American Journal of Epidemiology Copyright O 1992 by The Johns Hopkins University School of Hygiene and Pubc Health All rights reserved
Vol 136. No 9 Printed in U.S A
A BRIEF ORIGINAL CONTRIBUTION
Post-Challenge Glucose Concentration, Impaired Glucose Tolerance, Diabetes, and Cancer Mortality in Men
George Davey Smith,12 Matthias Egger,1 Martin J. Shipley,1 and M. G. Marmot1
The possibility that diabetes is associated with an elevated risk of cancer mortality has been discussed for many years. Recently, Levine et al. (Am J Epidemiol, 1990; 131:254-62) approached this issue by relating post-toad plasma glucose concentration to cancer mortality. For men, there appeared to be a positive association between postload glucose and mortality from cancer for all sites combined and for some specific sites. However, that analysis was based on only 298 cancer deaths among 11,521 men followed for 12 years. The current authors explored this issue in a cohort of 18,274 male civil servants, among whom there were 1,282 cancer deaths over 18-20 years of follow-up. There was no association between post-load glucose and cancer mortality, except for pancreatic cancer. A role for asymptomatic hyperglycemia in the etiology of cancer is not supported by the results of the present study. Am J Epidemiol 1992; 136:1110-14. diabetes mellitus; glucose; mortality; neoplasms; pancreatic neoplasms
The possibility that diabetes is associated with an elevated risk of cancer mortality has been discussed for many years. Development of pancreatic cancer and diabetes appear to be associated, while risk of lung cancer has been found to be reduced among diabetics, and the question of risk for cancer at all sites combined is unresolved (1-4). Theoretical reasons for why diabetes might be expected to increase the risk of carcinogenesis have also been advanced (5). Recently, Levine et al. (6) approached this Received fcx publication November 26, 1991, and in final form April 15, 1992 Abbreviation: ICD-8, International Classification of Diseases, Eighth Revision. ' Department of Epidemiology and Population Sciences, London School of Hygiene and Tropical Medicine, London WC1E 7HT, England. 2 Current address: Department of Pubic Health, University of Glasgow, Glasgow G12 8RZ, Scotland Reprint requests to Dr. George Davey Smith, Department of Public Health, University of Glasgow, 2 Uybank Gardens, Glasgow G12 8R2, Scotland.
issue by relating post-load plasma glucose concentration to cancer mortality over a 12-year follow-up period. For men, but not women, there appeared to be a positive association between post-load glucose and cancer mortality for all sites combined and for some specific sites. However, that analysis was based on only 298 cancer deaths among 11,521 men, and there was little power to explore the associations in detail. In the present study, we examined the risk of cancer mortality in a large cohort of male civil servants followed for 18-20 years. MATERIALS AND METHODS
In the Whitehall Study (7), conducted in London, England, 18,403 male civil servants aged 40-64 years were examined between 1967 and 1969. Participants completed a questionnaire regarding age and smoking habits. They attended the examination in the morning after an overnight fast; 50 g of
1110 Downloaded from https://academic.oup.com/aje/article-abstract/136/9/1110/81621 by INSEAD user on 29 May 2018
Glucose Tolerance and Cancer Mortality
oral glucose was administered, and 2 hours after this, a capillary blood sample was drawn for the measurement of glucose and cholesterol concentrations. Glucose concentration data were missing for 129 participants, leaving 18,274 men for this analysis. The 224 participants with a plasma glucose level of >200 ing/100 ml or with previously diagnosed diabetes constituted the diabetic group; the 999 nondiabetic men with glucose concentrations above the 95th percentiJe point (96-199 mg/100 ml) formed the group with impaired glucose tolerance, and the 17,051 remaining men were designated normoglycemic. This group was divided into quartiles for mortality analyses: 2 >5 >10
12,572 12,273 11,605
RR'.t
No. at risk
No. of deaths
RR
95% a *
34
1.0
4,350
22
1.86
1.08-3.17
33 28 21
1.0 1.0 1.0
4,296 4,182 3,939
20 16 9
1.74 1.65 1.26
1.00-3.13 0.89-3.05 0.58-2.75
• RR, rate ratio; Cl, confidence Interval, t Referent group.
TABLE 4. Age-adjusted rate ratios for pancreatic cancer in diabetic men and men with impaired glucose tolerance after the exclusion of the early follow-up periods, The Whitehall Study, 1967-1987 Start of foOowup period (years since baseline)
Al normoglycemic men
Men with IGT*
No. at risk
No. of deaths
RR*,t
No. at No. of risk deaths
Total
17,039
56
1.0
998
>2 >5
16,868 16,455 15,544
53 44 30
1.0 1.0 1.0
983 943 845
8
Diabetic men
95% cr
RR
No. at No. of risk deaths
RR
95% Cl
2.25 1.07-4.73 224
5.27 1.90-14.60
1.13-5.01 1.14-5.66 1.36-7.92
5.59 2.02-15.49 5.33 1.65-17.19 5.72 1.36-24.04
2.38 2.54 3.29
• IGT, impaired gtucose tolerance; RR, rate ratio; Cl, confidence interval, t Referent group. Downloaded from https://academic.oup.com/aje/article-abstract/136/9/1110/81621 by INSEAD user on 29 May 2018
220 208 166
1114
Davey Smith et al.
exclude virtually all participants with manifest tumors at baseline. Pancreatic cancer would need a long preclinical phase if the argument regarding preexisting cancers were extended to include a long lead-in time for undiagnosed cancer. A reduced risk of lung cancer for diabetic men has been seen in some (2, 10, 13) but not all (11) previous studies. The explanation usually advanced is a lower prevalence of smoking among diabetic men, but in the present study there was little difference in this regard. However, data regarding smoking were only available at baseline, and there could have been different rates of quitting smoking between the diabetic men and nondiabetic men. The reduced risk for lung cancer observed here was based on only two deaths (the expected number of deaths was 6.60) and was not statistically significant. Similarly, the increased risk of esophageal cancer among diabetics was also based on only two deaths (0.49 deaths were expected) and may well have been a chance finding. In normoglycemic men, the risk of pancreatic cancer mortality was elevated among those in the lowest post-challenge blood glucose quartile. There was no suggestion of a trend outside of this group, and the elevation in mortality was progressively attenuated with time from initial screening. This pattern would be the one anticipated if incipient pancreatic cancer led to a tendency toward post-challenge hypoglycemia. It is possible that the group of pancreatic cancers included a small number of malignant insulinomas, producing hypoglycemia. These tumors are rare (14), however, and the data available do not allow confirmation or refutation of this hypothesis. Since the finding was not hypothesized, it should be treated with extreme caution, although it would be interesting to see whether it could be reproduced in other data sets.
In conclusion, our data failed to confirm the findings of Levine et al. (6) regarding an association between post-load glucose and cancer risk, although this was a more powerful data set. Only if other studies produce positive results should consideration of a possible role of asymptomatic hyperglycemia in the etiology of cancer be given priority.
REFERENCES 1. Kessler II. Cancer and diabetes mellitus: a review of the literature. J Chronic Dis 1971;23:579-6OO. 2. O'Mara BA, Byers T, Schoenfeld E. Diabetes mellitus and cancer risk: a multisite case-control study. J Chronic Dis 1985;38:435-41. 3. Green A, Hougaard P. Epidemiological studies of diabetes mellitus in Denmark. V. Mortality and causes of death among insulin-treated diabetic patients. Diabetologia 1984,26:190-4. 4. Moss SE, Klein R, KJein BEK. Cause-specific mortality in a population-based study of diabetes. Am J Public Health 1991 ;81:1158-62. 5. Schumm DE, Matthews RH. Diabetes and cancer: a postulated relationship. Med Hypotheses 1979; 5:1353-61. 6. Levine W, Dyer AR, Shekelle RB, et al. Post-load plasma glucose and cancer mortality in middleaged men and women: 12-year follow-up findings of the Chicago Heart Association Detection Project in Industry. Am J Epidemiol 1990; 131:254-62. 7. Reid DD, Brett GZ, Hamilton PJS, et al. Cardiorespiratory disease and diabetes among middleaged male civil servants. Lancet 1974; 1:469-73. 8. Cox DR. Regression models and life tables (with discussion). J R Stat Soc B 1972;34:187-220. 9. Epidemiological Resources, Inc. EGRET: Epidemiological Graphics, Estimation, and Testing. Seattle, WA: Epidemiological Resources, Inc, 1988. 10. Kessler II. Cancer mortality among diabetics. J Natl Cancer Inst 1970;44:673-86. 11. Ragozzino M, Melton U, Chu C-P, et al. Subsequent cancer risk in the incidence cohort of Rochester, Minnesota, residents with diabetes mellitus. J Chronic Dis 1982;35:13-19. 12. Connolly MM, Dawson PJ, Michelassi F, et al. Survival in 1001 patients with carcinoma of the pancreas. Ann Surg 1987,206:366-70. 13. Armstrong B, Lea AJ, Adelstein AM, et al. Cancer mortality and saccharin consumption in diabetes. Br J Prev Soc Med 1976;30:151-7. 14. Boden G. Glucagonomas and insulinomas. Gastroenterol Clin North Am 1989;18:831—45.
Downloaded from https://academic.oup.com/aje/article-abstract/136/9/1110/81621 by INSEAD user on 29 May 2018
Vol 136. No 9 Printed in U.S A
A BRIEF ORIGINAL CONTRIBUTION
Post-Challenge Glucose Concentration, Impaired Glucose Tolerance, Diabetes, and Cancer Mortality in Men
George Davey Smith,12 Matthias Egger,1 Martin J. Shipley,1 and M. G. Marmot1
The possibility that diabetes is associated with an elevated risk of cancer mortality has been discussed for many years. Recently, Levine et al. (Am J Epidemiol, 1990; 131:254-62) approached this issue by relating post-toad plasma glucose concentration to cancer mortality. For men, there appeared to be a positive association between postload glucose and mortality from cancer for all sites combined and for some specific sites. However, that analysis was based on only 298 cancer deaths among 11,521 men followed for 12 years. The current authors explored this issue in a cohort of 18,274 male civil servants, among whom there were 1,282 cancer deaths over 18-20 years of follow-up. There was no association between post-load glucose and cancer mortality, except for pancreatic cancer. A role for asymptomatic hyperglycemia in the etiology of cancer is not supported by the results of the present study. Am J Epidemiol 1992; 136:1110-14. diabetes mellitus; glucose; mortality; neoplasms; pancreatic neoplasms
The possibility that diabetes is associated with an elevated risk of cancer mortality has been discussed for many years. Development of pancreatic cancer and diabetes appear to be associated, while risk of lung cancer has been found to be reduced among diabetics, and the question of risk for cancer at all sites combined is unresolved (1-4). Theoretical reasons for why diabetes might be expected to increase the risk of carcinogenesis have also been advanced (5). Recently, Levine et al. (6) approached this Received fcx publication November 26, 1991, and in final form April 15, 1992 Abbreviation: ICD-8, International Classification of Diseases, Eighth Revision. ' Department of Epidemiology and Population Sciences, London School of Hygiene and Tropical Medicine, London WC1E 7HT, England. 2 Current address: Department of Pubic Health, University of Glasgow, Glasgow G12 8RZ, Scotland Reprint requests to Dr. George Davey Smith, Department of Public Health, University of Glasgow, 2 Uybank Gardens, Glasgow G12 8R2, Scotland.
issue by relating post-load plasma glucose concentration to cancer mortality over a 12-year follow-up period. For men, but not women, there appeared to be a positive association between post-load glucose and cancer mortality for all sites combined and for some specific sites. However, that analysis was based on only 298 cancer deaths among 11,521 men, and there was little power to explore the associations in detail. In the present study, we examined the risk of cancer mortality in a large cohort of male civil servants followed for 18-20 years. MATERIALS AND METHODS
In the Whitehall Study (7), conducted in London, England, 18,403 male civil servants aged 40-64 years were examined between 1967 and 1969. Participants completed a questionnaire regarding age and smoking habits. They attended the examination in the morning after an overnight fast; 50 g of
1110 Downloaded from https://academic.oup.com/aje/article-abstract/136/9/1110/81621 by INSEAD user on 29 May 2018
Glucose Tolerance and Cancer Mortality
oral glucose was administered, and 2 hours after this, a capillary blood sample was drawn for the measurement of glucose and cholesterol concentrations. Glucose concentration data were missing for 129 participants, leaving 18,274 men for this analysis. The 224 participants with a plasma glucose level of >200 ing/100 ml or with previously diagnosed diabetes constituted the diabetic group; the 999 nondiabetic men with glucose concentrations above the 95th percentiJe point (96-199 mg/100 ml) formed the group with impaired glucose tolerance, and the 17,051 remaining men were designated normoglycemic. This group was divided into quartiles for mortality analyses: 2 >5 >10
12,572 12,273 11,605
RR'.t
No. at risk
No. of deaths
RR
95% a *
34
1.0
4,350
22
1.86
1.08-3.17
33 28 21
1.0 1.0 1.0
4,296 4,182 3,939
20 16 9
1.74 1.65 1.26
1.00-3.13 0.89-3.05 0.58-2.75
• RR, rate ratio; Cl, confidence Interval, t Referent group.
TABLE 4. Age-adjusted rate ratios for pancreatic cancer in diabetic men and men with impaired glucose tolerance after the exclusion of the early follow-up periods, The Whitehall Study, 1967-1987 Start of foOowup period (years since baseline)
Al normoglycemic men
Men with IGT*
No. at risk
No. of deaths
RR*,t
No. at No. of risk deaths
Total
17,039
56
1.0
998
>2 >5
16,868 16,455 15,544
53 44 30
1.0 1.0 1.0
983 943 845
8
Diabetic men
95% cr
RR
No. at No. of risk deaths
RR
95% Cl
2.25 1.07-4.73 224
5.27 1.90-14.60
1.13-5.01 1.14-5.66 1.36-7.92
5.59 2.02-15.49 5.33 1.65-17.19 5.72 1.36-24.04
2.38 2.54 3.29
• IGT, impaired gtucose tolerance; RR, rate ratio; Cl, confidence interval, t Referent group. Downloaded from https://academic.oup.com/aje/article-abstract/136/9/1110/81621 by INSEAD user on 29 May 2018
220 208 166
1114
Davey Smith et al.
exclude virtually all participants with manifest tumors at baseline. Pancreatic cancer would need a long preclinical phase if the argument regarding preexisting cancers were extended to include a long lead-in time for undiagnosed cancer. A reduced risk of lung cancer for diabetic men has been seen in some (2, 10, 13) but not all (11) previous studies. The explanation usually advanced is a lower prevalence of smoking among diabetic men, but in the present study there was little difference in this regard. However, data regarding smoking were only available at baseline, and there could have been different rates of quitting smoking between the diabetic men and nondiabetic men. The reduced risk for lung cancer observed here was based on only two deaths (the expected number of deaths was 6.60) and was not statistically significant. Similarly, the increased risk of esophageal cancer among diabetics was also based on only two deaths (0.49 deaths were expected) and may well have been a chance finding. In normoglycemic men, the risk of pancreatic cancer mortality was elevated among those in the lowest post-challenge blood glucose quartile. There was no suggestion of a trend outside of this group, and the elevation in mortality was progressively attenuated with time from initial screening. This pattern would be the one anticipated if incipient pancreatic cancer led to a tendency toward post-challenge hypoglycemia. It is possible that the group of pancreatic cancers included a small number of malignant insulinomas, producing hypoglycemia. These tumors are rare (14), however, and the data available do not allow confirmation or refutation of this hypothesis. Since the finding was not hypothesized, it should be treated with extreme caution, although it would be interesting to see whether it could be reproduced in other data sets.
In conclusion, our data failed to confirm the findings of Levine et al. (6) regarding an association between post-load glucose and cancer risk, although this was a more powerful data set. Only if other studies produce positive results should consideration of a possible role of asymptomatic hyperglycemia in the etiology of cancer be given priority.
REFERENCES 1. Kessler II. Cancer and diabetes mellitus: a review of the literature. J Chronic Dis 1971;23:579-6OO. 2. O'Mara BA, Byers T, Schoenfeld E. Diabetes mellitus and cancer risk: a multisite case-control study. J Chronic Dis 1985;38:435-41. 3. Green A, Hougaard P. Epidemiological studies of diabetes mellitus in Denmark. V. Mortality and causes of death among insulin-treated diabetic patients. Diabetologia 1984,26:190-4. 4. Moss SE, Klein R, KJein BEK. Cause-specific mortality in a population-based study of diabetes. Am J Public Health 1991 ;81:1158-62. 5. Schumm DE, Matthews RH. Diabetes and cancer: a postulated relationship. Med Hypotheses 1979; 5:1353-61. 6. Levine W, Dyer AR, Shekelle RB, et al. Post-load plasma glucose and cancer mortality in middleaged men and women: 12-year follow-up findings of the Chicago Heart Association Detection Project in Industry. Am J Epidemiol 1990; 131:254-62. 7. Reid DD, Brett GZ, Hamilton PJS, et al. Cardiorespiratory disease and diabetes among middleaged male civil servants. Lancet 1974; 1:469-73. 8. Cox DR. Regression models and life tables (with discussion). J R Stat Soc B 1972;34:187-220. 9. Epidemiological Resources, Inc. EGRET: Epidemiological Graphics, Estimation, and Testing. Seattle, WA: Epidemiological Resources, Inc, 1988. 10. Kessler II. Cancer mortality among diabetics. J Natl Cancer Inst 1970;44:673-86. 11. Ragozzino M, Melton U, Chu C-P, et al. Subsequent cancer risk in the incidence cohort of Rochester, Minnesota, residents with diabetes mellitus. J Chronic Dis 1982;35:13-19. 12. Connolly MM, Dawson PJ, Michelassi F, et al. Survival in 1001 patients with carcinoma of the pancreas. Ann Surg 1987,206:366-70. 13. Armstrong B, Lea AJ, Adelstein AM, et al. Cancer mortality and saccharin consumption in diabetes. Br J Prev Soc Med 1976;30:151-7. 14. Boden G. Glucagonomas and insulinomas. Gastroenterol Clin North Am 1989;18:831—45.
Downloaded from https://academic.oup.com/aje/article-abstract/136/9/1110/81621 by INSEAD user on 29 May 2018
