• cohort studies;
  • diet;
  • dietary carbohydrates;
  • epidemiology;
  • glycemic index;
  • glycemic load;
  • gastric cancer;
  • stomach cancer;
  • prospective studies


  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. References

The glycemic effects of diets high in refined grains and starchy foods might increase stomach cancer risk by affecting circulating glucose, insulin and insulin-like growth factor-I levels. No prospective data on the role of high glycemic load and glycemic index diets on stomach cancer risk have been reported. We therefore prospectively investigated dietary glycemic load, overall glycemic index and carbohydrate intake in relation to the incidence of stomach cancer among 61,433 women in the population-based Swedish Mammography Cohort. Diet was assessed at baseline (1987–1990) and again in 1997. During 903,586 person-years of follow-up, a total of 156 incident cases of stomach cancer were ascertained. We observed no material associations of dietary glycemic load, overall glycemic index and total carbohydrate intake with the risk of stomach cancer. The multivariate hazard ratios for the highest versus the lowest quintile were 0.76 (95% CI = 0.46–1.25) for glycemic load, 0.77 (95% CI = 0.46–1.30) for overall glycemic index and 0.85 (95% CI = 0.50–1.43) for carbohydrate intake. The associations did not vary according to body mass index. Lack of information on Helicobacter pylori infection status did not allow stratification by this potential effect modifier. Findings from this population-based prospective cohort of middle-aged and elderly women did not provide evidence of a positive association between glycemic load, glycemic index and carbohydrate intake with risk of stomach cancer. © 2006 Wiley-Liss, Inc.

Diet has been recognized as having a major role in the development of stomach cancer.1 In this regard, several case-control studies have found an increased risk of stomach cancer associated with high consumption of refined grains and starchy foods.1, 2, 3, 4, 5, 6 Most refined-grain products and starchy foods are rapidly digested in the gut and have a high glycemic index.7 The glycemic index is a measure used to rank carbohydrate-containing foods by their glycemic response and, hence, their effects on blood insulin concentrations.8, 9 Insulin has mitogenic properties and could stimulate tumor development by increasing bioactive insulin-like growth factor-I (IGF-I), which in turn stimulate cell proliferation and inhibit apoptosis.10 IGF-I has been linked to increased mitogenesis in gastric cancer cell lines,11 and higher IGF-I concentrations have been observed in patients with gastric cancer when compared with the healthy controls.12 In addition, a recent cohort study reported a direct relationship between fasting plasma glucose concentrations and risk of stomach cancer among Helicobacter (H.) pylori-seropositive subjects.13

To date, only 1 previous study14 has investigated glycemic index and dietary glycemic load (i.e., the product of the glycemic index of a specific food and its carbohydrate content) in relation to risk of stomach cancer. In that case-control study,14 a significant almost 2-fold increased risk of stomach cancer was observed when the top and bottom quartiles of dietary glycemic load were compared, but no association was observed with glycemic index. Because no prospective data are available, we investigated prospectively dietary glycemic load, overall glycemic index and total carbohydrate intake in relation to the incidence of stomach cancer in the Swedish Mammography Cohort with 18 years of follow-up and repeated measures of diet.

Material and methods

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. References

Study population

The present investigation is based on data from the Swedish Mammography Cohort (SMC), a prospective population-based cohort study established between 1987 and 1990.15 The source population consisted of all women who were born between 1914 and 1948 and were residents of Uppsala or Västmanland County in central Sweden from 1987 to 1990. Of 90,303 eligible women, 66,651 (74%) returned a completed questionnaire concerning diet, weight, height and education. In the autumn of 1997, a second questionnaire was sent to all participants who were still alive and residing in the study area to update information on diet and to ascertain data on other lifestyle factors.

From the initial cohort, we excluded women with incorrect or missing national registration number, those with implausible values for total energy intake (i.e., 3 standard deviations from the mean value for loge-transformed energy), and those with a cancer diagnosis (except nonmelanoma skin cancer) prior to baseline. This left 61,433 women eligible for analysis. This investigation has been approved by the regional ethics committee at the Karolinska Institutet (Stockholm, Sweden).

Dietary assessment

Diet was assessed with food-frequency questionnaires administered at baseline (including 67 food items) and in 1997 (96 food items). In these questionnaires, participants reported their average frequency of consumption of each food item during the previous year. We computed dietary glycemic load by multiplying the carbohydrate content of each food by its glycemic index, and then multiplying the product by the frequency of consumption and summing values from all foods. Glycemic index values of foods were obtained from international tables.7 The overall glycemic index value for each participant was calculated by dividing the participant's dietary glycemic load by the total carbohydrate intake. This variable represents the overall quality of the carbohydrates in the diet. In a subsample of 129 women from the cohort, the correlation coefficient between the average intakes assessed by 4 1-week weighted diet records (collected three to four months apart) and the baseline dietary questionnaire was 0.53 for total carbohydrate. High validity was observed for total carbohydrate intake as assessed by the 1997 dietary questionnaire (correlation coefficient, 0.73).16

Assessment of nondietary factors

Information on age, height, weight and education was obtained on the baseline and the 1997 questionnaires. We calculated body mass index (BMI) as weight in kilograms divided by the square of height in meters. Women reported on cigarette smoking in the 1997 questionnaire.

Case ascertainment and follow-up

Ascertainment of cancer incidence was achieved by linkage to the practically 100% complete national and regional Swedish Cancer registers.17, 18 Cases of stomach cancer were defined as International Classification of Diseases-9 codes 151.0–151.9. We received information on the dates of death and dates of migration by linkage to the Swedish Death and Population registers at Statistics Sweden.

Statistical analysis

Each woman contributed follow-up time from the date of entry into the cohort until the date of diagnosis of stomach or any other cancer, death, migration or December 31, 2004, whichever came first. We used Cox proportional hazards models19 to estimate hazard ratios (HRs) with 95% confidence intervals (CIs) according to quintiles of dietary glycemic load, glycemic index and total carbohydrate intake. All models were stratified by age in month and the year of entry into the cohort. In multivariate models, we also controlled for education, body mass index and intakes of total energy and alcohol. We examined the proportional hazards assumption by using the likelihood ratio test and by plotting the log of the cumulative hazards function. There was no evidence that the proportional hazards assumption was violated for any of the analyses.

To better represent participant's long-term diet, we used the baseline and the 1997 dietary intake measurements. In these analyses, dietary data from the baseline questionnaire were used for follow-up from 1987 through 1997, and the mean of dietary intakes from baseline and 1997 was used for the period from 1998 through 2004. Dietary glycemic load, glycemic index and total carbohydrate intake were adjusted for total energy intake using the residual method.20

Tests of trend were conducted by assigning the median value to each quintile and modeling this value as a continuous variable. We used the SAS software (version 9.1; SAS Institute, Inc. Cary, NC) for all statistical analyses. All p-values are 2-sided.


  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. References

In our cohort of 61,433 women during 903,586 person-years of follow-up from 1987 through December 2004, there were 156 incident cases of stomach cancer. The crude incidence rate of stomach cancer was 17 cases per 100,000 person-years. When compared with women who had a low dietary glycemic load, those with higher glycemic load consumed less alcohol, fat and protein, but more carbohydrates (Table I).

Table I. Age-Standardized Baseline Characteristics According to Quintiles of Energy-Adjusted Dietary Glycemic Load Among Women in The Swedish Mammography Cohort
CharacteristicsQuintiles of dietary glycemic load
Median glycemic load157167179190207
Mean age (y)54.053.453.353.454.2
Mean BMI (kg/m2)24.924.724.724.624.6
Post-secondary education (%)11.912.813.412.912.3
Mean daily intakes
Alcohol (g)
Fat (g)58.655.753.952.549.0
Protein (g)68.167.967.366.362.6
Carbohydrate (g)218225230234246
Dietary fiber (g)24.424.324.324.424.6

We observed no significant associations between dietary glycemic load, overall glycemic index or total carbohydrate intake with risk of stomach cancer after adjusting only for age, or after further adjusting for education, body mass index and intakes of total energy and alcohol (Table II). Additional adjustment for intakes of fat, protein and dietary fiber or for intakes of fruits, vegetables and red meat did not materially change the results presented in Table II. The results also remained essentially unaltered after excluding cases diagnosed within the first 2 years of follow-up (data not shown). When we used baseline diet only, rather than updating the information on diet during follow-up, there was no association. The null associations between glycemic load, glycemic index and carbohydrate intake with risk of stomach cancer persisted across strata defined by body mass index (data not shown).

Table II. Hazard Ratios and 95% Confidence Intervals (CIs) of Stomach Cancer According to Quintiles of Long-Term Dietary Glycemic Load, Overall Glycemic Index and Total Carbohydrate Intake
 Quintile of intakeptrend1
1 (lowest)2345 (highest)
  • 1

    P values were obtained from two-sided Wald tests using the median value for each quintile of as a continuous variable.

  • 2

    Multivariate hazard ratios (HRs) are adjusted for age (in months), education (less than high school, high school graduate, or more than high school), body mass index (<23.0, 23.0–24.9, 25.0–29.9 or ≥30 kg/m2) and intakes of total energy (continuous) and alcohol (quartiles).

Dietary glycemic load
 No. of cases3738213030 
 Person-years of follow-up182,110189,249186,369175,589170,269 
 Age-adjusted HR (95% CI)1.000.98 (0.62–1.55)0.53 (0.31–0.92)0.73 (0.45–1.20)0.73 (0.44–1.20)0.13
 Multivariate HR (95% CI)21.000.98 (0.61–1.55)0.53 (0.31–0.91)0.75 (0.45–1.23)0.76 (0.46–1.25)0.16
Overall glycemic index
 No. of cases3729352827 
 Person-years of follow-up187,338187,135181,291175,631172,191 
 Age-adjusted HR (95% CI)1.000.87 (0.53–1.43)1.02 (0.63–1.64)0.81 (0.48–1.36)0.77 (0.46–1.29)0.32
 Multivariate HR (95% CI)21.000.87 (0.53–1.43)0.98 (0.61–1.59)0.80 (0.47–1.33)0.77 (0.46–1.30)0.30
Total carbohydrate
 No. of cases2927293833 
 Person-years of follow-up173,468188,778175,188178,204187,948 
 Age-adjusted HR (95% CI)1.000.72 (0.43–1.23)0.79 (0.47–1.33)0.92 (0.56–1.51)0.77 (0.46–1.29)0.81
 Multivariate HR (95% CI)21.000.76 (0.44–1.29)0.83 (0.49–1.41)0.98 (0.59–1.61)0.85 (0.50–1.43)0.93

We conducted additional analyses using the 1997 dietary questionnaire as baseline. There was no association between increasing dietary glycemic load, overall glycemic index or total carbohydrate intake and risk of stomach cancer; the multivariate HRs (adjusted for the same covariates as in the multivariate model in Table II and also for cigarette smoking) comparing the highest with the lowest tertile (tertiles were used rather than quintiles because of a small number of cases; n = 52 cases) were 0.97 (95% CI = 0.49–1.94) for glycemic load, 1.32 (95% CI = 0.63–2.77) for glycemic index and 0.97 (95% CI = 0.48–1.98) for carbohydrate intake.


  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. References

Our data from a population-based prospective cohort with repeated diet measures and up to 18 years of follow-up do not show any evidence that high glycemic load and glycemic index diets increase the risk of stomach cancer. We also observed no association with carbohydrate intake.

We are aware of only 1 previous study that has examined the relation between dietary glycemic load and glycemic index and stomach cancer risk. In that hospital-based case-control study in Italy,14 risk of stomach cancer was significantly higher among individuals with a high glycemic load (OR = 1.94; 95% CI = 1.47–2.55, for highest compared with lowest quartile). No association was observed with glycemic index,14 consistent with our findings.

Diets with high glycemic load and glycemic index are associated with a high consumption of refined carbohydrates, which are rapidly absorbed and are capable to increase blood glucose and insulin concentrations to a greater extend than slowly absorbed carbohydrates.21 A number of epidemiologic studies, mainly case-control investigations, have reported an increased risk of stomach cancer associated with high consumption of starchy foods with high glycemic index, such as white bread, rice and potatoes.1, 2, 3, 4, 5, 6 Such foods are also relatively low in antioxidants and other micronutrients that might be protective in stomach carcinogenesis.

Our study has several major strengths. The prospective nature of the study design avoided the problem of biased recall of dietary habits because diet was measured before the diagnosis of stomach cancer. The virtually complete cohort follow-up by linkage to various population-based registers minimizes the possibility that differential loss to follow-up may have affected our results. Moreover, repeated assessments of diet were used, which provide a better measure of long-term intake than a single assessment of diet at baseline. The use of cumulative average updated dietary data also reduces random within-person measurement error.22

A potential limitation of our study is that dietary data are assessed with error. Errors in the measurement of carbohydrate intake may have resulted in misclassification, which would tend to attenuate any true relationships. Furthermore, the glycemic index values of some foods are currently based on results reported in only 1 or 2 studies, and those studies often had small sample sizes.7 Thus, misclassification in our study could also be caused by random variation in the estimated glycemic index values. Another limitation to our study is an inability to examine risk by H. pylori infection status. A recent study13 observed a statistically significant positive association between plasma glucose concentrations and risk of stomach cancer only among H. pylori-seropositive subjects. Future studies should assess whether the association of glycemic load and glycemic index with stomach cancer risk varies by H. pylori infection status.

In summary, our findings from a large prospective cohort study with long-term follow-up and repeated dietary measures do not support the hypothesis that diets with high glycemic load and glycemic index increase the risk of stomach cancer in middle-aged and elderly women. We cannot exclude the possibility that high-glycemic load/index diets increase the risk of stomach cancer in specific subgroups of the population, such as those infected by H. pylori.


  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. References
  • 1
    World Cancer Research Fund, American Institute for Cancer Research. Food, Nutrition and the Prevention of Cancer: A Global Perspective. Washington DC: American Institute for Cancer Research, 1997.
  • 2
    De Stefani E, Boffetta P, Deneo-Pellegrini H, Mendilaharsu M, Carzoglio JC, Ronco A. Carbohydrates and risk of stomach cancer in Uruguay. Int J Cancer 1999; 82: 61821.
  • 3
    Jansen MC, Bueno-de-Mesquita HB, Rasanen L, Fidanza F, Menotti A, Nissinen A, Feskens EJ, Kok FJ, Kromhout D. Consumption of plant foods and stomach cancer mortality in the seven countries study. Is grain consumption a risk factor? Seven Countries Study Research Group. Nutr Cancer 1999; 34: 4955.
  • 4
    Lissowska J, Gail MH, Pee D, Groves FD, Sobin LH, Nasierowska-Guttmejer A, Sygnowska E, Zatonski W, Blot WJ, Chow WH. Diet and stomach cancer risk in Warsaw, Poland. Nutr Cancer. 2004; 48: 14959.
  • 5
    Chatenoud L, La Vecchia C, Franceschi S, Tavani A, Jacobs DR,Jr, Parpinel MT, Soler M, Negri E. Refined-cereal intake and risk of selected cancers in Italy. Am J Clin Nutr 1999; 70: 110710.
  • 6
    Mathew A, Gangadharan P, Varghese C, Nair MK. Diet and stomach cancer: a case-control study in South India. Eur J Cancer Prev 2000; 9: 8997.
  • 7
    Foster-Powell K, Holt SH, Brand-Miller JC. International table of glycemic index and glycemic load values: 2002. Am J Clin Nutr 2002; 76: 556.
  • 8
    Wolever TM, Bolognesi C. Source and amount of carbohydrate affect postprandial glucose and insulin in normal subjects. J Nutr 1996; 126: 2798806.
  • 9
    Jenkins DJ, Kendall CW, Augustin LS, Franceschi S, Hamidi M, Marchie A, Jenkins AL, Axelsen M. Glycemic index: overview of implications in health and disease. Am J Clin Nutr 2002; 76: S26673.
  • 10
    Giovannucci E. Insulin, insulin-like growth factors and colon cancer: a review of the evidence. J Nutr 2001; 131: S310920.
  • 11
    Yi HK, Hwang PH, Yang DH, Kang CW, Lee DY. Expression of the insulin-like growth factors (IGFs) and the IGF-binding proteins (IGFBPs) in human gastric cancer cells. Eur J Cancer 2001; 37: 225763.
  • 12
    Franciosi CM, Piacentini MG, Conti M, Romano F, Musco F, Caprotti R, Rovelli F, Uggeri F. IGF-1 and IGF-1BP3 in gastric adenocarcinoma. Preliminary study. Hepatogastroenterology 2003; 50: 297300.
  • 13
    Yamagata H, Kiyohara Y, Nakamura S, Kubo M, Tanizaki Y, Matsumoto T, Tanaka K, Kato I, Shirota T, Iida M. Impact of fasting plasma glucose levels on gastric cancer incidence in a general Japanese population: the Hisayama study. Diabetes Care 2005; 28: 78994.
  • 14
    Augustin LS, Gallus S, Negri E, La Vecchia C. Glycemic index, glycemic load and risk of gastric cancer. Ann Oncol 2004; 15: 5814.
  • 15
    Wolk A, Bergstrom R, Hunter D, Willett W, Ljung H, Holmberg L, Bergkvist L, Bruce A, Adami HO. A prospective study of association of monounsaturated fat and other types of fat with risk of breast cancer. Arch Intern Med 1998; 158: 415.
  • 16
    Messerer M, Johansson SE, Wolk A. The validity of questionnaire-based micronutrient intake estimates is increased by including dietary supplement use in Swedish men. J Nutr 2004; 134: 18005.
  • 17
    Ekstrom AM, Signorello LB, Hansson LE, Bergstrom R, Lindgren A, Nyren O. Evaluating gastric cancer misclassification: a potential explanation for the rise in cardia cancer incidence. J Natl Cancer Inst 1999; 91: 78690.
  • 18
    Mattsson B, Wallgren A. Completeness of the Swedish Cancer Register. Non-notified cancer cases recorded on death certificates in 1978. Acta Radiol Oncol 1984; 23: 30513.
  • 19
    Cox DR, Oakes D. Analysis of Survival Data. London: Chapman and Hall, 1984.
  • 20
    Willett W, Stampfer MJ. Total energy intake: implications for epidemiologic analyses. Am J Epidemiol 1986; 124: 1727.
  • 21
    Ludwig DS. The glycemic index: physiological mechanisms relating to obesity, diabetes, and cardiovascular disease. JAMA 2002; 287: 241423.
  • 22
    Willett WC. Nutritional epidemiology, 2nd ed. New York: Oxford University Press, 1998.