Cost-Effectiveness Analysis of Empagliflozin in Comparison To

Cost-effectiveness Analysis of Empagliflozin in Comparison to Sitagliptin and Saxagliptin Based on Cardiovascular Outcomes Studies Mafalda Ramos 1, Volker Foos1, Anastasia Ustyugova², Nikco Hau3, Pranav Gandhi4, Mark Lamotte1 1IQVIA Real World Evidence Solutions, Zaventem, Belgium; ²Boehringer Ingelheim International GmbH, Ingelheim, Germany; 3Boehringer Ingelheim Ltd, Bracknell, UK; 4Boehringer Ingelheim Pharmaceuticals Inc., Ridgefield, CT, USA. 12 November 2018

Disclosure

• IQVIA, the employer of Mark Lamotte, Volker Foos and Mafalda Ramos, received consulting fees for adapting the IQVIA CDM for this project • Anastasia Ustyugova, Nikco Hau and Pranav Gandhi are/were full time employees of Boehringer Ingelheim

1 Several cardiovascular outcomes studies were conducted over the last years

• The EMPA REG OUTCOME® trial examined the effects of empagliflozin versus placebo, on top of standard of care, on CV morbidity and mortality in patients with T2D at high risk of CV events • Empagliflozin significantly reduced CV events and mortality compared with placebo

Significantly reduced CV mortality Significantly reduced Slowed progression of kidney hospitalisation for HF disease

CV death ↓ Hospitalisation for HF Incident or worsening 38% ↓ 35% nephropathy ↓ 39%

• Two other CVO trials conducted with DDP-4 (TECOS-CVO (sitagliptin) and SAVOR-TIMI CVO (saxagliptin)) did not show differences in outcomes with placebo

Zinman B et al. N Engl J Med 2015;373:2117; Zinman B. EASD 2015. Oral presentation 3

Objective

• To assess the long-term cost-effectiveness of empagliflozin added to standard of care (SoC) in adult T2D patients with established cardiovascular disease compared to sitagliptin + SoC and saxagliptin + SoC using the IQVIA Core Diabetes model

2 IQVIA Core Diabetes Model (CDM)

The CDM is the most used (90+ medline references) and validated diabetes1 model available online CDM inputs take surrogate endpoints and translate them into long-term health and economic outcomes that drives value for money in a diabetes INPUTS Economics Clinical

• HbA1c Management Surrogate • Blood pressure Treatment Endpoints • Weight • Lipids Cohort

OUTCOMES • Life expectancy, QALY • Macrovascular events Long-term • Microvascular events Endpoints • Costs (direct and indirect, combined) 1McEwan P et al. Validation of the IMS CORE Diabetes Model. Value Health. 2014 Sep;17(6):714-24

CDM version 9.0 has 9 risk equations available to predict CV risk

Prediction models End Point Risk engines for several complications: 1 UKPDS 689 Ischaemic heart disease (IHD) Myocardial infarction (MI), Heart failure (HF) and Stroke New set of equations based on longer follow up including more risk 2 UKPDS 82 factors

3 Framingham CHD, stroke, fatal events.

CVD=sudden cardiac death, fatal IHD, non fatal MI, unstable angina, 4 Swedish NDR1 percutaneous coronary intervention, CABG or fatal/non-fatal stroke

5 ADVANCE risk engine2 CVD = CHD or stroke

6 PROcam3 CHD event

7 ARIC4 CHD event

8 FREMENTLE5 CVD event = MI or stroke

RE 1 for stroke 9 HONG KONG6-8 RE 2 for HF.

1Kiadaliri et al. PLoS One. 2013 May 9;8(5); 2Kengne AP. Cardiovasc J Afr. 2013 Oct-Nov;24(9-10):376-81; 3Chien et al, Int J Cardiol. 2012 May 31;157(2):263-8; 4Folsom A et al. Diabetes Care 26:2777–2784, 2003; 5Davis WA et al. Intern Med J 2010;40:286e92; 6Yang X et al. Am J Cardiol. 2008 Mar 1;101(5):596-601; 7Yang X et al. Diabetes Care 2007 30:65–70; 8Yang X et al. Cardiovascular Diabetology 2008, 7:9; Stevens RJ et al. Group Clinical Science 2001;101:671-679 6

3 Since the available risk equations could not predict the outcomes of EMPA-REG OUTCOME the CDM had to be calibrated Relative Risk (RR) scores were derived via an iterative calibration approach and applied to each of the eight selected outcomes

Step 1 Step 2 Step 3 Step 4 Step 5 Step 6

Most suitable set of Initial RRs for After initial RR Based on outcomes CDM was re-run with Steps 4 and 5 were risk equations (RE) calibration were settings for all eight in step 3, RRs were the new RRs repeated until results was evaluated to calculated using the end points were updated using a (i.e. cumulative match the EMPA- ratio of observed assessed in Step 2, linear trend line incidence for each REG OUTCOME (EMPA-REG the CDM CVO was approach to predict outcome) at year 3 outcomes. OUTCOME) and run for a time the next best choice predicted by CDM Exploratory analysis predicted (CDM) horizon of for each CVO equaled results was run for 3 years cumulative incidence 3 years RR adjustment from the EMPA-REG for multiple (CI) outcomes OUTCOME study combinations of CV at year 3 RE and three approaches to Eight selected outcomes: predict mortality: • Primary and secondary MI UKPDS 82 was • Primary and secondary stroke selected. • Hospitalization for Unstable Angina • Hospitalization for Heart failure • CV death • Non CV death • Microalbuminuria • End stage renal disease

As a last step the relative risks obtained in the indirect comparison with DPP4 inhibitors were added

Hospitalization due to Non-Fatal Myocardial Empagliflozin vs. CV related mortality All-cause mortality Non-Fatal stroke Heart Failure Infarction 0.60 0.61 0.51 1.12 0.92 Saxagliptin (0.46, 0.80) (0.49, 0.76) (0.37, 0.70) (0.76, 1.63) (0.69, 1.20) 0.60 0.67 0.65 Sitagliptin -- -- (0.46, 0.79) (0.54, 0.83) (0.47, 0.90) *Balijepalli C, Shirali R, Kandaswamy P, Ustyugova A, Pfarr E, Lund SS, Druyts E. Cardiovascular Safety of Empagliflozin Versus Dipeptidyl Peptidase-4 (DPP-4) Inhibitors in Type 2 Diabetes: Systematic Literature Review and Indirect Comparisons. Diabetes Ther. 2018. https://doi.org/10.1007/s13300-018-0456-7

The 3-year EMPA-REG OUTCOME event rates (or calculated with indirect comparison) could be nicely replicated by using the calibrated CDM (internal validity) Empagliflozin Sitagliptin Saxagliptin From Trial* Calibrated RR** IDC*** Calibrated RR IDC Calibrated RR Death from any cause 5.82 5.78 0.42 8.69 8.67 0.51 9.70 9.78 0.72 Death from CV causes 3.72 3.68 0.51 6.20 6.13 0.80 6.10 6.14 0.73 Myocardial Infarction 5.04 5.05 0.75 5.04 5.00 0.74 5.46 5.44 0.81 Unstable Angina 3.00 3.01 1.63 3.00 3.01 1.63 3.00 3.04 1.61 Stroke 3.69 3.70 1.18 3.69 3.69 1.15 3.33 3.33 1.03 Heart Failure 2.82 2.83 2.11 4.34 4.36 3.13 5.53 5.57 3.99 Microalbuminuria 75.75 75.86 35.50 79.80 79.10 31.50 79.80 79.59 36.50 End Stage Renal Disease 0.30 0.30 0.81 0.63 0.54 1.09 0.63 0.63 1.52 *3-year event rates calculated based on the event rates per 1,000 patients ** RR = relative risk applied as correction factor in the CDM 8 ***IDC = indirect comparison

4 Base case settings used to run the cost-effectiveness analyses

Settings Input Comment

Perspective Payer perspective UK National Health Service (NHS) perspective

Discounting rate 3.5% annually for both cost and outcomes As recommended by NICE, 2014 A time horizon of 50 years was used in the base case analysis in order to Time horizon 50 years capture relevant long-term complications, and their associated costs and impact on quality-adjusted life expectancy (lifetime perspective). Population definition is considered same as described in the EMPA-REG Target population T2D patients at established cardiovascular disease OUTCOME study Empagliflozin compared with SoC, sitagliptin and After treatment switch, all patients receive basal bolus as next line of Comparators saxagliptin therapy Quality adjusted life years (QALYs) and life years All values for health-related quality of life (HRQoL) were obtained from Outcomes (LYs) published literature Including pharmacy costs, management cost (screening test, concomitant medication) and the costs of T2D complications (CVD complications, renal complications, acute events, eye disease, Cost data Direct medical costs neuropathy, foot ulcer and amputation). Unit costs were considered from published literature and UK national sources Treatment switch 8.5% HbA1c threshold HbA1c threshold of 8.5% for all the comparators

NICE: The National Institute for Health and Care Excellence; QALYs: Quality-adjusted life years; LYs: Life-years 9

Treatment effect on physiological parameters within the CDM

• The treatment benefit on physiological parameters was applied in the CDM in Table: First year treatment effects of first and second line treatments following ways: [1] *** *** [4] • 1st year benefit was programmed in Empagliflozin Sitagliptin Saxagliptin Basal Bolus HbA1c (%)* -0.58 -0.328[2] -0.4[3] -0.828 the treatment settings together with SBP (mmHg)* -3.9 -0.62 -0.62 0 the associated adverse event rates of DBP (mmHg)* -1.72 -0.78 -0.78 0 each treatment T-Chol (mg/dl)* 7.81 3.56 3.56 0 nd rd • For 2 and 3 years, the treatment HDL (mg/dl)* 1.81 -0.09 -0.09 0 effect followed the progression over LDL (mg/dl) * 4.79 1.42 1.42 0 time available in respective trials of TRIG (mg/dl)* 0 0 0 0 empagliflozin, sitagliptin and BMI (kg/m²)* -0.64 -0.04 -0.04 0.32 saxagliptin eGFR (ml/min/1.73m2)* -0.16 0.18 0.18 0 [3] • From the 4th year onwards, HbA1c NSHE rate** 13.62 13.98 34.91 2,566.83 SHE1 rate** 0.44 0.64 3.97[3] 23.81 progression from empagliflozin was SHE2 rate** 0.06 0.14 0.53[3] 3.19 applied for all comparators

HbA1c- Glycated haemoglobin; eGRF-Estimated glomerular filtration rate; SBP-Systolic Blood Pressure; DBP-Diastolic Blood Pressure; BMI-Body mass index; LDL-Low-density lipoprotein cholesterol; HDL-High-density lipoprotein cholesterol; TC-Total cholesterol; HR-Heart rate;NSHE-non severe hypoglycemic event; SHE1- severe hypoglyccemic event not needing medical attention; SHE2- severe hypoglyccemic event needing medical attention *Effect on the surrogate endpoints is applied on the first year of treatment; **Rate per 100 patient-year; *** Aside HbA1c effect, all other endpoints are assumed to be equal to SoC described in the EMPA-REG OUTCOME trial.

Sources: [1]Zinman et al 2016, Wanner et al 2016; [2]Green et al 2015 ; [3]Scirica et al 2013; [4]Riddle et al 2014 10

5 Progression of HbA1c over time in the different treatment arms based on initial effects and after treatment switch that occurs after 8 years

A1c progression

9.5

8.5

8 Empagliflozin

7.5 Sitagliptin 7

Saxagliptin 6.5

6 0 10 20 30 40 50 TIme horizon

Costs considered in the current analysis

Direct costs are considered in the current analysis

Treatment associated cost Management and complication cost • Drug cost (£), • A pragmatic research was done on the current • Needle cost (where applicable) and the costs care patterns and direct medical expenditure associated with self-monitoring of blood associated with T2D in the UK glucose (SMBG). • Costs of diabetes-related complications defined • Source: British National Formulary (BNF) as cost with either acute events (event costs) or chronic conditions of an ongoing disease (health states). Concomitant • All costs were updated to 2018. Drug Annual cost Drug Insulin cost cost 1st year 2nd year 1st year 2nd year Empagliflozin 477.30 339.35 356.67 816.65 833.97 Sitagliptin 433.86 165.98 234.48 599.84 668.35 Saxagliptin 412.21 298.76 309.35 710.97 721.56 12

6 Results: Model outcomes: improved cardiovascular outcomes with empagliflozin translate to long-term clinical benefits at an acceptable cost

CVO effects are applied up to 8 years (treatment switch)

Cost-utility measures Empa + SoC Sita + SoC Saxa + SoC

Life expectancy, years 10.3 9.5 9.2 QALY 5.7 5.4 5.2 Total Costs, £ 50,801 47,627 48,071 Incremental cost-effectiveness ratio Empa vs Sita Empa vs Saxa ICUR (QALY) 9,253 5,151 ICER (LY) 4,140 2,594

Looking at 5 year event rates the clinical benefits of empagliflozin are clearly shown

CVD incidence Mortality Empagliflozin 12.0 14.0 Sitagliptin Saxagliptin 10.0 12.0

10.0 8.0

8.0

6.0 11.1

6.0

Cumulative Cumulative Incidence (%)

Cumulative Cumulative Incidence (%) 12.2

4.0 8.7

7.2

7.1 7.1

4.0 6.8

6.4

6.1

5.8

6.7 6.4

2.0

4.5 3.7

2.0 3.7

3.5

3.4

3.8 3.0

0.0 0.0 CV-related Non-CV related PVD HF Angina Stroke MI

7 Probabilistic Sensitivity Analyses show the robustness of the projected economic benefit Versus sita Versus saxa

Conclusion

– The calibrated IQVIA Core Diabetes Model was able to replicate accurately the results of the EMPA- REG OUTCOME trial and its comparators in patients at high risk for cardiovascular disease – The short term results (time horizon of 5 years) demonstrated that empagliflozin reduces the risk of both CV death and non-CV death, as well as the risk of hospitalization for heart failure and end stage renal disease compared to DPP4 inhibitors. – Looking long term (lifelong time horizon) empagliflozin demonstrated added clinical value, by providing longer life expectancy and quality adjusted life years compared to DPP4 inhibitors with less cases of hospitalization for heart failure and renal failure. – This makes empagliflozin “good value for money” compared to DPP4 inhibitors both at short and long term using the NICE willingness to pay threshold of 20,000GBP/QALY

8 Any question?

Baseline characteristics of the patients in the EMPA-REG OUTCOME trial (n=7020)

Characteristics Mean SE/SD Units/Range Characteristics Mean SE/SD Units/Range Baseline Risk Factors Patient Demographics HbA1c 8.07 0.85 %-points Start age 63.10 8.60 years SBP 135.47 17.00 mmHg

Duration of Diabetes 9 0.00 years DBP 76.67 9.83 mmHg T-CHOL 162.90 43.80 mg/dL Prop. Male 0.71 [0-1] HDL 44.40 11.70 mg/dL Racial Characteristics LDL 85.60 35.70 mg/dL TRIG 170.50 126.90 mg/dL Prop. White 0.543 [0-1] BMI 30.62 5.26 kg/m2 Prop. Black 0.051 [0-1] eGFR 74.04 21.41 ml/min/1.73m2 HAEM 14.5 0 gr/dl Prop. Hispanic 0.180 [0-1] WBC 6.8 0 106/ml Prop. Native American 0.008 [0-1] Heart rate* 68.47/68.08 0 bpm WHR 0.93 0 (1 unit) Prop. Asian/Pacific Islander 0.218 [0-1] uAER 19.3 2.9 mg/mmol Prop. Australian Ser_Creat 1.1 0 mg/dl 0 [0-1] (south Europ.) Ser_alb 3.9 0 g/dl Prop. smoker 0.132 [0-1] Prop. Australian (Aboriginal) 0 [0-1] Cigarettes/day 3 *First value was applied in the clinical settings of the empagliflozin arm and second on the standard of care arm; Source: BI CTR, published literature, WHO 2002 and WHO 2004 Alcohol consumption 3 Oz/week

9 Complication costs (1/2)

Complication costs: The cost inputs are based on the desk research findings. Selection is based on the studied population consisting of type 2 Diabetes patients, most recent data and on the robustness of the data (quality assessment of the paper). All costs were updated to 2018 using the Hospital & Community Health Service (HCHS) Index from Personal Social Services Research Unit (PSSRU,2017)

Table: Costs inputs used in the CDM (£/GBP) (continued on next slide)

Variable Value Variable Value Variable Value Discounting Direct cost of complications Direct cost of complications Costs discount rate 3.5% Stroke (1st year) 7,375.23 MI (1st year) 6,913.54 Clinical outcomes discount rate 3.5% Stroke (2nd year) 1,219.27 MI (2nd year) 1,250.70 Management costs Stroke death within 30 days 4,285.33 Angina (1st year) 3,330.45 Annual statins treatment 11.36 PVD (1st year) 4,697.66 Angina (2nd year) 1,922.75 Annual aspirin treatment 9.25 PVD (2nd year) 1,009.97 CHF (1st year) 3,458.40 Annual ACE-I treatment 12.94 Haemodialysis (1st year) 43,359.47 CHF (2nd year) 1,596.44 Annual microalbuminuria screening 11.36 Haemodialysis (2nd year) 43,359.47 Annual gross proteinuria screening 11.31 Peritoneal Dialysis (1st year) 32,555.50 Stopping ACE-I due to side effects - Peritoneal Dialysis (2nd year) 32,555.50 Annual eye screening 34.94 Foot screening programme - Non-standard ulcer treatment -

Sources are detailed in appendix

Complication costs (2/2)

Table: Costs inputs used in the CDM (£/GBP)

Variable Value Variable Value Direct cost of complications Direct cost of complications Renal transplant (1st year) 27,222.46 Stroke (1st year) 7,375.23 Renal transplant (2nd year) 1,232.22 Stroke (2nd year) 1,219.27 Severe hypoglycaemia Type 2 (SHE2) 1,470.20 Stroke death within 30 days 4,285.33 Severe hypoglycaemia type 1 (NSHE1) 432.56 PVD (1st year) 4,697.66 Non-severe hypoglycaemia (NSHE) 4.13 PVD (2nd year) 1,009.97 Retinopathy laser treatment 1,175.52 Haemodialysis (1st year) 43,359.47 Cataract operation (1st year) 2,635.73 Haemodialysis (2nd year) 43,359.47 Cataract operation (2nd year) 178.21 Peritoneal Dialysis (1st year) 32,555.50 Gangrene treatment 3,693.81 Peritoneal Dialysis (2nd year) 32,555.50 After healed ulcer (annual) 279.54 Infected ulcer 2,117.84 Standard uninfected ulcer 857.38 Healed ulcer history of amputation (annual) 279.54

*Represents studies which did not differentiate costs of outcomes for type 2 diabetes (reported costs for diabetes OR generalised costs for type 1 and type 2 diabetes)

10 Quality of life data (explorative analysis)

QUALITY OF LIFE UTILITIES Beaudet et al 2014 U T2 no complications 0.7850 DisU MI event -0.0550 U post MI 0.7300 U angina 0.6950 U CHF 0.6770 DisU stroke event -0.1640 U post Stroke 0.6210 U PVD 0.7240 U MA 0.7850 U GRP 0.7370 U HD 0.6210 U PD 0.5810 U RT 0.7620 U BDR 0.7450 U BDR wrongly treated 0.7450 U PDR laser treated 0.7150 U PDR no Laser 0.7150 U ME 0.7450 U SVL 0.7110 U cataract 0.7690 U neuropathy 0.7010 U healed ulcer 0.7850 U active ulcer 0.6150 U post amputation 0.5050 Diminishing NSHE disutility yes DisU for SHE 1 (during daytime) -0.0205 DisU for SHE 1 (nocturnal) -0.0205 DisU for SHE 2 (during daytime) -0.0743 DisU for SHE 2 (nocturnal) -0.0743

Applying the CVO effects only for 3 years or decreasing the time horizon to 5 years doesn’t change the conclusions CVO effects are applied for 3 years Cost-utility measures Empa + SoC Sita + SoC Saxa + SoC Life expectancy, years 9.6 9.3 9.1 QALY 5.4 5.3 5.2 Total Costs, £ 44,592 44,592 48,846 Incremental cost-effectiveness ratio Empa vs Sita Empa vs Saxa ICUR (QALY) 8,331 5,462 ICER (LY) 3,057 2,749 Time horizon of 5 years Cost-utility measures Empa + SoC Sita + SoC Saxa + SoC Life expectancy, years 4.25 4.10 4.19 QALY 2.55 2.48 2.44 Total Costs, £ 13,973 13,148 13,717 Incremental cost-effectiveness ratio Empa vs Sita Empa vs Saxa

ICUR (QALY) 13,156 2,438 22

ICER (LY) 5,820 1,243