March 2014 • Volume 59 •

Diabetes

Voice

March 2014

 

•

 

Volume 59

 

•

 

Issue 1

40

coUntry perspectiVes

More and more frequently insulin 

is being recommended as an 

‘add-on’ to oral hypoglycaemic 

therapy for the achievement of 

blood glucose targets in people 

with established type 2  

diabetes. Indeed, there are 

now trials of insulin therapy in 

type 2 diabetes from diagnosis. 

Concerns have been raised in 

the recent medical literature 

that long-term insulin therapy 

in type 2 diabetes increases 

the risk of cardiovascular 

disease and some cancers. 

We have asked specialists in 

the fields of clinical diabetes 

and pharmacoepidemiology to 

comment on the question: 

DEBATE:  

long-term  

safety of insulin  

in type 2 diabetes

Insulin therapy in people with type 2 diabetes: 

is it safe in terms of the risk of cardiovascular 

disease, cancer and all-cause mortality?

NO

Sarah Holden and Craig Currie

Insulin has an unlimited potential 

to lower blood glucose and is a well-

established treatment for type 2  

diabetes. The International Diabetes 

Federation (IDF) recommends that 

it should be used as an optional 

third line when a combination of 

metformin, where indicated, and 

one other glucose-lowering medi-

cation has failed to adequately con-

trol blood glucose. ADA and EASD 

guidelines recommend a patient-

centred approach with the aim of 

achieving adequate glucose control 

while minimising side effects.

Two common side effects associated 

with insulin injections are weight 

clinical care

Diabetes

Voice

March 2014

 

•

 

Volume 59

 

•

 

Issue 1

41

coUntry perspectiVes

gain and hypoglycaemia. Weight gain 

is associated with an increased risk of 

cardiovascular disease and should be 

minimised in type 2 diabetes. Both 

insulin and hypoglycaemia may have 

vascular effects which are thought 

to be greatest in people with pre-

existing cardiovascular disease and 

diabetes.

1,2

 In addition, as a growth 

factor, insulin may affect cancer pro-

gression.

3

 However, this is a complex 

area where high glucose levels have 

also been linked to increased cancer 

risk.

4

 Some epidemiological studies 

have shown that the use of insulin is 

associated with an increased risk of 

cardiovascular events, cancer and all-

cause mortality in comparison with 

other glucose-lowering therapies.

5,6,7

In one of these studies (a retrospec-

tive cohort study using data from 

the UK General Practice Research 

Database)

6

 mortality and other dia-

betes-related outcomes for just under 

85,000 individuals were examined in 

relation to five diabetes therapies – 

monotherapy with either metformin, 

sulfonylurea or insulin, metformin 

plus sulfonylurea or insulin plus met-

formin. Treatment with insulin alone 

conferred a statistically significantly 

increased risk of a first major cardiac 

event or a first diagnosis of cancer 

ranging from 1.8 to 2.6 times the risk 

seen in those treated with metformin 

alone (the comparison group). This 

excess risk was lower in the group 

treated with insulin plus metformin 

than in the insulin monotherapy 

group, though was still significant at 

1.3 times the risk of the comparison 

group. However, even though these 

results are consistent and despite the 

use of statistical adjustment, obser-

vational data such as these should be 

interpreted with caution due to the 

risk of a form of analytical bias that 

is termed ‘confounding by indication’. 

This form of confounding is a com-

mon feature of studies of outcomes in 

relation to drug and other therapies 

simply because the reasons patients 

have been prescribed the therapies 

may themselves be related to the per-

ceived risks of any particular outcome 

occurring. In other words, people may 

be prescribed insulin partly because 

they are perceived of being at risk to 

adverse outcomes, including risk of 

cardiovascular disease.

In contrast, large randomised con-

trolled trials such as Action to Control 

Cardiovascular Risk in Diabetes 

(ACCORD) found no adverse safety 

signals associated with the use of in-

sulin.

8

 However, these studies were 

designed to assess the benefits of in-

tensive glucose control rather than the 

safety of insulin, and subjects could re-

ceive multiple glucose-lowering thera-

pies making individual comparisons 

difficult. Also, the Outcome Reduction 

with an Initial Glargine Intervention 

(ORIGIN) trial demonstrated that in-

sulin glargine had a neutral impact on 

cardiovascular outcomes and cancers 

compared with standard treatment.

9

 

However, it should be noted that peo-

ple included in ORIGIN were newly 

diagnosed with type 2 or impaired 

glucose tolerance, impaired fasting 

glucose or only using one glucose-

lowering therapy. In addition, by the 

end of the study, 65% of the insulin 

glargine group were also using other 

glucose-lowering agents, including 

47% using metformin.

When used in combination, metform-

in may attenuate any risks associated 

with insulin. Metformin is thought to 

protect against cancer and have car-

dio-protective effects that cannot be 

fully explained by its ability to lower 

blood glucose.

10

 When used in con-

junction with insulin, metformin has 

been associated with similar glucose 

control, but lower insulin doses and 

less weight gain.

11

 In addition, relative 

to the use of insulin alone, the use of 

metformin in combination with insu-

lin has been associated with a reduced 

risk of cardiovascular events, cancer 

and death from any cause.

6

 Current 

ADA, EASD and IDF guidelines ad-

vocate that, when starting insulin, 

it should be added to existing met-

formin therapy and not replace it.

Any potential risks associated with 

insulin therapy need to be seen in the 

context of its clear benefits for achiev-

ing good glucose control. However, the 

shortage of randomised controlled tri-

als examining the risks and benefits of 

using insulin on long term clinical out-

comes such as cardiovascular events, 

cancer and death from all causes needs 

to be addressed in order to provide 

more evidence on the safety of insulin 

in people with type 2 diabetes.

In the UK, the Medical and Healthcare 

Products Regulatory Agency will soon 

report on the safety of insulin in people 

with type 2 diabetes. Regardless, there 

are question marks about the safety 

of insulin in type 2 diabetes. There 

is, therefore, a need to show caution 

while our lack of understanding of 

this potential problem is addressed, 

and we can recommend injection of 

insulin in people with type 2 diabetes 

with confidence.

clinical care

Diabetes

Voice

March 2014

 

•

 

Volume 59

 

•

 

Issue 1

42

Insulin is an established glucose-lower-

ing therapy for both type 1 and type 2 

diabetes. However, insulin is a growth 

factor. It is administered in an un-phys-

iological manner and it is present in the 

circulation at levels that are far higher 

than in the non-diabetic population. For 

these reasons, concerns regarding insu-

lin safety have been long-standing and 

this has led to them being extensively in-

vestigated and, to my mind, repudiated.

First the concerns regarding cancer: 

these emerged following suggestions 

that the long-acting insulin analogue, 

glargine, increased the risk of cancer

12

 

and were supported by papers suggest-

ing that insulin or insulin secretagogues 

were associated with a higher cancer 

risk.

7

 However, slowly the picture be-

came less certain. Scrutiny of the orig-

inal report showed that the patients 

receiving glargine were on tiny doses 

of insulin, the cancer risk disappeared 

completely if they were on any other 

glucose lowering medications (includ-

ing other insulins) and the increased 

risk was only seen after an 'adjustment' 

by the authors.

12

 Attempts to replicate 

the original findings floundered, despite 

analyses of huge data sets. Finally a pro-

spective randomised clinical trial (RCT) 

was published which demonstrated that 

YES

Steve Bain

exogenous insulin therapy (with glar-

gine) over more than six years' follow-

up was associated with no evidence of 

increased risk of cancer.

9

 So, concerns 

raised by pharmacoepidemiology, which 

can only identify possible safety signals 

and generate hypotheses, were laid to 

rest by an RCT.

The case of cardiovascular disease has 

been longer and more convoluted. Once 

again, insulin had been implicated be-

cause of its potential to act as a growth 

factor and thereby promoting and/or 

enhancing the development of athero-

ma in the circulation. This hypothesis 

seemed to be supported by observations 

from epidemiological studies suggest-

ing an association between hyperinsu-

linaemia and cardiovascular mortality.

13

 

Subsequently, a meta-analysis of data 

from eleven different studies in non-

diabetic men and women concluded 

that hyperinsulinaemia was significantly 

associated with cardiovascular mor-

tality.

14

 However, this does not imply 

causality since the fasting insulinaemia 

seen in these studies may have been a 

consequence of insulin resistance, and 

hence an innocent surrogate.

An RCT was clearly needed – the United 

Kingdom Prospective Diabetes Study 

(UKPDS). In the seminal publication 

of 1998, the introduction clearly stated 

that ‘there is concern that sulphonylu-

reas may increase cardiovascular mor-

tality in patients with type 2 diabetes 

and that high insulin concentrations 

may enhance atheroma formation’. 

UKPDS conclusively demonstrated 

no such increase and, almost, showed 

a benefit from tight control using in-

sulin and the sulphonylureas.

15

 Now 

the hypothesis had changed leading to 

attempts to demonstrate a beneficial 

impact of tight glycaemic control on 

cardiovascular outcomes; ultimately 

this led to the controversy surrounding 

the ACCORD study.

8

ACCORD enrolled middle-aged and 

elderly people with type 2 diabetes and 

a very high risk of cardiovascular dis-

ease. To the researchers' surprise, near-

normal glucose control, achieved with 

the use of multiple drugs, was associated 

with increased all-cause mortality and 

cardiovascular mortality. At five years of 

follow-up, nonfatal myocardial infarc-

tion was reduced, but five-year mortality 

was increased in patients who received 

intensive glucose-lowering therapy. Of 

note, over 75% of the intensive group 

were using insulin at study end.

Meta-analyses were subsequently per-

formed to include all major studies ex-

amining the impact of tight glycaemic 

control on cardiovascular outcomes and 

drew the opposite conclusion from that 

reported in ACCORD.

16

 These trials did 

not compare insulin with non-insulin 

regimens, however, all had higher pro-

portions of insulin users in the intensive 

therapy arms. Given the large numbers 

involved in the studies, one might have 

expected that any intrinsic harmful ef-

fect of insulin would show up as a con-

sistently increased hazard ratio in the 

intensive arms of the trials but it did not.

Cue the pharmaco epidemiologists: 

Currie et al. conducted a retrospective 

database study of 84,622 primary care 

patients, defined a primary endpoint 

of all-cause mortality, incident can-

cer, or major cardiac adverse events, 

and reported hazard ratios (relative 

to metformin monotherapy) of 1.808 

for insulin monotherapy and 1.309 

for insulin plus metformin.

6

 Several 

other observational studies, based on 

databases, supported the association 

between increasing insulin use and 

serious events.

17

 Fortunately, an RCT 

coUntry perspectiVes

clinical care

Diabetes

Voice

March 2014

 

•

 

Volume 59

 

•

 

Issue 1

43

was already in progress which, for most 

people, would settle the argument.

The ORIGIN trial

9

 randomised 12,537 

people with cardiovascular risk factors 

as well as impaired fasting glucose, im-

paired glucose tolerance or type 2 dia-

betes, to receive standard care or insulin 

glargine. The aim was to identify any 

intrinsic benefit on cardiovascular out-

comes from early use of insulin and the 

population of ORIGIN included patients 

who were very well controlled (indeed, 

approximately 10% were non-diabetic), 

with HbA

1c

 values of 6.3% or less in the 

insulin glargine group and 6.5% or less 

in the standard care group. Early use of 

titrated basal insulin had no impact on 

cardiovascular outcomes compared with 

standard guideline-suggested glycae-

mic control. The ORIGIN investigators 

pointed out that a large between-group 

difference in insulin use was achieved, 

and only a small difference in HbA

1c

; 

the results were therefore relevant to 

the effect of insulin therapy rather than 

the effect of glucose lowering on cardio-

vascular outcomes. Once again, an RCT 

had come to the rescue.

The final question is whether the con-

troversies about cancer and cardiovas-

cular disease affect clinical practice. For 

people with type 1 diabetes, insulin is 

currently the only therapy option. For 

patients with type 2 diabetes, the pro-

gressive nature of the condition means 

that, given current therapies, everyone 

will eventually need insulin for symp-

tomatic relief of hyperglycaemia (as-

suming they survive the complications 

of the condition). In the UK at least, 

the average starting HbA

1c

 for insulin 

of over ~9.5%, suggests that neither 

patients nor clinicians feel that insulin 

is an easy or early treatment option. In 

this setting, the debates around safety 

can be seen as rather academic.

Sarah Holden, craig currie and Steve bain

Sarah Holden is PhD student at the Department of Primary Care and Public Health,  

School of Medicine, Cardiff University, UK.

Craig Currie is Professor of Applied Pharmacoepidemiology, The Pharma Research Centre,  

Cardiff Medicentre, Cardiff, UK.

Steve Bain is Professor of Medicine (Diabetes), College of Medicine, Swansea University,  

Swansea, UK.

references

1.    Rensing KL, Reuwer AQ, Arsenault BJ, et al. Reducing cardiovascular disease risk in patients with type 2  

diabetes and concomitant macrovascular disease: can insulin be too much of a good thing?  

Diabetes Obes Metab 2011; 13: 1073-87.

2.    Nordin C. The case for hypoglycaemia as a proarrhythmic event: basic and clinical evidence.  

Diabetologia 2010; 53: 1552-61.

3.    Pollak M. Insulin and insulin-like growth factor signalling in neoplasia. Nat Rev Cancer 2008; 8: 915-28.

4.    Jee SH, Ohrr H, Sull JW, et al. Fasting serum glucose level and cancer risk in Korean men and women.  

JAMA 2005; 293: 194-202.

5.    Currie CJ, Peters JR, Tynan A, et al. Survival as a function of HbA1c in people with type 2 diabetes:  

a retrospective cohort study. Lancet 2010; 375: 481-9.

6.    Currie CJ, Poole CD, Evans M, et al. Mortality and other important diabetes-related outcomes with insulin 

vs other antihyperglycemic therapies in type 2 diabetes. J Clin Endocrinol Metab 2013; 98: 668-77.

7.    Currie CJ, Poole CD, Gale EAM. The influence of glucose-lowering therapies on cancer risk in type 2  

diabetes. Diabetologia 2009; 52: 1766-77.

8.    Action to Control Cardiovascular Risk in Diabetes Study Group, Gerstein HC, Miller ME, et al. 

Effects of intensive glucose lowering in type 2 diabetes. N Engl J Med 2008; 358: 2545-59.

9.    ORIGIN Trial Investigators, Gerstein HC, Bosch J, Dagenais GR, et al. Basal insulin and cardiovascular  

and other outcomes in dysglycemia. N Engl J Med 2012; 367: 319-28.

10.  Zakikhani M, Dowling R, Fantus IG, et al. Metformin is an AMP kinase-dependent growth inhibitor for  

breast cancer cells. Cancer Res 2006; 66: 10269-73.

11.  Goudswaard AN, Furlong NJ, Rutten GE, et al. Insulin monotherapy versus combinations of insulin with oral 

hypoglycaemic agents in patients with type 2 diabetes mellitus. Cochrane Database Syst Rev 2004: CD003418.

12.  Hemkens LG, Grouven U, Bender R, et al. Risk of malignancies in patients with diabetes treated with human  

insulin or insulin analogues: a cohort study. Diabetologia 2009; 52: 1732-44.

13.  Despres JP, Lamarche B, Mauriege P, et al. Hyperinsulinemia as an independent risk factor for ischemic  

heart disease. N Engl J Med 1996; 334: 952-7.

14.  DECODE Insulin Study Group. Plasma insulin and cardiovascular mortality in non-diabetic European men 

and women: a meta-analysis of data from eleven prospective studies. Diabetologia 2004; 47: 1245-56.

15.  UK Prospective Diabetes Study (UKPDS) Group. Intensive blood-glucose control with sulphonylureas or  

insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes  

(UKPDS 33). Lancet 1998; 352: 837-53.

16.  Ray KK, Seshasai SR, Wijesuriya S, et al. Effect of intensive control of glucose on cardiovascular outcomes  

and death in patients with diabetes mellitus: a meta-analysis of randomised controlled trials.  

Lancet 2009; 373: 1765-72.

17.  Östgren CJ, Sundström J, Svennblad B, et al. Associations of HbA

1c

 and educational level with risk of  

cardiovascular events in 32,871 drug-treated patients with type 2 diabetes: a cohort study in primary care.  

Diabet Med 2013; 30: e170-7.

coUntry perspectiVes

clinical care