Fish sauce and gastric cancer: an ecological study in Fujian Province, China

Abstract

AIM: To explore the relationship between consumption of fish sauce and the risk of gastric cancer in Fujian Province.

METHODS: An ecological study was carried out. A total of 11000 subjects from 55 townships were randomly selected from 10 counties within Fujian Province. All subjects were local residents who had been living in Fujian Province for more than 20 years, within the age group of 45-74 years. Trained interviewers conducted face-to-face interviews with a standardized questionnaire, which covered the frequency and amount of food intake, dietary habit, tobacco and alcohol consumption and history of chronic gastric diseases. Univariate and multivariate analyses were performed using Epi-info and SAS statistical packages, respectively.

RESULTS: A significant correlation between monthly consumption of fish sauce and mortality of gastric cancer was found. Pearson’s coefficient of correlation was statistically significant with r = 0.7356 for males, r = 0.5246 for females (P < 0.01). In the multivariate analysis, consumption of fish sauce still showed an association with the risk of gastric cancer. No significant positive correlation between esophagus cancer, liver cancer, colon cancer and consumption of fish sauce were observed.

CONCLUSION: Long-term intake of fish sauce may be related to high mortality of gastric cancer. Consumption of fish sauce might be one of important and unique etiologic factors of gastric cancer in Fujian Province. Further studies are needed to confirm this ecological study.

Key Words: stomach neoplasm/etiology, food habits, risk factors, nitrosamines, epidemiology, fish products, fish sauce, carcinogens

INTRODUCTION

Fujian Province is a high-risk area of gastric cancer in China[1]. The standardized mortalities of gastric cancer during 1986-1988 were 37-72 per 100000 for men and 14-90 per 100000 for women, accounting for 25.31% of total cancer mortality. In certain areas, the annual incidence of gastric cancer is more than 50 per 100000 persons[2]. The health of general population in Fujian Province is most seriously threatened by gastric cancer.

A number of potential risk factors for gastric cancer have been examined in previous studies[3]. Among these, the dietary hypotheses are of particular interest[4,5]. N-nitroso compounds derived from the consumption of preserved food may be associated with the risk of the disease[6].

In the investigation of dietary factors, we have reported the relationship between high salt intake and gastric cancer in the high-risk area of Fujian Province[7-9]. The salted food includes fish sauce, which is particularly favored by Fujian local residents. Fish sauce is usually produced from several kinds of sea fishes after prolonged fermentation processes. Its mutagenicity has been reported by several experimental studies[10-12]. However, there have been few reports from population-based epidemiological study on the relationship between fish sauce and risk of gastric cancer. Therefore, this ecological study was carried out to explore the relationship between them in the area of Fujian Province.

MATERIALS AND METHODS

Study areas, subjects and methods

Gastric cancer is the most common cancer among the major types of cancer in Fuji an Province. However, even within the province the death rate distribution of gastric cancer varies greatly among different areas. This ecological study included 55 townships in 10 counties. The retrospective investigations of death cause, which was a part of a national program, had been concluded in these areas.

In reference to the national gastric cancer death rate of 17.30/100000 (95% confidence interval: 9.15/100000-25.45/100000, calculated by Poisson-normal-approximation method), the death rate among these 10 counties can be divided into three groups: ① High incidence area: Changle, Putian, and Fuzhou; ② Medium incidence area: Changtai, Huian, Shaxian, Sanming, Yongding, and Zhangpu; ③ Low incidence area: Fuan (Figure 1).

Figure 1 Gastric cancer mortality in 10 counties, Fujian, China, 1991.

A total of 200 subjects in each town (sampling spot) were interviewed. The 55 townships were randomly selected from 10 counties which include 11000 subjects in this study (Figure 2).

Figure 2 Not available

All subjects of this study are local resident of more than 20 years, within the age of 45-74 years, who can clearly answer the questions asked by the interviewers. Specially trained interviewers conducted face-to-face interviews according to a standardized questionnaire, which covers the frequency and amount of food intake. The investigations also include demographic factors, dietary habit, smoking and alcohol consumption, and history of chronic gastric diseases.

Statistical analysis

The database was established using Epi-info, and the gastric cancer mortality for 45-74 age group was calculated based on world truncated standardized population constitution. The correlation coefficient and multiple regression analysis were carried out using SAS software.

RESULTS

Univariate analysis

The highest Pearson’s correlation coefficient was found between gastric cancer mortality and average monthly consumption of uncooked fish sauce. Similar results were observed for both males and females, with Pearson’s correlation coefficient for males of 0.7356 and 0.5246 for females (P < 0.01) (Table 1). Positive correlation were found between gastric death rate and fish sauce-related variables (Fish sauce consumption, population ratio who consume fish sauce, population ratio who consume uncooked fish sauce, and uncooked fish sauce consumption). Further, compared with other factors, larger Pearson’s correlation coefficients were observed for all 4 variables (χ¹-χ⁴). Smoking index: (amount of smoking/day × years of smoking)/age of starting smoking.

Table 1 Correlation analysis between exposure factors and standardized truncated mortality of gastric cancer (55 townships).

Factors		Pearson’s coefficient of correlation
		Male	Female
χ1	Fish sauce average consumption (0.5 kg/month)	0.5170a	0.4261a
χ2	Population ratio who consume fish sauce	0.3576a	0.2720a
χ3	Population ratio who consume uncooked fish sauce	0.6904a	0.5086a
χ4	Uncooked Fish sauce average consumption (0.5 kg/month)	0.7356a	0.5246a
χ5	Salted fermented sea foods average consumption (0.5 kg/year)	0.4863a	0.3653a
χ6	Population ratio of crapulence	0.4887a	0.3198a
χ7	Population ratio of taking moldy foods	0.5538a	0.2582
χ8	Fresh vegetables consumption (0.5 kg/year)	-0.4742a	-0.2307
χ9	Fresh meat, fish, egg average consumption (0.5 kg/year)	0.4254a	0.1298
χ10	Average ratio of coarse food grain to grain ration	0.3900a	0.2107
χ11	Prevalence of chronic gastric diseases	0.2973b	0.1107
χ12	Average index of smoking	-0.2158	0.1737
χ13	Average cumulative alcohol consumption (0.05 kg)	0.1703	-0.1942
χ14	Population ratio of taking irregular dinners	0.2172	0.1443
χ15	Average salted vegetables consumption (0.5 kg/year)	-0.1488	-0.0079
χ16	average salt consumption (0.5 kg/month)	-0.2455	-0.2160
χ17	Average fresh fruit consumption (0.5 kg/year)	0.0978	-0.0581
χ18	Average bean products consumption (0.5 kg/year)	-0.1475	-0.0422

^aP < 0.01,

^bP < 0.05.

Gastric cancer mortality for 45-74 age group was calculated based on world truncated standardized population constitution.

Multi-variables analysis

Considering the possible confounding effects among factors, multivariate regression analyses were performed on other factors (χ⁵-χ¹⁸) and each of fish sauce variables (χ¹-χ⁴) respectively. MAXR method from REG of SAS package was used. The optimum subset of variables was established which also includes fish sauce as one of variables. The consumption of fish sauce still shows association with the risk of gastric cancer in analysis after controlling the possible confounding factors (Table 2, Table 3, Table 4, Table 5). The corresponding multiple correlation coefficients indicate that the goodness of fit tests was significant.

Table 2 The results of multivariate regression analysis on χ¹, χ⁵-χ¹⁸ and mortality of gastric cancer for men.

Factor	β	Sβ	Standard error	t	P
χ1	187.3639	0.4678	43.1674	4.340	0.0001
χ13	0.0306	0.3105	0.0101	3.018	0.0041
χ11	861.7824	0.5713	241.9891	3.561	0.0008
χ7	338.2060	0.1819	160.7826	2.103	0.0407
χ14	421.5166	0.3329	132.5871	3.179	0.0026
χ8	-0.8042	-0.6127	0.2250	-3.574	0.0008
χ17	-2.6339	-0.2527	1.1036	-2.387	0.0210
Source of variation	DF	Sum of squares	Mean Square	F value	P
Regression	7	4802841.3249	686120.1893	47.482	0.0001
Error	48	693599.3209	14449.9859
Total	55	5496440.6458

R² = 0.8738 Adjusted R² = 0.8554

Table 3 The results of multivariate regression analysis on χ⁴, χ⁵-χ¹⁸ and mortality of gastric cancer for men.

Factor	β	Sβ	Standard error	t	P
χ4	271.5961	0.4491	45.4883	5.971	0.0001
χ13	0.0248	0.2521	0.0092	2.696	0.0096
χ11	671.6803	0.4453	223.1106	3.011	0.0041
χ7	285.9537	0.1538	143.8708	1.988	0.0526
χ14	366.5093	0.2894	119.2315	3.074	0.0035
χ8	-0.5394	-0.4109	0.2124	-2.539	0.0144
χ17	-1.2599	-0.1209	0.8569	-1.470	0.1480
Source of variation	DF	Sum of squares	Mean Square	F value	P
Regression	7	4942225.6156	706052.2308	61.149	0.0001
Error	48	554215.0301	11546.1465
Total	55	5496440.6456

R² = 0.8992 Adjusted R² = 0.8845

Table 4 The results of multivariate regression analysis on χ¹, χ⁵-χ¹⁸ and mortality of gastric cancer for women.

Factor	β	Sβ	Standard error	t	P
χ1	81.2327	0.5020	24.7179	3.286	0.0019
χ11	417.1922	0.43288	122.1212	3.416	0.0013
χ14	132.1827	0.25882	44.0834	2.998	0.0015
χ5	1.2702	0.1639	0.9209	1.379	0.1739
χ17	-1.8558	-0.4045	0.5958	-3.115	0.0030
Source of variation	DF	Sum of squares	Mean Square	F value	P
Regression	5	646897.4448	129379.4890	34.871	0.0001
Error	50	185511.7385	3710.2348
Total	55	832409.1833

R² = 0.7771 Adjusted R² = 0.7549

Table 5 The results of multivariate regression analysis on χ⁴, χ⁵-χ¹⁸ and mortality of gastric cancer for women.

Factor	β	Sβ	Standard error	t	P
χ4	100.2382	0.4090	26.0132	3.853	0.0003
χ11	448.4859	0.4658	124.4522	3.604	0.0007
χ14	153.9328	0.3007	43.5068	3.538	0.0009
χ15	2.3826	0.3074	0.9565	2.491	0.0161
χ18	-0.9343	-0.4078	0.3853	-2.425	0.0190
Source of variation	DF	Sum of squares	Mean Square	F value	P
Regression	5	657252.0786	31450.4157	37.52	0.0001
Error	50	175157.1047	3503.1421
Total	55	832409.1833

R² = 0.7896 Adjusted R² = 0.7685

Correlation analysis between exposure factor and other digestive tract cancers

The Pearson’s correlation coefficients between each factor and truncated standardized mortality for esophagus cancer, liver cancer, and colon cancer were calculated separately and results are shown in Table 6, Table 7, Table 8. There are no significant positive correlation among 4 variables for fish sauce consumption and any of the mortality rates among three other digestive tract cancers. In multivariate regression analysis, the variables that reflect fish sauce consumption were not selected into the model when analyzing relationship between fish sauce and mortality of the other three cancers.

Table 6 Correlation analysis between exposure factors and truncate d standardized mortality of esophagus cancer.

Factors	Male		Female
	Pearson’s r	P	Pearson’s r	P
χ1	-0.2511	0.0644	-0.3079	0.0222
χ2	-0.2538	0.0615	-0.3647	0.0062
χ3	-0.2170	0.1115	-0.2519	0.0636
χ4	-0.1714	0.2109	-0.2218	0.1037
χ5	-0.0390	0.7778	-0.0988	0.4730
χ6	-0.0715	0.6042	0.0360	0.7939
χ7	0.3025	0.0248	0.2130	0.1184
χ8	-0.3618	0.0066	-0.2274	0.0950
χ9	0.0609	0.6590	-0.1751	0.2010
χ10	0.3571	0.0074	0.1731	0.2062
χ11	-0.3493	0.0090	0.1252	0.3625
χ12	0.5330	0.0001	0.0592	0.6677
χ13	-0.0686	0.6187	0.3725	0.0051
χ14	0.3292	0.0141	0.2771	0.0406
χ15	0.2606	0.0546	0.4740	0.0003
χ16	0.6384	0.0001	0.3743	0.0049
χ17	-0.2569	0.0584	-0.2759	0.0415

Esophagus cancer mortality for 45-74 age group was calculated based on world truncated standardized population constitution.

Table 7 Correlation analysis between exposure factors and truncated standardized mortality of liver cancer.

Factors	Male		Female
	Pearson’s r	P	Pearson’s r	P
χ1	-0.0913	0.5074	-0.1407	0.3055
χ2	-0.1420	0.3010	-0.1808	0.1865
χ3	-0.0237	0.8639	-0.1017	0.4599
χ4	0.0061	0.9648	-0.0970	0.4813
χ5	0.2226	0.1028	0.1915	0.1614
χ6	0.0679	0.6221	0.2546	0.0607
χ7	0.4113	0.0018	0.2141	0.1166
χ8	-0.3280	0.0145	-0.1681	0.2199
χ9	0.1200	0.3830	0.0737	0.5927
χ10	0.3627	0.0065	0.1609	0.2407
χ11	0.0570	0.6794	0.1594	0.2451
χ12	0.1352	0.3249	0.0595	0.6661
χ13	-0.0120	0.9412	-0.2453	0.0710
χ14	-0.0024	0.9863	0.1569	0.2527
χ15	0.0630	0.6477	0.2936	0.0296
χ16	0.1967	0.1501	0.0946	0.4922
χ17	-0.1065	0.4389	-0.0053	0.9693

Liver cancer mortality for 45-74 age group was calculated based on world truncated standardized population constitution.

Table 8 Correlation analysis between exposure factors and truncated standardized mortality of colon cancer.

Factors	Male		Female
	Pearson’s r	P	Pearson’s r	P
χ1	0.2461	0.0701	0.1077	0.4341
χ2	0.2356	0.0833	0.1055	0.4433
χ3	0.2428	0.0741	0.0535	0.6982
χ4	0.2250	0.0986	0.0103	0.9408
χ5	0.0035	0.9796	-0.0223	0.8718
χ6	-0.0504	0.7148	-0.0127	0.9266
χ7	0.0964	0.4838	-0.1242	0.3662
χ8	-0.1605	0.2418	0.1274	0.3539
χ9	0.0833	0.5215	0.0652	0.6362
χ10	-0.0974	0.4915	-0.2482	0.0677
χ11	0.3843	0.0038	0.2251	0.0985
χ12	-0.2815	0.0374	0.0277	0.8412
χ13	-0.1895	0.1659	-0.1819	0.1839
χ14	0.2185	0.1091	-0.0611	0.6579
χ15	-0.2899	0.0318	-0.0554	0.6879
χ16	-0.1654	0.2274	-0.2247	0.0990
χ17	0.2053	0.1327	0.1787	0.1919

Colon cancer mortality for 45-74 age group was calculated based on world truncated standardized population constitution.

DISCUSSION

Gastric cancer is the second common cancer in the world today[13]. It remains the leading cause of cancer death in China[14]. In the studies of etiology and epidemiology on gastric cancer, the associations between gastric cancer risk and Helicobacter pylori (Hp) infection[15-18], occupational exposures[19], diet and life factors[20-25], lack of essential trace elements[26] have been observed. The most widely reported protective factors are dietary factor[27-33]. It has been reported that a large amount of nitrite exists in most salted fishes and smoking meat, and may be related to the high gastric cancer incidence along coastal area and among Japanese or Chinese population.

Each local resident in Fujian Province consumed daily about 30 mL fish sauce as one kind of condiments. Fish sauce is a liquid product of small marine fish and table salt (7:3). The fishes are completely liquefied after fermentation for 1-2 years. The salted fermented fish products may contain many precursors of N-nitroso compound derived from the proteins content of the fishes, which could synthesize N-nitrosamides under simulated human stomach conditions[34]. N-nitrosamides is suggested as a major initial cause of gastric cancer[35]. Zhang et al[36] had analyzed the N-nitroso compound precursors from 49 fish sauce samples obtained from high risk area of gastric cancer in Fujian Province, and detected 630 μM/mL of 17 free amino acids. After nitrosifiction the total amount of N-nitroso compound in samples was as high as 2.95 μM/mL. The concentration of N-nitrosylamine from ethyl acetate extraction solution was 0.06 μM/mL. Fish sauce is rich in creatinine and other nitrosamide precursors. It was mutagenic and carcinogenic for the glandular stomach of Wistar rats, after it was nitrosified under simulated human stomach condition[37]. All these experiments have shown that fish sauce is potentially carcinogenic.

High incidence of gastric cancer may be attributed, to some extent, to some unique dietary habits in the high-risk area. This ecological study showed the positive correlationship between gastric cancer mortality and fish sauce favored by Fujian residents. The relationship between gastric cancer and uncooked fish sauce intake is particularly obvious. Among four digestive tract cancers, only gastric cancer had a positive correlation with fish sauce consumption, which suggests that the association may be specific to gastric cancer. This study indicates that long-term use of fish sauce may be related to high gastric cancer mortality in the Fujian area.

This ecological comparison study may provide a possible hypothesis for the etiology of gastric cancer. Fish sauce has both N-nitroso compounds and a high salt content, which might partially explain the high gastric cancer mortality along the Fujian coast. Further studies are needed to confirm our results. Recently, a population-based case-control study was conducted in Changle County, Fujian Province. The residents in Changle County had a high prevalence of H. pylori infection. However, no statistical significant difference was found in respect to presence of H. pylori infection between gastric cancer patients and controls. The results showed that fish sauce intake, deficiency in fresh vegetables were risk factors in gastric cancer[38].

The development of human gastric cancer is a multistep and multifactorial process[39]. A number of molecular events are involved in gastric carcinogenesis[40]. Coordinate prevention and treatment measures must be taken, including changing the habit of taking fish sauce, nutrition guidance and cancer prevention education for general public. This would undoubtedly have positive effect on reducing the risk of gastric cancer in Fujian Province.