Time trends in colorectal cancer incidence in four regions of Latin America: 1983-2012 Tendências temporais na incidência do câncer colorretal em quatro regiões da América Latina: 1983-2012 Tendencias temporales en la incidencia del cáncer colorrectal en cuatro regiones latinoamericanas: 1983-2012

This study aimed to assess time trends in colorectal cancer incidence from 1983 to 2012 in Latin America. This was an ecological time-series study whose population consisted of individuals aged 20 years or older diagnosed with colorectal cancer. Data from population-based cancer registries in Cali (Colombia), Costa Rica, Goiânia (Brazil), and Quito (Ecuador), were used for rates estimation, while time trends estimations were proceeded by the Joinpoint Regression Program. The study showed an increase in colorectal cancer incidence in men and women in Cali (2.8% and 3.2%, respectively), Costa Rica (3.1% and 2.1%, respectively), and Quito (2.6% and 1.2%, respectively), whereas in Goiânia, only women showed an increase in colorectal cancer rates (3.3%). For colon cancer, we observed an increasing trend in incidence rates in men and women in Cali (3.1% and 2.9%, respectively), Costa Rica (3.9% and 2.8%, respectively), and Quito (2.9% and 1.8%). For rectal cancer, we observed an increasing trend in incidence in men and women in Cali (2.5% and 2.6%, respectively), Costa Rica (2.2% and 1%, respectively), and Goiânia (5.5% and 4.6%, respectively), while in Quito only men showed an upward trend (2.8%). The study found increases in colorectal cancer, colon cancer, and rectal cancer in four Latin America regions. This findings reflect lifestyle, such as dietary changes, following the economic opening, and the prevalence variations of colorectal cancer risk factors by sex and between the four studied regions. Finally, the different strategies adopted by regions for colorectal cancer diagnosis and screening seem to influence the observed variation between anatomical sites. Colorectal Cancer; Incidence; Time Series; Latin America Correspondence T. C. Carvalho Escola Nacional de Saúde Pública Sergio Arouca, Fundação Oswaldo Cruz. Rua Leopoldo Bulhões 1480, Rio de Janeiro, RJ 21041-210, Brasil. thayanacalixto@gmail.com 1 Escola Nacional de Saúde Pública Sergio Arouca, Fundação Oswaldo Cruz, Rio de Janeiro, Brasil. 2 Instituto Nacional de Câncer José Alencar Gomes da Silva, Rio de Janeiro, Brasil. This article is published in Open Access under the Creative Commons Attribution license, which allows use, distribution, and reproduction in any medium, without restrictions, as long as the original work is correctly cited. Carvalho TC et al. 2 Cad. Saúde Pública 2021; 37(10):e00175720 Introduction Along with the third-highest cancer incidence (excluding non-melanoma skin cancer), colorectal cancer is considered a public health problem worldwide, especially in Latin America, ranking as the most common digestive system neoplasm in 2020 1. Distribution varies widely between different regions of the world according to the level of development. Regions with high human development index present the highest incidence rates, varying from 33.6 per 100,000 person-years in Northern Europe to 18.5 per 100,000 person-years in South America, and 10.4 per 100,000 person-years in Central America in 2020 1. However, evidences suggest growing incidence in countries with economic transition 2, alongside stabilization or reduction in developed countries’ rates in the last 15 years 3. Although several studies have analyzed colorectal cancer incidence worldwide, only a few were carried out in Latin America 4. Thus, the behavior of this neoplasm incidence in developing countries is not well defined. Some studies performed so far have assessed the trend in incidence rates for only 10 years 4,5 or were conducted 20 years ago 5, besides not assessing the colon and rectal cancer patterns separately 6. Arnold et al. 2 conducted a complete study to date, using the International Agency for Research on Cancer (IARC) data from 1980 to 2007, and found three distributions of colorectal cancer: increasing incidence and mortality rates in recent decades in countries with rapid economic transition, including Brazil, Costa Rica, and Colombia; increasing incidence and decreasing mortality in high Human Development Index (HDI) countries, including Canada and the United Kingdom; and decreases in both rates in countries with very high HDI as Australia, Japan, and France. However, the study did not analyze the rates by anatomical site, nor did it include Ecuador among the studied countries. Also, Arnold et al.’s study focused on socioeconomic differences between developed and developing countries, without considering the political, social, and economic changes within these countries throughout the study period. Thus, we aimed to assess time trends in incidence rates for colorectal cancer in Cali (Colombia), Goiânia (Brazil), Quito (Ecuador), and Costa Rica from 1983 to 2012, according to sex and anatomical site.


Introduction
Along with the third-highest cancer incidence (excluding non-melanoma skin cancer), colorectal cancer is considered a public health problem worldwide, especially in Latin America, ranking as the most common digestive system neoplasm in 2020 1 . Distribution varies widely between different regions of the world according to the level of development. Regions with high human development index present the highest incidence rates, varying from 33.6 per 100,000 person-years in Northern Europe to 18.5 per 100,000 person-years in South America, and 10.4 per 100,000 person-years in Central America in 2020 1 . However, evidences suggest growing incidence in countries with economic transition 2 , alongside stabilization or reduction in developed countries' rates in the last 15 years 3 .
Although several studies have analyzed colorectal cancer incidence worldwide, only a few were carried out in Latin America 4 . Thus, the behavior of this neoplasm incidence in developing countries is not well defined. Some studies performed so far have assessed the trend in incidence rates for only 10 years 4,5 or were conducted 20 years ago 5 , besides not assessing the colon and rectal cancer patterns separately 6 . Arnold et al. 2 conducted a complete study to date, using the International Agency for Research on Cancer (IARC) data from 1980 to 2007, and found three distributions of colorectal cancer: increasing incidence and mortality rates in recent decades in countries with rapid economic transition, including Brazil, Costa Rica, and Colombia; increasing incidence and decreasing mortality in high Human Development Index (HDI) countries, including Canada and the United Kingdom; and decreases in both rates in countries with very high HDI as Australia, Japan, and France. However, the study did not analyze the rates by anatomical site, nor did it include Ecuador among the studied countries. Also, Arnold et al.'s study focused on socioeconomic differences between developed and developing countries, without considering the political, social, and economic changes within these countries throughout the study period. Thus, we aimed to assess time trends in incidence rates for colorectal cancer in Cali (Colombia), Goiânia (Brazil), Quito (Ecuador), and Costa Rica from 1983 to 2012, according to sex and anatomical site.

Study design and population
We proceeded an ecological time-series study whose population was consisted of individuals aged 20 to 79 years diagnosed with colorectal cancer reported by population-based cancer registries (PBCR) from 1983 to 2012 in Cali (n = 5,528) and Costa Rica (n = 8,595), and from 1988 to 2012 in Goiânia (n = 3,856) and Quito (n = 2,463). Colorectal cancer incidence in the age group up to 19 years old is very low 1 , and the majority is genetic-related cancer cases 7,8 . Therefore, we focused on adult individuals (20+), aiming to raise hypotheses regarding the political, social, and economic changes that could have influenced the incidence rates of colorectal cancer.

Data sources
Cancer Incidence in Five Continents (CI5) series, volumes VI to XI, published by the IARC (https:// ci5.iarc.fr/), were the source of data on the number of colorectal cancer cases in each region, year and age at diagnosis, and size of at-risk populations. Discrepancies on the size of at-risk population in Goiânia between CI5 series and the Brazilian Institute of Geography and Statistics (IBGE) led us to work with the IBGE data for 2003 to 2007.
The CI5 data were obtained from each country or region's PBCR and underwent rigorous quality assessment. Four chosen PBCRs met the criteria of the IARC editorial process, attaining high quality. They were also the only PBCRs in Latin America with at least 20 years of uninterrupted data.
All data provided by the CI5 were coded according to the International Classification of Diseases for Oncology, 3rd edition (ICD-O-3), and converted to the International Classification of Diseases, 10th revision (ICD-10). This process guarantees the use of the same verifications of validity to all data from different regions.

Data analysis
For each region, crude incidence rates were calculated for colorectal cancer (C18-21), colon cancer (C18), and rectal cancer (C19-21) by years, and expressed per 100,000 at-risk person-years, stratified by sex. After this procedure, the incidence rates were standardized by age (ASRs) with the truncated method, using the world population as the standard, as proposed by Segi et al. 9 and modified by Doll et al. 10 . Data on colorectal cancer incidence included a proportion of cases with the unknown age bracket. To address such issue, a correction factor was obtained via the product of the sum of the number of reported cases in a given year, including those with unknown age group and the same sum excluding those with the unknown age group 11 . Thus, this factor corrected the standardized incidence rates of colorectal, colon, and rectal cancer.
The software Joinpoint Regression Program, version 4.5.0.1 (https://surveillance.cancer.gov/ joinpoint/), was used to calculate the average annual percent change (AAPC) for the entire period and the annual percentage change (APC) for each follow-up and the respective 95% confidence intervals (95%CI), using a t-test to determine whether the AAPCs and APCs differed statistically from zero. The adequate log-linear regression models were selected to identify the occurrence of possible inflection points where significant changes in the trends had occurred, allowing a minimum number of joinpoints necessary to adjust the data when the APCs changed significantly. The best model was selected based on the Montecarlo permutation test. Statistical significance was set at p < 0.05.

Discussion
This study found an upward trend in colorectal cancer, colon cancer, and rectal cancer for both sex in Cali, Costa Rica, Goiânia, and Quito during the period assessed. These results are consistent with the increase in colorectal cancer incidence reported in other countries in economic transition, such as Argentina, Chile, China, Czech Republic, and Slovakia 4,12 . Although the incidence increased in the four regions of Latin America, it occurred differently between the regions, with Goiânia showing the highest rates, both in men and women (32.2/100,000 men and 28.6/100,000 women). On the other hand, Quito showed the lowest rates (13.9/100,000 men and 14.6/100,000 women). Results also showed differences in the magnitude of the rates by sex. While women in Cali and Quito exhibited higher incidence rates of colorectal cancer, colon cancer, and rectal cancer, in Goiânia and Costa Rica, the highest rates were in men. These findings corroborate with the study by Sierra & Forman 4 in 13 countries of Central and South America from 1985 to 2007, who found higher incidence rates in men than in women (male/female ratio between 1 and 1.8:1). The increasing rates seem to reflect pertinent questions of population aging, resulting from the demographic transition over the years in Latin American countries, besides lifestyle changes related to economic opening and monetary stabilization in recent decades. This process led to the expansion of the consumer market and increased exposure to risk factors for this neoplasm 13 . The literature shows that the main risk factors for colorectal cancer are lifestyle issues, especially inadequate diet (diets rich in red and processed meat) 14 and low consumption of fruits and vegetables 15 , physical inactivity 16 , obesity 17 , smoking 18 , and alcohol consumption 18 . Thus, differences in the prevalence of these risk factors may have affected the observed variations in the incidence curve slope for this neoplasm in different regions and between the sex.
In Brazil, Mielke et al. 19 19 . Monteiro et al. 20 found a time trend in the relative contribution of ultraprocessed foods in Brazil's metropolitan areas in 1986-1987, 1995-1996, and 2002-2003. In the last three decades, the authors found that consumption of unprocessed and minimally processed foods and processed culinary ingredients has been replaced steadily by consuming ultra-processed foods in both high-income and low-income groups 20  neoplasm are inter-sector approaches, mainly primary prevention such as strengthening food systems that simultaneously promote prosperity, equity, environmental sustainability, and health 23 . Latin America has witnessed experiences such as taxation of soda drinks in Mexico 24 , new labeling with nutritional traffic lights in Ecuador 25 , and the publication of food guides adopting foods classification based on their processing level in Brazil 26 and Uruguay 27 . However, most commitments assumed by public management were not translated into real measures 23 . Such reality is evident because the prevalent Brazilian dietary pattern includes high consumption of red and processed meats, oils and fats, and ultra-processed foods 20 .
Besides promoting the change of dietary habits as a public health program, primary prevention efforts should include promoting physical activity, which is a preventive factor for obesity and, therefore, a preventive factor for colorectal cancer 17,18,28 . According to Continuous Update Project (CUP) data from World Cancer Research Fund (WCRF) 29 , there is strong evidence that being physically active decreases colon cancer risk. Physical activity reduces body fatness and, therefore, has a beneficial effect on colorectal cancer risk, possibly reducing insulin resistance and inflammatory processes 17,18,29 . Other mechanisms by which physical activity may reduce colorectal cancer risk include stimulating digestion and reducing transit time through the intestine 16,29 .
Regional differences in patterns of increasing incidence of colorectal cancer can also be affected by the number of screenings for this neoplasm 30 . Screening programs for colorectal cancer are essential measures to assess the global burden of colorectal cancer 12 . Screening may lead to a shortterm increase in incidences of colorectal cancer, through higher detection of prevalent cases, besides reducing the long-term incidence of the disease by removing premalignant lesions 12 . Thus, differences in the implementation of screening programs for this neoplasm may explain the variation in colorectal cancer incidence in these regions, for example Costa Rica has no national guidelines recommending colorectal cancer screening 30 .
Whereas the Brazilian Ministry of Health has recommended, since 2002, that individuals aged over 50 years should have an annual fecal occult blood test. If the result is positive, the recommendations include a colonoscopy or rectosigmoidoscopy 31 . The opportunistic screening model is currently in operation in the majority of the country 32 . However, it is necessary to consider regional differences in Brazil when action planning, with a focus on cancer care, considering decentralizing these actions to ensure their effectiveness 33,34 .
According to the anatomical site, the analysis showed a significant increase of AAPC for rectal and anal cancer in Goiânia (5.5% for men and 4.6% for women), whereas, in the other regions, the increase was around 2.5%. These results suggest that the increased incidence of colorectal cancer in Goiânia correlates with the increasing rectal cancer incidence, since a remarkable rectal cancer increase was highlighted in men from 1991 to 2005 (6.4% to 12.7%), followed by a non-significant decrease (-2.8%) from 2005 to 2012. Among women, the increase was smaller (2.4%), but constant. Malignant neoplasms of the colon and rectum/anus have distinct etiologies 35 , and the differences in both the prevalence of risk factors and screening and diagnostic strategies could influence the slope in the incidence curve in different anatomic sites 35 .
The HPV infection is a significant risk factor for rectal and anal cancer development 36 . The Brazilian Ministry of Health, in partnership with the Moinho de Ventos Hospital used a cross-sectional study, called POP-Brasil, to assess the prevalence of HPV (types, 6, 11, 16, and 18) in 26 state capitals and the Federal District in 2016 37 . The study included 5,812 sexually active women and 1,774 sexually active men aged 16 to 25 years, and estimated HPV prevalence using genital and oral samples with DNA extraction, followed by genotyping based on PCR amplification and hybridization. The estimated overall HPV prevalence in Brazil was 54.6%, and HPV types 16 and 18 -with a high risk of epithelial cancer -were present in 38.4% of the population 37 . In Goiânia, the estimated HPV prevalence was 54.1%, with 35.1% presenting high-risk HPV 37 . Evidence in the literature suggests that countries with high HPV prevalence have a higher incidence of neoplasms related to this infection 38 . Another possible explanation for differences observed between the sexes is that the implementation of cervical cancer screening programs in Brazil 36 since 1998 may have contributed to the increase in the diagnosis of rectal cancer in women, as the female perineal area was frequently examined by health professionals 39 .
Cad. Saúde Pública 2021; 37(10):e00175720 The different screening and diagnostic strategies adopted in each region may also explain the observed variations according to anatomical site. Neoplasms of the colon and rectum have distinct characteristics and require different screening methods and diagnosing lesions 40 . For example, flexible sigmoidoscopy is limited to diagnosing rectal or distal colon cancer, since it does not allow visualizing most of the colon during the examination 41,42 . Thus, methods of fecal occult blood testing, digital rectal examination, and flexible sigmoidoscopy are targeted to diagnosing rectal cancer, whereas for diagnosis of colon cancer, the recommendations include colonoscopy 41,42 .
The first limitation of this study was the data from PBCRs, which have inherent secondary data problems. Nevertheless, the use of PBCR data that are considered high-quality by the IARC allowed generating information that adds knowledge to the field, raising hypotheses on the potential factors that could lead to increased incidence of colorectal cancer in Latin American countries. Notably, the inclusion in this study of Cali, Costa Rica, Goiânia, and Quito only implies that the findings are not necessarily representative of Latin America as a whole. However, the PBCRs in these regions are high quality according to the IARC criteria, lending credibility to the data accuracy. Although we had found discrepancies in the population of Goiânia, IBGE data corrected this information, by population reference.
Although the rectosigmoid junction miss-classification may limit disaggregated analysis by primary colon sites, the frequency of neoplasms at the rectosigmoid junction is small (2.9%) 43 . Thus, miss-classification in such a cancer site would poorly influence the estimates. Besides, colon and rectum tumors have distinct etiologies, that can be influenced by social, lifestyle (diet, physical inactivity, so on), political, and economic aspects 3,13,29 . For example, the practice of physical activity, that plays a protective role in colon cancer, but not in rectal cancer 29 . Furthermore, this limitation is minimized by the fact that we also presented data on colorectal cancer.
Meanwhile, this study displayed essential advantages. As far as we know, this study was the first to assess time trends in colorectal cancer according to sex and anatomical sites in different regions of Latin America. Our study also assessed the evolution in the incidence rates of colorectal cancer up to 2012, the most recent period in the data published by the IARC, considering the political, social, and economic changes in these countries throughout the period.
The study found increases in colorectal cancer, colon cancer, and rectal cancer in four Latin-America regions, especially rectal cancer in Goiânia. The findings reflect lifestyle, such as dietary changes, following 1980s economic opening in Costa Rica, and the early 1990s in Brazil, Colombia, and Ecuador, and the prevalence variations of colorectal cancer risk factors by sex and between the four studied regions. Finally, the different strategies adopted by regions for colorectal cancer diagnosis and screening seem to influence the observed variation between anatomical sites.
This study findings suggest the need for specific measures to control the risk factors associated with colorectal cancer, besides effective screening and control programs for these neoplasms in the four countries. These measures can contribute to colorectal cancer incidence control, especially among men and women in Goiânia.
Cad. Saúde Pública 2021; 37(10):e00175720 Contributors T. C. Carvalho contributed to the acquisition, analysis and interpretation of data and wrote the first draft of the manuscript. A. K. M. Borges and I. F. Silva made substantial contributions to the conception and design of the work, and guided all the analyses of the study. R. J. Koifman critically reviewed the study. All authors approved the final version of the manuscript.