Abstract
Background: Prostate cancer is the leading cancer in men in sub-Saharan Africa (SSA) regarding incidence and mortality. Published data from a few registries in SSA suggest that the rates are still rising, but there is little comprehensive information on the time trends of prostate cancer incidence.
Methods: We analyzed registry data on 13,170 incident prostate cancer cases in men aged 40 years or above, from 12 population-based cancer registries in 11 SSA countries, with at least a 10-year time span of comparable data.
Results: We observed an increase in cumulative risks (CR) and age-standardized incidence rates (ASR) over time in all registries (statistically significant in all but one). The highest values of CR were found in Seychelles and Harare (Zimbabwe). The highest annual increase in the ASRs was seen in Seychelles and Eastern Cape (South Africa), whereas the lowest was seen in Mauritius. We mainly found a steady increase in incidence with age and during successive periods.
Conclusions: This analysis reveals that prostate cancer incidence rates are rising in many populations in SSA—often very rapidly—which is in contrast to recent observations worldwide. We acknowledge that the reasons are multifactorial and largely remain unclear, but believe that they are primarily associated with improvements in health care systems, for example, a broader use of prostate-specific antigen testing.
Impact: This study is the first to compare population-level data on time trends of prostate cancer incidence between multiple countries of SSA, presenting the different rates of increase in 11 of them.
Introduction
With an estimated 1.3 million new cases and 359,000 associated deaths, prostate cancer was the second most common cancer in men worldwide and the fifth leading cause of cancer-related death in 2018 (1).
Although prostate cancer incidence rates have been on the rise worldwide during the last decades, it has been reported that the trends have started to stabilize or even to decline in many countries in recent years (2, 3). This observation has been described to be more pronounced in high-income countries and is associated with changes in the recommendations concerning prostate-specific antigen (PSA) screening of asymptomatic men, and a subsequent decrease of its usage. Although men of African descent have been shown to be disproportionally affected by prostate cancer, and that it is the leading cause of cancer (and cancer-related deaths) among sub-Saharan African men, there is little information on the time trends of prostate cancer incidence in the region (4, 5). In Kampala (Uganda) Wabinga and colleagues (6) reported increasing trends for the period of 1991–2010 with an average annual percentage change (AAPC) of 5.2% (although they may have stabilized in the period 2008–2012; ref. 3). An AAPC of 6.4% has been observed in Harare (Zimbabwe; 1991–2010; ref. 7) and there has been a 2.5-fold increase in Maputo (Mozambique) in the period of 1956–2017 (8).
More population-based evidence is needed, to get a better insight into the burden of the disease in this region, to examine changes in incidence rates, anticipate future problems for local health care systems and possibly even help to better understand this disease, which affects so many men worldwide, but for which the etiology is still relatively obscure (9).
In this analysis, we describe and compare time trends in prostate cancer incidence rates for 12 populations from eleven sub-Saharan African countries, by examining data from population-based cancer registries of the African Cancer Registry Network (AFCRN).
Materials and Methods
Data inputs
The African Cancer Registry Network (AFCRN; https://afcrn.org/) is the regional hub for cancer registration in sub-Saharan Africa (SSA) of the International Agency for Research on Cancer's (IARC) Global Initiative for Cancer Registry Development in Low- and Middle-Income Countries (GICR). The project's aim is to improve cancer surveillance in the region and thereby provide a solid basis for the planning and evaluation of local cancer control programs.
At the time of our study, 32 population-based cancer registries were members of the network. According to the network's membership criteria, a population-based cancer registry must achieve at least 70% coverage of their target population within 3 years of admission.
All registries collect data on incident cancer cases within their catchment population, usually by active methods. They enter the data electronically to the AFCRN's database, by using IARC's CanReg-5 software (10). Cancer site and histology are coded according to the International Classification of Diseases for Oncology (ICD-O) following the AFCRN's Standard Procedure Manual (11, 12).
We included 9 registries with at least 10 years of continuous data, during which period there appeared to have been no obvious changes in completeness of registration (13).
For Eastern Africa: Nairobi (Kenya, 2003–2014), Blantyre (Malawi, 1995–2009), Mauritius (2001–2016), Seychelles (2005–2018), Kampala (Uganda, 1991–2015), and Harare (Zimbabwe, 1990–2016).
For Southern Africa: Eastern Cape (South Africa, 1998–2015).
For Western Africa: Bamako (Mali, 1999–2018) and Ibadan (Nigeria, 1996–2010).
We also reviewed changes in incidence from 3 registries with recently published incidence rates, for which comparable data from older publications were available: Abidjan (Côte d'Ivoire, 1995–1998, 2012–2015); Bulawayo (Zimbabwe, 1963–1972, 2011–2015) and Maputo (Mozambique, 1956–1960, 2015–2017).
All other AFCRN registries did not meet our inclusion criteria.
The registries of Mauritius and the Seychelles cover national populations, that of Eastern Cape in South Africa a rural population, whereas the other registries all cover urban populations.
From January to March 2020, we extracted completely anonymized data on incident cases of prostate cancer (International Classification of Diseases, ICD-10 C61) from the AFCRN's database.
Annual estimates of the population-at-risk, by 5-year age group, for each registry area and time period were prepared, using national census data, and assuming a logarithmic growth (within age groups) between the censuses. For the period after the most recent census, we assumed a linear growth adding the average annual increase in numbers of cases (by sex and age group) for the years between the preceding censuses.
For Bulawayo (Zimbabwe), we included data only on the black (African) population, because historic data (for the 1960s) were available only for this racial group.
For a few registries, there were years with a known reduction of registration activities due to political or socio-economic circumstances. Those were the years 2015 for Bamako (Mali) 2007–2009 for Harare (Zimbabwe), and 1999 for Ibadan (Nigeria). We excluded these years from our analyses.
Data analysis
We used RStudio Version 1.2.5033 for data analyses (14). We estimated the proportions of prostate cancer cases registered only on the basis of information on a death certificate (death certificate only, DCO), on the basis of a histological or cytological examination (morphologically verified, MV), or on the basis of clinical examination/imaging with and without the additional evidence of an elevated PSA. These proportions can be used to indicate the data quality of a population-based cancer registry (15).
We included all incident prostate cancer cases (regardless of their basis of diagnosis, i.e., MV, DCO, or clinical) and grouped them by year and registry in 5-year age groups. Carcinoma of the prostate is very rare before age 40. Although there were around 0.7% of such cases in the datasets, they were omitted from analyses, because of concerns that they represented possible errors in coding (of site, or age). Because the population at risk data were often truncated at age group “75+,” we also truncated our case grouping at age group “75+.” We redistributed cases with missing age into the different age groups according to the distribution of the known cases.
We calculated age-specific and age-standardized incidence rates (ASR), together with cumulative risk (CR; 0–74), per registry and year. We used the world standard population to obtain the ASRs (16). We calculated the same parameters for three approximately even time periods per registry.
We examined temporal trends in annual ASR by using the Joinpoint Regression Software (17, 18). The maximum possible joinpoints in the model was set at three. For registries without continuous registration activities, we set the maximum possible joinpoints at zero. We report the AAPC over the whole time of examination, together with the 95% confidence intervals for each registry.
We present the trends of the annual ASRs graphically as centered 3-year moving averages.
We performed an exploratory Age-Period-Cohort analysis for Kampala (Uganda) and Harare (Zimbabwe), each of which contributed to four successive volumes of Cancer Incidence in Five Continents (CI5: http://ci5.iarc.fr/Default.aspx). For this analysis we used the “Rcan”-package, Version 1.3.81, for “R” from Laversanne and colleagues (19). Because there were very few cases in the age groups “40–44” and “45–49,” resulting in very unstable rates, these age groups were omitted from the analyses.
Ethics
The AFCRN research committee approved this study (July 2019), as well as the respective registries. We conducted the study in accordance to the Declaration of Helsinki. The study used routinely collected, anonymized data, therefore no special ethical approval was needed.
Results
In Fig. 1, the registries (white circles) and their corresponding sub-Saharan African countries are highlighted. The table shows the national population (males) in 2010 and our estimates of the population of males in the areas covered by the registries, as a total, and a percentage of the national population (20).
Map of sub-Sahara Africa (Wikimedia Commons, 2019) and population of included registries. Included countries are in black; white circles indicate locations of registries; population estimates for the year 2010 are shown.
The coverage of the national population for urban, non-national registries ranged from 1.7% in Ibadan (Nigeria) to 18.8% in Abidjan (Côte d'Ivoire). For the single rural registry of Eastern Cape (South Africa) it was around 2 %. Mauritius and Seychelles both have national registries.
Table 1 shows the registries (by sub-Saharan African region and country), as well as the time periods included in the analysis and corresponding numbers of cases. A total of 13,170 cases ages 40 or more were registered during a total of 24,300,079 person-years at risk [excluding the historic data from Maputo (Mozambique), because person-years at risk were not available].
Description of included registries.
The proportion of microscopically verified (MV%) prostate cancer cases ranged from 94.7% in Mauritius to 46.2% in Blantyre (Malawi).
For Harare, Zimbabwe and the recent period of Maputo (Mozambique), the proportion of DCO cases was around 15%, whereas in all other cases the proportion was below 8%.
Mean age at diagnosis ranged from 66.5 years in Blantyre (Malawi) to 72.0 years in Harare [Zimbabwe; although in Bulawayo (Zimbabwe), the mean age increased from 63.9 in the early period to 74.1 in the more recent period]. Excluding the few cases recorded as being <40 years of age, the mean age at diagnosis for all 12 populations was 70.1 years.
Table 2 shows age standardized incidence rates, as well as the number of cases, in three time periods for each registry with continuous data, as well as the AAPC of the ASRs over the whole period studied. For the registries with data from two (discontinuous) time periods, we show the number of cases and ASRs in each, and the estimated AAPC between them. CRs and ASRs varied widely between and within sub-Saharan African regions, but had increased over time in all registries (not statistically significant in Ibadan, Nigeria).
Cumulative risks, age-standardized incidence rates by registry and period of time, together with the average annual percentage change.
We observed the highest values of CR in Seychelles and Harare (Zimbabwe), where 1 in 9 and 1 in 10 men would develop prostate cancer by the age of 74 under exclusion of competing risks of death. The lowest CRs in the most recent periods are seen in Ibadan (Nigeria) and Mauritius.
The annual increase in the ASRs (AAPC) was highest in Seychelles and Eastern Cape (South Africa), at around 10% per year, and the lowest in Bulawayo (Zimbabwe) and Mauritius, as well as in Maputo (Mozambique).
Figure 2 shows annual ASRs in each registry, as well as a line representing three-year moving averages. We observe increasing incidence rates in all registries, with variation in the magnitude of the slopes between the registries.
Age-standardized incidence rate by registry and year of registration for registries with continuous data: Trend lines show 3-year moving averages, dots represent observed values.
The joinpoint analysis revealed that the trends are well explained for all registries by the simplest model without any joinpoints (see AAPCs in Table 2 and Supplementary Fig. S1A and S1B).
Figure 3 depicts the age-specific incidence rates by registry for the same periods as in Table 2. Because case numbers in age groups below 50 years were very low, the corresponding rates are omitted from the graphs. Bulawayo (Zimbabwe) and Maputo (Mozambique) are also excluded because of the very sparse data in the older period.
Prostate cancer age-specific incidence rates by time period and registry.
We observe a steady increase in incidence with age, and the highest incidence rates in the oldest age group (with a few exceptions). For, Blantyre (Malawi), Bamako (Mali) and for the earliest periods in Ibadan (Nigeria), and Seychelles there are fluctuations in incidence. As well as being the result of small numbers [especially in younger age groups, and smaller registries (Blantyre (Malawi), Seychelles)], this relates to digit preference in given age of older men, with an excess of cases ages exactly 50, 60, and 70 (and corresponding higher rates in the age groups containing these digits; see Supplementary Fig. S2)
For all registries, these trends show an increase in age-specific incidence rates during these successive time periods, with, in general, a rather greater increase in older age groups.
Figure 4 shows trends of age-specific incidence rates by birth cohort and period of diagnosis in Harare (Fig. 4A) and Kampala (Fig. 4B). There is an increase in the age-specific incidence rates in both according to period of diagnosis and period of birth.
Prostate cancer age-specific incidence rates by birth cohort and period of diagnosis in Harare (Zimbabwe; A) and Kampala (Uganda; B).
In Harare, the rate of increase is greater in the older age groups (as seen also in Fig. 3). In Kampala, there is more fluctuation in successive cohorts and the picture is not as clear as for Harare. For the most recent cohort of birth year and diagnosis (2011–2015), there seems to be a small decline in all age groups (except for the youngest).
Discussion
Prostate cancer incidence rates have been increasing in all 12 sub-Saharan African populations during the periods of observation [although the estimated annual 2.5% increase in Ibadan (Nigeria) was non-significant]. CR, age-standardized incidence rates, and the annual average percentage change varied up to 7-fold between the populations. The highest CRs and ASRs in recent periods were observed in the Seychelles and Harare (Zimbabwe). The lowest values were reported from Ibadan (Nigeria) and Mauritius. The Seychelles (2005–2018) and the rural area of Eastern Cape province in South Africa (1998–2017) have seen the steepest increases in the annual ASRs with an average annual increase of around 10%.
In a recent international study of incidence rates in 44 countries (and mortality in 76), prostate cancer rates were found to have stabilized in most, and decreased in a few, since 2008–2012 (3). This is in contrast with our findings in SSA, where incidence has been rising, with decreases seen only in the most recent observation years of Ibadan (Nigeria; Fig. 2) and Kampala (Uganda; Figs. 2 and 4). The only population in common to the two studies was Kampala, where Culp and colleagues (3) observed stabilizing rates in the years 2008–2012 (non-significant AAPC), but a significant single trend with an annual percentage change of 1.8% during the period 1993–2012 (close to the single trend we found for the years 1991–2015).
Studies in migrants have suggested that environmental factors may affect prostate cancer risk. For example, in Japanese migrants to the United States and in particular in their descendants, an increase in their originally low risk of prostate cancer has been observed, leading to research on environmental or exogenous (lifestyle) factors, thought to be able to influence the prostate cancer risk (21). These factors comprise among others metabolic syndrome, obesity, body size, and dietary factors. However, apart from obesity and body size, the evidence is still poor and controversial (22–24).
Furthermore, the risks associated with these putative risk factors are not large, and it is unlikely that there has been a change in their prevalence large enough to account for such extensive increases in incidence. Rather, we believe that our results most likely reflect changes in the health care systems, linked to increasing socioeconomic development. Internationally, incidence of prostate cancer is higher in countries with higher socioeconomic development (25). In our study, the Seychelles has the highest human development index (HDI), followed by Mauritius (26). Although in the Seychelles, we observe correspondingly high ASRs of prostate cancer in the most recent period (97.5 per 100,000 men-years), Mauritius does not fulfil this criteria (15.5 per 100,000 men-years) and has one of the lowest ASRs of all countries under observation. We suggest that this might be due to a population of predominantly non-African ancestry, resulting in ASRs comparable with Asian countries with similar HDI like Malaysia or Thailand (1, 26). The biggest influence on reported incidence of prostate cancer has been the introduction of early detection programs through PSA testing in asymptomatic men (27, 28). However, in SSA, there has been no widespread general PSA screening activity. This is consistent with our observations that there have been no abrupt increases and successive declines, as for example, seen in the US, Canada, Australia or Sweden in the PSA screening era (2, 27, 29). Nevertheless, an ad hoc survey of AFCRN populations has confirmed that the PSA test is available in laboratories throughout Africa, and is widely used for diagnostic purposes, although there is no information on the trends in numbers of tests performed over time. It seems likely that at least some of the increase in incidence represents better detection (and diagnosis) of prostate cancers in middle-aged and elderly men with urinary symptoms.
In addition, there might also be an increase in the availability and consequently in the performance of trans-urethral resections of the prostate to treat urinary retention suspected to be caused by benign prostatic hyperplasia. This could lead to more incidental carcinomas and rising incidence rates, as has been described by Potosky and colleagues (30) for the United States in the pre-PSA screening era, during the years 1973 to 1986.
Limitations
The value of population-based cancer incidence data mainly relies on the accuracy of two parameters: First, the success of the registration activities, especially the completeness (and accuracy) of ascertainment of cancers in the targeted population. And second, the quality of the related population censuses and the derived population estimates.
This leads to the limitations of our descriptive study. Functioning cancer surveillance needs a relatively stable political and socio-economic environment. Guaranteeing high levels of completeness and constancy in registration activities over a longer period of time are accordingly challenging in low-resource settings. For example, for Harare (Zimbabwe), there have been registration problems reported in the 2007–2009 period (7). We initially allowed for this by only including 12 of the 32 AFCRN registries, with the presumably most consistent data on prostate cancer incident cases reflected in relatively constant annual case registrations, as well as the coherence of indicators like MV% and DCO%. Although the rate of MV% varies widely from Blantyre (Malawi) with 46.2% (urban) to Mauritius (national) with 94.7%, this does not necessarily reflect any incompleteness—indeed, inclusion of clinically diagnosed (and DCO) cases is a means to maximize completeness (13), rather it reflects different access to, and use of, diagnostic services such as pathology and imaging (CT scans and MRI) in the registry areas.
We excluded years with a clear reduction in registration activities in the corresponding registries from our analyses. Likewise, we had to exclude recent data from Ibadan (Nigeria); because the most recent population census was performed in the year 2006 there has been some discussion about its accuracy, and post-censal estimates are even less secure.
Conclusions
This registry data analysis presents trends in overall and age-specific prostate cancer incidence rates in 12 populations of SSA. Overall, we observe rising trends, and believe that, although the reasons are multifactorial, they are primarily due to improvements in the healthcare system with, among others, a broader use of PSA testing. Taking account of the fact that cancer is a growing health problem in SSA and prostate cancer is the top cancer in men, in both incidence and mortality, more studies on the patterns of diagnosis, including the prevalence of PSA testing for diagnostic or early detection purposes, on genomic and environmental factors, as well as the maintenance and improvement of population-based cancer registration, would help to better understand the reasons behind this observation and might eventually enlighten some important questions on the etiology of this disease.
Authors' Disclosures
No disclosures were reported.
Disclaimer
The funders/sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the article; or decision to submit the article for publication.
Authors' Contributions
T.P. Seraphin: Conceptualization, data curation, software, formal analysis, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. W.Y. Joko-Fru: Conceptualization, data curation, software, formal analysis, investigation, writing–review and editing. B. Kamaté: Resources, investigation, writing–review and editing. E. Chokunonga: Resources, investigation, writing–review and editing. H. Wabinga: Resources, investigation, writing–review and editing. N.I.M. Somdyala: Resources, investigation, writing–review and editing. S.S. Manraj: Resources, investigation, writing–review and editing. O.J. Ogunbiyi: Resources, investigation, writing–review and editing. C.P. Dzamalala: Resources, investigation, writing–review and editing. A. Finesse: Resources, investigation, writing–review and editing. A. Korir: Resources, investigation, writing–review and editing. G. N'Da: Resources, investigation, writing–review and editing. C. Lorenzoni: Resources, investigation, writing–review and editing. B. Liu: Data curation, project administration. E.J. Kantelhardt: Formal analysis, supervision, funding acquisition, validation, writing–original draft, project administration, writing–review and editing. D.M. Parkin: Conceptualization, resources, data curation, formal analysis, supervision, validation, methodology, writing–original draft, project administration, writing–review and editing.
Acknowledgments
The authors gratefully acknowledge the great work of the staff of all the contributing registries of the African Cancer Registry Network. T.P. Seraphin was recipient of a 7-month Halle-Oxford exchange fellowship grant within the EU/ESF-funded research, International research network biology of disease and molecular medicine ZS/2016/08/80642 from Martin-Luther-University Halle-Wittenberg. International Agency for Research on Cancer and American Cancer Society provided financial support for the extra data collection activities. The Commonwealth Scholarship, funded by the UK Government, is funding W.Y. Joko-Fru's PhD study at the University of Oxford.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Footnotes
Note: Supplementary data for this article are available at Cancer Epidemiology, Biomarkers & Prevention Online (http://cebp.aacrjournals.org/).
Cancer Epidemiol Biomarkers Prev 2021;30:158–65
- Received July 1, 2020.
- Revision received September 1, 2020.
- Accepted October 1, 2020.
- Published first October 8, 2020.
- ©2020 American Association for Cancer Research.
References
Volume 30, Issue 1
