污染和环境因素与我国食管癌的关系
|
摘要
背景和目的:
食管癌(EC)是人类常见的一种恶性消化道肿瘤。中国是世界上食管癌死亡率最高的国家,几十年来发病率和死亡率相对稳定。中国食管癌死亡率呈明显的地理聚集现象,提示特定的环境因素对其发生起着重要作用。氮循环病因假说认为,在特定的气候条件下,农肥、污水使饮用水中两类亚硝胺前体物(二级胺和硝酸盐、亚硝酸盐)增加,在体内合成亚硝胺引发食管癌。水污染必然造成食物污染。环境因素主要通过饮食和生活习惯作用于人体。中国各地氮污染程度、环境条件和社会经济状况等多因素的联合作用与食管癌高发的关系至今尚不清楚。氮循环病因假说仍需要更多实验数据、流行病学和统计学证据的支持。因此,环境氮污染-亚硝胺-食管癌关系的验证对食管癌病因研究及预防具有重要的理论意义和应用价值。本研究主要探讨我国环境氮污染综合相关危险因素与食管癌风险及其聚集性的关系。
方法:
资料来源:中国肿瘤数据库提供了全国各县70年代(1973-1975年)和90年代(1990-1992年)监测点食管癌标化死亡率(SMR,1964年中国标准人口构成标化),中国自然数据库提供了90年代(1990-1996年)全国环境、气候、农业和社会经济数据,数据权威可靠。本课题拟探讨食管癌死亡率是否具有聚集性;筛选全国食管癌环境氮污染及其相关危险因素;寻找食管癌聚集性的可能因素;评价相关因素的估计危险度(odds ratio, OR);以筛选出的相关因素作为预测变量,用反证法验证这些危险因素的准确性。拟合中国食管癌死亡率空间概率分布;应用因子分析逐步回归法筛选全国食管癌环境氮污染及其相关危险因素,重点研究特定气象条件下氮污染(肉产量及猪存栏数、氮肥使用量、秸秆产量)和农药污染(农药使用量)与食管癌死亡率的关系;通过混合线性模型(MLM)研究食管癌的聚集因素;利用负二项回归评价相关危险因素的估计危险度(odds ratio, OR);采用人工神经网络模型(BPNN)对食管癌死亡率进行预测。研究采用SPSS、DPS和STATA等软件进行。
结果:
河南省食管癌死亡率相关因素研究结果河南省食管癌死亡率空间分布符合负二项分布(其χ2值为最小,其中男性χ2=44.6378、女性χ2=58.2165),提示EC死亡率存在着地理聚集性。因子分析—负二项逐步回归分析结果显示,河南省男性食管癌流行因素因子分析提取出3个公因子,分别是:耕地猪肉因子、湿度因子和人密度菜因子。这三个因子的估计相对危险度分别是1.319、0.859和0.793。河南省女性食管癌流行因素因子分析也提取出同样3个公因子,分别是:耕地猪肉因子、湿度因子和人密度菜因子。这三个因子的估计相对危险度分别是1.249、0.784和0.806。河南省男性食管癌死亡率与女性食管癌死亡率具有共同的流行因素,但是流行因素对食管癌死亡率的影响程度不同。
我国食管癌死亡率相关因素及聚集因素趋势面分析结果显示我国70年代和90年代男性食管癌死亡率地理分布特征在空间结构上均呈聚集分布。因子分析—负二项逐步回归分析结果显示,我国70年代人群中食管癌流行因素是干燥因子、肥药因子、耕地猪秸秆因子和人密度耕地因子。其中70年代男性食管癌流行因素是:干燥因子(OR = 1.31)、肥药因子(OR = 1.17)、人密度耕地因子(OR = 1.12)和耕地猪秸秆因子(OR = 1.05)。70年代女性食管癌流行因素因子分析也提取出相同4个公因子,分别是:干燥因子(OR = 1.28)、肥药因子(OR = 1.19)、人密度耕地因子(OR = 1.15)和耕地猪秸秆因子(OR = 1.07)。
混合线型模型结果显示,我国70年代人群中食管癌流行因素是耕地猪秸秆因子和肥药因子,造成EC地理性聚集的因素是经济菜瓜因子、干燥因子、人密度耕地因子和旱涝指数因子。其中耕地猪秸秆因子(方差估计值Estimate= 0.2404, 95%CI = 0.0735～0.4074, p = 0.012)和肥药因子(Estimate = 0.2228, 95%CI = 0.0726～0.3731, p = 0.005)对我国70年代男性SMR均有固定影响;经济菜瓜因子(协方差估计值Covariance estimate(CE) = 21.667, 95%CI = 8.9642～52.3723, p = 0.026)、干燥因子(CE = 0.3151, 95%CI = 0.1579～0.6288, p = 0.026)、人密度耕地因子(CE = 0.2690, 95%CI = 0.1101～0.6570, p = 0.028)和旱涝指数因子(CE = 0.037, 95%CI = 0.0145～0.0831, p = 0.025)等4个因子是导致我国省内男性食管癌聚集的4个因素。耕地猪秸秆因子(Estimate = 0.2580, 95%CI = 0.0465～0.4695, p = 0.027)和肥药因子(Estimate = 0 .2104, 95%CI = 0 .0635～0.3572, p = 0.008)对我国70年代女性SMR均有影响;经济菜瓜因子(CE = 19.5861, 95%CI = 7.2817～52.6821, p = 0 .048)、干燥因子(CE = 0 .4312, 95%CI = 0.2162～0.8603, p = 0 .005)、人密度耕地因子(CE = 0.3353, 95%CI = 0. 1428～0.7869, p = 0 .022)和旱涝指数因子(CE = 0 .0420, 95%CI = 0.0176～0.1000, p = 0.024)等4个因子是导致省内女性食管癌聚集的4个因素。
我国90年代人群中食管癌的共同流行因素是蒸发量因子、人口密度因子、氮相关因子和经济菜瓜因子。其中男性食管癌流行因素是:蒸发量因子(OR = 1.30)、人口密度因子(OR = 1.29)、氮相关因子(OR = 1.15)和经济菜瓜因子(OR = 0.75)。女性食管癌流行因素是:蒸发量因子(OR = 1.40)、人口密度因子(OR =1.25)、氮相关因子(OR =1.140)和经济菜瓜因子(OR = 0.72)。BP神经网络分析得出的各因素对于我国90年代人群食管癌死亡率影响的重要性依次是:氮相关因子、人口密度因子、经济菜瓜因子和蒸发量因子;模型的男性食管癌死亡率预测精度为90%;模型的女性食管癌死亡率预测精度为87%。
因此,我国70年代和90年代男性食管癌死亡率与女性食管癌死亡率均具有共同的流行因素。
结论:
1.河南省食管癌死亡率的相关因素
在河南省,单位耕地猪存栏数和肉产量越大,相对湿度越低和年均降水越低,可能是食管癌的危险因素;菜地比例可能是食管癌的保护因素。环境畜牧粪便污染可能是影响食管癌SMR的重要危险因素之一。
2.我国食管癌死亡率相关因素及地理性聚集的影响因素
1.高单位耕地猪存栏数和肉产量、高干燥度和大化肥农药用量等可能是食管癌的危险因素。环境畜牧粪便污染可能是影响我国食管癌SMR的重要危险因素之一。
2、较高的相对湿度、菜瓜比例和经济水平,可能是我国食管癌的保护因素。
3.经济水平、菜瓜比例、蒸发量、年均降水量、相对湿度、人口密度、耕地比例和旱涝指数是造成我国EC地理性聚集的影响因素。
总之,本研究结果支持氮循环假说,在特定的气候条件下,农肥、污水使饮用水中两类亚硝胺前体物(二级胺和硝酸盐、亚硝酸盐)增加,在体内合成亚硝胺引发食管癌。
Objective:
Esophageal cancer (EC) is a common malignant tumor in human digestive tract. China is the country with the highest mortality rate of EC. The morbidity and mortality rates are relatively stable in the past several decades. The phenomena of significantly geographic aggregation of EC in China suggest that specific environmental factors may play an important role in its incidence. A hypothesis of the nitrogen cycle regarding EC causes proposed that agricultural fertilizer and sewage under special condition of weather are the sources of N-nitrosamine precursors (secondary amine, nitrite and nitrate) in drinking water. The precursors synthesize N-nitroso compounds(NOCs)in vivo to cause esophageal cancer. Water pollution will inevitably result in food contamination. Environmental factors affect the human health mainly by the diet and lifestyle. As yet, the relationship among the high incidence of EC and the integrative effects of multiple factors involved in the nitrogen pollution levels and the conditions of weather, environment and socio-economics in the various regions of China is unclear. The evidences of experiment, epidemiology and statistic for the hypothesis of the nitrogen cycle are still need. Therefore, the verification of linkage of environmental nitrogen pollution - NOCs - EC has great theoretical and practice significance for research and prevention of EC in the etiology. The study mainly explores the relationship of interaction of nitrogen pollution and environmental factors, and EC risk and its aggregation in special region.
Method:
The standardized mortality rate(SMR)of EC of the counties, adjusted by China standard population in 1964, were provided by China Cancer Database in 1970’s (1973-1975) and in 1990’s (1990-1992) (only data of monitoring counties). The environmental and weather data, agriculture and socio-economic data were also provided by Chinese Natural Resources Database in 1990’s (1990-1996). The data are complete and reliable. This study aimed to understand the local aggregation of EC and identify potential factors involved in the high risk of EC and its aggregation, especially the environmental factors of nitrogen pollution. The estimated relative risk (odds ratio, OR) of relevant factors was evaluated. The accuracy of these factors was also verified by reduction to absurdity. First, the space probability distributions of EC mortality in China were fitted by various distributive models. After the components were extracted by principal component analysis, a stepwise regression analysis was performed to identify relevant epidemic factors of EC that focus on the pollution factors of nitrogen (i.e., meat yield, hog population , the usage of nitrogen fertilizer and straw yield) and pesticide (the usage of pesticides) under specific weather conditions. Then, the local factors regarding aggregation of EC were studied by a Mixed-Linear Model (MLM). When a regression analysis of negative binomial was performed to evaluate the estimated relative risk of relevant factors, BP Neural Network Model (BPNN) was used to predict SMR of EC by the potential factors. We analyzed the data by SPSS, STATA and DPS.
Result:
The factors related to the mortality rate of EC in Henan Province
Theχ2 value of the negative binomial distribution was smallest (maleχ2 = 44.6378, womenχ2 = 58.2165), so the distribution on SMR of EC in Henan was a negative binomial distribution, which suggested that there was a geographic aggregation of mortality rate. Factor-negative binomial regression analysis showed that three components (3 Factors) were identified from 6 principal components (independent variables). These Factors related to the epidemic of male EC in Henan were named Hog-Meat-Farmland-Factor, Humidity-Factor and Population-Vegetable- Factor. The estimated relative risks (ORs) of three Factors were 1.319, 0.859 and 0.793, respectively. In addition, we also identified 3 Factors involved in the epidemic of female EC in Henan, which was the same as the male. The ORs of Hog-Meat-Farmland-Factor, Humidity-Factor and Population-Vegetable-Factor in female were 1.249, 0.784 and 0.806, respectively. In Henan Province, the mortality rates of EC in male and female had common endemic factors. However, their impacts on the mortality rates of EC were different.
The factors related to the mortality rate of EC in china
The result of trend surface analysis showed a character of the geographic aggregated distribution of mortality rates of EC in 1970’s or 1990’s. In 1970’s, Factor-negative binomial regression analysis revealed that the epidemic factors of male EC were Dry-Factor(OR = 1.31), Fertilizer-Pesticide-Factor(OR = 1.17),Population-Farmland-Factor(OR = 1.12) and hog-Straw-Farmland-Factor (OR = 1.05). Four Factors involved in the epidemic of female EC in 1970’s, which was also the same as the male, were Fertilizer-Pesticide-Factor(OR = 1.28), Dry-Factor(OR = 1.28), Fertilizer-Pesticide-Factor(OR = 1.19), Population-Farmland-Factor(OR = 1.15)and hog-Straw-Farmland-Factor(OR = 1.07).
Mixed linear model showed that Hog-Straw-Farmland-Factor and Fertilizer-Pesticide- Factor were risk factors of EC whilst Economic-Vegetable-Factor, Dry-Factor, Population-Farmland-Factor and Drought-Flood-Factor might cause the aggregation of EC in a Province in 1970’s. Hog-Straw-Farmland-Factor (Variance estimate = 0.2404 with 95%CI = 0.0735-0.4074, p = 0.012)and Fertilizer-Pesticide-Factor(Variance estimate = 0.2228 with 95%CI = 0.0726-0.3731, p = 0.005) were the risk factors of male EC in 1970’s. In addition, four Factors might cause the aggregation of male EC in a Province in 1970’s. They were Economic-Vegetable- Factor (Covariance estimate (CE) = 21.6674 with 95%CI = 8.9642-52.3723, p = 0.026), Dry-Factor (CE = 0.3151 with 95%CI = 0.1579-0.6288, p = 0.026), Population-Farmland-Factor (CE = 0.2690 with 95%CI = 0.1101-0.6570, p = 0.028)and Drought-Waterlogged-Factor (CE = 0.037, 95%CI = 0.0145-0.0831, p = 0.025). Hog-Straw-Farmland-Factor(Variance estimate = 0.2580 with 95%CI=0.0465-0.4695, p = 0.027)and Fertilizer-Pesticide-Factor(Variance estimate = 0.2104 with 95%CI = 0 .0635-0.3572, p = 0.008) were also the risk factors of female EC in 1970’s. Economic-Vegetable-Factor (CE = 19.5861 with 95%CI = 7.2817-52.6821, p = 0.048), Dry-Factor (CE = 0.4312 with 95%CI = 0.2162-0.8603, p = 0.005), Population-Farmland-Factor (CE = 0.3353 with 95%CI = 0. 1428-0.7869, p = 0.022)and Drought-Waterlogged -Factor (CE = 0.0420 with 95%CI = 0.0176-0.1000, p = 0.024)might also cause the aggregation of female EC in a Province in 1970’s.
In 1990’s, the epidemic factors of EC in China were Evaporation-Factor, Population density-Factor, Nitrogen-Factor and Economic-Vegetable-Factor. The male OR of Evaporation- Factor, Population density-Factor, Nitrogen-Factor and Economic-Vegetable-Factor were 1.30, 1.29, 1.15 and 0.75, respectively, whilst the female OR of these Factors were 1.40, 1.25, 1.14 and 0.72, respectively. According to the relative importance of inputs of BPNN analysis in China in 1990’s, the order of priority were Nitrogen-Factor, Population density-Factor, Economic-Vegetable-Factor and Evaporation-Factor. The prediction accuracies of EC-SMR by BPNN were 90% in male and 87% in female.
Therefore, in China, there were common epidemic factors of EC in the people of different genders and times.
Conclusion:
The related factors of mortality rate of EC in Henan Province
In Henan Province, the more hog population and meat yield per hectare of the arable land, the lower humidity and average annual rainfall might be the risk factors of EC. But, the bigger proportion of vegetable land may be the protective factors. The environmental pollution of animal excrement may be one of important reasons for the regional difference of EC-SMR.
Factors related to EC mortality rate and its geographic aggregation in China
1. The more hog population and meat yield per hectare of the arable land, the greater aridity and the more usage of chemical fertilizer and pesticide might be the risk factors of EC. The environmental pollution of animal excrement may be one of important reasons for the regional difference of EC-SMR in China.
2. The higher the relative humidity, the economic level and the ratio of the vegetable field in arable land might be the protective factors.
3. Economic level, population density, the proportion of vegetable field and arable land, evaporation, average annual precipitation, relative humidity and Drought-Waterlogged Index might cause the geographic aggregation of EC in China.
The result supports the view that agricultural fertilizer and sewage under special condition of weather are the sources of the precursors of N-nitrosamine which might synthesize N-nitroso compounds in vivo to cause EC in China.
食管癌(EC)是人类常见的一种恶性消化道肿瘤。中国是世界上食管癌死亡率最高的国家,几十年来发病率和死亡率相对稳定。中国食管癌死亡率呈明显的地理聚集现象,提示特定的环境因素对其发生起着重要作用。氮循环病因假说认为,在特定的气候条件下,农肥、污水使饮用水中两类亚硝胺前体物(二级胺和硝酸盐、亚硝酸盐)增加,在体内合成亚硝胺引发食管癌。水污染必然造成食物污染。环境因素主要通过饮食和生活习惯作用于人体。中国各地氮污染程度、环境条件和社会经济状况等多因素的联合作用与食管癌高发的关系至今尚不清楚。氮循环病因假说仍需要更多实验数据、流行病学和统计学证据的支持。因此,环境氮污染-亚硝胺-食管癌关系的验证对食管癌病因研究及预防具有重要的理论意义和应用价值。本研究主要探讨我国环境氮污染综合相关危险因素与食管癌风险及其聚集性的关系。
方法:
资料来源:中国肿瘤数据库提供了全国各县70年代(1973-1975年)和90年代(1990-1992年)监测点食管癌标化死亡率(SMR,1964年中国标准人口构成标化),中国自然数据库提供了90年代(1990-1996年)全国环境、气候、农业和社会经济数据,数据权威可靠。本课题拟探讨食管癌死亡率是否具有聚集性;筛选全国食管癌环境氮污染及其相关危险因素;寻找食管癌聚集性的可能因素;评价相关因素的估计危险度(odds ratio, OR);以筛选出的相关因素作为预测变量,用反证法验证这些危险因素的准确性。拟合中国食管癌死亡率空间概率分布;应用因子分析逐步回归法筛选全国食管癌环境氮污染及其相关危险因素,重点研究特定气象条件下氮污染(肉产量及猪存栏数、氮肥使用量、秸秆产量)和农药污染(农药使用量)与食管癌死亡率的关系;通过混合线性模型(MLM)研究食管癌的聚集因素;利用负二项回归评价相关危险因素的估计危险度(odds ratio, OR);采用人工神经网络模型(BPNN)对食管癌死亡率进行预测。研究采用SPSS、DPS和STATA等软件进行。
结果:
河南省食管癌死亡率相关因素研究结果河南省食管癌死亡率空间分布符合负二项分布(其χ2值为最小,其中男性χ2=44.6378、女性χ2=58.2165),提示EC死亡率存在着地理聚集性。因子分析—负二项逐步回归分析结果显示,河南省男性食管癌流行因素因子分析提取出3个公因子,分别是:耕地猪肉因子、湿度因子和人密度菜因子。这三个因子的估计相对危险度分别是1.319、0.859和0.793。河南省女性食管癌流行因素因子分析也提取出同样3个公因子,分别是:耕地猪肉因子、湿度因子和人密度菜因子。这三个因子的估计相对危险度分别是1.249、0.784和0.806。河南省男性食管癌死亡率与女性食管癌死亡率具有共同的流行因素,但是流行因素对食管癌死亡率的影响程度不同。
我国食管癌死亡率相关因素及聚集因素趋势面分析结果显示我国70年代和90年代男性食管癌死亡率地理分布特征在空间结构上均呈聚集分布。因子分析—负二项逐步回归分析结果显示,我国70年代人群中食管癌流行因素是干燥因子、肥药因子、耕地猪秸秆因子和人密度耕地因子。其中70年代男性食管癌流行因素是:干燥因子(OR = 1.31)、肥药因子(OR = 1.17)、人密度耕地因子(OR = 1.12)和耕地猪秸秆因子(OR = 1.05)。70年代女性食管癌流行因素因子分析也提取出相同4个公因子,分别是:干燥因子(OR = 1.28)、肥药因子(OR = 1.19)、人密度耕地因子(OR = 1.15)和耕地猪秸秆因子(OR = 1.07)。
混合线型模型结果显示,我国70年代人群中食管癌流行因素是耕地猪秸秆因子和肥药因子,造成EC地理性聚集的因素是经济菜瓜因子、干燥因子、人密度耕地因子和旱涝指数因子。其中耕地猪秸秆因子(方差估计值Estimate= 0.2404, 95%CI = 0.0735～0.4074, p = 0.012)和肥药因子(Estimate = 0.2228, 95%CI = 0.0726～0.3731, p = 0.005)对我国70年代男性SMR均有固定影响;经济菜瓜因子(协方差估计值Covariance estimate(CE) = 21.667, 95%CI = 8.9642～52.3723, p = 0.026)、干燥因子(CE = 0.3151, 95%CI = 0.1579～0.6288, p = 0.026)、人密度耕地因子(CE = 0.2690, 95%CI = 0.1101～0.6570, p = 0.028)和旱涝指数因子(CE = 0.037, 95%CI = 0.0145～0.0831, p = 0.025)等4个因子是导致我国省内男性食管癌聚集的4个因素。耕地猪秸秆因子(Estimate = 0.2580, 95%CI = 0.0465～0.4695, p = 0.027)和肥药因子(Estimate = 0 .2104, 95%CI = 0 .0635～0.3572, p = 0.008)对我国70年代女性SMR均有影响;经济菜瓜因子(CE = 19.5861, 95%CI = 7.2817～52.6821, p = 0 .048)、干燥因子(CE = 0 .4312, 95%CI = 0.2162～0.8603, p = 0 .005)、人密度耕地因子(CE = 0.3353, 95%CI = 0. 1428～0.7869, p = 0 .022)和旱涝指数因子(CE = 0 .0420, 95%CI = 0.0176～0.1000, p = 0.024)等4个因子是导致省内女性食管癌聚集的4个因素。
我国90年代人群中食管癌的共同流行因素是蒸发量因子、人口密度因子、氮相关因子和经济菜瓜因子。其中男性食管癌流行因素是:蒸发量因子(OR = 1.30)、人口密度因子(OR = 1.29)、氮相关因子(OR = 1.15)和经济菜瓜因子(OR = 0.75)。女性食管癌流行因素是:蒸发量因子(OR = 1.40)、人口密度因子(OR =1.25)、氮相关因子(OR =1.140)和经济菜瓜因子(OR = 0.72)。BP神经网络分析得出的各因素对于我国90年代人群食管癌死亡率影响的重要性依次是:氮相关因子、人口密度因子、经济菜瓜因子和蒸发量因子;模型的男性食管癌死亡率预测精度为90%;模型的女性食管癌死亡率预测精度为87%。
因此,我国70年代和90年代男性食管癌死亡率与女性食管癌死亡率均具有共同的流行因素。
结论:
1.河南省食管癌死亡率的相关因素
在河南省,单位耕地猪存栏数和肉产量越大,相对湿度越低和年均降水越低,可能是食管癌的危险因素;菜地比例可能是食管癌的保护因素。环境畜牧粪便污染可能是影响食管癌SMR的重要危险因素之一。
2.我国食管癌死亡率相关因素及地理性聚集的影响因素
1.高单位耕地猪存栏数和肉产量、高干燥度和大化肥农药用量等可能是食管癌的危险因素。环境畜牧粪便污染可能是影响我国食管癌SMR的重要危险因素之一。
2、较高的相对湿度、菜瓜比例和经济水平,可能是我国食管癌的保护因素。
3.经济水平、菜瓜比例、蒸发量、年均降水量、相对湿度、人口密度、耕地比例和旱涝指数是造成我国EC地理性聚集的影响因素。
总之,本研究结果支持氮循环假说,在特定的气候条件下,农肥、污水使饮用水中两类亚硝胺前体物(二级胺和硝酸盐、亚硝酸盐)增加,在体内合成亚硝胺引发食管癌。
Objective:
Esophageal cancer (EC) is a common malignant tumor in human digestive tract. China is the country with the highest mortality rate of EC. The morbidity and mortality rates are relatively stable in the past several decades. The phenomena of significantly geographic aggregation of EC in China suggest that specific environmental factors may play an important role in its incidence. A hypothesis of the nitrogen cycle regarding EC causes proposed that agricultural fertilizer and sewage under special condition of weather are the sources of N-nitrosamine precursors (secondary amine, nitrite and nitrate) in drinking water. The precursors synthesize N-nitroso compounds(NOCs)in vivo to cause esophageal cancer. Water pollution will inevitably result in food contamination. Environmental factors affect the human health mainly by the diet and lifestyle. As yet, the relationship among the high incidence of EC and the integrative effects of multiple factors involved in the nitrogen pollution levels and the conditions of weather, environment and socio-economics in the various regions of China is unclear. The evidences of experiment, epidemiology and statistic for the hypothesis of the nitrogen cycle are still need. Therefore, the verification of linkage of environmental nitrogen pollution - NOCs - EC has great theoretical and practice significance for research and prevention of EC in the etiology. The study mainly explores the relationship of interaction of nitrogen pollution and environmental factors, and EC risk and its aggregation in special region.
Method:
The standardized mortality rate(SMR)of EC of the counties, adjusted by China standard population in 1964, were provided by China Cancer Database in 1970’s (1973-1975) and in 1990’s (1990-1992) (only data of monitoring counties). The environmental and weather data, agriculture and socio-economic data were also provided by Chinese Natural Resources Database in 1990’s (1990-1996). The data are complete and reliable. This study aimed to understand the local aggregation of EC and identify potential factors involved in the high risk of EC and its aggregation, especially the environmental factors of nitrogen pollution. The estimated relative risk (odds ratio, OR) of relevant factors was evaluated. The accuracy of these factors was also verified by reduction to absurdity. First, the space probability distributions of EC mortality in China were fitted by various distributive models. After the components were extracted by principal component analysis, a stepwise regression analysis was performed to identify relevant epidemic factors of EC that focus on the pollution factors of nitrogen (i.e., meat yield, hog population , the usage of nitrogen fertilizer and straw yield) and pesticide (the usage of pesticides) under specific weather conditions. Then, the local factors regarding aggregation of EC were studied by a Mixed-Linear Model (MLM). When a regression analysis of negative binomial was performed to evaluate the estimated relative risk of relevant factors, BP Neural Network Model (BPNN) was used to predict SMR of EC by the potential factors. We analyzed the data by SPSS, STATA and DPS.
Result:
The factors related to the mortality rate of EC in Henan Province
Theχ2 value of the negative binomial distribution was smallest (maleχ2 = 44.6378, womenχ2 = 58.2165), so the distribution on SMR of EC in Henan was a negative binomial distribution, which suggested that there was a geographic aggregation of mortality rate. Factor-negative binomial regression analysis showed that three components (3 Factors) were identified from 6 principal components (independent variables). These Factors related to the epidemic of male EC in Henan were named Hog-Meat-Farmland-Factor, Humidity-Factor and Population-Vegetable- Factor. The estimated relative risks (ORs) of three Factors were 1.319, 0.859 and 0.793, respectively. In addition, we also identified 3 Factors involved in the epidemic of female EC in Henan, which was the same as the male. The ORs of Hog-Meat-Farmland-Factor, Humidity-Factor and Population-Vegetable-Factor in female were 1.249, 0.784 and 0.806, respectively. In Henan Province, the mortality rates of EC in male and female had common endemic factors. However, their impacts on the mortality rates of EC were different.
The factors related to the mortality rate of EC in china
The result of trend surface analysis showed a character of the geographic aggregated distribution of mortality rates of EC in 1970’s or 1990’s. In 1970’s, Factor-negative binomial regression analysis revealed that the epidemic factors of male EC were Dry-Factor(OR = 1.31), Fertilizer-Pesticide-Factor(OR = 1.17),Population-Farmland-Factor(OR = 1.12) and hog-Straw-Farmland-Factor (OR = 1.05). Four Factors involved in the epidemic of female EC in 1970’s, which was also the same as the male, were Fertilizer-Pesticide-Factor(OR = 1.28), Dry-Factor(OR = 1.28), Fertilizer-Pesticide-Factor(OR = 1.19), Population-Farmland-Factor(OR = 1.15)and hog-Straw-Farmland-Factor(OR = 1.07).
Mixed linear model showed that Hog-Straw-Farmland-Factor and Fertilizer-Pesticide- Factor were risk factors of EC whilst Economic-Vegetable-Factor, Dry-Factor, Population-Farmland-Factor and Drought-Flood-Factor might cause the aggregation of EC in a Province in 1970’s. Hog-Straw-Farmland-Factor (Variance estimate = 0.2404 with 95%CI = 0.0735-0.4074, p = 0.012)and Fertilizer-Pesticide-Factor(Variance estimate = 0.2228 with 95%CI = 0.0726-0.3731, p = 0.005) were the risk factors of male EC in 1970’s. In addition, four Factors might cause the aggregation of male EC in a Province in 1970’s. They were Economic-Vegetable- Factor (Covariance estimate (CE) = 21.6674 with 95%CI = 8.9642-52.3723, p = 0.026), Dry-Factor (CE = 0.3151 with 95%CI = 0.1579-0.6288, p = 0.026), Population-Farmland-Factor (CE = 0.2690 with 95%CI = 0.1101-0.6570, p = 0.028)and Drought-Waterlogged-Factor (CE = 0.037, 95%CI = 0.0145-0.0831, p = 0.025). Hog-Straw-Farmland-Factor(Variance estimate = 0.2580 with 95%CI=0.0465-0.4695, p = 0.027)and Fertilizer-Pesticide-Factor(Variance estimate = 0.2104 with 95%CI = 0 .0635-0.3572, p = 0.008) were also the risk factors of female EC in 1970’s. Economic-Vegetable-Factor (CE = 19.5861 with 95%CI = 7.2817-52.6821, p = 0.048), Dry-Factor (CE = 0.4312 with 95%CI = 0.2162-0.8603, p = 0.005), Population-Farmland-Factor (CE = 0.3353 with 95%CI = 0. 1428-0.7869, p = 0.022)and Drought-Waterlogged -Factor (CE = 0.0420 with 95%CI = 0.0176-0.1000, p = 0.024)might also cause the aggregation of female EC in a Province in 1970’s.
In 1990’s, the epidemic factors of EC in China were Evaporation-Factor, Population density-Factor, Nitrogen-Factor and Economic-Vegetable-Factor. The male OR of Evaporation- Factor, Population density-Factor, Nitrogen-Factor and Economic-Vegetable-Factor were 1.30, 1.29, 1.15 and 0.75, respectively, whilst the female OR of these Factors were 1.40, 1.25, 1.14 and 0.72, respectively. According to the relative importance of inputs of BPNN analysis in China in 1990’s, the order of priority were Nitrogen-Factor, Population density-Factor, Economic-Vegetable-Factor and Evaporation-Factor. The prediction accuracies of EC-SMR by BPNN were 90% in male and 87% in female.
Therefore, in China, there were common epidemic factors of EC in the people of different genders and times.
Conclusion:
The related factors of mortality rate of EC in Henan Province
In Henan Province, the more hog population and meat yield per hectare of the arable land, the lower humidity and average annual rainfall might be the risk factors of EC. But, the bigger proportion of vegetable land may be the protective factors. The environmental pollution of animal excrement may be one of important reasons for the regional difference of EC-SMR.
Factors related to EC mortality rate and its geographic aggregation in China
1. The more hog population and meat yield per hectare of the arable land, the greater aridity and the more usage of chemical fertilizer and pesticide might be the risk factors of EC. The environmental pollution of animal excrement may be one of important reasons for the regional difference of EC-SMR in China.
2. The higher the relative humidity, the economic level and the ratio of the vegetable field in arable land might be the protective factors.
3. Economic level, population density, the proportion of vegetable field and arable land, evaporation, average annual precipitation, relative humidity and Drought-Waterlogged Index might cause the geographic aggregation of EC in China.
The result supports the view that agricultural fertilizer and sewage under special condition of weather are the sources of the precursors of N-nitrosamine which might synthesize N-nitroso compounds in vivo to cause EC in China.
引文
[1]林东昕.中国食管癌分子流行病学研究[J].中华流行病学杂志,2003,24(10):939-943.
[2]蔡鹤生,刘存富,周爱国.饮用水NO 3污染与食管癌发病率及死亡率相关性探讨[J].地质科技情报, 2002,21(1):91-94.
[3]徐致祥,谭家驹,陈凤兰.农家肥料污染水源诱发鸡咽食管癌胃癌和肝癌[J].中华肿瘤杂志,2003 ,25(4):344-347.
[4] Ellen G ,Schuller PL, Bmijn E, et al. Volatile N-nitrosamines: nitrate and nitrite in urine and saliva of healthy volunteers after administration of large amounts of nitrate[J]. IARC Sci Publ, 1982, (41):365-378.
[5] Tenovuo J. the biochemistry of nitrates, nitrites, nitrtosamines and other potential carcinogens in human saliva[J]. Oral Pathol, 1986, 15(6):303-307.
[6]杨琰,蔡鹤生,刘存富.林州—安阳地区食管癌高发区地水氮污染及防治措施[J].水文地质工程地质,2002,6 :8-11.
[7] Lin K, Shen W, Shen Z, et al. Estimation of the potential for nitrosamine and its inhibition in subjects from high- and low-risk areas for esophageal cancer in southern China[J]. Int J Cancer, 2003, 107(6):891-895.
[8] Lin K, Shen W, Shen Z, et al. Dietary exposure and urinary excretion of total N-nitro’s compounds, nitrosamine acids and volatile nitrosamine in inhabitants of high- and low-risk areas for esophageal cancer in southern China[J]. Int J Cancer, 2002, 102(3):207-211.
[9] Lin K, Shen ZY, Lu SH,et al. Intake of Volatile N-nitrosamines and Their Ability to Exogenously Synthesize in the Diet of the Inhabitants from High-Risk Area of Esophageal Cancer in Southern China[J]. Biomed environ sci, 2002, 15(4):277-282.
[10]林昆,于世江,张建军.等.食物中总亚硝基化合物及其相关危险因素研究[J].卫生研究, 2005,34(3): 350-352.
[11]陆士新,林培中,王国清.食管癌的综合防治研究进展[J].中国肿瘤, 1999, 8(9):407-408
[12]韩建英,连士勇,徐致祥.改水对林州市食管癌发病率和死亡率的影响[J].环境与健康杂志, 2005, 22(3):200-202.
[13]徐志祥.农肥、污水与食管癌[M].北京:科学出版社,2003,1版,p28.
[14] Ahuja LR, Rojas KW, Hanson JD, et al. Root Zone Water Quality Model: ModelingManagement Effects on Water Quality and Crop Production [M]. USA: Water Resources, 1999, Publication: 372.
[15] Suzuki H, Iijima K, Scobie G. McCall Nitrate and nitrosamine chemistry within Barrett’s esophagus during acid reflux [J].Gut, 2005, 54: 1527–1535.
[16]方正祥,张水华,高建华.一些果蔬对活性自由基和亚硝酸盐的清除作用[J].食品与发酵工业,1995,(2):31-35.
[17]王方浩,马文奇,窦争霞.中国畜禽粪便产生量估算及环境效应[J].中国环境科学,2006,26(5):614-617.
[18]段诚中.养猪业生态环境存在的问题及建议[J].养猪,2006,6,61-62.
[19]杨文燕,宋长春,张金波.烧荒对土壤性质的影响[J].水土保持研究,2004,11(2):187-188.
[20] Hughes R F, Kauffman J B, Cummings D L. Dynamics of above ground and soil carbon and nitrogen stocks and cycling of available nitrogen along a land-use Gradient in Rondonia, Brazil [J]. Ecosysteims, 2002, 5(6): 244- 259.
[21]张明峰.世界农业污染及其防治概况[J].家畜生态,1995,16(3):45-49.
[22]乔光建,张均玲,唐俊智.地下水氮污染机理分析及治理措施[J].水资源保护,2004,3:9-12.
[23]孟祥琴,乔光建,鲍其钢.长期施用化肥对地下水水质的影响[J].南水北调与水利科技,2006,4:55-57.
[24]刁兴才,蔺青,杨荣光.小麦—土壤氮素循环的研究进展[J].安徽农业科学,2006,34(12):2806- 2808.
[25] Bardbury N J, Whitmore A P, Hart P B S, et al. Modeling the fate of nitrogen in crop and soil in the years following application of 15N-labelled fertilizer to winter wheat. J Agric Sci , 1993, 121:363-379.
[26] Dutt R G, Shaffer M J, Moore W J. Computer simulation model of dynamic bio-physicochemical processes in soils. Arizona Agric Exp Sta Technol Bull, 1972, 128-96
[27] Tisdale S L, Melson W L.肥力与肥料[M].金继运,刘荣乐,译.北京:中国农业科技出版社,1998. 288-291.
[28]徐德忠,李良寿,张秀玲.林州市食管癌病因及其变化的一次病例对照研究[J].中华流行病学杂志,2002,23(4):320-321.
[29]金丕焕.医用统计方法[M].上海:上海医科大学出版社,1993,1版,p189.
[30]金丕焕.医用统计方法[M].上海:上海医科大学出版社,1993,1版,p195.
[31]杨艳芳,李会庆,周英智.肥城市食管癌死亡率负二项拟合回归分析方[J].中国医院统计,2007,23(8):906-909.
[32]余松林.混合线型模型的应用[J].中国公共卫生,2003,13(1):70-76.
[33]岳勇,田考聪,汪洋. BP神经网络在结核疑似病患者就诊延迟影响因素分析中的应用探讨[J].中国公共卫生,2007,24(6):590-593.
[1] Tenovuo J, The biochemistry of nitrates, nitrites, nitrosamines and other potential carcinogens in human saliva [J]. Oral Pathol, 1986, 15(6): 303 -307.
[2]刁兴才,蔺青,杨荣光.小麦—土壤氮素循环的研究进展[J].安徽农业科学,2006,34(12):2806- 2808.
[3]黄树辉,曾光辉.土壤裂缝产生过程中双氢胺对氮溶质运移的影响[J].水土保持学报,2006,20(3):25 - 27.
[4]李向心,武周虎,孔德玉.人工湿地污水处理研究与进展[J].青岛建筑工程学院学报,2004,25 (4):56 - 60.
[5]徐致祥,谭家驹,陈凤兰.农家肥料污染水源诱发鸡咽食管癌胃癌和肝癌[J].中华肿瘤杂志,2003 ,25(4):344 - 347.
[6]段诚中.养猪业生态环境存在的问题及建议[J].养猪,2006,6,61-62.
[7]张明峰.世界农业污染及其防治概况[J].家畜生态,1995,16(3):45-49.
[8]张秀兰,候浚,梁索元.磁县食管癌防治区饮用水污染[J].肿瘤防治研究,1998,25(6):494 - 496.
[9]徐志祥.农肥、污水与食管癌[M].北京:科学出版社, 2003,1版,p28.
[10]方浩,马文奇,窦争霞.中国畜禽粪便产生量估算及环境效应[J].中国环境科2006,26(5):614-617.
[11]杨文燕,宋长春,张金波.烧荒对土壤性质的影响[J].水土保持研究,2004,11(2 ):187 - 188.
[12]陈玉成,李章平.城市生活垃圾渗沥水的污染及其垒过程控制[J].环境科学动态,1995,4:15 -18.
[13]蔡鹤生,刘存富,周爱国.饮用水NO 3污染与食管癌发病率及死亡率相关性探讨[J].地质科技情报, 2002,21(1):91 - 94.
[14]万厚民,张应丽,王震宇.亚硝基化合物与消化道肿瘤研究进展[J].医学综述,2004,10(3):161 -162.
[15]蔡鹤生,刘存富,周爱国.饮用水NO3-污染与食管癌发病率及死亡率相关性探讨[J].地质科技情报,2002,21(1):91 - 94.
[16]陆建邦,连士勇,孙喜斌.林州食管癌发病因素病例对照研究[J].中华流行病学杂志,2000,21(6):434-436.
[17] Destefani E, Deneo P H, Ron A L, etal. Food groups and risk of squamous cell carcinoma of the esophagus: a case-control study in Uruguay [J]. Br J Cancer, 2003, 89(7): 1209 - 1214
[18] Lin K, Shen W, Shen Z, Wu Y, etal. Dietary exposure and urinary excretion of total N-nitroso compounds, nitrosaminoacids and volatile nitrosamine in inhabitants of high- and low-risk areas for esophageal cancer in southern China[J]. Int-J-Cancer, 2002, 102(3): 207 - 211.
[19]刘桂亭,刘天舟.河南食管癌病因学研究[J].中国肿瘤,2000,9(5):220.
[20]林昆,沈文英,吴永宁.南方食管癌高低发区人群尿总N-亚硝基化合物与N-亚硝基氨基酸关系研究[J].卫生研究,2002,31(5):372 - 374.
[21] Suzuki H, iijima K, scobie G, etal. Nitrate and nitrosative chemistry within Barrett’s esophagus during acid reflux [J]. Gut, 2005, 54: 1527 - 1535.
[22] Lin K, Shen W, Shen Z, etal. Estimation of the potential for nitrosation and its inhibition in subjects from high- and low-risk areas for esophageal cancer in southern China [J]. Int J Cancer, 2003, 107(6): 891 - 895.
[2]蔡鹤生,刘存富,周爱国.饮用水NO 3污染与食管癌发病率及死亡率相关性探讨[J].地质科技情报, 2002,21(1):91-94.
[3]徐致祥,谭家驹,陈凤兰.农家肥料污染水源诱发鸡咽食管癌胃癌和肝癌[J].中华肿瘤杂志,2003 ,25(4):344-347.
[4] Ellen G ,Schuller PL, Bmijn E, et al. Volatile N-nitrosamines: nitrate and nitrite in urine and saliva of healthy volunteers after administration of large amounts of nitrate[J]. IARC Sci Publ, 1982, (41):365-378.
[5] Tenovuo J. the biochemistry of nitrates, nitrites, nitrtosamines and other potential carcinogens in human saliva[J]. Oral Pathol, 1986, 15(6):303-307.
[6]杨琰,蔡鹤生,刘存富.林州—安阳地区食管癌高发区地水氮污染及防治措施[J].水文地质工程地质,2002,6 :8-11.
[7] Lin K, Shen W, Shen Z, et al. Estimation of the potential for nitrosamine and its inhibition in subjects from high- and low-risk areas for esophageal cancer in southern China[J]. Int J Cancer, 2003, 107(6):891-895.
[8] Lin K, Shen W, Shen Z, et al. Dietary exposure and urinary excretion of total N-nitro’s compounds, nitrosamine acids and volatile nitrosamine in inhabitants of high- and low-risk areas for esophageal cancer in southern China[J]. Int J Cancer, 2002, 102(3):207-211.
[9] Lin K, Shen ZY, Lu SH,et al. Intake of Volatile N-nitrosamines and Their Ability to Exogenously Synthesize in the Diet of the Inhabitants from High-Risk Area of Esophageal Cancer in Southern China[J]. Biomed environ sci, 2002, 15(4):277-282.
[10]林昆,于世江,张建军.等.食物中总亚硝基化合物及其相关危险因素研究[J].卫生研究, 2005,34(3): 350-352.
[11]陆士新,林培中,王国清.食管癌的综合防治研究进展[J].中国肿瘤, 1999, 8(9):407-408
[12]韩建英,连士勇,徐致祥.改水对林州市食管癌发病率和死亡率的影响[J].环境与健康杂志, 2005, 22(3):200-202.
[13]徐志祥.农肥、污水与食管癌[M].北京:科学出版社,2003,1版,p28.
[14] Ahuja LR, Rojas KW, Hanson JD, et al. Root Zone Water Quality Model: ModelingManagement Effects on Water Quality and Crop Production [M]. USA: Water Resources, 1999, Publication: 372.
[15] Suzuki H, Iijima K, Scobie G. McCall Nitrate and nitrosamine chemistry within Barrett’s esophagus during acid reflux [J].Gut, 2005, 54: 1527–1535.
[16]方正祥,张水华,高建华.一些果蔬对活性自由基和亚硝酸盐的清除作用[J].食品与发酵工业,1995,(2):31-35.
[17]王方浩,马文奇,窦争霞.中国畜禽粪便产生量估算及环境效应[J].中国环境科学,2006,26(5):614-617.
[18]段诚中.养猪业生态环境存在的问题及建议[J].养猪,2006,6,61-62.
[19]杨文燕,宋长春,张金波.烧荒对土壤性质的影响[J].水土保持研究,2004,11(2):187-188.
[20] Hughes R F, Kauffman J B, Cummings D L. Dynamics of above ground and soil carbon and nitrogen stocks and cycling of available nitrogen along a land-use Gradient in Rondonia, Brazil [J]. Ecosysteims, 2002, 5(6): 244- 259.
[21]张明峰.世界农业污染及其防治概况[J].家畜生态,1995,16(3):45-49.
[22]乔光建,张均玲,唐俊智.地下水氮污染机理分析及治理措施[J].水资源保护,2004,3:9-12.
[23]孟祥琴,乔光建,鲍其钢.长期施用化肥对地下水水质的影响[J].南水北调与水利科技,2006,4:55-57.
[24]刁兴才,蔺青,杨荣光.小麦—土壤氮素循环的研究进展[J].安徽农业科学,2006,34(12):2806- 2808.
[25] Bardbury N J, Whitmore A P, Hart P B S, et al. Modeling the fate of nitrogen in crop and soil in the years following application of 15N-labelled fertilizer to winter wheat. J Agric Sci , 1993, 121:363-379.
[26] Dutt R G, Shaffer M J, Moore W J. Computer simulation model of dynamic bio-physicochemical processes in soils. Arizona Agric Exp Sta Technol Bull, 1972, 128-96
[27] Tisdale S L, Melson W L.肥力与肥料[M].金继运,刘荣乐,译.北京:中国农业科技出版社,1998. 288-291.
[28]徐德忠,李良寿,张秀玲.林州市食管癌病因及其变化的一次病例对照研究[J].中华流行病学杂志,2002,23(4):320-321.
[29]金丕焕.医用统计方法[M].上海:上海医科大学出版社,1993,1版,p189.
[30]金丕焕.医用统计方法[M].上海:上海医科大学出版社,1993,1版,p195.
[31]杨艳芳,李会庆,周英智.肥城市食管癌死亡率负二项拟合回归分析方[J].中国医院统计,2007,23(8):906-909.
[32]余松林.混合线型模型的应用[J].中国公共卫生,2003,13(1):70-76.
[33]岳勇,田考聪,汪洋. BP神经网络在结核疑似病患者就诊延迟影响因素分析中的应用探讨[J].中国公共卫生,2007,24(6):590-593.
[1] Tenovuo J, The biochemistry of nitrates, nitrites, nitrosamines and other potential carcinogens in human saliva [J]. Oral Pathol, 1986, 15(6): 303 -307.
[2]刁兴才,蔺青,杨荣光.小麦—土壤氮素循环的研究进展[J].安徽农业科学,2006,34(12):2806- 2808.
[3]黄树辉,曾光辉.土壤裂缝产生过程中双氢胺对氮溶质运移的影响[J].水土保持学报,2006,20(3):25 - 27.
[4]李向心,武周虎,孔德玉.人工湿地污水处理研究与进展[J].青岛建筑工程学院学报,2004,25 (4):56 - 60.
[5]徐致祥,谭家驹,陈凤兰.农家肥料污染水源诱发鸡咽食管癌胃癌和肝癌[J].中华肿瘤杂志,2003 ,25(4):344 - 347.
[6]段诚中.养猪业生态环境存在的问题及建议[J].养猪,2006,6,61-62.
[7]张明峰.世界农业污染及其防治概况[J].家畜生态,1995,16(3):45-49.
[8]张秀兰,候浚,梁索元.磁县食管癌防治区饮用水污染[J].肿瘤防治研究,1998,25(6):494 - 496.
[9]徐志祥.农肥、污水与食管癌[M].北京:科学出版社, 2003,1版,p28.
[10]方浩,马文奇,窦争霞.中国畜禽粪便产生量估算及环境效应[J].中国环境科2006,26(5):614-617.
[11]杨文燕,宋长春,张金波.烧荒对土壤性质的影响[J].水土保持研究,2004,11(2 ):187 - 188.
[12]陈玉成,李章平.城市生活垃圾渗沥水的污染及其垒过程控制[J].环境科学动态,1995,4:15 -18.
[13]蔡鹤生,刘存富,周爱国.饮用水NO 3污染与食管癌发病率及死亡率相关性探讨[J].地质科技情报, 2002,21(1):91 - 94.
[14]万厚民,张应丽,王震宇.亚硝基化合物与消化道肿瘤研究进展[J].医学综述,2004,10(3):161 -162.
[15]蔡鹤生,刘存富,周爱国.饮用水NO3-污染与食管癌发病率及死亡率相关性探讨[J].地质科技情报,2002,21(1):91 - 94.
[16]陆建邦,连士勇,孙喜斌.林州食管癌发病因素病例对照研究[J].中华流行病学杂志,2000,21(6):434-436.
[17] Destefani E, Deneo P H, Ron A L, etal. Food groups and risk of squamous cell carcinoma of the esophagus: a case-control study in Uruguay [J]. Br J Cancer, 2003, 89(7): 1209 - 1214
[18] Lin K, Shen W, Shen Z, Wu Y, etal. Dietary exposure and urinary excretion of total N-nitroso compounds, nitrosaminoacids and volatile nitrosamine in inhabitants of high- and low-risk areas for esophageal cancer in southern China[J]. Int-J-Cancer, 2002, 102(3): 207 - 211.
[19]刘桂亭,刘天舟.河南食管癌病因学研究[J].中国肿瘤,2000,9(5):220.
[20]林昆,沈文英,吴永宁.南方食管癌高低发区人群尿总N-亚硝基化合物与N-亚硝基氨基酸关系研究[J].卫生研究,2002,31(5):372 - 374.
[21] Suzuki H, iijima K, scobie G, etal. Nitrate and nitrosative chemistry within Barrett’s esophagus during acid reflux [J]. Gut, 2005, 54: 1527 - 1535.
[22] Lin K, Shen W, Shen Z, etal. Estimation of the potential for nitrosation and its inhibition in subjects from high- and low-risk areas for esophageal cancer in southern China [J]. Int J Cancer, 2003, 107(6): 891 - 895.