Does Insulin Therapy Promote, Reduce, or Have a Neutral Effect on Cancers?

Because up to 10% of adults have diabetes and because a substantial proportion will ultimately be prescribed insulin, reports that insulin therapies may be associated with cancer are of significant concern.^{1 - 2} This Commentary highlights limitations of the literature and notes that the evidence is also consistent with the hypothesis that insulin therapy targeting good glycemic control might actually reduce cancers.

Type 2 diabetes is an independent risk factor for several cancers,³ and this relationship parallels the relationship between obesity and cancer. For type 1 diabetes, prospective studies report a modest 20% increased risk of stomach, endometrial, and cervical cancer and a 2-fold higher risk of pancreatic cancer. Prospective studies also suggest that glucose levels above normal predict cancers. For example, in a 10-year follow-up of more than 1 million Korean adults, there was a progressive relationship among fasting glucose levels above normal, cancers, and cancer death.⁴ Moreover, in a prospective British study, hemoglobin A_1c levels above normal were related to an increased risk of incident colorectal cancer.⁵

Explanations include a cancer-promoting effect of insulin, confounding of some causal factor with glycemia and cancer, and the elevated glucose level.

Endogenous or Exogenous Hyperinsulinemia. Insulin has mitogenic effects that may promote cancer growth among individuals with high insulin levels due to insulin resistance. Insulin binding to insulin and insulinlike growth factor (IGF) receptors (overexpressed on many cancer cells) may promote these effects. Moreover, high insulin levels reduce IGF binding proteins, thereby increasing IGF-1 levels, and also may promote angiogenesis within tumors. However, there is no evidence that insulin promotes de novo tumor development.

Confounding of the Causal Factor With Both Dysglycemia and Cancers. If patients with high glucose levels have other diseases or are taking therapies that may be associated with cancer, studies will identify dysglycemia as a cancer risk factor because of its link with the putative causal factor (ie, other drugs or disorders). That is, the causal factor would be confounded with both dysglycemia and cancers. Whereas confounding is typically managed by statistical adjustment, such an approach is only possible if a particular variable is postulated to be a confounder and if data are available regarding that variable. An inability to satisfy both conditions may lead to the conclusion that a risk factor is causally related to an outcome when in reality it is related to the unmeasured or “residual” confounder.

For example, a recent analysis of an administrative database¹ suggested that after adjustment for age and sex, the use of glargine insulin (an insulin analog) compared with other insulin preparations was associated with a significant 14% lower risk of incident cancer (95% confidence interval, 6%-21%). After additional adjustment for insulin dose, however, there was a significant 14% higher risk of cancer with glargine insulin (95% confidence interval, 5%-24%). Because obesity is linked to both higher insulin dose and cancer, an analysis that also adjusted for obesity may have eliminated or attenuated any relationship with insulin. Absence of data on obesity in the database meant that it could not be included in the analysis. This example highlights the importance of interpreting data from administrative databases as hypothesis generating as opposed to hypothesis testing.

Elevated Glucose Level. The cancer literature has traditionally focused on the role of molecular signaling (eg, insulin receptor binding) and cancer. However, without energy, even powerful signals will not promote cancer growth. Bioenergetics (ie, the need of cancer cells for energy from adenosine triphosphate [ATP]) represents an important yet somewhat neglected aspect of cancer biology.

In nonproliferating cells, a glucose molecule is converted into pyruvate through glycolysis, which in the presence of oxygen enters the mitochondrion and undergoes oxidative phosphorylation, ultimately yielding approximately 36 molecules of ATP. Without oxygen, pyruvate is mainly converted to lactate, yielding only 2 molecules of ATP.⁶ In 1924, Otto Warburg, MD, observed that cancer cells primarily depend on the glycolytic pathway for ATP production and convert most of the pyruvate to lactate under aerobic conditions. This may allow synthesis of sufficient amino acids, nucleotides, and lipids for cell replication at the expense of ATP production (because only approximately 4 molecules of ATP are produced per molecule of glucose). Moreover, because cancer cells tolerate acidic environments, the presence of lactic acid may promote cancer growth.

Cancer cells may therefore not thrive without a reliable, constant glucose supply. The avidity of cancer cells for glucose is highlighted by the use of fluorodeoxyglucose to image cancers and to follow their response to therapy using positron emission tomography.

When Insulin Is Used vs Not Used. Although database analyses have associated insulin therapy with cancer, the possibility of residual confounding cannot be excluded. For example, in a recent analysis reporting a relation of insulin use (vs metformin) with pancreas cancer, colon cancer, or both,⁷ patients prescribed insulin had already “failed” therapy with oral hypoglycemic agents, whereas patients prescribed other agents were drug naive. Confounding of insulin use with unmeasured factors associated with oral agent failure may have accounted for the observed association.

Glargine vs Other Insulin. Because glargine insulin has an increased affinity for IGF-1 and insulin receptor compared with human insulin, some investigators have hypothesized that this drug may promote cancer.¹ However, analyses of administrative databases² did not consistently identify a relationship. Moreover, a meta-analysis of all randomized trials with glargine insulin in patients with type 1 or type 2 diabetes did not identify any relationship with cancer,⁸ and no relationship was identified in a 4.2-year study of 1017 patients with type 2 diabetes who were randomized to receive glargine vs neutral protamine Hagedorn (NPH) insulin.⁹

Insulin to Lower Glucose Levels. Several large trials of different glycemic targets in patients with diabetes (Table) have not reported any effect on cancer deaths or hospitalizations. More insulin was used in the treatment group compared with the control group in all of these trials. Moreover, a recent independent data monitoring committee review of data from an ongoing large randomized trial using glargine insulin did not identify any safety concerns.¹⁰

Cancer cells are obligate glucose users and dysglycemia is a cancer risk factor. Randomized trials have not identified a consistent relationship between insulin therapy and cancers and are compatible with the hypothesis that glucose lowering by insulin, insulin analogs, or both may reduce, increase, or have no effect on cancer risk. Incident cancers should therefore be measured in prospective trials of glucose-lowering drugs in patients with dysglycemia.