Trends in colorectal cancer incidence: a period and birth-cohort analysis in a well-defined French population

In France, colorectal cancer is currently the third most common cancer in men and the second in women [1]. In 2005, the number of new cases was estimated to be 37,413 for both genders with age-standardized rates of 37.7/100,000 in males and 24.5/100,000 in females [1]. Incidence in France is comparable to that found in other high-risk areas of Western Europe, North America, Australia/New Zealand, and Japan [2], whereas it remains lower in Africa and Asia. Colorectal cancers were long considered to be a homogeneous entity, but several studies suggest that their characteristics differ by anatomical subsites [3, 4]. Temporal trends in incidence are related to three main factors: age, period of diagnosis and birth cohort. Study of colorectal cancers using age-period cohort models is commonly used in order to better understand the observed trends and aetiological factors connected with them [5, 6]. These models also allow estimation of the cumulative risk of developing a colorectal cancer for a given birth cohort [7]. The aim of this paper is to provide updated temporal trends in colorectal cancer incidence over a 30-year period in a well-defined French population, and to specifically differentiate the influences of period of diagnosis and birth cohort.

Over a 30-year period, the overall incidence of colorectal cancer in France has slightly increased in males and remained stable in females. In the youngest birth cohorts the cumulative risk over the age range 0-74 years slightly decreased in males whereas it slightly increased in females. The most impressive aspect of our study was the marked increase in the incidence rates of right and left colon cancers both by time period and birth cohort. The cumulative risk of sigmoid and rectal cancers decreased in males while it remained relatively stable in females.

The Burgundy registry is the most longstanding population-based digestive cancer registry in France. The multiplicity of information sources allows us to assume that nearly all newly diagnosed cancers are recorded. Thus, it is an appropriate tool for observing time trends in incidence rates. When analyzing 30 years of incidence, it is necessary to evaluate the comparability of data over such a long period. The strength of our results lies in the fact that the registration scheme and the coding rules remained the same over the study period. Colonoscopy has progressively replaced barium enema as a diagnostic procedure. However, this was unlikely to explain time trends in incidence.

Five volumes of the publication Cancer Incidence in Five Continents provide data on time trends on colorectal cancer incidence over the period covered by this paper [11‐15]. Different trends in CRC incidence can be seen around the world. In Western Europe, incidence rates were stable or slightly increased, with the exception of Norway and Spain where the increase in incidence was more pronounced. Incidence is generally on the increase in East European countries, Central and South America and in Asia. The most prominent feature was the sharp increase in incidence in Japan, which now stands among high-risk areas [16]. After an increase in incidence until the late 1980s, a decrease in incidence was reported in the USA, while it remained stable in Canada. Incidence rates were also stable in Australia and New Zealand [17].

Availability of incidence rates by subsites is more limited. Evidence exists for different time trends, in particular among high-risk areas. In the USA, stable incidence rates for right colon cancer, and declining rates for left colon (descending and sigmoid) and rectal cancer were reported over the 1992-2001 period [18]. Data covering the 1997-2006 period indicates that all CRCs are now significantly declining in the US [19]. Among available data in Europe, an increase in all colon subsites was reported in Denmark [4, 20], Norway [21] and Italy [22, 23], whereas in Burgundy, the increase was limited to right and left colon cancers. A recent study from Norway reveals deceleration in the rate of increase of colon cancer subsites [24]. In most high-risk areas, the incidence of rectal cancer decreased (USA, Denmark, France) or was stable (Italy, Australia), whereas it increased in Norway [24‐27]. In low-risk areas of Asia, Central and South America, there was a striking increase in incidence of all colon subsites, while it was moderate for rectal cancers [11‐15].

Most studies on trends in colorectal cancer incidence are limited to analysis by period of diagnosis. Trends by birth cohort bear different implications and are thus complementary. Effects of both period of diagnosis and birth cohort were found in this study, as well as in other studies [21, 26, 27]. However for sigmoid cancers the birth cohort effect was more marked in males and the period of diagnosis effect in females. Exposure to early-stage risk factor or protective factor with long-term latency periods over different generations will introduce a cohort effect. A period of diagnosis effect can be attributed to risk factors or protective factors involved in the late stage of carcinogenesis affecting all age groups, or to changes in screening practices. The protective role of vegetables reported in case control studies (and not in cohort studies) could be attributed to a period effect. Case control studies, concerning recent intakes, provide data on dietary factors involved in the last phase of colorectal carcinogenesis, while cohort's studies provide data on the early phase of this colorectal carcinogenesis. Dietary factors affecting the last step of the adenoma-carcinoma sequence can modify cancer risk in a relatively short period of time. An example of this period effect is given by the westernisation of diet in Japan. It became a high-risk country for colorectal cancer in about 15 years following changes in dietary behaviour [28]. A similar effect was observed in Japanese migrants to the USA [29]. There are discrepancies between case-control and cohort studies, but these are not surprising. Cohort studies collecting data on diet a long time before the appearance of the cancer are able to identify the main dietary factors related to the cohort effect, while case-control studies that take recent diet into account reveal factors related to the period effect. There is a tendency to consider that in case of discrepancies, cohort studies are more representative of the truth. As factors involved in adenoma appearance or growth differ, at least partly, from those involved in adenoma transformation, then cohort and case control studies should provide somewhat different results.

Data from the United States suggest an impact of screening on the decline in mortality and incidence from CRC. The 2010 US annual report to the Nation on the status of cancer provides the results of a simulation model for interpreting mortality trends for CRC. The results suggest that 35% of the observed mortality decline can be explained by changes in risk factors, 53% by screening and 12% by improvement in treatment [19]. However, the impact on incidence rates can only be seen after a long time lag [30]. In a cohort of patients with adenomas measuring more than 1 cm in diameter that remained in the colon, the cumulative colon cancer rate was 8% after 10 years and 24% after 20 years [31]. The marked decrease in the incidence of CRC in the US, between 1985-1995 cannot be attributed to screening, which was not widespread in the 15-20 years preceding that period. Changes in risk factors or protective factors probably played the most important role. A significant positive effect of screening on CRC mortality or incidence can be expected only during the most recent period. The decline in CRC incidence could be accelerated further by higher utilisation of screening. Trends in incidence in Europe cannot be related to screening practices because of too recent development of screening programmes. For instance Norway has not undertaken CRC screening programmes and less than 5% of colonoscopies performed are related to CRC screening [24].

The different time trends by subsite within areas, and among areas with similar overall CRC incidence, suggest that dietary factors involved in CRC carcinogenesis along the length of the colon might differ, at least partly, and that the protecting or enhancing factors may differ from one area of the world to another. For instance meat consumption is a more important risk factor in North America or Australia where consumption is higher than in Europe, and refined cereal consumption (in particular pasta and rice) are risk factors only in Latin European countries where consumption of these products is high [32]. There are also data indicating that aetiological factors may differ along the large bowel. In a case-control study performed in the Côte-d'Or area, the protective effect of high vegetable intake was restricted to the sigmoid colon. Factors that might be more linked to right colon cancer include body mass index. It has increased over the past few decades in France. The increased incidence of right colon cancer could be related to the increase in obesity.

The strength of our study lies in the fact that it is a population-based study, using high-quality cancer registry data collected over a 30-year period, with time trends analysed both by period and birth cohort. Incidence rates in Burgundy are similar to those reported at the national level [1]. So we feel that our findings can be generalised to the French population overall. This study shows that the incidence increased in males and was stable in females over the 1976-2005 period. In revealing different time trends in incidence for subsites and genders, our results suggest that CRCs have, at least in part, different aetiological factors. Differences are also reported among cancer registries. Since colorectal cancer is one of the most common cancers it is important to conduct future studies that take into account environmental, genetic and molecular factors. The shift towards right colon cancers reported here has implications for the choice of colorectal cancer screening methods.