Insulin/Insulin-Like Growth Factor-1 Pathway in Barrett's Carcinogenesis

The epidemiology of esophageal cancer has shifted dramatically over the past three decades.1 The incidence of esophageal adenocarcinoma (EAC) has risen dramatically in recent years, whereas the incidence of esophageal squamous cell carcinoma has declined.2, 3 This shift in relative incidence of the two histological subtypes of esophageal cancer has been partly attributed to the obesity epidemic and the metabolic changes that accompany accumulation of excess body weight. Obesity is accompanied by high serum levels of insulin, insulin resistance, altered levels of adipokines as well as proinflammatory mediators.4 High levels of insulin can disrupt the delicate balance of proliferation and apoptosis and promote tumorigenesis. Hyperinsulinemia may be the mechanistic link between excess body weight and obesity-driven carcinogenesis in multiple organs, including esophagus.5, 6, 7 The proliferative actions of insulin may be mediated through the phosphatidyl inositol 3-kinase (PI 3-K) pathway. Insulin and insulin growth factor-1 (IGF-1) bind to the IGF-1 receptor (IGF-1R). Ligand binding to the extracellular α subunit causes conformational changes in the IGF-1R, which activate the tyrosine kinase domain of the intracellular β subunit. Activation of the tyrosine kinases leads to rapid phosphorylation of insulin receptor substrate (IRS) 1 through 4 as well as many other signaling molecules. IRS proteins then mediate the metabolic and anti-apoptotic effects of insulin signaling through the activation of PI 3-K, protein kinase B (AKT) and mammalian target of rapamycin (mTOR).5, 8 Studies have shown activation of the PI 3-K/Akt pathway in a number of obesity-driven cancers such as breast,9, 10 pancreas,11, 12 lung,13 colorectum,14, 15 liver,16 and gallbladder.17 To our knowledge, only limited number of studies have explored the activation of the PI 3-K pathway in Barrett’s esophagus (BE) based on human tissue samples.18 Further, it was of interest to compare activation of this pathway across the spectrum of esophageal carcinogenesis. We wanted to assess the PI 3-K pathway activation by means of immunohistochemistry of biopsy samples derived from normal squamous esophageal tissue, areas of intestinal metaplasia, that is, BE, low-grade dysplasia (LGD) and high-grade dysplasia (HGD), or EAC. We hypothesized that tissues in the latter spectrum of esophageal carcinogenesis would be more likely to show intense staining of the mediators of the PI 3-K pathway such as p-IRS, mTOR, and AKT. Our study group consisted of 75 patients. Sixty-five patients had BE, the remainder were diagnosed with EAC. Dysplasia was seen in 21 (32.3%) of 65 BE cases. In all, 44 out of the 65 cases (67.7%) had long segment BE, short segment BE was seen in the rest. The mean age of the study participants was 64.7±11.8 years. Study subjects were predominantly male (80%). Majority of study subjects (94.7%) were Caucasian, reflecting known demographics of BE. Mean BMI of the study participants was 30.46±s.d. 5.51 kg/m2. There were no significant differences in mean BMI of subjects with long vs. short segment BE (30.4 vs. 30.6 kg/m2). Mean BMI of cases with EAC was 29.7 kg/m2, which was not significantly different from cases with BE, P=0.7. Mean waist to hip ratio did not significantly differ between BE cases and those with ECA (1.02±s.d. 0.22 vs. 1.01±s.d. 0.04, P=0.86). Mean serum insulin levels were not significantly different between BE cases and those with ECA (12.1±s.d. 9.51 vs. 9.3±s.d. 6.5 μIU/ml (P=0.37), respectively). Mean IGF-1 of Barrett’s cases was higher than in those with ECA (191.5±s.d. 80.9 vs. 150.9±s.d. 69.7 ng/ml), this difference showed a trend toward statistical significance (P=0.14). Mean IGFBP-3 was lower in patients with ECA (3076±s.d. 886.4 vs. 2789±s.d. 580.6 ng/ml), however, this difference was not statistically significant (P=0.33). Mean levels of insulin, IGF-1, and IGFBP-3 did not differ in BE cases with and without dysplasia (Figure 1). The results of this study support activation of the insulin/IGF-1 axis in a proportion of cases of non-dysplastic BE, dysplastic BE, and EAC. We demonstrated intense staining with the phosphorylation of IRS-1, a specific indicator for activation of the insulin and IGF-1Rs, in 44% of BE, 47% of dysplastic BE, and 70% of EACs. Furthermore, the presence of phosphorylated IRS-1 showed weak-to-moderate correlation with the downstream mediators of the insulin/IGF-1 axis, phosphorylated AKT, and mTOR. Activation/phosphorylation of these molecules was closely correlated with cell proliferation as assessed by Ki-67. As increased cell proliferation is believed to contribute to carcinogenesis, this study suggests that this pathway may possibly have a role in the progression from metaplasia to dysplasia to carcinoma in BE. Systemic measurements of insulin, IGF-1, and its binding proteins did not show any correlation with tissue phosphorylation of IRS-1 assessed by immunohistochemical staining. This may be due to the fact that pIRS-1 may also be activated by signaling molecules other than insulin and IGF-1, including the IGF-1R.19, 20, 21 Further, pIRS-1 may also cause activation of alternate proliferative pathways implicated in carcinogenesis, including the extracellular signal-regulated kinase mitogen-activated protein kinase pathway. The lack of correlation is not totally surprising given that insulin and IGFs are under complex physiologic control mechanisms, which closely regulate circulating and tissue levels of these hormones. The development of IGF knockout models and identification of patients with IGF-1 gene defects have provided evidence on the important role of circulating and locally produced IGF-1 in growth and development.22, 23, 24 Data on relative contribution of endocrine and paracrine/autocrine regulation of IGF-1 in esophageal carcinogenesis is sparse25 but is emerging for cancers of other body sites.26 It is possible that the systemic levels of insulin, IGF-1, and its growth factor family are altered at the tissue level by locally derived paracrine factors, which then determine how the signal is transmitted intracellularly. Systemic levels of insulin and IGF-1 may correlate positively or negatively with tissue derived mRNA levels for pIRS. Cancer cell lines may also have increased affinity to IGF-1 stimulation than normal, non-cancerous, tissue. This has been observed in prostate tissue where IGF-1 stimulation led to 10-fold upregulation of IGF-1R expression.27 We opted to use IHC assays in this mainly exploratory study because of their specific ability to identify phosphorylated proteins in BE and its associated lesions. A potential limitation of this study is that tissue mRNA expression or protein activation by western blot analysis was not measured. Quantative PCR analysis performed on laser captured BE epithelium could provide further support for this insulin/IGF-1 hypothesis in future experiments but will not be able to uniquely determine phosphorylation status. Initial studies were performed with antibodies to phosphorylated IGF receptors. However, we found that these antibodies also cross-reacted with other growth factor receptors such as PDFGR when assessed by western blots (data not shown). The phosphorylated IRS-1 antibodies provided the best specificity and allowed us to assay the state of both the IGF-1R and IR, as both receptors phosphorylate IRS-1 on binding of substrate. However, this study cannot determine whether the insulin receptor, the IGF-1R, or both are active in Barrett’s epithelium and EAC. Further support for a potential role of the insulin/IGF-1 axis in esophageal carcinogenesis comes from a recent Australian case–control study demonstrating the association of diabetes mellitus and EAC.28 This study found that the risk of adenocarcinoma was increased 1.5 times in diabetics compared with population controls, although the risk was attenuated after adjusting for BMI. In addition, specific IGF-1 polymorphisms have been found to be associated with the reflux–metaplasia–dysplasia–cancer progression in another population-based study from Ireland.29 A more recent study also found evidence for activation of the IGF-1R in human adenocarcinomas and demonstrated that adenocarcinoma cell lines proliferate in response to IGF-1.25 Our study indicates that activation of this axis is already present at the precursor stage of BE. Future studies will have to determine whether subjects with BE who have more active cellular proliferation in response to activation of the insulin/IGF-1 pathways are at increased risk for progression to malignancy. Finally, the size of this study was small, which could have prevented us from detecting some of the meaningful correlations. Several of the correlations observed in this study were above the P value of 0.05 and thus statistically insignificant, yet still smaller than a P value of 0.15, which would be indicative of a potentially significant trend. One possible approach to solve this problem would be to increase the study sample size and retest the null hypothesis of insulin–IGF-1 pathway involvement in esophageal carcinogenesis. Increasing the study sample size could, however, also increase the chance of achieving a false-positive result. In summary, our study provides evidence that the insulin/IGF pathway is activated in a proportion of BE and a major proportion of EACs. In conjunction with our prior study that showed increased systemic levels of insulin and IGF-1 in subjects with BE,6 we propose that this pathway may partly mediate obesity-associated carcinogenesis in EAC. Increasing evidence indicates that this is an important pathway for further research. The effects of interventions that promote weight loss and medications that decrease insulin or IGF-1 need to be studied to determine whether they may prevent the progression from BE to cancer. Guarantor of the article: A. Chak, MD. Specific author contributions: Dr K.B. Greer collected data and performed statistical analyses. A. Kresak supervised tissue preparation and immunohistochemical staining. Dr J. Willis reviewed and interpreted all tissue sections. Dr A. Chak was responsible for study concept and analysis. All authors participated in the preparation of the manuscript. Study support: The study was supported by grants R21 CA135692 and U54 CA163060. Potential competing interests: None.