Results

Poster presented at the 21st Annual International Meeting of the International Society For Pharmacoeconomics and Outcomes Research (ISPOR), May 21–25, 2016, Washington, DC, USA

Conclusions, Other

Considerations

TREATMENT AREA: Hyperglycemia, type-2 diabetes

BACKGROUND AND PURPOSE: Oral agents in the sodium-glucose co-tranporter-2 inhibitor class are an alternative to DPP-4 inhibitors like sitagliptin (SITA) as add-on therapy when first-line metformin no longer brings patients to goal.1, 2 Few RCTs provide direct comparisons of SGLT2s vs. SITA in this context. How do SGLT2s compare to SITA in controlling glycated hemoglobin (HbA1c), body weight and blood pressure, when used as an add-on therapy to metformin?

APPROACH: Systematic review and meta-analysis of randomized controlled trials (RCTS) in T2D patients to 9-30-2014, aggregating across results for canagliflozin (CANA), dapagliflozin (DAPA) and empagliflozin (EMPA) as add-on therapy to metformin (MET). All included studies compared outcomes for add-on therapy to outcomes for a MET+ placebo control. We provide adjusted indirect comparisons of outcomes for SGLT2s + MET to SITA + MET.

Objectives

Methods SGLT2 + MET trials

Trial n Drug/ Dose Age % Male % HbA1c BMI

Bailey, 2010 137 DAPA, 5 mg 54.3 50 8.2 31.4

135 DAPA, 10 mg 52.7 57 7.9 31.2

137 MET (Control) 53.7 56 8.1 31.8

Häring, 2014 217 EMPA, 10 mg 55.5 58 7.9 29.1

214 EMPA, 25 mg 55.6 56 7.9 29.7

207 MET (Control) 56.0 56 7.9 28.7

L-G 2013 368 CANA, 100 mg 55.5 47 7.9 32.4

367 CANA, 300 mg 55.3 45 7.9 31.4

183 MET (Control) 55.3 51 8.0 31.1

Rosenstock, 64 CANA, 100 mg 51.7 56 7.8 31.7

2012 64 CANA, 300 mg 52.3 56 7.7 31.6

65 MET (Control) 53.3 48 7.8 30.6

Rosenstock, 71 EMPA, 10 mg 59.0 47 7.9 31.4

2013 70 EMPA, 25 mg 59.0 53 8.1 31.5

71 MET (Control) 60.0 47 8.0 31.3

SITA + MET trials

Trial n Drug/ Dose Age % Male % HbA1c BMI

Charbonnel, 464 SITA, 100 mg 54.4 56 8.0 --

2006 237 MET (Control) 54.7 60 8.0 --

L-G 2013 366 SITA, 100 mg 55.5 47 7.9 32.0

183 MET (Control) 55.3 51 8.0 31.1

Nauck, 2014 315 SITA, 100 mg 54.0 48 8.1 31.0

177 MET (Control) 55.0 51 8.1 31.0

Raz,2008 96 SITA100 mg 53.6 51 9.3 30.1

94 MET (Control) 56.1 42 9.1 30.4

Rosenstock, 65 SITA, 100 mg 51.7 44 7.6 31.6

2012 65 MET (Control) 53.3 58 7.8 30.6

Scott, 2008 94 SITA, 100 mg 55.2 48 7.8 30.3

92 MET (Control) 55.3 59 7.7 30.0

Yang, 2012 197 SITA, 100 mg 54.1 47 8.5 25.3

198 MET (Control) 55.1 55 8.5 25.3

SYSTEMATIC REVIEW: We searched the Cochrane Central Register and MEDLINE via PubMed, using a sensitive search algorithm for RCTs.3, 4 Figure 1 depicts the search and selection process for SGLT2s. The search for SITA studies yielded 964 results. A similar process retained 7 trials that met inclusion criteria, out of 520 de-duplicated records.

CRITERIA: Included trials were phase 2b or 3 double-blind parallel-group RCTs ≥ 12 weeks long, testing CANA, DAPA, EMPA or SITA in dual therapy with MET, in T2D patients no longer well-controlled on MET monotherapy. All studies had a MET + placebo (PBO) control. • Patients were 18-80 years old, on appropriate

therapeutic levels of MET monotherapy for at least 8 weeks prior to baseline, with a baseline HbA1c

between 7 and 10.5 %. • Trials in special populations were not included • Pre-specified outcomes were extracted by

treatment condition, along with details of the trial and sample.

• Extension-trial data were not included, since this would violate assumptions of independence.5

META-ANALYSIS: Main steps were to: • Pool trial means and standard deviations for three

interventions: FDA-approved ‘low’ and ‘high’-dose SGLT2 and SITA 100 mg, plus the active (MET + PBO) control.

• Obtain (inverse-variance weighted) pooled mean change in outcome from baseline for patients receiving SGLT2 or SITA add-on therapy vs. controls, using random-effects models.6, 7

• Use the pooled, adjusted outcomes to gauge whether SGLT2 + MET therapy yielded similar or better HbA1c, body-weight and blood-pressure (BP) control compared to SITA + MET, in indirect comparisons.8 The same control data were used to estimate effects of SGLT2LOW and SGLT2HIGH.

HbA1c: Effect of add-on treatment

adjusted for MET-only control:

SGLT2 Trial Identification and

Selection

FDA-Approved SGLT2s

Included trials were randomized, double-blind con-trolled studies, at low risk of bias. Using the Cochrane Collaboration risk of bias tool however, all studies would be considered at high risk due to manufacturer sponsorship and use of a “last observation carried forward” method to replace missing data.19-21

Health is all we do.

Figure 2. Control-adjusted HbA1c means for patients randomized to receive “low”- or “high”- dose SGLT2 or a 100-mg QD dose of SITA added to MET relative to PBO+ MET

Bias Risk

SGLT2 Inhibitors vs. Sitagliptin as Add-on to Metformin

in Type-2 Diabetes: A Systematic Review and Meta-analysis

Christine A. Sevald, Joseph P. Jackson and John F. McAna

-0.57

-0.67 -0.65

-0.75

-0.55

-0.35

-0.15

Low-dose SGLT2

High-dose SGLT2

SITA 100 mg

% H

bA

1c

Indirect Comparison SGLT2+ MET vs. SITA + MET: Pooled HbA1c

SGLT2 + MET Trials

Author, year Intervention Total n Length (weeks)

Bailey, 20109 DAPA 5, 10 mg 409 24

Häring, 201410 EMPA 10, 25 mg 638 24

L-G, 201311 CANA 100, 300 mg 918 26

Rosenstock, 201212 CANA 100, 300 mg 193 12

Rosenstock, 201313 EMPA 10, 25 mg 212 12

Total n, SGLT2 + MET v MET = 2370

SITA + MET Trials

Author, year Intervention Total n Length (weeks)

Charbonnel, 200614 SITA 100 mg 701 24

L-G 2013 SITA 100 mg 549 26

Nauck, 201415 SITA 100 mg 492 26

Raz, 2008 16 SITA 100 mg 190 18

Rosenstock, 2012 SITA 100 mg 130 12

Scott, 200817 SITA 100 mg 186 18

Yang, 201218 SITA 100 mg 395 24

Total n, SITA + MET v MET = 2643

-1.77

-2.09

0.17

-3.00

-2.00

-1.00

0.00

Low-dose SGLT2

High-dose SGLT2

SITA 100-mg

Weig

ht

Lo

ss (

kg

)

Indirect Comparison SGLT2 + MET vs. SITA + MET: Pooled Weight Loss

SITA 100 mg

ACKNOWLEDGEMENTS This research received no commercial support. The first author thanks the Jefferson College of Population Health at Thomas Jefferson University for funding to attend this meeting. Contact Chris Sevald @ Christine.Sevald@alumni.Jefferson.edu.

REFERENCES 1. Inzucchi et al. Diabetologia, 2012.; 2. Inzucchi, et al. Diabetes Care, 2015; 3. Haynes et al. BMJ, 2005; 4. Higgins & Green Cochrane Handbook, 2011; 5. Hedges & Vevea Psych. Methods, 1998; 6. DerSimonian & Laird Contr Clin Trials,1986; 7. StatsDirect v3: Meta-analysis of Effects; 8. Bucher et al. J Clin Epidemiol. 1997; 9. Bailey et al. Lancet, 2010; 10. Häring et al. Diabetes Care, 2014; 11. Lavalle-González et al. J Diabetol, 2013; 12. Rosenstock et al. Diabetes Care, 2012; 13. Rosenstock et al. Diab Obes Metab, 2013; 14. Charbonnel et al. Diabetes Care, 2006; 15. Nauck et al. Diabetes Care, 2014; 16. Raz et al. Curr Med Res & Opinion, 2008; 17. Scott et al. Diab Obes Metab, 2008; 18. Yang et al. J Diabetes, 2012; 19. Higgins et al. BMJ, 2011; 20. Vasilakou et al. Ann Intern Med, 2013; 21. AHRQ, 2012 #12-EHC047-EF; 22. Mozaffarian et al. Circulation, 2016; 23. UKPDS 34 Lancet, 1998; 24. UKPDS 35 BMJ, 2000; 25. UKPDS 75 Diabetologia, 2006; 26. Seshasai et al. NEJM, 2011; 27. Zinman et al. NEJM, 2015; 28 Abdul-Ghani et al. Diabetes Care, 2016; 29. Neal et al. Am. Heart J, 2013; 30.DECLARE -TIMI58, NCT01730534, clinicaltrials.gov

Figure 3. Pooled HbA1c, n = 1502, 1489, and 2622

Figure 1. Trials retained at each step in search and selection

Figure 4. Pooled Weight Loss (kg), n = 1514, 1502 and 1339

Sample characteristics

Included Trials

Generic Brand Approved: 'Low Dose'

'High Dose‘

Canagliflozin Invokana Mar, 2013 100 mg 300 mg

Dapagliflozin Farxiga Jan,2014 5 mg 10 mg

Empagliflozin Jardiance Aug, 2014 10 mg 25 mg

Figures 3 and 4 show adjusted indirect comparisons of outcomes. Body weight data were available in all SGLT2 and four out of seven SITA + MET trials. SBP data were available in all SGLT2 and three out of seven SITA trials. Only two SITA trials reported DBP. BP data for SITA + MET trials were too sparse for bias assessment or to allow indirect comparison.

• There were no statistical differences in impact on HbA1c between ‘low’ or ‘high’-dose SGLT2 and SITA 100 mg (Z = 1.66 p = 0.10; Z = .43, p = .67).

• There was a 1.7 to 2.1 kg weight loss among patients in the SGLT2 conditions (12-26 weeks), compared to a <.2 kg weight gain with SITA (Z = 9.46 p < .00001; Z = 9.48, p < .00001).

Results (cont’d)

SGLT2, high dose

SGLT2, low dose

SITA, 100 mg

• Indirect comparisons showed parity between the SGLT2s and SITA in blood-glucose control. There was a nearly 3% difference in impact on body weight in favor of the SGLT2s – a small but potentially clinically meaningful difference. Excess blood glucose is associated with increased cardiovascular risk, typically doubled in T2D.22-26 Overweight is associated with a diagnosis of T2D.

• A recent two-year long RCT in 7020 T2D patients with prior CV disease found EMPA monotherapy was associated with a 38% lower risk of CV death,32% lower all-cause mortality and 35% fewer heart-failure hospitalizations compared to placebo.27 Lower CV risk is thought to be linked to more than reduced weight and BP however, because of its rapid onset during the trial.28

• Trials to assess long-term CV outcomes with other SGLT2s are currently underway. The results of these studies, expected in 2Q 2017 and 2019, will help clarify whether CV protection is a class effect.29, 30

Duplicates removed

(n=156)

Records excluded:

1. Not primary research

(review, commentary,

opinion) n=100

2. Wrong study population

(healthy/ non-T2D subjects;

wrong indication) n=46

3. Wrong study design in

T2D subjects: (PK/PD or

proof-of-concept) n=34

Studies included in

quantitative

analysis

(n=5)

CANA, DAPA, or

EMPA Records

identified in

database search

(n=407)

Records after

duplicates removed

(n=251)

Records screened,

title and abstract or

full-text articles

assessed for

eligibility (n=251)

Full-text articles

assessed for

eligibility (Phase 2

or 3 studies in T2D

subjects) (n=71)

4). Monotherapy or comb.

therapy without MET n=47

5) Therapy with MET,

excluded for other pre-

specified reason: (triple

combination, initial dual

combination, FDC; lacked

MET- plus-placebo control

OR extension of included

trial n=19