Contrast between Cancers of Colon and Rectum
One purpose of this chapter is to examine demographic characteristics associated with the incidence and mortality of colon and rectal cancers to identify any differences in these variables between these two sites. Theories regarding the etiology or course of cancers of the intestinal tract should conform to the basic demographic characteristics of cancers. If these characteristics differ between cancers of the colon and rectum, then studies of etiology or course of the cancers should examine the two sites separately. The analyses of incidence rates based on data from 41 NPCR and SEER registries indicate a male predominance in rates for both sites, but it is much more pronounced in the rectum than in the colon. Incidence rates are higher in blacks than in whites for colon cancer. Conversely, whites have slightly higher rates of rectal cancers than blacks. Asians and Hispanics have lower rates of colon cancer than other races or ethnic groups, but their rates of rectal cancer are similar to other races. Both the shapes and slopes of the log-linear curves of age-specific rates differ between the colon and rectum. These observations suggest that these cancers are not similar with respect to basic population characteristics. It is probable that we are looking at different cancers perhaps with different risk factors that must be examined separately.
Another question is whether the current anatomical designations used to define these cancers reflect similar distributions by race and sex. Examination of subsite data indicates that these different demographic characteristics of colon and rectum cancers actually begin in the lower colon. High black-to-white and low Asian-to-white ratios occur in the proximal colon. However, these ratios change at about the sigmoid for blacks and at the descending colon for Asians, generally reaching rates that are more similar to those of whites in the lower colon, and these relationships continue through the rectum. The male/female ratio shows a small excess in males in the upper colon, but the excess abruptly rises at the splenic flexure to the 1.5- to 1.7-fold excess commonly seen in the rectum. Thus, separation of the intestinal tract into the traditional split between colon and rectum at the rectosigmoid junction would include cancers with rectal cancer characteristics among all colon cancers. The current division is solely based on the anatomy of the tract and may not be the optimum separation for epidemiologic investigations of etiology. For such studies the division should be based on grouping by observed demographic characteristics.
Interpretation of Demographic Data
Many personal characteristics such as excess weight, physical inactivity, and poor diet are considered risk factors for colorectal cancers.13, 14 The question is whether the demographic characteristics of these cancers support the role of these factors as risks. For example, theoretically, if diet were a risk factor for CRC, then the incidence rates for males and females within a race should be more similar than the rates for different races. Yet, the predominant demographic characteristic for every race, especially for cancers below the splenic flexure, appears to be the consistently higher rates in males. Only in the proximal and transverse colon do the rates of colon cancer differ by race, with higher rates in blacks and very low rates in Asians as compared with whites. Therefore, although differences in rates by race in the upper colon support the hypothesis that diet may play a role in the risk of colon cancer, the inconsistencies in incidence by race in the rectum and descending colon, the abrupt changes in risk by race at colon subsites, and the high ratios for sex compared with race do not fit with that theory.
Many studies have noted changes in demographic characteristics of CRC based on subsite.4–8, 15 The studies have used different anatomic locations to define proximal and distal cancers. However, the demographic characteristics by subsite deserve particular attention. Normally, the expectation would be that changes in the distribution of demographic characteristics of cancers would occur gradually by subsites, but the changes seem to occur abruptly. Previous studies have suggested that changes in M/F ratios from cecum to rectum have occurred gradually.5, 16 Only Wu et al. noted abrupt changes in B/W ratios for subsites in the colon.5 The abruptness of the change in incidence for both race and sex by subsite in the lower colon in the current analysis might suggest an effect from screening, but this is unlikely to be the only reason. If screening were playing a major role, then we would expect W/B ratios in the lower colon and rectum to have increased over the past 15 or 20 years with the more widespread use of screening, especially in whites. Instead, SEER data for 1975–2001 indicate that the W/B ratio was high for rectal cancer in 1975, but by 2000 rates were almost equal for the two races.15 Furthermore, screening cannot explain the abrupt changes in racial ratios as high as the descending colon or the sex-specific changes at the splenic flexure given that screening using instruments to detect lesions above the sigmoid is a recently introduced procedure.
Evidence supports the theory that cancers might differ by site because of differences in the biologic characteristics of lesions at points along the intestinal tract.9, 10 Tumors of the upper colon have a higher proportion of MIN cancers than do cancers in the lower colon.9, 10 However, for the entire colorectal tract, the frequency of MIN cancers is only 20% of all cancers. Thus, their unequal distribution would have a greater impact on cancer rates in the upper tract than in total CRC. MIN cancers also are reported to occur more frequently in blacks and in women and to have poorer survival than CIN cancers.9 MIN cancers do not occur in adenomatous polyps, the target of screening. Therefore, increased screening with removal of polyps could potentially increase the proportion of MIN cancers among all cancers and change the ratio of cancer types at lower colorectal tract levels. Thus, increasing the proportion of MIN cancers through screening would result in poorer survival and higher B/W ratios especially in the lower colon and rectum. This outcome has not yet occurred. Still, by 2000, B/W ratios were approaching 1.00 for rectal cancers, whereas ratios had been below 1.00 about 25 years previously.15 Certainly, the existence of two cancers with major differences in genetic characteristics that are related to race and sex would suggest varying etiologic factors for these different cancer types.
The next obvious question is whether the predominance in males is related to basic physiological or other differences between the sexes. Such factors might include diet and obesity as well as drug therapies, aspirin, and hormones. The influence of many of these factors cannot be tested with the use of surveillance data. One such difference that could be examined, however, is the effect of natural premenopausal estrogens in women that might suppress CRC at early ages. If these hormones do play a preventive role, then the expectation is that age-specific incidence data would indicate that, at postmenopausal ages, women should show an increase in incidence as seen with other diseases, but none occurs. The literature also provides little support for this theory except for one study that has shown that use of postmenopausal estrogen may suppress MIN cancers specifically.17 Since these cancers represent only 20% of all CRC, any protective effect of estrogen on this tumor type may not be apparent when total cancers are studied.
Colon and rectal cancers have differences in the shapes of the age-specific incidence curves by race that may relate to the etiology or course of CRC. The results show that the slopes of incidence curves for each sex for both cancers are parallel by race for ages 40 to 60 years. The striking difference is that although the rates for ages 40–60 differ by race; for rectal cancers, all races have similar rates for this age range. Thus, the difference in incidence of colon cancer by race is primarily due to starting with incidence rates in blacks that are higher under age 40. This may represent differences related to susceptibility or to exposures in the young by race. The higher incidence rates of rectal cancers in whites occur generally because of higher rates after age 65 as compared with other races. The shapes of the age-specific incidence curves by race for both sexes are similar for colon cancer, but the slope of rectal cancer incidence is not as steep in females. These observations emphasize the need to examine these sites separately. For example, studies to explain differences in sites by race for colon cancer must focus on the youngest age group, whereas for rectal cancers the oldest age group is the major source of racial differences. In addition, the overall excess in M/F ratios for colon cancer is represented by small and varying differences by sex for each age. Rectal cancers, on the other hand, have consistently lower age-specific rates in females at all ages, which might indicate a physiologic or exposure difference in risk by sex.
For both colon and rectal cancers in males and females, the age-specific rate curves for blacks at about age 65+ have slopes that are less steep than those for whites. This change results in age-specific rates that become equal for colon cancer and that become higher in whites for rectal cancer at these older ages. Several explanations are possible for the slowing in rates of increase in incidence of colon and rectal cancers by age in blacks. This finding could result from lower rates of cancer detection, slower growth rates of cancer, or birth cohort differences in older blacks compared with whites. If birth cohort differences are the explanation for lower rates at older ages, then this would suggest a later introduction of risk factors into the black compared with the white population. Evidence in the literature suggests that CRC mortality rates for whites compared with blacks were higher around the middle of the 20th century, after which rates rose in blacks to exceed whites in the past three or more decades.2 The true explanation for the decline in the slopes of black incidence at older ages is unclear, but the differences in B/W ratios in older compared with younger ages do not occur because of an increase in the slopes of the incidence curves at age 60 in the white population.
As noted in the introduction, CRC mortality has changed its demographic characteristics over the past 40 to 50 years. In addition to changing trends in mortality, SEER incidence data from 1975 to 2001 can help track these cancer changes and determine the predicted directions of future rates.15 The published NCI data tables of joinpoint analyses permit an examination of trends from 1975 to 2001 with an evaluation of whether the rates over time are represented by a curve with a single slope or one with multiple segments having different slopes. The death rates in whites indicate a progressively steeper declining curve from 1975 to the present. In blacks, the death rates actually increased to 1990 after which there was a decline in rates that was less than half of that in whites. This pattern has resulted in the differences in mortality seen in the current analysis in which the B/W ratio for colon cancer mortality is much higher than the ratio for the incidence. For rectal cancers, there is actually a reversal in the racial ratio for incidence versus mortality with higher incidence in whites than in blacks but higher mortality in blacks. This combined effect of increasing mortality in blacks for the first 15 years of the period while whites showed declining mortality rates and subsequently a smaller decline in rates for blacks when compared with whites after that period has led to the curves becoming more divergent over time. This is a major concern because it represents a serious racial disparity that will only increase in the future if the current trends continue.
Differences in incidence trends by race have also added to the greater burden of CRC in blacks when compared with that in whites. The 27-year joinpoint analysis for 1975–2001 indicated that the incidence of colon cancer rose in both races for about the first 10 years with an increase that was three times higher in blacks than in whites. Both colon and rectal cancer incidence rates then began to decrease in whites, but there has been no decline in incidence rates for either cancer in blacks. Therefore, the reported trends indicate that the cancers are occurring at about the same incidence over the past 20 years in blacks accompanied by a mortality that rose until the last 10 years of the period and is now falling much more slowly than that of whites. The result has been high incidence rates of these cancers in blacks in recent years when compared with results from about 30 or more years ago, and death rates that show an even higher predominance in blacks.