MC Perrin Department of Psychiatry, School of Medicine, New York University, 550 1st Avenue, New York, NY 10017, USA and Department of Epidemiology, Mailman School of Public Health, Columbia University, 722 West 168th Street, New York, NY 10032, USA , MB Terry Department of Epidemiology, Mailman School of Public Health, Columbia University, 722 West 168th Street, New York, NY 10032, USA , K Kleinhaus New York State Psychiatric Institute, 1051 Riverside Avenue, New York, NY 10032, USA , L Deutsch Unit of Epidemiology, The Hebrew University-Hadassah School of Public Health, Ein Kerem, Jerusalem, 91120, Israel , R Yanetz Unit of Epidemiology, The Hebrew University-Hadassah School of Public Health, Ein Kerem, Jerusalem, 91120, Israel , E Tiram Unit of Epidemiology, The Hebrew University-Hadassah School of Public Health, Ein Kerem, Jerusalem, 91120, Israel , R Calderon Unit of Epidemiology, The Hebrew University-Hadassah School of Public Health, Ein Kerem, Jerusalem, 91120, Israel , Y Friedlander Unit of Epidemiology, The Hebrew University-Hadassah School of Public Health, Ein Kerem, Jerusalem, 91120, Israel , O Paltiel Unit of Epidemiology, The Hebrew University-Hadassah School of Public Health, Ein Kerem, Jerusalem, 91120, Israel , S Harlap Department of Psychiatry, School of Medicine, New York University, 550 1st Avenue, New York, NY 10017, USA and Department of Epidemiology, Mailman School of Public Health, Columbia University, 722 West 168th Street, New York, NY 10032, USA

Discussion

To our knowledge this is the first study to relate pancreatic cancer in women to gestational diabetes mellitus. The results suggest that gestational diabetes mellitus could be an important risk factor for pancreatic cancer. Unlike type 2 diabetes mellitus, gestational diabetes mellitus has been little studied as a risk factor for cancer. In a population-based case control study that examined pregnancy characteristics and breast cancer, it was reported that the risk associated with gestational diabetes mellitus was reduced in the first five years after pregnancy but somewhat increased after five years [15]. A prospective cohort study in Scotland studied indices of sub-clinical glucose intolerance among women recruited during pregnancy, relating them to cancer 20 years later [16]. The risk of any neoplasm increased with higher fasting plasma glucose levels. Similar results were found for breast cancer. Though that study was quite small (N = 753), it suggested that even subclinical gestational glucose intolerance might be an important determinate of cancer risk.

The conversion rate for gestational diabetes mellitus to type 2 diabetes mellitus can be as high as 70% (range 2.6–70%) depending on the length of follow-up [19]. Women who have had gestational diabetes mellitus are often more insulin resistant than their normal counterparts after pregnancy and have also been found to have a defect in β-cell function (reviewed in Buchanan [7]). Increasing insulin resistance coupled with a decrease in β-cell function over time could lead to hyperglycemia characteristic of type 2 diabetes mellitus.

Several studies have investigated the role of glucose levels and risk of pancreatic cancer. A prospective study (follow-up 12 years) in Chicago found that mean post-load plasma glucose levels were lower among women who subsequently died of pancreatic cancer compared to survivors, though there were only six deaths reported [20]. Another study in the same population in Chicago (average follow-up of 25 years), examined post-load plasma glucose levels in 15183 women, relating them to pancreatic cancer mortality [21]. The authors reported a non-significant increase in deaths from this cause associated with post-load plasma glucose levels above 119 mg/dl. In a prospective study among 468615 Korean women enrolled in a health plan and followed for up to 10 years, mortality from pancreatic cancer was associated with a significant increase in risk among women with fasting serum glucose levels above 90 mg/dl (p = 0.01) [22]. The relative risk (compared to < 90 mg/dl) of fasting serum glucose in the normal range (90–109 mg/dl) was 1.5 (95% confidence interval (CI) 1.2–1.8) and up to 2.1 (95% CI 1.4–2.9) for those ≥ 126 mg/dl; the findings were similar when examining the association between fasting serum glucose levels and the incidence of pancreatic cancer. The results of the study in Korea suggest, as did the study in Scotland [16], that even glucose levels in the upper range of normal could be associated with an increased risk of some cancers, including pancreatic cancer.

There is a pathway by which hyperglycemia could increase the risk of pancreatic cancer in persons with either type 1 or type 2 diabetes mellitus. Hyperglycemia increases the generation of reactive oxygen species (ROS) that are hypothesized to be linked to some of the more common morbidities associated with diabetes [23]. Persons with type I and type 2 diabetes mellitus have been reported to generate more reactive oxygen species (ROS) than controls [24]. If anti-oxidant mechanisms are overwhelmed and DNA is damaged, this could lead to the loss of function in critical proteins such as those involved in tumor suppression. However, it should be considered that in our sample only women diagnosed with gestational diabetes mellitus went on to develop pancreatic cancer whereas no cases of pancreatic cancer were found among women diagnosed with type 1 diabetes mellitus.

Gapstur and colleagues have proposed a mechanism by which hyperinsulemia, associated with type 2 diabetes mellitus, might act to increase the risk of pancreatic cancer [21]. They postulate that in the hyperinsulemic state, the exocrine cells of the pancreas are exposed to extremely high levels of insulin. Insulin at high levels binds to the insulin-like growth factor I (IGF1) receptor [25] and downregulates the IGF binding protein 1, (IGFBP1) [26], thus the amount of bioavailable IGF-1 could increase. Insulin has been reported to increase cell growth in pancreatic cancer cell lines [27], while IGF1 has been shown to increase pancreatic cancer cell growth [28].

For many years there has been a controversy whether diabetes is causally associated with pancreatic cancer or is merely a consequence of tumor growth. Different studies have addressed this question by examining the length of time between diagnosis of diabetes and pancreatic cancer. In a recent meta-analysis, time between diagnoses was examined; it was reported that there was a significant 50% increase in risk associated with type 2 diabetes mellitus diagnosed 5 or more years earlier [6]. In the current study the time between diagnosis of gestational diabetes mellitus and that of pancreatic cancer ranged from 14–35 years. As disease progression is aggressive in pancreatic cancer (the 1-year survival rate was 30% in this cohort) it makes it unlikely that gestational diabetes diagnosed 14 or more years earlier is a consequence of tumor growth [6].

An intriguing observation is the strong association of pancreatic cancer with heart disease. Although based on only three cases, it is unlikely to be due to chance. We have no information on the exact diagnosis; however, heart disease recorded at the time of the birth would have been due to congenital defect, rheumatic heart disease or cardiomyopathy.

Our study has the advantage of a prospective design, long follow-up and complete obstetric history on all births. The data on gestational diabetes mellitus and other birth complications were taken from medical records at the time of birth and did not rely on self-report. Further, the ascertainment of pancreatic cancer is likely to be complete and unbiased relative to the obstetric history. Unlike many studies of diabetes, we were able to distinguish between women with type 1 diabetes mellitus and those with gestational diabetes mellitus; however, we do not have an exact definition of diabetes and modern criteria for gestational diabetes mellitus were not applied in that era. Data were available only on observed pregnancies, and there might be some misclassification of exposure, but it was likely random and would only tend to weaken the association. We were unable to control for smoking history and body mass index in the analysis; however, the prevalence of current smoking and obesity in a subgroup of women interviewed postpartum was only 13.2% and 2.1% respectively. Though there were only 54 cases of pancreatic cancer ascertained through the follow-up period, given the strength of the association our results are unlikely to be due to chance or confounding, although it cannot be ruled out as the RR at the low end of the confidence interval was 2.8.

Conclusion

The prevalence of gestational diabetes mellitus is reported to be increasing [29,30], not surprisingly given the prevalence of overweight and obesity in the US [31]. Other than type 2 diabetes mellitus, the sequelae of gestational diabetes mellitus remains largely unknown. If the results of the current study are confirmed, then it would be expected that the incidence of pancreatic cancer could increase. Research is urgently needed to ascertain this, and other possible health sequelae of gestational diabetes mellitus, so that appropriate interventions can be developed and implemented.