diabetes research and clinical practice 1 1 8 ( 2 0 1 6 ) 1 7 2 1 8 2

Contents available at ScienceDirect

Diabetes Research
and Clinical Practice
journa l home page: www.e lse vier.com/locate/diabres

Probiotics for the management of type 2 diabetes

mellitus: A systematic review and meta-analysis

Syamimi Samah a, Kalavathy Ramasamy a,b, Siong Meng Lim a,b, Chin Fen Neoh a,b,*
a
Faculty of Pharmacy, Universiti Teknologi MARA (UiTM), 42300 Bandar Puncak Alam, Selangor Darul Ehsan, Malaysia
b
Collaborative Drug Discovery Research (CDDR) Group, Pharmaceutical and Life Sciences Community of Research, Universiti
Teknologi MARA (UiTM), 40450 Shah Alam, Selangor Darul Ehsan, Malaysia

A R T I C L E I N F O A B S T R A C T

Article history: Aims: To systematically review evidence of probiotic interventions against type 2 diabetes
Received 18 January 2016 mellitus (T2DM) and analyse the effects of probiotics on glycaemic control among T2DM
Received in revised form patients.
23 May 2016 Methods: Electronic search using five electronic databases was performed until October
Accepted 6 June 2016 2015. Relevant studies were identified, extracted and assessed for risk of bias. The primary
Available online 18 June 2016 outcomes of this review were glycated haemoglobin (HbA1c) and fasting blood glucose
(FBG). Fasting plasma insulin, homeostasis model assessment-insulin resistance,
C-reactive protein, interleukin-6 and malondialdehyde, were identified as the secondary
Keywords:
outcomes. Mean differences (MD) between probiotics and control groups for all outcomes
Probiotics
were pooled using either Fixed- or Random-Effect Model. Statistical heterogeneity was
Glycaemic
assessed using I2 and Chi2 tests.
Type 2 diabetes mellitus
Results: Six randomised controlled trials (RCTs) were included in the systematic review,
Review
whereas only five were included in meta-analysis. Most RCTs were presented with low or
unclear risk of bias. When compared to placebo, FBG was significantly lower with probiotic
consumption (MD =
0.98 mmol/L; 95% CI:
1.17, 0.78, p < 0.00001), with moderate but
insignificant heterogeneity noted. Insignificant changes between the groups were also
noted for HbA1c and other secondary outcomes.
Conclusions: A moderate hypoglycaemic effect of probiotics, with a significantly lower FBG
was noted. Findings on HbA1c, anti-inflammatory and anti-oxidative effects of probiotics in
the clinical setting, however, remain inconsistent. The findings imply the need for
well-designed clinical studies to further assess the potential beneficial effects of probiotics
in management of T2DM.
2016 Elsevier Ireland Ltd. All rights reserved.

1. Introduction insensitivity of insulin or lack of insulin secretion [1]. It is

the leading cause of cardiovascular disorders, blindness,
Type 2 diabetes mellitus (T2DM), a metabolic disease end-stage renal failure, amputations and hospitalisation [2].
characterised by hyperglycaemia, is associated with either The incidence of T2DM is increasing. In 2011, 366 million

* Corresponding author at: Faculty of Pharmacy, Universiti Teknologi MARA (UiTM), 42300 Bandar Puncak Alam, Selangor Darul Ehsan,
Malaysia.
E-mail address: neohchinfen@puncakalam.uitm.edu.my (C.F. Neoh).
http://dx.doi.org/10.1016/j.diabres.2016.06.014
0168-8227/ 2016 Elsevier Ireland Ltd. All rights reserved.
 diabetes research and clinical practice
adults worldwide were estimated to suffer from diabetes mel-
litus. The number is projected to increase to 592 million by
year 2035 [3]. In addition, the management of T2DM is costly.
In 2010, the global estimation of diabetes expenditures was
USD376 billion; the expenditure is predicted to increase to
USD490 billion in the next 20 years [4]. Given its substantial
health and socio-economic burden, identifying an optimal
therapy for T2DM is important.
The underlying mechanisms of the abnormal rise of blood
glucose level are complex and multi-factorial [5]. T2DM asso-
ciated risk factors that have already been identified include
age, genetic predisposition, sedentary lifestyle, diet patterns
and stress [6]. Recently, emerging data suggested that gut
microbiota may have an important role in progression and
development of T2DM. It was reported that when the micro-
biome balance is shifted in favour of the unhealthy ones,
the level of metabolic endotoxin will increase and potentially
trigger a chronic, low-grade inflammation [7]. The release of
inflammatory cytokines can cause oxidative stress [8], and
ultimately, lead to destruction of b-cells in the pancreas. It
was found that probiotic consumption could increase the
amount of beneficial bacteria (i.e. Bifidobacteria) in the gut,
which could in turn reduce intestinal permeability towards
lipopolysaccharide (LPS) produced by unhealthy gut microor-
ganisms, and altogether attenuate systemic inflammation
responses [9]. As such, modulating gut microbiota through
dietary interventions (e.g. probiotic intake) may be useful in
the prevention and control of inflammatory metabolic disor-
ders including T2DM.
Probiotics are live microorganisms that when adminis-
tered in sufficient amounts can confer health benefit to their
host [10]. Lactobacilli and bifidobacteria are the two most
common types of probiotics [5]. The global probiotic market
was estimated at USD33.19 billion in 2015 and it is projected
to reach USD46.55 billion by 2020; dominated by AsiaPacific
market [11]. The applications of probiotics as alternative bio-
therapeutics have been successfully demonstrated in treat-
ment of respiratory infection [12], inflammatory bowel
disease [13], antibiotic associated diarrhoea [14] and ulcera-
tive colitis [15]. In spite of the booming in vitro and in vivo pro-
biotic studies against metabolic diseases such as T2DM, their
application at clinical settings remains scarce [16,17]. In fact,
the findings from the very few clinical trials that have been
conducted were inconsistent. The most recent systematic
review and meta-analysis on probiotic use in glycaemic con-
trol comprised of 17 trials, involving a broad range of study
populations [i.e. healthy participants, T2DM patients, patients
with hypercholesterolaemia, non-alcoholic steatohepatitis or
metabolic syndrome, obese populations and patients with
gestational diabetes mellitus (GDM)]. The findings, however,
were limited by substantial inter-study clinical heterogeneity
[18]. In addition, the effects of probiotics against vital param-
eters like glycated haemoglobin (HbA1c), anti-inflammatory
and anti-oxidative markers were not assessed [18]. This sys-
tematic review and meta-analysis focused mainly on clinical
studies that have investigated the efficacy of probiotics in
T2DM patients, highlighting areas that were not usually
addressed by previous review (Supplementary Table 1): the
effect of probiotic on HbA1c and fasting blood glucose (FBG),
anti-inflammation and anti-oxidation.
1 1 8 ( 2 0 1 6 ) 1 7 2 1 8 2 173
2. Material and methods
2.1. Literature search strategy
A search of electronic databases, including EMBASE, PubMed/
Medline, SpringerLink, the Cochrane Library and trial registry
website (ClinicalTrials.gov) was performed. The final search
was carried out in October 2015, using combinations of search
terms which included diabetes mellitus, probiotics,
lactobacilli, bifidobacter, streptococcus, microbiota,
microflora, microbiome, gut hormone, insulin sensitivity,
anti-oxidant and anti-inflammatory. In addition, citations
in each paper were also used to further identify relevant
papers that were not found in electronic databases.
2.2. Study selection
The inclusion criteria for this systematic review and meta-
analysis were as follow: (a) randomised controlled trials
(RCT), (b) adult T2DM patients (i.e. age of 18 and above), and
(c) the use of probiotics (of any form, including capsule,
yogurt and kefir) as an intervention. On the other hand, stud-
ies presented only as abstracts with no subsequent full report
of findings, on-going clinical studies, quasi-randomised study
design, in vitro or in vivo studies, review papers, non-English
literatures, studies involving patients with GDM, Type 1 dia-
betes mellitus (T1DM), any other metabolic diseases such as
obesity or hypercholesterolaemia, studies with no probiotic
genus/strains reported and studies using synbiotics (i.e. pro-
biotics combined with prebiotics) were all excluded. Whilst
the primary outcomes for meta-analysis were HbA1c and
FBG, the secondary outcomes included fasting plasma insu-
lin, homeostasis model assessment of insulin resistance
(HOMA-IR), inflammatory markers [i.e. C-reactive protein
(CRP), interleukin-6 (IL-6)] and anti-oxidative or oxidative
stress markers [i.e. malondialdehyde (MDA)]. The eligibility
of all potential studies identified for inclusion was indepen-
dently assessed by two review authors (SS and CFN). Discrep-
ancies on study inclusion were resolved through discussion
and consensus.
2.3. Data extraction and collection
Data (e.g. probiotic strains, dose, duration of intervention,
dosage forms, sample size, baseline and post-intervention
values for the aforementioned outcome measures) obtained
from the included studies were independently extracted by
two authors (SS and CFN), using a standardised, electronic
abstraction form.
2.4. Assessments of risk bias and publication bias
Risk bias for the included studies were independently
assessed (SS and CFN) using criteria as outlined in the
Cochrane Handbook for Systematic Reviews of Interventions
[19]. The assessment included selection bias (method for ran-
dom sequence generation and allocation concealment), per-
formance bias (blinding of participants and personnel),
detection bias (blinding of outcome assessment), attrition
174 diabetes research and clinical practice
bias (incomplete outcome data), reporting bias (selective
reporting) and other sources of bias. In addition, sample size
calculation and funding declaration associated with each
clinical trial were also assessed. Any disagreement was
resolved by discussion. Potential publication bias was
assessed using visual inspection of Funnel Plots (if RCTs
included in meta-analysis <10) and Eggers Test (if RCTs
included in meta-analysis >10).
2.5. Data analysis
Statistical analyses were performed by employing the Review
Manager Software version 5.3 (The Nordic Cochrane Centre,
Copenhagen, Denmark). Mean differences (MD) between
intervention (probiotics) and control group for each afore-
mentioned outcome were first pooled using the Fixed-Effect
Model. The statistical heterogeneity in each meta-analysis
was assessed using I2 and Chi2 statistics. Heterogeneity was
regarded as substantial if I2 > 50% or I2 > 25% with a low p
value of <0.10 in the Chi2 test. In this case, the Random-
Effect Model was performed. In order to explore the source
of heterogeneity, subgroup analyses were conducted by com-
paring the mean difference for each outcome measure as fol-
low: (a) probiotic dose [<109 colony forming unit (CFU) versus
P109 CFU], (b) duration of probiotic intervention (66 weeks
versus >6 weeks), (c) probiotic dosage forms (capsule versus
yogurt or kefir), and (d) probiotic genus (single versus multiple
genus). Also, sensitivity analyses were performed to examine
the robustness of meta-analysis findings. This was achieved
by excluding studies with small sample size (i.e. <20 patients)
for each intervention and control group.
3. Results
3.1. Description of included studies
Out of the 260 records that have been identified through the
search (after removal of duplicates), 254 records were
excluded based on the pre-specified criteria (Fig. 1). Table 1
outlines the details of all eligible studies and their findings.
Whilst a total of six eligible studies that have investigated
the use of probiotics in T2DM patients were included in the
qualitative synthesis (systematic review) [8,2024], only five
were included in quantitative synthesis (meta-analysis)
[8,2023]; all studies were published within the past five years.
The study by Andreasen et al. [24] was excluded from the
meta-analysis because findings for outcomes of interest
exclusive for T2DM patients were not reported. There were
two trials that used probiotics in yogurt form [8,20], one in
the form of kefir [23] and the remaining in capsule form
[21,22,24]. Except for one Denmark-based study [24], all stud-
ies [8,2023] were conducted in Middle East countries (i.e.
Iran). The majority of the trials (n = 4) used blends of various
probiotic genera, including lactobacilli and bifidobacteria
[8,20,22,23]. No changes in study participants diabetic medi-
cation regimens (e.g. metformin, glibenclamide, glitazones)
was reported in two of the studies trials [8,23]. Two trials
[20,22] were categorised as industry sponsored whereas four
studies explicitly stated no competing interests [8,21,23,24].
1 1 8 ( 2 0 1 6 ) 1 7 2 1 8 2
3.2. Risk bias and publication bias
Fig. 2(A) and (B) summarise the authors judgements on each
risk of bias item for all six RCTs included in this systematic
review. In general, most RCT were presented with either low
or unclear risk of bias. Publication bias was not assessed
using Eggers regression given that RCTs included in this
meta-analysis were less than 10 studies; the Funnel Plots
for all outcome measures were approximately symmetric.
3.3. Hypoglycaemic effects
3.3.1. HbA1c and FBG
Fig. 3(A) and (B) illustrate the Forest Plots of the pooled esti-
mates of probiotics on HbA1c and FBG, respectively. All six
studies [8,2024] investigated the hypoglycaemic effects of
probiotics; four studies documented significant decrease
(p < 0.05) in HbA1c and/or FBG levels. In the current meta-
analysis, no significant difference was noted in HbA1c
(MD =
0.11%, 95% CI:
1.00, 0.78, p = 0.81) between the probi-
otic and control groups [Fig. 3(A)]. Significant substantial
inter-study heterogeneity was noted (I2 = 91%, p < 0.00001).
Subgroup analyses (Table 2) revealed that the extreme incon-
sistency among four trials in HbA1c (I2 = 91%) was reduced to
I2 = 55% when differences in probiotic dose and dosage form
are accounted for. A significant increase (p < 0.05) in HbA1c
was noted in trials using probiotic dose of P109 CFU and cap-
sule dosage form. The findings, however, remain inconclusive
as the number of trials included in these subgroups was
small. No significant difference in HbA1c lowering effect of
probiotics was observed in trials with administration of probi-
otics >6 weeks (p = 0.79). Subgroup analysis for probiotic
genus was not performed as all trials included for HbA1c anal-
ysis used multiple probiotic genus products.
In contrast, a significantly lower FBG at 0.98 mmol/L (95%
CI:
1.17,
0.78, p < 0.00001), but with non-significant moder-
ate inter-study heterogeneity was noted (I2 = 32%, p = 0.21).
Further to removal of study with small sample size (i.e.
n < 20 for each group) and the use of Random-Effect Model,
no change in the significance of the pooled estimates of FBG
(MD =
0.98 mmol/L; 95% CI:
1.18,
0.78, p < 0.00001 and
MD =
0.95 mmol/L; 95% CI:
1.52,
0.39, p = 0.001) was
noted, respectively. Subgroup analysis revealed that a signifi-
cantly lower FBG was noted in those trials with administra-
tion of probiotics >6 weeks; greater consistency was
observed within the subgroups as well (Table 2). The use of
probiotic capsules has resulted in a significantly lower FBG
with low heterogeneity (I2 = 0%, p = 0.55). Significantly lower
FBG was noted in both subgroups receiving probiotic dose
<109 CFU and P109 CFU. No significant difference in FBG low-
ering effect of probiotics was observed in trials using single or
multiple probiotic genus.
3.3.2. HOMA-IR and fasting plasma insulin
Three studies investigated the effects of probiotics against
insulin resistance [21,22,24] but only one [22] with positive
findings. Forest Plot of the pooled estimates of probiotics on
HOMA-IR [Supplementary Fig. 1(A)] revealed a non-
significant difference between the groups (MD =
0.79, 95%
 diabetes research and clinical practice
Fig. 1 Study selection process.
CI:
2.71, 1.14, p = 0.42) but with significant substantial inter-
study heterogeneity observed (I2 = 85%, p = 0.009). Subgroup
analyses was not performed given the number of studies
included was small (n = 2).
There were only three studies that documented the effects
of probiotics on fasting insulin concentration [2022]. There
was, however, no significant difference in fasting plasma
insulin (MD = 0.13 lIU/mL, 95% CI:
3.02, 3.28, p = 0.93)
between the probiotic and control groups. The findings were
presented with significant substantial inter-study hetero-
geneity (I2 = 95%, p < 0.00001) [Supplementary Fig. 1(B)]. Sub-
sequent removal of the study with small sample size [21]
had led to no change in the significance of the pooled esti-
mates of fasting plasma insulin (MD =
1.04 lIU/mL, 95% CI:

3.62, 1.54, p = 0.43). Subgroup analyses (Table 2) of trials
using probiotic dose P109 CFU and >6 weeks revealed signifi-
cant reductions of fasting plasma insulin. Also, significant
increase of fasting plasma insulin was noted in the trial using
single probiotic genus; the findings, however, remain incon-
clusive as the number of trials included in these subgroups
was small. The extreme inconsistency among three trials in
fasting plasma insulin (I2 = 95%) was reduced to I2 = 52% when
differences in probiotic dose and duration of trial were
1 1 8 ( 2 0 1 6 ) 1 7 2 1 8 2 175
accounted for. There was no significant association between
the pooled estimates of probiotics on fasting plasma insulin
and probiotic dosage form.
3.4. Anti-inflammatory effects
A total of four studies reported on anti-inflammatory effects
of probiotics, and all were with mixed responses [8,21,22,24].
Both Forest Plots indicated that there were no significant dif-
ferences in CRP [Supplementary Fig. 2(A)] (MD =
0.47 mg/L,
95% CI:
1.25, 0.32, p = 0.25) and IL-6 [Supplementary Fig. 2
(B)] (MD =
0.57 pg/mL, 95% CI:
3.26, 2.12, p = 0.68) levels,
respectively. Significant inter-study heterogeneity was
observed in the overall analyses for CRP (I2 = 76%, p = 0.02)
and IL-6 (I2 = 87%, p = 0.005) levels. The removal of a study
with small sample size [21], however, led to significant reduc-
tion of CRP level (MD =
0.93 mg/L; 95% CI:
1.20,
0.66,
p < 0.00001) and low heterogeneity (I2 = 0%, p = 0.34). Subgroup
analyses of trials (Table 2) revealed significant reductions of
CRP in groups using probiotic dose P109 CFU and multiple
probiotic genus for >6 weeks; the findings, however, remain
inconclusive as the number of trials included in these sub-
groups was small. The inconsistency among three trials in
 176
Table 1 Clinical studies on the efficacy of probiotics in type 2 diabetes mellitus patients (n = 6).
Author, Country Probiotic Strains (Dose) Study Population Intervention Study Findings
(n) Period
Hypoglycaemic Anti-inflammatory Anti-oxidative

Mohamadshahi Lactobacillus delbrueckii subsp. T2DM patients 8 weeks ;HbA1cb, FBG b

;TNF-a , IL-6, CRP
et al. (9), Iran bulgaricusa (n = 54)
Streptococcus thermophilusa
Bifidobacterium animalis subsp.
lactis Bb12 (3.7  106 CFU/g)

diabetes research and clinical practice

L. acidophilus (3.7  106 CFU/g)
Ejtahed et al. (64), L. acidophilus La5 T2DM patients 6 weeks ;HbA1cb, FBGb "SODb, GPxb, TAS
Iran (7.23  106 CFU/g) (n = 64) ;MDA
B. lactis Bb12
(6.04  106 CFU/g)
L. bulgaricusa
Streptococcus thermophilusa
Mazloom et al. L. acidophilusa T2DM patients 6 weeks ;FBG ;IL-6 ;MDA
(65), Iran L. bulgaricusa (n = 40) "CRP
L. bifiduma
L. caseia
Asemi et al. (69), L. acidophilus (2  109 CFU/g) T2DM patients 8 weeks ;HbA1c, FBGb ;CRPb "GSHb, TAC
Iran L. rhamnosus (1.5  109 CFU/g) (n = 54)
L. bulgaricus (2  108 CFU/g)
L. casei (7  109 CFU/g)
S. thermophilus (1.5  109 CFU/g)

1 1 8 ( 2 0 1 6 ) 1 7 2 1 8 2
B. longum (7  109 CFU/g)
B. breve (2  1010 CFU/g)
Ostadrahimi et al. L. casei (2  106 CFU/mL) T2DM patients 8 weeks ;HbA1cb, FBGb
(67), Iran L. acidophilus (3  106 CFU/mL) (n = 60)
B. lactis (0.5  106 CFU/mL)
Andreasen et al. L. acidophilus NCFM T2DM patients 4 weeks No changes in No changes in
(68), Denmarkc (1  1010 CFU/g) (n = 18) HbA1c TNF-a
Impaired glucose ;FBG "IL-6, IL-1rA
tolerance (n = 5) ;CRP
Healthy subjects
(n = 22)
CRP = C-reactive protein; FBG = fasting blood glucose; GPx = glutathione peroxidase; GSH = gluthathione; HbA1c = glycated haemoglobin; IL-6 = interleukin-6; IL-1rA = IL-1 receptor antagonist;
MDA = malondialdehyde; SOD = superoxide dismutase; TAC = total anti-oxidant capacity; TAS = total anti-oxidant status; TNF-a = Tumor Necrosis Factor-a
a
No dose or strength was reported in this study.
b
Significant difference was noted between the intervention and placebo groups (post-intervention).
c
Excluded from the meta-analysis.
 diabetes research and clinical practice 1 1 8 ( 2 0 1 6 ) 1 7 2 1 8 2 177

Fig. 2 Risk of bias (A) graph, and (B) summary.

CRP (I2 = 75%) was reduced when differences in probiotic dose
(I2 = 28%), probiotic genus (I2 = 0%) and duration of trial
(I2 = 0%) were accounted for. There was no significant evi-
dence for an association between reduction of CRP level and
the probiotic dosage form. Subgroup analyses for IL-6 was
not performed given the small number of studies included
(n = 2).
3.5. Anti-oxidative effects
There were only three studies that documented the anti-
oxidative properties of probiotics [2022]. As there was only
one single study that investigated the changes of SOD, GPx,
CAT, total antioxidant status [20] and GSH [22] after probiotic
consumption, no pooled analysis was performed for the
abovementioned parameters. Forest Plot of the pooled esti-
mate of probiotics on MDA (Supplementary Fig. 3) revealed
non-significant difference between the groups (p = 0.54), with
low inter-study heterogeneity noted (I2 = 0%, p = 0.90).
4. Discussion
The current meta-analysis is the first to examine the pooled
estimate of probiotics on HbA1c. The use of HbA1c is a gold
standard in clinical management of T2DM [25] and it permits
comparability among the published studies. A 1% reduction in
HbA1c has been associated with 21% risk reduction of diabetes
related-end points and 37% risk reduction for microvascular
complications [26]. Hence, a 1% reduction in HbA1c is consid-
ered to be clinically relevant. The findings from the current
meta-analysis revealed no significant difference in HbA1c
between the probiotic and the control groups; substantial
heterogeneity, however, was noted across the clinical studies.
Even though results of the subgroup analyses indicated that
178 diabetes research and clinical practice
Fig. 3 Forest plot of the effect of probiotics on (A) HbA1c, and (B) FBG.
there were significant associations between the effects of pro-
biotics on HbA1c and the probiotic dose or capsule dosage
form, but not for the duration of probiotic intervention, these
findings were limited by the small number of studies included
in the subgroup analyses. One of the common limitations in
all the studies included in this review is the fact that none
investigated the long term effects of probiotics; all these stud-
ies were conducted in a period ranging from 4 to 8 weeks.
Accordingly, the findings on HbA1c from these studies were
doubtful given that the changes in HbA1c could only be
detected every three months as per the life cycle of red blood
cells [27]. In addition, only one study [23] in this review deter-
mined sample sizes based on the reduction in HbA1c between
intervention and placebo groups; the other two studies which
reported a power calculation were aimed to detect the differ-
ences in pro-inflammatory markers [i.e. hypersensitive (hs)-
CRP] [22] or anti-oxidative enzymes (i.e. CAT) [20]. Of note,
having sample size based on the primary outcome provides
a better focus path for trial as well as adding quality to the
results obtained [28,29].
Consistent with Ruan et al. [18], a significantly lower FBG
upon probiotic intake when compared to the control group
was documented. A strength of the current findings is the fact
that insignificant moderate inter-study heterogeneity was
noted, as opposed to the findings by Ruan et al. [18]. Of note,
the pooled estimate of probiotics on FBG in the current meta-
analysis was primarily driven by the findings of Asemi et al.
[22], in which the intake of probiotics exerted a preventive
effect on the rise of FBG rather than resulting in a lower
FBG level, as compared to placebo group. In addition, our sub-
group analysis revealed that the lowering effect of FBG was
significant in those trials with intervention period >6 weeks.
These clinical findings were consistent with the in vivo studies
using diabetic mice and rat models, which reported signifi-
cant reductions of glycaemic parameters (i.e. FBG and 2 h post
prandial glucose), and improvement in glucose tolerance
[3032]. The intervention period in in vivo studies were usually
1 1 8 ( 2 0 1 6 ) 1 7 2 1 8 2
longer (i.e. 8 to 20 weeks) [3033], in comparison to the clinical
studies (i.e. 4 to 8 weeks). The specific mechanisms underly-
ing the hypoglycaemic effects of probiotics, however, remain
unclear. Some researchers described these hypoglycaemic
effects as the positive consequences of probiotic-mediated
decrease in systemic inflammation or oxidative stress [5,18].
Earlier meta-analysis by Ruan et al. [18] did not examine
the pooled anti-inflammatory effects of probiotics. In the cur-
rent meta-analysis, no significant differences in CRP and IL-6
levels, however, were noted in the probiotic group, from which
significant heterogeneity observed. Our findings suggested
that high probiotic dose, longer duration of probiotic interven-
tion and the use of multiple probiotic genuses associated sig-
nificantly with better anti-inflammatory effects of probiotics.
Of note, the removal of study with small sample size [21]
had led to significant reduction of CRP level and low hetero-
geneity among those in the probiotic group. The inconsistency
of results had also been observed in vivo. Although a signifi-
cant decrease of TNF-a and IL-6 levels were noted among
colitis-induced rats fed with probiotics [34], yet another study
revealed a statically indifferent insignificant decrease of IFN-c
[35]. It is important to note that these in vivo studies used dif-
ferent inflammatory markers, probiotic genus/strain and
feeding duration, making it difficult to directly compare and
assess the anti-inflammatory effects of probiotics.
Similar to systemic inflammation, the levels of free radical
and oxidative stress in T2DM patients were also reported to be
higher than the healthy subjects [3638]. It has been postu-
lated that probiotic consumption would increase the level of
anti-oxidative enzymes (i.e. SOD, GPx, CAT) which are able
to rapidly scavenge reactive oxygen species, and therefore,
reduce the rate of oxidative stress in T2DM patients [20,21].
In agreement with the clinical studies [20,22], the majority
of the in vivo studies observed significantly increase anti-
oxidative enzyme activities (i.e. SOD and GSH-Px) in hyperlip-
idaemic and diabetic rats [39,40] after consumption of L. casei
and L. acidophilus (7.56  1072  1010 CFU/mL) for one to four
 diabetes research and clinical practice 1 1 8 ( 2 0 1 6 ) 1 7 2 1 8 2 179

weeks. The current meta-analysis, however, revealed a non-

Pheterogeneity
significant difference in MDA, with low inter-study hetero-

0.006
0.24

0.34

0.34
geneity noted.

Contrary to the findings by Ruan et al. [18], the current

I2 (%)
study revealed no significant difference in HOMA-IR in groups

28

87
0

that have received probiotics. It is interesting to note that

<0.00001

<0.00001

<0.00001
Ruan et al. [18] indicated an opposing pooled effect of

0.24

0.52

0.50
0.31

0.52
probiotics on fasting plasma insulin in which a significant
P
Anti-inflammatory Effects

0.97 (
1.25,
0.69) reduction was observed, particularly in the hyperglycaemic

0.93 (
1.20,
0.66)

0.93 (
1.20,
0.66)

0.07 (
0.82, 0.67)

0.41 (
1.62, 0.79)

0.49 (
1.44, 0.46)
0.27 (
0.56, 1.10)

0.27 (
0.56, 4.10)

subgroup. The current study, however, suggested otherwise.
MD (95%CI)

The pooled effect of probiotics on fasting plasma insulin in

the current meta-analysis was sensitive to the omission of
CRP (mg/L)

RCT with small sample size. The inconsistencies in the pooled

findings of fasting plasma insulin and insulin resistance
n

2
1

1
2

2
1

1
2
between the current work and Ruan et al. [18] have hindered
Pheterogeneity

<0.00001

meaningful conclusions to be made and these data should be

0.15

0.15

0.02

interpreted with caution given the substantial inter-study

heterogeneity observed in both meta-analyses.

I2 (%)

Also, there remain gaps in the knowledge albeit the

52

52

97

83

increase of number of trials that investigate the efficacy of

<0.00001

<0.00001

0.0003

probiotics in T2DM in clinical settings. Inconsistencies in

0.07

0.07

1.00
0.65

0.43
P

the findings were noted across the clinical studies included

2.19 (
2.90,
1.48)

2.19 (
2.90,
1.48)

in this meta-analysis [8,2022,24]. The possible reasons

0.01 (
4.34, 4.32)

1.04 (
3.62, 1.54)

1.55 (
0.13, 3.23)

1.55 (
0.13, 3.23)

0.47 (
1.58, 2.52)

underlying this observation remain to be elucidated. It is

2.23 (1.01, 3.45)
Insulin (lIU/mL)

MD (95%CI)

believed that the beneficial effects of probiotics (i.e. hypogly-

Fasting Plasma

caemic, anti-inflammatory or anti-oxidative properties) could

be highly strain specific. The use of a single strain, such as L.
acidophilus NCFM in the study by Andreasen et al. [24] and Lac-
n

2
1

2
1

2
1

1
2

tobacillus spp. in the study by Mazloom et al. [21], could be the

Pheterogeneity

factor explaining the negative outcome observed. Marteau

0.13

0.97
0.25

0.55
0.07

0.14

[41] claimed that the beneficial outcome of probiotic adminis-

tration could be most likely achieved with the combined use

I2 (%)

47

28

63

46

of multiple probiotic strains. Different strains or species

0

seemed to exhibit different mechanisms of action, and how

<0.00001

<0.00001

<0.00001

these differences influence the study outcome (i.e. glycaemic

0.03

0.71

0.09

0.86
0.14
P

control) are yet to be determined in clinical trials. Lactobacilli

0.81 (
1.55,
0.07)

0.99 (
1.19,
0.79)

1.02 (
1.22,
0.81)

0.98 (
1.19,
0.78)

1.01 (
1.67,
0.35)

(i.e. L. bulgaricus, L. plantarum, L. acidophilus), for instance, have

0.18 (
1.09, 0.74)

1.22 (
2.62, 0.19)

0.22 (
2.74, 2.30)

Table 2 Results of Subgroup Analyses of RCT in Meta-analysis.

been shown to be useful in lowering total cholesterol [42],

MD (95%CI)

increasing high-density lipoprotein level [43] and exhibiting

FBG (mmol/L)

anti-carcinogenic effects [44]. The bifidobacteria (i.e. B.

longum, B. breve, B. adolescentis), on the other hand, are benefi-
n

4
1

2
3

2
3

1
4

cial in reducing the secretion of pro-inflammatory cytokines

Pheterogeneity

(i.e. IL-12, TNF-a) [45], minimising the risk of colon cancer

<0.00001

and reducing Helicobacter pylori infection symptoms [46]. Nev-

0.11

0.11

ertheless, due to great variability of probiotic strains used in

the reported RCT and the relatively small number of the exist-
I2 (%)

55

92

55

ing RCT, subgroup analyses for assessment of the efficacy of

each probiotic strain on hypoglycaemic, anti-inflammatory

<0.00001

<0.00001

and anti-oxidant effects were not able to be performed in

0.17

1.00
0.79

0.17
P

the current meta-analysis.

0.45 (
1.10, 0.19)

0.19 (
1.57, 1.19)

0.45 (
1.10, 0.19)

Hypoglycaemic effects

Another important consideration for interpreting the

0.00 (
0.50, 0.50)
0.88 (0.69, 1.07)

0.88 (0.69, 1.07)

inconsistencies of the study findings is the homogeneity of

MD (95%CI)

the study populations. The study by Andreasen et al. [24],

HbA1c (%)

for instance, have included three different groups of subject

participants [i.e. T2DM patients (n = 18), subjects with

Duration of intervention
n

3
1

1
3

1
3

0
4
Probiotic dosage forms

impaired glucose tolerance (n = 5) and healthy subjects

P10 billion CFU
<10 billion CFU

Yogurt or kefir
Probiotic genus

(n = 22)]. The heterogeneity of the subjects in this study [24]

Probiotic dose
Subgroups

66 weeks

could be the possible reason for the absent effect on inflam-

>6 weeks

Multiple
Capsule

Single

matory parameters. The heterogeneity of subjects or the

absence of standardisation in subject population could lead
180 diabetes research and clinical practice
to variation in results, overestimation and bias [47]. In addi-
tion, neither of these studies has profiled the changes of gut
microbiota and evaluated the quality of life of T2DM patients
upon administration of probiotics. No published studies have
investigated the role of probiotics on gut hormone regulation
as well. Well-designed clinical trials that address the afore-
mentioned limitations and considerations are needed to
determine the efficacy of probiotics in T2DM setting.
The marked increase of clinical studies on the application
of probiotics in T2DM patients over the last two years is an
acknowledgement of the potential usefulness of probiotics in
managing glycaemic control [16]. Nevertheless, the quality of
reporting for the majority of the RCTwas low, lacking adequate
information to facilitate understanding of the trials design,
conduct, analysis and interpretation, and to assess the validity
of its findings. As revealed in the risk of bias graph [Fig. 2(A)]
and summary [Fig. 2(B)], most of the RCT have unclear risk of
bias. Of the six clinical studies included in this systematic
review, three [8,21,24] did not report on the power calculation
when determining their sample sizes; most RCT randomised
a small number of participants (i.e. 3460) [8,2024]. As such,
it remains unknown if the effect of a given size could be
detected in these studies [48]. Four [8,2123] RCT did not men-
tion the concealment of their treatment allocation and the
steps taken to conceal the sequence of intervention that were
assigned. In RCT, it is vital to make sure that treatment alloca-
tion is conducted to eliminate all potential selection bias [49].
Most studies reported only the genus and species [21,23], but
not the strains of probiotics that were given to the intervention
arm. Additionally, there was one study that did not reveal the
dose of probiotics at all [21]. In all the clinical studies identified,
none has indicated if an ITT analysis was performed [8,2024].
Also, all did not explicitly state the primary and secondary out-
comes of the clinical trials. In addition, it is apparent that none
of the studies reported on adverse events of the probiotic prod-
ucts used in their trials; therefore, the safety profile of these
probiotics remain unknown.
The present work has several limitations. Firstly, signifi-
cant substantial unexplained inter-study heterogeneity was
noted in most analyses except for FBG and MDA. Due to the
small number of RCT included in this meta-analysis for cer-
tain outcome measures, most subgroup analyses were under-
powered in assessing confounding factors (e.g. the efficacy of
certain probiotic strains or blends) that would have influenced
glycaemic control in T2DM patients. The lack of assessment of
probiotic specific adverse events was another limitation.
5. Conclusion
The current meta-analysis suggested moderate beneficial
hypoglycaemic effects of certain probiotics, with significantly
lower FBG. Findings on HBA1c, anti-inflammatory and anti-
oxidative effects of probiotics in the clinical setting, however,
remain inconsistent, and thus, merits further investigations
in future clinical studies. Based on the current literature, well
designed, prospective clinical studies investigating the effects
of probiotic administration on glycaemic control in T2DM
patients remain scarce. Existing clinical trials are limited by
their variation in the study design, quality and depth of the
1 1 8 ( 2 0 1 6 ) 1 7 2 1 8 2
methodologies employed, and thus hindering meaningful
conclusions to be made.
Declaration of interest
No competing interest declared by the authors.
Authors contribution
Each author contributed equally in this study.
Acknowledgements
The authors thank the Ministry of Education Malaysia for
financial support under the Fundamental Research Grant
Scheme [600-RMI/FRGS 5/3 (22/2014)].
Appendix A. Supplementary data
Supplementary data associated with this article can be found,
in the online version, at http://dx.doi.org/10.1016/j.diabres.
2016.06.014.
R E F E R E N C E S
[1] Boada CAC, Martinez-Moreno JM. Pathophysiology of
diabetes mellitus type 2: beyond the duo insulin resistance-
secretion deficit. Nutr Hosp 2013;28:7887.
[2] Inzucchi SE, Bergenstal RM, Buse JB, Diamant M, Ferrannini E,
Nauck M, et al. Management of hyperglycemia in type 2
diabetes: a patient-centered approach: position statement of
the American Diabetes Association (ADA) and the European
Association for the Study of Diabetes (EASD). Diabetes Care
2012;35:136479.
[3] Guariguata L, Whiting DR, Hambleton I, Beagley J,
Linnenkamp U, Shaw JE. Global estimates of diabetes
prevalence for 2013 and projections for 2035. Diabetes Res
Clin Pract 2014;103:13749.
[4] Zhang P, Zhang X, Brown J, Vistisen D, Sicree R, Shaw J, et al.
Global healthcare expenditure on diabetes for 2010 and 2030.
Diabetes Res Clin Pract 2010;92:301.
[5] Panwar, Rashmi HM, Batish VK, Grover S. Probiotics as
potential biotherapeutics in the management of type 2
diabetes-prospects and perspectives. Diabetes Metab Res Rev
2013;29:10312.
[6] Hu FB, Manson JE, Stampfer MJ, Colditz G, Liu S, Solomon CG,
et al. Diet, lifestyle, and the risk of type 2 diabetes mellitus in
women. N Engl J Med 2001;345:7907.
[7] Zhang Y, Zhang H. Microbiota associated with type 2 diabetes
and its related complications. Food Sci Hum Wellness
2013:16772.
[8] Mohamadshahi M, Veissi M, Haidari F, Shahbazian H,
Kaydani G-A, Mohammadi F. Effects of probiotics yogurt
consumption on inflammatory biomarkers in patients with
type 2 diabetes. BioImpacts 2014;4:838.
[9] Jayashree B, Bibin YS, Prabhu D, Shanthirani CS,
Gokulakrishnan K, Lakshmi BS, et al. Increased circulatory
levels of lipopolysaccharide (LPS) and zonulin signify novel
biomarkers of proinflammation in patients with type 2
diabetes. Mol Cell Biochem 2014;388:20310.
[10] Khani S, Hosseini HM, Taheri M, Nourani MR, Fooladi AAI.
Probiotics an alternative strategy for prevention and
 diabetes research and clinical practice
treatment of human disease: a review. Inflamm Allery Drug
2012;11:7989.
[11] Markets and Markets. Probiotic Ingredients Market by Function [27]
(Regular, Preventative, Therapy), Application (Food & Beverage,
Dietary Supplements, & Animal Feed), End Use (Human &
Animal Probiotics), Ingredient (Bacteria & Yeast), and by Region
Global Trends & Forecast to 2020, 2015. Retrieved 18 February, [28]
2016, from <http://www.marketsandmarkets.com/Market-
Reports/probiotic-market-advanced-technologies-and-global-
market-69.html>. [29]
[12] Shida K, Sato T, Iizuka R, Hoshi R, Watanabe O, Igarashi T,
et al. Daily intake of fermented milk with Lactobacillus casei
strain Shirota reduces the incidence and duration of upper
respiratory tract infections in healthy middle-aged office [30]
workers. Eur J Nutr 2015:19.
[13] Kim E-H, Hong H, Hong K-S, Choi K-S, Han Y-M, Kangwan N,
et al. High concentrated probiotics improve inflammatory
bowel diseases better than commercial concentration of
probiotics. J Food Drug Anal 2012;20:2925. [31]
[14] Wong S, Jamous A, ODriscoll J, Sekhar R, Weldon M, Yau CY,
et al. A Lactobacillus casei Shirota probiotic drink reduces
antibiotic-associated diarrhoea in patients with spinal cord
injuries: a randomised controlled trial. Br J Nutr [32]
2014;111:6728.
[15] Goldin BR, Gorbach SL. Clinical indications for probiotics: an
overview. Clin Infect Dis 2008;46.
[16] Gomes AC, Bueno AA, Souza RGMD, Mota JF. Gut microbiota, [33]
probiotics and diabetes. Nutr J 2014;13:113.
[17] Carvalho BM, Saad MJA. Influence of gut microbiota on
subclinical inflammation and insulin resistance. Mediat [34]
Inflamm 2013:113.
[18] Ruan Y, Sun J, He J, Chen F, Chen R, Chen H. Effect of
probiotics on glycemic control: a systematic review and
meta-analysis of randomized, controlled trials. PLoS One [35]
2015:10.
[19] Higgins J, Green S, editors. Cochrane Handbook for
Systematic Reviews of Interventions Version 5.1.0. The
Cochrane Collaboration; 2011 [updated March 2011]. [36]
[20] Ejtahed H, Mohtadi-Nia J, Homayouni-Rad A, Niafar M,
Asgari-Jafarabadi M, Mofid V. Probiotic yogurt improves
antioxidant status in type 2 diabetic patients. Nutrition [37]
2012;28:53943.
[21] Mazloom Z, Yousefinejad A, Dabbaghmanesh MH. Effect of
probiotics on lipid profile, glycemic control, insulin action,
oxidative stress and inflammatory markers in patients with
Type 2 Diabetes: a clinical trial. Iran J Med Sci 2013;38: [38]
3843.
[22] Asemi Z, Zare Z, Shakeri H, Sabihi S-S, Esmaillzadeh A. Effect
of multispecies probiotic supplements on metabolic profiles, [39]
hs-CRP and oxidative stress in patients with type 2 diabetes.
Ann Nutr Metab 2013;63:19.
[23] Ostadrahimi A, Taghizadeh A, Mobasseri M, Farrin N, [40]
Payahoo L, Ghheshlaghi ZB, et al. Effect of probiotic
fermented Milk (Kefir) on glycemic control and lipid profile in
type 2 diabetic patients: a randomized double-blind placebo- [41]
controlled clinical trial. Iran J Public Health 2015;44:22837.
[24] Andreasen AS, Larsen N, Pedersen-Skovsgaard T, Berg RM,
Moller K, Svendsen KD, et al. Effects of Lactobacillus
acidophilus NCFM on insulin sensitivity and the systemic [42]
inflammatory response in human subjects. Br J Nutr
2010;104:18318.
[25] Nazaimoon WMW, Isa SHM, Mohamad WBW, Khir AS,
Kamaruddin NA, Kamarul IM, et al. Prevalence of diabetes in [43]
Malaysia and usefulness of HbA1c as a diagnostic criterion.
Diabet Med 2013:14.
[26] Stratton IM, Adler AI, Andrew H, Neil W, Matthews DR, [44]
Manley SE, et al. Association of glycaemia with
macrovascular and microvascular complications of type 2
1 1 8 ( 2 0 1 6 ) 1 7 2 1 8 2 181
diabetes (UKPDS 35): a prospective observational study. Br
Med J 2000;321:40512.
Soranzo N. Genetic determinants of variability in glycated
hemoglobin (HbA1c) in humans: review of recent progress
and prospects for use in diabetes care. Curr Diab Rep
2011;11:5629.
Noordzij M, Tripepi G, Dekker FW, Zocalli C, Tanck MW, Jager
KJ. Sample size calculations: basic principles and common
pitfalls. Nephrol Dial Transplant 2010;24:138893.
Inzucchi SE, Bergenstal RM, Buse JB, Diamant M, Ferrannini
E, Nauck M, et al. Management of hyperglycemia in Type 2
diabetes: a patient-centered approach. Diabetes Care
2012;35:136479.
Alsayadi M, Jawfi YA, Belarbi M, Soualem-Mami Z, Merzouk
H, Sari DC, et al. Evaluation of anti hyperglycemia and anti
hyperlipidemic activities of water kefir as probiotic on
streptozotocin-induced diabetic wistar rats. J Diabetes
Mellitus 2014;4:8595.
Stenman LK, Waget A, Garret C, Klopp P, Burcelin R, Lahtinen
S. Potential probiotic Bifidobacterium animalis ssp. lactis 420
prevents weight gain and glucose intolerance in diet-induced
obese mice. Beneficial Microbes 2014;5:43745.
Yun SI, Park HO, Kang JH. Effect of Lactobacillus gasseri
BNR17 on blood glucose levels and body weight in
a mouse model of type 2 diabetes. J Appl Microbiol
2009;107:16816.
Andersson U, Branning C, Ahrne S, Molin G, Alenfall J, Onning
G, et al. Probiotics lower plasma glucose in the high-fat fed
C57BL/6J mouse. Beneficial Microbes 2010;1:18996.
Dai C, Zhen C-Q, Meng FJ, Zhou Z, Sang L-X, Jiang M. VSL#3
probiotic exerts the anti inflammatory activity via P13k/Akt
and NF-kB pathway in rat model of Dss-induced colitis. Mol
Cell Biochem 2013;374:111.
Valladares R, Sankar D, Li N, Williams E, Lai K-K, Abdelgeliel
AS, et al. Lactobacillus johnsonii N6.2 mitigates the
development of type 1 diabetes in BB-DP rats. PLoS One
2010;5:19.
Henriksen EJ, Diamond-Stanic MK, Marchionne EM.
Oxidative stress and the etiology of insulin resistance and
type 2 diabetes. Free Radical Biol Med 2011;51:9939.
Faghihi T, Radfar M, Barmal M, Amini P, Qorabani M,
Abdollahi M, et al. A randomized, placebo controlled trial of
selenium supplementation in patients with type2 diabtes:
effect on glucose homeostasis, oxidative stress and lipid
profile. Am J Ther 2014;21:4915.
Al-Rawi NH. Oxidative stress, antioxidant status and lipid
profile in the saliva of type 2 diabetics. Diab Vasc Dis Res
2011;8:228.
Zhang Y, Du R, Wang L, Zhang H. The antioxidative effects of
probiotic Lactobacillus casei Zhang on the hyperlipidemic rats.
Euro Food Res Technol 2010;231:1518.
Harisa GI, Taha EI, Khalil AF, Salem MM. Oral administration
of Lactobacillus acidophilus restores nitric oxide level in
diabetic rats. Aust J Basic Appl Sci 2009;3:29639.
Marteau P. Evidence of probiotic strain specificity makes
extrapolation of results impossible from a strain to another,
even from the same species. Ann Gastroentol Hepatol
2011:13.
Nguyen TDT, Kang JH, Lee MS. Characterization of
Lactobacillus plantarum PH04, a potential probiotic bacterium
with cholesterol-lowering effects. Int J Food Microbiol
2007;113:35861.
Ranasinghe J, Silva S, Herath N. Changes of serum lipids and
protein during probiotics feeding and its exposure. Int J Sci
Res Publ 2013;3.
Wollowski I, Rechkemmer G, Pool-Zobel BL. Protective role of
probiotics and prebiotics in colon cancer. Am J Clin Nutr
2001;73:451s5s.
182 diabetes research and clinical practice
[45] Abd El-Gawad IA, El-Sayed EM, Hafez SA, El-Zeini HM, Saleh
FA. The hypocholesterolaemic effect of milk yoghurt and soy-
yoghurt containing bifidobacteria in rats fed on a cholesterol-
enriched diet. Int Dairy J 2005;15:3744.
[46] Myllyluoma E, Veijola L, Ahlroos T, Tynkkynen S, Kankuri E,
Vapaatalo H, et al. Probiotic supplementation improves
tolerance to Helicobacter pylori eradication therapy a
placebo-controlled, double-blind randomized pilot study.
Aliment Pharmacol Ther 2005;21:126372.
1 1 8 ( 2 0 1 6 ) 1 7 2 1 8 2
[47] Noto H, Goto A, Tsujimoto T, Noda M. Cancer risk in diabetic
patients treated with metformin: a systematic review and
meta-analysis. PLoS One 2012;7:19.
[48] Button KS, Ioannidis JP, Mokrysz C, Nosek BA, Flint J,
Robinson ESJ, et al. Power failure: why small sample size
undermines the reliability of neuroscience. Nat Rev Neurosci
2013;14:36576.
[49] Schulz KF, Grimes DA. Allocation concealment in randomised
trials: defending against deciphering. Lancet 2002;359:6148.