Colorectal cancer (CRC) is the third most common cancer in men and the second most common cancer in women worldwide [1,2]. The high CRC incidence and magnitude make it a real public health concern [3]. Therefore, it is essential to determine its risk factors as a basis for evidence-based prevention strategies. The risk factors of CRC are complex and involve genetic and environmental factors [4,5]. Indeed, 30% to 70% of CRCs are considered to be because of environmental factors such as lifestyle and eating habits [6-8].

According to the recent World Cancer Research Fund (WCRF)/American Institute for Cancer Research (AICR) 2018 report, high red and processed meat intake, high body fatness, high consumption of alcoholic drinks, and low levels of physical activity have been associated with increased CRC risk, whereas diets high in antioxidants, in particular fruits and vegetables, or fibers have been associated with reduced risk [9]. A high body mass index is associated with an increased risk of occurrence of large adenomas at the level of the colon and rectum [10].

CRC incidence is indeed particularly high in Western countries with high meat consumption, such as Australia and the United States [11], whereas lower in Mediterranean countries [1] which could potentially be attributed to the quality of the Mediterranean diet, including high consumption of fruits, vegetables, and olive oil [12].

Regarding genetic factors, epidemiological studies conclude that the risk of developing CRC increases with increasing first-degree family history [13]. Several recent studies attempt to show a potential interaction between genetic predisposition and dietary factors [14]. One of these interactions could involve the acetylator status [15]. Indeed, rapid acetylating subjects (rapid phenotype of N acetyl transferase [NAT2]) are more likely to develop CRC, particularly when consuming overgrilled meat, through greater activation of heterocyclic amines [15]. Another example demonstrating the importance of considering genetic status is the analysis of certain mutations that occur in the early stages of CRC such as Kirsten rat sarcoma (KRAS) and proto-oncogene B-Raf (BRAF) mutations. These mutations were associated with certain foods, nutrients, and micronutrients, and it has been suggested that the risk of KRAS and BRAF mutations may depend on dietary and lifestyle factors [13,16].

The majority of studies investigating CRC risk factors have been conducted in countries with high CRC prevalence [11]. Results from these high-prevalence studies are sometimes controversial and not always applicable to low-prevalence countries such as Morocco, which have started feeling the effects of the triple demographic, epidemiological, and nutritional transition [17-19].

According to the cancer incidence data from Globocan 2018, CRC ranks as the third most common cancer in Morocco after lung and prostate cancer in men and after breast and cervical cancer in women (8.7% and 7% of all types of cancer, respectively). Furthermore, this number of new cancer cases seems to increase each year [20,21]. According to the Cancer Registry of the Greater Casablanca Region (2005 to 2012), age-standardized incidence rates raised from 3.8 to 8.4 and from 2.6 to 7.4 per 100,000 in Moroccan men and women, respectively [20,21]. At the same time, cancer risk factors and behaviors of the Moroccan population are progressing rapidly owing to profound societal and industrial changes. This progression would also be linked to the nutritional transition that shows local specificities so far scarcely documented [22].

Moroccan food habits are characterized by their tradition and culture but also by profound changes to a Western lifestyle [22]. Increasingly, the westernization of food fads that is characterized by more consumption of red meat, processed meat, alcohol, and junk food is observed, all this in a context of a sedentary and stressful lifestyle [11]. Given that the majority of these changes are considered to be CRC risk factors, this could explain part of the increasing incidence rates in a country supposed to be low in CRC incidence.

In Morocco, to our knowledge, no previous study has investigated associations between these risk factors and CRC. Therefore, we designed a multicenter case-control study to evaluate the relationship of the Moroccan diet, physical activity, and other lifestyle habits with CRC risk. This study will also describe the genetic profiles of CRCs in Morocco and their interaction with food intakes. Finally, we will also select CRC index cases that would identify family cancers in Morocco.

This case-control study was conducted between September 2009 and February 2017 in 5 major public health hospitals in Morocco, namely, the University Hospital Center (UHC) Hassan II of Fez, UHC Ibn Sina of Rabat, UHC Mohammed VI of Oujda, UHC Ibn Rushd of Casablanca, and UHC Mohammed VI of Marrakech. The sample size estimation was based on red meat consumption as one of the main exposures of interest. According to the National Survey of Dietary Habits in Morocco, the proportion of Moroccan adults eating red meat at least twice a week was 62.7% [23]. The sample size was calculated using the following formula specific for individual-matched case-control studies (Figure 1) [24], considering a type I error (Cronbach alpha) equal to 5%, a statistical power of 90% (beta=.10), and a minimum difference in terms of risk of 43% as reported by the WCRF/AICR report [5].

The number of pairs required for the study was 1496, that is, in each of the 2 groups 1496, which was rounded to 1500.

Cases were patients who had newly confirmed CRC diagnosis by histopathology less than 3 months before the interview and who did not start any therapeutic protocol (chemotherapy, radiation therapy, hormone therapy, or surgery) yet at the time of inclusion. However, for some patients included in the study, the diagnosis was done at the time of surgery. For the exclusion criteria, patients who had received chemotherapy, radiation therapy, or hormone therapy were excluded as they were not considered as newly diagnosed, and the therapy may affect their lifestyle habits.

Each case was matched with a control subject by age (±5 years), sex, and center. Other eligibility criteria included the following: at least 18 years old, no prior history of diabetes mellitus, capability to give consent, and ability to communicate and carry out the interview. Extra exclusion criteria were patients under exclusive palliative treatment and patients confined to their chairs or beds.

Controls were selected from the same local population and in the same hospitals as cases, among healthy subjects accompanying other patients or visitors. Eligibility criteria were the same as for cases, except that the controls should not have any personal history of CRC or any other type of cancer. Unfortunately, we did not have the funding to carry out a fecal occult blood test or colonoscopy for controls. However, to avoid (or minimize) the inclusion of false controls, questions about their medical history and clinical symptoms related to CRC were systematically asked to all controls.

Data were collected through face-to-face interviews conducted by trained investigators. All participants were asked to answer questionnaires on the following topics.

Dietary data were collected using a validated semiquantitative Food Frequency Questionnaire (FFQ) that was developed to assess food intake in the Moroccan population [27]. It was inspired from the Global Allergy and Asthma European Network (GA²LEN) [28] and validated in the Moroccan context [27].

Multimedia Appendix 1 depicts this FFQ that included 255 foods and the following 32 food groups: (1) bread, (2) breakfast with grains, (3) couscous (one of the traditional staple foods of Maghreb countries’ cuisine prepared by durum wheat semolina), (4) pasta, (5) cake, (6) rice, (7) sugar, (8) sweets without chocolate, (9) chocolate, (10) vegetable oil, (11) margarine and vegetable fat, (12) butter and animal fat, (13) dried fruit, (14) legumes, (15) vegetables, (16) potatoes, (17) fruits, (18) juice, (19) nonalcoholic beverages, (20) coffee/tea, (21) beer, (22) wine, (23) other alcoholic beverages, (24) red meat and processed meat, (25) poultry, (26) sekat (offal and brain), (27) fish, (28) eggs, (29) milk of cow/soya, (30) cheese, (31) other dairy products, and (32) miscellaneous foods (Multimedia Appendix 1).

Participants, both cases and controls, recorded their food consumption of the past year before the interview. Frequency of food consumption was recorded in 8 different categories (never, 1-3 times per month, once a week, 2-4 times per week, 5-6 times per week, once a day, 2-3 times per day, and ≥4 times/day). Regarding seasonal foods, participants were asked to answer the question based on intakes during periods/seasons when these foods are available. The daily intake of the foods was then calculated according to the number of months per year that each seasonal food was available. To convert foods into nutrient intakes, we essentially used food composition data from Tunisia [29] and Morocco [30]. The nutrient and energy intakes were calculated by multiplying the daily intakes of each food item by the nutrient and calorie content (per 100 g) of all food items.

In addition, the reproducibility and validity of this FFQ were evaluated among 105 healthy Moroccan adults. The results showed a good relative validity (deattenuated correlations ranging from 0.24 for fiber to 0.93 for total monounsaturated fatty acids) and a good reproducibility (intraclass correlation coefficient ranging from 0.69 for fat to 0.84 for Vitamin A) [27].

Pathology tumor samples were collected for 170 patients, whose anatomopathological tests were done in the UHCs and whose block tumors were available. In fact, the majority of patients are used to do their biopsies outside the university hospitals, in private sectors, which are difficult to access and not always willing to participate in research projects. For this reason, biopsies done in the private sector were not included in this study. In addition, only 3 out of 5 public health hospitals provided consent to participate in the genetics part of the study and because of financial difficulties, only a subsample of pathological tumor samples was obtained. The tumor samples were collected between February 2016 and July 2017 from 3 public health hospitals in Morocco (Casablanca, Oujda, and Rabat). In addition, all molecular analyses were done in the laboratory of genetics at the University Hospital Hassan II in Fez. A pathologist doctor classified the samples embedded in paraffin and registered and coded using consecutive and unique identification numbers. These stored paraffin-embedded tissues were collected from the 3 UHCs included in this study. DNA was extracted using an Invitrogen RNA/DNA isolation kit by manually scraping tissue from unstained slides. The BRAF and KRAS mutations will be determined by direct sequencing and analyzed by methylation-specific polymerase chain reaction.

In total, the study recruited 3032 subjects (1516 cases and 1516 controls). Table 1 depicts the data cleaning and handling by reporting the exclusions. Exclusions before starting statistical analysis included participants with unspecified primitive cancer (n=7), cases with old biopsies (6 cases), participants with missing dietary data because the FFQ was not well filled (n=10), duplicate records (n=2), and unmatched records (n=8).

The participation rate in this study was 97% (1516/1555) for cases and 0.75% (1516/2000) for controls. Table 2 depicts the missing data of all variables for the final sample included in this study (1483 cases and 1483 controls).

The protocol of this study has been reviewed and approved by the ethics Committee at the University of Fez in September 2009. Written informed consent was obtained from all participants before enrollment. Confidentiality of data is secured by removing personal identifiers from the datasets.

The descriptive information of the categorical variables was presented as the frequency of each category and for the continuous variables by means and standard deviation. Differences between continuous variables were examined by using the student t test (2 tailed) for matched samples. Chi-square tests (McNemar) were used to examine differences among categorical variables.