This study models some risk factors for some topographies of cancer in South West of Nigeria (Osun, Ondo, Ogun, Ekiti, Oyo and Lagos states). Data on  various topographies of cancer are collected by transcription from cancer registry and patients’ case note of different hospitals (teaching hospitals and medical  centres) across the six states. Binary logistic regression is used for modelling and probability of a patient suffering for a typical cancer is obtained with all the  significant risk factors identifies in each model. Due to insignificant of frequency observed for Prostate, Rectum, and Pancreas, only models for Breast, Cervix,  Colon, and Ovarian cancers are examined. Age, Marital Status, Age at first Menstruation, Use of Birth Control Pill, Consumption of High Fat Diet, Alcohol,  Obesity and having multiple sexual partners are all significant factors for breast cancer. Age at first Menstruation and Consumption of High Fat Diet are the  most significant factors at 5% level. Significant risk factors for cervical cancer based on the result of analyses are: Religion, Job, and Age at first Menstruation  while Age, Marital Status, and Educational Status are significant factors for colon cancer. For ovarian cancer, significant risk factors are Educational Status,  Residence (Urban or Rural), and Age at first menstruation, and Obesity. 

Cancer;Oncology;Topography;Risk factors;Binary logistics.

Iyiola RO, Aliu AA, Obafemi OS, Ajao IO, Adetunji AA*, et al. (2017) Modelling Cancer Risk Factors for Vital Topographies in the South Western States of Nigeria. J Cancer Biol Res 5(3): 1104.

Report [1] revealed that about 14.1 million new cancer cases excluding skin cancer occur a year with about 8.8 million deaths which is approximately 15.7% of human deaths [2]. Cancer has been predicted [3] to be an important cause of morbidity and mortality in the next few decades, especially in low and middleincome countries (mostly in Africa). It belongs to a group of diseases that involve abnormal cell growth with the possibility of spreading to different location in the body [4]. It forms a set of neoplasm’s (tumor) which is a group of cells with unregulated growth and often form a mass or lump, but may be distributed diffusely [5,6]. These tumors show various hallmark of cancers which are required to produce a malignant tumor. These include:

• Cell growth and division absent the proper signals

• Continuous growth and division even given contrary signals

• Limitless number of cell divisions

• Promoting blood vessel construction

• Avoidance of programmed cell death

• Invasion of tissue and formation of metastases [7]

In report by [8], Nigeria in the last few years is witnessing a tsunami in cancer incidence with about 102,000 new cases per annum along with the mortality rates of about 75,000 deaths per annum. This is largely due to associated problems with cancer care in the country and the general neglect of the country’s health care system. In descending order of frequency [9] showed the following cancers: breast, cervical, prostate, liver, and colorectal are prominent in Nigeria while [8] reported that prostate and liver cancers are most common in males while breast and cervical cancer are the most frequent ones in females. There are over 50 million women whose ages are 15 years and above in Nigeria with record showing over 14,000 of them diagnosed with cervical cancer on yearly basis and over 8,000 recorded deaths [10].

Little is still being done in area of diagnosis and treatment of cancer in Nigeria. This is largely due to lack of well-equipped personnel (oncologists); finances and modern equipment [11]. There is still no known any national policy on cancer in the country apart from various non-governmental organizations usually run by wives of political office holders. Among several other factors militating against cancer in Nigeria are poverty, poor health management, ignorance, and illiteracy [8,12,13].

While assessing the knowledge, attitudes, and practices of women concerning breast cancer in Jos and environs [14] reported that about half of 395 respondents do not have knowledge of signs and symptoms of breast cancer with majority of them having no idea of initiating self-breast examination.

In Nigeria, survival rate of cancer is next to nothing and cancer diagnosis is taken as death sentence. Data on cancer survival is mostly nonexistence and most of diagnosed patients prefer seeking spiritual assistance when they run out of cash for prescribed treatment. Most of the oncology departments of various teaching hospitals lack up-to-date record on cancer register. Those that use Can Reg (A cancer registry software provided by WHO) usually have several unknown cases in various records. Patients are usually found of reneging from treatment, making it difficult to have complete record on their treatment progression.

In Nigeria, survival rate of cancer is next to nothing and cancer diagnosis is taken as death sentence. Data on cancer survival is mostly nonexistence and most of diagnosed patients prefer seeking spiritual assistance when they run out of cash for prescribed treatment. Most of the oncology departments of various teaching hospitals lack up-to-date record on cancer register. Those that use Can Reg (A cancer registry software provided by WHO) usually have several unknown cases in various records. Patients are usually found of reneging from treatment, making it difficult to have complete record on their treatment progression.

Nigeria has a total surface area of 923,768 km2 . The study area is South-Western region of Nigeria (Figure 1).

Figure 1 Map of Nigeria showing the study areas.

South-western Nigeria consists of Lagos, Ogun, Oyo, Osun, Ondo and Ekiti states. The area lies between longitude 2° 311 and 6° 001 East and Latitude 6° 211 and 8° 371 N with a total land area of 77,818 km2 . The study area is bounded in the East by Edo and Delta states, in the North by Kwara and Kogi states, in the West by the Republic of Benin and in the south by the Gulf of Guinea.

Data used in this study were collected from cancer registry and patients’ case note of different hospitals (teaching hospitals and medical centres) across the six states of the South-West of Nigeria. Risk factors examined are Age, Sex (66 males and 510 females), Marital Status,Educational Status,Residence,Religion,Nature ofJob,Age atfirst ofmenstruation,Use ofBirthControlPills,Consumption of high fat diet, Alcohol, Physical Exercise, Obesity before diagnosis, Smoking, Number of Sexual Partner, Sexual Activities. Cancer topographies examined are limited to Breast, Cervix, Prostrate, Colon, Rectal, Ovarian, and Pancreatic. The age of the respondents are categorised as less than 35 years, 35–39, 40–44, 45–49, and 50 and above. Marital Status are categorised as Single, Married, Divorced, Separated and Widow. Respondents’ residences are grouped as Rural and Urban.

Most common method to analyse binary response data is the Logistic regression. It is used to model relationships between the response variable and several explanatory variables, which may be discrete or continuous. This is used for the situation where the response (Y) can only take one of two possible values usually alive/dead, or present/absent in practice. Logistic regression is useful in situations where the interest is to predict the presence or absence of a characteristic or outcome based on values of a set of predictor variables. This situation also arises frequently in medical trials, where at the end of the trial period, the patient has either recovered or has not. It is convenient to denote the two levels by 0 and 1 and to refer to the categories as a “failure” or a “success”. Statistical software used is STATA version 12 and SPSS IBM version 22. STATA especially was employed in the analysis due its enormous ability to accept external epidemiological plug ins to carry out some important calculations and creation of vital tables of results from statistical tool used, and SPSS for ease of coding. In order to ease the various computations necessary before obtaining the result, the response variables were made to have categories of two levels (dichotomous), hence the use of logistic regression to model the data.

Logistic regression model

To fit a binary logistic regression model, a set of regression coefficients that predict the probability of the outcome of interest are estimated. The same logistic model can be written in different ways. The version that shows what function of the probabilities results in a linear combination of parameters is

The log is the log of the odds that an event occurs. The odds that an event occurs are the ratio of the number of people who experience the event to the number of people who do not. This is obtained when the probability that the event occurs is divided by the probability that the event does not occur.

The general linear logistic regression model is defined as:

Where  are continuous measurements corresponding covariates and/or dummy variables corresponding to factor levels and   are the parameters.

is the probability the ith case experiences the event of interest.

is the jth predictor for the ith case

P is the number of predictors

In this research, logistic regression is used to model various characterised risk factors for cancer topography. The intent is to know how significant these factors are on various forms of cancer. All observed factors are modelled for each form of cancer.

Due to insignificant of frequency observed for Prostate, Rectum, and Pancreas, only models for Breast, Cervix, Colon, and Ovarian cancers are examined.

The classification table for cross tabulations of Table 1 shows

the values of the sensitivity and specificity of 67.39% and 71.54% respectively, and the percentage of correct classification is 64.71%. The area under the ROC curve is 0.7453 (74.53%) which is a strong predictive power.

From the Table 2,

Age, Marital Status, Age at first Menstruation, Use of Birth Control Pill, Consumption of High Fat Diet, Alcohol, Obesity and having multiple sexual partners are all significant factors for breast cancer. The odds ratio is the ratio-change in the odds of the event of interest for a one-unit change in the predictor. For example, in the Table 2, the odds for Consumption of high fat is 2.410, which means that the odds of default for a person who consumes high fat diet to have breast cancer is 2.410 times the odds of default for a person who does not consume such diet, all other things being equal. Based on the result obtained in Table 2 above, the probability that a patient will be diagnosed of Breast Cancer is expressed as:

The figures show the specificity and sensitivity (Figure 2)

Figure 2 Sensitivity and specificity by predicted probability.

of the classification table that produced Table 2 and the ROC curve for the cut-off point (Figure 3).

It reveals that the greater percentage of those correctly classified (sensitivity and specificity) is above the cut-off 0.5.

Classification table for the cross tabulation of Table 3 shows

the values of the sensitivity and specificity of 23.08% and 100.00% respectively while there is 89.58% of correct classification. The ROC curve is 0.7470 (74.70%) which is a strong predictive power. Significant risk factors for cervical cancer based on the result of Table 3 above are: Religion, Job, and Age at first Menstruation. It can also be observed from the Table 4

above that for example, the odds for Number of sexual partners and cervical cancer is 1.606, this implies that the odds of default for a person who has more than 1 sexual partners to have cervical cancer is 1.606 times the odds of default for a person who has only one person, all other things being equal. The probability that a patient will be diagnosed of Cervical Cancer can is obtained as:

The figures show the specificity and sensitivity (Figure 4)

Figure 4 Sensitivity and specificity by predicted probability.

of the classification table that produced Table 3 and the ROC curve for the cut-off point (Figure 5).

Figure 5 ROC curve for the best model.

It reveals that the greater percentage of those correctly classified (sensitivity and specificity) is above the

For the classification of data for Table 4, the values of the sensitivity and specificity of 33.33% and 98.70% respectively, and there is 93.98% of correct classification with the ROC curve is 0.9481 (94.81%) which is a strong predictive power. Table 4 gives significant risk factors for colon cancer as: Age, Marital Status, and Educational Status. The table also revealed for example that, the odds for a married person is 21.657, this implies that the odds of default for a person who is married to have cervical cancer is 21.657 times the odds of default for a person who is not, all other things being equal. The probability that a patient will be diagnosed of Colon Cancer is:

Figure 6 shows

Figure 6 Sensitivity and specificity by predicted probability.

that greater percentage of those correctly classified (sensitivity and specificity) is above the cut-off 0.5 while the ROC curve Figure 7 shows

Figure 7 ROC curve for the best model.

that all observed points are above the cut-off point.

Classification table for the reports obtained in Table 5 shows

the values of the sensitivity and specificity of 70.00% and 98.51% respectively with 94.81% of correct classification and a strong predictive power (ROC) of 92.69%. Table 5 shows that significant risk factors for ovarian cancer: Educational Status, Residence (Urban or Rural), and Age at first menstruation, and Obesity. The table also revealed for example that, the odds for someone who starts menstruation earlier in life is 251.660, this implies that the odds of default for a person who starts menstruating early is 251.660 times the odds of default for a person who does not, all other things being equal. The probability that a patient will be diagnosed of Ovarian Cancer can be is obtained as:Figure 8 shows

Figure 8 Sensitivity and specificity by predicted probability.

 that greater percentage of those correctly classified (sensitivity and specificity) is above the cut-off 0.5 while the ROC curve Figure 9 shows

Figure 9 ROC curve for the best model.

that all observed points are above the cut-off point.

The results from the various analyses carried out show that the risk factors determining the topographies of cancer on human differ. Since most of the cases handled are breast, cervix, colon, and ovarian cancers, it behaves the researchers to model the ones most reported in order to give room for concise prediction of probability of having a case or not having it. Almost all the factors considered are significant for breast cancer, though the odd ratios give more information on the significant factors. The specificity, sensitivity and predictability of the tests are presented in Figures 2-9, show various reliability of all the tables of cancers’ topography. Also the ROC curve was used for determination of the cut-off point and the largest area occupied.

From various analysis carried out, obtained results indicate that Age, Marital Status, Age at first Menstruation, Use of Birth Control Pill, Consumption of High Fat Diet, Alcohol, Obesity and having multiple sexual partners are all significant factors for breast cancer. Most of these factors have earlier been identified by [16]. Other reported risk factors for breast cancer by [17,18] include westernized diet, low fibre intake, family history of breast cancer and presence of benign breast disease. For cervical cancer significant risk factors are: Age at first Menstruation and Consumption of High Fat Diet while Religion, Job, and Age at first Menstruation are significant factors for cervical cancer. For colon cancer, Age, Marital Status, and Educational Status are significant risk factors and Educational Status, Residence (Urban or Rural), Age at first menstruation, and Obesity are significant for ovarian cancer.

The authors appreciate the Tertiary Education Trust Fund’s (TET Fund) Institutional Based Research (IBR) Intervention line for the sponsorship of this research and the Committee of Research (Centre for Research and Innovative Development, CRID) of the Federal Polytechnic, Ado-Ekiti for the approval.

Readers are strictly advice to consult the original database and original articles before making decisions based on any document. Authors are not responsible for action taken by readers based on any information in this report. People seeking individual medical advice are referred to their physician. This document is not intended as personal health advice.

1. Mucci LA, Wedren S, Tamimi RM, Trichopoulos D, Adami HO. The role of gene-environment interaction in the aetiology of human cancer: examples from cancers of the large bowel, lung and breast. J Intern Med. 2001; 249: 477-493.

2. Bray F, Moller B. Predicting the future burden of cancer. Nat Rev Cancer. 2006; 6: 63-74.

3. Cancer incidence and mortality worldwide. International Agency for Research on Cancer. 2011.

4. Cancer Fact sheet N° 297. World Health Organization. 2014.

5. Longnecker MP, Newcomb PA, Mittendorf R, Greenberg ER, Clapp RW, Bogdan GF, et al. Risk of breast cancer in relation to lifetime alcohol consumption. J Natl Cancer Inst. 1995; 87: 923-929

6. Stickel F, Schuppan D, Hahn EG, Seitz HK. Cocarcinogenic effects of alcohol in hepatocarcinogenesis. Gut. 2002; 51: 132–139. 7. Seitz HK, Poschl G, Simanowski UA. Alcohol and cancer. Recent Dev Alcohol. 1998; 14: 67-95

8. Ferlay J, Shin HR, Bray F. Estimates of worldwide burden of cancer in 2008. Int J Cancer. 2010; 127: 2893-2917.

9. Abdulkareem F. Epidemiology and incidence of common cancers in Nigeria. Cancer Reg and Epidemiology Workshop. 2013.

10. Nigeria Human Papillomavirus and Related Cancers, Fact Sheet 2017, ICO Information Centre on HPV and Cancer. 11.Adebamowo C. Cancer Care in Nigeria Part 1: The Social Cost of Cancer. American Soc Clin Oncol. 2013.

12. Harford JB. Barriers to overcome for effective cancer control in Africa. Lancet Oncol. 2015; 16: 385-393.

13. Eze JN, Emeka-Irem EN, Edegbe FO. A six-year study of the clinical presentation of cervical cancer and the management challenges encountered at a state teaching hospital in southeast Nigeria. Clin Med Insights Oncol, 2013; 7: 151-158.

14. Jimam NS, David S, Ejisun GO. Breast cancer in Jos, Nigeria: An audit on knowledge, attitude, and practices. Environ Dis. 2017; 2: 22-26.

15. Ima-Obong A. Ekanem. History of Cancer Registration in the World and Nigeria, presented at Institute of Human Virology Abuja. 2009.

16. Kolawole AO. Feasible Cancer Control Strategies for Nigeria: Mini-Review. Am J Trop Med Public Health. 2011; 1: 1-10.

17. World Cancer Report. World Health Organization. Geneva: 2014.

18. Cancer Facts and Figures. American Cancer Society. 2016