Epidemiologic studies linking insulin glargine and glucose-lowering therapies to cancers and n-3 fatty acids to cancer prevention have not been confirmed. We aimed to assess the effect of insulin glargine and n-3 fatty acids on incident cancers within the context of the ORIGIN (Outcome Reduction with Initial Glargine Intervention) trial.
The ORIGIN trial is an international, long-term, randomized two-by-two factorial study comparing insulin glargine with standard care and n-3 fatty acids with placebo (double blind) in people with dysglycemia at high risk for cardiovascular events. The primary outcome measure (cancer substudy) was the occurrence of any new or recurrent adjudicated cancer. Cancer mortality and cancer subtypes were also analyzed.
Among 12,537 people (mean age 63.5 years, SD 7.8; 4,388 females), 953 developed a cancer event during the median follow-up of 6.2 years. In the glargine and standard care groups, the incidence of cancers was 1.32 and 1.32 per 100 person-years, respectively (P = 0.97), and in the n-3 fatty acid and placebo groups, it was 1.28 and 1.36 per 100 person-years, respectively (P = 0.39). No difference in the effect of either intervention was noted within predefined subgroups (P for all interactions ≥0.17). Cancer-related mortality and cancer-specific outcomes also did not differ between groups. Postrandomization HbA1c levels, glucose-lowering therapies (including metformin), and BMI did not affect cancer outcomes.
Insulin glargine and n-3 fatty acids have a neutral association with overall and cancer-specific outcomes, including cancer-specific mortality. Exposure to glucose-lowering therapies, including metformin, and HbA1c level during the study did not alter cancer risk.
Both type 2 diabetes mellitus (hereafter referred to as diabetes) and cancer are common diseases that are rising in incidence and prevalence throughout the world. Moreover, epidemiologic data suggest that diabetes is associated with an increased risk of several different cancers (1). Reasons for this association may include metabolic features typical of diabetes such as hyperglycemia, insulin resistance, and/or hyperinsulinemia; glucose-lowering oral agents or exogenous insulin; various cardioprotective therapies used frequently in people with diabetes (including blood-pressure-lowering, lipid-lowering, and antiplatelet therapies and various “health” supplements); or antecedent factors that may increase the risk of both diabetes and cancer such as obesity or environmental or genetic factors. Although no clear explanation has emerged, interest in the diabetes–cancer relationship was recently heightened by some epidemiologic analyses linking insulin in general and basal insulin glargine in particular to incident cancers (2), linking sulfonylureas to the development of cancers, and linking metformin to protection from cancers (3). Other epidemiologic findings have suggested that n-3 polyunsaturated fatty acid intake may reduce the risk of colorectal, prostate, breast, or other cancers (4,5). However, these epidemiologic analyses are unable to reliably determine whether the observed relationships were due to these drugs or to confounding factors associated with both the use of these drugs and cancers. Such a distinction is best made within a large randomized trial, as the randomization process distributes all suspected and unsuspected confounders equally between the groups that are allocated to either receive or not receive the study drug.
The ORIGIN (Outcome Reduction with Initial Glargine Intervention) trial allocated 12,537 people with either prediabetes or diabetes to receive either titrated basal insulin glargine targeting a normal fasting plasma glucose ≤95 mg/dL (5.3 mmol/L) or standard care; 12,536 of these individuals were also allocated to either 1 g of ethyl esters of n-3 fatty acid or placebo. Both interventions had a neutral effect on the primary and other cardiovascular outcomes (6,7). They also had a neutral association with incident cancers (6,7). Herein we present a detailed analysis of the association of both interventions with cancer outcomes overall and within key subgroups (prespecified analyses) and explore the potential effects of glycemic control and other glucose-lowering therapies on these outcomes.
Research Design and Methods
Detailed descriptions of the ORIGIN trial design and results have been published previously (6–8). ORIGIN was designed and conducted by an international steering committee of academic investigators in 40 countries and was funded by Sanofi, which also provided regulatory support, site monitoring, and insulin glargine (Lantus). Pronova BioPharma Norge supplied the n-3 fatty acid supplements and placebo. The study was approved by the ethics committee at each study site, and all participants provided written informed consent.
Eligibility criteria included age of at least 50 years; a diagnosis of diabetes on no or one oral glucose-lowering drug, impaired glucose tolerance, or impaired fasting glucose based on one oral glucose tolerance test; and either a prior cardiovascular event or evidence of vascular disease (angina with ischemia, albuminuria, left ventricular hypertrophy, coronary, carotid, or lower-limb arterial stenosis of ≥50% or an ankle-brachial index <0.9). People with expected survival of less than 3 years for noncardiovascular causes such as cancer were excluded. Eligible participants were randomized to the two interventions using a 2 × 2 factorial design between September 2003 and December 2005 and then seen at 0.5, 1, 2, and 4 months and then every 4 months until the study ended after a median (interquartile range) follow-up of 6.2 (5.8–6.7) years. At each visit, the protocol was reinforced and all primary, secondary, and other outcomes were ascertained.
The two coprimary outcomes for the glargine trial were 1) cardiovascular death, nonfatal myocardial infarction, or nonfatal stroke and 2) any of these or heart failure hospitalization or myocardial revascularization; the primary outcome for the n-3 fatty acid trial was cardiovascular death. All cancers requiring hospitalization were ascertained from the time of randomization until the end of the trial. Moreover, starting January 2010, all participants were asked about any cancer events that had occurred since the time of randomization that did not require hospitalization and were then asked about incident cancers at each subsequent visit. Supporting documentation for all cancers since the time of randomization was sent for adjudication according to prespecified definitions by a committee whose members were unaware of study group assignments (i.e., masked to treatments). Cancers were classified as either definite or probable or possible, by primary anatomic site, by status (i.e., new or recurrent), and by clinical consequence (i.e., death or hospitalization). The first occurrence of any cancer since the date of randomization using all adjudicated cancer cases, including definite and probable cancer cases, was used for the analyses. Detailed definitions of cancer outcomes are provided in the supplement to the main paper (6,7).
Cancer data were analyzed using SAS software (version 9.1 for Solaris) according to an intention-to-treat approach. Tests for differences in baseline characteristics between individuals who did and did not develop a new or recurrent cancer were performed with χ2 and t tests for discrete and continuous variables, respectively. Time-to-event curves were constructed and compared using stratified log-rank tests. Hazard ratios were calculated with the use of Cox regression models with factorial allocation, baseline diabetes status, and a history of a cardiovascular event before randomization as covariates. The effect of the intervention in baseline subgroups defined according to age, gender, concomitant pharmacotherapy at baseline (metformin, sulfonylurea, acetylsalicylic acid, other antiplatelet), BMI, prior cardiovascular event, smoking history, and factorial allocation was calculated using similar Cox regression models, and differences in effect by subgroup were assessed by including an interaction term in the model. The effect of postrandomization addition of metformin or metformin plus a sulfonylurea and the effect of the metformin dose and total daily dose of any insulin, postrandomization HbA1c, and weight were assessed by including these as time-varying covariates in Cox regression models that also included current smoking at baseline as a covariate. P values <0.05 were considered as nominally significant, with no adjustments for multiple tests.
As reported previously, the 12,537 participants’ mean (SD) age was 63.5 (7.8) years, and 35% were female; 48% of participants were recruited in Europe, 31% in Latin America, and 11% in North America. The 82% of all participants with a prior diagnosis of diabetes had a mean (SD) duration of 5.4 (6.0) years of diabetes. In participants randomized to glargine, the median insulin dose was 0.31 units/kg body weight (interquartile range, 0.19–0.46) at 1 year and 0.40 units/kg (interquartile range, 0.27–0.56) at 6 years. By the end of the trial, insulin glargine was permanently discontinued in 19.4% of the study population (7). Participants randomized to n-3 fatty acid ingested a 1-g capsule containing at least 900 mg (90% or more) of ethyl esters of n-3 fatty acids daily. Baseline dietary eicosapentaenoic–docosahexaenoic acid intake was similar between groups (median 210.0 vs. 209.3 mg/day in the n-3 fatty acids and placebo groups, respectively). The rates of adherence to n-3 fatty acids/placebo were similar in the two groups (6).
Cancer Outcomes Overall and According to Glargine and n-3 Fatty Acid Allocation
Of all participants, 953 (7.6%) developed a new or recurrent cancer during the median follow-up of 6.2 years at an incidence of 1.32/100 person-years. The majority of cancer cases were new (n = 906 cases) as opposed to recurrent (n = 47 cases). Compared with those who did not develop cancer during follow-up (Table 1), those who did develop a cancer were older, had a higher frequency of smoking, had a higher frequency of alcohol intake, had a previous cardiovascular disease event, had a new diagnosis of diabetes, had a lower HbA1c level, and used statin, aspirin, and oral glucose-lowering agents more often.
Kaplan–Meier curves for all cancer, breast cancer, and colorectal cancer are summarized in Fig. 1A–C for glargine and Fig. 1D–F for n-3 fatty acids. Based on randomization allocation, all Kaplan–Meier curves were overlapping (all P ≥ 0.39). The unadjusted incidence and adjusted hazard of cancers in participants on glargine versus standard care and n-3 fatty acids versus placebo are shown in Fig. 2A and B, respectively. Insulin glargine had a neutral association with any cancers (new cancers or recurrent cancers), cancer-related mortality, and all the various types of cancer (all P ≥ 0.27) analyzed, with no evidence of different effects according to key clinical subgroups (interaction P values all ≥0.17). Similar results were noted for n-3 fatty acids (all P ≥ 0.06) and when analyses were repeated with baseline smoking status in the Cox models.
Effect of Postrandomization HbA1c Levels, Glucose-Lowering Therapies, and Weight
Metformin was used at baseline in 27.0% and 27.8% of participants in the glargine and standard care groups, respectively, and in 46.5% and 59.7%, respectively, of these groups by the end of the trial. The mean (SD) daily dose of metformin was 1,372.89 (625.46) and 1,358.66 (607.83) mg at randomization in the glargine and standard care groups, respectively, and 1,376.92 (666.81) and 1,558.76 (705.83) mg in the glargine and standard care groups, respectively, by the end of the trial. Inclusion of the use or dose of metformin either at baseline or after randomization in the Cox models did not affect the estimate of the effect of either glargine (Table 2) or n-3 fatty acids (data not shown) on incident cancers, on breast cancers in women, or on colorectal cancers. Similar findings were noted in models that included the total daily dose of any insulin, the use of metformin plus a sulfonylurea, the baseline and postrandomization HbA1c levels, and the baseline and postrandomization weight. As noted in Table 2, neither metformin use, HbA1c level, nor weight affected the incidence of cancer outcomes in this population.
Perspectives and Clinical Implications
The ORIGIN study provides timely and important data on the risk of cancer in patients with impaired fasting glucose, impaired glucose tolerance, or diabetes and at high risk for major cardiovascular events. In this study, daily exposure to a dose of glargine sufficient to normalize fasting plasma glucose levels for a median of 6.2 years had a neutral association with any cancers, new cancers, recurrent cancers, cancer-related mortality, and various types of cancer, with no evidence of different effects according to key clinical subgroups. These results are consistent with meta-analyses of small trials of insulin glargine that showed no relationship between the use of glargine and development of cancers (9), with a recent French cohort study based on national administrative databases (10) and with a trial of insulin glargine’s effect on diabetic retinopathy (11). They are also consistent with the lack of effect of intensive glucose lowering on cancers reported in large outcomes trials in which more insulin was used in the intensive groups (12,13). These findings of the ORIGIN trial provide no support for previous suggestions of a link between cancer and short-term use of either basal insulin in general or basal insulin glargine in particular (3,14–16).
In the ORIGIN trial, long-term exposure to n-3 fatty acid did not influence cancer outcomes, including types of cancers, new cancers, recurrent cancers, or cancer-related mortality, and there was no evidence of different effects within key clinical subgroups. Data on the use of n-3 fatty acids and cancers have been inconsistent and were mainly based on experimental models and observational studies (5). Data from several preclinical models suggest that n-3 fatty acids have an antitumor effect exerted through a variety of mechanisms such as inhibition of cellular proliferation and apoptosis. Attenuation of inflammatory pathways has also been implicated (17,18). Epidemiological data are inconsistent. Observational studies suggest a protective effect of n-3 fatty acids against colon, prostate, and other cancers (19,20). Its role in breast cancer is unclear. In a recent meta-analysis, higher consumption of dietary marine n-3 fatty acids was associated with a lower risk of breast cancer (21). The ORIGIN trial provides evidence that n-3 fatty acids are neither beneficial nor detrimental to the development and progression of cancers in general. Future studies should address the role of n-3 fatty acids on specific cancers or in at-risk groups.
It is notable that no relationship between HbA1c levels (a measure of glycemic control) and cancer outcomes was observed despite evidence that elevated blood glucose concentration may mediate cancer risk (22). This could be, in part, due to the fact that in the ORIGIN trial, glucose levels were closely monitored and optimized, and as such, HbA1c was narrowly distributed throughout the study, limiting the ability to detect any effect of HbA1c on cancer outcome. It is also notable that exposure to metformin during the ORIGIN trial did not modulate cancer outcomes. Biguanide metformin, the most commonly used oral glucose-lowering agent, has been associated with a reduced risk of cancer (3,23–26) and cancer mortality (27) in a number of observational human studies. In contrast, a recent meta-analysis of available randomized clinical trials data did not find a statistically significant beneficial effect of metformin on cancer outcomes and all-cause mortality in adults allocated to metformin (28).The role of metformin in the prevention and treatment of cancer is currently under study (29,30). Although body weight and BMI have been independently associated with an increased risk of incident cancer (31), the ORIGIN trial revealed no effect of body weight on cancer outcomes.
These analyses have several strengths. They include the prospective randomized nature of the trial, large sample size, 6.2 years of follow-up, and large number of cancer outcomes. Indeed, with 953 cancer outcomes, this trial had 90% power to detect a 20% or greater decrease or increase in the incidence of cancers with either therapy. Strengths also include the standardized approaches to data collection, blinded adjudication of cancer outcomes, and detailed regular collection of potential confounding factors. Limitations include the low incidence of specific types of cancer and a median follow-up period of 6.2 years; more prolonged follow-up of these participants during the ongoing passive follow-up phase of ORIGIN (entitled ORIGIN and Legacy Effects [ORIGINALE]) should help address these concerns. Finally, the absence of a normoglycemic control group limits the ability to assess the effect of potential mediators of cancer risk such as HbA1c, weight, and drug use on cancers and makes it impossible to compare cancer outcomes to those in the general population.
The ORIGIN trial’s finding that insulin glargine has a neutral association with cancer outcomes is reassuring in light of the growing need for basal insulin therapy throughout the world. The finding that supplementation with n-3 fatty acid also had a neutral association with all cancer outcomes plus previously reported findings that it had a neutral effect on other clinically important outcomes should lead to a reevaluation of the widespread use of these supplements.
Clinical trial reg. no. NCT00069784, clinicaltrials.gov.
Duality of Interest. The ORIGIN trial was funded by Sanofi. N.Y. received grant support from Sanofi to her research institution. G.R.D. received payment for serving on a data and safety monitoring board from Sanofi and payment for serving on a steering committee from Eli Lilly. J.R. received consulting fees and research grants from Sanofi, Eli Lilly, and Novo Nordisk. J.P. received contract support and payment for serving an advisory committee for Sanofi. L.E.R. received lecture and/or consultation fees from Roche, Sanofi, Bristol-Myers Squibb, and AstraZeneca and grant support from the Swedish Heart-Lung Foundation, AFA Insurance, Karolinska Institutet, and Roche. V.P. received a research grant from the Latvian Council of Science and lecture and/or consultation fees from Merck Sharp & Dohme (MSD), Takeda, and Glenmark Pharmaceuticals. G.A.S. received research support from Novo Nordisk and performed advisory board activities with AstraZeneca/Bristol-Myers Squibb, Medtronic, Lilly, MSD, Novo Nordisk, Novartis, Daiichi Sankyo, and Sanofi. K.I.B. received lecture and/or consultation fees from Novo Nordisk, Eli Lilly, Sanofi, Bristol-Myers Squibb, Merck, Novartis, Pfizer, and Boehringer Ingelheim. R.E.R. is a member of the steering committee of the ORIGIN trial. M.K. is a member of the steering committee of the ORIGIN trial and received travel support and honorarium from Sanofi. M.C.R. received grant or research support from AstraZeneca, Amylin, Bristol-Myers Squibb, Eli Lilly, Sanofi; consultation fees from AstraZeneca, Amylin, Bristol-Myers Squibb, Eli Lilly, Sanofi, Valeritas, and Elcelyx; and lecture honoraria from Sanofi. S.Y. received consulting and lecture fees and grant support from Sanofi. H.C.G. received consulting fees from Sanofi, Novo Nordisk, Lilly, Bristol-Myers Squibb, Roche, AstraZeneca, Novartis, GlaxoSmithKline, and Bayer; lecture fees from Sanofi and Bayer; and support for research or continuing education through his institution from Sanofi, Lilly, Merck, Novo Nordisk, Boehringer Ingelheim, Bristol-Myers Squibb, and AstraZeneca. No other potential conflicts of interest relevant to this article were reported.
Sanofi had no influence on the design and conduct of the study; collection, management, analysis, and interpretation of the data; or preparation, review, or approval of the manuscript.
Author Contributions. L.B. and H.C.G. developed the study concept and design, performed the analysis and interpretation of data, drafted the manuscript, performed critical revision of the manuscript for important intellectual content, and performed statistical analysis. N.Y., G.R.D., J.P., L.E.R., M.C.R., J.B., and S.Y. developed the study concept and design and performed critical revision of the manuscript for important intellectual content. J.R., P.C.Y., V.P., G.A.S., K.I.B., R.E.R., J.A.M.-N., and M.K. performed critical revision of the manuscript for important intellectual content. L.B. and H.C.G. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Prior Presentation. This study was presented in abstract form at the 73rd Scientific Sessions of the American Diabetes Association, Chicago, IL, 21–25 June 2013.