Trends in the incidence of colorectal cancer in sub‐Saharan Africa: A population‐based registry study

In sub‐Saharan Africa, colorectal cancer (CRC) has historically been considered a rare disease, although some previous studies have suggested that the incidence is increasing. We examine time trends in the incidence of CRC using data from 12 population‐based cancer registries in 11 countries of sub‐Saharan Africa that were able to provide time series data for periods of 12 or more years, or with earlier data with which recent rates may be compared. Age‐standardized incidence rates were highest in the higher‐income countries, and were increasing in all of the populations studied, and these increases were statistically significant in all but three. Current evidence has suggested a link between the increased adoption of western lifestyle habits with colorectal cancer, and along with increasing urbanization of African populations, there is an increase in body weight, as well as evidence of increasing consumption of meat, sugars, and alcohol.

and along with increasing urbanization of African populations, there is an increase in body weight, as well as evidence of increasing consumption of meat, sugars, and alcohol.

K E Y W O R D S
cancer registry, colorectal cancer, incidence, sub-Saharan Africa, trends What's new?
The highest rates of colorectal cancer are in highly developed countries, but rates are on the rise in low-and middle-income countries.Here, the authors analyze trends in CRC incidence in sub-Saharan Africa.Using data from 12 population-based registries in 11 countries covering a period of at least 12 years, they found a steady rise in CRC across all regions, possibly due to wider adoption of a western lifestyle, including consumption of more meat, sugars, and alcohol.This highlights the need for prevention, early diagnosis, and management strategies in these regions.

| INTRODUCTION
Worldwide, colorectal cancer (CRC) is the third most commonly diagnosed cancer with an estimated 1.9 million new cases in 2020, and the second most common cause of death from cancer, with an estimated 935,000 deaths. 1 The distribution of CRC burden varies widely, with more than two-thirds of all cases and about 60% of all deaths occurring in countries with a high or very high human development index (HDI).However, CRC incidence and mortality rates are rising rapidly in many low-income and middle-income countries, while in high-income countries rates have stabilized or are even decreasing, although they remain among the highest in the world. 2 Several lifestyle factors have been linked to an increase in the risk of CRC, including overweight and obesity, alcohol drinking, and consumption of red and processed meat, 3 a diet low in fiber, 4 tobacco smoking, 5 and lack of physical activity. 3In sub-Saharan Africa (SSA), the incidence of CRC is relatively low, 6 and although it is likely that adverse changes in these lifestyle factors would result in rising rates, there are in fact relatively few studies of trends in incidence (or mortality) from populations in SSA.Increasing incidence rates have been documented in South Africa, 7 Kampala (Uganda) 8 Harare (Zimbabwe), 9 and, in males, in Bulawayo (Zimbabwe). 10e African Cancer Registry Network (AFCRN) through collaborations with its member population-based cancer registries in Africa has contributed to improving cancer registration on the continent and through its activities has generated data that can be used to estimate the burden of various cancers in SSA with implications for cancer control in the region. 6In this present study, we investigate trends of incidence of CRC in 12 cancer registries in three sub-Saharan African regions, for periods of between 14 and 60 years, to provide important information for the development of cancer prevention and control strategies, and as a benchmark for monitoring of the effectiveness of such programs.We also examine trends in the prevalence of the known risk factors for large bowel cancer, to assess whether the adoption of western lifestyle practices may be responsible for changes in the incidence of CRC in sub-Saharan Africa.

| Data sources
Completely anonymised data were obtained on incident cases of CRC (International Classification of Diseases, ICD-10 C18-C20) from the database of the AFCRN (http://afcrn.org).This is a network of sub-Saharan African population-based registries that facilitates homogenization of registration activities, collaboration, advocacy, and research. 11Member registries of AFCRN must have attained a minimum of 70% coverage of their target population.For the current study, we used data from 9 with at least 10 years of continuous data on incidence of CRC, and three for which data on CRC incidence was available for a period in the past (Figure 1).Population at-risk data for each registry area were obtained from population censuses, from their respective National Statistics Office reports.To obtain the population-at-risk for each year, inter-censal interpolations were made by sex, and within 5-year age groups assuming a constant logarithmic increase.Post-censal projections were made assuming a linear rate of increase at the rates observed following the preceding census.

| Statistical analyses
For each registry, the proportion of cases registered based only on a death certificate-death certificate only (DCO)-and of morphologically verified (MV) cases were estimated.These proportions are used as indicators of the data quality for each registry. 12 estimated age-specific (5-year age groups) and crude incidence rates per 100,000 persons at risk by registry for each year and for the whole period available.Age standardization was performed by the direct method, using the World Standard population. 13r the continuous time series, temporal trends were examined by fitting a regression line to the log-transformed age-standardized incidence rates.From this, we calculated the average annual percentage change (AAPC) as the slope of the regression line, together with its 95% confidence interval. 14e-standardized incidence was calculated for the historic data of Abidjan (1995-1997), 15 Maputo (at the time called Laurenco Marques) (1956-1960), 16,17 andBulawayo (1963-1972). 18,19The ratio of the ASRs in the recent and historic datasets was calculated, with their 95% confidence intervals. 14

| Time trends in age-standardized incidence rates
Table 1 shows, for each registry area, the period for which continuous data was available, the total number of CRC cases, the median age of the cases, the MV% and DCO%, and the age-standardized incidence rate (ASR) for the whole period.The change in incidence over the period is shown as the AAPC in the ASR, with 95% confidence limits.
Mauritius and Seychelles, both island countries with national coverage, have over 90% of morphology-verified cases, while in the Gambia, only one-third of the cases were morphology-verified. 12% of cases in Harare (Zimbabwe) had been registered from information on a DCO.The median age of the cases was around 55 for most registry series, while in Mauritius and Seychelles, the cases were older (median 64 and 65, respectively).
The average ASR ranged from 1.2 per 100,000 in Brazzaville (Republic of the Congo) to 22.0 per 100,000 in Seychelles. Figure 2 shows the trends in the ASR by year for each registry area, with the best-fitting line for linear trend, for both sexes combined.Preliminary analyses revealed no significant differences in trends by sex, therefore ASRs were calculated for both males and females together.
We examined the trends according to age (less than 55 or 55 years or older at diagnosis).In Mauritius and Seychelles, the increase in incidence was greater under the age of 55, although the differences were non-significant.In the other seven registries with continuous time series, the increases in incidence were more marked in the older age T A B L E 1 Colorectal cancer cases, age-standardized rates (ASR) and average annual percentage change (AAPC) in period.F I G U R E 2 Age-standardized incidence by registry and year of registration registries with continuous data, with best fitting regression line.
group; indeed, in five of them, the changes in ASR in the younger age group (under age 55) were not statistically significant.
Table 2 compares, for three cancer registries, the recent ASR of CRC for males and females combined, with data from the same registry at an earlier period, together with the rate ratio and its 95% confidence interval.The 46% increase in incidence in Maputo (Mozambique) over a period of almost 60 years is not statistically significant, partly because of the uncertainty of the estimate for 1956-60, based on only 6 cases.

| DISCUSSION
Changes in the risk or incidence of cancer in the population can only inferred data coming from cancer registries that are populationbased. 20Changes in numbers of cases, or the relative frequency of a cancer in hospital or laboratory series, may result from patterns of referral or availability of diagnostic services, and, for relative frequencies, trends in the numbers of other cases being included.The registries in the current study are all members of AFCRN and, as such, must have demonstrated relatively complete case ascertainment for the respective populations (a minimum of 70% 12 ).The data from several have been published in the "Cancer Incidence in Five Continents" series of IARC, 21 and data from several of them, for rather earlier time periods, have been reproduced in reviews of time trends of CRC in Africa. 22,23ere are currently 33 population-based member registries within the network, and almost all the information on incidence of CRC in SSA derives from this source, 6 although, at the time of this study, only 9 had been established sufficiently long-term to provide time series data for 12 or more years.
Temporal trends in age-standardized incident rates of CRC show consistent increases in incidence across all catchment areas during the study period.These findings are consistent with reviews suggesting that incidence of CRC in this region has been on the rise.This increase is thought to be likely due to the increased adoption of western lifestyle habits, as these reviews have suggested.
Lifestyles in Africa are undoubtedly changing with increasing urbanization of the population.In 1990, 27% of the population of sub-Saharan Africa lived in urban areas, and this had risen to 40% by 2018 and is projected to be 58% by 2050. 24Urban populations have diets that are lower in staple starches, fiber, and plant protein and higher in meat and processed animal products and in energy-dense  snacks, beverages, and alcohol.[27] Trends in BMI for sub-Saharan African populations are available from the Non-Communicable Diseases Risk Factor Collaboration group (NCD-RisC). 28The NCD-RisC group used data from all available population-level datasets to estimate trends in BMI over this period.
Figure 3 shows mean body mass index (BMI) in men and women in 1980 and 2014 in the 11 countries included in this study.There are rapid increases in BMI in all of these nations and with higher mean BMI in countries with higher HDI such as Seychelles, South Africa, and Mauritius.
Guthold et al. 29  The results of this study have important public health implications.
Although by global standards, the incidence of CRC remains relatively low, 1 the observed increase in incidence of CRC across all catchment areas indicates that CRC is becoming a health issue of concern for sub-Saharan African countries.Studies have found that late diagnosis of CRC leads to poorer health outcomes and higher rates of CRCassociated mortality. 30Effective treatment for early stage (I and II) CRC is probably the most cost-effective approach to control, 32 although most patients present at later stages of the disease, and even those in early stages may fail to receive guideline-concordant therapy in African settings. 33There have been efforts to test screening interventions in SSA settings, in an effort to achieve earlier diagnosis, 34,35 The time series datasets included incident cases of CRC from the AFRCN database for 9 registries from 9 countries: Brazzaville (Republic of the Congo) (1999-2020), The Gambia (1986-2016), Nairobi (Kenya) (2003-2016), Mauritius (2001-2019), Ibadan (Nigeria) (1995-2015), Seychelles (2008-2020), Kampala (Uganda) (1990-2017), Harare (Zimbabwe) (1990-2017), and Eastern Cape (South Africa) (1998-2017) (Figure 1).The population-based registries of The Gambia, Mauritius, and Seychelles cover the national territory, while all the others cover urban areas, except for the Eastern Cape registry of South Africa, which covers a rural population.The three registries with historical data available were Abidjan (Cote d'Ivoire), Maputo (Mozambique), and Bulawayo (Zimbabwe) (Figure 1); data were abstracted from the AFCRN database for the periods 2012-2016, 2015-2017, and 2013-2017 respectively.

F I G U R E 1
Map of sub-Sahara Africa and population of included registries.Included countries are in black; light circles indicate locations of registries; population estimates for the year 2020 are shown.

F I U R E 3
Mean body mass index (BMI) (kg/m 2 ) in males and females in the 11 counties studied, in 1980 and 2014 (Source: Reference[28]).
but although feasible, the low yield of cancers (and precancerous adenomas) and the non-availability of follow-up of positive screening tests by colonoscopy means that this is currently not a realistic approach.Public health interventions to mitigate adverse lifestyle practices 36 are also worthwhile as a means of reducing the risk of the disease.AUTHOR CONTRIBUTIONS Conceptualization: DMP.Data curation: LM, DMP.Formal Analysis: LM DMP.Funding acquisition: MP.Investigation: LM, PB, TC, EC, AF, AK, BoL, CFL, SSM, GN, AWSO, OO, NIMS, DMP.Methodology: LM, DMP.Project administration: BiL.Resources: PB, TC, EC, AF, AK, BoL, CFL, SSM, GN, AWSO, OO, NIMS.Supervision: DMP, BiL.Validation: LM, DMP.Visualization: LM, DMP, BiL.Writing-original draft: LM, DMP.Writing-review and editing: All.The work reported in the paper has been performed by the authors, unless clearly specified in the text.ORCID Nontuthuzelo I. M. Somdyala https://orcid.org/0000-0001-6564-8754Donald M. Parkin https://orcid.org/0000-0002-3229-1784 T A B L E 2 Colorectal cancer cases and age standardize incidence rates (ASR) from two separate time periods.
31milial adenomatous polyposis, can increase the risk of CRC; it is estimated that approximately 30% of all cases are an inherited form of the disease,31and most such cases occur at young ages.This may explain why there are only small increases in incidence in younger age groups.
by country, with the most marked changes in higher income countries, such as South Africa and Mauritius.One might have expected changes in incidence due to changing lifestyles to have affected younger age groups (more recent generations), while we observed that trends were more marked in older age groups.A family history or certain genetic conditions, such as Lynch syndrome and