Ageing Research Reviews 40 (2017) 31–44

Contents lists available at ScienceDirect

Ageing Research Reviews

journal homepage: www.elsevier.com/locate/arr

Metformin reduces all-cause mortality and diseases of ageing

independent of its effect on diabetes control: A systematic review and
meta-analysis
Jared M. Campbell a,b,∗ , Susan M. Bellman a , Matthew D. Stephenson a , Karolina Lisy. c
a
The Joanna Briggs Institute, The University of Adelaide, Adelaide, South Australia, Australia
b
Centre for Nanoscale BioPhotonics, Macquarie University, Sydney, New South Wales, Australia
c
Peter MacCallum Cancer Centre, Melbourne, Victoria, Australia

a r t i c l e i n f o a b s t r a c t

Article history: This systematic review investigated whether the insulin sensitiser metformin has a geroprotective effect
Received 6 July 2017 in humans. Pubmed and Embase were searched along with databases of unpublished studies. Eligible
Received in revised form 3 August 2017 research investigated the effect of metformin on all-cause mortality or diseases of ageing relative to non-
Accepted 3 August 2017
diabetic populations or diabetics receiving other therapies with adjustment for disease control achieved.
Available online 10 August 2017
Overall, 260 full-texts were reviewed and 53 met the inclusion criteria. Diabetics taking metformin had
significantly lower all-cause mortality than non-diabetics (hazard ratio (HR) = 0.93, 95%CI 0.88–0.99),
Keywords:
as did diabetics taking metformin compared to diabetics receiving non-metformin therapies (HR = 0.72,
Metformin
ageing
95%CI 0.65–0.80), insulin (HR = 0.68, 95%CI 0.63–0.75) or sulphonylurea (HR = 0.80, 95%CI 0.66–0.97). Met-
Insulin sensitizer formin users also had reduced cancer compared to non-diabetics (rate ratio = 0.94, 95%CI 0.92–0.97) and
Lifespan cardiovascular disease (CVD) compared to diabetics receiving non-metformin therapies (HR = 0.76, 95%CI
Longevity 0.66–0.87) or insulin (HR = 0.78, 95%CI 0.73–0.83). Differences in baseline characteristics were observed
Geroprotection which had the potential to bias findings, although statistical adjustments were made. The apparent reduc-
tions in all-cause mortality and diseases of ageing associated with metformin use suggest that metformin
could be extending life and healthspans by acting as a geroprotective agent.
© 2017 Elsevier B.V. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2. Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.1. Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.2. Inclusion criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.2.1. Population . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.2.2. Exposure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.2.3. Comparator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.2.4. Outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.2.5. Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.3. Search strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.4. Appraisal and extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.5. Data synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.1. All-cause mortality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.2. Diseases of ageing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.2.1. Cancer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.2.2. Cardiovascular disease . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

∗ Corresponding author at: The Joanna Briggs Institute, The University of Adelaide, Adelaide, South Australia.
E-mail address: Jared.campbell@mq.edu.au (J.M. Campbell).

http://dx.doi.org/10.1016/j.arr.2017.08.003
1568-1637/© 2017 Elsevier B.V. All rights reserved.

Downloaded for Budi Lakukua (googledotbudi@yahoo.co.id) at Universitas Tarumanagara from ClinicalKey.com by Elsevier on October 02, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
32 J.M. Campbell et al. / Ageing Research Reviews 40 (2017) 31–44

3.2.3. Other diseases of ageing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

4. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
Funding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
Appendix A. Supplementary data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

1. Introduction cial effects of metformin overcoming their generally worse health

status. A weakness of this strategy is that people taking metformin
Ageing is characterised by progressive damage at a cellular and could be observed to have worse or the same outcomes as non-
organ level as prevention and repair mechanisms begin to break diabetics in the presence of an actual beneficial effect of metformin
down (Finkel and Holbrook, 2000; Garinis et al., 2008; Mitchell, due to it being masked by the detrimental effects of diabetes. The
2008). Diseases of ageing – such as cardiovascular disease (CVD), second strategy is to compare diabetics taking metformin to diabet-
kidney failure and cancer – are the result of the accumulation of this ics managing their diabetes through other means while statistically
damage and ultimately responsible for organisms dying of old age. controlling for the effects of the different therapies on diabetes
With the ageing population of the developed world the extension management (i.e. HbA1c or diabetes related doctor visits (Lin et al.,
of healthspan – the period that an individual is functional and free 2015)). This strategy reasons that any difference in outcomes seen
of chronic diseases (Barzilai et al., 2016) – has been identified as an after adjusting for the relative effects on disease management can
important goal both for the minimisation of human suffering and be attributed to metformin’s activity beyond diabetes control, and
to enable healthcare systems to cope. are therefore generalisable. This strategy’s weaknesses are that sta-
One drug which has been subjected to significant research as tistical adjustment is an imperfect process – it is always possible
a geroprotective factor is the insulin sensitiser metformin. Met- that residual confounding remains that could cause any differences
formin has been shown to extend the lifespans of model organisms observed − and that other antidiabetic drugs could cause harms
(including mice and C. elegans (Anisimov et al., 2011; Anisimov which in comparison would appear to be beneficial effects by met-
et al., 2010a; Cabreiro et al., 2013)) and a number of potential mech- formin.
anisms for its effects have been discussed, including decreased However, although additional controlled, experimental
insulin and IGF-1 signalling (Liu et al., 2011), inhibition of mTOR research will need to be carried out, there exists a wealth of
(Kickstein et al., 2010; Nair et al., 2014), reducing the levels of observational data that can offer evidence on whether metformin
reactive oxygen species (ROS)(Batandier et al., 2006; Zheng et al., could be applied as a geroprotective agent in humans. As such,
2012), lowering inflammation (Saisho, 2015), reducing DNA dam- this systematic review and meta-analysis has been undertaken to
age (Algire et al., 2012; Na et al., 2013), and the activation of identify and synthesise all studies where the effects of metformin
AMP-activated protein kinase (AMPK)(Zhou et al., 2001). Its effect on all-cause mortality or diseases of ageing have been compared
on AMPK has received particular attention as it models the intra- to the general or non-diabetic population or to diabetics managing
cellular mechanisms of caloric restriction (Gillespie et al., 2016; Lee their diabetes through other means with adjustment for disease
and Min, 2013) – another intervention which has been found to be control. Numerous systematic reviews have been carried out
capable of extending the lifespans of animal models – but which which nominally address this review’s outcomes of interest, how-
is not feasible for widespread implementation in humans (Roth ever those which investigate the association between metformin
and Ingram, 2016). This, among other factors, has led to metformin and all-cause mortality do so in the context of severe diseases
being described as a caloric restriction mimetic (CRM) (Gillespie in addition to diabetes (usually various forms of cancer) (Coyle
et al., 2016; Lee and Min, 2013; Roth and Ingram, 2016). et al., 2016; Gash et al., 2017; Li et al., 2017; Meng et al., 2017;
One aspect of metformin which makes it particularly promising Tang et al., 2017; Zhou et al., 2017a) and their results therefore
for development as a geroprotective agent is that it is already being cannot be generalised, while those that investigate the incidence of
widely used in humans for a different purpose. As an insulin sen- diseases of ageing do so without attempting to adjust for diabetes
sitiser metformin is a first line therapy for diabetes (Nathan et al., control (Liu et al., 2017; Ma et al., 2017; Pladevall et al., 2016; Tang
2006). Through this use its potential for adverse effects and con- et al., 2017; Zhou et al., 2017b). As such, the question of whether
traindications have been well characterised, and it has been found metformin could act as a geroprotective agent in humans has yet
to have a broad safety profile (Rojas and Gomes, 2013) which would to be addressed in a systematic review.
make it significantly quicker and easier to implement as an inter-
vention for ageing than a previously unutilised drug. 2. Methods
Due to the widespread application of metformin for the manage-
ment of diabetes, a large amount of data has already been collected 2.1. Objective
on its effects on mortality and diseases of ageing. However, extrap-
olation of this research evidence to the general population is This systematic review aimed to identify and synthesise all
invariably confounded by the fact that the people who are cur- research on the effect of metformin on all-cause mortality and
rently receiving metformin have diabetes and are taking it for diseases of ageing which had the potential to give evidence on
the treatment thereof. As such, any benefits seen could be due whether it could be used as a geroprotective factor to extend life
to improvements in diabetes control. Two analytic strategies have and healthspan in the general population. It followed an a priori
been attempted to overcome this limitation and gain some insight protocol pre-registered with PROSPERO (CRD42016036098).
as to whether metformin could be used to improve the healthspan
of the general population. 2.2. Inclusion criteria
The first strategy is the comparison of the outcomes of dia-
betics taking metformin to the general or non-diabetic population 2.2.1. Population
(Bannister et al., 2014). This approach reasons that as diabetics are Studies on all-cause mortality were eligible for inclusion if they
on average less healthy than non-diabetics, superior outcomes in reported on adults with a minimum age of 40 years or a mean age of
the metformin/diabetic group should be a consequence of benefi- 50 years. This age restriction was applied to increase the proportion

Downloaded for Budi Lakukua (googledotbudi@yahoo.co.id) at Universitas Tarumanagara from ClinicalKey.com by Elsevier on October 02, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
 J.M. Campbell et al. / Ageing Research Reviews 40 (2017) 31–44 33

of deaths which would be attributable to age and was not applied 2.4. Appraisal and extraction
to studies on diseases of ageing. Studies were included where met-
formin was being given to patients for the treatment of a particular Duplicate critical appraisal was carried out using the stan-
disease (diabetes in all cases although this was not an inclusion dardised Joanna Briggs Institute (JBI) appraisal checklists for
criterion), however where populations were specifically selected experimental, cohort, case control and cross-sectional studies (The
to include a second serious illness (i.e. patients with cancer who Joanna Briggs Institute, 2014) by independent reviewers work-
were also being treated for diabetes), studies were not eligible for ing separately, with consensus reached through discussion in all
inclusion. cases. Data was extracted for all studies using an extraction tem-
plate which included fields for; study methodology, study design,
inclusion criteria, data source, country, disease control, exposure,
2.2.2. Exposure
comparator, sample size/events, follow up, outcome(s), outcome
Studies were included where a group of patients had been
data, and statistical adjustments. Where additional data or clarifi-
treated with metformin of any dosage taken on an ongoing basis.
cation was needed, attempts were made to contact corresponding
authors by email.
2.2.3. Comparator
Any controls were eligible for inclusion provided they were not 2.5. Data synthesis
receiving metformin. This included members of the general or non-
diabetic population as well as people who were receiving other Where possible, data was pooled through meta-analysis. The
therapies for diabetes. In the latter case, studies were only included inverse variance method with a random effects model from
if data was adjusted for the effect of diabetes on disease control (i.e. RevMan was used, unless otherwise specified. Data included in
HbA1c levels or number/rate of diabetes related healthcare visits). the meta-analyses was always from the analyses with the greatest
Where patients entered into studies were using metformin on an adjustment. Where summary measures differed, outcomes were
ongoing basis their data was included if diabetes control at baseline converted to hazard ratios (HRs) using the equations described in
was adjusted for, as this reflected how well their diabetes was being Tierney et al. (2007). Data were analysed separately based on com-
managed by metformin. However, if patients were enrolled with parators utilised, and heterogeneity was assessed using the Chi2
incident metformin use and the only adjustment for disease control and I2 tests. Two sets of sensitivity analyses were carried out where
was made at baseline, the study was excluded as this data would poor quality studies (<70% on critical appraisal) were removed from
not reflect the effect of metformin. meta-analyses, and where studies that did not adjust for duration
of diabetes and/or comorbidities were removed. Analysis of pub-
2.2.4. Outcomes lication bias through Funnel plots and Egger’s test was planned if
The two primary outcomes were all-cause mortality and the sufficient studies (at least 10) were found for any outcomes, how-
incidence, onset or prevalence of diseases of ageing. Studies on ever none were. Where statistical pooling was not possible data is
disease specific mortality were not eligible for inclusion. Any dis- presented narratively.
eases of ageing were eligible for inclusion with the exception of
diabetes as any effect metformin had on diabetes incidence could 3. Results
more likely be attributed to it directly treating pre-diabetes as an
insulin sensitiser than extending healthspan. Diseases ultimately The initial search identified a total of 21,027 studies, which was
included were cancer (including breast, lung, colorectal, pancreatic, reduced to 19,408 following the removal of duplicates. Follow-
prostate, oesophageal, thyroid, renal, hepatocellular, and head and ing title/abstract screening, 260 studies were included for full-text
neck cancer), CVD (including stroke, myocardial infarction, heart review. Of these, 207 were excluded and 53 were included. Full
failure, coronary heart disease, macrovascular morbidity and ven- search details and reasons for exclusion are given in Fig. 1. Overall,
tricular dysfunction), kidney failure, fracture, open angle glaucoma, 13 included studies reported data on all-cause mortality while 49
cognitive impairment, and carpal tunnel syndrome. reported on diseases of ageing.

3.1. All-cause mortality

2.2.5. Studies
Both experimental and observational studies (including cohort,
Of the studies that investigated the effect of metformin on all-
case-control and analytical cross-sectional studies) were eligible
cause mortality, four compared diabetic patients being treated
for inclusion in this systematic review. Case report and case series
with metformin to the general population or non-diabetic patients
studies were ineligible.
(Bannister et al., 2014; Berard et al., 2011; Bo et al., 2012; Claesen
et al., 2016), while nine compared diabetic patients being treated
2.3. Search strategy with metformin to diabetic patients managing their diabetes
through other means. Subgroups within the latter group included
The search for published literature included Pubmed and insulin therapy (Ekstrom et al., 2012; Ghotbi et al., 2013), diet
Embase, while the search for unpublished studies included the interventions (Bo et al., 2012; Sullivan et al., 2011), sulphonylurea
International Clinical Trials Registry Platform, ProQuest Disserta- therapy (Evans et al., 2006; Kahler et al., 2007; Sullivan et al., 2011;
tions and Theses Global, and OpenThesis. No date restrictions were Wang et al., 2014), and non-metformin therapies (studies assigned
applied, however only English language articles were eligible for to this group included all patients receiving any other oral hypo-
inclusion. The search was carried out in March 2016. Full details of glycemic agent, but may have also included those receiving insulin,
the structure and content of each search undertaken are reported diet, or no therapy as well) (Ekstrom et al., 2012; Ghotbi et al.,
in supplementary material 1. Titles and abstracts were screened 2013; Gosmanova et al., 2008; Libby et al., 2009; Pratipanawatr
followed by the retrieval of potentially relevant full-texts. These et al., 2010). Full details of interventions, controls and other study
were then carefully compared to the inclusion criteria to identify characteristics are included in Table 1. For the metformin versus
eligible studies. The reference lists of included full-texts were then non-diabetic comparison three of the cohorts were matched by
reviewed, however no additional studies that met the inclusion age (Bannister et al., 2014; Bo et al., 2012; Claesen et al., 2016)
criteria were identified. while in the fourth (Berard et al., 2011) metformin users were

Downloaded for Budi Lakukua (googledotbudi@yahoo.co.id) at Universitas Tarumanagara from ClinicalKey.com by Elsevier on October 02, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
 34
Downloaded for Budi Lakukua (googledotbudi@yahoo.co.id) at Universitas Tarumanagara from ClinicalKey.com by Elsevier on October 02, 2018.

Table 1
Characteristics of studies investigating the effect of metformin on all cause mortality.

Study Exposure Comparator Population Diabetes control N/events and Follow up Covariates adjusted for in
adjustment multivariable analysis

Metformin compared to Non-Diabetic or General Population

For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.

Bannister et al. (2014) Metformin initiation Non diabetic matched for United Kingdom. Cohort. N/a Metformin = 78,241/2663 Age, modified Charlson index,
age, gender, same general Incident diabetes and Non- gender, smoking status, prior
practice, prior cancer exposed to glucose diabetic = 78,241/2669 antiplatelet therapy, prior
status and smoking status lowering therapy for a deaths lipid-lowering therapy, prior
minimum of 180 days Mean: 2.8 years antihypertensive therapy, year of
(excluded if any record of study index date and study arm.
secondary diabetes).
Berard et al. (2011) Metformin use at baseline Non diabetic at baseline France. Cohort. Middle N/a Metformin = 40/9 Duration of diabetes, history of
aged men and women with Non-diabetic = 3162/213 diabetes complications, area of
diabetes. Exposure 14 years residence, age, gender,
(hypoglycaemic drug use, educational level, alcohol

J.M. Campbell et al. / Ageing Research Reviews 40 (2017) 31–44

presence of metformin) consumption, smoking, blood
was assessed at entry only. pressure, LDL and HDL
cholesterol.
Bo et al. (2012) Metformin use General population based Italy. Cohort. Patients with N/a Metformin = NR/104 None.
on age and sex type 2 diabetes. General
population = NR/128
Mean = 4.5 years
Claesen et al. (2016) Metformin use at baseline Non users of glucose Belgium. Cohort. Patients N/a Metformin = 42,900/3389 Adjusted for age, associated
(censored if discontinued lowering agents older than 18 prescribed Non therapies, and gender.
for more than 9 months) metformin and non users diabetic = 214,500/16,517
of glucose lowering agents 5 years
matched 5 to one on age,
gender, cardiovascular
history, associated
therapies and year of start
of follow up.

Metformin compared to other diabetes therapies, controlling for diabetes management

Bo et al. (2012) Metformin use Management of diabetes Italy. Cohort. Patients with HbA1c Metformin = 1479/136 Adjusted for propensity score
through diet alone type 2 diabetes. Diet = 620/68 which included: propensity score
Mean: 4.5 years which included: age, sex,
diabetes duration, HbA1c,
smoking, BMI, retinopathy,
nephropathy, coronary or
peripheral artery disease, other
comorbidities, use of
hypertensives, use of
acetylsalicylic acid
Ekstrom et al. (2012) Metformin use Insulin use continuously Sweden. Cohort. Men and HbA1c at baseline (at least Metformin = 14,697/1734 Age, sex, diabetes duration,
continuously for 12 for 12 months before women aged 40–85 with 1 year after treatment Insulin = 12,291/2389 HbA1c, non-HDL-C, BMI,
months before baseline baseline type 2 diabetes on commencement) Other OHA = 5171/020 smoking, eGFR, multidose
Or continuous glucose Insulin and other dispensation, previous
Other OHA lowering treatment that OHA = 1365/NR hospitalisation, history of CVD
Or filed at least 3 prescriptions Mean: 3.9 years and CHF, microalbuminuria, and
Insulin and other OHA for their treatment (or 18 treatment with antihypertensive
fills of multi-dose agents, lipid lowering agents,
dispensed drugs) in the 12 and cardiac glycosides
months before baseline.
Downloaded for Budi Lakukua (googledotbudi@yahoo.co.id) at Universitas Tarumanagara from ClinicalKey.com by Elsevier on October 02, 2018.

Evans et al. (2006) Metformin monotherapy Sulfonylurea monotherapy United Kingdom. Cohort. Average HbA1c for the Metformin = 2286/107 Gender, Age, Diabetes duration,
throughout study period throughout study period Patients with type 2 study period Sulphonylurea = 3331/597 previous CV hospitalisation,
(initiated at baseline, (initiated at baseline, diabetes newly prescribed NR Smoking, mean HbA1c, mean
censored from date of censored from date of OHAs. BMI, mean systolic blood
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.

switch to new therapy) switch to new therapy) pressure, mean diastolic blood
pressure, mean cholesterol, using
aspirin, using statins, using beta
blockers, using ACE inhibitors.
Ghotbi et al. (2013) Metformin monotherapy Insulin monotherapy Worldwide (16 countries). Change in HbA1c values Metformin = 1631/96 Age, smoking habits, diabetes
Or OR Cohort. Patients who were And/or HbA1c levels Metformin with duration, congestive heart
Metformin with insulin Any other therapy (Diet, 55 years or over with BMI insulin = 1000/88 failure, history of hypertension,
insulin, sulphonylureas) 27–45 with diabetes. Insulin = 1116/138 BMI, sex, history of CVD, tobacco
alone or in any Any other therapy = NR use, HDL concentrations, LDL
combination concentrations, change in HbA1c
values and/or HbA1c levels, heart
rate, actual systolic and diastolic
blood pressure, sibutramine

J.M. Campbell et al. / Ageing Research Reviews 40 (2017) 31–44

usage
Gosmanova et al. (2008) Metformin (combination or Not receiving metformin USA. Cohort. Veterans with HbA1c at baseline Metformin = 1207/266, Age, race, eGFR, HbA1c, use of
montherapy) type 2 diabetes receiving (treatment was ongoing) Non-metformin = 999/253 insulin, ACEI/ARBs, or statins.
long term care at veterans Metormin = 62 months,
affair medical center Non-metformin = 61
treated with a prescription months
claim for oral hypoglyemic
agents
Kahler et al. (2007) Metformin montherapy Sulfonylurea monotherapy USA. Cohort. People with HbA1c at baseline Metformin = 2988/82 Age, diabetes suration, HbA1C,
Metformin with diabetes at least 18 years of (treatment was ongoing) Metofrmin with sulphony- creatinine, diabetes related
sulphonylurea age who survived at least lurea = 13,820/468 physician visits and also a
1 year after drug Sulphonylurea = 19,053/1005 propensity score constructed
assessment. NR from 48 variables.
Libby et al. (2009) In receipt of a metformin No record of metformin United Kingdom. Cohort. HbA1C during the study Metformin = 1085/609 Age, sex, smoking, deprivation,
prescription use, matched by year of Patients diagnosed with period Non- BMI, A1C, insulin use,
diagnosis type 2 diabetes aged 35 or metformin = 4085/1422 sulphoylurea use.
over. NR
Pratipanawatr et al. (2010) Metformin use Non use of metformin Thailand. Cohort. Diabetic HbA1C at baseline Overall = 9370/424 Age, sex, HbA1c, serum
patients receiving clinic (treatment was ongoing) 3 years creatinine, healthcare plan,
treatment education staus, smoking status,
previous history of coronary
artery disease and
cerebrovascular disease, lipid
lowering medication, insulin.
Sullivan et al. (2011) Metformin montherapy Diet alone International. Cohort. HbA1c at baseline Metformin = 1746/NR Age, sex, duration of diabetes,
Or Patients with type 2 (treatment was ongoing) Diet = 1632 smoking, waist-hipratio, systolic
Sulphonylurea diabetes Sulphonylurea = 1632 blood pressure, total cholesterol,
monotherapy 5 years HDL cholesterol, HbA1c, ACR
group, history of CVD, presence
ofmicrovascular disease,
creatinine and peripheral
damage to feet.
Wang et al. (2014) Metformin prescription as Sulfonylurea prescription USA. Cohort. Veterans aged HbA1c during the study Metformin = 307/NR Age, race, diabetes duration, age
sole class of glucose as sole class of glucose 65–89 years with type 2 period Sulphonylurea = 2108 adjusted Charlson co-morbidity,
lowering medication for lowering medication for diabetes without history of Mean = 5.6 years smoking cessation status, mean
≥180 days ≥180 days liver, renal diseases, or LDL across study period, mean
cancers HbA1c across study period and
propensity score.

35
36 J.M. Campbell et al. / Ageing Research Reviews 40 (2017) 31–44

could not be carried out as no studies were of poor quality and only
one study adjusted for diabetes duration and/or comorbidity.
The finding that metformin reduces mortality was supported by
analyses that compared diabetic patients receiving metformin to
diabetic patients receiving other therapies (Fig. 3), where diabetes
control had been adjusted for.
The majority of subgroups, including studies with non-
metformin controls (HR = 0.72, 95%CI 0.65–0.80, p < 0.00001),
insulin treated controls (HR = 0.68, 95%CI 0.63–0.75, p < 0.00001),
and sulphonylurea treated controls (HR = 0.80, 95%CI 0.66–0.97,
p = 0.02), found that even after adjusting for metformin’s effect on
diabetes control, diabetics taking it had lower all-cause mortality
than diabetics who were not. The one exception was the subgroup
where the controls managed their diabetes through diet alone. In
this case one study (Bo et al., 2012) found a significant benefit to all-
cause mortality for metformin and the other study (Sullivan et al.,
2011) found a non-significant disadvantage resulting in an over-
all non-significant finding (HR = 0.91, 95%CI 0.68–1.22, p = 0.53).
An overall meta-analysis could not be performed as the same
studies were included across multiple subgroups where they pre-
sented their metformin data relative to different controls. However,
when the duplicate data sets were excluded (removing those with
the strongest positive findings: Ekstrom Non-Metformin, Ghotbi
Non-Metformin and Sullivan Sulphonylurea), diabetics taking met-
formin had significantly lower mortality compared to controls
after adjusting for disease control (HR = 0.75, 95%CI 0.69–0.82,
Fig. 1. PRISMA flow chart. p < 0.00001). Sensitivity analysis for the non-metformin control
subgroup where two studies which did not adjust for duration
of diabetes and/or comorbidity were excluded (Gosmanova et al.,
2008; Libby et al., 2009) did not affect results. Sensitivity analysis
older (mean ± SD age 56.2 ± 6.6) than non-diabetics (50.1 ± 8.3); for study quality could not be performed as no studies were poor
a disparity which could be expected to bias the results towards quality.
a finding of metformin being associated with increased mortality,
although age was adjusted for. In the majority of cases when met- 3.2. Diseases of ageing
formin users were compared to other diabetics, metformin users
were younger (Ekstrom et al., 2012; Evans et al., 2006; Ghotbi et al., The main diseases of ageing investigated in the included studies
2013; Gosmanova et al., 2008; Kahler et al., 2007; Sullivan et al., were cancer (Supp Table S1) and cardiovascular disease (CVD, Supp
2011; Wang et al., 2014). Differences between mean ages ranged Table S2). However, other age related conditions were investigated
from 0.4 years (Gosmanova et al., 2008) to 5.9 years (Ekstrom et al., in a smaller number of studies including renal failure, fracture, open
2012). In one study the metformin group was older than the control angle glaucoma, cognitive impairment and carpal tunnel syndrome
by <1 year (Bo et al., 2012), and in two mean ages were not reported (Supp Table S3).
(Libby et al., 2009; Pratipanawatr et al., 2010), although Libby et al.
did match by year of diagnosis so age is likely to have been close. 3.2.1. Cancer
The trend of metformin users being younger than other diabetics Overall, 25 studies investigated the association between met-
has the potential to bias results towards a finding of decreased mor- formin use and cancer (Supp Table S1). Where people taking
tality, however all studies adjusted for age, with the exception of metformin were compared to the general or non-diabetic pop-
Bo et al. The appraisal of study quality found that overall the stud- ulation the majority of studies did not report the ages of these
ies were conducted rigorously, with notable deficiencies being that groups. The two that did found that people taking metformin were
none reported similar characteristics in their respective popula- significantly older (Andersson et al., 2012; But et al., 2014), poten-
tions at baseline and less than half had adequate follow up times tially biasing the results to increased risk of cancer in metformin
for our observations of interest (defined as ≥5 years) with thorough users although age was adjusted for in both analyses. Amongst the
reporting on reasons for loss to follow up (Supplementary material remainder of studies half were case-controls which matched for age
2). and are therefore unlikely to have meaningful hidden bias (Becker
Meta-analysis of the four studies that compared people manag- et al., 2014, 2016; Bodmer et al., 2012a; Bodmer et al., 2012b; Lu
ing their diabetes with metformin to non-diabetics or the general et al., 2015; Walker et al., 2015) and the remainder adjusted for
population (Bannister et al., 2014; Berard et al., 2011; Bo et al., 2012; age (Baur et al., 2011; Becker et al., 2013; Nordström et al., 2015;
Claesen et al., 2016) found that people taking metformin had sig- Tseng, 2011, 2012a, 2012b). For people taking metformin compared
nificantly lower mortality compared to those who were not (Fig. 2 to diabetics controlling their diabetes through other means, only
HR = 0.93, 95%CI 0.88–0.99, p = 0.03). One included study was an one study reported relative ages for the comparison (metformin
outlier in that it reported a non-significant increase in mortality for users were younger biasing results towards a lower risk of can-
people taking metformin (Berard et al., 2011), however it was by far cer although age was adjusted for (Goossens et al., 2015)). Nine
the smallest study, with just nine events in the metformin group. studies were case-controls which matched for age and therefore
Hazard ratios (HRs) used in the meta-analysis represent adjusted have limited risk of bias in this regard (Azoulay et al., 2011; Becker
values with two exceptions where the HR could only be calculated et al., 2013; Bodmer et al., 2010; Bosco et al., 2011; Chen et al.,
for crude data (Bannister et al., 2014) and where only crude analyses 2013; Mazzone et al., 2012; Sehdev et al., 2015; Smiechowski et al.,
were performed for this data (Bo et al., 2012). Sensitivity analysis 2013a, 2013b), while two cohort studies adjusted for age but did not

Downloaded for Budi Lakukua (googledotbudi@yahoo.co.id) at Universitas Tarumanagara from ClinicalKey.com by Elsevier on October 02, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
 J.M. Campbell et al. / Ageing Research Reviews 40 (2017) 31–44 37

Fig. 2. All-cause mortality hazard ratios in people using metformin for diabetes compared to the general or non-diabetic population.

Fig. 3. All-cause mortality hazard ratios in people using metformin for diabetes compared people using other therapies for diabetes. 2.1.2 ‘All Non-Metformin Control’ includes
studies where the controls were any diabetic people receiving who were not being treated with metformin (or controls which were receiving a broad range of therapies).
2.1.3 ‘Diet Control’ includes studies where controls were managing their diabetes though diet. 2.1.4 ‘Insulin Control’ includes studies were controls were managing their
diabetes using insulin. 2.1.5 ‘Sulphonylurea’ includes studies where controls were managing their diabetes using sulphonylurea.

report it (Libby et al., 2009; Redaniel et al., 2012). Critical appraisal which received 99% of the weight. The third study (Baur et al., 2011)
showed that cohort studies compared groups with significant dif- was cross-sectional and found a non-significant decrease in the
ferences at baseline, tended to lack adequate follow up and did prevalence of cancer in diabetics receiving metformin monother-
not report on reasons for loss to follow up with similar deficien- apy compared to the non-diabetic population, and a non-significant
cies observed in case-control studies (supplementary material 2). increase in diabetics receiving metformin as a combination therapy.
Of the included studies, three investigated the incidence of any can- A fourth study investigated the effect of metformin on cancer inci-
cer for people taking metformin for diabetes control compared to dence but compared it to matched diabetics receiving any other oral
non-diabetics. Two of these studies could be combined in meta- hypoglycaemic agent rather than non-diabetics (Libby et al., 2009).
analysis (Fig. 4A, (Andersson et al., 2012; But et al., 2014)) and It reported a significant reduction for metformin (HR = 0.63, 95%CI
showed that diabetics taking metformin had significantly lower 0.53–0.75, p < 0.05), supporting the findings of the meta-analysis.
cancer incidence (Rate ratio = 0.94, 95%CI 0.92–0.97, p = 0.0003). In Three studies investigated the effect of taking metformin on the
this case rate ratio was used in the meta-analysis rather than HR as development of pancreatic cancer in people with diabetes com-
both studies used it as the summary measure to report their data. pared to the general (Bodmer et al., 2012b; Lu et al., 2015) or
It should be noted that although there was essentially no hetero- non-diabetic population (Walker et al., 2015). However, findings
geneity between the two studies, the results of the meta-analysis were mixed and the result of the meta-analysis was non-significant
drew almost entirely on the findings of (Andersson et al., 2012) (Fig. 4B, all studies reported their outcomes as odds ratios (ORs),

Downloaded for Budi Lakukua (googledotbudi@yahoo.co.id) at Universitas Tarumanagara from ClinicalKey.com by Elsevier on October 02, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
38 J.M. Campbell et al. / Ageing Research Reviews 40 (2017) 31–44

Fig. 4. A) Cancer incidence in people with diabetes taking metformin compared to the general or non-diabetic population. B) Pancreatic cancer incidence in people with
diabetes taking metformin compared to the general or non-diabetic population.

therefore this was the summary measure used in the meta- betics being treated with metformin to diabetics being treated with
analysis). Sensitivity analysis could not be carried out as no studies sulphonylurea (Redaniel et al., 2012) and found a non-significant
were of poor quality and only one study adjusted for duration of increase in breast cancer for metformin. Interestingly, they also
diabetes and/or comorbidity. compared people taking metformin in combination with sulphony-
The effect of metformin on the incidence of colorectal can- lurea to people taking sulphonylurea alone and found a significant
cer was investigated in four studies, three of which were similar reduction in the incidence of breast cancer in the people treated
enough to be combined in meta-analysis (Libby et al., 2009; Sehdev with metformin (HR = 0.66, 95%CI 0.50–0.88, p < 0.05).
et al., 2015; Smiechowski et al., 2013a). These studies compared the The effect of metformin on the incidence of lung cancer was
incidence of colorectal cancer in diabetic patients taking metformin investigated in four studies. Three studies compared the incidence
to non-metformin diabetic controls. Although a trend for met- of lung cancer in diabetic patients treated with metformin to dia-
formin use being associated with a reduced incidence of colorectal betic patients not treated with metformin (Libby et al., 2009;
cancer was found, the results were ultimately non-significant Mazzone et al., 2012; Smiechowski et al., 2013b) and could be
(Fig. 5A, HR = 0.85, 95%CI 0.72–1.02, p = 0.08). As heterogeneity was combined in meta-analysis Fig. 5C. The combined results showed
non-significant (Chi2 = 0.15, I2 = 48%) and all studies suggested a that taking metformin was associated with a decreased incidence
benefit, a meta-analysis using the fixed effects model was trialled. of lung cancer in diabetic patients (HR = 0.80, 95%CI 0.65–0.98,
A significant reduction in colorectal cancer incidence for the met- p = 0.03). Sensitivity analysis could not be carried out as no stud-
formin group was found (HR = 0.91, 95%CI 0.84–0.98, p = 0.009). ies were of poor quality and only one study adjusted for duration
Sensitivity analysis where one study which did not adjust for dura- of diabetes and/or comorbidity. The fourth study (Bodmer et al.,
tion of diabetes and/or comorbidity was excluded (Libby et al., 2012a) compared diabetics taking metformin to the general popu-
2009) did not have a large effect on the estimate but resulted in lation. It did not find any significant effects for short term (1–14
a significant finding by the random effects model (HR = 0.92, 95%CI prescriptions) or long term (≥40 prescription) metformin use,
0.85–0.99, p = 0.02). Sensitivity analysis for study quality could not however medium term (15–39 prescriptions) metformin use was
be performed as no studies were poor quality. An additional study contradictorily associated with a significant increase in the odds of
not included in the meta-analysis (Tseng, 2012a) compared the developing lung cancer (OR = 1.24, 95%CI 1.03–1.50, p < 0.05).
incidence of colorectal cancer in diabetics taking metformin to Additional cancers that were investigated included prostate,
the general population and found that metformin was associated bladder, thyroid, renal, head and neck, oesophageal, and hepatocel-
with a significant reduction (relative risk = 0.73, 95%CI 0.58–0.92, lular cancer. Neither of the two studies on prostate cancer found any
p = 0.008), supporting the conclusion that metformin may reduce significant association with metformin use (Azoulay et al., 2011;
colorectal cancer incidence. Nordström et al., 2015). Nor did the studies on bladder cancer
Breast cancer incidence was investigated in four studies, three (Goossens et al., 2015; Tseng, 2011), thyroid cancer (Tseng, 2012b),
of which compared the incidence in diabetic patients taking renal cell carcinoma (Becker et al., 2016), head and neck cancer
metformin to diabetic patients who were not taking metformin (Becker et al., 2014), or oesophageal cancer (Becker et al., 2013).
(Bodmer et al., 2010; Bosco et al., 2011; Libby et al., 2009). These However, metformin users had significantly lower odds of devel-
were combined in meta-analysis Fig. 5B which showed a signifi- oping hepatocellular cancer compared to other diabetics (OR = 0.79,
cant reduction in the incidence of breast cancer (HR = 0.71, 95%CI 95%CI 0.75–0.85, p < 0.0001).
0.54–0.92, p = 0.01). Sensitivity analysis where one study which
did not adjust for duration of diabetes and/or comorbidity was 3.2.2. Cardiovascular disease
excluded (Libby et al., 2009) did not have a large effect on the Fifteen studies were found that investigated the effect of met-
estimate but resulted in a non-significant finding being made formin on the development of CVDs (Supp Table S2). Critical
(HR = 0.72, 95%CI 0.50–1.04, p = 0.08). Sensitivity analysis for study appraisal showed a similar pattern in quality as observed for studies
quality could not be performed as no studies were poor quality. on all-cause mortality and cancer – groups had significant differ-
The fourth study compared the incidence of breast cancer in dia- ences at baseline, follow-up was frequently shorter than desirable

Downloaded for Budi Lakukua (googledotbudi@yahoo.co.id) at Universitas Tarumanagara from ClinicalKey.com by Elsevier on October 02, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
 J.M. Campbell et al. / Ageing Research Reviews 40 (2017) 31–44 39

Fig. 5. A) Colorectal cancer incidence for people with diabetes taking metformin compared to diabetics not taking metformin. B) Breast cancer incidence for people with
diabetes taking metformin compared to diabetics not taking metformin C) Lung cancer incidence for people with diabetes taking metformin compared to diabetics not taking
metformin.

to observe the onset of disease, and loss to follow up was inade- any CVD between diabetics being treated with metformin to dia-
quately described (Supplementary material 2). Additionally, over betics treated with insulin (Ekstrom et al., 2012; Ghotbi et al., 2013),
half of the cohort studies included participants who already had where a significant decrease in incidence with metformin use was
some form of CVD at baseline and half of the case-control studies did found (Fig. 6, HR = 0.78, 95%CI 0.73–0.83, p < 0.00001). However, the
not adequately control for potential confounding factors. No stud- meta-analysis for sulphonylurea controls, which also included two
ies compared people taking metformin for diabetes management studies (Evans et al., 2006; Sullivan et al., 2011), did not show a
to non-diabetics or the general population. For those that com- significant effect (Fig. 6). One study was found that investigated
pared people taking metformin to people controlling their diabetes the effect of metformin on incidence of CVD compared to diabetics
through other means, six included analyses where the group receiv- managing their diabetes through diet and showed a non-significant
ing metformin was younger than those who were not (Ekstrom increase in the metformin group (Sullivan et al., 2011). An overall
et al., 2012; Evans et al., 2006; Ghotbi et al., 2013; McAlister et al., meta-analysis was prevented as the subgroups contained data from
2008; Sullivan et al., 2011) (the difference ranged from 1.1 years) the same participants. However, when duplications were removed
(Sullivan et al., 2011) to 5.9 years (Ekstrom et al., 2012), while in two (again excluding the data that showed the strongest positive effect:
studies people receiving metformin were younger than compara- (Ekstrom Non-metformin, Ghotbi Non Metformin, Sullivan Diet),
tors (2.8 years (Nichols et al., 2005) and 5 years (Kooy et al., 2009)). the result was a significant effect for metformin reducing the inci-
Three further studies were case-controls which matched for age dence of CVD (HR = 0.83, 95%CI 0.73–0.94, p = 0.004).
(Floyd et al., 2016; Hartung et al., 2005; Koro et al., 2005) while Three studies investigated the effect of metformin on the
four did not report on relative ages but did adjust for age in their incidence of stroke. Two compared the incidence of stroke in met-
analyses (Gejl et al., 2015; Giorda et al., 2011; Jansson et al., 2014; formin users to other diabetics not using metformin (Floyd et al.,
Peters et al., 2013). Five studies comparing the incidence of any 2016; Jansson et al., 2014). Meta-analysis of their findings (Fig. 7A)
CVD in diabetics taking metformin to diabetics not receiving met- showed a significant reduction in stroke for metformin users
formin could be combined in meta-analysis (Ekstrom et al., 2012; (HR = 0.70, 95%CI 0.53–0.93, p = 0.01). The third study (Sullivan
Gejl et al., 2015; Ghotbi et al., 2013; Jansson et al., 2014; Peters et al., 2011) compared metformin use to diabetics managing their
et al., 2013). This showed a significant reduction in CVD for people diabetes through diet alone or with sulphonylurea, and found a
taking metformin (Fig. 6, HR = 0.76, 95%CI 0.66–0.87, p < 0.0001). non-significant increase in the incidence of stroke with metformin
Sensitivity analysis where studies of low quality were excluded use in both cases.
from meta-analysis did not have a large impact on the estimate Myocardial infarction incidence was examined in just two stud-
(HR = 0.73, 95%CI 0.53–1.00, p = 0.05). No studies did not adjusted ies (Floyd et al., 2016; Jansson et al., 2014) and although (Jansson
for duration of diabetes and/or comorbidity. The finding of a reduc- et al., 2014) showed a strong significant effect in favour of met-
tion was supported by two studies that compared the incidence of formin, the result of the meta-analysis was a non-significant

Downloaded for Budi Lakukua (googledotbudi@yahoo.co.id) at Universitas Tarumanagara from ClinicalKey.com by Elsevier on October 02, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
40 J.M. Campbell et al. / Ageing Research Reviews 40 (2017) 31–44

Fig. 6. Incidence of any CVD in diabetic patients treated with metformin compared to diabetic controls receiving other therapies. 2.5.1 ‘All Non-Metformin Control’ includes
studies where the controls were any diabetic patients who were not being treated with metformin (or controls who were receiving a broad range of therapies). 2.5.2‘Diet
Control’ includes studies where controls were managing their diabetes though diet. 2.5.3 ‘Insulin Control’ includes studies where controls were managing their diabetes
using insulin 2.5.4 ‘Sulphonylurea control’ includes studies where controls were managing their diabetes using sulphonylurea.

Fig. 7. A) Stroke incidence for diabetics taking metformin compared to diabetics not taking metformin. B) Myocardial infarction incidence for diabetics taking metformin
compared to diabetics not taking metformin.

decrease in myocardial infarction for metformin users compared to and diabetics treated with sulphonylurea. Another study (Nichols
diabetics not using metformin (Fig. 7B, HR = 0.63, 95%CI 0.28–1.42, et al., 2005) showed a non-significant decrease in congestive heart
p = 0.27). failure for diabetics treated with metformin compared to those
Heart failure was investigated in four studies, however due treated with sulphonylurea. It also found a significant decrease
to differences in study design and outcomes reported no meta- in heart failure with metformin compared to insulin use (rate
analysis could be performed. One study (Hartung et al., 2005) ratio = 0.39, 95%CI 0.19–0.80, p < 0.05). The final study (McAlister
compared metformin users to other oral hypoglycaemic agent users et al., 2008) compared the incidence of heart failure in metformin
with regards to hospitalisations resulting from heart failure, but treated patients to those treated with sulphonylurea and found a
found no significant effect. Another study (Koro et al., 2005) found non-significant decrease in heart failure for those receiving met-
non-significant increases for metformin on incidence of conges- formin.
tive heart failure compared to diabetics with no drug exposure

Downloaded for Budi Lakukua (googledotbudi@yahoo.co.id) at Universitas Tarumanagara from ClinicalKey.com by Elsevier on October 02, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
 J.M. Campbell et al. / Ageing Research Reviews 40 (2017) 31–44 41

Results for coronary heart disease also could not be meta- ≥500 DDD (OR = 0.81, 95%CI 0.70–0.93, p < 0.05) consumed from the
analysed. One study (Peters et al., 2013) found a non-significant study start date to the date of censoring.
reduction for users of metformin compared to other diabet- Additional studies on diseases of ageing investigated open angle
ics who were not receiving metformin, while another (Sullivan glaucoma (Lin et al., 2015), which had a significantly lower inci-
et al., 2011) found significant increases with metformin use com- dence in diabetics taking metformin compared to other diabetics
pared to patients being treated with diet alone or sulphonylurea (HR = 0.75, 95%CI 0.59–0.95, p < 0.05), cognitive impairment (Ng
monotherapy. Other CVD outcomes investigated were left ventricu- et al., 2014), which was significantly less likely to develop in
lar dysfunction, which was significantly more prevalent in diabetics patients taking metformin compared to other diabetics (OR = 0.49,
being treated with metformin compared to those not taking met- 95%CI 0.25–0.95, p < 0.05; a finding which remained significant in
formin (OR = 1.62, 95%CI 1.09–2.40, p = 0.16) – although the authors users who took metformin for more than 6 years before baseline
of this cross-sectional study concluded that this outcome was likely (OR = 0.27, 95%CI 0.12–0.60, p < 0.05), but not those who took it for 6
due to confounding due to numerous large between group dif- years or less), and carpal tunnel syndrome (Geoghegan et al., 2004),
ferences (Giorda et al., 2011) – and macrovascular morbidity and which was not significantly associated with metformin use relative
mortality (Kooy et al., 2009) which was significantly lowered by the to the general population.
addition of metformin to insulin therapy compared to the addi-
tion of a placebo to insulin therapy (HR = 0.34, 95%CI 0.21–0.56,
p = 0.001). 4. Discussion

3.2.3. Other diseases of ageing This systematic review has shown through meta-analysis that
Seven studies investigated the effect of metformin on diseases diabetics taking metformin have a lower rate of all-cause mor-
of ageing, other than cancer and CVD, although none could be com- tality than non-diabetic people and the general population. Our
bined in meta-analysis (Supp Table S3). Critical appraisal of studies results suggest that metformin could be an effective intervention
(Supplementary material 2) showed differences between groups at to extend the lifespans of people who do not have diabetes. This is
baseline as well as short follow up with no reasons given for losses. supported by additional meta-analyses showing that diabetics tak-
Two studies compared outcomes for people using metformin to ing metformin had lower rates of all-cause mortality than diabetics
the general or non-diabetic population; neither reported on dif- receiving other therapies after adjusting for relative disease con-
ferences in ages, although one was a case-control study which trol. This was true for metformin compared to diabetics receiving
matched for age (Geoghegan et al., 2004) and the other adjusted any non-metformin therapy, insulin, or sulphonylurea, however,
for age (Vestergaard et al., 2005). Amongst the five studies which the result was non-significant for diet therapy. This finding could
compared people receiving metformin to those controlling their be due to diet therapy being a more likely option for people with
diabetes through other means there was heterogeneity in report- early or less severe diabetes, or it could alternatively be explained
ing; two studies included a cohort of metformin patients who were by improved diet and lifestyle having an effect beyond disease con-
younger than their comparators (Hung et al., 2013; Masica et al., trol comparable to metformin. With only two studies contributing
2013), one reported that one subgroup of metformin users was to the meta-analysis further research is needed for conclusions to
younger and another older but did not report the overall com- be drawn.
parison (Ng et al., 2014), one reported that metformin users were Other results suggest that metformin’s effect on all-cause mor-
slightly older than patients assigned to a different medication (in an tality could be due to it having a geroprotective effect of delaying or
RCT design) (Kahn et al., 2008), and one did not report relative ages preventing diseases of ageing, particularly cancer and CVD which
(Lin et al., 2015). All studies adjusted for age in their analyses with are two of the leading causes of death and disability worldwide
the exception of the RCT. Two studies investigated kidney failure (Mathers and Loncar, 2006). We show that diabetic people taking
or decline. One study (Hung et al., 2013) found a non-significant metformin had a lower rate of developing any cancer compared
decrease for the incidence of a glomerular filtration event or end with the general population, and had a lower risk of developing
stage renal disease for diabetic patients treated with metformin colorectal, breast or lung cancer compared with diabetics manag-
compared to diabetic patients treated with sulphonylurea, however ing their diabetes through non-metformin therapies after adjusting
this decrease was significant in a subgroup of patients where urine for disease control. The rate of any form of CVD was similarly lower
protein at baseline was known and adjusted for (HR = 0.78, 95%CI for diabetic people taking metformin compared to those manag-
0.64–0.97). The same analyses were performed for patients treated ing their diabetes through any non-metformin therapy or insulin
with metformin in addition to sulphonylurea compared to sulpho- therapy after adjusting for disease control, however findings for
nylurea alone, however no significant differences were found. The sulphonylurea were non-significant. The incidence of stroke was
second study found a non-significant reduction in the decline of also reduced with metformin use compared to any non-metformin
estimated glomerular filtration rate to <60 ml/min/1.73 m2 within therapy; however there was no significant effect for metformin on
one year for metformin users (Masica et al., 2013) compared to myocardial infarction.
sulphonylurea users, however there was a significant reduction in These findings rest on the credibility of two assumptions: that
risk of developing proteinuria (HR = 0.71, 95%CI 0.53–0.95, p < 0.05). improvements in survival and reduced onset of diseases of ageing
These comparisons were repeated for treatment with thiazolidine- seen in diabetics above the outcomes of people who do not have
diones compared to metformin, and no significant differences were diabetes can be generalised to the non-diabetic population, and
found. that benefits seen between populations receiving different thera-
The effect of metformin on fracture risk was investigated in pies for diabetes (which are still present after adjusting for disease
two studies. One study (Kahn et al., 2008) showed that metformin control) are independent on the therapies’ effects on diabetes and
significantly decreased risk of fracture relative to treatment with are likewise generalisable. Of the two, the former is the more robust
rosiglitazone (HR = 0.64, 95%CI 0.46–0.88, p = 0.007). Stratified anal- assumption, although the potential for confounding, as with any
ysis showed that the effect persisted in women but not in men. observational research, does exist. As such, the findings for all-
The other study (Vestergaard et al., 2005) found the metformin cause mortality and cancer − which are backed by comparisons
users had lower risks of fracture compared to the general popula- to the non-diabetic population − should be considered more reli-
tion for <150 defined daily doses (DDD) (OR = 0.87, 95%CI 0.86–0.96, able than those for CVD − which are based solely on comparisons
p < 0.05), 150–499 DDD (OR = 0.81, 95%CI 0.71–0.94, p < 0.05), and to diabetics treated with other therapies. However, taken together

Downloaded for Budi Lakukua (googledotbudi@yahoo.co.id) at Universitas Tarumanagara from ClinicalKey.com by Elsevier on October 02, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
42 J.M. Campbell et al. / Ageing Research Reviews 40 (2017) 31–44

these results provide strong evidence, from a combined population (Ekstrom et al., 2012; Vasisht et al., 2010), metformin is not risk
of 417,316 for all-cause mortality compared to the non-diabetic or free. Although studies in mice have generally been positive, some
general population alone, to support the hypothesis that metformin have shown increased mortality when taken at higher concentra-
has a geroprotective effect in humans, slowing the progression of tions (Martin-Montalvo et al., 2013), and others have suggested that
diseases of ageing and in doing so extending both health and lifes- its effects may be gender exclusive (favouring females (Anisimov
pans. et al., 2010b)), an observation which is supported by some subgroup
Beyond the assumptions discussed, limitations of this research analyses performed in included studies (Bodmer et al., 2012b; Kahn
include that none of the meta-analyses included sufficient stud- et al., 2008). As such, there is a risk that people taking metformin as
ies for any investigation of publication bias, study designs were a geroprotective agent will experience harm or no effect even if it is
entirely observational (with two exceptions that showed benefits ultimately found to be beneficial in some circumstances. Addition-
for metformin on fracture (Kahn et al., 2008) and macrovascu- ally, recent work by Wang et al. − who showed in an included study
lar morbidity and mortality (Kooy et al., 2009)), and none of that the presence of frailty significantly attenuated the effect of
the data for diseases of ageing beyond cancer and CVD could be metformin on mortality (Wang et al., 2014) −found that older men’s
combined through meta-analysis despite promising findings for propensity for diseases of ageing significantly interacted with met-
kidney failure, fracture risk, open angle glaucoma and cognitive formin’s effect on the development of these diseases (Wang et al.,
impairment. With regards to cognitive impairment, in the con- 2017).
duct of this review’s search, numerous additional studies were The findings reported in this systematic review remain
found which investigated the effect of metformin on measures preliminary generalisations, primarily making use of existing
of dementia. However, these studies were not eligible for inclu- observational evidence collected for other purposes to investi-
sion due to investigating average level of cognitive ability (rather gate the credibility of the hypothesis that the insulin sensitiser
than the incidence or prevalence of impairment) or not adjusting metformin may extend the health and lifespans of people from
for disease control. An additional systematic review with broader the non-diabetic population. Differences in baseline characteristics
inclusion criteria relating to dementia outcomes is now planned were found which had the potential to bias results both towards
(Campbell et al., 2017). Diabetes, despite being a major chronic dis- positive findings (where metformin users were compared to other
ease which becomes progressively more common with age, was not diabetics) as well as away from (where metformin users were com-
investigated by this review as metformin being an intervention for pared to non-diabetics). While they should not be overstated, the
diabetes was too much of a complicating factor. However, experi- apparent association with reductions in all-cause mortality and
mental studies have directly investigated the effect of prophylactic diseases of ageing found through meta-analysis do support this
metformin for the prevention of diabetes (in high BMI populations) hypothesis, and metformin should be investigated as an interven-
and shown that it significantly reduced its development (Andreadis tion for ageing in future clinical trials.
et al., 2009; Knowler et al., 2009).
The primary issue with study quality in this systematic review, Funding
highlighted by critical appraisal, was that studies rarely compared
groups that did not have meaningful differences at the beginning This research did not receive any specific grant from funding
of the observation period. Where diabetics receiving metformin agencies in the public, commercial, or not-for-profit sectors.
were compared to general or non-diabetic populations, this aspect
is essentially a feature of the study design. However, for metformin Appendix A. Supplementary data
compared to other diabetic therapies it is more concerning, as it
reflects a trend for people receiving metformin to be younger with Supplementary data associated with this article can be found, in
shorter durations of disease, which has the potential to bias results the online version, at http://dx.doi.org/10.1016/j.arr.2017.08.003.
towards finding a protective effect for metformin. Critical appraisal
also showed that studies were generally thorough in adjusting for References
potential confounding factors, however, and a sizeable proportion
of studies utilised matching. As such, the potential impact of this Algire, C., Moiseeva, O., Deschenes-Simard, X., Amrein, L., Petruccelli, L., Birman, E.,
Viollet, B., Ferbeyre, G., Pollak, M.N., 2012. Metformin reduces endogenous
source of bias is at least mitigated. Additionally, the general pat-
reactive oxygen species and associated DNA damage. Cancer Prev. Res. (Phila.)
tern of findings for metformin compared to other diabetic therapies 5, 536–543.
being reproduced by findings for metformin compared to the gen- Andersson, C., Vaag, A., Selmer, C., Schmiegelow, M., Sørensen, R., Lindhardsen, J.,
eral or non-diabetic population suggests that they are real and not Gislason, G.H., Køber, L., Torp-Pedersen, C., 2012. Risk of cancer in patients
using glucose-lowering agents: a nationwide cohort study of 3.6 million
the result of confounding. Studies frequently had follow up peri- people. BMJ Open 2.
ods less than five years, which affects the likelihood of age related Andreadis, E.A., Katsanou, P.M., Georgiopoulos, D.X., Tsourous, G.I., Yfanti, G.K.,
outcomes having an opportunity to occur, however the use of sum- Gouveri, E.T., Diamantopoulos, E.J., 2009. The effect of metformin on the
incidence of type 2 diabetes mellitus and cardiovascular disease risk factors in
mary measures which incorporate time (i.e. hazard and rate ratios) overweight and obese subjects–the Carmos study. Exp. Clin. Endocrinol.
will have accommodated this. Finally, studies did not report on rea- Diabetes 117, 175–180.
sons for (or the existence of) loss to follow up which, if known, may Anisimov, V.N., Egormin, P.A., Piskunova, T.S., Popovich, I.G., Tyndyk, M.L., Yurova,
M.N., Zabezhinski, M.A., Anikin, I.V., Karkach, A.S., Romanyukha, A.A., 2010a.
have revealed or precluded potential sources of bias. Metformin extends life span of HER-2/neu transgenic mice and in combination
Future research in this area should include controlled trials. Two with melatonin inhibits growth of transplantable tumors in vivo. ABBV Cell
existing trials on the effects of metformin on ageing − Metformin in Cycle 9, 188–197.
Anisimov, V.N., Piskunova, T.S., Popovich, I.G., Zabezhinski, M.A., Tyndyk, M.L.,
Longevity Study (MILES NCT02432287) and Targeting Ageing with
Egormin, P.A., Yurova, M.V., Rosenfeld, S.V., Semenchenko, A.V., Kovalenko, I.G.,
Metformin (TAME) (Barzilai et al., 2016) − are currently planned. Poroshina, T.E., Berstein, L.M., 2010b. Gender differences in metformin effect
The findings of this systematic review suggest that studies on the on aging, life span and spontaneous tumorigenesis in 129/Sv mice. Aging
(Albany NY) 2, 945–958.
effect of metformin in older people on the incidence of chronic dis-
Anisimov, V.N., Berstein, L.M., Popovich, I.G., Zabezhinski, M.A., Egormin, P.A.,
eases (healthspan) and longevity (lifespan) over the long term are Piskunova, T.S., Semenchenko, A.V., Tyndyk, M.L., Yurova, M.N., Kovalenko, I.G.,
warranted. It must be stressed, however, that any immediate use of Poroshina, T.E., 2011. If started early in life, metformin treatment increases life
metformin as an intervention for ageing by the general population span and postpones tumors in female SHR mice. Aging (Albany NY) 3, 148–157.
Azoulay, L., Dell’Aniello, S., Gagnon, B., Pollak, M., Suissa, S., 2011. Metformin and
is not supported. Despite being a relatively safe drug, with feared the incidence of prostate cancer in patients with type 2 diabetes. Cancer
effects on lactic acidosis and kidney function emerging to be minor Epidemiol. Biomarkers Prev. 20, 337–344.

Downloaded for Budi Lakukua (googledotbudi@yahoo.co.id) at Universitas Tarumanagara from ClinicalKey.com by Elsevier on October 02, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
 J.M. Campbell et al. / Ageing Research Reviews 40 (2017) 31–44 43

Bannister, C.A., Holden, S.E., Jenkins-Jones, S., Morgan, C.L., Halcox, J.P., Ghotbi, A.A., Kober, L., Finer, N., James, W.P., Sharma, A.M., Caterson, I., Coutinho,
Schernthaner, G., Mukherjee, J., Currie, C.J., 2014. Can people with type 2 W., Van Gaal, L.F., Torp-Pedersen, C., Andersson, C., 2013. Association of
diabetes live longer than those without? A comparison of mortality in people hypoglycemic treatment regimens with cardiovascular outcomes in
initiated with metformin or sulphonylurea monotherapy and matched, overweight and obese subjects with type 2 diabetes: a substudy of the SCOUT
non-diabetic controls. Diabetes Obes. Metab. 16, 1165–1173. trial. Diabetes Care 36, 3746–3753.
Barzilai, N., Crandall, J.P., Kritchevsky, S.B., Espeland, M.A., 2016. Metformin as a Gillespie, Z.E., Pickering, J., Eskiw, C.H., 2016. Better living through chemistry:
tool to target aging. Cell Metab. 23, 1060–1065. caloric restriction (CR) and CR mimetics alter genome function to promote
Batandier, C., Guigas, B., Detaille, D., El-Mir, M.Y., Fontaine, E., Rigoulet, M., Leverve, increased health and lifespan. Front. Genet. 7, 142.
X.M., 2006. The ROS production induced by a reverse-electron flux at Giorda, C.B., Cioffi, G., De Simone, G., Di Lenarda, A., Faggiano, P., Latini, R., Lucci, D.,
respiratory-chain complex 1 is hampered by metformin. J. Bioenerg. Maggioni, A.P., Tarantini, L., Velussi, M., Verdecchia, P., Comaschi, M., 2011.
Biomembr. 38, 33–42. Predictors of early-stage left ventricular dysfunction in type 2 diabetes: results
Baur, D.M., Klotsche, J., Hamnvik, O.P., Sievers, C., Pieper, L., Wittchen, H.U., Stalla, of DYDA study. Eur. J. Cardiovasc. Prev. Rehabil. 18, 415–423.
G.K., Schmid, R.M., Kales, S.N., Mantzoros, C.S., 2011. Type 2 diabetes mellitus Goossens, M.E., Buntinx, F., Zeegers, M.P., Driessen, J.H., De Bruin, M.L., De Vries, F.,
and medications for type 2 diabetes mellitus are associated with risk for and 2015. Influence of metformin intake on the risk of bladder cancer in type 2
mortality from cancer in a German primary care cohort. Metabolism 60, diabetes patients. Br. J. Clin. Pharmacol. 80, 1464–1472.
1363–1371. Gosmanova, E.O., Canada, R.B., Mangold, T.A., Rawls, W.N., Wall, B.M., 2008. Effect
Becker, C., Meier, C.R., Jick, S.S., Bodmer, M., 2013. Case-control analysis on of metformin-containing antidiabetic regimens on all-cause mortality in
metformin and cancer of the esophagus. Cancer Causes Control 24, 1763–1770. veterans with type 2 diabetes mellitus. Am. J. Med. Sci. 336, 241–247.
Becker, C., Jick, S.S., Meier, C.R., Bodmer, M., 2014. Metformin and the risk of head Hartung, D.M., Touchette, D.R., Bultemeier, N.C., Haxby, D.G., 2005. Risk of
and neck cancer: a case-control analysis. Diabetes Obes. Metab. 16, 1148–1154. hospitalization for heart failure associated with thiazolidinedione therapy: a
Becker, C., Jick, S.S., Meier, C.R., Bodmer, M., 2016. Metformin and the risk of renal medicaid claims-based case-control study. Pharmacotherapy 25, 1329–1336.
cell carcinoma: a case-control analysis. Eur. J. Cancer Prev. 26 (3), 257–262. Hung, A.M., Roumie, C.L., Greevy, R.A., Liu, X., Grijalva, C.G., Murff, H.J., Griffin, M.R.,
Berard, E., Bongard, V., Arveiler, D., Dallongeville, J., Wagner, A., Montaye, M., 2013. Kidney function decline in metformin versus sulfonylurea initiators:
Ruidavets, J.B., Ferrieres, J., 2011. 14-year risk of all-cause mortality according assessment of time-dependent contribution of weight, blood pressure, and
to hypoglycemic drug exposure in a general population. Eur. Heart J. 32, glycemic control. Pharmacoepidemiol. Drug Saf. 22, 623–631.
973–974. Jansson, S.P., Svardsudd, K., Andersson, D.K., 2014. Effects of fasting blood glucose
Bo, S., Ciccone, G., Rosato, R., Villois, P., Appendino, G., Ghigo, E., Grassi, G., 2012. levels and blood pressure and treatment of diabetes and hypertension on the
Cancer mortality reduction and metformin: a retrospective cohort study in incidence of cardiovascular disease: a study of 740 patients with incident Type
type 2 diabetic patients. Diabetes Obes. Metab. 14, 23–29. 2 diabetes with up to 30 years’ follow-up. Diabet. Med. 31, 1055–1063.
Bodmer, M., Meier, C., Krahenbuhl, S., Jick, S.S., Meier, C.R., 2010. Long-term Kahler, K.H., Rajan, M., Rhoads, G.G., Safford, M.M., Demissie, K., Lu, S.E., Pogach,
metformin use is associated with decreased risk of breast cancer. Diabetes Care L.M., 2007. Impact of oral antihyperglycemic therapy on all-cause mortality
33, 1304–1308. among patients with diabetes in the Veterans Health Administration. Diabetes
Bodmer, M., Becker, C., Jick, S.S., Meier, C.R., 2012a. Metformin does not alter the Care 30, 1689–1693.
risk of lung cancer: a case-control analysis. Lung Cancer 78, 133–137. Kahn, S.E., Zinman, B., Lachin, J.M., Haffner, S.M., Herman, W.H., Holman, R.R.,
Bodmer, M., Becker, C., Meier, C., Jick, S.S., Meier, C.R., 2012b. Use of antidiabetic Kravitz, B.G., Yu, D., Heise, M.A., Aftring, R.P., Viberti, G., 2008.
agents and the risk of pancreatic cancer: a case-control analysis. Am. J. Rosiglitazone-associated fractures in type 2 diabetes: an analysis from a
Gastroenterol. 107, 620–626. diabetes outcome progression trial (ADOPT). Diabetes Care 31, 845–851.
Bosco, J.L.F., Antonsen, S., Sorensen, H.T., Pedersen, L., Lash, T.L., 2011. Metformin Kickstein, E., Krauss, S., Thornhill, P., Rutschow, D., Zeller, R., Sharkey, J.,
and incident breast cancer among diabetic women: a population-based Williamson, R., Fuchs, M., Kohler, A., Glossmann, H., Schneider, R., Sutherland,
case-control study in Denmark. Cancer Epidemiol. Biomark. Prev. 20, 101–111. C., Schweiger, S., 2010. Biguanide metformin acts on tau phosphorylation via
But, A., Wang, H., Mannisto, S., Pukkala, E., Haukka, J., 2014. Assessing the effect of mTOR/protein phosphatase 2A (PP2A) signaling. Proc. Natl. Acad. Sci. U. S. A.
treatment duration on the association between anti-diabetic medication and 107, 21830–21835.
cancer risk. PLoS One 9, e113162. Knowler, W.C., Fowler, S.E., Hamman, R.F., Christophi, C.A., Hoffman, H.J.,
Cabreiro, F., Au, C., Leung, K.Y., Vergara-Irigaray, N., Cocheme, H.M., Noori, T., Brenneman, A.T., Brown-Friday, J.O., Goldberg, R., Venditti, E., Nathan, D.M.,
Weinkove, D., Schuster, E., Greene, N.D., Gems, D., 2013. Metformin retards 2009. 10-year follow-up of diabetes incidence and weight loss in the Diabetes
aging in C elegans by altering microbial folate and methionine metabolism. Prevention Program Outcomes Study. Lancet 374, 1677–1686.
Cell 153, 228–239. Kooy, A., De Jager, J., Lehert, P., Bets, D., Wulffelé, M.G., Donker, A.J.M., Stehouwer,
Campbell, J.M., Stephenson, M.D., de Courten, B., Chapmean, I., Bellman, S.M., C.D.A., 2009. Long-term effects of metformin on metabolism and
Aromataris, E., 2017. Metformin and Alzheimer’s disease, dementia and microvascular and macrovascular disease in patients with type 2 diabetes
cognitive impairment: a systematic review protocol. JBI Database Syst. Rev. mellitus. Arch. Intern. Med. 169, 616–625.
Implement Rep. 117 (2), 210–219. Koro, C.E., Bowlin, S.J., Weiss, S.R., 2005. Antidiabetic therapy and the risk of heart
Chen, H.P., Shieh, J.J., Chang, C.C., Chen, T.T., Lin, J.T., Wu, M.S., Lin, J.H., Wu, C.Y., failure in type 2 diabetic patients: an independent effect or confounding by
2013. Metformin decreases hepatocellular carcinoma risk in a dose-dependent indication. Pharmacoepidemiol. Drug Saf. 14, 697–703.
manner: population-based and in vitro studies. Gut 62, 606–615. Lee, S.H., Min, K.J., 2013. Caloric restriction and its mimetics. BMB Rep. 46, 181–187.
Claesen, M., Gillard, P., De Smet, F., Callens, M., De Moor, B., Mathieu, C., 2016. Li, Y., Hu, L., Xia, Q., Yuan, Y., Mi, Y., 2017. The impact of metformin use on survival
Mortality in individuals treated with glucose-lowering agents: a large, in kidney cancer patients with diabetes: a meta-analysis. Int. Urol. Nephrol. 49,
controlled cohort study. J. Clin. Endocrinol. Metab. 101, 461–469. 975–981.
Coyle, C., Cafferty, F.H., Vale, C., Langley, R.E., 2016. Metformin as an adjuvant Libby, G., Donnelly, L.A., Donnan, P.T., Alessi, D.R., Morris, A.D., Evans, J.M., 2009.
treatment for cancer: a systematic review and meta-analysis. Ann. Oncol. 27, New users of metformin are at low risk of incident cancer: a cohort study
2184–2195. among people with type 2 diabetes. Diabetes Care 32, 1620–1625.
Ekstrom, N., Schioler, L., Svensson, A.M., Eeg-Olofsson, K., Miao Jonasson, J., Lin, H.C., Stein, J.D., Nan, B., Childers, D., Newman-Casey, P.A., Thompson, D.A.,
Zethelius, B., Cederholm, J., Eliasson, B., Gudbjornsdottir, S., 2012. Effectiveness Richards, J.E., 2015. Association of geroprotective effects of metformin and risk
and safety of metformin in 51 675 patients with type 2 diabetes and different of open-angle glaucoma in persons with diabetes mellitus. JAMA Ophthalmol.
levels of renal function: a cohort study from the Swedish National Diabetes 133, 915–923.
Register. BMJ Open, 2. Liu, B., Fan, Z., Edgerton, S.M., Yang, X., Lind, S.E., Thor, A.D., 2011. Potent
Evans, J.M., Ogston, S.A., Emslie-Smith, A., Morris, A.D., 2006. Risk of mortality and anti-proliferative effects of metformin on trastuzumab-resistant breast cancer
adverse cardiovascular outcomes in type 2 diabetes: a comparison of patients cells via inhibition of erbB2/IGF-1 receptor interactions. ABBV Cell Cycle 10,
treated with sulfonylureas and metformin. Diabetologia 49, 930–936. 2959–2966.
Finkel, T., Holbrook, N.J., 2000. Oxidants, oxidative stress and the biology of ageing. Liu, F., Yan, L., Wang, Z., Lu, Y., Chu, Y., Li, X., Liu, Y., Rui, D., Nie, S., Xiang, H., 2017.
Nature 408, 239–247. Metformin therapy and risk of colorectal adenomas and colorectal cancer in
Floyd, J.S., Wiggins, K.L., Christiansen, M., Dublin, S., Longstreth, W.T., Smith, N.L., type 2 diabetes mellitus patients: a systematic review and meta-analysis.
McKnight, B., Heckbert, S.R., Weiss, N.S., Psaty, B.M., 2016. Case-control study Oncotarget 8, 16017–16026.
of oral glucose-lowering drugs in combination with long-acting insulin and the Lu, Y., Garcia Rodriguez, L.A., Malgerud, L., Gonzalez-Perez, A., Martin-Perez, M.,
risks of incident myocardial infarction and incident stroke. Lagergren, J., Bexelius, T.S., 2015. New-onset type 2 diabetes, elevated HbA1c,
Pharmacoepidemiol. Drug Saf. 25, 151–160. anti-diabetic medications, and risk of pancreatic cancer. Br. J. Cancer 113,
Garinis, G.A., van der Horst, G.T., Vijg, J., Hoeijmakers, J.H., 2008. DNA damage and 1607–1614.
ageing: new-age ideas for an age-old problem. Nat. Cell Biol. 10, 1241–1247. Ma, S., Zheng, Y., Xiao, Y., Zhou, P., Tan, H., 2017. Meta-analysis of studies using
Gash, K.J., Chambers, A.C., Cotton, D.E., Williams, A.C., Thomas, M.G., 2017. metformin as a reducer for liver cancer risk in diabetic patients. Medicine
Potentiating the effects of radiotherapy in rectal cancer: the role of aspirin, (Baltimore) 96, e6888.
statins and metformin as adjuncts to therapy. Br. J. Cancer 117 (2), 210–219. Martin-Montalvo, A., Mercken, E.M., Mitchell, S.J., Palacios, H.H., Mote, P.L.,
Gejl, M., Starup-Linde, J., Scheel-Thomsen, J., Gregersen, S., Vestergaard, P., 2015. Scheibye-Knudsen, M., Gomes, A.P., Ward, T.M., Minor, R.K., Blouin, M.J.,
Risk of cardiovascular disease: the effects of diabetes and anti-diabetic drugs – Schwab, M., Pollak, M., Zhang, Y., Yu, Y., Becker, K.G., Bohr, V.A., Ingram, D.K.,
a nested case-control study. Int. J. Cardiol. 178, 292–296. Sinclair, D.A., Wolf, N.S., Spindler, S.R., Bernier, M., de Cabo, R., 2013.
Geoghegan, J.M., Clark, D.I., Bainbridge, L.C., Smith, C., Hubbard, R., 2004. Risk Metformin improves healthspan and lifespan in mice. Nat. Commun. 4, 2192.
factors in carpal tunnel syndrome. J. Hand Surg. Br. 29, 315–320. Masica, A.L., Ewen, E., Daoud, Y.A., Cheng, D., Franceschini, N., Kudyakov, R.E.,
Bowen, J.R., Brouwer, E.S., Wallace, D., Fleming, N.S., West, S.L., 2013.

Downloaded for Budi Lakukua (googledotbudi@yahoo.co.id) at Universitas Tarumanagara from ClinicalKey.com by Elsevier on October 02, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
44 J.M. Campbell et al. / Ageing Research Reviews 40 (2017) 31–44

Comparative effectiveness research using electronic health records: impacts of Sehdev, A., Shih, Y.C.T., Vekhter, B., Bissonnette, M.B., Olopade, O.I., Polite, B.N.,
oral antidiabetic drugs on the development of chronic kidney disease. 2015. Metformin for primary colorectal cancer prevention in patients with
Pharmacoepidemiol. Drug Saf. 22, 413–422. diabetes: a case-control study in a US population. Cancer 121, 1071–1078.
Mathers, C.D., Loncar, D., 2006. Projections of global mortality and burden of Smiechowski, B., Azoulay, L., Yin, H., Pollak, M.N., Suissa, S., 2013a. The use of
disease from 2002 to 2030. PLoS Med. 3, e442. metformin and colorectal cancer incidence in patients with type ii diabetes
Mazzone, P.J., Rai, H., Beukemann, M., Xu, M., Jain, A., Sasidhar, M., 2012. The effect mellitus. Cancer Epidemiol. Biomark. Prev. 22, 1877–1883.
of metformin and thiazolidinedione use on lung cancer in diabetics. BMC Smiechowski, B.B., Azoulay, L., Yin, H., Pollak, M.N., Suissa, S., 2013b. The use of
Cancer 12. metformin and the incidence of lung cancer in patients with type 2 diabetes.
McAlister, F.A., Eurich, D.T., Majumdar, S.R., Johnson, J.A., 2008. The risk of heart Diabetes Care 36, 124–129.
failure in patients with type 2 diabetes treated with oral agent monotherapy. Sullivan, D., Forder, P., Simes, J., Whiting, M., Kritharides, L., Merrifield, A.,
Eur. J. Heart Fail. 10, 703–708. Donoghoe, M., Colman, P.G., Graham, N., Haapamaki, H., Keech, A., 2011.
Meng, F., Song, L., Wang, W., 2017. Metformin improves overall survival of Associations between the use of metformin, sulphonylureas, or diet alone and
colorectal cancer patients with diabetes: a meta-analysis. J Diabetes Res. 2017, cardiovascular outcomes in 6005 people with type 2 diabetes in the FIELD
5063239. study. Diabetes Res. Clin. Pract. 94, 284–290.
Mitchell, G.F., 2008. Effects of central arterial aging on the structure and function of Tang, Y.L., Zhu, L.Y., Li, Y., Yu, J., Wang, J., Zeng, X.X., Hu, K.X., Liu, J.Y., Xu, J.X., 2017.
the peripheral vasculature: implications for end-organ damage. J. Appl. Metformin use is associated with reduced incidence and improved survival of
Physiol. (1985) 105, 1652–1660. endometrial cancer: a meta-analysis. BioMed Res. Int. 2017.
Na, H.J., Park, J.S., Pyo, J.H., Lee, S.H., Jeon, H.J., Kim, Y.S., Yoo, M.A., 2013. The Joanna Briggs Institute, 2014. The Joanna Briggs Institute Reviewer’s Manual:
Mechanism of metformin: inhibition of DNA damage and proliferative activity 2014 edition, Australia.
in Drosophila midgut stem cell. Mech. Ageing Dev. 134, 381–390. Tierney, J.F., Stewart, L.A., Ghersi, D., Burdett, S., Sydes, M.R., 2007. Practical
Nair, V., Sreevalsan, S., Basha, R., Abdelrahim, M., Abudayyeh, A., Rodrigues methods for incorporating summary time-to-event data into meta-analysis.
Hoffman, A., Safe, S., 2014. Mechanism of metformin-dependent inhibition of Trials 8, 16.
mammalian target of rapamycin (mTOR) and Ras activity in pancreatic cancer: Tseng, C.H., 2011. Diabetes and risk of bladder cancer: a study using the National
role of specificity protein (Sp) transcription factors. J. Biol. Chem. 289, Health Insurance database in Taiwan. Diabetologia 54, 2009–2015.
27692–27701. Tseng, C.H., 2012a. Diabetes, metformin use, and colon cancer: a population-based
Nathan, D.M., Buse, J.B., Davidson, M.B., Heine, R.J., Holman, R.R., Sherwin, R., cohort study in Taiwan. Eur. J. Endocrinol. 167, 409–416.
Zinman, B., 2006. Management of hyperglycemia in type 2 diabetes: a Tseng, C.H., 2012b. Thyroid cancer risk is not increased in diabetic patients. PLoS
consensus algorithm for the initiation and adjustment of therapy: a consensus One 7.
statement from the American Diabetes Association and the European Vasisht, K.P., Chen, S.C., Peng, Y., Bakris, G.L., 2010. Limitations of metformin use in
Association for the Study of Diabetes. Diabetes Care 29, 1963–1972. patients with kidney disease: are they warranted? Diabetes Obes. Metab. 12,
Ng, T.P., Feng, L., Yap, K.B., Lee, T.S., Tan, C.H., Winblad, B., 2014. Long-term 1079–1083.
metformin usage and cognitive function among older adults with diabetes. J. Vestergaard, P., Rejnmark, L., Mosekilde, L., 2005. Relative fracture risk in patients
Alzheimers Dis. 41, 61–68. with diabetes mellitus, and the impact of insulin and oral antidiabetic
Nichols, G.A., Koro, C.E., Gullion, C.M., Ephross, S.A., Brown, J.B., 2005. The medication on relative fracture risk. Diabetologia 48, 1292–1299.
incidence of congestive heart failure associated with antidiabetic therapies. Walker, E.J., Ko, A.H., Holly, E.A., Bracci, P.M., 2015. Metformin use among type 2
Diabetes Metab. Res. Rev. 21, 51–57. diabetics and risk of pancreatic cancer in a clinic-based case-control study. Int.
Nordström, T., Clements, M., Karlsson, R., Adolfsson, J., Grönberg, H., 2015. The risk J. Cancer 136, E646–E653.
of prostate cancer for men on aspirin, statin or antidiabetic medications. Eur. J. Wang, C.P., Lorenzo, C., Espinoza, S.E., 2014. Frailty attenuates the impact of
Cancer 51, 725–733. metformin on reducing mortality in older adults with type 2 diabetes. J.
Peters, K.E., Chubb, S.A., Davis, W.A., Davis, T.M., 2013. The relationship between Endocrinol. Diabetes Obes. 2.
hypomagnesemia, metformin therapy and cardiovascular disease complicating Wang, C.P., Lorenzo, C., Habib, S.L., Jo, B., Espinoza, S.E., 2017. Differential effects of
type 2 diabetes: the Fremantle Diabetes Study. PLoS One 8, e74355. metformin on age related comorbidities in older men with type 2 diabetes. J.
Pladevall, M., Riera-Guardia, N., Margulis, A.V., Varas-Lorenzo, C., Calingaert, B., Diabetes Complications 31, 679–686.
Perez-Gutthann, S., 2016. Cardiovascular risk associated with the use of Zheng, Z., Chen, H., Li, J., Li, T., Zheng, B., Zheng, Y., Jin, H., He, Y., Gu, Q., Xu, X., 2012.
glitazones, metformin and sufonylureas: meta-analysis of published Sirtuin 1-mediated cellular metabolic memory of high glucose via the
observational studies. BMC Cardiovasc. Disord. 16, 14. LKB1/AMPK/ROS pathway and therapeutic effects of metformin. Diabetes 61,
Pratipanawatr, T., Rawdaree, P., Chetthakul, T., Bunnag, P., Ngarmukos, C., 217–228.
Benjasuratwong, Y., Leelawatana, R., Kosachunhanun, N., Plengvidhya, N., Zhou, G., Myers, R., Li, Y., Chen, Y., Shen, X., Fenyk-Melody, J., Wu, M., Ventre, J.,
Deerochanawong, C., Suwanwalaikorn, S., Krittiyawong, S., Mongkolsomlit, S., Doebber, T., Fujii, N., Musi, N., Hirshman, M.F., Goodyear, L.J., Moller, D.E., 2001.
Komoltri, C., 2010. Thailand Diabetic Registry cohort: predicting death in Thai Role of AMP-activated protein kinase in mechanism of metformin action. J.
diabetic patients and causes of death. J. Med. Assoc. Thai. 93 (Suppl. (3)), Clin. Invest. 108, 1167–1174.
S12–S20. Zhou, P.T., Li, B., Liu, F.R., Zhang, M.C., Wang, Q., Li, Y.Y., Xu, C., Liu, Y.H., Yao, Y., Li,
Redaniel, M.T., Jeffreys, M., May, M.T., Ben-Shlomo, Y., Martin, R.M., 2012. D., 2017a. Metformin is associated with survival benefit in pancreatic cancer
Associations of type 2 diabetes and diabetes treatment with breast cancer risk patients with diabetes: a systematic review and meta-analysis. Oncotarget 8,
and mortality: a population-based cohort study among British women. Cancer 25242–25250.
Causes Control 23, 1785–1795. Zhou, X.L., Xue, W.H., Ding, X.F., Li, L.F., Dou, M.M., Zhang, W.J., Lv, Z., Fan, Z.R.,
Rojas, L.B., Gomes, M.B., 2013. Metformin: an old but still the best treatment for Zhao, J., Wang, L.X., 2017b. Association between metformin and the risk of
type 2 diabetes. Diabetol. Metab. Syndr. 5, 6. gastric cancer in patients with type 2 diabetes mellitus: a meta-analysis of
Roth, G.S., Ingram, D.K., 2016. Manipulation of health span and function by dietary cohort studies. Oncotarget.
caloric restriction mimetics. Ann. N. Y. Acad. Sci. 1363, 5–10.
Saisho, Y., 2015. Metformin and inflammation: its potential beyond
glucose-lowering effect. Endocr. Metab. Immune Disord. Drug Targets 15,
196–205.

Downloaded for Budi Lakukua (googledotbudi@yahoo.co.id) at Universitas Tarumanagara from ClinicalKey.com by Elsevier on October 02, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.