Time trends of cancer incidence in Setif, Algeria, 1986–2010: an observational study

Background Incidence rates of various cancers are increasing in Arab countries and are expected to reach those of industrialized ones in few decades. This paper aimed to describe the incidence rates of most common cancers - and/or of those cancer preventable through modifiable behaviors - recorded in the province of Setif, Algeria from 1986 through 2010. Methods Cancer diagnoses for the 1986–2010 period were provided by the population-based Cancer Registry of Setif, disentangled by site, morphology, age (quinquennia), sex, and calendar period. The corresponding population was obtained from the Algerian Institute of Statistics. Age-standardized rates (world population) (ASR-WR) were computed by calendar period (five quinquennias from 1986–1990 to 2006–2010), while annual percent changes (APCs) were computed for the period 1996–2010. Results During the 2006–2010 period, ASR-WR for all cancer sites were 106.4/100,000 in men and 110.3 in women. The four leading cancers were: lung (18.0%); colon-rectum (9.6%); bladder (9.1%); and prostate (6.5%) in men; breast (36.4%); colon-rectum (8.5%); cervix uteri (6.0%); and thyroid (6.0%) in women. Between 1996–2010, overall cancer incidence increased statistically significantly (p < 0.05) in both men (APC = +2.5%) and women (APC = +3.7%). Statistically significant decreasing trends were observed for nasopharyngeal carcinoma (APC = -3.4%) in men, and for cervical (APC = -4.2%) and gallbladder (APC = -3.2%) cancers in women. Statistically significant increasing trends were observed for most common cancers both in men (lung:+1.8%, colon-rectum:+5.4%, prostate:+4.3%, liver:+8.9%, and bladder:+5.9%) and women (breast:+8.2%, colon-rectum:+4.5%, lung:+10.0%, liver:+5.4%, thyroid:+5.3%, and larynx:+13.8%). Conclusions International recommendations against cancer must be strongly promoted in Setif after taking into account epidemiological transition, lifestyle, and environmental changes.


Background
Cancer rates in northern Arab countries lying on the African side of the Mediterranean Sea are substantially lower than those reported in European Mediterranean countries such as Spain, France or Italy [1]. However, some increases in cancer incidence in northern African countries (e.g., Algeria, and Tunisia) have been recently observed for some cancer sites [2][3][4][5]. Although still one third or one half lower than those of industrialized countries, incidence rates are expected to rise and to reach similar levels in few decades. Furthermore, variations in overall rates among those countries were by far smaller than the differences between this group of countries and Europe [6]. A substantial proportion of these increasing cancer rates is potentially avoidable by primary and secondary prevention programs, while improvements in diagnosis and anticancer treatments would increase survival. In Algeria, cancer has become a national priority bringing to a 2014-20 national cancer plan under aegis of the President of the Republic.
The cancer registry of Setif (CR-S), Algeria, has been recording incidence data for the whole Wilaya (province) of Setif, the second in size after the Wilaya of Algiersthe capital city of Algeria. An accurate description of temporal trends by cancer site can contribute to a better understanding of cancer incidence patterns and their potential etiology and/or prevention. We took advantage of the availability of computerized incidence data recorded in CR-S over the 1986-2010 period to describe trends in incidence for the most common cancer sites and/or for the potentially preventable ones.

Methods
Data regarding all incident cancer cases diagnosed during the period 1986-2010 derived from the populationbased cancer registry of Setif, established in 1989 in collaboration with the International Agency for Research on Cancer (IARC). The Director of the Cancer Registry of Setif (Prof. Hamdi Cherif ) approved the use of the registry data for the purposes of the study. Nowadays, the CR-S covers a population of about 1.5 million inhabitants (as attested by 2008 national census), which is nearly 5% of the Algerian population. Collection of data is primarily active.
The structure of the resident population by sex, age, and calendar year was abstracted from the Algerian Institute of Statistics database [www.ons.dz].
Incidence data were grouped into five quinquennia of calendar years (1986-1990, 1991-1995, 1996-2000, 2001-2005, and 2006-2010). Age-standardized incidence rates per 100,000 (to the world standard population) [10] (ASR-WR) were calculated in both sexes by the five quinquennia described -or for each calendar year from 1996 to 2010 -according to selected cancers or groups of cancers. Incidence rates and their corresponding standard errors were calculated using SEER*Stat (Surveillance Research Program, National Cancer Institute SEER*Stat software, seer.cancer.gov/seerstat, version 8.1.2).
The computation of annual percent change (APC) [11,12] of incidence rates was restricted to the 1996-2010 period, to specifically quantify the recent impact of cancer incidence on the Setif population. APCs were calculated separately by age groups (15-44 years, 45-64 years, and 65+ years) and in the total population. APCs were estimated by fitting a linear regression line to the natural logarithm of annual incidence rates using calendar year as a regressor variable. This calculation assumes that the incidence rates changed at a constant rate over the entire calendar-year interval examined, and the validity of this assumption was checked by merely examining plotted curves. APCs were computed using the Joinpoint software and their statistical significance (p < 0.05) was calculated by means of a Student's t-distribution [12].
Throughout the study period (25 years), overall cancer incidence rates steadily increased in both sexes (Table 1). In men, rates (per 100,000) increased from 67.  Table 2). These positive APCs were consistent across all age groups   years, 45-64 years, and 65+ years) with statistically significant (p < 0.05) APCs in the three age groups in women and in the 65+ age group in men. In men, stomach cancer incidence rates decreased continuously during the whole period. Nasopharyngeal incidence rates increased up to 1996-2000 and declined thereafter. Non-Hodgkin's lymphomas rates increased up to 1996-2000 and leveled-off thereafter. Cancers of gallbladder and kidney, Hodgkin lymphomas, leukemias and ill-defined cancer showed somewhat stable trends during the 25 years examined. Liver cancer rates increased since 1996-2000 after a decrease. Conversely, increased trends were observed for cancer of the colon-rectum, larynx, lung, prostate, bladder and thyroid (since 1996-2000).
In women, stomach cancer rates declined up to 1996-2000 and leveled-off thereafter, whereas rates of invasive cervical cancer increased up to 1991-1995 and declined thereafter. Gallbladder rates increased up to 2001-2005 and declined thereafter while lung cancer incidence rates increased after 2001-2005. Cancers of nasopharynx, larynx, ovary, kidney, and bladder, Hodgkin's lymphomas, leukemias and ill-defined cancers showed somewhat stable trends during the 25 years examined. Liver cancer rates increased since 1996-2000 after a decrease. By contrast, cancer of the colon-rectum, breast, and thyroid, and non-Hodgkin's lymphomas increased throughout the whole period examined.
We investigated cancer incidence trends recorded by the Cancer Registry of the Wilaya of Setif, Algeria, across a 25 year period . To the best of our knowledge, this is the first report to investigate a large number of different types of cancer according to age for a quarter century in a northern African population. Decreasing trends were observed for nasopharyngeal carcinoma in men, and for cervical and gallbladder cancers in women, whereas increasing ones were observed mainly in leading cancers both in men (lung, colonrectum, prostate, liver, and bladder) and women (breast, colon-rectum, lung, liver, thyroid and larynx).
Leading cancers were similar to those reported by other North-African studies [2][3][4]6,13,14]. In contrast with the United States and other Western Countries, lung, colon-rectum, female breast, and prostate cancers showed in Setif increasing temporal trends [9]. The observed temporal trends give clues about large scale changes in exposure to risk factors, and to improvements in opportunistic early diagnosis.
Lung cancer incidence is largely determined by prevalence and duration of smoking. In Algeria, between 1970 and 1980, prevalence of smoking increased both for blond (34 times) and for black (+50%) tobacco. In addition between 1978 and 2005 prevalence of smoking increased from 7.7% to 28.6% [3,15,16]. Smoking is also strongly associated with the occurrence of bladder, kidney, and larynx carcinomas. The increasing trends in incidence of all these cancers in men can thus be ascribed to the increased prevalence of smoking over time. In women, the smoking effect is noticeable only for lung and larynx cancers and an increase of incidence of bladder and kidney cancers is likely expected. The sharp temporal variations observed in laryngeal cancer in women across age groups are compatible with random variations linked to small number of incident cases.
Nasopharyngeal carcinoma (NPC) is a rare malignancy in most regions of the world, but it is endemic in Algeria and more generally in North Africa. Major risk factors include Epstein-Barr virus infection, genetic susceptibility, consumption of salt-preserved fish or other preserved foods, a family history of NPC, and tobacco smoking [17]. In Setif, traditionally preserved food such as "harissa", pickled fruits and vegetables, dried and salted meat or fat, rancid butter, and domestic fumes have been described as NPC risk factors [16,[18][19][20]. By contrast XRCC1 and hOGG1 genes are unlikely to play a role in the susceptibility to NPC in North Africans [21]. We observed that NPC decreased in men but not in women. However, which determinants reflected this different pattern across sexes remains outside the scope of an observational study.
Changes in viral hepatitis prevalence rather than alcohol consumption seemed to explain the rise of liver cancer incidence because the other alcohol-related cancers (oral cavity, pharynx or esophagus cancers) were stable, with ASR-WRs below 1/100.000 (not shown). The prevalence of hepatitis B virus in blood donors was around 3.6% in 1989 [3]. To the best of our knowledge, no other information such as prevalence of hepatitis C virus is yet available.
Breast cancer, the most common female cancer diagnosed in Setif, dropped in all age groups. The reasons for this shift are unknown, but they may include increases in the prevalence of risk factors that modify endogenous levels of sex hormones such as early menarche, late menopause, late childbearing, having fewer children, and low prevalence of breastfeeding, and unhealthy diets, physical inactivity or obesity [2,22]. For instance, total fertility (average number of children per woman) declined in the last 30 years in Algeria (7.18 children per woman in the 1975-80 period to 2.72 in 2005-2010 period) [23]. Moreover, in two neighboring countries of Algeria, Morocco and Tunisia, obesity is significantly higher among women than among men (22.7% vs. 6.7% in Tunisia and 18% vs. 5.7% in Morocco) and prevalence of obesity among women has tripled over the past 20 years [24]. Half of all women are overweight or obese (BMI > 25) with 50.9% in Tunisia and 51.3% in Morocco [24]. BRCA1 and BRCA2 mutations have been identified in Algerian patients. The pathogenic mutations observed are mainly similar to those described in European counties except one found in a family of Kabylia, a northern neighboring region of Setif [25]. Noteworthy, endometrial and ovarian cancers, which share risk factors with breast cancer, did not display increases. In particular, ovarian cancer displayed a decrease in the 45-64 years age group. It is possible that opportunistic mammography anticipated the diagnoses of breast cancer in younger age groups.
Invasive cervical cancer (ICC) is the third most common female cancer in Setif, and a sharp decline in ICC incidence was observed mainly in the 15-44 age group. A potential shift of ICC diagnoses to uterus, unspecified (near 16% of all uterine cancers in 2006-2010) cannot explain this decline. The development of cervical cancer is causally associated to infection by high-risk Human Papilloma Viruses (HPV). In Algeria, the HPV16 in single-or multiple-type infection has been shown to be the predominant HPV type (61.4%) in ICC, followed by HPV18 (15.8%), and HPV45 (6.4%) [26]. The prevalence of HPV16 is similar to the prevalence observed in European or North-American populations [26]. In Setif, the prevalence of high-risk HPV was 75%. Moreover, in all age groups a drop of the squamous cell carcinoma (SCC) histologic subtype (89% of all ICC diagnoses in 1986-1990, vs. 53% in 2006-2010) is observed. This decrease can thus be ascribed to opportunistic early detection by cytological screening that is more effective in detecting SCC than adenocarcinomas [27]. A similar pattern was observed in the Center of Tunisia after the introduction of regional cytological screening program [4].
Thyroid cancer (TC) incidence increased in women in 15-44 and 45-64 age groups. In 2006-2010, the most common form of TC was the papillary sub-type and was related to most of the increased incidence (19% of all TC diagnoses in 1986-1990, vs. 49% in 2006-2010). This is thought to be largely due to increased detection of small papillary cancers associated to widespread use of ultrasound and fine needle biopsies [28].
Prostate cancer is the third most common cancer in men in Algeria. A number of risk factors detected in Western Algeria can explain the increased trend in prostate cancer incidence, such as high consumption of red meat, and dairy products, or low consumption of olive oil, cauliflower, and intake of fruits [29]. Presently prostate cancer incidence is closely linked to increased uptake of serum prostate-specific antigen (PSA) testing in western countries, but no data on PSA testing are yet available to verify this hypothesis in Algeria.
Reduction of stomach cancer incidence is generally attributed to diets rich in fruit and vegetables, improved preservation of foods, and reduced prevalence of Helicobacter pylori [22]. Some discrepant trends across age groups between gastric cancer and NPC that share a common risk factor (preserved foods) may suggest a decreasing exposure to Helicobacter pylori in the 15-44 age group.
Colon-rectum cancer increased in 45-64 and 65+ age groups and is the 2 nd most common cancer in both sexes. The adoption of western lifestyles and behaviors such as consumption of high-fat diets and physical inactivity could explain the increase of these cancers together with genetic predisposition [3]. It is possible that opportunistic early detection contributed to the increase of incidence in older age groups.
Gallbladder is the 5 th most common cancer in Setif women. Algerian women show the highest rates of gallbladder cancer as compared to the other Arab countries where this cancer is generally rare [2]. The temporal trends showed a favorable pattern in the 15-44 age group suggesting that differential risk factors are acting.
The selective increase of non-Hodgkin's lymphomas in the 15-44 age group in women is intriguing. Moreover, further studies are required to elucidate the increase of leukemias, particularly lymphoid leukemias, in the 15-44 age group.
Incidence data have potential limitations in accuracy and completeness that should be made clear. Firstly, cancer and comorbidities are extremely common in elderly people and the percentage of histologically confirmed cancers is generally lower among them than in younger age groups. Although under-diagnosis and underregistration seem to be unlikely, the complete identification of true cases in older age groups may be difficult. Secondly, the accuracy of DCOs depends on coding practices. The low number of DCO should have limited the possibility of differential practices influencing death certification procedures. Thirdly, no mortality/incidence ratios were available (as in all North-African cancer registries). This information would have given further confidence in the observed results. Fourthly, due to the relatively small population covered (1.5 million, i.e. 5% of Algeria's population), more accurate results from a bigger cohort are needed.
Among the strengths of the study, MV% reached good or acceptable levels in the studied cancer sites. Moreover MV% did not change significantly over time, thus, any estimated trend in incidence is unlikely to be the consequence of changes in the degree of ascertainment. Moreover, a small proportion of cancer cases were reported by DCOs. Finally, the availability of complete morphology data provided the opportunity to associate some trends to potential early detection initiatives.

Conclusions
Some main conclusions can be drawn from the trends of cancer incidence observed in the last 25 years in Setif. Besides ageing of the population, temporal variations in caseload indicate cohort effects due to differential exposure to risk factors and improved detection methodology. In the light of international recommendations against cancer [30] three leading priorities can be advocated. Firstly, in the light of the observed variation in incidence that may be linked to participation in opportunistic screenings, it is reasonable to promote organized populationbased screening programs aimed at reducing breast, colon-rectum, and cervical cancers. Secondly, primary prevention intervention at population level must be established to attempt a reduction in factors such as tobacco smoking and/or overweight. Thirdly, efforts should focus on the reduction of vaccine-preventable cancers. HPV vaccination for the prevention of cervical cancer should be introduced. The use of hepatitis B virus (HBV) vaccine for prevention of hepatocellular carcinoma should also be expanded. However, the application of these international recommendations against cancer in Setif must take into account epidemiological transition, lifestyle and environmental changes, poor health education, and difficult access to health care facilities. Thus, using evidence-based medicine, AROME project defined the "minimal requirements" guidelines (vs. standard of care guidelines) to form a basis for the development of practices and policies for diagnosis and treatment in limited-resource countries. The guidelines included cancers of prostate, bladder, lung, breast, head and neck, colon-rectum, and cervix [31].