A Case-Control Study on Familial Aggregation of Colorectal Cancer

A Case-control study was conducted in 110 patients with colorectal cancer and, as control groups, 84 with gastric cancer, 50 with other cancer and 111 with benign gastroenterological disease. In this study we try to make it clear whether colorectal cancer has really familial aggregation and whether a positive family history can be used as one of screening procedures, by getting information more exactly about family history and by taking account of family size into analysis. 9.1% (OR 1.20) of colorectal cancer group and 11.3% (OR 1.70) of colon cancer group had positive family history of colorectal cancer, as compared with the benign disease group. 1.8% of their first-degree relatives were affected by colorectal cancer. No significant difference was seen, although these proportions were higher than those in 3 control groups. These results suggest familial aggregation of colorectal cancer is not so large as reseachers have mentioned so far, although a tendency for familial aggregation was found in colon cancer. At the present stage in Japan, we should not perform screening for the first-degree relatives only by positive family history from the reason of efficiency, though those of patients with clinically diagnosed or suspected HNPCC should be screened. J Epidemiol, 1995; 5 : 165-169.

taking and analysis.
It is not obvious whether colorectal cancer has familial aggregation in Japan.
In this study we try to make it clear whether colorectal cancer has really familial aggregation and whether a positive family history can be used as one of screening procedures, by getting information more exactly about family history and by taking account of family size into the analysis.
MATERIALS AND METHODS 651 patients were admitted at the Department of Gas-troenterology in Tokyo Metropolitan Komagome Hospital from July 1990 to May 1991. Of those, study population was 327 patients taken family history exactly as mentioned below. These 327 patients included 24 with double cancers and 2 with tripple ones. This led the total study population to 355 patients. These 355 patients consisted of 110 with colorectal cancer (colorectal cancer group ; colon cancer 80 cases, rectal cancer 35 cases,including overlapping) and, as control groups, 84 with gastric cancer (gastric cancer group), 50 with other cancers (other cancer group) and 1ll with benign gastroenterological disease (benign disease group). The mean age was about 64 years old in the three malignancy groups, while 55 in benign disease group. The number of males was larger than that of females in all the groups ( Table 1).
The first-degree relatives (parents and siblings, except  Table 2). The age distribution of the first-degree relatives was shown in Table 3. Dead cases accounted for about a half in each group. This is because parents had already been dead in many cases.
In taking family history, we used an original questionnaire, defining objects as the first-degree relatives, and confirming the patient and/or his family whether or not to have past history of cancers one by one mainly in the gastroenterological system. This study on familial aggregation was performed between the colorectal cancer group and the three control groups, according to the observation, first of family units and, secondly of individual family members. The former was a case-control study and we observed adjusted Odds ratio by using Mantel-Haenszel method. Age was classified into three ; <50 years, 50~69 years,>70+. In addition, we divided colorectal cancer group into colon and rectal cancer group, and compared them with benign disease group.
Besides, we also examined the frequency of hereditary non-polyposis colorectal cancer (HNPCC). Criteria for HNPCC are defined as below : 1) a case with 3 or more colorectal cancer cases within the first-degree relatives, 2) a case with 2 or more colorectal cancer cases within the first-degree relatives and with any of the following ; a) age at onset of colorectal cancer is younger than 50 years old, b) the right colon involvement, c) multiple colorectal cancers, d) association with extracolorectal malignancies. Table 3.
Age Distribution of first degree relatives.

RESULTS
We first investigated a positive family history for cancers according to the observation of family units ( Table 4). The highest proportion of a positive family history for colorectal cancer was 9.1% in the colorectal cancer group. No significant correlation was observed between the colorectal cancer group and the gastric cancer group (OR 1.73, 95% CI 0.5~5.9), the other cancer group (OR 1.35, 95% CI 0.37~4.9) or the benign disease group (OR 1.20, 95% CI 0.46~3.1). Analyzing colon and rectal cancer separately, the frequency of positive family history of colorectal cancer was 11.3% (OR 1.70,95% CI 0.83-3.48) in the colon cancer group and 5.7% (OR 0.81, 95% CI 0.26~2.53) in the rectal cancer group (Table 4). On the other hand, the highest proportion of that for gastric cancer was 29.8% in the gastric cancer group.
The HNPCC patients were 4 (3.6%) out of 110 with colorectal cancer.
Secondly we investigated the frequency of colorectal cancer among the first-degree relatives ( Table 5). The cases noted to suffer from colorectal cancer were 12 out of 679 (1.8%) first-degree relatives in the colorectal cancer group, 6 out of 500 (1.2%) in the gastric cancer group, 4 out of 304 (1.3%) in the other cancer group and 7 out of 615 (1.1%) in the benign disease group. No significant difference was seen among them, although the highest propor-tion of colorectal cancer was found in the first-degree relatives of the colorectal cancer group. The highest proportion of gastric cancer was 5.6% in the first-degree relatives of the gastric cancer group. Similarily the highest proportion of other cancer was 10.2% in those of other cancer group.

DISCUSSION
This case-control study suggests familial aggregation of colorectal cancer is not so evident as researchers have shown so far.
Studies on familial aggregation of colorectal cancer were advanced mainly by reseachers in the Western countries. In 1958, Woolf1) reported that colorectal cancer had hereditary background. In 1960, Macklin2) reported that gastric cancer had familial aggregation of gastric cancer and colorectal cancer had that of colorectal cancer. The frequency of positive family history of colorectal cancer has been reported as follows : 11.3% (2.36-fold as high as that in the control group) by Bonelli Table 5. Frequency of cancers in first degree relatives of subjects.
FDR : First degree relatives report has suggested that the frequency of positive family history was more than 2-fold of that in the control group.
We could not find many reports about familial aggregation of colorectal cancer in Japan. According to the report of the Japanese Research Society for Cancer of the Colon and Rectums), the frequence of positive family history (including the second-degree relatives) was 6.5% in 15369 cases of colorectal cancer. Yoshida et al.10) reported that the proportion was 8.2% (OR: male 2.18, female 2.15), being significantly higher than that in the control group. It was 5.9% by Murata et al.11). Kato et al. 12) also reported that the relative risk was 2.8-fold in the colon cancer group and 1.8-fold in the rectal cancer group compared with the other cancer group, though their data included the seconddegree relatives. Most of the reports in Japan have also suggested that positive family history is one of the risk factors for colorectal cancer. The only exceptional report is the one by Watanabe et al.13) They showed it was not evident whether positive family history might be a risk factor of colorectal cancer.
In our study those who have a positive family history in the colorectal cancer group was 10 out of 110 cases (9.1%) according to the observation of family units. This proportion was higher than that of the previous Japanese reports9- 11,14), but odds ratio was only 1.2 compared with the benign control group. If HNPCC might be excluded from colorectal cancer group, the proportion would be 5.7% and lower than that of control group. According to the observation of individual family members, 1.8% of their first-degree relatives of the colorectal cancer group were affected by colorectal cancer. Although these proportions were also higher than those in 3 control groups, there was no significant difference. This study suggests that familial aggregation of colorectal cancer may not be, even if any, so large as researchers in the Western countries have mentioned so far.
On the other hand, analyzing colon and rectal cancer groups separately, Odds ratio of the former increased to 1.70, and that of the latter decreased to 0.82. Although colon cancer may have a tendency for familial aggregation, it was not so large as researcher have mensioned so far. Rectal cancer may not have a tendency for familial aggregation.
We can not deny a tendency toward familial aggregation of colorectal cancer, because it is now evident that colorectal cancer is derived from accumulation of gene disorders. Since common colorectal cancers may be derived from gene disorders in the somatic cell line except special cases such as familial adenomatous polyposis, HNPCC and so on, such disorders would not necessarily be transmitted to the next generation. Recently, Fuchs et al.15) reported by a prospective study that the relative risk of the colorectal cancer for persons with affected first-degree relatives was 1.72 (colon cancer : 1.99, rectal cancer : 0.86) as compared with those without a family history of the disease. They also said the relative risk increased to 5.37 especially for persons under the age of 45 years with affected first-degree relatives. This suggests the age is one of the significant factors. In addition John et al.7) reported, by a case-control study, that odds ratio was 1.8 in those who have one affected first-degree relative and that it increased to 5.7 in those who have two. It suggests a patient should be regarded as a high risk group based on not only a positive family history but also on his age and on the age at diagnosis and number of affected first-degree relatives.
HNPCC should be obviously regarded as a high risk factor. It is evident HNPCC has hereditary susceptibility, and its gene disorder is now revealed 16,17). Although genetic diagnosis may be possible in future, at present we should perform screening examinations positively for the first-degree relatives of the patients clinically diagnosed HNPCC.
We must recognize to have some problems in this study. One is that sample size was not so large. Secondly sex and age in some of family members were unknown. The last is that we chose benign gastroenterological disease, of which familial aggregation might be a little higher than that of general population, as a control group.
Two things are pointed out on analysis of family history. One is the way of getting information and another is the way to analyze it. In taking family history, one of the biggest problems is uncertainty that it may be influenced by personal memory. As the means of settling it, three conditions as below are thought to be needed. They are 1) to restrict the objects only to the closest possible relatives, 2) to interview patients and their family directly, instead of self-recording, 3) to confirm the name of the disease by death certificates or medical records as far as possible. The only report which met all of these three conditions was presented by Leon et al.8) among all reports in Japan and the Western countries within the limits of our investigation. We could find nothing satisfying the conditions of only 1) and 2) among the Japanese reports.
In our study we restricted the objects of the questionnaire to the first-degree relatives and interviewed the patient and/or his family directly presenting them the concrete name of disease and confirming them about its existence one by one. In this way, we could obtain much more exact and detailed family history.
Another problem is the way to analyze family history . Family size has not generally been considered , when thinking about the frequency of positive family history of cancers. In studying family history, it is pointed out we should take family size into consideration18) . We could find such a considerataion in a few reports by Macklin2) , John et al.7), Leon et al.8), and Watanabe et al. 19) within the limits of our investigation.
As a matter of course, it depends on the family size how often colorectal cancer occurs in the family of patients with colorectal cancer.
When the number of the first-degree relatives may be decreased by a decline of birth rate, the prevalence of colorectal cancer with positive family history would reduce. However, according to the observation of individual family members, an error of the prevalence might be fewer because a base population could also reduce at the same time. In addition, when more than 2 first-degree relatives had colorectal cancer in the same family, more correct prevalance could be presented by treating the first-degree relatives as a denominator.
Consequently, it is necessary to use frequency of cancers among the first-degree relatives in analyzing family history.
In conclusion, we have shown that familial aggregation of colorectal cancer is not so large as researchers especially in the Western countries have mentioned so far, although a tendency for familial aggregation was found in colon cancer. At the present stage in Japan, we should not perform screening examinations for the first-degree relatives only by positive family history from the reason of efficiency, though those of patients with clinically diagnosed or suspected HNPCC should be screened.