Effects of Long-Acting Insulin Analogues on Breast and Colon Cancer Promotion

Transcription

1 Effects of Long-Acting Insulin Analogues on Breast and Colon Cancer Promotion by Eunhyoung Ko A thesis submitted in conformity with the requirements for the degree of Master of Science Department of Physiology University of Toronto Copyright by Eunhyoung Ko (2014)

2 Effects of Long-Acting Insulin Analogues on Breast and Colon Cancer Promotion Eunhyoung Ko Master of Science Department of Physiology University of Toronto 2014 Abstract Insulin glargine, a long-acting insulin analogue, has been implicated in increased cancer risk by epidemiological studies and has been reported to increase mitogenicity in in vitro studies. However, studies in preclinical models confirming the tumor promoting effect of glargine are lacking. Methylnitrosourea and azoxymethane treated female Sprague-Dawley rats were randomly assigned to receive saline, neutral protamine Hagedorn (NPH, unmodified insulin), glargine, or detemir (long-acting insulin analogues) for 6 weeks (30/group). Rats treated with NPH had significantly higher mammary tumour multiplicity and a tendency to increased tumour incidence. Mammary tumour burden tended to increase in all insulin-treated rats compared to control. There was no effect of insulin on aberrant crypt foci, a precursor of colon cancer, compared to control. These data suggest that glargine does not promote breast or colon cancer to a greater extent than unmodified insulin. This is important information in the management of patients with diabetes. ii

3 Acknowledgments I would first and foremost like to thank my supervisor, Dr. Adria Giacca, for all the assistance that she has provided during this project. The tasks were not always easy, but with her guidance and patience we were able to succeed. I could not have finished this research without her and I am deeply grateful for the incredible opportunity- she provided me to develop both academically and personally. I feel privileged to have such a considerate and caring supervisor. Furthermore, I would like to thank all my colleagues in the Giacca lab including Simon Chang, June Gua, Alex Ivovic, Sandra Peirera, Tejas Desai, Khajag Koulajian, Tiffany Yu, Linda Qu, Stuart Wiber, and Yuri Crispim de Moraes. The project students, Linda and Stuart, and a medical student, Yuri, assisted me on taking metabolic measurements and sacrificing rats. Your advice and encouragement throughout my project truly made my experience enjoyable and memorable. I would also like to thank Loretta Lam for all her technical assistance and guidance. Another special thank you to Rudolf Furrer, who injected the rats with the carcinogens, taught me the techniques of chemical carcinogenesis, and supported me from the very beginning with words of wisdom and encouragement. I am also very grateful to my supervisory committee members Dr. Young-In Kim and Dr. George Fantus for their suggestions and guidance throughout my project. They continued to encourage me and support me on my project and future career. Finally, I would like to thank all my friends and family for all their support throughout the years. Your continued encouragement and understanding helped make difficult times easier to get through and good times all the more enjoyable. For that I am forever grateful. iii

4 Table of Contents Abstract......ii Acknowledgements......iii Table of Contents......iv List of Abbreviations......vii List of Tables......ix List of Figures......x Chapter 1: Introduction Diabetes Insulin Treatment Regular Human Insulin Long-Acting Insulin Analogues Insulin Receptor (IR), Insulin-like Growth Factor-1 Receptor (IGF-1R), and Hybrid receptors Diabetes, Obesity, Insulin Resistance, Inflammation, and Cancer Insulin Therapy, Insulin-like growth factor (IGF)-1 and Cancer Risk: Focusing on Breast and Colon Cancer Insulin, IGF-1, and Cancer Epidemiological studies In Vitro studies In Vivo studies Insulin Analogues and Cancer B10Asp Long-Acting Insulin analogues Epidemiological studies In Vitro Studies In Vivo Studies Multistage Model of Chemically-Initiated Carcinogenesis Initiation Promotion Malignant Conversion...21 iv

5 1.6.4 Progression Breast and Colon Cancer Animal Models using Carcinogen Breast Cancer Model - Methylnitrosourea (MNU) Colon Cancer Model - Azoxymethane (AOM) Aberrant Crypt Foci (ACF) Dual-Organ Carcinogenesis Model Rationale and Significance Hypothesis Chapter 2: Materials and Methods Procedure Experimental Animal Model and Sacrifice Procedures Animals Carcinogen Preparation Injections Carcinogens Insulin Rat Monitoring Sacrifice Tumour Evaluation Methods Mammary Tumour Analysis Western Blots ACF Scoring Calculations Power Calculations Statistical Analysis...30 Chapter 3: Results Metabolic Measurements Rat Weight Food Consumption Blood Glucose Levels...32 v

6 3.2 Mammary Tumour Analysis Mammary Tumour Incidence, Multiplicity, Burden, Growth Rate Histology Western Blot Colon Cancer (ACF) Results Chapter 4: Discussion Mammary Cancer Colon Cancer General Discussion and Future Direction References Appendix A vi

7 List of Abbreviations DMBA 7,12-Dimethylbenz(a)anthracene ACF Aberrant Crypt Foci ADA American Diabetes Association ANOVA analysis of variance AOM Azoxymethane Bcl-2 B-cell CLL/lymphoma 2 BMI Body Mass Index JNK c-jun N-terminal Kinases CI Confidence Interval PIP2 Di-phosphorylated Phosphoinositide EGF Epidermal Growth Factors EPIC European Prospective Investigation into Cancer and Nutrition F344 Fischer 344 FDA Food and Drug Administration Grb-2 Growth Factor Receptor-bound Protein 2 HCT116 Human Colorectal Cancer Cell Line HR Hazard ratio HEAL Health, Eating, Activity, and Lifestyle HFD High Fat Diet HIF-1 Hypoxia-Inducible Factor-1 IKK IkB kinase IKB Inhibitor of NF-kB IR Insulin Receptor IRS-1 Insulin Receptor Substrate-1 ITT Insulin Tolerance Test IGF-1R Insulin-like Growth Factor-1 Receptor IL-1 Interleukin-1 IL-6 Interleukin-6 i.p. intraperitoneal i.v. intravenous LID Liver-specific IGF-1 Deficient msos mammalian Son of Sevenless MCF-7 Michigan Cancer Foundation-7 (Human breast cancer cell line) MKR Skeletal Muscle-specific loss of IGF-1 Receptor MNU Methylnitrosourea MAPK Mitogen Activated Protein Kinases MDF Mucin-Depleted Foci NPH Neutral Protamine Hagedorn PTEN Phosphatase and tensin homolog vii

8 PBS PI3K PVDF RIA RIPA RR SEM s.c. PIP3 TNF-α WHI WHO Phosphate Buffered Solution Phosphoinositide 3-kinase Polyvinylidene Fluoride Radioimmuno Assay Radioimmunoprecipitation Assay Relative Risk Standard Error of Mean subcutaneous Tri-phosphorylated Phosphoinositide Tumour Necrosis Factor-α Women's Health Initiative World Health Organization viii

9 List of Tables Table 1 Summary of epidemiological studies reporting the association between the use of insulin glargine and all-cancer risks Table 2 Summary of epidemiological studies reporting the association between the use of insulin glargine and breast cancer risk Table 3 Diet Information ix

10 List of Figures Figure 1 Regular Insulin...3 Figure 2 Insulin Glargine (Lantus)...4 Figure 3 Insulin Detemir (levemir- 14C) and Insulin Degludec (16C)...5 Figure 4 Schematic Overview of Insulin-IGF System Figure 5 Post-receptor Signaling Pathways of Insulin Receptor (IR) and Insulin-like Growth Factor-1 Receptor (IGF-1R) Figure 6 Metabolite Formation of Insulin Glargine after Injection...19 Figure 7 Multi-stage Chemical Carcinogenesis Figure 8 Overall Schematic Animal Protocol Figure 9 Weekly Average Body Weight Figure 10 Weekly Average Food Consumption...32 Figure 11 Blood Glucose Before Insulin Injection...33 Figure 12 Blood Glucose at 4 hours Post-insulin Injection...33 Figure 13 Blood Glucose at 8 hours Post-insulin Injection...34 Figure 14 Glycated Hemoglobin Level Figure 15 Mammary Tumour Incidence...35 Figure 16 Mammary Tumour Incidence over time Figure 17 Mammary Tumour Multiplicity Figure 18 Mammary Tumour Burden Figure 19 Mammary Tumour Growth rate/day Figure 20 Histology of Mammary Tumours in each group...38 Figure 21 Preliminary Western blots...39 Figure 22 ACF Analysis...39 x

11 1 Chapter 1 Introduction 1.1 Diabetes Diabetes Mellitus, a metabolic disorder mainly characterized by an elevated blood glucose levels (hyperglycemia), is one of the most prominent disorders worldwide. The World Health Organization (WHO) estimates 347 million people have diabetes around the globe (Accessed on January 2013). In Canada alone, it is estimated that over 9 million people have diabetes or pre-diabetes (Canadian Diabetes Association, Accessed on January 2013), which is about a third of the Canadian population. There are three main types of diabetes mellitus: Type 1, Type 2, and gestational diabetes. Type 1 diabetes is caused by an autoimmune response toward β-cells, which secrete insulin. In contrast, Type 2 diabetes is a consequence of lack of insulin sensitivity or insulin resistance in insulin target organs, including muscle, liver, and adipose tissues. Compensatory hyperinsulinemia (elevated level of insulin in blood) follows insulin resistance and eventually results in dysglycemia due to β-cell exhaustion in predisposed individuals. Lastly, gestational diabetes occurs in pregnant women when insulin action is impaired due to pregnancy related factors such as human placental lactogen. Type 2 diabetes accounts for about 90% of all diabetes cases. The well-known symptoms and signs of diabetes include polyuria (frequent urination), polydipsia (frequent thirst) and polyphagia (frequent hunger) but can also include fatigue, weight loss and blurred vision. There are some risk factors that contribute to development of Type 1 and Type 2 diabetes. Main risk factors for Type 1 diabetes include race, family history, and certain viral infections in childhood. For Type 2 diabetes, the core risk factors are: diet, race, family history, age, Body Mass Index (BMI), and sedentary lifestyle. Diabetic patients are at risk of developing complications such as retinopathy, nephropathy, and cardiovascular disease (1). Moreover, multiple epidemiological studies have demonstrated that there is an association between diabetes and risk for cancer, including breast and colorectal cancer (2-7). Currently, the incidence of diabetes is increasing and no cure is available; however, the disease can be managed by diet, exercise, and medications such as metformin, insulin secretagogues, or exogenous insulin.

12 2 1.2 Insulin Treatment Regular Human Insulin The discovery of insulin by Banting and Best in the 1920s was a breakthrough in the diabetes community. At first, insulin was extracted from bovine or porcine pancreas. With time, biosynthetic human insulin became available by mass production using biotechnology (8) (Figure 1). Even though insulin is not a cure for diabetes, the hormone has saved millions of diabetic patients and bestowed a better quality of life; however, a difficulty in controlling blood glucose still remains. Regular human insulin has relatively short-duration of action with unstable absorption and therefore diabetic patients are required to receive injections multiple times a day. Moreover, if not controlled carefully, the patients may be at risk of hypo- or hyperglycemia. In order to prolong the action of the insulin, unmodified human insulin called Neutral Protamine Hagedorn (NPH) was developed. By adding neutral protamine (an arginine rich nuclear protein group), the absorption of insulin from the subcutaneous injection site is retarded (9). Prolonged action-profile of NPH allowed diabetic patients to establish a bed time basal insulin level but due to its peak release at 4-8 hours post-injection and an insufficient duration, high risk of nocturnal hypoglycemia, and morning hyperglycemia were experienced (9). In order to overcome these problems, long-acting insulin analogues were developed.

13 3 Figure 1: Proinsulin and Insulin Insulin is a peptide hormone that contains 51 amino acids. It is composed of an A-chain and a B-chain, which are linked by disulphide bonds. Proinsulin, which is the precursor molecule to insulin, is cleaved by prohormone convertases (at cleavage sites) to produce a C-peptide and insulin. An additional removal of 4 amino acids (denoted by red colors) by carboxypeptidase E reproduces a mature insulin. C-peptide and insulin molecules are stored in granules to be released in response to stimuli, such as elevated blood glucose Long-Acting Insulin Analogues Insulin analogues are modified insulin molecules at a primary structural level by recombinant technology. They were developed to best mimic endogenous insulin secretion by altering the pharmacokinetics. Some analogues, such as insulin aspart and lispro, are modified so that the molecules are absorbed much faster than regular insulin and remain in circulation for shorter duration. These analogues are suitable for administration prior to meal time. In contrast, other analogues are modified to have a prolonged time-action profile. Long-acting insulin analogues, such as insulin glargine (Lantus) and detemir (Levemir), are used to establish a stable basal insulin level for diabetic patients. The long-acting insulin analogues greatly reduce nocturnal hypoglycemia when compared to NPH and last for sufficient duration overnight.

14 4 Insulin glargine, which is the first long-acting insulin analogue produced by recombinant technology, has a glycine substitution for asparagine at A21 position and two arginine additions on the C-terminal end of B chain of the insulin molecule (10) (Figure 2). These modifications shift the isoelectric point of the insulin glargine molecules from ph 5.4 to ph 6.7 and thus, glargine precipitates under the neutral condition of the subcutaneous tissue (11). Precipitated glargine molecules are released slowly into the bloodstream, which protracts the duration of action to about 20 hours without any prominent peak. Insulin detemir, another long-acting insulin analogue, is modified using a different strategy. Amino acid threonine at B30 is removed and a 14-carbon fatty acid chain is acylated on B29 lysine (Figure 3). These modifications permit the molecules to bind albumin and to be released over a prolonged period of time (12). Insulin detemir lasts about 18 hours. Lastly, insulin degludec, an ultra long-acting insulin analogue, utilizes a similar approach as insulin detemir. It also has a deletion of B30 threonine and an addition of hexadecadionic (C16) fatty acid chain (13) (Figure 3). An exceptionally prolonged duration of insulin degludec (up to 48 hours) results from formation of multi-hexamer chains by the molecules in the subcutaneous tissue. While it is currently approved by the European Union, Japan, and Mexico, the Food and Drug Administration (FDA) rejected its approval in February 2013 due to a lack of cardiovascular outcome (Medscape News; accessed on April 13, 2013). Figure 2: Insulin Glargine (Lantus) Insulin glargine has a substitution to glycine for asparagine in position 21 on the A-chain. Also, it has additions of two arginine molecules at the end of the B-chain.

15 5 Figure 3: Insulin Detemir (Levemir- 14C) and Insulin degludec (16C) Both insulin detemir and degludec are missing threonine in position 30 on B-chain and have either a 14-C or 16-C fatty acid chain attached onto lysine 29 on the B-chain respectively. 1.3 Insulin Receptor (IR), Insulin-like Growth Factor-1 Receptor (IGF-1R), and Hybrid receptors IR and IGF-1R are members of the receptor tyrosine kinase family that share 47-67% amino acid homology (14). They are activated by their cognate ligands, insulin and IGFs respectively, to "turn-on" the downstream signaling pathways. IGF-1R and IR resemble each other structurally to an extent that their designated ligands can bind to each other's receptor with low affinity (15) (Figure 4). IRs can exist in two isoforms, IR-A and IR-B, depending on the presence or absence of exon 11. IR-A, which lacks exon 11, is expressed ubiquitously in low levels but predominantly in central nervous system. It is also highly expressed during fetal growth where it may be activated by insulin (16). Thus, it mainly activates a mitogenic signaling pathway. In contrast, IR-B retains exon 11 and is expressed dominantly in classical insulin targets: liver, muscle, and adipose tissue (16). A metabolic signaling pathway is predominantly activated by IR-B when bound to its ligand, insulin. While insulin can bind to IR-A, IGF-2 can also bind and activate IR-A to mediate a mitogenic effect (15). Once IR is activated by insulin, it mainly activates the PI3K/Akt pathway (Figure 5). Activated IR phosphorylates insulin receptor substrate (IRS), which in turn activates phosphoinositide 3-kinase (PI3K). PI3K phosphorylates di-

16 6 phosphorylated phosphoinositide (PIP 2 ) to tri-phosphorylated phosphoinositide (PIP 3 ), which then activates Akt or protein kinase B. The activation of Akt by PI3K is inhibited by phosphatase and tensin homolog (PTEN) dephosphorylation of PIP 3. Activated Akt is an important mediator that acts on various cellular processes, including glucose metabolism, cellular growth, and survival. IGF-1R is well-documented to induce mitogenic signaling and thus, it has been implicated in cancer development (17). Upon stimulation by its ligands, IGF-1 and 2, Shc protein is phosphorylated by the beta-subunit of the receptor. Then a growth factor receptorbound protein 2 (Grb 2)-mammalian Son of Sevenless (msos) complex is recruited. The activated msos loads GTP onto small G protein Ras, then the series of kinase activities occur downstream to mediate cell proliferation (18). Both IGF-1R and IR share very similar intracellular signaling pathways (19). Thus, it is a general consensus that IR and IGF-1R have overlapping functions besides their own distinguishing functions. Insulin can also act as a weak agonist for IGF-1R. While the mitogenic effect derived from insulin (at a physiological concentration) by activation of IGF-1R is expected to be almost negligible in normal tissues when compared to IGF-1, overexpression of IGF-1R displayed in breast cancer cells (20) and colorectal polyps (21) may allow the mitogenic effect of insulin to manifest in breast and colon cancer promotion, especially in hyperinsulinemic condition (19). Moreover, many malignant cells also overexpress IR-A, which appears to mediate growthpromoting effect over metabolic effect. Hyperactivation of IR-A by increased levels of insulin may facilitate malignant cell growth and survival (19). Intriguingly, due to highly homologous structure shared by IR and IGF-1R, half receptors from each type can heterodimerize to form an IR/IGF-1R hybrid receptor. As IRs can exist in two forms, IR-A and IR-B, the hybrid receptors can also exist in two isoforms, IR- A/IGF-1R and IR-B/IGF-1R. A study by Pandini et al. evaluated the differences in signaling and biological roles of two hybrid isoforms in engineered cells (22). The author reported an upregulation of IGF signaling system (mitogenic pathway) by IR-A/IGF-1R hybrid receptors by binding to IGF-1 and IGF-2 with similar affinity and to insulin with slightly lower affinity albeit in a physiological range. It is noteworthy that insulin, resembling IGF-1 and IGF-2, activated IGF-1R downstream pathway through IR-A/IGF-1R hybrid receptors. In contrast, IR- B/IGF-1R hybrid receptors bound to IGF-1 and IGF-2 with 6-fold lower affinity, and to insulin with negligible binding affinity. Thus, cells containing predominantly the IR-A/IGF-1R hybrid

17 7 receptors are more effectively stimulated by both insulin and IGFs than cells mainly composed of IR-B/IGF-1R hybrid receptors. Figure 4: Schematic overview of insulin-igf system Binding of ligands (one of insulin, IGF- 1, and IGF-2) to their designated receptor activates either metabolic or mitogenic effects. Thick arrows represent strong affinity or signal, and thin arrows represent weak binding or signal. The figure is modified from Pollak et al. (23) and the figure is not meant to be all-inclusive.

18 8 Figure 5: Post-receptor signaling pathways of insulin receptor (IR) and insulin-like growth factor-1 receptor (IGF-1R) Both insulin and IGF-1 bind to IR and IGF-1R, while the affinity is much higher toward their cognate receptor. Once corresponding ligand binds to the receptor, a series of phosphorylation event occurs downstream. Phosphorylation through the PI3K/Akt pathway results not only in metabolic effects, but also anti-apoptosis and protein synthesis. The MAPK pathway results in mitogenesis (cell proliferation). IR (especially IR-A) can also activate the MAPK pathway (Erk1/2). Thick arrows represent strong downstream signaling pathways whereas thin arrows represent weaker signal. The figure is modified from Tognon and Sorensen (24) and the figure is not meant to be all-inclusive. 1.4 Diabetes, Obesity, Insulin Resistance, Inflammation, and Cancer Many epidemiological studies have shown associations between diabetes and increased incidence of various types of cancers. Although prostate cancer risk seems to have a reduced among diabetic patients (25), other major types of cancer risks including colorectal (26), pancreatic (27), breast (28), hepatic (29), and endometrial (4) are increased. Moreover, diabetes

19 9 has been associated with increased cancer mortality (30). In Canada, a slightly but significantly increased relative risks (RR) of 1.1 to 1.2 were shown for breast (5;6) and colorectal (3) cancer in diabetes patients. This increase may appear to be negligible, however due to the high prevalence of diabetes and breast and colorectal cancer, diabetes is a risk factor in a sizeable population of cancer patients. The type of diabetes that is associated with cancer is type 2, whereas this association is not clear in type 1 (31). Obesity has been historically known as a risk factor for both diabetes and cancer. Additionally, a decreased cancer incidence and mortality rates were reported for obese patients who underwent bariatric surgery (32-34).Obesity is also independently associated with cancer. Solid epidemiological studies have linked overweight and obesity to certain types of cancer, such as breast and colorectal (35-40). In fact, a large meta analysis based on 30 European countries reported that 2.5% of new cancer cases in men and 4.1% of new cancer cases in women (or over 70,000 in absolute terms) were attributable to excess BMI (BMI 25 kg/m 2 ) over 10 years (36). Among all the cancer types, endometrial (33,421) cancer was the largest attributable new cancer, followed by post-menopausal breast (27,770) and colorectal (23,720) cancer. With a generally accepted association between overweight/obesity and increased cancer risks, a few mechanisms may explain how obesity increases cancer prevalence and cancer deaths: inflammation and insulin resistance (41). Obesity is associated with metabolic inflammation, which is characterized as lowgrade, chronic inflammatory state induced by excess nutrients (42). Although the initiating mechanism of metabolic inflammation is unclear, one mechanism proposed is hypoxia. Due to adipose tissue expansion, some cells distant from blood vessels become poorly oxygenated, resulting in local hypoxia (43). This hypoxia condition activates hypoxia-inducible factor (HIF)-1α to mediate infiltration of macrophages and monocytes into adipose tissue (44). It has been shown that both macrophages and monocytes secrete cytokines such as tumour necrosis factor (TNF)-α after their infiltration into the adipose tissue in obese individuals (45). TNF-α is implicated in cell survival (46), growth, and differentiation (47;48); these effects are believed to be through inhibiting the inhibitor of NF-kB (IkB). Inactivation of IkB activates NF-kB to establish its anti-apoptotic effect, thus enhancing cancer cell survival (46). Furthermore, NF-kB activation in cancer cells promotes cell cycling through c-myc and cyclin D1 to increase proliferation and growth (46;47). Thus, obesity-induced inflammation may facilitate carcinogenesis.

20 10 Obesity is also highly associated with the development of insulin resistance, a pathological condition that reduces insulin sensitivity in the body. An increased rate of lipolysis due to expanded adipose tissues in obese conditions results in excess free fatty acids, which can trigger the inflammatory pathways to impair insulin signaling (49). Additionally, an increased secretion of pro-inflammatory cytokines, including interleukin (IL)-6, IL-1β, and TNF-α, by adipose tissue can activate serine kinases such as IkB kinase (IKK) and c-jun N-terminal Kinases (JNK). These kinases phosphorylate serine residues of the IRS-1 to impair downstream cascades (49) and thus induce insulin resistance. Inflammation is induced and a consequence of insulin resistance, as cytokines impair insulin signaling and insulin has an anti-inflammatory effect (50). Insulin resistance is a key pathogenic factor of type 2 diabetes and is associated with augmented circulating insulin levels (51). This compensatory hyperinsulinemia exposes tissues to elevated insulin and insulin signaling. Since insulin resistance mainly occurs in the PI3K dependent metabolic effects of insulin in the target tissues (52), the MAPK dependent proliferative effect remains intact. Various studies have reported a proliferative effect elicited by elevated insulin levels. For example, insulin was found to promote proliferation by shortening the G1 phase in the Michigan Cancer Foundation-7 (MCF-7) human breast cancer cell line (53). These studies will be detailed in section Insulin Therapy, Insulin-like growth factor (IGF)-1 and Cancer Risk: Focusing on Breast and Colon Cancer Insulin, IGF-1, and Cancer In addition to its well-known metabolic effect, insulin also increases cell growth as well as proliferation of many cell types (54), the most common target being malignant cells (55;56). The stimulation of proliferation by insulin differs from that of IGF-1. While insulin exhibits mitogenic properties, IGF-1 is not a mutagen but may favor either spontaneous mutation or dedifferentiation (57) Epidemiological studies The increased cancer risks such as breast and colorectal cancer in type 2 diabetic patients can be independent of obesity (19;58). A number of epidemiological studies have demonstrated a positive association between serum insulin and breast/colon cancer risks. For example, the Women's Health Initiative (WHI) revealed that postmenopausal women with

21 11 higher levels of serum insulin were more prone to develop colorectal cancer (59). Moreover, an increase in C-peptide levels was shown to be positively associated with higher breast and colorectal cancer risk by a few studies, including the European Prospective Investigation into Cancer and Nutrition (EPIC) (60;61). Furthermore, a meta-analysis in 2008 that included prospective and case-control studies reported a 26% higher breast cancer risk (RR of 1.26; 95% Confidence Interval (CI); ) in patients with higher C-peptide levels (62). Health, Eating, Activity, and Lifestyle (HEAL) study by the National Cancer Institute, which was a prospective, multiethnic cohort study, reported a 35% increased breast cancer-specific deaths in patients with a 1ng/ml increase in C-peptide levels (63) (range of C-peptide: 0.25 to 9.7 ng/ml). Case-control as well as cohort studies also indicated a positive association between fasting insulin levels and breast cancer incidence with a RR of 2 to 3 (64;65). C-peptide is also linked with colorectal cancer risk. Studies displayed an association between elevated C-peptide levels and increased colorectal cancer risk (60). Also, higher levels of C-peptide is reported to be a predictive factor for higher colorectal neoplasia susceptibility according to a meta-analysis (66). Not only is endogenous insulin associated with cancer risk, but exogenous insulin therapy has been reported to be linked to cancer risk in type 2 diabetes by most studies (67-73) with a few exceptions (57;74). Although one meta-analysis observed that insulin use is not significantly associated with increased breast cancer risk, it was reported to be linked with increased colorectal risk (75). The association between hyperinsulinemia and cancer may be partly explained by IGF-1. Insulin can increase IGF-1 production from the liver by up-regulating growth hormone receptor and its downstream signaling (76). In addition to increased production of IGF-1, insulin simultaneously diminishes the level of Insulin-like Growth Factor-Binding Proteins (IGFBPs), which results in increased bioavailable IGF-1 (77;78). Higher levels of serum IGF-1 is correlated with increased breast and colorectal cancer risks. A meta-analysis involving 96 studies found that higher circulating IGF-1 levels elevated cancer risks, including breast and colorectal cancer (79). Moreover, another study reported that increased serum IGF-1 levels were associated with higher risk of cancer mortality in older men (80). Thus, the association between cancer risks and insulin levels in blood, whether it is direct or indirect, shows that insulin may play a role in cancer development.

22 In Vitro studies Both insulin and IGF-1 have been shown to stimulate proliferation and migration in cancer cell lines including breast and colorectal cancer cells (81-84). In breast tumour tissues, IR is overexpressed compared to normal breast tissue; this difference results in greater insulin induced cell proliferation in comparison to normal tissue (85). Increased proliferation by insulin was found to be mediated through IGF-1R and partly through IR (86) via both the MAPK and PI3K/Akt signaling pathways (87). Similarly, IGF-1 induced increased proliferation in MCF-7 breast cancer cells by activating MAPK and PI3K/Akt pathways (88). These pathways were activated by IGF-1 also in colorectal cancer cell lines (89;90). In both breast and colorectal cancer cells, IGF-1 activates Src kinase that consequently transactivates other growth factor receptors, such as epidermal growth factors (EGF) (81;91). Moreover, IGF-1 was found to oppose chemotherapy-induced death of breast cancer cells by increasing proliferation and inhibiting apoptosis (92). In addition to stimulation of proliferation and inhibition of apoptosis of cancer cell lines, IGF-1R can induce assumption of some malignant in vitro characteristics (such as anchorage-independent colony formation and loss of cell polarity) of non-cancerous breast (82) and colorectal (81) cell lines. The potential of IR to induce these characteristics is controversial (93;94) In Vivo studies In vivo data shows some heterogeneity in the link between insulin and cancer risk. A study by Tran et al. demonstrated that exogenous insulin administrations promoted colon tumour (tubular adenoma) formation in azoxymethane (AOM)-treated male Fischer 344 (F344) rats (95). They injected rats with doses of insulin (porcine NPH) of 15U/kg 5 times per week for 17 weeks. The insulin levels achieved are pathophysiological, i.e. can be seen in very insulin resistant rats. Another group did not find a significant increase in the number of aberrant crypt foci (ACF) in AOM-treated female F344 rats by exogenous insulin injections (bovine NPH, 20U/kg 5 times/wk for 14 weeks), but the crypt multiplicity was significantly increased (96), which indicates advanced dysplasia. A tumour promoting effect of insulin was also seen in mammary tumours. A study by Heuson et al. reported that exogenous insulin injection (dosage of 25U/kg Lente insulin (species not indicated but likely either bovine or porcine because the study was carried out before the introduction of human insulin) 6 times per week for 6 weeks significantly increased mammary tumour growth (in cm 2 ) in 1,2 dimethylbenz(a)anthracene (DMBA)-treated female Sprague-Dawley rats compared to control rats (55). Moreover, the

23 13 same author also reported mammary tumour growth dependence on insulin by showing regression of existing mammary tumour in DMBA-induced female Sprague-Dawley rats by treatment with alloxan (97) or streptozotocin (98), which destroys beta-cells in pancreas. Additionally, streptozotocin-induced diabetes resulted in regression of MCF-7 orthografts, an effect reversed by insulin treatment (99). On the other hand, other studies did not find the promoting effect of hyperinsulinemia. For instance, when female MNU-treated Sprague- Dawley rats were injected with insulin (same insulin type, dosage, and number of injections per week as Tran et al.) for 25 weeks, no differences in mammary tumour incidence were observed (100). There are also other studies that tested diabetic animal model. A study by Novosyadlyy et al. used muscle creatine kinase promoter/human IGF-I receptor (MKR) mice, which are lean, insulin resistant, and hyperglycemic, to assess the effect of insulin on breast cancer. The hyperinsulinemic state in MKR mice (about three-fold higher insulin level than the wild-type mice) caused hyperplasia in mammary gland and enhanced development of precancerous and cancerous mammary gland lesion. Both normal and malignant breast tissues from MKR mice exhibited augmented phosphorylation of IR and IGF-1R and a downstream biomarker, Akt (101). Also, after inoculation with orthotopic mammary tumours in MKR mice, the tumour growth was significantly greater than in inoculated wild-type mice. Expectedly, the use of selective IR/IGF-1R inhibitor (BMS ) decreased the tumour growth (101). As cell culture studies have shown, in vivo studies also confirmed the association between IGFs and increased breast and colon cancer. Liver-specific IGF-1 deficient (LID) mice, which exhibit only 25% of the serum IGF-1 level observed in wild-type, are protected from genetic and carcinogen-induced mammary and colon cancer (102;103). Another study reported that both male and female LID mice had significant 25% reduction in colon tumour size even though tumour incidence and multiplicity were not (104). These findings suggest that IGF-1 is an independent cancer promoter, which augments the growth of tumour Insulin Analogues and Cancer B10Asp While insulin analogues are preferred over regular human insulin owing to their favorable kinetics, some of these new insulin analogues may exhibit an increased mitogenic potential compared to regular insulin. This mitogenicity may have derived from either higher IGF-1R binding affinity or slower dissociation-rate from IR or both. These characteristics were observed with the first insulin analogue developed, B10Asp insulin (105). It is a rapid-acting

24 14 insulin analogue with an absorption rate twice as fast as human insulin. It had a substitution of histidine on B10 position with aspartic acid. B10Asp not only exhibited a higher binding affinity for IGF-1R, but it also phosphorylated IR and Akt to a greater extent and for more prolonged time than regular insulin. It was demonstrated to induce spontaneous mammary cancer in female Sprague-Dawley rats (105). Although its carcinogenic potential in rats prevented it from becoming commercialized, the mitogenic potency of B10Asp raised concerns about the carcinogenic potential of insulin analogues Long-Acting Insulin analogues Epidemiological studies In 2009, a strong concern about the safety of insulin glargine arose from a German study that demonstrated a dose-dependent association between all cancer risk and the use of insulin glargine among diabetic patients (70) (Table 1). However, it is noteworthy that the higher overall cancer incidence among the patients using insulin glargine over human insulin was found only after adjustments for insulin dose, whereas with the unadjusted data the opposite finding was observed (70). Another study found three other European studies followed up to address this issue (69;106;107). None of the three studies found an association between allcancer risk and the use of insulin glargine. While two studies reported an increased breast cancer risk with the usage of glargine at least in subsets of patients (106;107), one study found no association (69). The study by Jonasson et al. showed a breast cancer RR of 1.99 (95% CI ) over 2 years ( ) (107) but no increase in breast cancer risk was observed with the use of insulin glargine in the follow-up study with the identical group in third year (2008) (108). The study by Colhoun et al. showed increased breast cancer risk in a subset of patients using insulin glargine alone (106) (Table 2). Echoing the Jonasson's and Colhoun's data, another large population-based cohort study that included almost 20,000 patients reported a significant positive association between the use of insulin glargine and breast cancer risk ([Hazard ratio (HR)] Hazard ratio is the ratio of the hazard in treatment group versus control group. It differs from relative risk (RR) because HR is a cumulative risk of the entire study with a defined endpoint while RR is an at one time point of the study (usually the end), 95% CI ) (109).

25 15 Author Year Study type Patient Number (N) All-Cancer Risks (95% Confidence Interval) Hemkens et al Cohort study 127,031 HR= 1.19 ( ) at 30U and HR= 1.31 ( ) at 50U glargine vs human insulin Mannucci et al. Currie et al. Gerstein et al. Chang et al. Ljung et al Case-Control 1,340 OR= 1.33 ( ) glargine 0.3U/kg/day vs human insulin 2009 Cohort study 62,809 HR= 1.24 ( , p=0.19) glargine alone vs basal human insulin alone 2012 Randomized controlled study 12,537 HR= 1.00 ( ) glargine vs other treatment 2010 Cohort study 59,443 HR= 0.86 ( ) glargine vs intermediate/long-acting human insulin 2011 Cohort study HR= 1.10 ( ) glargine only vs other types of insulin Blin et al Cohort study 6,649 HR= 0.59 ( ) glargine vs human insulin Fagot et al Cohort study 70,027 HR= 1.01 ( ) glargine vs other basal insulin Table 1: Summary of epidemiological studies reporting the association between the use of insulin glargine and all-cancer risks However, this study observed a lower all cancer risk with the use of insulin glargine (HR 0.75, 95% CI ). The association between insulin glargine and breast cancer risk was also observed in a study by Suissa et al., where the authors found an increased breast cancer risk with the usage of insulin glargine after 5 years (first 5 years: HR 0.9; 95% CI , after 5 years: HR 1.8; 95% CI ), and a significant increase for the women who had insulin prior to taking insulin glargine (HR 2.7; 95% CI ) (110).

26 16 Author Year Study type Colhoun et al Cohort study Jonasson et al Cohort study Suissa et al Cohort study Ruiter et al Cohort study Habel et al Cohort study Currie et al Cohort study Chang et al Cohort study Fagot et al Cohort study Patient Breast Cancer Risk (95% Number (N) Confidence Interval) 36,254 HR= 3.39 ( ) glargine only vs non-glargine insulin alone 114,841 RR= 1.97 ( ) glargine only vs other insulin 15,227 HR= 1.8 ( ) glargine vs other insulin after 5 years HR= 2.7 ( ) glargine vs other insulin for women who had been on insulin before switching to glargine 19,377 HR= 1.58 ( ) glargine vs human insulin 128,175 HR= 1.3 ( ) ever used glargine vs NPH 62,809 HR= 0.86 ( ) glargine vs other insulin 59,443 HR= 0.62 ( ) glargine vs human insulin 70,027 HR= 0.97 ( ) glargine vs other basal insulin Grimaldi- Bensouda et al Casecontrol 3,825 HR= 1.04 ( ) glargine vs other insulin Sturmer et al Cohort study 52,453 HR= 1.12( ) initiating glargine vs initiating NPH Table 2: Summary of epidemiological studies reporting the association between the use of insulin glargine and breast cancer risk.

27 17 Other studies with a shorter follow-up period also yielded a positive association between glargine and breast cancer in subgroups (108;111;112). There are a few studies that reported no association between insulin glargine use and breast cancer risk (69; ). Thus, results in epidemiological studies for the effect of insulin glargine and breast cancer risk are still under debate, and longer follow-up is required. As far as all-type cancer is concerned, the majority of studies, except for one study (70;117), do not confirm the increased cancer risk by glargine (69;71; ). For instance, a study by Blin et al. used the French National Healthcare Insurance Database to report an all-type cancer HR of 0.59 (95% CI 0, ) among the type 2 diabetic patients (118). In fact, this study demonstrated a lower cancer risk with using insulin glargine compared to regular insulin. There are two recent randomized control studies, which also found no adverse carcinogenic effect of insulin glargine (119;120). However, it must be noted that the cancer incidence was secondary endpoint in these studies In Vitro Studies Numerous in vitro studies investigated the mitogenic effect of insulin glargine on malignant cell lines, including breast (121) and colorectal cancer cells (89). Not only MCF-7 breast cancer cells treated with insulin glargine showed a maximum of 3.1-fold increased proliferation compared to untreated cells, but also significantly higher proliferation than cells treated with human insulin (121). The author stated that the higher proliferative effect by insulin glargine in MCF-7 cells derived from a strong activation of IGF-1R and MAPK pathway. Consistently, a greater binding affinity (6- to 8-fold higher) and a higher activation of IGF-1R by insulin glargine compared to regular insulin or insulin detemir was reported by other studies (11; ). A study reported that insulin glargine can induce a significantly higher proliferative effect on MCF-7 cells at 1.5nM and 15nM concentration compared to regular insulin, but no proliferation enhancing effect was observed on benign mammary cell line MCF- 10A (123). The mitogenicity pertaining to insulin analogues may be partly linked with the IGF- 1R:IR ratio of the cell, considering the ratio of 4:1 in MCF-7 cell line and 0.8:1 in MCF-10A (125). Despite having a higher binding affinity towards IGF-1R than human insulin, insulin glargine has a faster dissociation rate from IR compared to regular insulin (122). It is thus thought that the higher mitogenic potency of insulin glargine is derived from an increased activation of IGF-1R signaling. However, it was reported that insulin analogues exhibiting slower dissociation rate from IR have higher mitogenic potency versus regular insulin

28 18 (122;126), thus suggesting a possible attenuation of the mitogenic potency of insulin glargine through IR. A study by Sciacca et al. compared mitogenic effects (cell proliferation and colony formation) of regular insulin, long-acting insulin analogues, and IGF-1 using engineered cell cultures that express only one of IR-A, IR-B, or IGF-1R to assess the contribution of each receptor (127). They found that both long-acting insulin analogues, insulin glargine, and detemir, induced a significantly higher Erk phosphorylation through IR-A receptor and IGF-1R compared to regular insulin. The author also assessed cell proliferation induced by both longacting insulin analogues. Coinciding with the Erk phosphorylation data, both insulin glargine and detemir induced a greater cell proliferation in IR-A, IR-B, and IGF-1R expressing cells (127). In other study, the activation of IR and IGF-1R by insulin glargine was seen human cancer cell lines. Insulin glargine at 50 ng/ml concentration induced phosphorylation of both IR and IGF-1R HCT 116 cells, and showed a robust phosphorylation of Akt signaling pathways, whereas Erk phosphorylation was comparable to regular insulin and IGF-1 (89). Insulin glargine had significantly higher proliferative effect and anti-apoptotic activities. Increased phosphorylation of Akt in MCF-7 cells is reported by Teng et al. when treating cells with insulin glargine (128). They also found that anti-apoptotic B-cell CLL/lymphoma 2 (Bcl-2) in up-regulated while Bax, a competitive antagonist of Bcl-2, is down-regulated. The author concluded that insulin glargine can exert a growth-promoting effect on breast cancer cells by enhanced anti-apoptotic signal activation through the Akt pathway In Vivo Studies As mentioned above, B10Asp insulin analogue was found to promote spontaneous mammary cancer in female Sprague-Dawley rats. Although both B10Asp insulin and insulin glargine exhibit higher binding affinity for IGF-1R, insulin glargine does not promote spontaneous mammary cancer in vivo. A study by Stammberger et al. reported that there was no difference in mammary tumour incidence in both rats and mice when given a daily subcutaneous injection with various doses of insulin glargine or human NPH (129). One other study compared insulin glargine and NPH (Human, doses from 5 to 20 U/kg/day) for 18 weeks on colonic mucosa cell proliferation and aberrant crypt foci formation in 1,2 dimethylhydrazinetreated female db/db mice to see the effects of insulin glargine on colon cancer promotion (a widely used mouse model of type 2 diabetes) (130). In this study, both NPH and insulin glargine resulted in a higher colonic epithelial proliferation and aberrant crypt foci formation

29 19 versus saline control. However, insulin glargine did not induce increased colonic epithelial proliferation or aberrant crypt foci formation when compared to NPH. Figure 6: Metabolite formation of insulin glargine after injection Insulin undergoes metabolism to form M1, missing two arginine molecules at the C-terminus, and M2, missing two arginine molecules and a threonine at the C-terminus of B chain. In vivo, it was also discovered that insulin glargine rapidly undergoes proteolytic degradation to form two metabolites called M1 and M2 after its subcutaneous injection in rats, dogs, and humans (11;131). It was reported that insulin glargine and M1/M2 are present in a ratio of 50:50 in human serum (131) and in vitro experiment using the serum of patients under glargine treatment had shown a 47% to 98% (mean of 72%) metabolism of insulin glargine into M1 (132). M1 has been found to be the predominant form in the circulation (Figure 6). While both M1 and M2 show similar binding affinity for IR as insulin glargine, they have weaker binding affinity to and activation of IGF-1R, and mitogenicity that is comparable to regular insulin (11). Intriguingly however, one pilot cross over randomized study reported that serum of type 1 diabetic patients under insulin glargine treatment evoked higher proliferation in MCF-7 cells than the serum of patients under NPH treatment (133). This finding suggests that in vivo, the concentration of glargine may be enough to induce proliferative effect at least in some

30 20 individuals that presumably metabolize glargine slowly. Most recently, an abstract presented at the American Diabetes Association (ADA) 2013 reported that unlike IGF-1 treatment, there was no increase in mammary tumour growth versus control using glargine treatment (12.5U/kg) for 2 weeks in MKR mice grafted with two types of mouse mammary tumour cells (134). 1.6 Multistage Model of Chemically-Initiated Carcinogenesis Multi-stage chemical carcinogenesis can be divided into four stages: tumour initiation, tumour promotion, malignant transformation, and tumour progression (Figure 7). Figure 7: Multi-stage chemical carcinogenesis DNA damage due to chemical exposure can lead to activation of proto-oncogenes and/or inhibition of tumour-suppressor genes. The figure is adapted from: Holland-Frei Cancer Medicine, 5th edition (2000) Initiation Tumour initiation involves irreversible damage to DNA. Tumour initiation by chemical carcinogen is brought about from formation of adducts between the carcinogen and nucleotides in DNA. The modification of DNA structure caused by carcinogen-dna adducts can lead to mutation during DNA synthesis. In general, a positive correlation is observed with the amount of adducts detected and the number of tumours formed in an animal model (135). Carcinogen- DNA adduct formation is believed to be necessary for tumour initiation and more specifically, formation of a DNA adduct that results in mutation activating a proto-oncogene or inhibiting a

31 21 tumour-suppressor gene is considered to be a tumour initiating event. Proto-oncogenes are genes involved in growth that, when mutated, can become oncogenes, i.e. cause uncontrolled growth. Conversely, tumour-suppresor genes suppress growth of the cells and when inactivated, can cause uncontrolled growth Promotion Tumour promotion stage comprises a selective clonal expansion of initiated cells. Since the accumulation of mutation is proportional to the rate of cell division, selective cellular expansion of initiated cells produces cells with increased risk of further genetic changes (because proliferating cells are more susceptible to mutations) and malignant conversion. Tumour promoters, such as IGF-1 or insulin, are non-mutagenic and non-carcinogenic by themselves. The contribution of cancer promoters to the process of carcinogenesis is the expansion of initiated cell population, which will then be more susceptible to malignant conversion. These agents are characterized by their ability to reduce tumour latency period of initiated cells or to increase tumour number formed Malignant Conversion With accumulation of genetic mutations in initiated cells, a fraction of these cells undergo malignant conversion, which is the transformation of preneoplastic cells into ones that express a malignant phenotype, defined based on genetic instability, uncontrolled growth, and invasion Progression Tumour progression refers to expression of malignant phenotype and more aggressive characteristics in malignant cells with time. Metastasis may occur if the tumour cells exhibit an ability to secrete proteases that permits penetration beyond their primary site. During tumour progression, further genetic mutations can occur to activate proto-oncogenes or inhibit tumoursuppressor genes. 1.7 Breast and Colon Cancer Animal Models using Carcinogen Breast Cancer Model - Methylnitrosourea (MNU) Methylnitrosourea (MNU) is a water-soluble alkylating agent that induces mammary tumours in rats when injected subcutaneously and in contrast to 7,12- Dimethylbenz(a)anthracene (DMBA - another widely used mammary carcinogen), metabolic activation is not required (136). MNU decomposes spontaneously to result in methyl diazonium

32 22 ion, which is believed to be the ultimate carcinogen (137). Due to its direct carcinogenesis, MNU is one of the widely used carcinogens for inducing mammary tumours. About 90% of the rat mammary tumours induced by MNU are known to contain activating H-ras mutations (members of the ras family are proto-oncogenes that increase cell proliferation), caused by guanine to adenine transition (138). Rat mammary tumours induced by MNU are estrogen receptor-positive and locally invasive, but metastasis is uncommon (139). Thus, MNU tumours are good models for estrogen receptor positive tumours. However, they are not good models for HER2/neu receptor positive and triple negative human mammary tumour. Also, ras mutations are not commonly found in human mammary tumour (<10%) (140). A study by Isaacs et al. investigated the mammary tumour incidence by subcutaneous (s.c.) injection of MNU (dosage 50mg/kg of body weight) in 50-days old Sprague-Dawley female rats (141). Among 20 rats tested, 95% tumour carcinoma incidence, 3.9 average tumour number, and 68 day latency period were observed. The same study also compared the efficacy between s.c. and intravenous (i.v.) injections of MNU in Sprague-Dawley rats and found no significant difference between the two modes of delivery. Thus, the author recommended the s.c. injection since it has the advantage of being easier and faster to perform and allows a reproducible treatment of a large quantity of animals by a minimal number of technicians Colon Cancer Model - Azoxymethane (AOM) Azoxymethane (AOM) is an alkylating agent that is potent in inducing colon tumours (mostly adenocarcinoma) in rats and mice. AOM is a derivative of 1,2-dimethylhydrazine, which is as potent as AOM at inducing colon carcinogenesis. However, 1,2-dimethylhydrazine undergoes a side reaction that consequently results in a highly toxic asymmetric hydrazine derivative (142). Thus, although 1,2-dymethylhydrazine is cheaper and more readily available, AOM is preferred for inducing colon tumours in animals. After administration, AOM is metabolized by CYP2E1 into methylazoxymethanol, which is the active agent that causes DNA mutation (143). AOM-initiated colon tumours often contain mutations in K-ras and B-catenin similar to human colon tumours (144). However, unlike human tumours, Apc gene is seldom mutated and p53 gene is never mutated. In addition, there is a low tendency for AOM-treated rat colon tumour to metastasize (145). Azoxymethane is found to give rise to aberrant crypt foci (ACF), colon preneoplastic lesions, as early as 3 weeks post-injection (146) and induce tumours (adenoma and adenocarcinoma) after 36 weeks post-aom injection in rats (147).

33 Aberrant Crypt Foci (ACF) ACF are generally thought to be the precancerous lesions that may develop into colon tumours ( ). Studies found a positive association between the number of ACF and colon tumour development. It was suggested that increasing number of crypts in ACF (increased ACF multiplicity) displayed more advanced preneoplastic state (152). ACF are found in both humans and rodents. A study by Pretlow et al. reported the presence of ACF in the colonic mucosa of colon cancer patients (149). They observed a much higher number of ACF in the mucosa from colon cancer patients compared to the colon from non-colon cancer patients. Despite the acknowledgement of ACF as the preneoplastic lesions of colon tumours, ACF may not be a reliable predictor of colon tumours. For example, a study of colon cancer prevention found a significantly increased ACF multiplicity in AOM-treated male F344 rats while colon tumour incidence and multiplicity were significantly decreased (153). This study also found that crypt foci with absent or minimal mucous production called mucin-depleted foci (MDF) may be a better predictor because the total number of MDF and crypt number/mdf significantly decreased, showing a consistent result as with the colon tumour incidence. Moreover, MDF were found to be more dysplastic than ACF when their histological dysplasia were graded according to dysplasia parameters (such as number of mitosis, nuclear crowding, and increase in nuclear:cytoplasmic ratio) at 7 and 15 weeks. MDF detection was also confirmed in human colon cancer patients by two studies (154;155) Dual-Organ Carcinogenesis Model A study by Shivapurka et al. demonstrated a potential way to examine both mammary and colon cancer in a single animal model, the female Sprague-Dawley rat (156). Three-week old rats underwent acclimatization and quarantine for a week and were then provided with high fat diet (HFD). The authors gave either a s.c. injection of MNU at a dosage of 50 mg/kg or an injection of saline at the end of two weeks on HFD. For the following week 3 and 4, the rats were given one s.c. injection of AOM at a dosage of 15 mg/kg or an injection of saline per week. MNU and AOM showed a relative specificity for their respective organs. Rats injected with MNU alone had only mammary tumours and rats injected with AOM alone had colon tumours. At the end of the protocol (32 weeks), the author reported similar mammary tumour incidence between MNU+AOM treated rats and rats treated with MNU alone (100% vs 95%). The colon tumour incidence was also similar between rats treated with MNU+AOM and AOM

34 24 alone (70% vs 65%). The authors also used a short-term protocol (11 weeks), which resulted in identical mammary tumour incidence in both MNU+AOM-treated rats and rats treated with MNU (80% vs 80%). In this short-term protocol, only ACF but no colon tumours were detected. The number of ACF formed was not significantly different as rats treated with MNU+AOM had 64.9±8.38 and rats treated with AOM had 63.6±7.52. This short-term protocol is the protocol followed in this thesis. 1.8 Rationale A large body of epidemiological studies and in vitro studies point to the potential mitogenic actions of insulin glargine. However, the epidemiological studies only show association, which may not be causal, and the in vitro studies cannot address the pharmacokinetics, such as peak action by NPH and proteolytic metabolism of insulin glargine. Currently, there is a paucity of preclinical animal studies that investigate the effect of insulin glargine on breast and colon cancer promotion. Moreover, only a dearth of available data regarding the effect of insulin detemir on cancer promotion under in vivo condition is available. Thus, the current study investigates the effect of insulin glargine and detemir on breast and colon cancer promotion in carcinogen-treated female Sprague-Dawley rats. 1.9 Hypothesis Insulin glargine treatment will promote breast and colon cancer to a greater extent than treatment with unmodified insulin (NPH) or other insulin analogues. To test this hypothesis, we used a carcinogen-induced mammary and colon tumour model in rats.

35 25 CHAPTER 2 Materials and Methods 2.1 Procedure Experimental animal model and sacrifice procedures Figure 8: Overall Schematic Animal Protocol 120 female Sprague-Dawley rats, age of 3 weeks, were used. Following a week of acclimatization to the facility, a high fat diet (HFD) was provided. After two weeks, one s.c. MNU injection (50mg/kg) was given and for the following two weeks, one s.c. AOM injection (15mg/kg) was given per week. 4 days later, the dark cycle was anticipated by 6 hours (12 hour light/dark cycle beginning at 7am to 12 hour light/dark cycle beginning at 1am) (the reason for anticipating the dark cycle is explained in the text). Then at age of 9 weeks, all rats were randomly allocated into four groups (Saline control, NPH, glargine, or detemir) and injected subcutaneously with assigned treatments 5 times/week for 6 weeks. When the rats were 15 weeks old, they were sacrificed for further analysis. The carcinogen injection protocol used was a well-established rapid dual-organ carcinogen model protocol (156). The insulin injection protocol used was that of Tran et al. (95) and Lu et al (100).

36 Animals The rapid dual carcinogenesis model of Shivapurkar et al. (156) was used. A hundred and twenty 3 week old female Sprague-Dawley rats weighing 50-60g were obtained from Charles River Laboratory International Inc. They were exposed to a 12 h light/dark cycle and caged in static cages in the animal facility (Department of Comparative Medicine, Medical Science Building, University of Toronto). The temperature of the animal facility was 22 o C and 60% humidity. After 1 week of acclimatization during which they were fed a regular pelleted rodent chow, the diet was switched to the HFD used in Shivapurkar et al. (156). The HFD was purchased from Dyets Inc. in Bethlehem, Pennsylvania, USA (Table 3). The purpose of the HFD was to induce insulin resistance (157) and also increase the tumour incidence (156). Table 3: Diet information The nutrient breakdown of the High Fat Diet (HFD) used in the experiment. The table was taken from Shivapurkar et al. (156). The diet contains roughly 40% fat by calorie Carcinogen Preparation According to the protocol in Shivapurkar et al. (156) the mammary tumours were induced by MNU (Sigma). MNU was dissolved in 0.9% NaCl saline (vehicle) at a

37 27 concentration of 10mg/ml. ACF were induced by AOM (Sigma). AOM was also diluted in 0.9% NaCl saline at a concentration of 10mg/ml Injections Carcinogens One s.c. injection of MNU at a dosage of 50mg/kg was given to 6 week old rats to induce mammary tumours. For the following two weeks, one s.c. injection of AOM at a dosage of 15mg/kg was given each week to induce ACF as per Shivapurkar et al. (156) (Figure 7). These injections were performed by Rudolf Furrer, who is a senior technician as well as an expert in animal cancer models Insulin The insulin treatment was same as in Tran et al. (95) and Lu et al. (100). Three days before insulin injection began, the light/dark cycle was anticipated by 6 hours (from 7am- 7pm(light) 7pm-7am(dark) to 1am-1pm(light) 1pm-1am(dark)). This was done in preparation for insulin injections because we wished the rats, which eat in the dark, once injection to eat at the peak of insulin effect in order to avoid hypoglycemia. S.c. insulin injections were given 5 times a week (from Monday to Friday) between 11am-noon for 6 weeks starting when the rats were 9 weeks old. Rats were randomly allocated into four groups: control (saline), NPH, insulin glargine, and detemir (N=30 rats/group) (Figure 7). NPH is an unmodified human insulin, and insulin glargine and detemir are long-acting insulin analogues. Rats were injected subcutaneously with the assigned treatments. They were injected with increasing dosage for the first week as follows: 5U/kg first two days, 10U/kg next two days, and 15U/kg for the remainder of the experiment, which according to blood glucose levels obtained at the peak of insulin action (4-8 hours after injection) achieved a moderate degree of glucose lowering (glucose level of 4-5mM). The insulin dosage was recalculated weekly according to the body weight measurements Rat Monitoring Metabolic measurements, including food consumption, weight, and blood glucose were obtained twice a week (Monday and Thursday) before the injection of insulin. After insulin injections commenced, blood glucose levels determined prior to every insulin injection (11am) and 4-5 hours after the insulin injection (3-4pm). In a subgroup of rats, blood glucose was also determined at 7pm (i.e. 8 hours after the insulin injections). Blood glucose was obtained by a nick in the tail and glucose levels were measured by a glucose meter (The OneTouch Ultra 2

38 28 Meter). The glycated hemoglobin (HbA1c level of each rat was determined just before sacrifice using the A1C NOW kit from Bayer Sacrifice When the rats were 15 weeks old, they were sacrificed. As the rats had to be sacrificed between 4-5 hours after the insulin injection, the assigned insulin injection was given to each rat every 15 minutes beginning at 11am. The sacrifice began at 3pm. As mentioned, the HbA1c level was measured prior to euthanasia for each rat. The rats were euthanized by ketamine injection at a dosage of 1 ml/kg. The euthanized rats were palpated for any presence of mammary tumours and the tumours were excised if there were any. Thus, we might have missed small non-palpable tumours. The excised mammary tumours were cut in half. Each mammary tumour was weighed and subsequently, one half of the tumour was fixed in formalin for histological analysis and the other half was frozen in liquid nitrogen for Western Blot. Following the mammary tumour excision, the colon was cut out from the anus to just before the caecum. Then the colon was washed thoroughly with phosphate buffered solution (PBS). Any fat and connective tissues that were attached to the colon were removed. The clean colon was cut into three pieces and laid on a Petri dish paper and these pieces were cut open along the longitudinal median axis and spread out flat. Then the paper was placed in a Petri dish to be fixed in 10% formalin. The Petri dishes containing all the colon samples were stored in airsealed containers. 2.2 Tumour Evaluation Methods Mammary Tumour Analysis Mammary glands were palpated and mammary tumours were measured (length and width) using a caliper twice a week. At sacrifice, the excised tumours were weighed. A few (n=) mammary tumours in each group underwent histological analysis by a pathologist Dr.R.Renlund) after hematoxylin & eosin staining performed by the pathology lab Western Blots Frozen mammary tumours were collected as described in and homogenized with a homogenizer in radioimmunoprecipitation assay (RIPA) buffer containing protease inhibitor cocktail (Roche Diagnostics, Laval, QC). The cell lysates were then spun at 12,000rpm for 10 minutes and the supernatant mixed with the dye was loaded onto 10% gel (Biorad, Hercules, CA) and transferred onto polyvinylidene fluoride (PVDF) membrane using the Turbo Blotter Transfer System (Biorad, Hercules, CA). The membrane was cut longitudinally along the 55

39 29 kda protein ladder line to separately detect for Akt and Erk1/2. The membrane that contained the protein size bigger than 55 kda was probed with rabbit anti-phospho Akt (60 kda) antibody (Cell Signaling) at a 1:500 dilution and developed on a film after detection with anti-rabbit secondary antibody (1:1000). Then the identical membrane was stripped and probed for Total Akt with rabbit anti-akt antibody (Cell Signaling) at a 1:1000 dilution, which was developed again on a film. The other membrane that retained protein size smaller than 55kDa was used for detection of Erk1/2 (42 and 44 kda) by using rabbit anti-phospho Erk1/2 antibody (Cell Signaling) at a 1:500 dilution and developed on a film after detection with anti-rabbit secondary antibody (1:1000). The actin was also detected using the corresponding membrane with goat anti-actin at a 1:1000 dilution and developed on a film after detection with anti-goat secondary antibody (1:1000). The images were then quantified using Image J software ACF Scoring Fixed colons were first washed with PBS and placed on microscope slides. Then each colon was stained with 0.1% methylene blue solution (methylene blue powder was dissolved in PBS and filtered to remove any debris) for 10 minutes. Each colon was washed again with PBS and scored for ACF with a light microscope. 2.3 Calculations Power calculations Initially, a sample size calculation was done based on the following equation extracted from a paper by Florey (158): n= (( ) 2 (p l q l +p 2 q 2 ))/True difference 2 where p 1 and p 2 are the percentages of rats that had mammary tumour (tumour incidence) and q 1 and q 2 are the percentages of rats that did not have any mammary tumour (100-p 1 ) in the NPH and control groups respectively. The true difference is the percentage difference in mammary tumour incidence between control and insulin-treated rats. We used the incidence value obtained in pilot data (33.3% difference between control and NPH). The value 1.96 is a multiplier when α = 5% or 0.05 and is a multiplier when β = 20% or 0.2, giving a power of 80%. With the power of 80%, we obtained a n=30/group. Based on our pilot data that showed a significant difference in tumour multiplicity, we also performed a power calculation for tumour multiplicity using the following equation (158): n=2( ) 2 s 2 /d 2

40 30 where d is the difference between the groups and s is the standard deviation of the variable. the value 1.96 is a multiplier when α = 5% or 0.05 and is a multiplier when β = 20% or 0.2, giving a power of 80%. With the power of 80%, we obtained a n=28/group. Thus, we used 30 rats per group for this study Statistical Analysis Continuous data are presented as means +/- standard error (SEM). Statistics were done with one way nonparametric analysis of variance (ANOVA) followed by Tukey's posthoc test for continuous data, Fisher's exact test followed by Bonferroni's correction was used for categorical data (incidence). Calculations were performed using statistical analysis software (SAS) (Cary, NC).

41 31 Chapter 3 Results All results refer to 30/group, unless otherwise specified. 3.1 Metabolic Measurements Rat Weight The weekly average weight of each group is plotted throughout the protocol. Although it was expected that the weight of insulin-treated groups would be higher than that of the control, the average weight between four groups showed no difference (Figure 9). Figure 9: Weekly average body weight The body weight of each group did not differ among groups. Data are presented as means +/- SEM Food Consumption The weekly average food consumption of HFD (HFD consumption began at 4 weeks of age) of each rat was measured and showed that all four groups consumed similar amount of diet. A slight decrease in food consumption in all groups between 6-8weeks may be due to the stress arisen from carcinogen injections (Figure 10). Also, it is possible that a transient peak at 5 week is due to the 'novelty' of the HFD.

42 32 Figure 10: Weekly average food consumption Food consumption did not differ among groups. Data are presented as means +/- SEM Blood Glucose Levels When measured at 11am (before the insulin treatment), the blood glucose levels of all groups remained relatively constant throughout the experiment. However, soon after the commencement of insulin injections in three groups (NPH, glargine, and detemir), we observed that the insulin treated groups maintained slightly but significantly higher blood glucose levels than the control group at 11am (Figure 11), consistent with a glucose rebound presumably due to counter-regulation. While the blood glucose level of control group remained constant when measured at 3pm (peak insulin release: 4 hours after the insulin treatment), the three insulin treated groups (NPH, glargine, and detemir) showed significantly decreased blood glucose levels at the same time (Figure 12). In a few rats (n in figure legend) blood glucose levels were measured at 7pm (end of peak release: 8 hours after the insulin treatment) (Figure 13). Insulin glargine and NPH groups had significantly lower blood glucose levels compared to control and insulin detemir showed significant decrease during last week. To determine the average blood glucose concentration over the protocol duration, HbA1 C levels were measured at sacrifice (Figure 14). As expected, all three insulin treated groups had significantly lower glycated hemoglobin levels compared to the control.

43 33 Figure 11: Blood glucose before insulin injection Blood glucose measured at 11 am (Prior to insulin injection) of each group. Data are presented as means +/- SEM. Figure 12: Blood glucose at 4 hours post-insulin injection Blood glucose measurements 4 hours after insulin injections for each group. Data are presented as means +/- SEM. *P<0.05 vs. control

44 34 Figure 13: Blood glucose at 8 hours post-insulin injection Blood glucose measurements 8 hours after insulin injections for each group. Control n=12, NPH n=10, Glargine n=11, and Detemir n=8. Data are presented as means +/- SEM. *P<0.05 vs. control Figure 14: Glycated hemoglobin level Glycated hemoglobin level taken at sacrifice for each group. Data are presented as means +/- SEM. *P<0.05 vs. control

45 Mammary Tumour Analysis Mammary Tumour Incidence, multiplicity, burden, growth rate At sacrifice, the mammary tumour incidence of insulin glargine and detemir groups showed a comparable result as control (57% and 63 vs. 60%). NPH displayed a tendency for an increase in tumour incidence compared to the control (80% vs. 60%) (Figure 15). In a few rats (n in figure) where we had data to show the cumulative incidence overtime and the mammary tumour incidence overtime of all insulin treated groups tended to have shorter latency period compared to control group (Figure 16). Insulin glargine and detemir showed a strong tendency towards increased tumour multiplicity (average number of tumours per rat) vs. control (1.53 and 1.43 vs. 0.8), but the difference was not significant. NPH group had a significantly higher tumour multiplicity versus control (1.77 vs. 0.8) (Figure 17). Finally, while the increase was not significant, all insulin-treated groups tended to have higher tumour burden compared to control (Figure 18). In a few rats (n in Figure), where we had longitudinal data on tumour growth, we did not find significant differences among groups (Figure 19). Figure 15: Mammary tumour incidence Mammary tumour incidence (% of tumour positive of all rats) was determined at sacrifice and analyzed. n=30/group for all graphs unless specified

46 36 Figure 16: Mammary tumour incidence over time cumulative mammary tumour incidence in rats following the 50mg/kg MNU injection. Insulin injection began when rats were 9 weeks old. Figure 17: Mammary tumour multiplicity Tumour multiplicity (number of tumours/rat) of each group. Value in non-tumour-bearing rats was counted as 0. Data are presented as means +/- SEM P<0.05 vs. control.

47 37 Figure 18: Mammary tumour burden The total tumour weight is the additions of all tumour weights per rat. Data are presented as means +/- SEM. Figure 19: Increase in tumour area/day Overtime measurement of mammary tumour growth rate/day. Data are presented as means +/- SEM

48 Histology One representative tumour from each group underwent histological analysis. The tumours exhibited abnormal architecture including reduced tubule formation, hyperchromia and mitosis, and irregular size and shape of the nuclei, which are all indicators of carcinoma (Figure 20). A pathologist (Dr.Renlund) confirmed that these tumours were adenocarcinomas. Figure 20: Histology of Mammary tumours in each group. H&E stained mammary tumours of a representative tumour from NPH group at 40x magnification. Black and white arrows denote mitotic figures and prominent nucleoli respectively Western Blot Pilot Western blot results showed a tendency for greater p-akt phosphorylation in the NPH group (Figure 21).

49 39 p-akt p-erk1/2 Actin Actin Figure 21: Preliminary Western blots results Expression levels of phospho-akt and phospho Erk1/2, which are markers of Akt and Erk activity respectively, were analyzed. As a loading control, actin levels were also measured. 3.3 Colon Cancer Precursors (ACF) Results ACF, the putative preneoplastic lesions, were scored and analyzed (Figure 22). The average total number of ACF in each rat and the average number of crypts per ACF were determined. All three insulin treated groups displayed no significant difference compared to control group in the total number of ACF and the number of crypts per ACF. Figure 22: ACF analysis a) Analysis of total ACF number per rat b) ACF multiplicity. Data are presented as means +/- SEM