Subsite-Specific Incidence Rate And Stage Of Disease in Colorectal Cancer by Race, Gender, and Age Group in the United States
Abstract & Commentary
Synopsis: To better understand the influence of screening strategies, this study examined stage of colorectal cancer at different anatomic subsites. The relation between stage and site-specific data from 344,775 patients diagnosed between 1992-1997 was analyzed by race, gender, and age group. Overall, the percentage of localized disease increased from 31.9% in the proximal colon to 41.5% in the distal colorectum, implying that screening strategies have had a strong preventive influence on distal disease. Black patients were more likely than white to receive a diagnosis of advanced disease for every subsite and more likely to receive a diagnosis of proximal colon than distal colorectal cancer. Male-to-female rate ratios monotonically increased from the proximal colon to the distal colorectum. With increasing age, ratios of proximal-to-distal disease increased.
Source: Wu XC, et al. Cancer. 2001;92:2547-2554.
Variation in subsite-specific incidence rates of colorectal cancer has been demonstrated between genders, among different age groups, and among racial groups.1-3 In this population-based study, Wu and colleagues analyzed variations in stage of colorectal cancer at different anatomic subsites, stratified by race, gender, and age group, in order to explore the inherent screening implications.
Data were derived from 28 central cancer registries representing 40% of the US population. Population estimates from 1992-1997 were based on 1999 US Bureau of Census estimates, as reported by the Surveillance, Epidemiology, and End Results program. Age adjustment rates were based on the 1970 US population. Specific racial and ethnic information was only adequate to determine black-white differences.
For staging, 11 registries used the SEER program’s Extent of Disease system, which converts to a Summary Stage. The remaining 17 registries use the Summary Stage directly. Internal analysis by Wu et al revealed 95% agreement between the 2 systems, which justified combining stage data for all registries. Staging collapsed into 4 major stage categories: localized (confined to the colon or rectum), regional (extension to adjacent tissues or regional lymph nodes), distant (metastastic to other organs or distant areas of the body), and unknown.
The colon was divided into 4 major subsites: proximal colon (cecum, ascending colon, hepatic flexure, transverse colon, splenic flexure), descending colon, distal colorectum (sigmoid, rectosigmoid, rectum), and others (appendix and NOS).
Among the 344,775 colorectal cancer patients older than age 50 whose data were available for analysis, 88.3% were white, 8.3% were black, 2.8% were other races, and 0.6% were unknown. Consistent with other epidemiologic studies, 41.2% of tumors were located in the proximal colon, 4.7% in the distal colon, and 49.3% in the distal colorectum. Stage information was available for 93.1% of tumors overall.
For all race, gender, and age groups, localized cancer rates increased from 31% in the proximal colon to 37% in the descending colon to 41.5% in the distal colorectum. Conversely, percentages of regional disease were much higher proximally and decreased more distally in the colorectum. Percentages of distant disease remained similar throughout anatomic subsites. Taken together and given the relatively easy access to the distal colon and colorectum for polypectomy and early tumor identification, these data imply a beneficial effect of screening for the population overall.
For both genders, age-adjusted incidence rates for proximal and descending colon cancers were higher in the black population than the white. However, as in previous reports,1-3 cancer incidence rates in the distal colorectum were highest for white males. Also consistent with prior research, males had a higher incidence of cancer at all anatomic subsites than females. Moreover, male-to-female rate ratios increased with more distal progression throughout the colon. Some have suggested the gender difference in incidence rate may result from a protective estrogen effect; however, the mechanism of site-specific effects remains to be clarified.4
Examining subsite-specific stage, in contrast to incident disease, revealed that blacks were more likely than whites to have advanced disease and less likely to have localized disease at all anatomic subsites. For example, in the distal colorectum, blacks had 35.3% localized tumors, 41.4% regional tumors, and 23.2% distant (metastatic) tumors, while rates for whites were 41.9%, 40.7%, and 17.4%, respectively. The most pronounced black-white differential among anatomic subsites was that for localized (early) stage tumors in the distal colorectum. These data suggest both that black patients may benefit from more extensive screening regimens (that include the right colon) and that black patients are not receiving any screening as frequently as white patients.
Lastly, among all races and genders, incidence rates increased with increasing age group and the rate of increase was steepest for proximal tumors. Considering that, as above, localized disease rates increased from proximal to distal sites (that is, proportionately more advanced stage disease was found more proximally), these data suggest that disease may be detected earlier or prevented more effectively in elderly patients who undergo full colonoscopy screening rather than flexible sigmoidoscopy.
In summary, Wu et al found that women, elderly patients, and black patients appear to comprise subpopulations at higher risk for proximal tumors. Furthermore, black patients are at higher risk for more advanced stage of colorectal cancer at presentation. They conclude that screening efforts specifically directed at proximal tumors for these patient groups may be more effective in identifying earlier stage, and therefore more curable, colorectal cancer.
Comment by Arden Morris, MD
When different subpopulations present with dissimilar sites of incident disease, the potential implications range from differential disease behavior, to discrepancies in patient or provider behavior, to inequitable access to systems of care. Wu et al investigated these dissimilarities in colorectal cancer incidence further by examining the relationship between stage of disease and anatomic subsite, stratified by race, age, and gender. Because colorectal cancer screening and prevention are uniquely combined, such a study allows exploration of the etiologic influence of differential screening strategies. For example, previous work has documented higher prevalence of right-sided polyps in women.2 Since polyp formation is an early step in malignant transformation, one might deduce that women are more susceptible to right-sided colon cancers. Furthermore, these cancers may be prevented by polypectomy if full colonoscopy screening were performed but wouldn’t be identified by flexible sigmoidoscopy. Determining whether all women older than age 50 should undergo full colonoscopy screening every 10 years is a complex issue; risks and costs must also be considered. However, population-based outcome studies like this one add to our knowledge about which subpopulations may benefit from more extensive screening efforts.
Wu et al collapsed the colon into 4 anatomic subsites: proximal colon (cecum to splenic flexure), descending colon, distal colorectum (sigmoid to rectum), and other. The division was both anatomic, with mid-gut to hind-gut blood supply transition at the splenic flexure, and also relevant to screening strategies—a flexible sigmoidoscope typically extends 60 cm to the splenic flexure. Dissent in the literature regarding whether to include the splenic flexure in the right (proximal) colon vs. left (distal) colon was not discussed; however, given that this study sought to address screening issues, one might argue that the splenic flexure should have been grouped with distal tumors rather than proximal.
Staging was also collapsed into 4 major categories: localized (AJCC Stage I or II), regional (AJCC Stage II or III), distant (AJCC Stage IV), and unknown. For accurate data merging this was probably necessary, although using an AJCC or Dukes staging regimen might have greater resonance for clinicians.
For a cohort of this magnitude (n = 344,775), even tiny percentage variations carry statistical significance. Therefore, confidence intervals or P values were appropriately not reported and, instead, only clinically meaningful data were highlighted.
The study found an increasing percentage of localized disease with anatomic progression through the colon. This finding persisted for both genders, and among all racial groups and age groups. Within each anatomic subsite, men had increased incidence of disease but both genders appeared to have similar stages of disease. In contrast, blacks had an increased incidence rate of proximal disease and a more advanced stage of disease at every anatomic subsite than whites.
Wu et al did not clarify whether there was any racial variation in patients whose cancers had available stage information. For example, 8% of the cohort was black and 7% had missing stage information. Were black patients disproportionately represented in the missing category? Such a scenario may have a misleading affect on the ultimate conclusions, although this may be mitigated by special biostatistical techniques such as multiple imputations.
In conclusion, population differences in subsite-specific stages lend support to Wu et al’s argument for targeted screening of susceptible subgroups. However, further evaluation of risks, costs, and other possible etiologies for advanced stage of disease (e.g., differential tumor biology or exposure variables) is warranted before broad scale implementation of full colonoscopy as the most effective screening strategy.
Dr. Morris is the Robert Wood Johnson Clinical Scholar, University of Washington, Seattle, WA.
References
1. Nelson RL, et al. Cancer. 1997;80(2):193-197.
2. McCashland TM, et al. Am J Gastroenterol. 2001; 96(3):882-886.
3. Demers RY, et al. Cancer. 1997;79(3):441-447.
4. Grodstein F, et al. Ann Intern Med. 1998;128(9): 705-712.