The bidirectional interation between pancreatic cancer and diabetes

Pancreatic cancer (PanCa) ranks fourth in the United States and eighth to ninth worldwide among cancer-related deaths. From 1999 to 2008, the incidence rate of PanCa increased among men and women. During the same period, death rates increased for pancreatic cancer among both men and women. PanCa is more common in elderly persons aged 60 years and older, with fewer than 7% of patients presenting with localized, potentially curable tumors and approximately 53% of patients presenting with distant metastases. The overall five-year survival rate among patients with pancreatic cancer is <6% [1, 2].

The bilateral causality between PanCa and diabetes has been widely documented. Type 2 diabetes could be either a cause, and/or possibly a consequence of PanCa [3].

Epidemiological studies have confirmed a higher incidence of PanCa in diabetic patients. A meta-analysis of cohort studies reported that individuals in whom diabetes had only recently been diagnosed (<1 year) had greater risk of malignancy in both males and females and that diabetes mellitus (DM) is both an early manifestation and an etiologic factor of pancreatic cancer [4]. Li et al. explored the association between diabetes and risk of PanCa in 2,192 cases and 5,113 controls in three large case–control studies using the method of multivariable analyses and found that diabetes was associated with a 1.8-fold risk of PanCa [5]. In a cohort study, Ogunleye et al. found the risk ratio to be significantly increased for pancreatic, liver and colon cancer [6]. Magruder et al. also reported that long-standing diabetes increases the risk of pancreatic cancer by 40% to 100%. Recent-onset diabetes is associated with a 4- to 7-fold increase in risk, such that 1% to 2% of patients with recent-onset diabetes will develop pancreatic cancer within three years [7]. Johnson et al. performed a population-based retrospective cohort study and found that within three months following diabetes onset, diabetics had a significantly increased risk of PanCa [8]. A population-based cohort study of 37,926 child-bearing Israeli women with a 28- to 40-year follow-up for diabetes and PanCa showed a 7-fold increase in the relative risk for PanCa development in women with a history of gestational diabetes mellitus. Furthermore, glucose levels in the upper range of normal have been associated with an increased risk for the development of some cancers, including PanCa [9]. Several studies have confirmed type II diabetes as a risk factor for PanCa. But rare research about the relationship between dyslipidemia and PanCa is reported. Tseng presented a new viewpoint that dyslipidemia was a significant risk factor for PanCa with a random sample of 1 million subjects, which would be helpful for clarifying the mechanisms of metabolic syndrome on PanCa [10].

A regional disparity was also found in diabetes-associated PanCa. For example, Hispanic men and Asian individuals had a higher risk of diabetes-associated PanCa than did white and black individuals, although the differences were not statistically significant [5]. Ogawa et al. found that in Japan the prevalence of PanCa in a group of individuals diagnosed with recent-onset diabetes (within three years) was 13.6%, which was significantly greater compared with the group of individuals who were diagnosed with diabetes more than three years prior (2.0%) [11]. Hong et al. reported a prevalence of PanCa among diabetic patients of 1.6% and a prevalence of diabetes in PanCa patients of 40.6% in Korea [12]. Chari et al. reported that approximately 1% of diabetics aged ≥50 years will be diagnosed with pancreatic cancer within three years of first meeting the criteria for diabetes. Furthermore, in 2,122 individuals in the USA, 10 of 18 (56%) were diagnosed <6 months after first meeting the criteria for diabetes [13]. An additional study in Americans analyzed a total of 1,172,496 case and control subjects and found that the frequency of pancreatic cancer in subjects with diabetes was 0.9% in comparison to subjects without diabetes (0.3%) [14].

Stevens et al. conducted a systematic review of the risk of pancreatic cancer in people with type 1 and young-onset diabetes, and found an increased risk of pancreatic cancer in type 1 diabetics, based on 39 cases of pancreatic cancer in young-onset and type 1 diabetes [15].

Tumor removal in PanCa patients improved glucose metabolism, which may suggest the presence of diabetogenic effect of PanCa [16]. Diabetes, especially new onset, is more frequent in PanCa patients and has been considered an early manifestation for asymptomatic PanCa. Approximately 80% of PanCa patients have glucose intolerance or frank diabetes [17]. In a population-based case–control study of PanCa, participants with PanCa were more likely to report a history of diabetes (13%) and had a shorter duration of diabetes and a larger proportion of insulin users than the control groups [18]. Diabetes has a high (50%) prevalence in PanCa and is frequently new onset (<36 months before index) in subjects with PanCa compared with subjects without PanCa [19]. At the time of diagnosis, 80% of PanCa patients have either impaired glucose tolerance or diabetes [20]. The incidence of diabetes is 34.63% among PanCa patients in China, of which 74.56% experience diabetes onset within two years of cancer diagnosis [21]. Matsubayashi et al. found the patients with familial PanCa were more likely to have diabetes mellitus (32.5%) than sporadic cases [22]. However, Bartsch et al. reported there were no statistical significant differences between familial and sporadic PanCa cases regarding the presence of diabetes and that diabetes was not a risk factor for pancreatic cancer [23, 24].

There has been a paradigm shift in the approach to understanding the relationship between PanCa and diabetes, as research has progressed into identifying the potential mechanisms that diabetes drive PanCa (Table 1).

Recently, there has emerged a growing understanding that new-onset diabetes type 2, which begins up to 24 months before cancer diagnosis, can be an early manifestation of asymptomatic PanCa [60]. The prevalence of diabetes is higher (40%) and more frequently of the new-onset type in PanCa than in control groups [61]. It has been suggested that screening for PanCa should be performed in patients with new-onset hyperglycemia and diabetes [62]. Because new-onset diabetes occurs 50 to 100 times more commonly than pancreatic cancer, screening efforts to detect occult pancreatic cancer in patients with new-onset diabetes require a feasible method to detect type 3 diabetes as the cause of impaired glucose tolerance [7]. Although the diabetes observed in patients with PanCa seems similar to type 2 diabetes in impaired glucose tolerance and insulin resistance, it affects only 1% to 5% of all diabetic patients [63]. Further support for the concept that the growing pancreatic tumor produces substances that alter glucose metabolism comes from experimental studies of glucose utilization and lactate formation in mouse myoblasts [44].

PanCa-associated diabetes has some specific characters in the clinic, which may be helpful for differentiation of PanCa-associated diabetes from type 2 diabetes (Table 4). Type 2 diabetes is common in the elderly and pancreatic cancer populations. In elderly subjects within two years of meeting criteria for diabetes, who have relatively low premorbid body mass index, experience weight loss, and have no family history of diabetes, a suspicion of pancreatic cancer is justified [64]. Impaired glucose tolerance or diabetes is often the first sign of PanCa and occurs shortly before other clinical symptoms occur, such as weight loss, malaise and pain [65]. Hart et al. compared body weight and fasting blood glucose at diabetes onset, one to two years before and at index date, and found that weight loss frequently precedes onset of PanCa-related diabetes, whereas new-onset primary type 2 diabetes is typically associated with weight gain, which may be an important clue to understanding the pathogenesis of PanCa-related diabetes [66]. Middle-aged diabetic men and women aged 45 to 65 years reportedly have the greatest risk of malignant neoplasms of the pancreas [67]. The prevalence of diabetes in PanCa did not differ by tumor stage or site. Median survival was similar in PanCa with and without diabetes [68]. Lin et al. found that PanCa should be considered in the list of precipitants for diabetic ketoacidosis in type 2 diabetes [69].

Biomarkers may be a useful strategy for the early diagnosis of PanCa, which can differentiate PanCa-associated diabetes from the more common type 2 diabetes and have differential expression the various stages in PanCa (Table 5).

Perhaps the strongest evidence for the tumor itself as a source for diabetogenic factor(s) is derived from the observation that glucose metabolism is improved after resection of the PanCa [70]. The hypothesis that a tumor-associated factor elicits the development of diabetes in PanCa is also supported by the fact that the onset of diabetes predates the first symptoms of PanCa by as much as two years, that is, well before the tumor is large enough to induce degradation of pancreatic islets [71]. However, the use of a biomarker may help to differentiate, PanCa-associated diabetes from the more common type 2 diabetes [61]. IAPP causes hyperglycemia by inhibiting muscle glycogen synthesis and stimulating glucose release from the muscles [72]. Permert et al. found that plasma glucagon and IAPP were significantly increased at 12 and 27 weeks, respectively, demonstrating that islet hormone changes accompany the early development of pancreatic tumors in the hamster pancreatic model [29]. However, Saruc et al. compared the IAPP-expressing cells between cancer area and tumor-free area, and found a decrease rather than an increase in the number of IAPP-expressing cells in PanCa. The development of diabetes in subjects prone to pancreatic cancer could be a red flag for malignancy [73]. Katsumichi et al. also found that the number of IAPP-expressing cells was significantly lower in diabetes and in the tumor area but not in the tumor-free region [74].

Kolb et al. found significant differences in the serum glucagon/insulin ratio in patients with PanCa-associated diabetes and type 2 diabetes, with a borderline cut-off of 7.4 ng/mU glucagon/insulin [75]. Aggarwal et al. found that adrenomedullin (AM) is overexpressed in PanCa tissue and plasma, especially in association with diabetes. Additionally, plasma AM is elevated in PanCa. Further studies in a larger cohort of patients are planned to validate AM as a potential biomarker of PanCa-associated diabetes [76]. Using a novel translational bioinformatics approach, Sharaf et al. identified fatty acid binding protein-1 (FABP-1) as having a significant positive association with PanCa on tissue microarrays, which was further strengthened by the presence of diabetes [77]. Using microarray analysis and RT-PCR, Huang et al. demonstrated that the combination of vanin-1 and matrix metalloproteinase 9 may be used as a novel blood biomarker panel for the discrimination of pancreatic cancer-associated diabetes from type 2 diabetes [78]. In addition, Basso et al. isolated a 14-amino-acid peptide from S100A8 in PanCa tissue from PanCa patients with diabetes. As S100A8 impairs the catabolism of glucose by myoblasts in vitro and may cause hyperglycemia in vivo, it might be helpful in diagnosing PanCa in patients with recent-onset diabetes mellitus [39]. Selective stimulation of amylin secretion was demonstrated when isolated pancreatic islet cells were incubated in media conditioned with Panc-1 as compared to unconditioned media [79]. Cancer antigen (CA) 19–9 is used in the diagnosis of pancreatic cancer but is also a marker of pancreatic tissue damage that might be caused by diabetes. Thus, the diagnosis of new-onset diabetes combined with higher CA 19–9 and/or carcinoembryonic antigen (CEA) might be regarded as a useful tool to screen early pancreatic cancer [80].

Amadori-glycated phosphatidylethanolamine (Amadori-PE), known as a reliable indicator of lipid glycation in vivo, is a nonenzymatically glycated lipid formed under hyperglycemic conditions. Using the streptozotocin (STZ)-induced diabetic rat model, Sookwong et al. found high levels of Amadori-PE in the blood and in organs that are strongly affected by diabetes, such as the kidney with a significant increase in STZ rats seven days after STZ treatment, suggesting that Amadori-PE may be a useful predictive marker for early-stage diabetes [81]. Diabetes mellitus could be a risk factor for PanCa. Eitsuka et al. found that Amadori-PE-enhanced cellular telomerase, which contributed to the infinite replicative potential of Panc-1 cancer cells in a time- and dose-dependent manner by upregulating human telomerase reverse transcriptase (hTERT) expression through induction of c-myc. These results provide experimental evidence for a novel role of Amadori-PE in linking diabetes and PanCa [82].

Navaglia et al. introduced the technique of surface-enhanced laser desorption and ionization time-of-flight mass spectrometry, which permits identification of new peptides that, in addition to CA 19–9, enable the correct classification of the vast majority of patients with pancreatic cancer, who can be distinguished from patients with chronic pancreatitis or type 2 diabetes mellitus [83]. To identify biomarkers for early PanCa, Fukamachi et al. established transgenic rats carrying a mutated H-ras or K-ras gene as a PanCa model, and found that even in rats with very small microscopic ductal carcinoma lesions, elevated serum Erc/Mesothelin can be detected [84].

Pelaez-Luna et al. determined the resectability of PanCa on abdominal computed tomography (CT) scans performed prior to clinical diagnosis and correlated resectability with onset of diabetes. They found that PanCa is frequently undetectable or resectable on CT scans performed ≥ 6 months prior to clinical diagnosis. At onset of diabetes, pancreatic cancers are generally resectable [85]. Several studies have reported the risk of antidiabetic drugs on cancer incidence. Chang et al. examined cancer incidence associated with use of insulin glargine, and found usage was positively associated with pancreatic and prostate cancers in men [86]. To elucidate the effect of pharmacotherapy for diabetes in cancer patients, Feng et al. found that insulin and glucose promoted cancer cell proliferation and contributed to chemoresistance, with metformin and rosiglitazone suppressing cancer cell growth and inducing apoptosis of four human cancer cell lines in vitro. Both drugs affected signaling in the protein kinase B (AKT)/mammalian target of rapamycin pathway, which provides experimental evidence to support further investigation of metformin and rosiglitazone as first-line therapies for type 2 diabetes in cancer patients [87]. Univariate and multivariate analyses were performed to determine whether the medications were predictive of early mortality in patients undergoing resection for PanCa.

Chagpar et al. found that patients with PanCa being treated for pre-existing diabetes with insulin therapy have an increased risk of early postoperative mortality [88]. To explore the ability of metformin to protect against cancer risk in Orientals, the possible metformin effect on total, esophageal, gastric, colorectal, hepatocellular and pancreatic cancers was examined in a Taiwanese cohort. In diabetics not taking anti-hyperglycemic medication, cancer incidence density increased at least twofold in total, including pancreatic cancers. The metformin dosage needed to observe a significant decrease in cancer incidence was ≤500 mg/day [89].

In addition to its ability to reduce insulin resistance, the antidiabetic drug, metformin has shown antitumor properties and is increasingly being considered as a drug for the prevention and treatment of obesity-related cancers [90]. Currie et al. reported that the effect of metformin might be tumor-specific, in that its use was associated with a reduced risk of cancer of the colon and pancreas, but not of cancer of the breast or prostate [91]. The antitumor effect of metformin seems to be mediated via its ability to increase the AMP-activated protein kinase (AMPK) signaling pathway [92]. Metformin negatively regulates mammalian target of rapamycin (mTOR) complex 1 (mTORC1) and the crosstalk between insulin/IGF-1 receptor and G-protein coupled receptor (GPCR) signaling, thus inhabiting the development of certain types of cancer [93].

Treatment with metformin induced striking and sustained increase in the phosphorylation of AMPK at Thr(172), and a selective AMPK inhibitor (compound C, at 5 μmol/L) reversed the effects of metformin on (Ca(2+))(i) and DNA synthesis, indicating that metformin acts through AMPK activation [94].

Using Cox proportional hazards models and controlling for smoking and alcohol use, a ten-year prospective cohort study reported that the stratum with the highest fasting serum glucose had higher death rates from all cancers combined compared with the stratum with the lowest level. By cancer site, the association was strongest for pancreatic cancer when comparing the highest and lowest strata in men and in women in Korea [95].