Accepted Manuscript Title: The Efficacy and Safety of DPP4 Inhibitors compared to Sulfonylureas as add-on Therapy to Metformin in Patients with Type 2 diabetes: A Systematic Review and Meta-analysis Author: Basem M. Mishriky Doyle M. Cummings Robert J. Tanenberg PII: DOI: Reference:
S0168-8227(15)00255-7 http://dx.doi.org/doi:10.1016/j.diabres.2015.05.025 DIAB 6401
To appear in:
Diabetes Research and Clinical Practice
Received date: Revised date: Accepted date:
29-9-2014 29-1-2015 2-5-2015
Please cite this article as: B.M. Mishriky, D.M. Cummings, R.J. Tanenberg, The Efficacy and Safety of DPP4 Inhibitors compared to Sulfonylureas as add-on Therapy to Metformin in Patients with Type 2 diabetes: a Systematic Review and Meta-analysis, Diabetes Research and Clinical Practice (2015), http://dx.doi.org/10.1016/j.diabres.2015.05.025 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
HIGHLIGHTS Title: The Efficacy and Safety of DPP4 Inhibitors compared to Sulfonylureas as add-on Therapy to Metformin in Patients with Type 2 diabetes: a Systematic Review and Meta-analysis
ip t
Meta-analysis comparing DPP-4 inhibitors vs. Sulfonylureas added to metformin
Sulfonylureas lower HbA1c more early; this difference is lost at one year follow-up
cr
Sulfonylureas are less expensive but have a greater risk of hypoglycemia
Ac ce p
te
d
M
an
us
DPP-4 inhibitors cause little hypoglycemia and weight loss instead of weight gain
1 Page 1 of 39
The Efficacy and Safety of DPP4 Inhibitors compared to Sulfonylureas as add-on Therapy to Metformin in Patients with Type 2 diabetes: a Systematic Review and Meta-analysis
ip t
Basem M. Mishriky, MD; Department of Internal Medicine, East Carolina University, Greenville, NC, United States
cr
Doyle M. Cummings, PharmD, FCP, FCCP;
us
Department of Family Medicine, East Carolina University, Greenville, NC, United States
an
Robert J. Tanenberg, MD, FACP;
M
Division of Endocrinology, East Carolina University, Greenville, NC, United States
d
Corresponding author:
te
Doyle M. Cummings, Pharm.D., Professor, Department of Family Medicine, Brody School of
Ac ce p
Medicine at East Carolina University, 101 Heart Drive, MS #654, Greenville, NC 27834; 252744-2607; [email protected]
2 Page 2 of 39
Abstract: There is no consensus on the selection of specific drug therapies when metformin fails in type 2 diabetes (T2D). This meta-analysis was performed to determine the efficacy and safety of
ip t
Dipeptidyl peptidase-4 inhibitors (DPP4-I) compared to sulfonylurea (SU) as add-on therapy to
cr
metformin in inadequately controlled T2D patients. We searched MEDLINE, CENTRAL,
EMBASE, and CINAHL for randomized trials comparing DPP4-I to SU as add-on therapy to
us
metformin and reported a change in hemoglobin A1c (HbA1c). Sixteen articles were included.
an
There was a significantly greater reduction in HbA1c from baseline to 12 weeks with SU versus DPP4-I (MD[95% CI]=0.21%(2mmol/mol) [0.06,0.35]) but no significant difference at 52 and
M
104 weeks (MD[95% CI]=0.06%(-1mmol/mol) [-0.03,0.15] and 0.02%(-1mmol/mol) [0.13,0.18] respectively). SU was associated with weight gain and DPP4-I with weight loss at all
d
time-points. The incidence of hypoglycemia at 12, 52, and 104 weeks was significantly greater
te
with SU (20%, 24%, and 27% respectively) compared to DPP4-I (6%, 3%, and 4% respectively).
Ac ce p
The proportion of patients with a HbA1c 6.5% (48 mmol/mol) and < 7.5% (58 mmol/mol) [1-4]. However, there is no consensus on the choice of add-on therapy if metformin monotherapy fails to achieve the
an
therapeutic goal within the initial 3 months [3, 5].
Sulfonylureas (SU), are the oldest agents available that act as an insulin secretagogue [4].
M
Although they are less expensive and moderately effective in controlling blood sugars with an expected reduction in HbA1c of up to 2%, they are associated with weight gain, hypoglycemia,
te
[3, 4].
d
drug-drug interactions, and controversy exists regarding earlier drug failure with sulfonylureas
Ac ce p
Dipeptidyl peptidase-4 inhibitors (DPP4-I) are one of the newer anti-diabetic medications that were investigated as monotherapy and as add-on therapy [6]. This class includes sitagliptin, vildagliptin, linagliptin, saxagliptin, alogliptin, and dutogliptin [6, 7]. These medications prolong glucagon-like peptide-1 and glucose-dependent insulinotropic polypeptide activity via the reversible inhibition of the DPP4 enzyme [6]. Even when T2D is initially controlled, many patients will ultimately become inadequately controlled secondary to the progressive nature of the disease [3]. This raises the challenging question regarding which second agent should be added. While there are reviews [8] comparing DPP4-I to SU, none specifically examined these medications as add-on therapy to metformin. 4 Page 4 of 39
Furthermore, durability over time was not evaluated. Therefore we performed a systematic review and meta-analysis to assess the efficacy and safety of DPP4-I compared to SU as add-on
Ac ce p
te
d
M
an
us
cr
ip t
therapy to metformin in inadequately controlled T2D.
5 Page 5 of 39
2. Methods
ip t
This systematic review and meta-analysis followed the recommendations of the PRISMA statement [9]. Data sources and search:
cr
2.1
us
We searched MEDLINE (1966-2014), the Cochrane Central Register of Controlled trials (CENTRAL), EMBASE (1947-2014), and CINAHL. The terms “sitagliptin”, “vildagliptin”,
an
“linagliptin”, “saxagliptin”, “alogliptin”, and “dutogliptin” were searched and combined with the
M
terms “glimepiride”, “glipizide”, “gliclazide”, “glibenclamide”, “glyburide”, and “gliguidone” without language or date restrictions. The date of the last computer search was November 26th,
Study selection:
te
2.2
d
2014.
Ac ce p
We included studies that compared DPP4-I to SU in patients with T2D who were inadequately controlled on metformin monotherapy and reported a change in HbA1c as mean and standard deviation (SD) from baseline to a minimum of 12 weeks. Articles investigating change in HbA1c from baseline to less than 12 weeks were excluded as changes in HbA1c usually reflects glycemic control over the previous two to three months. Reviews, abstracts, letters to the editors, and retrospective studies were excluded. 2.3
Data Extraction and Quality Assessment:
Articles meeting the inclusion criteria were assessed separately by two independent reviewers (BM and DC) using the risk of bias tables suggested by the Cochrane Collaboration [10] 6 Page 6 of 39
evaluating selection bias (randomization sequence generation and allocation concealment), performance bias (blinding of participants and personnel), detection bias (blinding to outcome assessment), attrition bias (incomplete outcome data), and reporting bias (selective reporting).
ip t
For each parameter, studies were assessed as low, unclear, or high risk of bias. In addition, we
discussion between the reviewers until consensus was reached.
cr
used the seven-point modified Oxford scale [11, 12]. Any discrepancies were resolved by
us
A data collection sheet was created including: 1) author name; 2) duration of the study; 3)
an
HbA1c eligibility; 4) inclusion/exclusion criteria; 5) DPP4-I/SU drug name and dose; 6) primary outcomes of the trial; 7) baseline age, HbA1c, fasting plasma glucose (FPG), and weight; 8)
M
definition of hypoglycemia; 9) definition of adherence; 10) change in HbA1c from baseline to 12, 52, and 104 weeks; 11) change in FPG from baseline to 12, 52, and 104 weeks; 12) change in
d
body weight from baseline to 12, 52, and 104 weeks; 13) proportion of patients achieving HbA1c
te
< 7% (53 mmol/mol) at 12, 52, and 104 weeks as well as those achieving that target without any
Ac ce p
hypoglycemic episodes; 14) proportion of patients with one or more episodes of hypoglycemia at 12, 52, and 104 weeks; and 15) side effects reported. Data were extracted from graphs if not reported in the text. We also searched clinicaltrial.gov for any missing data not reported in articles. 2.4
Data Synthesis and statistical Analysis:
The primary outcome of this meta-analysis was the pooled mean change in HbA1c from baseline to 12, 52, and 104 weeks follow-up. Secondary outcomes were the changes in FPG and body weight over time, the proportion of patients achieving HbA1c < 7% (53 mmol/mol), and the proportion of patients experiencing more than one episode of any hypoglycemia. Regarding
7 Page 7 of 39
the proportion of patients experiencing hypoglycemia, we used data for any hypoglycemia episode rather than selecting data for only severe hypoglycemia. Articles investigating any DPP4-I versus any SU were pooled together. If an article had different
ip t
arms, we only included data for the DPP4-I and SU arms. If an article was investigating multiple
cr
doses of linagliptin, we only included data for linagliptin 5 mg as this dose caused inhibition of 86% of the enzymes activity and it was shown that inhibition of ≥80% is adequate for maximum
us
effect [13]. However for other agents, we combined different doses together. If an article did not
an
report the mean change in HbA1c but reported the HbA1c at different time points, we calculated the mean change in HbA1c using equations recommended by the Cochrane [10]. We converted
M
HbA1c reported in mmol/mol to percent (%) and converted FPG reported in mg/dL to mmol/l. In articles reporting data at 18 or 24 weeks but not 12 weeks, we included those results in the 12
d
weeks data. In articles including the same patients in two different follow-up periods, we
te
included the number of events from the entire study period.
Ac ce p
Continuous data were summarized as mean difference (MD) between treatment arms with 95% confidence interval (CI). If the 95% CI included a value of 0, we considered that the difference between the DPP4-I group and SU group was not statistically significant. Dichotomous data were summarized as relative risk (RR) with 95% CI. If the 95% CI included a value of 1, we considered that the difference between the DDP4-I and SU groups was not statistically significant. Analyses were performed using the Review Manager (RevMan, Version 5.3, Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2011). A random effects model was used. Number needed to harm (NNH) was calculated for any statistically significant side effect in dichotomous outcome data [14-16]. We considered heterogeneity to be present if the I² test was > 50 %. Forest plots were used to graphically represent and evaluate 8 Page 8 of 39
treatment effects. A sensitivity analysis was performed according to the risk of bias by excluding studies with at least one parameter with a high risk of bias or a modified Oxford scale of 4 or
Ac ce p
te
d
M
an
us
cr
HbA1c was calculated rather than reported by the original article.
ip t
less. Another sensitivity analysis was performed by excluding articles where the mean change in
9 Page 9 of 39
3. Results Study Characteristics:
ip t
3.1
The initial search yielded 820 studies; however, 806 were excluded (Figure 1). Two articles were
cr
excluded as they were duplicate publications [17, 18] while another study was excluded as the
us
duration of the study was
S0168-8227(15)00255-7 http://dx.doi.org/doi:10.1016/j.diabres.2015.05.025 DIAB 6401
To appear in:
Diabetes Research and Clinical Practice
Received date: Revised date: Accepted date:
29-9-2014 29-1-2015 2-5-2015
Please cite this article as: B.M. Mishriky, D.M. Cummings, R.J. Tanenberg, The Efficacy and Safety of DPP4 Inhibitors compared to Sulfonylureas as add-on Therapy to Metformin in Patients with Type 2 diabetes: a Systematic Review and Meta-analysis, Diabetes Research and Clinical Practice (2015), http://dx.doi.org/10.1016/j.diabres.2015.05.025 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
HIGHLIGHTS Title: The Efficacy and Safety of DPP4 Inhibitors compared to Sulfonylureas as add-on Therapy to Metformin in Patients with Type 2 diabetes: a Systematic Review and Meta-analysis
ip t
Meta-analysis comparing DPP-4 inhibitors vs. Sulfonylureas added to metformin
Sulfonylureas lower HbA1c more early; this difference is lost at one year follow-up
cr
Sulfonylureas are less expensive but have a greater risk of hypoglycemia
Ac ce p
te
d
M
an
us
DPP-4 inhibitors cause little hypoglycemia and weight loss instead of weight gain
1 Page 1 of 39
The Efficacy and Safety of DPP4 Inhibitors compared to Sulfonylureas as add-on Therapy to Metformin in Patients with Type 2 diabetes: a Systematic Review and Meta-analysis
ip t
Basem M. Mishriky, MD; Department of Internal Medicine, East Carolina University, Greenville, NC, United States
cr
Doyle M. Cummings, PharmD, FCP, FCCP;
us
Department of Family Medicine, East Carolina University, Greenville, NC, United States
an
Robert J. Tanenberg, MD, FACP;
M
Division of Endocrinology, East Carolina University, Greenville, NC, United States
d
Corresponding author:
te
Doyle M. Cummings, Pharm.D., Professor, Department of Family Medicine, Brody School of
Ac ce p
Medicine at East Carolina University, 101 Heart Drive, MS #654, Greenville, NC 27834; 252744-2607; [email protected]
2 Page 2 of 39
Abstract: There is no consensus on the selection of specific drug therapies when metformin fails in type 2 diabetes (T2D). This meta-analysis was performed to determine the efficacy and safety of
ip t
Dipeptidyl peptidase-4 inhibitors (DPP4-I) compared to sulfonylurea (SU) as add-on therapy to
cr
metformin in inadequately controlled T2D patients. We searched MEDLINE, CENTRAL,
EMBASE, and CINAHL for randomized trials comparing DPP4-I to SU as add-on therapy to
us
metformin and reported a change in hemoglobin A1c (HbA1c). Sixteen articles were included.
an
There was a significantly greater reduction in HbA1c from baseline to 12 weeks with SU versus DPP4-I (MD[95% CI]=0.21%(2mmol/mol) [0.06,0.35]) but no significant difference at 52 and
M
104 weeks (MD[95% CI]=0.06%(-1mmol/mol) [-0.03,0.15] and 0.02%(-1mmol/mol) [0.13,0.18] respectively). SU was associated with weight gain and DPP4-I with weight loss at all
d
time-points. The incidence of hypoglycemia at 12, 52, and 104 weeks was significantly greater
te
with SU (20%, 24%, and 27% respectively) compared to DPP4-I (6%, 3%, and 4% respectively).
Ac ce p
The proportion of patients with a HbA1c 6.5% (48 mmol/mol) and < 7.5% (58 mmol/mol) [1-4]. However, there is no consensus on the choice of add-on therapy if metformin monotherapy fails to achieve the
an
therapeutic goal within the initial 3 months [3, 5].
Sulfonylureas (SU), are the oldest agents available that act as an insulin secretagogue [4].
M
Although they are less expensive and moderately effective in controlling blood sugars with an expected reduction in HbA1c of up to 2%, they are associated with weight gain, hypoglycemia,
te
[3, 4].
d
drug-drug interactions, and controversy exists regarding earlier drug failure with sulfonylureas
Ac ce p
Dipeptidyl peptidase-4 inhibitors (DPP4-I) are one of the newer anti-diabetic medications that were investigated as monotherapy and as add-on therapy [6]. This class includes sitagliptin, vildagliptin, linagliptin, saxagliptin, alogliptin, and dutogliptin [6, 7]. These medications prolong glucagon-like peptide-1 and glucose-dependent insulinotropic polypeptide activity via the reversible inhibition of the DPP4 enzyme [6]. Even when T2D is initially controlled, many patients will ultimately become inadequately controlled secondary to the progressive nature of the disease [3]. This raises the challenging question regarding which second agent should be added. While there are reviews [8] comparing DPP4-I to SU, none specifically examined these medications as add-on therapy to metformin. 4 Page 4 of 39
Furthermore, durability over time was not evaluated. Therefore we performed a systematic review and meta-analysis to assess the efficacy and safety of DPP4-I compared to SU as add-on
Ac ce p
te
d
M
an
us
cr
ip t
therapy to metformin in inadequately controlled T2D.
5 Page 5 of 39
2. Methods
ip t
This systematic review and meta-analysis followed the recommendations of the PRISMA statement [9]. Data sources and search:
cr
2.1
us
We searched MEDLINE (1966-2014), the Cochrane Central Register of Controlled trials (CENTRAL), EMBASE (1947-2014), and CINAHL. The terms “sitagliptin”, “vildagliptin”,
an
“linagliptin”, “saxagliptin”, “alogliptin”, and “dutogliptin” were searched and combined with the
M
terms “glimepiride”, “glipizide”, “gliclazide”, “glibenclamide”, “glyburide”, and “gliguidone” without language or date restrictions. The date of the last computer search was November 26th,
Study selection:
te
2.2
d
2014.
Ac ce p
We included studies that compared DPP4-I to SU in patients with T2D who were inadequately controlled on metformin monotherapy and reported a change in HbA1c as mean and standard deviation (SD) from baseline to a minimum of 12 weeks. Articles investigating change in HbA1c from baseline to less than 12 weeks were excluded as changes in HbA1c usually reflects glycemic control over the previous two to three months. Reviews, abstracts, letters to the editors, and retrospective studies were excluded. 2.3
Data Extraction and Quality Assessment:
Articles meeting the inclusion criteria were assessed separately by two independent reviewers (BM and DC) using the risk of bias tables suggested by the Cochrane Collaboration [10] 6 Page 6 of 39
evaluating selection bias (randomization sequence generation and allocation concealment), performance bias (blinding of participants and personnel), detection bias (blinding to outcome assessment), attrition bias (incomplete outcome data), and reporting bias (selective reporting).
ip t
For each parameter, studies were assessed as low, unclear, or high risk of bias. In addition, we
discussion between the reviewers until consensus was reached.
cr
used the seven-point modified Oxford scale [11, 12]. Any discrepancies were resolved by
us
A data collection sheet was created including: 1) author name; 2) duration of the study; 3)
an
HbA1c eligibility; 4) inclusion/exclusion criteria; 5) DPP4-I/SU drug name and dose; 6) primary outcomes of the trial; 7) baseline age, HbA1c, fasting plasma glucose (FPG), and weight; 8)
M
definition of hypoglycemia; 9) definition of adherence; 10) change in HbA1c from baseline to 12, 52, and 104 weeks; 11) change in FPG from baseline to 12, 52, and 104 weeks; 12) change in
d
body weight from baseline to 12, 52, and 104 weeks; 13) proportion of patients achieving HbA1c
te
< 7% (53 mmol/mol) at 12, 52, and 104 weeks as well as those achieving that target without any
Ac ce p
hypoglycemic episodes; 14) proportion of patients with one or more episodes of hypoglycemia at 12, 52, and 104 weeks; and 15) side effects reported. Data were extracted from graphs if not reported in the text. We also searched clinicaltrial.gov for any missing data not reported in articles. 2.4
Data Synthesis and statistical Analysis:
The primary outcome of this meta-analysis was the pooled mean change in HbA1c from baseline to 12, 52, and 104 weeks follow-up. Secondary outcomes were the changes in FPG and body weight over time, the proportion of patients achieving HbA1c < 7% (53 mmol/mol), and the proportion of patients experiencing more than one episode of any hypoglycemia. Regarding
7 Page 7 of 39
the proportion of patients experiencing hypoglycemia, we used data for any hypoglycemia episode rather than selecting data for only severe hypoglycemia. Articles investigating any DPP4-I versus any SU were pooled together. If an article had different
ip t
arms, we only included data for the DPP4-I and SU arms. If an article was investigating multiple
cr
doses of linagliptin, we only included data for linagliptin 5 mg as this dose caused inhibition of 86% of the enzymes activity and it was shown that inhibition of ≥80% is adequate for maximum
us
effect [13]. However for other agents, we combined different doses together. If an article did not
an
report the mean change in HbA1c but reported the HbA1c at different time points, we calculated the mean change in HbA1c using equations recommended by the Cochrane [10]. We converted
M
HbA1c reported in mmol/mol to percent (%) and converted FPG reported in mg/dL to mmol/l. In articles reporting data at 18 or 24 weeks but not 12 weeks, we included those results in the 12
d
weeks data. In articles including the same patients in two different follow-up periods, we
te
included the number of events from the entire study period.
Ac ce p
Continuous data were summarized as mean difference (MD) between treatment arms with 95% confidence interval (CI). If the 95% CI included a value of 0, we considered that the difference between the DPP4-I group and SU group was not statistically significant. Dichotomous data were summarized as relative risk (RR) with 95% CI. If the 95% CI included a value of 1, we considered that the difference between the DDP4-I and SU groups was not statistically significant. Analyses were performed using the Review Manager (RevMan, Version 5.3, Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2011). A random effects model was used. Number needed to harm (NNH) was calculated for any statistically significant side effect in dichotomous outcome data [14-16]. We considered heterogeneity to be present if the I² test was > 50 %. Forest plots were used to graphically represent and evaluate 8 Page 8 of 39
treatment effects. A sensitivity analysis was performed according to the risk of bias by excluding studies with at least one parameter with a high risk of bias or a modified Oxford scale of 4 or
Ac ce p
te
d
M
an
us
cr
HbA1c was calculated rather than reported by the original article.
ip t
less. Another sensitivity analysis was performed by excluding articles where the mean change in
9 Page 9 of 39
3. Results Study Characteristics:
ip t
3.1
The initial search yielded 820 studies; however, 806 were excluded (Figure 1). Two articles were
cr
excluded as they were duplicate publications [17, 18] while another study was excluded as the
us
duration of the study was
