Optimizing clinical outcomes resulting from glucose‐lowering therapies in type 2 diabetes: increased confidence about the DPP‐4 inhibitors and continued concerns regarding sulphonylureas and exogenous insulin

Abstract

We each decide to take medication after making a judgement about the relative benefits and risks associated with the drug. People with life-threatening diseases tolerate potentially fatal drugs after weighing up the available information characterizing the risks and benefits. No chemical is completely safe; even excessive intake of water is potentially lethal. With regard to new classes of drug, at product launch we have only a rudimentary insight into the risks and benefits of new medicines but we definitely do not at that time understand with confidence their relative merits. The beneficial and adverse consequences of treatment can take a long time to elicit. For instance, an adverse event may occur with very low incidence, or there may be a long time lag between exposure and adverse outcome (or benefit). In diabetes, we conventionally now infer long-term benefit by accepting a proxy measure of outcome; typically a short-term reduction in glucose control. Evidence-based medicine in diabetes is still largely related to glucose control. With regard to the most commonly prescribed glucose-lowering medications in type 2 diabetes, how has evidence-based medicine impacted prescribing? We prescribe sulphonylurea (SU) in very large quantities, even though it has been argued that SUs cause relative harm compared with metformin 1-3. SUs carry a ‘black box’ warning in relation to increased risk of cardiovascular events in the United States. Simultaneously we are stopping or reducing the amount of prescribing for glucose-lowering drugs that are arguably relatively good, such as pioglitazone 4-6. In our own data, rosiglitazone is a better alternative to SU in combination with metformin 7. The glitazones are now off-patent in most countries, so cost considerations are less of a factor in a decision to prescribe this class of drugs. In this edition of Diabetes, Obesity and Metabolism, Mogensen and colleagues publish findings from an important study of outcome resulting from treatment with alternative regimens of combination, glucose-lowering therapies in people with type 2 diabetes 8. Their study utilized national data from Denmark. The Scandinavian countries have excellent population data with which to study drug-related outcomes. However, population data are never perfect in this regard – no data are – and the analytical techniques that we use have inherent flaws and complexities. There are now two incretin mimetic classes of glucose-lowering drugs available; the DPP-4 inhibitors and the GLP-1 agonists. Long-term population data are only now becoming available that provide a general picture of the overall risk balance of these products; the most important outcome of all being all-cause mortality. Data from Mogensen and colleagues, as well as from our own related study in this journal 9 add to other data to suggest that not only are the DPP-4 inhibitors safe in that they do not increase overall harm, but they may also confer survival benefit. The Danish investigators report an adjusted relative risk (RR) of all-cause mortality for the DPP-4 inhibitors in combination with metformin versus SU plus metformin of 0.65 (0.54–0.80). The most directly comparable, corresponding, adjusted value from our related study was 0.74 (0.58–0.93). It is always reassuring when these novel findings are resonant. Importantly, all-cause mortality as an outcome singularly takes into account the impact of all of the benefits and postulated adverse events relating to this class of drug, such as pancreatitis 10 and thyroid cancer 11. Their findings for the comparison between GLP-1 agonist plus metformin and SU plus metformin were inconclusive in terms of showing any overall difference in outcomes (adjusted RR of all-cause mortality 0.77, 0.51–1.17). Why the apparent beneficial impact on mortality of combination therapy with the DPP-4 inhibitors plus metformin which is evident in observational population data is not being reflected in phase IV randomized controlled trial (RCT) data remains a matter of conjecture. Of more concern in the report from Mogensen and colleagues are their data relating to treatment with exogenous insulin plus metformin in those with type 2 diabetes. They report an adjusted RR of all-cause mortality when treated with insulin plus metformin combination therapy versus SU plus metformin of 1.95 (1.70–2.25): essentially a doubling of mortality risk. In a previous observational study we reported a comparable adjusted value of 1.46 (1.34–1.59) 12. In the same statistical models, a high relative risk of all-cause mortality associated with insulin plus metformin in comparison with GLP-1 agonist plus metformin was also reported. This comparison would be regarded by some commentators as involving more directly comparable, late-stage glucose-lowering regimens. Similarly, distressing findings have been reported by Klein and colleagues in relation to cardiovascular adverse outcomes when using data from studies of exenatide versus insulin (adjusted RR of stroke or myocardial infarction 0.55, 0.40–0.76) 13. A pattern of increased risk of adverse outcome in people with type 2 diabetes treated with exogenous insulin in type 2 diabetes has been reported by us and in other studies 14-19. There is also evidence of a dose–response association 20, although metformin attenuates mortality risk versus insulin only regimens in these patients. The reasons why insulin may cause adverse outcome in type 2 diabetes are many-fold, and include increased weight gain 21, increased risk of hypoglycaemia 22, arrhythmias 23, increased insulin resistance 24, hyperinsulinaemia 25, increased vascular inflammation 26, changes in vascular haemodynamics 27, 28, coagulopathy 29 and mitogenicity 30. The launch of insulin degludec has been delayed in the United States because of cardiovascular safety concerns 31. Unfortunately, the report by Mogensen and colleagues omits details of these important findings relating to the use of insulin from the title, abstract and the discussion. Most recently, an article in JAMA also reports that people treated with metformin plus insulin combination therapy versus metformin plus SU combination therapy subsequent to metformin monotherapy had increased risk of cardiovascular events and all-cause mortality 32. Much of the historic data concerning insulin has been distilled in a number of meta-analyses published in recent years relating to cardiovascular disease and cancer in type 2 diabetes. Two meta-analyses have investigated the association between insulin and all-cause mortality or cardiovascular events. Monami and colleagues combined five RCTs and found that people with type 2 diabetes using SUs had an increased risk of all-cause mortality [odds ratio (OR) 1.80, 95% CI 0.45–7.26] and cardiovascular death (OR 1.73, 0.38–7.88) compared with insulin but these results were not statistically significant 33. Two RCTs were combined to determine that the risk of major cardiovascular events (MACE) was also not significantly greater for people using SUs versus insulin (OR 0.98, 0.80–1.20) 33. A Cochrane systematic review of RCTs found that, compared with insulin, there was no significant difference in the risk of all-cause mortality (1.18, 95% CI 0.88–1.59, and 0.96, 0.79–1.18, for first and second generation SUs, respectively) and cardiovascular death (1.36, 0.68–2.71, and 0.96, 0.73–1.28, for first and second generation SUs, respectively) 34. Six meta-analyses investigated the association between insulin use and the risk of cancer in type 2 diabetes 34-39. However, the Outcome Reduction with Initial Glargine Intervention (ORIGIN) trial (published later) is one of the few RCTs to report the effect of insulin on the risk of cancer and cardiovascular outcomes. ORIGIN found that compared with standard treatment, the use of insulin glargine over a six-year period had a neutral effect on cardiovascular outcomes [1.02, 95% CI 0.94–1.11 for cardiovascular death, non-fatal myocardial infarction or non-fatal stroke] and cancer (1.00, 0.88–1.13) 40. In terms of site-specific cancers, insulin glargine was found to have a neutral effect on the risk of breast, colon and prostate cancers 40. A Cochrane systematic review of RCTs conducted by Hemmingsen and colleagues also found no significant increased risk of cancer associated with insulin compared with SUs in type 2 diabetes (RR 0.81, 95% CI 0.29–2.27, and 0.95, 0.61–1.49, for first and second generation SUs, respectively) 34. The increased risk of cancer and cancer-related death associated with insulin use was also not significant in the meta-analysis conducted by Janghorbani and colleagues 38. In a meta-analysis of cohort and case–control studies, Colmers and colleagues found that new or ever-use of insulin was not associated with an increased risk of breast, prostate or colorectal cancer versus no insulin use 37. The lack of an association between insulin and the risk of prostate cancer is not that surprising considering that diabetes is associated with a reduced risk of prostate cancer, possibly because of lower circulating levels of testosterone 41, 42. Conversely, Wang and colleagues analysed four studies (one case–control and three cohort studies) and found that, compared with no insulin use, insulin use in type 2 diabetes was associated with an increased risk of colorectal cancer 36. In addition, a significantly increased risk of hepatocellular cancer was found by Singh and colleagues (RR 2.61, 95% CI 1.46–4.65) 36. However, ORIGIN did not investigate pancreatic and hepatocellular cancers specifically. The results from the meta-analyses from Monami and colleagues and Hemmingsen and colleagues determined that, compared with SUs, the risks of death from all causes, cardiovascular death and cardiovascular events were not significantly different for insulin, a finding similar to the one reported from ORIGIN 33, 34. However, it is important to note that the comparison groups were not identical (SUs versus standard care). Almost all of the existing meta-analyses characterize SUs versus insulin because trials that look at other alternatives do not exist. Furthermore, the insulin dose of patients included in the ORIGIN study was relatively low after six years (median 0.40 U/kg; interquartile range: 0.27–0.56) and there was considerable cross-contamination of other glucose-lowering therapies between groups (47% of the insulin glargine group also received metformin and 11% of patients receiving standard care were treated with insulin) 40. SUs stimulate insulin secretion and can cause weight gain and hypoglycaemia and may therefore adversely affect cardiovascular risk. Some SUs may also impair a cardioprotective mechanism known as ischaemic preconditioning 43. In addition, epidemiological data have shown that, compared with metformin, all-cause mortality and MACE were raised for both SUs (adjusted hazard ratio 1.75, 95% CI 1.64–1.86 and 1.39, 1.25–1.55, respectively) and insulin monotherapy (2.20, 1.98–2.43 and 1.74, 1.44–2.09, respectively) 14. Therefore a similar risk of cancer, all-cause mortality and cardiovascular events associated with insulin and SUs cannot and should not be used as an indication that insulin is safe. One meta-analysis by Franciosi and colleagues 39 compared metformin with insulin and found that the use of metformin was associated with a non-significant reduction in the risk of colorectal cancer (OR 0.75, 95% CI 0.43–1.31) and a significant reduction in the risk of pancreatic cancer (OR 0.24, 0.18–0.32). However, further RCTs comparing insulin use with other glucose-lowering agents are necessary. What now? UKPDS demonstrated that the intensive control of blood glucose in type 2 diabetes using SUs, insulin or metformin reduced the risk of microvascular but not macrovascular complications 44. However, only after the 10 year follow-up was a statistically significant reduction in the risk of myocardial infarction and all-cause mortality observed 45. Since then, other RCTs have been conducted to investigate the benefits of intensive blood glucose control but have failed to show cardiovascular benefit 46-48. The benefits associated with improved glycaemic control not only rely on age at diagnosis and HbA1c levels but also on a patients view of the burdens of treatment 49. In addition, epidemiological evidence has shown that the risk of all-cause mortality is associated with either too low or too high blood glucose levels 12, 50. Therefore, as the benefits of intensive glucose control remain uncertain, there remains a need for large RCTs to investigate the effect of insulin on hard endpoints such as cancer, all-cause mortality and cardiovascular events rather than HbA1c levels. Despite this, available data from all facets of the biomedical sciences, from population data to cellular/molecular biology 51, are sufficient to promote this matter as a serious cause for concern. Although conducting a randomized study raises ethical questions 52, it is now vital that studies are somehow carried out to provide a conclusive understanding of this matter. ORIGIN was never designed to answer this specific question 40. We have reproduced very similar findings to those reported from ORIGIN, and we can show that people with low-dose exogenous insulin plus metformin combination therapy have the same level of mortality risk as those treated with metformin plus SU combination therapy (full data under review but a related presentation at the EASD is available online) 53. Therefore, there is a way of using exogenous insulin relatively safely in type 2 diabetes. However, a study designed specifically to look at the overall safety of insulin in people with type 2 diabetes would look very different to ORIGIN. The issue of insulin and SU safety is contentious: esteemed scientists with an interest in diabetes have conflicting opinions 54 and there are discordant findings from observational and randomized studies. It is therefore vital that further investigations be carried out and that we strive to understand where and why such differences in outcome occur. It is unacceptable that there remains so much uncertainty about an issue of such importance. C. C. is a director of and S. H. is employed by Pharmatelligence, a research consultancy receiving funding from pharmaceutical companies. S. H. receives a fully funded PhD studentship from Cardiff University and is employed by Alliance Boots. C. C. has received research grants from various health-related organizations, including Abbott, ALK, Astellas, Diabetes UK, the Engineering and Physical Sciences Research Council, the EASD, Ferring, GSK, Jenson (Internis), Lilly, the Medical Research Council, Medtronic, MSD, the National Health Service, Norgine, Pfizer, Sanofi Aventis, Shire and Wyeth, and has consulted for Amylin, Aryx, Astellas, Boehringer Ingelheim, BMS, Diabetes UK, Eisei, Ferring, Grunenthal, GSK, Ipsen, Lilly, Norgine, Medtronic, MSD, Pfizer, Sanofi-Aventis and Takeda.