that AA have a higher prevalence of KRAS mutations, and those with these mutations have poorer survival [50]. While personalized medicine will certainly focus effective treatment on a colon tumor’s molecular vulnerabilities, it may be some time before this knowledge can fully explain the disparities we observed in the present study. Future studies that integrate detailed molecular genetic/epigenetic markers along with detailed treatment information may be of particular value.

The persistence of racial differences in survival within the same stage suggests that treatment differences may be at least part of the explanation for differences in survival [3, 51]. Historically, many [3, 11, 51, 52], but not all, studies [14] have found that AA patients are less likely to receive standard recommended therapy, and refuse therapy at a higher rate. We observed no statistically significant differences in the initiation of chemotherpy by race, but did see striking improvements in survival among EA patients after the introduction of combination chemotherpy in 2000, and again with the introdution of bevacizumab to the standard of care in 2004. Treatment utilization is strongly influenced by access to high-quality medical care. Medically underserved populations (including AA) are less likely to have equal access to the state-of-the art care, especially younger AA without health insurance [51]. As such, it may be that there is a racial disparity in the dissemination of the newer chemotherpies although a recent report did not observe this [15]. In health care settings where patients have equal access to care, the racial disparity in treatment utilization and outcomes is attenuated. In the Veterans Administration, for example, no significant differences by race were reported in rates of surgical resection, radiation therapy, or chemotherapy use. Although AA as compared to EA had modestly poorer survival [53], the disparity is less pronounced than commonly observed in population-based studies. However, treatment efficacy may also differ by race. Sanoff and colleagues [8] analyzed outcomes to combination chemotherapy within the context of a clinical trial and found that survival by race differed by type of treatment regimen. AA patients on FOLFOX or FOLFIRI had lower survival compared to EA, yet AA on IROX (i.e., no 5-FU) appeared to have better survival than EA. On the other hand, recently, Polite et al. [15] studied metastatic CRC patients who received bevacizumab within a community based setting and observed no difference in survival by race. In both of these studies, AA had significantly lower response rates. Because treatment regimen was not detailed in our data, it is difficult to assess the role of treatment use versus efficacy on survival by race. Furthermore, because we do not have information on the treatment medication administered to patients we cannot be certain that AA and EA actually received the same multi-drug regimens which, if different by race, could potentially contribute to some of the differences in survival.

Advantages of our study include a large, racially diverse population of patients with advanced stage colorectal cancer with careful characterization of demographic and pathologic characteristics and vital status. However, our study had a few limitations. First, we had no data on patient-level factors (such co-morbid conditions or lifestyle and behaviors) or detailed treatment regimen data, which confound or modify the association between race and CRC survival. Second, we did not have information on CRC specific deaths only on all cause mortality; however, because all of the patients had advanced stage CRC we expect that most of these people died a result of their disease. Finally, the setting in South Carolina may differ from other locales in ways that limit the generalizability of the study findings. For example, South Carolina has a high proportion of rural residents, which may introduce substantive differences in results compared to populations that are largely urban. In the future, it will be essential to see if our findings can be replicated in different populations, and future studies that integrate patient-level, clinical, molecular, and treatment-related data hold promise for advancing understanding of the racial disparity in CRC survival.