Mortality From Site-Specific Malignancies in Type 2 Diabetic Patients From Verona

The design of the study (Verona Diabetes Study) has been described previously (19,20). In brief, 7,148 known type 2 diabetic patients residing in the Social Health Unit of Verona were identified on 31 December 1986 using three different sources (diabetes clinic, family physicians, and the drug prescription database). A total of 4,047 patients (56.6%) were ascertained from the records of two diabetes clinics of the area; in addition, 2,165 patients (30.3%) were identified by family physicians, and 936 patients (13.1%) were identified from the drug prescription database. A 76.5% completeness of ascertainment was estimated by the capture-recapture method (21). The diagnosis of diabetes was made according to National Diabetes Data Group criteria (22).

In 1986, BMI (kg/m²) of each study subject was computed as a measure of adiposity from weight and height recorded in indoor clothing without shoes. This variable was available in 6,012 patients (84.1% of the whole cohort), i.e., in those patients identified by the first two sources (diabetes clinics and family physicians). To allow comparison between the present cohort and the general population (23), a patient was considered overweight when BMI ranged from 25.0 to 29.9 kg/m² and obese when BMI was >30.0 kg/m².

Information on smoking habits was acquired by reviewing clinical records for those patients attending the Diabetes Center (n = 4,047).

Patients were followed prospectively from 1 January 1987 to 31 December 1996 to assess all-cause and cause-specific mortality. The life status was ascertained on 31 December 1996 by reviewing death certificates of the Registry Office and mortality records of the Social Health Unit of Verona. Of the initial cohort, 446 patients (6.2%) who had moved from Verona were not traced and were arbitrarily considered alive at the end of the follow-up.

To evaluate whether the prevalence of overweight and obesity in the type 2 diabetic cohort exceeded the prevalence in the general population, indirect standardization was performed using the prevalence values recorded in Northeastern Italy as a reference (23).

The standardized mortality ratio (SMR) was computed using the PYRS (person-years) software, version 1.3 (WHO, IARC, Lyon, France) (24), using the mortality rates in the general population of the administrative area of Verona as a reference. Exact confidence limits were computed for SMRs when observed deaths were fewer than 30 cases; otherwise, Byar’s approximation was used (25). Significance of the SMR was computed by the Poisson distribution after setting the level of significance at 0.05.

Gail’s test and the χ² test for trend (25) were used to evaluate changes in the observed/expected ratio (obs/exp) according to BMI: <25 (n = 1,759), 25–29.9 (n = 2,785), and ≥30 kg/m² (n = 1,468).

The main demographic and clinical characteristics of the cohort under study are described in Table 1. The cohort was rather old; the mean age was 67 ± 11 years, and the mean time since diagnosis was 10 ± 7 years. Sexes were almost equally represented; women were slightly older than men. Most patients (>80%) were treated with oral agents; men were treated with diet more frequently than women, who, in turn, were more frequently treated with insulin. A total of 36% of the men smoked, whereas this habit was rather uncommon among women (7%). More than 70% of the diabetic patients were either overweight or obese. The prevalence of overweight was slightly higher in the diabetic patients than that observed in the general population: the observed/expected ratio was 1.09 (95% CI 1.03–1.14) in men and 1.07 (1.01–1.13) in women. The prevalence of obesity was much higher than expected, particularly in women (obs/exp = 2.91, 2.73–3.11), but also in men (obs/exp = 1.98, 1.82–2.15).

During follow-up, 641 patients died of malignancies (Table 2), which corresponds to 22.1% of all deaths (n = 2,896). Deaths from malignancies were 378 in men and 263 in women (28.5 and 16.8% of all deaths, respectively). A small increase in overall mortality from malignancies was observed in the diabetic cohort and reached statistical significance in women (obs/exp = 1.16, 95% CI 1.02–1.30; P = 0.019) but not in men (obs/exp = 1.07, 0.97–1.19; P = 0.163) (Table 3). The most common causes of death among malignancies were digestive tumors both in men (140 of 378, 37.0%) and women (105 of 263, 39.9%), respiratory tumors in men (103 of 378, 27.2%), and tumors of the reproductive system in women (79 of 263, 30.0%) (Table 2).

Excess mortality from digestive tumors was observed in the diabetic cohort (SMR = 1.19, 1.05–1.35; P = 0.005). Of note, mortality from hepatic cancer was nearly doubled in both sexes (SMR = 1.86, 1.44–2.38; P < 0.001) and mortality from pancreatic cancer was doubled in women (obs/exp = 1.78, 1.13–2.67; P = 0.011). In addition, diabetic women experienced a higher mortality from malignancies of the reproductive system (obs/exp = 1.25, 0.99–1.55; P = 0.050), particularly breast tumors (obs/exp = 1.40, 1.06–1.81; P = 0.011) (Table 3).

The obs/exp ratio for the other site-specific cancers were not statistically significant. Of note, a large decrease in mortality was observed for cancer of the larynx (SMR = 0.51, 1.19–1.12; P = 0.080).

In women, the excess mortality from malignancies of the reproductive system and from breast cancer was confined to obese patients (BMI ≥30 kg/m²) (Fig. 1). Accordingly, the obs/exp ratios for malignancies of the reproductive system in women increased significantly with BMI (χ² for trend, P = 0.009). Also, cancer of the biliary tract was significantly increased in obese diabetic patients (χ² for trend, P = 0.020), and a similar trend, although not significant, was also observed for liver cancer (Fig. 1).

The main findings of the present study are as follows:

1. Mortality from malignancies is slightly but significantly increased in type 2 diabetic women compared with the general population.

2. In particular, mortality from breast and pancreatic tumors is significantly higher in women than men.

3. Mortality from hepatic cancer is increased in both sexes.

4. Obesity seems to be an important risk factor for cancer of the reproductive system in women, particularly breast cancer and cancer of the biliary tract.

These findings are partly in agreement with the current literature. In particular, incidence of hepatic cancer (11), breast cancer (13), and pancreatic cancer (12) has been reported to be higher in patients with diabetes, although the causal role of diabetes in the pathogenesis of the latter cancer has been recently questioned (26).

There is no straightforward explanation for the excess mortality from hepatic cancer in diabetic patients, but several hypotheses can be suggested. First, mortality from cirrhosis is nearly three times higher in patients with type 2 diabetes than in the general population (19). Hepatitis infections could have contributed to the increased mortality from both cirrhosis and hepatic cancer, as a high prevalence of the hepatitis C virus has been reported in diabetic patients (27). Also, alcohol consumption could contribute to the excess mortality from hepatic cancer: in the Paris Prospective Study (28), risk of death from alcohol-related diseases was high in men with diabetes. Finally, nonalcoholic steatohepatitis, which has been shown to occur in diabetic patients, could play a role, especially in overweight individuals (29). Indeed, mortality tended to be higher in obese patients, i.e., in those patients with a BMI >30 kg/m². Finally, another interesting hypothesis can be put forward: the increased mortality from hepatic cancer could somehow be related to hyperinsulinemia. Indeed, in a recent study, a relation between insulin and hepatic cancer in men has been reported (30).

Breast cancer and diabetes share many risk factors, such as high dietary fat intake, overweight, and sedentary lifestyle (16,17). Accordingly, the excess mortality from breast cancer was confined to overweight and obese diabetic women.

Obesity can be linked to breast cancer through the aromatase activity (p-450 aromatase) of the adipose tissue (31). In fact, especially after menopause, the adipose tissue becomes the primary endogenous source from aromatization of adrenal and ovarian androgens (32). The connection between breast cancer and estrogens has been known for many years, and in postmenopausal women there is a positive correlation between weight and risk of breast tumors. A striking correlation also emerged between BMI and biliary tract cancers, which are rare neoplasms, the most frequent being gallbladder cancer. Obesity and cholelithiasis, both common in diabetes, are considered risk factors for this type of tumor. It is interesting to note that the Verona diabetic cohort, and especially women with type 2 diabetes, presented a larger proportion of obese subjects than the general population (23); therefore, better control of body weight in diabetic patients seems necessary not only to reduce cardiovascular risk but also to prevent cancer from certain sites.

The present study did not support the view that diabetic patients present a higher mortality from cancer of the colon and rectum (10), endometrium, (9) and kidney (14) or a reduced mortality from lung (7) and prostate cancer (15).

In the present study, mortality from colorectal cancer, although not significantly different between the diabetic cohort and the general population, tended to increase in men and decrease in women. Similarly, a recent study from the American Cancer Society found that the direct relation between obesity and mortality from colorectal cancer is stronger in men than in women (33). Concerning endometrial cancer, the lack of a significant association with diabetes could be attributed to the rather low statistical power of the present investigation (expected event = 9.4). Mortality from lung cancer has been reported to be decreased in Scandinavian diabetic patients, possibly for educational intervention (7). In the present study, diabetic men presented a 10% reduction in mortality from lung cancer and a 50% reduction in mortality from larynx cancer; however, these were not statistically significant.

In conclusion, in the present study we found that some site-specific cancers behave differently in terms of mortality in diabetic patients than in the general population. These observations can have important implications both in evaluation and management of cancer risk factors in diabetes. In particular, the control of body weight needs to be improved.

This work was supported by grants from the Italian National Research Council, the Italian Ministry of University and Scientific and Technological Research, the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), and Regione Veneto.

We thank the following pharmaceutical companies for their support: Astra, Bayer, Boehringer Mannheim, Fournier Pierrel Pharma, GlaxoSmithKline, Laboratori Guidotti, Hoechst Marion Russel, Servier Italia, A. Menarini, Merck Sharp & Dohme, Molteni Farmaceutici, Zeneca, Pfizer Italiana, Roering Farmaceutici Italiana, Parke-Davis, Novo Nordisk Farmaceutici, Neopharmed, and LifeScan.