Correspondence Analysis And Its Application In Oncology

Objective:In China, tumors are extremely harmful and very terrible diseases, ones of which, malignant tumors, are the second highest mortality diseases only next to cardiovascular desease. In the world malignant tumor has also become a major killer of mankind. A lot of manpower, material and financial resources have been put into the tumor research including both basic and clinical research at home and abroad. And we have made certain progress. This article introduces correspondence analysis method and explore its application in oncology. We hope it can play an active role in the oncology development.Methods:The data come from following several aspects:(1) China health statistics annual in 2009; (2) A statistical table from an Analysis of the Constitution of the top 10 Malignant Tumors of In-patients from 2000 to 2004 written by Li Lixing. (3) A two-dimensional table from Statistical Analysis of Malignant Tumors of In-patients in Boertala Autonomous Prefecture Hospital written by Dong Su-hong and Liu Li-li. (4) A crosstabulation table from a Survey on Life Satisfaction in Cancer Patients written by Li An-le, Lu Yong-liang and Han Xue. (5) A contingency table from Expression of DAPK in Cervical Lesions in Uygur Women in Xinjiang written by Liu Xiao-wan, Mayinuer Niyazi, Abulizi Abudula et al. Study the relationship between age and tumor, between the tumors and the races, between demographic characteristics and life satisfaction, between cervical lesions of the women and death-associated protein kinase(DAPK). Adopting correspondence analysis method to study the relationship between them.Correspondence analysis has been recognized to be an important method in dealing with a large undifferentiated set of data. It is mainly employed in analyzing the relationship between row and column variables for 2-dimension data matrix.One of its commonest uses in practice is to scale a set of objects or individuals of large sample on the basis of the attributes that they possess. The other major use of the method is to classify variables. The method can simultaneously analyze the importance of all kinds of variables which are relevant and which are redundant, as well as relationship between the variables and objects or individuals of sample. It has intuitive, simple and convenient advantages. It has been widely applied in fields of botany, environmental science, psychology, medicine and hygiene, market research, etc. It is a better prospect of multivariate statistical analysis method. The statistical analysis software used in this research is SAS9.0 and SPSS 13.0. We explain the correspondence analysis method in detail through analog data. We introduce the basic idea of correspondence analysis and relate the principle, method, steps and calculation ot correspondonce analysis in detail. The SAS procedure of correspondence analysis is provided in this article. It can be used directly by practical operators.Results:From the datum of China health statistics annual in 2009, the results of this study indicate that age groups 60-,65-,70,75-and lung cancer, esophageal cancer are closely related with each other, age group 80-and gastric carcinoma are related with each other, age group 85-and colon cancer related, age groups 0-,55-and leukemia, bladder cancer,cervical carcinoma, breast cancer, nasopharyngeal carcinoma, liver cancer, benign tumor, the other tumors are related.From the datum of an Analysis of the Constitution of the top 10 Malignant Tumors of In-patients from 2000 to 2004, the results of this study show that the number of less than 30-year-old people suffering from leukemia is more than other age group, the number of people suffering from lung cancer, liver cancer, stomach cancer, esophageal cancer, intestinal cancer, and malignant lymphoma, over 60-year-old, is larger, while the number of patients suffering from breast cancer, cervical cancer and nasopharyngeal carcinoma, age groups from 30-to 50-is larger.From the datum of Statistical Analysis of Malignant Tumors of In-patients in Boertala Autonomous Prefecture Hospital, the results of this research indicate that the Han nationality and lung cancer, liver cancer, rectal cancer, breast cancer are more closely associated, Uygur ethnic group(Uygur national minority) and Mongolian nationality and gastric cancer, leukemia are more closely associated, Kazak nationality and cervical cancer,urinary bladder cancer are more closely associated, other nationality and the tumors are not obviously associated. For a Survey on Life Satisfaction in Cancer Patients, because the data of these two lines with male and female are not statistically significant, and the data sheet which is composed of the two rows by four columns, the dimension of which is one, is not analyzed. And the data table which is composed of marital status, the data of which are not statistically significant, is not analyzed, too. The results of correspondence anlysis show that college or above and not too satisfactory are more closely related, high school or technical secondary school and average satisfactory are related, junior middle school and very satisfactory are related, elementary school and very unsatisfactory are related, illiteracy and other categories or variables are not obviously related.Suffering from a cancer for 3-years and not too satisfactory are more closely related. Suffering from a cancer for 4-years and very satisfactory are linked with each other. Suffering from a cancer for 1-year and average satisfactory are linked with each other. Suffering from a cancer for age groups< 1 and 2-and very unsatisfactory are linked. Suffering from a cancer for>=5 and other categories or variables are not obviously linked.Retirement and very satisfactory are closely linked. Incumbent and average satisfactory are linked. Jobless and very unsatisfactory are linked. Other employment status and very unsatisfactory are also linked.From the datum of Expression of DAPK in Cervical Lesions in Uygur Women in Xinjiang, according to correspondence analysis, the control group and DAPK protein expression strong positive(+++) are closely associated. Group CIN I and expression positive are closely associated. Group CINII+ III and weakly positive(+) are closely associated. Group SCC and negative (-) are closely associated.Comparing the results from correspondence analysis method and clustering analysis method, we can see correspondence analysis can analyzes the data with more 0; and correspondence analysis has advantages of comprehensive evaluation. It can classify the data with the variables and the objects or individuals of sample. But the cluster analysis is very difficult to deal with them.Conclusions:This research has discussed the practical application of correspondence analysis method in five aspects of oncology. We have obtained the desired effect.1. Patients aged 60-to 75-(Groups from 60-to 75-years old of patients) probably died of lung cancer, esophageal cancer. Patients aged 80-probably died of gastric carcinoma. Patients aged 85-probably died of colon cancer. Patients aged from 0-to 55-died of leukemia, bladder cancer, cervical cancer, breast cancer, nasophaiyngeal cancer, liver cancer, benign tumors and other tumors more likely.2. For women patients from Zhaoqing city, Guangdong province, the patients aged under 30 probably suffer from leukemia. The patients aged over 60 probably suffer from liver cancer, stomach cancer, esophageal cancer or malignant lymphoma. The patients aged from 30-to 50-suffer from breast cancer, cervical cancer or nasopharyngeal carcinoma more likely.3. For Boertala Autonomous Prefecture, the patients suffering from tumors have national aggregation. Han residents are predisposed to lung cancer, liver cancer, rectal cancer or breast cancer. Uighur and Mongolian residents are predisposed to leukemia. Kazak residents are predisposed to cervical cancer or bladder cancer.4. Cancer patients whose education is elementary school, Suffering from a cancer for<1 and 2-years, whose employment status is jobless or other employment conditions, are very unsatisfactory to life. Cancer patients whose education is college or above, Suffering from a cancer for 3-, are not too satisfactory to life. Cancer patients whose education is high school or technical secondary school, Suffering from a cancer for 1-, whose employment status is incumbent, are average satisfactory to life. Cancer patients whose education is junior middle school, whose employment status is retirement, are very satisfactory to life.5. DAPK gene and (cervical lesions of women)women cervical disease have stronger correlation.6. Correspondence analysis can analyzes the data with more 0; and correspondence analysis has advantages of comprehensive evaluation. It can classify the data with the variables and the objects or individuals of sample. But the cluster analysis is very difficult to deal with them.