Single Technology Appraisal

Dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648]

Committee Papers

NATIONAL INSTITUTE FOR HEALTH AND CARE EXCELLENCE

SINGLE TECHNOLOGY APPRAISAL

Dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648]

Contents:

The following documents are made available to stakeholders:

Access the final scope and final stakeholder list on the NICE website.

1. Company submission from AstraZeneca UK Ltd

2. Clarification questions and company responses

3. Patient group, professional group and NHS organisation submissions from:

• a. UK Clinical Pharmacy Association – Heart Failure Committee

4. Expert personal perspectives from:

• a. Sarah Worsnop – patient expert, nominated by Pumping Marvellous Foundation

• b. Nick Hartshorne-Evans – patient expert, nominated by Pumping Marvellous Foundation

• c. Lisa Anderson – clinical expert, nominated by British Society for Heart Failure

• d. Andrew Ludman – clinical expert, nominated by British Cardiovascular Society

5. External Assessment Report prepared by BMJ-TAG

6. External Assessment Report – factual accuracy check

7. External Assessment Report addendum prepared by BMJ-TAG

8. External Assessment Group summary of direct and indirect treatment effects prepared by BMJ-TAG

Any information supplied to NICE which has been marked as confidential, has been redacted. All personal information has also been redacted.

NATIONAL INSTITUTE FOR HEALTH AND CARE EXCELLENCE

Single technology appraisal

Dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648]

Document B

Company evidence submission

September 2022

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 1 of 134

Instructions for companies

This is the template for submission of evidence to the National Institute for Health and Care Excellence (NICE) as part of the single technology appraisal (STA) process. Please note that the information requirements for submissions are summarised in this template; full details of the requirements for pharmaceuticals and devices are in the user guide.

This submission must not be longer than 150 pages, excluding appendices and the pages covered by this template. If it is too long it will not be accepted.

Companies making evidence submissions to NICE should also refer to the NICE health technology evaluation guidance development manual.

In this template any information that should be provided in an appendix is listed in a box.

Highlighting in the template (excluding the contents list)

Square brackets and **** highlighting are used in this template to indicate text that should be replaced with your own text or deleted. These are set up as form fields, so to replace the prompt text in ***** ************* with your own text, click anywhere within the highlighted text and type. Your text will overwrite the highlighted section.

To delete grey highlighted text, click anywhere within the text and press DELETE.

Grey highlighted text in the footer does not work as an automatic form field, but serves the same purpose – as prompt text to show where you need to fill in relevant details. Replace the text highlighted in ****** in the header and footer with appropriate text. (To change the header and footer, double click over the header or footer text. Double click back in the main body text when you have finished.)

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 2 of 134

Contents

Instructions for companies ............................................................................................................... 2 Contents ........................................................................................................................................... 3 Tables and figures............................................................................................................................ 4 Abbreviations ................................................................................................................................... 7 B.1. Decision problem, description of the technology and clinical care pathway .......................... 11 B.1.1. Decision problem ............................................................................................................. 11 B.1.2. Description of the technology being evaluated ............................................................... 15 B.1.3. Health condition and position of the technology in the treatment pathway .................... 16 B.1.4. Equality considerations ................................................................................................... 27 B.2. Clinical effectiveness .............................................................................................................. 28 B.2.1. Identification and selection of relevant studies ............................................................... 28 B.2.2. List of relevant clinical effectiveness evidence ............................................................... 29 B.2.3. Summary of methodology of the relevant clinical effectiveness evidence ..................... 32 B.2.4. Statistical analysis and definition of study groups in the relevant clinical effectiveness evidence 40 B.2.5. Critical appraisal of the relevant clinical effectiveness evidence .................................... 43 B.2.6. Clinical effectiveness results of the relevant studies: DELIVER ..................................... 44 B.2.7. Subgroup analysis ........................................................................................................... 53 B.2.8. Prespecified analysis: estimated benefits with long-term treatment with dapagliflozin .. 55 B.2.9. Meta-analysis .................................................................................................................. 56 B.2.10. Indirect and mixed treatment comparisons ................................................................... 61 B.2.11. PRESERVED-HF trial outcome summary .................................................................... 61 B.2.12. Adverse reactions .......................................................................................................... 68 B.2.13. Ongoing studies ............................................................................................................ 75 B.2.14. Interpretation of clinical effectiveness and safety evidence .......................................... 75 B.3. Cost effectiveness .................................................................................................................. 79 B.3.1. Published cost-effectiveness studies .............................................................................. 80 B.3.2. Economic analysis ........................................................................................................... 80 B.3.3. Clinical parameters and variables ................................................................................... 85 B.3.4. Measurement and valuation of health effects ............................................................... 101 B.3.5. Cost and healthcare resource use identification, measurement and valuation ............ 103 B.3.6. Severity .......................................................................................................................... 107 B.3.7. Uncertainty .................................................................................................................... 108 B.3.8. Summary of base case analysis inputs and assumptions ............................................ 108 B.3.9. Base case results .......................................................................................................... 116 B.3.10. Exploring uncertainty ................................................................................................... 117 B.3.11. Subgroup analysis ....................................................................................................... 123 B.3.12. Benefits not captured in the QALY calculation ........................................................... 123 B.3.13. Validation ..................................................................................................................... 123 B.3.14. Interpretation and conclusions of economic evidence ................................................ 126 References ................................................................................................................................... 127 B.4. Appendices ........................................................................................................................... 134

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 3 of 134

Tables and figures

Table 1: The decision problem ...................................................................................................... 11 Table 2: Technology being evaluated ............................................................................................ 15 Table 3: Classifications of HF across major international HF guidelines ...................................... 17 Table 4: NYHA classification criteria .............................................................................................. 21 Table 5: KCCQ questionnaire domains and summary scores ...................................................... 22 Table 6: Clinical effectiveness evidence ........................................................................................ 30 Table 7: Summary of trial methodology: DELIVER ....................................................................... 32 Table 8: Characteristics of participants in DELIVER across treatment groups ............................. 36 Table 9: Summary of statistical analyses in DELIVER .................................................................. 40 Table 10: Critical appraisal of DELIVER ....................................................................................... 43 Table 11: Analysis of the composite endpoint of CV mortality and recurrent HF events in DELIVER ........................................................................................................................................ 47 Table 12: Summary of HF events and CV mortality – number of events per patient in DELIVER 47 Table 13: Summary of adjudicated death classification in DELIVER[a] .......................................... 49 Table 14: Change in KCCQ parameters at Month 1, Month 4 and Month 8 ................................. 50 Table 15: Analysis of first occurrence of hospitalisation from any cause in DELIVER ................. 52 Table 16: Effect of dapagliflozin versus placebo on NYHA functional class over time[a] ................ 53 Table 17: Primary composite endpoint (CV mortality and HF events) in patients with HFimpEF compared with those with HF and an LVEF consistently >40% .................................................... 55 Table 18: Baseline characteristics of the patients included in the pooled analysis of DELIVER and DAPA-HF by LVEF category .................................................................................................. 58 Table 19: Summary of trial methodology: PRESERVED-HF ........................................................ 62 Table 20: Characteristics of participants in PRESERVED-HF across treatment groups .............. 65 Table 21: Primary and secondary endpoints at 12 weeks after treatment initiation in PRESERVED-HF ........................................................................................................................... 67 Table 22: Number of patients with AEs in any category in DELIVER – on treatment................... 70 Table 23: Number of patients with SAEs (≥ 0.5%) by preferred term in DELIVER – On treatment ........................................................................................................................................................ 71 Table 24: Safety analysis in PRESERVED-HF ............................................................................. 72 Table 25: Adverse drug reactions reported in the SmPC for dapagliflozin: adverse reactions in placebo-controlled clinical studies[a] and postmarketing experience .............................................. 74 Table 26: Summary of the key differences in modelling approaches between TA679 versus this appraisal ......................................................................................................................................... 80 Table 27: Key features of the economic analysis .......................................................................... 83 Table 28: Patient demographic characteristics incorporated in the base case economic analysis ........................................................................................................................................................ 85 Table 29: Patient clinical characteristics incorporated in the base case economic analysis ........ 85 Table 30: Patient medical history incorporated in the base case economic analysis ................... 86 Table 31: Patient demographic characteristics based on the UK CPRD dataset used in a scenario analysis ............................................................................................................................ 86 Table 32: Patient clinical characteristics based on the UK CPRD dataset used in a scenario analysis .......................................................................................................................................... 87 Table 33: Patient medical history based on the UK CPRD dataset used in a scenario analysis . 87 Table 34: Monthly KCCQ-TSS transition matrix ............................................................................ 88 Table 35: AIC and BIC values of the unadjusted parametric survival model distributions for CV mortality derived from the DELIVER ITT population ..................................................................... 91 Table 36: AIC and BIC values of the unadjusted parametric survival model distributions for all-

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 4 of 134

cause mortality derived from the DELIVER ITT population ........................................................... 91 Table 37: AIC and BIC values of the adjusted parametric survival model distributions for CV mortality derived from the DELIVER ITT population ..................................................................... 92 Table 38: AIC and BIC values of the adjusted parametric survival model distributions for allcause mortality derived from the DELIVER ITT population ........................................................... 92 Table 39: Age and sex-stratified mortality rates derived from WHO global health estimates ....... 97 Table 40: Adjusted GEEs predicting HHF events ......................................................................... 98 Table 41: Adjusted GEEs predicting UHFV events ....................................................................... 99 Table 42: Unadjusted GEE coefficients derived from the DELIVER trial .................................... 100 Table 43. Adverse event frequency observed in DELIVER ........................................................ 100 Table 44: Annual probability of AEs ............................................................................................. 100 Table 45: Health state utility values used in the base case economic analysis .......................... 101 Table 46: Utility decrements used for HF events......................................................................... 102 Table 47. Utility decrements used for AEs .................................................................................. 102 Table 48: Summary of utility values used in base case economic analysis ................................ 103 Table 49: Annual drug acquisition costs applied within the cost-effectiveness analysis ............ 104 Table 50: Event costs for transient events and mortality............................................................. 105 Table 51: Health state resource use and frequency and unit costs ............................................ 105 Table 52: Unit costs used for health state costs .......................................................................... 106 Table 53: Unit costs for adverse events ...................................................................................... 106 Table 54: Summary features of QALY shortfall analysis ............................................................. 107 Table 55: Summary of health state benefits and utility values for QALY shortfall analysis ........ 107 Table 56: Summary of QALY shortfall analysis ........................................................................... 108 Table 57: Summary of variables applied in the economic model ................................................ 108 Table 58: Summary of assumptions in the base case economic analysis .................................. 114 Table 59: Base case economic analysis results – ICERs ........................................................... 116 Table 60: Base case economic analysis results – NHB .............................................................. 116 Table 61: Base case PSA results ................................................................................................ 117 Table 62: Summary of scenario analyses ................................................................................... 120 Table 63. Statistics from the internal validation of event rates .................................................... 125 Figure 1: Summary of NICE NG106 diagnostic pathway for HF ................................................... 20 Figure 2: Summary of pharmacologic treatments for patients with HF and an LVEF >40% recommended in NG106 ................................................................................................................ 23 Figure 3: Current treatment pathway for patients with HF and an LVEF >40% in NG106 and proposed positioning of dapagliflozin............................................................................................. 26 Figure 4: DELIVER trial design ...................................................................................................... 35 Figure 5: Testing procedure for DELIVER ..................................................................................... 42 Figure 6: Patient disposition in DELIVER ...................................................................................... 43 Figure 7: KM plot of the primary composite endpoint (CV mortality and HF events) in DELIVER 45 Figure 8: Forest plot of the primary composite endpoint (CV mortality and HF events) and the individual components in DELIVER[a] .............................................................................................. 46 Figure 9: KM plot of CV mortality in DELIVER .............................................................................. 48 Figure 10: KM plot of all-cause mortality in DELIVER ................................................................... 49 Figure 11: Mean changes in KCCQ domains over time by treatment allocation[a,b] ....................... 50 Figure 12: Responder analyses of clinically meaningful change in KCCQ domains at 8 months with dapagliflozin versus placebo[a] ................................................................................................. 51 Figure 13: Forest plot of the primary composite endpoint (CV mortality and HF events) by subgroups in DELIVER .................................................................................................................. 54

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 5 of 134

Figure 14: Effect of dapagliflozin on key clinical outcomes in pooled DAPA-HF and DELIVER dataset ............................................................................................................................................ 60 Figure 15: Effect of randomised treatment on CV mortality according to the prespecified subgroups[a] ..................................................................................................................................... 61 Figure 16: Effects of dapagliflozin on the primary endpoint and its components .......................... 68 Figure 17: KM plot of the cumulative percentage of patients with premature permanent discontinuation of treatment in DELIVER[a] ..................................................................................... 72 Figure 18: Schematic of Markov state-transition model structure, health states, and possible transitions ....................................................................................................................................... 82 Figure 19: KM curves for CV mortality in the DELIVER trial, stratified by treatment .................... 89 Figure 20: KM curves for all-cause mortality in the DELIVER trial, stratified by treatment ........... 90 Figure 21: Adjusted survival model extrapolations for CV mortality[a] ............................................ 93 Figure 22: Adjusted survival model extrapolations for all-cause mortality[a] ................................... 94 Figure 23: Adjusted all-cause mortality predictions for patients receiving placebo in the DELIVER trial compared with meta-analysed 5-year survival reported in Jones et al . (2019)[116a] ................ 95 Figure 24: Adjusted all-cause mortality predictions for patients receiving placebo in the DELIVER trial compared with long-term survival reported in Shahim et al. (2021)[117a] ................................. 96 Figure 25: Cost-effectiveness scatter plot from PSA ................................................................... 117 Figure 26: Cost-effectiveness acceptability curve from PSA ...................................................... 118 Figure 27: ICER convergence plot from PSA .............................................................................. 118 Figure 28: Tornado plot of DSA results ....................................................................................... 119 Figure 29: Internal validation of survival for the DELIVER ITT population[a] ................................ 124 Figure 30: Internal validation of predicted versus observed event rates for the DELIVER ITT population[a] ................................................................................................................................... 125

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 6 of 134

Abbreviations

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 7 of 134

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 8 of 134

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 9 of 134

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 10 of 134

B.1. Decision problem, description of the technology and clinical care pathway

B.1.1. Decision problem

This submission aims to demonstrate the clinical and cost-effectiveness of dapagliflozin as a treatment for patients with chronic heart failure (HF) and a left ventricular ejection fraction (LVEF) >40%. This population is covered under the technology’s anticipated expanded marketing authorisation for this indication: ************* ** ********* ** ****** *** *** ********* ** *********** ******* *** Treatment with dapagliflozin for patients with symptomatic chronic HF and a reduced LVEF (HFrEF; LVEF ≤40%) has already been recommended by NICE (TA679; 2021).[1]

Table 1: The decision problem

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 11 of 134

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 12 of 134

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 13 of 134

Abbreviations : CKD: chronic kidney disease; HF: heart failure; HFrEF: HF with reduced ejection fraction; LVEF: left ventricular ejection fraction; NHS: National Health Service; NICE: National Institute for Health and Care Excellence; N/A: not applicable; PSS: Personal and Social Services; QALY: quality-adjusted life year; NYHA: New York Heart Association; SGLT2: sodium-glucose co-transporter-2; SoC: standard of care; T2DM: type 2 diabetes mellitus. Source : Dapagliflozin NICE final scope [ID1648].[11]

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 14 of 134

B.1.2. Description of the technology being evaluated

The draft summary of product characteristics (SmPC) for dapagliflozin that covers the indication of relevance to this submission (patients with HF and an LVEF >40%) is provided in Appendix C. Details of the technology being evaluated, including the method of administration, dosing and related costs, are provided in Table 2.

Table 2: Technology being evaluated

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 15 of 134

aCosting assumption: 365.25 days per year. Abbreviations: CKD: chronic kidney disease; HF: heart failure; HFrEF: Heart failure with a reduced ejection fraction; LVEF: left ventricular ejection fraction; MHRA: Medicines and Healthcare Products Regulatory Agency; T2DM: type 2 diabetes mellitus; SGLT2: sodium-glucose transporter-2; SmPC: summary of product characteristics.

B.1.3. Health condition and position of the technology in the treatment pathway

B.1.3.1. HF overview

HF is a complex clinical syndrome that occurs when the heart is unable to pump enough blood to maintain a cardiac output that meets the metabolic needs of the body either at rest or on exertion, or without a rise in intracardiac pressure.[2] Diagnosis of HF requires the presence of both cardiac dysfunction, as well as symptoms and signs of HF such as difficulty breathing, fatigue, ankle swelling, or oedema.[2, 3] Mortality associated with HF remains high, with approximately 50–75% of patients dying within 5 years of a HF diagnosis.[14]

Most commonly, HF is due to myocardial dysfunction, which may be systolic (reflecting contraction of the left ventricle of the heart), diastolic (reflecting relaxation and filling of the left ventricle of the heart), or both. However, valvular, pericardial or endocardial disease, as well as abnormalities of heart rhythm and conduction, can also cause or contribute to HF. The most common causes of myocardial dysfunction are ischaemic heart disease (IHD) and hypertension, although the cause in many patients is not known.[15]

Patients with HF often have other co-morbid conditions that may contribute to, or interact with, the severity of HF.[16] In addition to CV-related co-morbidities such as hypertension, coronary artery disease (CAD), atrial fibrillation (AF) and CKD, other HF co-morbidities include chronic obstructive pulmonary disease (COPD) and T2DM.[17-20] Co-morbidities, such as CKD and T2DM have important clinical implications on patient outcomes and healthcare costs,[15, 20-25] further accentuating the severity of disease burden with greater impact on mortality and morbidity.[26] A pooled analysis of studies with a follow-up of at least 6 months reported a 28% higher mortality risk in patients with HF and T2DM than patients with HF alone,[27] and over a 2-year period, a 12.4% decrease in survival was observed in patients with HF and CKD versus CKD alone.[28 ]

HF is usually classified based on measurement of LVEF, obtained from echocardiography (or other imaging modalities). LVEF is a means of quantifying the percentage of blood in the left ventricle that is pumped out with every contraction.[29] Based on this measurement of LVEF, individuals with HF can be broadly classified into those with a preserved LVEF (HFpEF), those with a mildly reduced LVEF (HFmrEF), those with a reduced LVEF (HFrEF) and those with an improved LVEF (HFimpEF; Table 3):[3, 15, 30]

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 16 of 134

• HFrEF: Up to half of patients with HF have a reduced LVEF ≤40%.[3] The underlying pathophysiology in these patients is systolic dysfunction.[15] Dapagliflozin in this indication has already been recommended by NICE in TA679.[1 ]

• HFmrEF: Patients with HF and an LVEF between 41% and 49% are described as having HF with a mildly reduced LVEF, to reflect the fact that in most patients, pathophysiologically, HFmrEF is more like HFrEF than HFpEF.[3]

• HFpEF: Patients with signs and symptoms of HF, with raised natriuretic peptides and evidence of structural abnormalities such as elevated left ventricular filling pressure at rest or during exercise but an LVEF ≥50% are described as having HFpEF.[3, 15]

• HFimpEF : Patients who had prior LVEF ≤40% with a follow-up measurement of LVEF >40%.[30]

HF has historically been categorised as per the four phenotypes above based on LVEF, mainly due to multiple HF clinical trials initially demonstrating significant outcomes for treatments in patients with HF and an LVEF ≤40%. It is therefore important to note that the overall clinical syndrome of HF includes patients across the entire range of LVEF, which is a normally distributed variable.[3]

Moreover, while there are four HF classifications, there are in effect two clinically distinct patient populations with HF in UK clinical practice; those with LVEF ≤40% and those with LVEF >40%. This is predominantly due to the lack of disease-modifying treatment options for patients with HF and an LVEF >40%, coupled with the availability of disease-modifying treatment options for HF and an LVEF ≤40% that are routinely commissioned in UK clinical practice. Therefore, HFmrEF and HFpEF are not usually considered as clinically distinct subgroups for the purposes of treatment decisions. As outlined in the NICE final scope and in the decision problem for this appraisal (Table 1), this submission is concerned with the treatment of patients with HF and an LVEF >40%.

Table 3: Classifications of HF across major international HF guidelines

aSigns may not be present in the early stages of HF (especially in HFpEF) and in optimally treated patients; bFor

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 17 of 134

the diagnosis of HFmrEF, the presence of other evidence of structural heart disease (e.g., increased left atrial size, LV hypertrophy or echocardiographic measures of impaired LV filling) makes the diagnosis more likely;[c] For the diagnosis of HFpEF, the greater the number of abnormalities present, the higher the likelihood of HFpEF; dFor example, elevated natriuretic peptide, non-invasive and invasive hemodynamic measurement. Abbreviations: ACC: American College of Cardiology; AHA: American Heart Association; ESC: European Society of Cardiology; HF: heart failure; HFA: Heart Failure Association of the European Society of Cardiology; HFSA: heart failure Society of America; HFimpEF: heart failure with an improved ejection fraction; HFmrEF: heart failure with a mildly reduced ejection fraction; HFpEF: heart failure with a preserved ejection fraction; HFrEF: heart failure with a reduced ejection fraction; HFSA: Heart Failure Society of America; JHFS: Japanese Heart Failure Society; LV: left ventricular; LVEF: left ventricular ejection fraction.

B.1.3.2. Disease burden

Summary of disease burden

There are currently no disease-modifying treatments routinely commissioned in UK clinical practice for patients with HF and an LVEF >40%, highlighting the urgent unmet need for easily accessible new treatments which can reduce mortality and

hospitalisation, and improve disease symptoms and quality of life for these patients

• Mortality associated with HF is high ;[14] following a hospitalisation for HF (HHF), the 5- year survival for patients with HFpEF is 35%, which is worse than many cancers.[32 ]

• For patients with HF and an LVEF >40%, the co-morbidity burden is substantial, and may contribute to, or interact with, patients’ HF severity, which can greatly impact health-related quality of life (HRQoL).[16, 33-36] Co-morbidities of HF include CAD, AF, CKD, COPD and T2DM.[17-20 ]

• Patients with HF and an LVEF >40% struggle with poor HRQoL similar to, or worse than, patients with HFrEF .[32] For instance, physical activity levels for these patients have been reported to be as suppressed as those observed in patients with moderate-to-severe COPD.[37]

• HF and an LVEF >40% is associated with a considerable economic burden, primarily driven by high hospitalisation rates .[38-42]

• The prevalence of HF is likely to rise in the future , due to factors such as the ageing population in the UK, and rising rates of obesity and T2DM.[14, 43, 44]

HF represents one of the most significant healthcare problems in the UK; one in five people over 40 years old are at risk of developing HF in their lifetime.[45] While the mortality associated with HF remains high with up to 75% of patients dying within 5 years of diagnosis,[14] for those with HF and an LVEF >40%, the 5-year survival rate following a HHF is 35%.[32] Cardiovascular disease (CVD) is believed to cause a quarter of all deaths in the UK and has been identified by the NHS in its Long Term Plan as the single biggest area where lives can be saved until 2029.[46] To address this, the NHS has set the objective to better support people with HF in primary care through the provision of multi-disciplinary teams working across primary and secondary care.[46] Optimising treatment outcomes in HF will help meet this long-term NHS goal.

For patients with HF and an LVEF >40%, the co-morbidity burden is substantial, and may contribute to, or interact with, patients’ HF severity, which can greatly impact HRQoL.[16, 33-36] Many risk factors and co-morbidities can contribute to HF and an LVEF >40% including CAD, AF, CKD, COPD and T2DM.[2, 29] There is a complex relationship between HF and its comorbidities; HF may also cause common co-morbidities, which can then adversely affect overall

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 18 of 134

patient outcomes.[47] For instance, HF is a known risk factor for the development of incident comorbidities such as CKD and T2DM,[47] both of which can negatively impact patient outcomes and healthcare costs,[15, 20-25] further accentuating the severity of the HF disease burden.[26] As previously mentioned, there is a 28% higher mortality risk in patients with HF and T2DM than patients with HF alone, and a 12.4% decrease in survival over a 2-year period in patients with HF and CKD versus patients with CKD alone.[27, 28] This emphasises the importance of managing comorbidities in the treatment of HF.

The HRQoL of patients with HF and an LVEF >40% is poor, similar to, or worse than, patients with HFrEF.[32] For instance, exercise intolerance is a hallmark feature in patients with HF and an LVEF >40%; cardiac and peripheral abnormalities as well as changes in body composition cause tissue congestion and disrupt oxygen delivery, resulting in exercise intolerance and physical inactivity.[48, 49] Physical activity levels for patients with HF and an LVEF >40% have been reported to be as suppressed as those observed in patients with moderate-to-severe COPD.[37] Improving exercise capacity and HRQoL is therefore a primary goal in the management of patients with HF and an LVEF >40%.[48-50]

HF and an LVEF >40% is associated with a substantial economic burden, primarily driven by high rates of hospitalisations.[38-42] A systematic review of the economic burden associated with HFpEF (2001–2020) reported that hospitalisations account for approximately 80% of total costs associated with HFpEF treatment.[51] HF is one of the leading causes of hospitalisations in people aged >65 years[52] and of rehospitalisation in the general population.[53] Thus, HF is associated with a high economic burden and costs the NHS up to 2% of its annual budget (~£3 billion).[43, 54] Reducing hospitalisations is therefore key to addressing the economic burden associated with HF. In addition to direct costs, HF also contributes substantial indirect costs as a result of mortality, lost productivity, and the need to provide long-term domiciliary of institutional care for some patients.[55]

The prevalence of HF is likely to rise in the future, due to factors such as the ageing population in the UK, and rising rates of obesity and T2DM.[14, 43, 44] Despite improvements in clinical care, many patients still experience disabling symptoms,[56, 57] and mortality rates are expected to remain high.[14, 57] Currently, there are no disease-modifying treatment options routinely commissioned by the NHS for patients with diagnosed HF and an LVEF >40% as, unlike HFrEF, several randomised controlled trials (RCTs) have failed to demonstrate improved outcomes in this patient population.[58] Thus, the current guideline on diagnosis and treatment of HF (NICE NG106) advises only on the treatment of underlying co-morbidities for patients with HF and an LVEF >40%, and to manage any congestion with diuretics.[59] There is consequently a substantial unmet need for easily accessible new treatments which can lower mortality, reduce hospitalisation rates, and improve symptoms and HRQoL for patients with HF and an LVEF >40%.

B.1.3.3. Epidemiology

The prevalence of HF is estimated to be 0.91% in England.[60] Therefore, based on 2021 population estimates,[61] there are approximately 423,000 adult patients with HF in England and Wales.

As this submission is concerned with patients with HF and an LVEF >40% from both outpatient and inpatient (acute) settings, utilisation of data from the Clinical Practice Research Datalink (CPRD) dataset can be considered more representative of UK clinical practice to estimate the size of this patient population than the National HF Audit 2022 which focusses solely on the

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 19 of 134

acute setting.[10, 57] The CPRD dataset includes patients with at least one relevant code for HF diagnosis in primary (SNOMED-CT) or secondary care (ICD-10) in the period of 1[st] January 2010 and 1[st] January 2020 in England.[10] Of ***** eligible patients with HF and an LVEF recorded, approximately ***** patients (*****) had a recorded LVEF >40%.[62] Applying this proportion to the 423,000 adult patients with HF in England and Wales means that there are approximately ******* patients with HF and an LVEF >40% in primary or secondary care settings in England and Wales.

It should be noted that some UK prevalence estimates are as high as 900,000 patients with HF,[63] highlighting that the prevalence of HF based on the above data is likely underestimated. Also, although echocardiography is recommended by NICE for the diagnosis of HF, the measurement of LVEF has not always been recorded well in Read codes,[64] which constitutes a barrier to accurately assessing HF epidemiology in UK clinical practice.

B.1.3.4. Diagnosis of HF

The heterogenous nature of HF and an LVEF >40% (e.g., different contributing conditions), and the high frequency of co-morbidities, (e.g., CAD, AF, CKD, COPD and T2DM) that may mimic or accompany the condition, can make diagnosis challenging.[2, 3] Current UK practice is consistent with the NICE HF guideline diagnostic pathway in England (NG106; Figure 1). Patients in whom there is clinical suspicion of HF receive a measurement of plasma N-terminal pro B-type natriuretic peptide (NT-proBNP). Where the NT-proBNP concentration is ≥400 ng/L, clinical assessment and transthoracic echocardiography should occur within 2 or 6 weeks to allow a diagnosis of HF to be established either by an HF specialist, or in some cases by a general practitioner (GP) following open access echocardiography.

Figure 1: Summary of NICE NG106 diagnostic pathway for HF

==> picture [452 x 253] intentionally omitted <==

Source: Adapted from NICE NG106.[59] Abbreviations : ECG: echocardiogram; HF: heart failure; NT-proBNP: N-terminal pro B-type natriuretic peptide.

Clinical guidelines for the diagnosis of HF from the European Society of Cardiology (ESC), updated in 2022, have similar recommendations.[3] Once a diagnosis of HF is confirmed, the

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 20 of 134

measurement of LVEF is used to further categorise the disease as either HFrEF (LVEF ≤40%), HFmrEF (LVEF 41–49%) or HFpEF (LVEF ≥50%).[3] Whilst the diagnostic pathway is the same for all HF patients irrespective of LVEF, in practice (as in guidelines) once a patient’s LVEF has been determined, the therapeutic pathways diverge for HFrEF versus HFmrEF/HFpEF (though notably diuretics are common to both pathways). As previously mentioned, the management of patients with HF and an LVEF >40% is the same for both HFmrEF and HFpEF as no diseasemodifying treatments are routinely commissioned in UK clinical practice for this patient population.

In UK clinical practice, HF symptom severity is routinely assessed using the New York Heart Association (NYHA) Functional Classification (Table 4), which is based on physical limitations due to symptoms. However, symptom severity does not correlate closely with LV function and patients with “mild symptoms” (NYHA class II) still have a substantial risk of hospitalisation and death.[3] While the NYHA tool remains useful as a brief description of a patient’s clinical status, it is highly subjective with an inter-rater concordance of 54–56% for mild to moderate symptoms,[65] poorly reproducible, including among trained cardiologists,[66] and not patient-centric as it is a clinician’s assessment of a patients’ functional limitations.[65] Moreover, input from UK clinical experts indicates that NYHA class has a limited impact on the treatments offered to patients in clinical practice, given the subjective nature of the classification criteria.[10]

Table 4: NYHA classification criteria

Abbreviations : HF: heart failure; NYHA: New York Heart Association.

The Kansas City Cardiomyopathy Questionnaire (KCCQ) has been demonstrated to be a reliable and valid patient-reported outcome measure to assess HRQoL in HFpEF,[50] and is considered to provide a more comprehensive and robust assessment of a patient’s health status and be more responsive to changes in health status than the NYHA classification.[67] The KCCQ score is composed of several domains such as physical limitations, symptoms, social limitations and QoL, as presented in Table 5.[67] Importantly, the KCCQ is a patient-reported outcome providing a more granular assessment of a patient’s symptoms and limitations. It is consequently a more robust measure of changes in a patient’s condition than NYHA class, particularly in clinical trials, and has established thresholds which indicate clinically relevant changes in health status.[68] Baseline KCCQ–Total Symptom Score (TSS) has been found to align with clinical outcomes, with patients with a worse KCCQ-TSS at baseline having higher mortality and higher rates of HHF.[68] As a result, KCCQ rather than NYHA class, has become the standard tool used in clinical trials to evaluate patient-reported health status and response to treatment in patients with HF.

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 21 of 134

Table 5: KCCQ questionnaire domains and summary scores

Sources : Spertus et al. (2020);[69] FDA (2020).[70] Abbreviations : CSS: Clinical Summary Score; HF: heart failure; KCCQ: Kansas City Cardiomyopathy Questionnaire; OSS: Overall Summary Score; QoL: quality of life; TSS: Total Symptom Score.

B.1.3.5. Current management of patients with HF and an LVEF >40%

As per NICE NG106, recommendations for pharmacological treatments in HF are stratified between HFrEF and HFpEF in UK clinical practice (Figure 2).[59] In this context, as the management of patients with HF and an LVEF >40% is the same for both HFmrEF and HFpEF, it is assumed that recommendations in this clinical guideline for the HFpEF population comprise both subpopulations i.e., patients with HF and an LVEF >40%.

Once the diagnosis of HF and an LVEF >40% has been confirmed on the basis of clinical assessment, natriuretic peptides and echocardiography, patients are typically offered loop diuretics for congestive symptoms and fluid retention, in addition to treatments for any comorbidities.[59] While patients with HF and an LVEF >40% may have multiple varying comorbidities for which they are separately treated, SoC for symptom management of HF and an LVEF >40% in UK clinical practice predominantly comprises treatment with loop diuretics (typically furosemide or bumetanide).[4] In addition, unless the condition is unstable, a personalised exercise cardiac rehabilitation programme is to be offered, though uptake of this is typically poor.[59] For those whose HF does not respond to this treatment, further specialist advice is needed.[59]

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 22 of 134

Figure 2: Summary of pharmacologic treatments for patients with HF and an LVEF >40% recommended in NG106

==> picture [446 x 261] intentionally omitted <==

aMeasure serum sodium, potassium and assess renal function before and after starting and after each dose increment. If eGFR is 30 to 45 ml/min/1.73 m[2] , consider lower doses or slower titration of ACEi or ARBs, MRAs, sacubitril valsartan and digoxin.[b] It is assumed that recommendations for the HFpEF population comprise both the HFpEF and HFmrEF populations; i.e., those with HF and an LVEF >40%. Source : Adapted from NICE NG106.[59]

Abbreviations : ACEi: angiotensin-converting enzyme inhibitor; ARB: angiotensin-receptor blocker; BB: betablocker; HF: heart failure; eGFR: estimated glomerular filtration rate; HFpEF: heart failure with a preserved ejection fraction; HFrEF: heart failure with a reduced ejection fraction; MRA: mineralocorticoid-receptor antagonist; TA: technology assessment.

B.1.3.6. Diagnosis and management of patients with HF in clinical practice

There are three main routes through which patients are diagnosed with HF and an LVEF >40% in the UK: by a specialist using echocardiography following GP referral due to raised NT-proBNP and HF symptoms (as per the NICE pathway), in general practice following NT-proBNP tests using open access echocardiography or following an emergency admission to hospital for an acute HF event.

Under the NICE diagnosis pathway, once HF symptoms are recognised and clinical suspicion of HF is raised, the patient is referred by their GP for further HF diagnostic tests, specifically echocardiography, performed by an HF specialist. As few as 24% of patients with recorded HF symptoms follow the NICE pathway to diagnosis, with only 4% completing the NICE pathway within its 6-week timeframe.[64] In an observational study using CPRD data between 2010 and 2013, from presenting with symptoms suggestive of HF in primary care to recorded relevant investigations either as an echocardiogram or NT-pro-BNP test, a median time of 9.5 months (292 days) was observed, and for a referral to a specialist, a median time of 7.7 months (236 days) was observed, substantially exceeding the NICE recommended timelines of 2–6 weeks.[71]

Alternatively, in some cases patients are referred by their GP for diagnostic tests performed in primary care through open access echocardiography. There are some limitations to this approach owing to variable expertise amongst GPs in interpreting the results for patients with HF and an LVEF >40%. Several other important parameters need to be measured and correctly

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 23 of 134

interpreted, including the presence of both cardiac dysfunction, as well as symptoms and signs of HF such as difficulty breathing, fatigue, ankle swelling, or oedema.[2, 3] There is the potential for misdiagnosis or HF misclassification with open access echocardiography, with particular risk that patients with hypertension and other comorbidities may be wrongly classified as having HF and an LVEF >40%.[10] This highlights the specialist-confirmed HF diagnosis following echocardiogram as per the NICE clinical pathway.

In UK clinical practice, the majority of patients (approximately 80%) only receive a formal diagnosis of HF following hospitalisation for acute decompensated HF.[64] According to UK clinical experts consulted by AstraZeneca, many of these patients would typically have been known to primary care as having suspected HF symptoms but are only formally coded as having HF following an acute admission. These patients, once diagnosed, tend to be quickly discharged back to primary care where they are then managed for chronic HF symptoms.

As well as incident cases, there are prevalent populations of patients already diagnosed with HF and an LVEF >40% through one of the three pathways outlined above that are predominantly managed in primary care due to a lack of specific HF services actively managing this population. Input from UK clinical experts indicates that limited resource availability (e.g., HF nurses, cardiologists) contributes to inequalities in patient access to relevant investigations and HF services in the UK.[10] Moreover, the measurement of LVEF has not always been recorded well in Read Codes, which constitutes a further barrier to effective management of the condition where early identification and classification of HF are key to avoid delays that can negatively impact morbidity and mortality.[72,64]

Once a diagnosis of HF and an LVEF >40% has been established, loop diuretics for congestive symptoms and fluid retention, namely furosemide and bumetanide, as well as treatments for comorbidities are to be offered according to NG106.[4, 59] For instance, in a contemporary, crosssectional study of patients with HFpEF in primary care, 80% were hypertensive, thus received treatment for this co-morbidity.[73] The majority of patients with HF and an LVEF >40% are managed in primary care and are either not referred to specialists or, if referred, are not provided with a treatment plan upon discharge.[73, 74, 75] Inputs from UK clinical experts indicate that, in UK clinical practice, only a few HF centres commission services to support patients with HF and an LVEF >40% after diagnosis, or offer specialised HFpEF clinics alongside their usual HF services, owing predominantly to the lack of therapeutic options available to this patient population to date.[10]

For all patients with HF and stable disease, a personalised exercise cardiac rehabilitation programme should be offered according to NG106,[59] which has been shown to improve outcomes after one year. However, in UK clinical practice few patients are referred, with just 12% of patients with HF referred for cardiac rehabilitation following a HHF in the 2022 National HF Audit.[57] This is due mainly to capacity challenges and the design of services being unsuitable for frail patients.[10]

In summary, patients diagnosed with HF and an LVEF >40% are predominantly managed by primary care physicians with a focus on HF symptom control to relieve congestion and oedema and managing common co-morbidities such as hypertension.[10] NG106 states that monitoring in primary care, including clinical assessment, renal assessment and medication review, should be individualised with a frequency based on co-morbidities, prescribed medications and clinical stability, but that this should be at least six-monthly.[54] Primary care physicians already have considerable clinical experience in the prescribing of dapagliflozin and could therefore initiate

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 24 of 134

treatment at the earliest opportunity for patients with new and existing diagnoses, or as part of routine check-up appointments in situations where there is insufficient capacity to proactively schedule a therapy review appointment. Although the availability of novel therapies for patients with HF and an LVEF >40% may result in a greater focus on specialist service provision, service re-design specifically for the prescribing of dapagliflozin in these patients would not be necessary as dapagliflozin does not require dose up-titration nor specific monitoring over and above what is already recommended for a patient with HF.

B.1.3.7. Proposed positioning of dapagliflozin in the treatment pathway for patients with HF and an LVEF >40%

In the pivotal DELIVER RCT, dapagliflozin administered in addition to SoC demonstrated a significant reduction in the primary composite endpoint of CV mortality and HF events (HHF or an urgent HF visit [UHFV] requiring IV diuretic therapy, hereafter jointly referred to as HF events for ease of reading) compared with placebo in addition to SoC (see Section B.2.6), along with a favourable safety profile and significant symptom benefit as measured by the KCCQ-TSS.[76] SoC consisted of the treatments recommended in NICE NG106, namely diuretics for decongestion and the management of co-morbidities.[59]

Positioning

The proposed positioning of dapagliflozin is in patients with a diagnosis of HF and an LVEF >40% confirmed by a specialist, as an add-on to current SoC, which predominantly comprises loop diuretics as illustrated in Figure 3 as part of the existing NICE NG106 treatment pathway.

This proposed positioning is based on UK clinical expert input and the clinical benefit demonstrated with dapagliflozin in addition to SoC at this place in the pathway in the DELIVER trial.[10, 76] Given the absence of disease-modifying treatment options in patients with HF and an LVEF >40%, dapagliflozin should be initiated as soon as the diagnosis is established, and irrespective of diuretic initiation depending on the specific signs of congestion. For patients with a documented diagnosis of HF and an LVEF >40% that are already managed in primary care (or those not routinely followed-up within specialist care), dapagliflozin could be initiated at the earliest opportunity, ideally following proactive invitation for a treatment optimisation review or alternatively, where capacity is a limitation, during their routine check-up appointment without the need for a specific or extended appointment. Having the HF diagnosis confirmed by a specialist mitigates the risk of potential misdiagnosis following misinterpretation of open access echocardiography and therefore removes the risk of over-treatment in patients with conditions that can mimic HF and an LVEF >40%.

In the context of the existing NICE clinical pathway adapted in the figure below, ‘HFpEF’ encompasses all patients with HF and an LVEF >40% for which clinical management and treatment are the same.

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 25 of 134

Figure 3: Current treatment pathway for patients with HF and an LVEF >40% in NG106 and proposed positioning of dapagliflozin

==> picture [454 x 266] intentionally omitted <==

aMeasure serum sodium, potassium and assess renal function before and after starting and after each dose increment. If eGFR is 30 to 45 ml/min/1.73 m[2] , consider lower doses or slower titration of ACEi or ARBs, MRAs, sacubitril valsartan and digoxin.[ b] It is assumed for the appraisal that recommendations for the HFpEF population comprise both the HFpEF and HFmrEF populations; those with HF and an LVEF >40%. Source : Adapted from NICE NG106.[59]

Abbreviations : ACEi: angiotensin-converting enzyme inhibitor; ARB: angiotensin-receptor blocker; BB: betablocker. HF: heart failure; eGFR: estimated glomerular filtration rate; HFpEF: heart failure with a preserved ejection fraction; HFrEF: heart failure with a reduced ejection fraction; MRA: mineralocorticoid-receptor antagonist; TA: technology assessment.

Comparators

As per NICE NG106 and clinical practice, the relevant comparator for dapagliflozin for the treatment of patients with HF and an LVEF >40% is placebo in addition to SoC (i.e., SoC alone). While patients with HF and an LVEF >40% may have varying multiple co-morbidities for which they are separately treated, due to the lack of disease-modifying treatment options routinely commissioned in UK clinical practice in this indication, SoC for these patients consists of loop diuretics for congestive symptoms and fluid retention.[59] The loop diuretics considered as SoC in UK clinical practice for the management of patients with HF and an LVEF >40% are furosemide and bumetanide.[4] While dapagliflozin is expected to be used in addition to SoC, including loop diuretics, other treatments are very much dependent on a patients’ underlying symptoms and comorbidities.

Treatment setting

As per TA679,[1] initiation of dapagliflozin in patients with HFrEF should be on the advice of a HF specialist, while monitoring is to be done by the most appropriate healthcare professional. It is proposed that treatment with dapagliflozin in patients with HF and an LVEF >40% could be initiated either in primary or secondary care, with confirmation of HF diagnosis by a specialist enabling the initiation of dapagliflozin in primary care without the need for further specialist advice. Given that patients may be discharged back to primary care following specialist diagnosis before a care plan is provided or treatment is initiated, it is both appropriate and optimal for the

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 26 of 134

patient that primary care physicians are able to initiate therapy autonomously. This is also critical to ensure that the management of patients already diagnosed with HF and an LVEF >40% who are managed in primary care is optimised, allowing dapagliflozin to be initiated at the earliest opportunity, ideally following proactive invitation for a treatment optimisation review or alternatively, where capacity is a limitation, during their routine check-up appointment without the need for a specific or extended appointment. In the case of both an incident and prevalent population with confirmed HF and an LVEF >40%, the requirement to seek additional specialist advice before treatment initiation would delay access and create additional resource constraints in both primary and secondary care amidst the large post-COVID back-logs still being experienced. As dapagliflozin is currently available across the primary and secondary care treatment settings for patients with T2DM,[5-7] CKD,[9] including those with co-morbid HF and an LVEF >40%, and HFrEF, [1] clinicians across care settings have considerable clinical experience with prescribing dapagliflozin. Therefore, the additional advice of a HF specialist seems unnecessary for the initiation of dapagliflozin after HF and an LVEF >40% has already been diagnosed, and delays could be costly in terms of morbidity and mortality.[72]

Finally, it should be noted that based on feedback from UK clinical experts consulted by AstraZeneca, the recommendation made by NICE in TA679 that dapagliflozin can be initiated to treat patients with HFrEF following the “advice of a HF specialist”,[1] has commonly been misinterpreted in UK clinical practice to be the same as “initiated by a specialist”, requiring an additional referral back to specialist services prior to the initiation of treatment. Misinterpretation of the NICE TA679 recommendation constitutes an additional barrier to access for many patients with HFrEF which AstraZeneca believes to contradict the intentions of the recommendations in TA679. This is especially worrisome considering the current post-COVID back-log for specialist review with estimates of over 275,000 people waiting for heart tests and treatment in September 2021 in England.[77] Therefore, empowering primary care physicians to initiate treatment with dapagliflozin after the appropriate diagnostic work-up is complete in patients with HF and an LVEF >40% is key to overcoming the barriers in access to care for these patients, including the inequalities associated with different levels of specialist provision across the country, as discussed below.

B.1.4. Equality considerations

Based on insights gathered by AstraZeneca in discussions with UK healthcare professionals, very few specialist centres review or actively manage patients with HF and an LVEF >40%. Most patients are managed in the primary care setting and, in some areas, there are no specialist-led or multidisciplinary clinics organised or commissioned to manage these patients.[10] Access to specialist care is even further restricted by the current post-COVID back-log.[77] Moreover, as dapagliflozin is already routinely commissioned and represents established clinical practice for treating T2DM, [5-7] CKD, [9] and HFrEF, [1] clinicians across both the primary and secondary care settings have considerable clinical experience in the prescribing of dapagliflozin. Therefore, enabling the initiation of dapagliflozin in both primary and secondary care for the treatment of patients with HF and a documented LVEF >40% would ensure consistent equality of access to efficacious therapies without relying on specialist care, which may not exist or have long waiting lists in some areas of the UK, and therefore would otherwise serve to drive unwarranted variation in care.

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 27 of 134

B.2. Clinical effectiveness

Summary of clinical effectiveness

• DELIVER was an international, multicentre, parallel-group, event-driven, double-blind RCT with a median follow-up of ** months which enrolled 6,263 patients and compared dapagliflozin (n=3,131) with placebo (n=3,132) for the treatment of patients with HF and an LVEF >40%, in addition to SoC.[76, 78]

• DELIVER is the first clinical trial in a patient population with HF and an LVEF >40% to include patients with improved LVEF (HFimpEF; prior LVEF ≤40% with improvement to >40% before study enrolment; ******** ***** of the full analysis set [FAS] population).[76, 79]

• Dapagliflozin in addition to SoC (referred to as dapagliflozin throughout Section B.2 for simplicity) was significantly superior to placebo in addition to SoC (referred to as placebo throughout Section B.2 for simplicity) in reducing the incidence of the primary composite endpoint of CV mortality or a HF event (hazard ratio [HR] 0.82; 95% confidence interval [CI]: 0.73, 0.92; p<0.001).[76 ]

• Pre-planned subgroup analysis of the primary efficacy outcomes was consistent across the prespecified subgroups, including those defined according to LVEF, with no attenuation in the highest LVEF group:[76]

  • Results were consistent across all LVEF groups: ≤49% (HR 0.87, 95% CI: 0.72, 1.04), 50–59% (HR 0.79, 95% CI: 0.65, 0.97), ≥60% (HR 0.78, 95% CI: 0.62, 0.98) *****

  • (p-value for interaction= ).[76,78 ]

  • Patients with HFimpEF experienced similar treatment benefits compared to those with HF and an LVEF consistently >40% (HR 0.74; 95% CI: 0.56, 0.97 versus HR *****

  • 0.84; 95% CI: 0.73, 0.95; p-value for interaction= ).[76, 79]

• Dapagliflozin was also superior to placebo in reducing the risk of the secondary composite endpoint of CV mortality and recurrent HF events (rate ratio [RR] 0.77; 95% CI: 0.67, 0.89; p<0.001), and in reducing recurrent HF events (RR 0.73; 95% CI: 0.62, 0.87; p=0.0003).[76, ] 78

• Both CV and all-cause mortality were reduced in patients treated with dapagliflozin compared with placebo although the differences were not statistically significant (HR 0.88; 95% CI: 0.74,1.05; p=0.1678 and HR 0.94; 95% CI: 0.83, 1.07; p=0.3425, respectively).[76,78 ]

• Dapagliflozin provided statistically significant improvements in symptom and physical function benefit as measured by KCCQ-TSS, -PLS, -CSS and -OSS at 8 months (mean difference in change from baseline 2.4,[76] 1.9, 2.3 and 2.1 points higher versus placebo; p<0.001, for all).[79]

• Based on UK clinical expert feedback, the baseline characteristics and background therapy profiles of the patients enrolled in the DELIVER trial were considered overall generalisable to those seen in UK clinical practice.[10]

B.2.1. Identification and selection of relevant studies

A systematic literature review (SLR) was conducted to identify relevant evidence of the clinical

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 28 of 134

efficacy and safety of treatments for patients with HF and an LVEF >40% in the form of RCTs.

The SLR was broad, and considered a range of possible treatments for patients with HF and an LVEF >40%, including SGLT2 inhibitors as well as loop diuretics, angiotensin-converting enzyme inhibitors (ACEi), angiotensin II receptor blockers (ARBs) and beta blockers. However, as described in B.1.1, placebo in addition to SoC represents the only comparator to dapagliflozin in this appraisal, and in UK clinical practice SoC comprises predominantly loop diuretics (e.g., furosemide or bumetanide). As such, only included studies conducted in patients receiving either dapagliflozin or loop diuretics were ultimately extracted for this submission, in line with the decision problem of this appraisal (see Appendix D).

The SLR was originally conducted in August 2018 and a subsequent update was conducted in June 2022, with adaptations made to the original SLR protocol to ensure alignment of the SLR with the decision problem of this appraisal. For instance, the original SLR considered studies in patients with HFrEF as well as observational study designs, which are not relevant to this submission.

In total, across the original SLR and the SLR update, 258 publications reporting 36 unique studies were included in the SLR. Of the 36 unique studies, 4 studies were identified in patients with HF and an LVEF >40% receiving either dapagliflozin or loop diuretics:

Two studies were identified that investigated dapagliflozin:

• DELIVER[80]

• PRESERVED-HF [81]

Two studies were identified that investigated loop diuretics:

• DROP-PIP [82]

• J-MELODIC [83]

The trials identified for dapagliflozin are discussed in more detail in B.2.2 below. The two studies investigating loop diuretics were not considered to provide more relevant evidence for SoC in comparison to the DELIVER trial (see Appendix D.4), and therefore are not considered further in this submission.

Full details on the SLR, including the detailed search terms, inclusion/exclusion criteria and a Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram detailing studies that were included and excluded at each stage of screening can be found in Appendix D.

B.2.2. List of relevant clinical effectiveness evidence

Two studies investigating the efficacy of dapagliflozin in patients with HF and an LVEF >40% were identified in the clinical SLR: DELIVER (N=6,263) and PRESERVED-HF (N=324).[76, 81] Of these, the clinical trial most relevant to this submission is DELIVER, the pivotal international, multicentre, parallel-group, event-driven, double-blind RCT for dapagliflozin in this indication that compared treatment with dapagliflozin in addition to SoC versus placebo in addition to SoC in patients with HF and an LVEF >40%.[76]

PRESERVED-HF is a smaller clinical trial that evaluated whether dapagliflozin in addition to SoC improved symptoms and physical limitations versus placebo in addition to SoC, as measured by

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 29 of 134

the KCCQ-CSS.[81] PRESERVED-HF was not used to populate the economic model for this submission due to its smaller sample size of 324 patients aged ≥19 with HF and an LVEF ≥45% (which differs from the population included in the DELIVER trial), its short duration of 12 weeks, and as it primarily evaluated HF disease-specific health status.[81] The results of this study support that dapagliflozin significantly improved patient-reported symptoms and physical limitations of patients with HF and an LVEF ≥45% compared with placebo, and was generally well tolerated.[81] Further details of the PRESERVED-HF trial are presented in B.2.11 for completeness. A brief summary of both trials is presented in Table 6.

Table 6: Clinical effectiveness evidence

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 30 of 134

Source : DELIVER CSR;[78] Solomon et al. (2022);[76] Solomon et al. (2022) – Supplementary Appendix.[85] Nassif et al. (2021);[81] ClinicalTrials.gov 2021 [NCT03030235].[84] Abbreviations : AE: adverse event; AKI: acute kidney injury; BP: blood pressure; CSS: Clinical Summary Score; CV: cardiovascular; DAE: adverse events leading to treatment discontinuation; eGFR: estimated glomerular filtration rate; EQ-5D-5L: EuroQol-5 Dimensions-5 Levels; HbA1c: haemoglobin A1c; HF: Heart failure; HHF: hospitalisation for heart failure; HFrEF: Heart failure with a reduced ejection fraction; KCCQ: Kansas City Cardiomyopathy Questionnaire; LVEF: left ventricular ejection fraction; MI: myocardial infarction; NT-proBNP: N- terminal pro B-type natriuretic peptide; NYHA: New York Heart Association; OSS: Overall Summary Score; QoL: quality of life; SAE: serious adverse event; TSS: Total Symptom Score; UHFV: urgent heart failure visit.

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 31 of 134

B.2.3. Summary of methodology of the relevant clinical

effectiveness evidence

B.2.3.1. Summary of trial methodology

DELIVER was an international, multicentre, parallel-group, event-driven, randomised, doubleblind Phase III study in patients with HF and an LVEF >40% and evidence of structural heart disease, evaluating the effect of dapagliflozin 10 mg versus placebo, given once daily in addition to background regional SoC therapy, including treatments for co-morbidities, in reducing the composite of CV mortality and HF events over a 28-month median follow-up period.[76] The methodology of DELIVER is summarised in Table 7 and Figure 4.

Table 7: Summary of trial methodology: DELIVER

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 32 of 134

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 33 of 134

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 34 of 134

aTypical symptoms associated with heart failure: breathlessness, orthopnoea, paroxysmal nocturnal dyspnoea, reduced exercise tolerance, fatigue, tiredness, increased time to recover after exercise, ankle swelling; Signs associated with HF: More specific: elevated jugular venous pressure, hepatojugular reflux, third heart sound (gallop rhythm), laterally displaced apical impulse; Less specific: weight gain (>2 kg/week), weight loss (in advanced HF), tissue wasting (cachexia), cardiac murmur, peripheral oedema (ankle, sacral, scrotal), pulmonary crepitations, reduced air entry and dullness to percussion at lung bases (pleural effusion), tachycardia, irregular pulse, tachypnoea, Cheyne-Stokes respiration, hepatomegaly, ascites, cold extremities, oliguria,narrow pulse pressure.[b] Left Atrial Enlargement defined by at least 1 of the following: left atrial (LA) width (diameter) ≥3.8 cm or LA length ≥5.0 cm or LA area ≥20 cm[2] or LA volume ≥55 mL or LA volume index ≥29 mL/m[2] . Left Ventricular Hypertrophy defined by septal thickness or posterior wall thickness ≥1.1 cm.

Source: DELIVER CSR.[78] Solomon et al. (2022);[76] Solomon et al. (2022) – Supplementary Appendix.[85] Abbreviations : AEs: adverse events; ARVC/D: arrhythmogenic right ventricular cardiomyopathy/dysplasia; BMI: body mass index; CABG: coronary artery bypass graft; COPD: chronic obstructive pulmonary disease; CKD-EPI: chronic kidney disease epidemiology; CV: cardiovascular; DAEs: adverse events leading to treatment discontinuation; DMC: data monitoring committee; EC: executive committee; ECG: echocardiogram; eGFR: estimated glomerular filtration rate; HF: heart failure; HHF: hospitalisation for heart failure; KCCQ: Kansas City Cardiomyopathy Questionnaire; LA: left atrial; LVEF: left ventricular ejection fraction; NT-proBNP: N-terminal pro B-type natriuretic peptide; NYHA: New York Heart Association; MI: myocardial infarction; MR: magnetic resonance; MRI: magnetic resonance imaging; PACD: primary analysis censoring date; PCI: percutaneous coronary intervention; SAEs: serious adverse events; SBP: systolic blood pressure; SGLT2: sodium-glucose cotransporter-2; SoC: standard of care; T2DM: Type 2 diabetes mellitus; TIA: transient ischemic attack; TSS: Total Symptom Score; UHFV: urgent heart failure visit.

Figure 4: DELIVER trial design

==> picture [460 x 130] intentionally omitted <==

Source: Solomon et al. (2022) – Supplementary Appendix.[85] Abbreviations : E: enrolment; HF: heart failure; IV: intravenous; LAE: left atrial enlargement; LVH: left ventricular hypertrophy; NYHA: New York Heart Association; PACD: primary analysis censoring date; R: randomisation: SCV: study closure visit.

B.2.3.2. Baseline characteristics and demographics

Patient characteristics at baseline in DELIVER are summarised in Table 8. Overall, 6,263 patients were randomised; 3,131 in the dapagliflozin group and 3,132 in the placebo group. In total, ***** of patients were female.[76, 78] The mean age was 71.7 years.[80] Demographic and other baseline patient characteristics were well balanced between treatment groups in the full study population.[76] Overall, ***** of patients had T2DM at baseline.[78] Median LVEF was *****, median NT-proBNP was 1,011.0 pg/mL, mean eGFR was 61.0 mL/min/1.73m[2] , and median systolic BP was ***** mmHg.[76,78, 80] Over 18% ********* of enrolled patients had HFimpEF, whereby their LVEF was ≤40% prior to study enrolment when it had increased to >40%.[76, 79] This population which was usually excluded from trials, tend to have worse outcomes than patients without a history of HF.[86] Also, outcomes tend to worsen for this patient population once a disease-

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 35 of 134

modifying treatment is discontinued.[86] Therefore, the DELIVER trial provides further evidence of the benefit that an SGLT2 inhibitor in addition to standard care may offer to those with residual symptoms of HF, thus HFimpEF.[86]

Most patients were diagnosed with HF <5 years before enrolment.[78, 80] A total of ***** patients ******* had a history of being hospitalised for HF prior to study enrolment.[78]

At randomisation, treatment of HF symptoms and co-morbidities was balanced between treatment groups.[76] In total, ***** ** ******** **** ******* **** * ********* **** ******** * **** ********* ***** **** * **** ******** ***** **** ** ************** *** ***** **** ** *** .[78] The high proportion of patients taking beta blockers, ACEi/ARB/ARNI and MRAs[80] which are not typically prescribed to treat HF with LVEF >40% is due to a combination of these being prescribed to treat comorbidities such as hypertension and the fact that the DELIVER trial contained over 18% of patients with HFimpEF, in whom clinical guidelines recommend to continue with treatments initiated to treat HFrEF even when their LVEF increases to >40%.[30] Compared with the cohort of real-world patients with HF and an LVEF >40% from the CPRD dataset,[62] the rates of treatment with these therapies was generally a little higher (DELIVER versus CPRD: ACEi: ***** versus *****; ARB: ***** versus *****; ARNI: **** versus ****; beta-blocker: ***** versus *****; MRA: ***** versus *****, respectively), but the same is true for the use of loop diuretics (DELIVER versus CPRD: ***** versus *****) which are the established SoC symptomatic treatments in these patients.[62, 78] This indicates that the DELIVER trial cohort represented a slightly better-treated group of patients compared with real-world clinical practice in the UK which is to be expected given the clinical trial setting.[10]

UK clinical experts consulted by AstraZeneca expressed confidence that the DELIVER trial characteristics at baseline were overall considered generalisable of the patients expected to receive dapagliflozin in UK clinical practice.[10]

Table 8: Characteristics of participants in DELIVER across treatment groups

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 36 of 134

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648]

© AstraZeneca (2022). All rights reserved

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 38 of 134

Source: Solomon et al. (2022);[76] Solomon et al. (2022);[80] Solomon et al. (2022) – Supplementary Appendix;[85] Vaduganathan et al. (2020);[87] Cunningham et al. (2022);[88] Ostrominski et al. (2022);[89] DELIVER CSR.[78] Abbreviations : ACEi: angiotensin converting enzyme inhibitor; ARB: angiotensin receptor blocker; ARNI:

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 39 of 134

angiotensin receptor neprilysin inhibitor; CV: cardiovascular; ECG: echocardiogram; eGFR: estimated glomerular filtration rate; HF: heart failure; HHF: hospitalisation for heart failure; max: maximum; min: minimum; LVEF: left ventricular ejection fraction; MRA: mineralocorticoid receptor antagonist; N: number of patients in treatment group; n: number of patients included in analysis; NT-proBNP: N-terminal pro-brain natriuretic peptide; NYHA: New York Heart Association; Q1: first quartile; Q3: third quartile; T2DM: type 2 diabetes mellitus.

B.2.4. Statistical analysis and definition of study groups in the

relevant clinical effectiveness evidence

An overview of the patient population analysis sets and details of the statistical analysis conducted in DELIVER are provided below.

B.2.4.1. Definitions of patient population analysis sets

Full analysis set (FAS): All patients who were randomised to treatment were included in the FAS, irrespective of their protocol adherence and continued participation in the study. Patients were analysed according to their randomised treatment assignment, irrespective of the treatment actually received. The FAS was considered the primary analysis set for the intention-to-treat (ITT) analysis of primary and secondary variables and for the exploratory efficacy variables. A subset of the FAS consisting of patients with a baseline LVEF <60% (i.e., the subpopulation with LVEF <60%) was analysed separately as part of the confirmatory statistical testing procedure.[85]

Safety analysis set (SAS): All randomised patients who received at least one dose of treatment were included in the SAS.[85]

B.2.4.2. Statistical analysis

A summary of the statistical analysis in DELIVER is provided in Table 9.

Table 9: Summary of statistical analyses in DELIVER

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 40 of 134

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 41 of 134

Source: Solomon et al. (2022) – Supplementary Appendix;[85] DELIVER CSR.[78] Abbreviations : CI: confidence interval; CV: cardiovascular; FAS: full analysis set; HF: heart failure; HHF: hospitalisation for heart failure; HR: hazard ratio; LVEF: left ventricular ejection fraction; LWYY: Lin Wei Yang Ying; NT-proBNP: N-terminal pro-brain natriuretic peptide; PACD: primary analysis censoring date: SoC: standard of care; T2DM: type 2 diabetes mellitus.

Figure 5: Testing procedure for DELIVER

==> picture [452 x 129] intentionally omitted <==

==> picture [452 x 128] intentionally omitted <==

Source: Solomon et al. (2022) – Supplementary Appendix.[85] Abbreviations : CV: cardiovascular; HF: heart failure; KCCQ: Kansas City Cardiomyopathy Questionnaire; LVEF: left ventricular ejection fraction; TSS: Total Symptom Score.

B.2.4.3. Participant flow in the relevant randomised controlled trials

Participant flow in DELIVER is summarised in Figure 6.

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 42 of 134

Figure 6: Patient disposition in DELIVER

==> picture [371 x 433] intentionally omitted <==

Source: DELIVER CSR.[78] Abbreviations : DKA: diabetic ketoacidosis; IP: investigational product; PACD: primary analysis censoring date.

B.2.5. Critical appraisal of the relevant clinical effectiveness

evidence

A critical appraisal of the DELIVER trial is provided in Table 10.

Table 10: Critical appraisal of DELIVER

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 43 of 134

B.2.6. Clinical effectiveness results of the relevant studies:

DELIVER

B.2.6.1. Primary efficacy outcome: composite of CV mortality and HF events

Dapagliflozin statistically significantly reduced the risk of the primary composite endpoint of CV mortality and HF events by 18% compared with placebo[76]

Dapagliflozin was statistically significantly superior to placebo in reducing the incidence of the primary composite endpoint of CV mortality or a HF event (HR 0.82; 95% CI: 0.73, 0.92; p<0.001; Figure 7).[76] Over a median duration of follow-up of 2.3 years, there were 512 and 610 patients with CV mortality or a HF event in the dapagliflozin and placebo groups, respectively, corresponding to event rates per 100 patient-years of 7.8 and 9.6, respectively.[76] This meant that ** fewer patients experienced either CV mortality or a HF event on treatment with dapagliflozin compared with placebo.[76,78] Of a total of 1,122 patients with a composite event, 300 patients had CV mortality as their first event.[87]

A Kaplan-Meier (KM) analysis of the composite of CV mortality or an HF event is presented in Figure 7.[76] The curves diverged early and the separation was maintained throughout the study.

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 44 of 134

Figure 7: KM plot of the primary composite endpoint (CV mortality and HF events) in DELIVER

==> picture [445 x 314] intentionally omitted <==

Source: Solomon et al. (2022).[76] Abbreviations : CI: confidence interval; CV: cardiovascular; Dapa: dapagliflozin; D: dapa 10mg; FAS: full analysis set; HF: heart failure; HR: hazard ratio; KM: Kaplan-Meier; N: number of patients; P: placebo.

All components of the primary composite endpoint ************ *********** to the treatment effect (Figure 8).[78]

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 45 of 134

Figure 8: Forest plot of the primary composite endpoint (CV mortality and HF events) and the individual components in DELIVER[a]

==> picture [452 x 254] intentionally omitted <==

Source: DELIVER CSR.[78]

Abbreviations : CI: confidence interval; CV: cardiovascular; Dapa: dapagliflozin; FAS: full analysis set; HF: heart failure; HHF: hospitalisation for heart failure; HR: hazard ratio; N: number of patients in treatment group; T2DM: type 2 diabetes mellitus; UHFV: urgent heart failure visit.

Sensitivity analysis of primary outcome

Results of the sensitivity analysis, ** ***** ****** *********** ** ************* ***** ** ****** ****

********** ** ****** *** ******** ** ******** ******* **** ********** **** ***** ** *** **** ******** .[78]

Results of the COVID-19 sensitivity analysis, in which patients were censored at the onset date of the first AE associated with COVID-19 infection, were also consistent with those of the main analysis.[76]

B.2.6.2. Secondary efficacy outcomes

Composite of CV mortality and recurrent HF events

Dapagliflozin statistically significantly reduced the risk of the secondary composite endpoint of CV mortality and recurrent HF events by 23% compared with placebo[76]

Dapagliflozin was statistically significantly superior to placebo in reducing the incidence of the composite of total (first and recurrent/ repeat) HF events and CV mortality (RR 0.77; 95% CI: 0.67, 0.89; p<0.001; Table 11). There were 815 and 1,057 events of the composite endpoint in the dapagliflozin and placebo groups, respectively, corresponding to event rates per 100 patientyears of 11.8 and 15.3, respectively.[76] Dapagliflozin provided a statistically significant reduction versus placebo in the incidence of recurrent HF events (RR 0.73; 95% CI: 0.62, 0.87; p=0.0003).[78] Dapagliflozin reduced the incidence of CV mortality although the difference was not statistically significant (HR 0.88; 95% CI: 0.74, 1.05; p=0.1678).[76,78]

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 46 of 134

Table 11: Analysis of the composite endpoint of CV mortality and recurrent HF events in DELIVER

Source: Solomon et al. (2022);[76] DELIVER CSR.[78] Abbreviations : CI: confidence interval; CV: cardiovascular; Dapa: dapagliflozin; HF: heart failure; HHF: hospitalisation for heart failure; LWYY: Lin Wei Yang Ying; N: number of patients in treatment group; RR: rate ratio; T2DM: type 2 diabetes mellitus; UHFV: urgent heart failure visit.

***** patients in the dapagliflozin group had ≥ 1 and ≥ 2 of the events included in the composite endpoint versus the placebo group (Table 12).[78]

Table 12: Summary of HF events and CV mortality – number of events per patient in DELIVER

Source: DELIVER CSR.[78]

Abbreviations : CV: cardiovascular; Dapa: dapagliflozin; HF: heart failure; HHF: hospitalisation for health failure; N: number of patients in treatment group; UHFV: urgent heart failure visit.

Results of the COVID-19 sensitivity analysis in which patients were censored at the onset date of the first AE associated with COVID-19 infection, were consistent with those of the main analysis.[85]

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 47 of 134

CV mortality

CV mortality was reduced in patients treated with dapagliflozin compared with placebo although the difference was not statistically significant[76, 78]

There were fewer CV deaths in the dapagliflozin group compared with the placebo group (231 versus 261), not reaching statistical significance (HR 0.88; 95% CI: 0.74, 1.05; p=0.1678); Figure 9).[76, 78]

Figure 9: KM plot of CV mortality in DELIVER

==> picture [452 x 315] intentionally omitted <==

Source: Solomon et al . (2022).[76] Abbreviations : CI: confidence interval; CV: cardiovascular; Dapa: dapagliflozin; D: dapa 10mg; HR: hazard ratio; KM: Kaplan-Meier; N: number of patients; P: placebo.

Mortality from any cause

All-cause mortality was reduced in patients treated with dapagliflozin compared with placebo although the difference was not statistically significant[76,78]

There were fewer deaths from any cause in the dapagliflozin group compared with the placebo group (497 versus 526 not reaching statistical significance (HR 0.94; 95% CI 0.83, 1.07; p= 0.3425; Figure 10).[76,78]

The hierarchical testing sequence stopped before the endpoint of time to death from any cause could be assessed. Hence, the analysis of this endpoint was not conducted as part of the confirmatory testing sequence.

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 48 of 134

Figure 10: KM plot of all-cause mortality in DELIVER

==> picture [443 x 305] intentionally omitted <==

Source: Solomon et al. (2022).[76] Abbreviations : CI: confidence interval; Dapa: dapagliflozin; D: dapa 10mg; HR: hazard ratio; KM: Kaplan-Meier; N: number of patients; P: placebo.

Adjudicated death causes are presented in Table 13. The most common adjudicated cause of mortality was CV death.

Table 13: Summary of adjudicated death classification in DELIVER[a]

Source: DELIVER CSR;[78] Solomon et al. (2022);[76] Vaduganathan et al. (2022).[87] Abbreviations : CV: cardiovascular; Dapa: dapagliflozin; N: number of patients in treatment group; PACD: primary analysis censoring date.

Change from baseline in Total Symptom, Clinical Summary, Overall Summary and Physical Limitation Scores of KCCQ[93]

Dapagliflozin provided significant patient-reported symptom benefits and physical limitation improvement versus placebo

At baseline, KCCQ data were available for ***** patients (***** of the overall trial population) with

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 49 of 134

a median KCCQ-TSS of **** **** **** * ****).[78, 93]

Dapagliflozin provided statistically significant improvements versus placebo in mean KCCQ-TSS, -PLS, -CSS and -OSS at 8 months (2.4, 1.9, 2.3 and 2.1 points higher versus placebo; p<0.001, **** ******** ** * ***** *** ********* **** **** for all).[76, 78, 93] Improvements .[78, 93] Mean changes over time in KCCQ-TSS, -PLS, -CSS and -OSS are presented in Table 14 and Figure 11.

Table 14: Change in KCCQ parameters at Month 1, Month 4 and Month 8

aTSS quantifies the symptom frequency and severity, PLS evaluates the physical function, CSS includes the symptoms and physical function domains, and OSS summarises all key domains (TSS, physical function, quality of life and social function). Scores are transformed to a range of 0–100, in which higher scores reflect better health status. Source: Solomon et al. (2022)[76] ; DELIVER CSR;[78] AstraZeneca UK Ltd. Data on File.[93] Abbreviations: CSS, Clinical Summary Score; OSS: Overall Summary score; PLS: Physical Limitation Score; TSS: Total symptom score.

Figure 11: Mean changes in KCCQ domains over time by treatment allocation[a,b]

==> picture [450 x 321] intentionally omitted <==

aIndividual graphs for KCCQ domain including KCCQ-TSS (Panel A), KCCQ-PLS (Panel B), KCCQ-CSS (Panel C) and KCCQ-OSS (Panel D);[b] TSS quantifies the symptom frequency and severity, PLS evaluates the physical function, CSS includes the symptoms and physical function domains, and OSS summarises all key domains

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 50 of 134

(TSS, physical function, quality of life and social function). Scores are transformed to a range of 0–100, in which higher scores reflect better health status.

Source: AstraZeneca UK Ltd. Data on File.[93]

Abbreviations: CSS: Clinical Summary Score; Dapa: dapagliflozin; OSS, Overall Summary Score; PLS: Physical Limitation Score; TSS: Total Symptom Score; wk: week.

The results of the responder analysis showed that * ******* ********** ** ******** in the dapagliflozin ************ * * ****** group compared with the placebo group by , which is the clinically significant improvement threshold.[93] A ******* ********** of patients in the dapagliflozin group compared with the placebo group had at least small (), moderate (), and large (**) ************ in KCCQTSS, PLS, CSS and OSS with all comparisons being statistically significant, except 15 point or greater improvement in KCCQ-TSS and 5 point or greater improvement in OSS; (Figure 12).[93]

Figure 12: Responder analyses of clinically meaningful change in KCCQ domains at 8 months with dapagliflozin versus placebo[a]

==> picture [471 x 342] intentionally omitted <==

aResponder analyses of clinically meaningful changes in KCCQ-TSS (Panel A), KCCQ-PLS (Panel B), KCCQCSS (Panel C) and KCCQ-OSS (Panel D). Source: AstraZeneca UK Ltd. Data on File.[93]

Abbreviations: CSS: Clinical Summary Score; Dapa: dapagliflozin; OR, odds ratio; OSS: Overall Summary Score; PLS: Physical Limitation Score; TSS: Total Symptom Score.

B.2.6.3. Exploratory endpoints

Exploratory outcomes, including time to first occurrence of hospitalisation from any cause, proportion of patients with worsened NYHA class from baseline to 8 months, and EQ-5D-5L analysis are presented in detail below. Other exploratory outcomes, including change in eGFR, body weight and systolic blood pressure from baseline, are presented in Appendix M, while the KCCQ clinical and overall scores, and domains are presented in Section B.2.6.2.

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 51 of 134

Hospitalisation from any cause

**All-cause hospitalisation was reduced in patients treated with dapagliflozin compared with placebo although the difference was *** ************* *************

Occurrence of hospitalisation from any cause is presented in Table 15. In the dapagliflozin group, ***** of patients ****** patients) had an occurrence of hospitalisation from any cause compared with ***** ****** patients) in the placebo group.[78]

Table 15: Analysis of first occurrence of hospitalisation from any cause in DELIVER

Source: DELIVER CSR.[78]

Abbreviations : CI: Confidence interval; HR: hazard ratio; N: Number of patients in treatment group; T2DM: type 2 diabetes mellitus.

Proportion of patients with worsened NYHA class from baseline to 8 months

Dapagliflozin provided early (4 weeks) and sustained net improvement in NYHA functional class through to Week 32 versus placebo[89]

The effect of dapagliflozin versus placebo on NYHA functional class over time is presented in Table 16.[89] Any improvements in NYHA class were experienced more often by patients on dapagliflozin than those on placebo by Week 4 (11.0% versus 8.7%), Week 16 (15.8% versus 13.2%) and Week 32 (18.7% versus 14.5%).[89] Also, dapagliflozin, at Weeks 4, 16 and 32, was associated with a lower likelihood of NYHA class deterioration.[89] There was a higher likelihood in patients treated with dapagliflozin versus placebo to experience an improvement rather than a worsening in NYHA class at Week 4 (OR 1.37, 95% CI: 1.17–1.60; p<0.001), Week 16 (OR 1.20, 95% CI: 1.05–1.38; p=0.007) through to Week 32 (OR 1.32, 95% CI: 1.16–1.51; p<0.001).[89]

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 52 of 134

Table 16: Effect of dapagliflozin versus placebo on NYHA functional class over time[a]

==> picture [450 x 266] intentionally omitted <==

aValues displayed as percentage of participants with any improvement or deterioration in NYHA functional class. Odds ratios (OR) represent OR for improvement rather than worsening NYHA functional class at each timepoint. Source: Ostrominski et al . (2022).[89] Abbreviations : CI: confidence interval; NYHA: New York Heart Association; OR: odds ratio.

EQ-5D-5L

In DELIVER, ********* ** *****, as estimated from EQ-5D-5L data **** ******** ** **** ***** *****[**] ************ *** ****** ** *** ************* ********* *** *** ******* ******** **** ****** *** *** ************* ************[**] ***** ******* *** ** **** **** ***** ******** **** * *********** ******** ***** ********** **** ** ***** ***** ** *** ********* *** ******** ****** ** *** ** ********** ** ******* ******** ********[**]

B.2.7. Subgroup analysis

Pre-planned subgroup analyses of the primary efficacy outcomes in DELIVER are presented in Figure 13. The benefit of dapagliflozin on the primary composite endpoint was consistent across the key prespecified subgroups, including age, sex and those defined by baseline LVEF (≤49%, 50%–59%, ≥60%), with no attenuation of treatment observed in patients with greater LVEF of 50%–59% and ≥60% (Figure 13).[76]

Baseline characteristics of patients in the DELIVER trial are described in Section B.2.3.2 with statistical methods summarised in B.2.4.

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 53 of 134

Figure 13: Forest plot of the primary composite endpoint (CV mortality and HF events) by subgroups in DELIVER

==> picture [452 x 529] intentionally omitted <==

The primary outcome was a composite of worsening heart failure, which was defined as either an unplanned HHF or an UHFV, or cardiovascular mortality. Race was reported by the investigators. The size of the boxes is proportional to the number of patients in the subgroup, and arrows on the CI bars indicate that the upper or lower boundary of the confidence interval is off the scale. One patient in the placebo group who had NYHA class I disease at baseline was not included in the analysis of NYHA class at enrolment. Source: Solomon et al. (2022).[76] Abbreviations : CI: confidence interval; CV: cardiovascular; Dapa: dapagliflozin; ECG: echocardiogram; eGFR: estimated glomerular filtration rate; HF: heart failure; HHF: hospitalisation for heart failure; HR: hazard ratio; LVEF: left ventricular ejection fraction; N: number of patients in treatment group; N#: number of patients in the subgroup; n: number of patients with event; NT-proBNP: N-terminal pro b-type natriuretic peptide; NYHA: New York Heart Association; SBP: systolic blood pressure.

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 54 of 134

Patients with HFimpEF

In this previously unstudied patient subpopulation, a total of ***** ******** ******* of patients enrolled in DELIVER had HFimpEF (prior LVEF ≤40%) and a prespecified analysis was conducted to investigate the efficacy of dapagliflozin in this subgroup of patients.[78, 79] Overall, event rates were similar in those with HFimpEF compared with patients with HF and an LVEF consistently >40%.[76] Treatment with dapagliflozin reduced the primary composite outcome in participants with HFimpEF (HR 0.74, 95% CI: 0.56, 0.97, *******) to a similar extent as in those with HF and an LVEF consistently over 40% (HR 0.84, 95% CI: 0.73, 0.95, *******; p- interaction=*****) (Table 17).[76, 79 ] Similarly, ** ************* *********** *********** was observed

between those with HFimpEF and those with HF and an LVEF >40% prior to enrolment in all other secondary outcomes.[79]

Table 17: Primary composite endpoint (CV mortality and HF events) in patients with HFimpEF compared with those with HF and an LVEF consistently >40%

a Per 100 patient years. Source: AstraZeneca UK Ltd. Data on File;[79 ] Solomon et al. (2022).[76] Abbreviations: CI: confidence interval; CV: cardiovascular; Dapa: dapagliflozin; HF: heart failure; HR: hazard ratio; LVEF: left ventricular ejection fraction; N: number of patients in treatment group; py: patient year.

B.2.8. Prespecified analysis: estimated benefits with long-term

treatment with dapagliflozin

The following section provides an overview of a prespecified analysis conducted to estimate the long-term benefits of treatment with dapagliflozin in patients with HF and an LVEF >40%.[87]

B.2.8.1. Objectives

To investigate the expected long-term benefits of dapagliflozin in patients with HF and an LVEF >40% beyond the timelines of the DELIVER trial.[87]

B.2.8.2. Summary of methodology

In this prespecified analysis, validated nonparametric age-based methods were used to extrapolate potential gains in event-free survival from the primary endpoint (composite of CV mortality and HF events) from the long-term use of dapagliflozin in patients with HF and an LVEF >40%.[87] Projected event-free survival using age at randomisation instead of time from randomisation as the time horizon, was estimated for every year between the ages of 55 and 85 years.[87] For each year of age in both treatment arms, the residual life span free from the primary endpoint was estimated based on area under the survival curve, up to a maximum of 100 years.[87]

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 55 of 134

B.2.8.3. Summary of results

A total of 1,122 events of the primary endpoint occurred over the median follow-up period of 2.3 years with an incidence rate of 8.7 (95% CI: 8.2, 9.2) per 100 patient-years.[87] Treatment gains in event-free survival from the primary endpoint for dapagliflozin versus placebo were 2.0 years (95% CI: -0.6, 4.6; p=0.14) at age 55 years, 2.3 years (95% CI: 0.9, 3.8; p=0.002) years at age 65 years (p=0.002), and 1.2 years (95% CI: -0.1, 2.4; p=0.063) at age 75 years.[87]

At age 65 years, event-free survival was greater with dapagliflozin than placebo across relevant subgroups examined. Treatment with dapagliflozin may extend event-free survival by 1.2 to 2.3 years for patients aged 55 years and older with HF and an LVEF >40%.[87]

B.2.9. Meta-analysis

DELIVER was not powered to test the effect of dapagliflozin on the individual components of the composite primary outcome or important secondary outcomes.[76] In order to examine the effects of dapagliflozin on key clinical outcomes in patients with HF across the full continuum of LVEF, a pooled analysis of the DELIVER and DAPA-HF trials was planned prior to DELIVER database lock, then conducted and published recently.[94] The population evaluated in this analysis is aligned with the anticipated update to the existing marketing authorisation for dapagliflozin for the treatment of chronic HF with LVEF ≤40%, ******* *** *********** ** **** ** ******** ** ** ******* *** *

Summary of the pooled analysis

• The pooled analysis (N=11,007) was a patient-level pooled meta-analysis of DELIVER and DAPA-HF (the pivotal RCT for dapagliflozin in addition to SoC versus placebo in addition to SoC in patients with HF and an LVEF ≤40%), and thus covered the full population of patients with HF irrespective of LVEF[94 ]

• In the pooled analysis of patients with HF irrespective of LVEF, dapagliflozin compared with placebo significantly:[94]

  • Reduced the risk of mortality from CV causes (HR 0.86, 95% CI: 0.76, 0.97; p=0.01)

  • Reduced the risk of mortality from any causes (HR 0.90, 95% CI: 0.82, 0.99; p=0.03)

  • Reduced total hospital admissions for HF (RR 0.71, 95% CI: 0.65, 0.78; p<0.001)

  • Reduced major adverse cardiovascular events (MACE; HR 0.90, 95% CI: 0.81, 1.00; p=0.045)

• The results of this pooled analysis therefore support the benefits of dapagliflozin in the full HF population, irrespective of LVEF[94]

B.2.9.1. Summary of methodology

The pooled analysis was a patient-level pooled meta-analysis of DELIVER and DAPA-HF to evaluate the efficacy of dapagliflozin across the full continuum of LVEF in patients with HF.[94] The pooled analysis was prespecified to examine the effect of treatment with dapagliflozin on endpoints which neither trial was sufficiently powered for. While both trials enrolled patients with diagnosed HF, functional limitation, and elevated natriuretic peptides, the main difference between the trials was that DAPA-HF enrolled patients with HF and an LVEF ≤40% whereas

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 56 of 134

DELIVER enrolled those with HF and an LVEF >40%.[94] In each trial, patients were randomised to receive either dapagliflozin 10mg once daily, or a matching placebo, in addition to SoC.[94] Both trials were event driven and used the primary composite endpoint of CV mortality, and HF events.[94]

The pooled analysis included the following endpoints:[94]

• CV mortality;

• Mortality from any cause;

• Total hospital admissions for HF;

• Composite of CV mortality, MI or stroke (“major adverse cardiovascular events” [MACE]).

B.2.9.2. Results

A total of 11,007 participants were included in the analysis.[94] Of these, 4,744 had HF and an LVEF ≤40% and 6,263 had HF and an LVEF >40%, with 5,503 randomised to placebo and 5,504 to dapagliflozin. The median LVEF was 44% (IQR: 34, 55).[94]

Baseline characteristics

Baseline characteristics for the patients included in the pooled analysis are presented in Table 18. Patients with a higher LVEF were older, more likely to be female, had higher blood pressure and a higher BMI than those with a lower LVEF.[94] It was more common for those with higher LVEF to have had a history of hypertension and AF than those with lower LVEF.[94] On the contrary, it was less common for those with higher LVEF to have had a history of MI than those with lower LVEF.[94] There was a lower proportion of patients in NYHA class III/IV amongst patients with higher LVEF. KCCQ scores were better in patients with lower LVEF than in those with higher LVEF.[94] NT-proBNP and eGFR levels were lower amongst patients with higher LVEF, as was the use of ACEis, ARBs, sacubitril/valsartan, beta-blockers, MRAs and ICDs.[94]

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 57 of 134

Table 18: Baseline characteristics of the patients included in the pooled analysis of DELIVER and DAPA-HF by LVEF category

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 58 of 134

Source : Jhund et al. (2022).[94] Abbreviations: ACE: angiotensin-converting enzyme; ARB: angiotensin receptor blocker; ARNI: angiotensin receptor neprilysin inhibitor; AF: atrial fibrillation; BMI: body mass index; CRT-D: cardiac resynchronisation therapy – defibrillator; CRT-P: cardiac resynchronisation therapy – pacemaker; eGFR: estimated glomerular filtration rate; HHF: hospitalisation for heart failure; ICD: implantable cardioverter defibrillator; KCCQ-TSS: Kansas City Cardiomyopathy Questionnaire – Total Symptom Score; LVEF: left ventricular ejection fraction; MI: myocardial infarction; MRA: mineralocorticoid receptor antagonist; NT-proBNP: N-terminal pro-brain natriuretic peptide; NYHA: New York Heart Association; T2DM: Type 2 diabetes mellitus

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 59 of 134

Outcomes

In the pooled analysis of DELIVER and DAPA-HF, dapagliflozin significantly reduced the risk of mortality and HHF versus placebo for patients with HF irrespective of LVEF[94]

The rate of each prespecified outcome was lower in the dapagliflozin group compared with the placebo group as shown on Figure 14.[94] Dapagliflozin compared with placebo reduced the risk of mortality from CV causes (HR 0.86, 95% CI: 0.76, 0.97; p=0.01), the risk of mortality from any cause (HR 0.90, 95% CI: 0.82, 0.99; p=0.03), total HHF (RR 0.71, 95% CI: 0.65, 0.78; p<0.001), and MACE (HR 0.90, 95% CI: 0.81, 1.00; p=0.045).[94]

Figure 14: Effect of dapagliflozin on key clinical outcomes in pooled DAPA-HF and DELIVER dataset

==> picture [452 x 245] intentionally omitted <==

a–fIncidence of: death from CV causes (a); death from all causes (b); the total number of hospital admissions for HF (c); time to first hospital admission for HF (d); death from CV causes, MI or stroke (e); and death from CV causes or hospital admission for HF (f), according to randomised therapy. Participants randomised to dapagliflozin are shown in blue and those randomised to placebo in red. All figures are Kaplan–Meier curves with an HR and 95% CI estimated from Cox’s model with two-sided p-values except for the total number of hospital admissions for HF, which was plotted using the Gosh and Lin method accounting for death from CV causes (the RR is estimated from the joint frailty model with a two-sided p-value). No adjustment for multiple comparisons was made. NNT indicates the number of patients who need to be treated over the median duration of follow-up to prevent one event (of the type in each panel). An NNT could not be calculated for the total number of hospital admissions for HF because this was an episode-based rather than a patient-based analysis (that is, patients may have had more than one hospital admission). ARRs and NNTs are shown with a 95% CI. Source : Jhund et al. (2022).[94]

Abbreviations : ACE: angiotensin-converting enzyme; ARB: angiotensin receptor blocker; ARNI: angiotensin receptor neprilysin inhibitor; ARR: absolute risk reduction; CI: confidence interval; CV: cardiovascular; ICD: implantable cardioverter defibrillator; HF: heart failure; HR: hazard ratio; MI: myocardial infarction; MRA: mineralocorticoid receptor antagonist; NNT: number needed to treat; RR: rate ratio.

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 60 of 134

Figure 15: Effect of randomised treatment on CV mortality according to the prespecified subgroups[a]

==> picture [452 x 329] intentionally omitted <==

aEstimates are HRs with error bars representing 95% CIs from Cox’s model and a two-sided p-value for interaction from Wald’s test of Cox’s model. No adjustment for multiple comparisons was made.[a] Not a prespecified subgroup. Source : Jhund et al. (2022).[94]

Abbreviations : BMI: body mass index; CI: confidence interval; CV: cardiovascular; Dapa: dapagliflozin; ECG: echocardiogram; eGFR: estimated glomerular filtration rate; HF: heart failure; HR: hazard ratio; LVEF: left ventricular ejection fraction; N: number of patients in treatment group; N#: number of patients in the subgroup; n: number of patients with event; NT-proBNP: N-terminal pro b-type natriuretic peptide; NYHA: New York Heart Association; SBP: systolic blood pressure; T2DM: type 2 diabetes mellitus.

B.2.10. Indirect and mixed treatment comparisons

Indirect and mixed treatment comparisons were not required as the relevant comparator, namely placebo in addition to SoC for patients with HF and an LVEF >40%, was included in the pivotal RCT DELIVER.[76]

B.2.11. PRESERVED-HF trial outcome summary

PRESERVED-HF supports that dapagliflozin significantly improved patient-reported symptoms and physical limitations in patients with HF and an LVEF ≥45% as well as being generally well tolerated.[81] Although PRESERVED-HF was not used to populate the economic model due to the reasons presented in Section B.2.2, it is presented for completeness as the outcomes observed in the trial were consistent with those from DELIVER.

B.2.11.1. Summary of trial methodology

PRESERVED-HF was a randomised, double-blind, placebo-controlled, multicentre Phase IV

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 61 of 134

study in patients with HF and an LVEF ≥45%, evaluating the effect of dapagliflozin 10 mg versus placebo, given once daily in addition to background regional SoC, including treatments for comorbidities, on disease-specific biomarkers (NT-proBNP and BNP), symptoms, health status, and QoL.[81] The methodology of PRESERVED-HF is summarised in Table 19.

Table 19: Summary of trial methodology: PRESERVED-HF

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 62 of 134

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 63 of 134

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 64 of 134

a For patients with permanent atrial fibrillation inclusion thresholds will be BNP ≥ 100 pg/mL or NT-proBNP ≥ 375 pg/mL. [b] For patients with permanent atrial fibrillation exclusion thresholds will be BNP<100 pg/mL and NTproBNP<375pg/mL.

Sources: Nassif et al. (2021);[81] ClinicalTrial.gov 2021 [NCT03030235].[84]

Abbreviations : ACEi: angiotensin-converting enzyme inhibitor; AEs: adverse events; ARB: angiotensin receptor blockers; BNP: B-type natriuretic peptide; BP: blood pressure; CABG: coronary artery bypass grafting; CCB: calcium channel blockers; COPD : chronic obstructive pulmonary disease; CRT: cardiac resynchronisation therapy; CSS: Clinical Summary Score; CV: cardiovascular; DKA: diabetic ketoacidosis; EF: ejection fraction; (e)GFR: (estimated) glomerular filtration rate; HF: heart failure; HHF: hospitalisation for heart failure; HOCM: hypertrophic obstructive cardiomyopathy; IUD: intrauterine device; KCCQ: Kansas City Cardiomyopathy Questionnaire; LA: left atrial; LVEDP: left ventricular end diastolic pressure; LVEF: left ventricular ejection fraction; LVH: left ventricular hypertrophy; MDRD: modification of diet in renal disease; MI: myocardial infarction; N: number of patients in treatment group; NYHA: New York Heart Association; NT-proBNP: N-terminal pro B-type natriuretic peptide; OSS: Overall Summary Score; PCI: percutaneous coronary intervention; SAEs: serious adverse events; SBP: systolic blood pressure; SGLT2: sodium-glucose co-transporter-2; SoC: standard of care; (N)STEMI: (Non) ST-elevation myocardial infarction.

B.2.11.2. Baseline characteristics

Patient characteristics at baseline for patients included in PRESERVED-HF are summarised in Table 20. Overall, baseline characteristics were well balanced between the two groups.

Table 20: Characteristics of participants in PRESERVED-HF across treatment groups

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 65 of 134

Values are shown as absolute numbers (percentages) and median (IQR) or mean ± sd. Sources: Nassif et al. (2021).[81] Abbreviations : 6MWT: 6-minute walk test; ACEi: angiotensin-converting enzyme inhibitor; AF: atrial fibrillation; ARB: angiotensin receptor blockers; ARNI: angiotensin receptor neprilysin inhibitor; BMI: body mass index; CSS: Clinical Summary Score; eGFR: estimated glomerular filtration rate; HF: heart failure; HHF: hospitalisation for heart failure; IQR: interquartile range; KCCQ: Kansas City Cardiomyopathy Questionnaire; MRA: mineralocorticoid receptor antagonists; NT-proBNP: N-terminal pro B-type natriuretic peptide; NYHA: New York Heart Association; OSS: Overall Summary Score; T2DM: Type 2 diabetes mellitus.

B.2.11.3. Summary of primary and secondary efficacy outcomes

Primary endpoint: KCCQ-CS

At 12 weeks, data for the primary endpoint was available for 304 (93.8%) patients with 152 (93.8%) patients in the dapagliflozin and placebo groups, respectively. Dapagliflozin was associated with an improvement in KCCQ-CSS (difference in mean change from baseline, 5.8 points [95% CI: 2.3, 9.2], p=0.001; Table 21), which was due to improvements in symptoms (difference in mean change from baseline for KCCQ-TSS, 5.8 points [95% CI: 2.0, 9.6], p=0.003) and physical limitations (difference in mean change from baseline for KCCQ-PLS, 5.3 points [95% CI: 0.7, 10.0], p=0.026; Figure 16). Consistent subgroup results were obtained (Figure 16).[81]

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 66 of 134

Secondary endpoints: 6-minute walk test (6MWT), KCCQ-OS, clinically meaningful changes in KCCQ-CS and KCCQ-OS, and changes in weight, natriuretic peptides, glycated hemoglobin and systolic blood pressure

At 12 weeks, data for the secondary endpoint of 6MWT were available for 291 (89.8%) patients with 148 (91.4%) patients in the dapagliflozin group and 143 (88.3%) in the placebo group.[81] An improvement in 6MWT in the dapagliflozin group was observed (effect size 20.1m [95% CI 5.6, 34.7]; p=0.007; Table 21). This effect was proportionally large (8.2%) considering the baseline value of 244.4m.[81]

Dapagliflozin also improved KCCQ-OSS versus placebo as demonstrated with the effect size of 4.5 points (95% CI: 1.1, 7.8; p=0.009; Table 21) and was associated with a greater number of patients in the dapagliflozin group versus placebo that had a 5-point or more improvement in KCCQ-OSS (45.4% versus 34.9%; adjusted OR 1.73; 95% CI: 1.05, 2.85; p=0.03).[81] Similarly, 49.4% of patients in the dapagliflozin group versus 38.2% of those in the placebo group had a 5- point or more improvement in KCCQ-CSS at 12 weeks (adjusted OR 1.64, 95% CI: 0.98, 2.75; p=0.06). Dapagliflozin was associated with greater weight loss (effect size 0.72 kg, 95% CI: 0.01, 1.42; p=0.046; Table 21).[81]

There were no significant differences between groups in other secondary endpoints, including NT-proBNP and BNP; proportion of patients with 20% or greater decrease in NT-proBNP; proportion of patients with both a 5-point or greater increase in KCCQ-CS and 20% or greater decrease in NT-proBNP; HbA1c; and systolic blood pressure at 12 weeks.[81]

Table 21: Primary and secondary endpoints at 12 weeks after treatment initiation in PRESERVED-HF

aValues are shown as adjusted means (95% CI) for continuous variables. bAdjusted for the corresponding baseline value, history of T2DM, sex, AF, baseline eGFR and LVEF.[ c] Ratio of dapagliflozin compared with placebo. Sources: Nassif et al. (2021).[81] Abbreviations : AF: atrial fibrillation; 6MWT: 6-minute walk test; BNP: B-type natriuretic peptide; CI: confidence interval; CSS: Clinical Summary Score; eGFR: estimated glomerular filtration rate; KCCQ: Kansas City Cardiomyopathy Questionnaire; LVEF: left ventricular ejection fraction; N: number of patients in treatment group; NT-proBNP: N-terminal pro B-type natriuretic peptide; OSS: Overall Summary Score; T2DM: type 2 diabetes.

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 67 of 134

Figure 16: Effects of dapagliflozin on the primary endpoint and its components

==> picture [452 x 387] intentionally omitted <==

a–dEffects of dapagliflozin on the primary endpoint and its components. Effects of dapagliflozin versus placebo at 12 weeks on KCCQ-CS (a), KCCQ-TS (b), KCCQ-physical limitations score (KCCQ-PL) (c) and KCCQ-CS by subgroup (d). Units for loop diuretic dose (d), mg furosemide equivalents. Data are presented as mean values with 95% CI. a–c, An F-test was used in the data analysis. All P values are two-sided, with no adjustments made for multiple comparisons.

Sources: Nassif et al. (2021).[81]

Abbreviations : AF: atrial fibrillation; BMI: body mass index; eGFR: estimated glomerular filtration rate; KCCQ: Kansas City Cardiomyopathy Questionnaire; KCCQ-CS: KCCQ clinical score; KCCQ-OS: KCCQ overall score; KCCQ-TS: KCCQ total symptom; KCCQ-PL: KCCQ physical limitation; LVEF: left ventricular ejection fraction; NT-proBNP: N-terminal pro B-type natriuretic peptide; NYHA: New York Heart Association.

B.2.12. Adverse reactions

Summary of safety of dapagliflozin

• The safety profile of dapagliflozin has been previously well reported in other indications, including T2DM, CKD and HFrEF.[8] In DELIVER and PRESERVED-HF, no new safety concerns with dapagliflozin were identified.[76, 81 ]

• In DELIVER, dapagliflozin was generally well tolerated in patients with HF and an LVEF >40%, consistent with the known safety profile.[76] Likewise, in PRESERVED-HF, dapagliflozin was generally well tolerated in patients with HF and an LVEF ≥45%.[81]

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 68 of 134

==> picture [451 x 402] intentionally omitted <==

----- Start of picture text -----
• Overall, the safety profile in DELIVER was associated with: [76, 78]
o Balanced proportions of patients with SAEs (dapagliflozin 43.5% versus placebo
45.5%) and patients with an AE with outcome of death ************** ***** ******
between treatment groups. [76, 78]
o Low and balanced proportion of patients with AE leading to discontinuation of
treatment (DAE) (dapagliflozin 5.8% versus placebo 5.8%) between treatment
groups. [76]
o Balanced proportion of patients with AE leading to interruption of treatment
(dapagliflozin 13.9% versus placebo 15.8%) between treatment groups as well. [76]
** ***
o proportions of patients with SAEs suggestive of volume depletion
****** **** ****** ******* ***** between treatment groups. DAEs suggestive of
volume depletion (************* **** ****** ******* ***** **** *** *** *********** ****** in the
dapagliflozin group. [76, 78]
o Balanced SAEs of renal events ************** **** ****** ******* ***** between
treatment groups. [76, 78]
**** ***
o Two patients with DKA events; and were in the dapagliflozin group. [76, 78]
o Low and balanced proportions of patients with major hypoglycaemic events
(dapagliflozin 0.2% versus placebo 0.2%) between treatment groups. [76]
o Balanced proportions of patients with amputations (dapagliflozin 0.6% versus
placebo 0.8%) between treatment groups. [76]
**********
o Balanced proportions of patients with cardiac ischaemic events
** ******* ***** and strokes ************** **** ****** ******* ***** between treatment
groups. [76, 78]
o No cases of Fournier’s gangrene. [76]
----- End of picture text -----

Extensive safety data already exist for dapagliflozin in other indications, and the safety profile of dapagliflozin has been previously well reported.[8] A summary of common and uncommon adverse drug reactions which have been experienced in these indications is therefore provided in B.2.12.3 based on the SmPC for dapagliflozin.[8]

B.2.12.1. Safety outcomes in DELIVER

In the DELIVER trial, safety and tolerability data were collected for all SAEs, AEs leading to discontinuation, amputation, AEs leading to amputation and potential risk factor AEs for amputations affecting lower limbs.[76]

An overall summary of AEs for patients on treatment is presented in Table 22, while an overall summary for SAEs is shown in Table 23. The proportions of patients with SAEs and of patients with ** ** **** ******* ** ***** were balanced between treatment groups.[76, 78] The proportions of patients with DAEs were low and balanced between treatment groups.[76] The proportions of patients with AEs leading to interruptions of treatment were balanced between treatment groups.[76] The frequency of discontinuation of treatment was *** *** ******** between treatment groups (Figure 17).[78]

The proportions of patients with SAEs suggestive of volume depletion were *** *** ********

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 69 of 134

between treatment groups, whereas DAEs suggestive of volume depletion **** *** *** *********** ****** in the dapagliflozin group.[78] SAEs or DAEs of renal events were balanced between treatment groups.[76]

There were 2 patients with adjudicated as definite DKA events in the dapagliflozin group compared with none in the placebo group; **** patients had T2DM and were treated with insulin.[76, 78] The proportions of patients with major hypoglycaemic events were low and balanced between treatment groups. The proportions of patients with amputations were balanced between treatment groups.[76]

Table 22: Number of patients with AEs in any category in DELIVER – on treatment

a Subjects with multiple events in the same category are counted only once in that category. Subjects with events in more than one category are counted once in each of those categories.[b] Possibly related to IP, as assessed by the Investigator.[c] Based on predefined list of preferred terms.[d] Events adjudicated as definite or probable diabetic

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 70 of 134

ketoacidosis.[e] AE with the following criteria confirmed by the Investigator: i) symptoms of severe impairment in consciousness or behaviour ii) need of external assistance iii) intervention to treat hypoglycaemia iv) prompt recovery of acute symptoms following the intervention reported by the investigator in CRF.[f] Reported by the investigator on the CRF amputation form, including surgical or spontaneous/non-surgical amputation, excluding amputation due to trauma.[g] Investigator-reported diagnosis from the cardiac ischaemic events CRF. hInvestigator-reported diagnosis from the cerebrovascular events CRF (haemorrhagic, ischaemic, undetermined). iIncludes ST elevation myocardial infarction (STEMI), Non-ST elevation myocardial infarction (NSTEMI), and Myocardial infarction, ST elevation status unknown.

This table includes AEs with an onset date on or after date of first dose of IP (on and off treatment), and on or after the first dose and up to and including 30 days following last dose of IP (on treatment). Percentages are based on the total numbers of patients in the treatment group (N). Source: Solomon et al. (2022);[76] DELIVER CSR.[78]

Abbreviations : AE: adverse event; CRF: case report form; DAE: AE leading to discontinuation of IP; Dapa: dapagliflozin; IP: investigational product; MI: myocardial infarction; N: number of patients in treatment group; SAE: serious AE.

Table 23: Number of patients with SAEs (≥ 0.5%) by preferred term in DELIVER – On treatment

a Number (%) of patients with SAEs, sorted by descending frequency of PT in Dapa 10 mg group. Subjects with multiple events in the same PT are counted only once in that PT. Subjects with events in more than

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 71 of 134

one PT are counted once in each of those PTs. This table includes SAEs with an onset date on or after date of first dose of IP, and up to and including 30 days following last dose of IP, with a frequency ≥ 0.5% in the dapagliflozin treatment group.

Source : Solomon et al. (2022).[85]

Abbreviations : COVID-19: coronavirus disease 2019; Dapa: dapagliflozin; IP: investigational product; MI: myocardial infarction; N: number of patients in treatment group; PT: preferred term; SAE: serious adverse event.

Figure 17: KM plot of the cumulative percentage of patients with premature permanent discontinuation of treatment in DELIVER[a]

==> picture [452 x 252] intentionally omitted <==

aN at risk is the number of patients at risk at the beginning of the period. One month corresponds to 30 days. Two-sided p-value is displayed.

Source : DELIVER CSR.[78]

Abbreviations : Dapa: dapagliflozin; D: dapa 10 mg; IP: investigational product; KM: Kaplan-Meier; N: number of patients; P: placebo.

B.2.12.2. Safety outcomes in PRESERVED-HF

Adverse events from the PRESERVED-HF trial are presented in Table 24. Overall, adverse events were similar between the dapagliflozin and placebo groups with 44 (27.2%) patients versus 38 (23.5%) patients experiencing adverse events, respectively.[81]

Table 24: Safety analysis in PRESERVED-HF

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 72 of 134

Values are shown as absolute numbers (percentages) for patients with events. Sources: Nassif et al . (2021).[81] Abbreviations : DKA: diabetic ketoacidosis; MI: myocardial infarction; N: number of patients in treatment group.

B.2.12.3. Adverse drug reactions reported in the Summary of Product Characteristics

A summary of common and uncommon adverse drug reactions which have been identified in the placebo-controlled clinical studies and post-marketing surveillance of dapagliflozin is provided in Table 25, based on the SmPC for dapagliflozin.

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 73 of 134

Table 25: Adverse drug reactions reported in the SmPC for dapagliflozin: adverse reactions in placebo-controlled clinical studies[a] and postmarketing experience

aThe table shows up to 24-week (short-term) data regardless of glycaemic rescue. bSee corresponding subsection of SmPC for additional information.[c] Vulvovaginitis, balanitis and related genital infections includes, e.g., the predefined preferred terms: vulvovaginal mycotic infection, vaginal infection, balanitis, genital infection fungal, vulvovaginal candidiasis, vulvovaginitis, balanitis candida, genital candidiasis, genital infection, genital infection male, penile infection, vulvitis, vaginitis bacterial, vulval abscess.[d] Urinary tract infection includes the following preferred terms, listed in order of frequency reported: urinary tract infection, cystitis, Escherichia urinary tract infection, genitourinary tract infection, pyelonephritis, trigonitis, urethritis, kidney infection and prostatitis. eVolume depletion includes, e.g., the predefined preferred terms: dehydration, hypovolaemia, hypotension. fPolyuria includes the preferred terms: pollakiuria, polyuria, urine output increased. gMean changes from baseline in haematocrit were 2.30% for dapagliflozin 10 mg versus-0.33% for placebo. Haematocrit values >55% were reported in 1.3% of the patients treated with dapagliflozin 10 mg versus 0.4% of placebo patients.[h] Mean percent change from baseline for dapagliflozin 10 mg versus placebo, respectively, was: total cholesterol 2.5% versus 0.0%; HDL cholesterol 6.0% versus 2.7%; LDL cholesterol 2.9% versus -1.0%; triglycerides –2.7% versus -0.7%. iSee section 4.4 of the SmPC. jAdverse reaction was identified through postmarketing surveillance. Rash includes the following preferred terms, listed in order of frequency in clinical studies: rash, rash generalised, rash pruritic, rash macular, rash maculo-papular, rash pustular, rash vesicular, and rash erythematous. In active- and placebo-controlled clinical studies (dapagliflozin, N=5936, All control, N=3403), the frequency of rash was similar

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 74 of 134

for dapagliflozin (1.4 %) and all control (1.4%), respectively.[k] Reported in the CV outcomes study in patients with type 2 diabetes (DECLARE). Frequency is based on annual rate.[l] Reported in ≥ 2% of patients and ≥ 1% more and at least 3 more patients treated with dapagliflozin 10 mg compared with placebo.

mReported by the investigator as possibly related, probably related or related to study treatment and reported in ≥ 0.2% of patients and ≥ 0.1% more and at least 3 more patients treated with dapagliflozin 10 mg compared with placebo.

Source : Forxiga 10 mg film-coated tablets [SmPC].[8]

Abbreviations : HDL: high density lipoprotein; LDL: low density lipoprotein; T2DM: type 2 diabetes mellitus.

B.2.13. Ongoing studies

There are no ongoing trials relevant to this appraisal.

B.2.14. Interpretation of clinical effectiveness and safety evidence

B.2.14.1. Principal outcomes from DELIVER and PRESERVED-HF highlighting the clinical benefits and harms of the technology

DELIVER is one of the first trials including patients with HF and an LVEF >40% that has demonstrated statistically significantly improved outcomes in this highly underserved patient population

To date, all but a few recently published RCTs have failed to demonstrate significant clinical benefits for treatments in patients with HF and an LVEF >40%.[58, 76, 95] As such, there are no targeted, disease-modifying treatments indicated or commissioned by the NHS to treat this patient population. Without an efficacious, well-tolerated treatment, patients with HF and an LVEF >40% experience poor clinical outcomes and HRQoL and face a life expectancy worse than patients with some cancers.[32] As such, clinical care is currently limited to symptomatic treatment and/or treatment for underlying co-morbidities, rather than treatments for HF and an LVEF >40%. There is therefore an urgent need for easily accessible new treatments which can reduce mortality and hospitalisation and improve disease symptoms and quality of life.

DELIVER (N=6,263) is one of a few RCTs to demonstrate statistically significantly improved outcomes in patients with HF and an LVEF >40%.[76, 81, 95] DELIVER was also the first trial to include patients with HFimpEF and to demonstrate a ************* *********** *********** in this patient subgroup. The treatment benefits of dapagliflozin versus placebo , when given in addition to SoC, in DELIVER demonstrate that dapagliflozin is a key opportunity to significantly reduce worsening of HF in patients with HF and an LVEF >40%.[76]

Consistent with other phase III RCTs of dapagliflozin, a statistically significant reduction in the risk of the primary composite endpoint of CV mortality and HF events was observed for dapagliflozin compared with placebo in DELIVER[76]

Dapagliflozin significantly reduced the incidence of the primary composite endpoint of CV mortality and HF events by 18% compared with placebo in the DELIVER trial (HR 0.82; 95% CI 0.73, 0.92; p<0.001).[76] The treatment benefit on the primary composite endpoint was consistent across the key prespecified subgroups.[76]

This significant reduction in the primary composite endpoint of CV mortality and HF events is consistent with outcomes from other dapagliflozin RCTs.[96] In the DAPA-HF trial, which enrolled patients aged ≥18 years with NYHA functional class ≥II and an LVEF ≤40% who were currently optimally treated for HFrEF, dapagliflozin reduced the relative risk of CV mortality or an HF event

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 75 of 134

by 26% (HR 0.74; 95% CI: 0.65, 0.85; p<0.001).[97] In the DAPA-CKD trial, which enrolled patients with CKD (eGFR ≥25 and ≤75 ml/min/1.73 m[2] , and uACR ≥200 mg/g to ≤5,000 mg/g [≥22.6 to ≤565 mg/mmol]), dapagliflozin was associated with a 29% reduction in the relative risk of HHF or CV mortality (HR 0.71; 95% CI: 0.55, 0.92; p=0.009).[98]

In the DELIVER trial, dapagliflozin also statistically significantly reduced the relative risk of the secondary composite endpoint of CV mortality and recurrent HF events by 23% compared with placebo[76]

In the DELIVER trial, dapagliflozin was statistically significantly superior to placebo in reducing the incidence of the of total (first and recurrent/ repeat) HF events and CV mortality (RR 0.77; 95% CI: 0.67, 0.89; p<0.001).[76] Dapagliflozin was also statistically significantly superior to placebo in reducing the incidence of recurrent HF events (RR 0.73; 95% CI: 0.62, 0.87; p=0.0003). Dapagliflozin reduced the incidence of CV mortality and all-cause mortality although the differences were not statistically significant (HR 0.88; 95% CI: 0.74, 1.05; p=0.1678 and HR 0.94; 95% CI: 0.83, 1.07; p=0.3425, respectively).[76, 78] Consistent with results from the DELIVER trial, as demonstrated by the pooled analysis for the DELIVER and DAPA-HF trials, dapagliflozin in HF irrespective of LVEF significantly reduced total hospital admissions for HF by 29% (RR 0.71; 95% CI: 0.65, 0.78; p<0.001).[94] These clinical benefits further demonstrate the value that dapagliflozin could offer patients and the NHS by improving outcomes and reducing the resource utilisation associated with HF and an LVEF >40%.

The treatment effect of dapagliflozin was highly consistent across prespecified subgroups including those defined by LVEF, with no attenuation of treatment effect in the highest LVEF group (>60%) or those with HFimpEF[76, 79]

The results observed in the DELIVER FAS were consistently reflected in key prespecified subgroups.[76] Importantly, when the population was stratified into LVEF ≤49%, 50%–59%, and ≥60%, the treatment effect with dapagliflozin remained consistent across the groups (p-value for interaction=*****),[78] suggesting no attenuation of treatment effect in patients with a higher LVEF.[76] This is a critical finding given that previous trials of treatments for patients with HF and an LVEF >40% appeared to show a trend towards an attenuation of treatment effect in those with a higher LVEF.[95]

Similarly, results were also consistent between patients with HFimpEF and those with HF and an ***** LVEF consistently >40% (p-value for interaction= ).[76, 79] This is an important finding since patients with HFimpEF were previously unstudied. Taken together, and considered alongside the results from DAPA-HF, these results demonstrate that initiating dapagliflozin in patients with HF is associated with a consistent treatment effect irrespective of LVEF.

In a pooled analysis of the DELIVER and DAPA-HF trials, dapagliflozin was associated with a statistically significant reduction in the risk of CV mortality and mortality from any cause of 14% and 10%, respectively, in patients with HF irrespective of LVEF[94]

In the pooled analysis of the DELIVER and DAPA-HF trials, dapagliflozin significantly reduced the risk of mortality from CV causes in HF irrespective of LVEF by 14% (HR 0.86; 95% CI: 0.76, 0.97; p=0.01), and the risk of mortality from any cause by 10% (HR 0.90; 95% CI: 0.82, 0.99; p=0.03).[94] These results are broadly consistent with reductions in both CV mortality and all-cause mortality reported in the DELIVER and DAPA-HF trials,[76, 97] with the higher power of the pooled analysis resulting in statistically significant differences being demonstrated.[94] Dapagliflozin offers a key opportunity to reduce mortality across the spectrum of HF regardless of LVEF, which is of critical importance given the high mortality rates associated with HF,[14] and the NHS Long Term

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 76 of 134

Plan having identified CVD as the single biggest area where lives can be saved by 2029 in England.[46]

Treatment with dapagliflozin provides meaningful symptom relief in patients with HF and an LVEF >40% based on both the DELIVER and the PRESERVED-HF trials[76, 81, 93]

Given the high morbidity burden associated with HF and an LVEF >40%,[32] improving disease symptoms in addition to treatment outcomes is a primary goal in the management of these patients. In the DELIVER trial, dapagliflozin provided symptom benefits over placebo as measured by KCCQ-TSS (mean difference in change from baseline 2.4 [95% CI: 1.5, 3.3] points higher versus placebo; p<0.001).[76, 79, 93] Similarly, in the PRESERVED-HF trial, 12-week treatment with dapagliflozin versus placebo was associated with statistically significant improvements in patient-reported symptoms, physical limitation and exercise function.[81]

Dapagliflozin was generally well tolerated, consistent with its known safety profile

In DELIVER, dapagliflozin showed a favourable tolerability profile compared with placebo; SAEs were numerically less frequent with dapagliflozin (43.5%) than with placebo (45.5%) and there was no difference in incidence of AEs leading to discontinuation between dapagliflozin (5.8%) and placebo (5.8%).[76] In DELIVER and PRESERVED-HF, no new safety concerns were identified.[76, 81] Dapagliflozin is already routinely commissioned to treat T2DM,[5-7] CKD,[9] and HFrEF,[1] thus clinicians across both primary and secondary care settings have considerable clinical experience in the prescribing of dapagliflozin. Therefore, the lack of new safety concerns from DELIVER and PRESERVED-HF provides reassurance about the known safety profile clinicians are already familiar with.

Dapagliflozin is a vital new treatment option for patients with HF and an LVEF >40%, with the potential to significantly reduce the burden of HF on patients and the healthcare system

The DELIVER results demonstrate that dapagliflozin is an effective and well tolerated treatment which can help ease the substantial burden of HF and an LVEF >40% to patients and the NHS.[76] Benefits associated with dapagliflozin in this patient population include improved outcomes, including mortality and patient-reported symptoms, and lowered healthcare resource use in HF, such as HF events, compared with placebo.[76, 81, 94]

Improved outcomes with dapagliflozin compared with SoC are key to tackling the current burden associated with HF and an LVEF >40%, including the high mortality,[14, 32] and poor HRQoL.[32] Improving care in HF will support achieving one of the priorities of the NHS Long Term Plan, in which CVD has been identified as the single biggest area where lives can be saved by 2029 in England.[46] Given that HF and an LVEF >40% is associated with a substantial economic burden, primarily driven by high rates of HHF,[38-42] dapagliflozin offers a key opportunity to reduce healthcare resource use in HF, including HF events, for the NHS. Although the availability of novel therapies for patients with HF and an LVEF >40% may result in a greater focus on specialist service provision, service redesign specifically for dapagliflozin would not be necessary as it does not require up-titration nor specific additional monitoring.

Given that there is substantial clinical experience in the prescribing of SGLT2 inhibitors in primary care, AstraZeneca believes that there is no clinical rationale for restricting the initiation of dapagliflozin for patients with HF and an LVEF >40% to advice from a HF specialist only. In this context, a specialist confirmed HF diagnosis is likely to remove uncertainty such as misdiagnosis

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 77 of 134

and associated over-treatment, and to increase confidence in primary care physicians to initiate treatment that they are familiar with in newly diagnosed patients and those recently discharged from HF specialist services.

For the prevalent population of patients with a diagnosis of HF and an LVEF >40% already managed in primary care or for those who are not routinely followed-up within specialist care, initiation of dapagliflozin could take place at the earliest opportunity, ideally following proactive invitation for a treatment optimisation review or alternatively, where capacity is a limitation, during their routine check-up appointment without the need for a specific or extended appointment. Initiating treatment for patients within primary care will support the NHS with its COVID-19 recovery plans, reducing wait times to outpatient services,[99] and will reduce unwarranted variations in care across England and Wales. Thus, enabling initiation of dapagliflozin in both primary and secondary care for the treatment of this patient population would ensure consistent equality of access without relying on specialist care, which may not exist in some areas for these patients.

B.2.14.2. Strengths and limitations of the clinical evidence base for the

technology

DELIVER was a large (N=6,263), Phase III, international, multi-centre, double-blind, placebocontrolled high quality RCT, which enrolled a patient population with a broad range of comorbidities, including patients with and without T2DM.[76] DELIVER was designed with broader inclusion criteria than those used in previous trials involving similar populations; it enrolled patients who were hospitalised or recently hospitalised, for whom evidence-based therapy is limited, as well as those with HF and an LVEF previously ≤40% prior to enrolment.[76] Data from DELIVER suggest that these understudied groups also benefit from dapagliflozin.[76]

Overall, demographic and other baseline patient characteristics were well balanced between treatment groups.[76] The outcome measures selected were those most relevant to patients with HF and an LVEF >40%, including CV mortality and HF events, with a composite of these outcomes as the primary efficacy measure.[76]

Moreover, the overall impact of the COVID-19 pandemic on the efficacy evaluation was assessed as low and the COVID-19 pandemic was judged not to have had a meaningful impact on the interpretation of results of the trial.[76]

Based on UK clinical expert feedback, the DELIVER trial is overall considered to be reflective of SoC used in UK clinical practice.[10] Although the trial did not enrol any UK patients, it included a large European and American cohort, where treatments are expected to be similar to those in UK clinical practice.[76] Discrepancies mentioned by UK clinical experts included the trial mean age of 71.7 years,[80] which appears slightly younger than in UK clinical practice as demonstrated by the average of **** years in the CPRD analysis (see Section B.3.3.2),[10, 62] the likelihood of an increased proportion of patients with NYHA class II HF in UK clinical practice than in the trial, and the potential to have higher proportion of patients from African and Caribbean background in some areas than in the trial.[10, 62, 76] While there are some differences between DELIVER and UK clinical practice, UK clinical experts generally agreed that the trial is broadly representative of UK clinical practice.[10] Nonetheless, AstraZeneca recognises these differences and have, therefore, performed a scenario analysis using the CPRD dataset in addition to using the DELIVER trial cohort in the base case cost-effectiveness analysis (see Section B.3.10.3).

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 78 of 134

B.3. Cost effectiveness

Summary of cost effectiveness

• A cost-utility model was developed to estimate the cost-effectiveness of dapagliflozin in addition to SoC (defined as loop diuretics, either furosemide or bumetanide) versus placebo in addition to SoC (hereafter referred to as SoC alone for ease of reading) for the treatment of adult patients with HF and an LVEF >40%.

• The model was a Markov cohort model with health states based on KCCQ-TSS scores. Disease progression was modelled through transitions between discrete health states characterised by KCCQ-TSS quartiles (scores of 0–<55, 55–<73, 73–<88, 88–100, where higher scores represent better health status), with health state-specific clinical event rates, costs and utility values.

• Baseline characteristics and clinical evidence for the efficacy of dapagliflozin in addition to SoC and SoC alone were derived directly from the DELIVER trial, and applied in the economic model as transition probabilities, survival equations and risk equations. These were used to model clinical events, including HF events, CV mortality and all-cause mortality, as well as any relevant AEs.

• Health state utility values and clinical event disutility values were derived from the DELIVER trial and AE utility decrements were sourced from the published literature.

• The analysis was consistent with the NICE reference case and took an NHS and PSS perspective. Costs and benefits were discounted at a rate of 3.5% and a lifetime time horizon was adopted.

• In the deterministic base case economic analysis, treatment with dapagliflozin in addition to SoC, compared with SoC alone, was associated with increased life years (+0.369 per patient) and increased QALYs (+0.250 per patient), at an incremental cost of +£1,880 per patient. As a result, dapagliflozin in addition to SoC was highly cost-effective compared with SoC alone, with an ICER of £7,507/QALY gained.

• The probabilistic base case economic analysis results were similar to the deterministic base case results, demonstrating that the cost-effectiveness of dapagliflozin is robust to uncertainties associated with the model input parameters. The probabilistic sensitivity analysis (PSA) showed that the probabilities of cost-effectiveness for dapagliflozin at willingness-to-pay (WTP) thresholds of £20,000/QALY and £30,000/QALY gained were 89.0% and 92.3%, respectively.

• The most influential factors of the deterministic sensitivity analysis (DSA) were the annual probability of amputation for dapagliflozin in addition to SoC and SoC alone and the event cost of HHF. Overall,, dapagliflozin in addition to SoC remained highly cost-effective compared with SoC alone with ICERs below £9,000/QALY gained for all upper and lower input values varied in the DSA.

• Similarly, all scenario analyses demonstrated the base case economic analysis to be robust, with probabilistic ICERs remaining below £12,500/QALY gained in all scenarios.

• In conclusion, the economic analysis shows dapagliflozin to represent a highly cost-effective use of NHS resources, as an add-on therapy to SoC for the treatment of adults with HF and an LVEF >40%.

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648]

© AstraZeneca (2022). All rights reserved

B.3.1. Published cost-effectiveness studies

An economic SLR was conducted in June 2022 to identify any relevant published costeffectiveness analyses, utilities studies, or cost and resource use studies in patients with HF and an LVEF >40%. Full details of the methodology and results of the economic SLR are presented in Appendix G, H and I.

In total, only one economic evaluation was identified in the cost-effectiveness analyses SLR: a cost per outcome study (Tsaban et al. [2021]),[100] which evaluated the annual number needed to treat to prevent the composite outcome of HF hospitalisation and CV mortality for either spironolactone or sacubitril/valsartan.

The study is summarised in Appendix G, but was not considered to provide relevant evidence to the decision problem of this submission, or any relevant assumptions that could be leveraged for the economic analysis of this submission, and was therefore not considered further.

B.3.2. Economic analysis

In the absence of identifying any previously conducted cost-effectiveness studies relevant to the decision problem of this submission in the economic SLR, a de novo economic model was developed for this submission, based on the modelling approach adopted in previous economic models in HFrEF (TA388 and TA679) which have been accepted by NICE.[1, 101]

In particular, the model structure used in this appraisal is closely aligned with the model used in the previous appraisal for dapagliflozin as a treatment for HFrEF (TA679), as discussed with the EAG and NICE prior to this submission.[1]

A summary of the key differences between the underlying model structure and methodology in TA679 versus the economic model developed for this submission is provided in Table 26. It should be noted that in addition to these differences, the model inputs used in TA679 were reviewed and updated to include inputs from the DELIVER trial, or those from the published literature considered most appropriate to this appraisal, as detailed in the sections below.

Table 26: Summary of the key differences in modelling approaches between TA679 versus this appraisal

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 80 of 134

Abbreviations : CV: cardiovascular; NICE: National Institute for Health and Care Excellence; T2DM: type 2 diabetes mellitus; TA: technology appraisal; UHFV: urgent heart failure visit.

B.3.2.1. Patient population

In line with the expected licensed indication and the decision problem for the current submission, the base case economic analysis evaluated adult patients with HF and an LVEF >40%. This is aligned to the population investigated in the DELIVER trial which is the pivotal study for dapagliflozin in addition to SoC versus placebo in addition to SoC (SoC alone) in this indication (see Section B.2.2).

B.3.2.2. Model structure

A Markov state-transition model was developed whereby disease progression was modelled through transitions between discrete health states characterised by KCCQ-TSS quartiles with the following scores, with higher scores representing better health status:

• Q1: 0–<55

• Q2: 55–<73

• Q3: 73–<88

• Q4: 88–100

As discussed in Section B.1.3.4, the KCCQ-TSS is an extensively validated and established self-

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 81 of 134

administered instrument for quantifying HF-related symptoms, function, and HRQoL in patients with HF.[67] As a specifically designed patient-reported measure of HF health status, reflective of patient utility, KCCQ-TSS quartiles were considered appropriate for defining health states in the economic model. The inclusion of KCCQ-TSS quartiles for health states also has a precedent in economic modelling in HF, in line with the previous model structure accepted for the NICE appraisal for dapagliflozin for HFrEF in TA679.[1]

A schematic overview of the economic model structure is presented in Figure 18. Each health state was assigned health state-specific utility values. This represents one of the advantages of this model structure, as the KCCQ-TSS health states enable the impact of disease severity to be captured in the health state utility values and in the risk of events, and therefore allow the impact of disease severity to be more accurately modelled.

Additionally, the model captured the incidence of HF events as transient events. Patient mortality (i.e., transition to the absorbing dead state) was modelled through the application of parametric survival equations describing CV mortality and all-cause mortality.

At each cycle, the proportion of patients who died from CV causes was estimated and the costs associated with CV mortality were applied. The non-CV mortality rate was estimated as the difference between the all-cause mortality rate and the CV mortality rate, which was also applied to all KCCQ-TSS health states. The transition probability matrix for the different KCCQ-TSS quartiles was then applied to the remaining number of patients alive, to calculate the health state distribution in the next cycle (see Section B.3.3.3).

Patients had a per-cycle probability of discontinuing treatment with dapagliflozin due to intolerability or other reasons, based on the DELIVER trial, as detailed in Section B.3.3.4 below.[78] Patients discontinuing treatment with dapagliflozin in addition to SoC were then modelled to experience the same event rates as patients receiving SoC alone.

Figure 18: Schematic of Markov state-transition model structure, health states, and possible transitions

==> picture [404 x 218] intentionally omitted <==

Abbreviations: CV: cardiovascular; HHF: hospitalisation for heart failure; KCCQ: Kansas City Cardiomyopathy Questionnaire; UHFV: urgent heart failure visit.

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 82 of 134

Justification of model structure

The implementation of a Markov state-transition model was considered appropriate as the heterogeneity between patients with HF and an LVEF >40% with respect to important disease characteristics can be captured by a tractable number of mutually exclusive and exhaustive health states. The use of a Markov model structure is aligned with the model structure used in the previous NICE appraisal for dapagliflozin in patients with HFrEF (TA679), and prior discussion with the NICE and EAG indicated that a similar model structure would be suitable for this appraisal.[1]

HF is a chronic and progressive disease associated with an increased risk of mortality over time. As such, the model incorporated a lifetime horizon in line with the NICE Methods Guide.[105] Consistent with UK 2017–2019 life tables, it was assumed that all patients died upon reaching 101 years old.[106]

The cycle length was one month, and a half-cycle correction was applied, in line with TA679.[1]

A summary of the key model characteristics is presented in Table 27.

Table 27: Key features of the economic analysis

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 83 of 134

Abbreviations: KCCQ-TSS: Kansas City Cardiomyopathy Questionnaire Total Symptom Score; NHS: National Health Service; NICE: National Institute for Health and Care Excellence; PSS: Personal Social Services; QALY: quality-adjusted life year.

B.3.2.3. Intervention technology and comparators

The intervention technology is oral dapagliflozin (10 mg) once daily. In line with the proposed positioning of dapagliflozin in the treatment pathway for patients with HF and an LVEF >40%, dapagliflozin is to be given as an add-on therapy to current SoC. Therefore the intervention arm of the economic analysis comprised dapagliflozin in addition to SoC.

The principal comparator to dapagliflozin in addition to SoC in this submission is placebo in addition to SoC (hereafter, referred to as SoC alone).

Dapagliflozin plus SoC

Dapagliflozin was modelled in line with the SmPC at a dose of 10 mg orally once daily until treatment discontinuation,[8] while SoC for patients receiving dapagliflozin was modelled in line with the modelling approach for SoC alone, detailed below.

A constant probability of dapagliflozin treatment discontinuation was included in the model, and once patients discontinued treatment with dapagliflozin they became subject to the same risks, costs and utility decrements as patients in the SoC arm of the model (see Section B.3.3.4).

SoC

The principal comparator to dapagliflozin in addition to SoC in this submission is SoC alone. There are currently no disease-modifying treatment options for patients with HF and an LVEF >40% and in UK clinical practice, SoC for this patient population consists of loop diuretics prescribed for symptom relief (see Section B.1.3.5).

SoC within the base case economic analysis was modelled as the cost of loop diuretics, assumed to be comprised of a weighted average of 80% furosemide (40 mg orally once daily) and 20% bumetanide (1 mg orally once daily), based on UK clinical expert feedback that these are the most commonly used loop diuretics in UK clinical practice.[110, 111] The costs of SoC were applied to both arms of the model (see Section B.3.5.1). No discontinuation of SoC was assumed within the model.

The modelling of further additional therapies to treat comorbidities was not included, given the use of these therapies is expected to be the same for patients receiving dapagliflozin in addition to SoC and those receiving SoC alone. As such, any differences in the costs associated with further additional therapies to treat comorbidities was assumed to be negligible, and it was not

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 84 of 134

considered necessary to explicitly model these therapies.

B.3.3. Clinical parameters and variables

B.3.3.1. Incorporation of clinical data within the model

Data from the DELIVER trial were incorporated directly into the dapagliflozin economic model to inform: patient baseline characteristics, KCCQ-TSS quartile health state transition probabilities, survival curves for mortality, incidence of HF events, incidence of AEs and probability of treatment discontinuation. Additionally, health state utility values and utility decrements for HF events were also derived directly from the DELIVER trial (see Section B.3.4.1 and Section B.3.4.5, respectively).

The treatment effect of dapagliflozin was incorporated into the economic model as coefficients for the survival equations and risk equations for all-cause mortality, CV mortality and HF events. Additionally, *** ************* *********** ********* ****** with respect to change in KCCQ-TSS from baseline in the DELIVER trial was incorporated in the economic model as treatment-specific KCCQ-TSS quartile transition probabilities.[78]

B.3.3.2. Baseline characteristics

DELIVER ITT population

The patient baseline characteristics informing the economic model were derived from the DELIVER trial, and are summarised below in Table 28 (demographic characteristics), Table 29 (clinical characteristics) and Table 30 (medical history). The patient baseline characteristics determined the initial distribution of the modelled cohort across the alive health states and influenced the rates of all-cause mortality, CV mortality and HF events estimated by the covariate-adjusted survival equations and covariate-adjusted risk equations.

Table 28: Patient demographic characteristics incorporated in the base case economic analysis

Abbreviations : BMI: body mass index; SE: standard error. Source: Solomon et al. (2022);[80] DELIVER CSR.[78]

Table 29: Patient clinical characteristics incorporated in the base case economic analysis

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 85 of 134

Abbreviations : BMI: body mass index; eGFR: estimated glomerular filtration rate; KCCQ: Kansas City Cardiomyopathy Questionnaire; LVEF: left ventricular ejection fraction; NT-proBNP: N-terminal pro-B-type natriuretic peptide; NYHA: New York Heart Association; SBP: systolic blood pressure; SE: standard error. Source: DELIVER CSR.[78]

Table 30: Patient medical history incorporated in the base case economic analysis

Abbreviations : AFF: atrial fibrillation/flutter; CKD: chronic kidney disease; HF: heart failure; HHF: hospitalisation for heart failure; SE: standard error; T2DM: type 2 diabetes mellitus. Source: DELIVER CSR.[78]

UK CPRD dataset

In a scenario analysis (see Section B.3.10.3), patient baseline characteristics were incorporated in the economic model based on the UK CPRD dataset for patients with HF and an LVEF >40% in the UK, as detailed in Table 31, Table 32 and Table 33 below.[62] Where baseline characteristics were not available from the UK CPRD, the inputs from the DELIVER trial were used, as denoted above.[78]

Table 31: Patient demographic characteristics based on the UK CPRD dataset used in a scenario analysis

Abbreviations : BMI: body mass index. Source: UK CPRD dataset.[62]

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 86 of 134

Table 32: Patient clinical characteristics based on the UK CPRD dataset used in a scenario analysis

Abbreviations : BMI: body mass index; eGFR: estimated glomerular filtration rate; KCCQ: Kansas City Cardiomyopathy Questionnaire; LVEF: left ventricular ejection fraction; NT-proBNP: N-terminal pro-B-type natriuretic peptide; NYHA: New York Heart Association; SBP: systolic blood pressure; SE: standard error. Source: UK CPRD dataset.[62]

Table 33: Patient medical history based on the UK CPRD dataset used in a scenario analysis

Abbreviations : AFF: atrial fibrillation/flutter; CKD: chronic kidney disease; HF: heart failure; HHF: hospitalisation for heart failure; SE: standard error; T2DM: type 2 diabetes mellitus. Source : UK CPRD dataset.[62]

B.3.3.3. Health state transitions

Health state membership within the economic model was fully determined by time-dependent transition probabilities between health states. The transition probabilities between health states defined by KCCQ-TSS quartiles were derived using monthly transition count data from the DELIVER trial, assuming last observation carried forward (i.e., patients were assumed to remain in a KCCQ-TSS quartile until an observation indicating they had moved elsewhere).[78] Transition counts have a multinomial likelihood, which was combined with a flat Dirichlet prior distribution using Gibbs sampling to obtain the posterior probability distribution of the KCCQ-TSS transition matrix.

Treatment-specific transition probabilities were derived for the dapagliflozin in addition to SoC and placebo in addition to SoC arms of the DELIVER trial, respectively, as a statistically significant change in KCCQ-TSS was observed in the DELIVER trial (win ratio 1.11 [95% CI: 1.03, 1.21], p=0.009).[76] Given that KCCQ-TSS quartiles are used in this analysis to capture

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 87 of 134

disease progression, this result indicated an associated difference in disease progression between treatment with dapagliflozin in addition to SoC versus SoC alone, thereby validating the separation of transition probabilities.

Based on previous methods for modelling HF and an LVEF <40% (including dapagliflozin in the DAPA-HF trial on which the present analyses were based), disease progression trajectories were split between a phase spanning the first four months and a separate phase covering the remainder of the trial.[1, 9, 112] The monthly probably of transition between health states defined by KCCQ-TSS quartiles is shown in Table 34.

Table 34: Monthly KCCQ-TSS transition matrix

Abbreviations : KCCQ-TSS : Kansas City Cardiomyopathy Questionnaire-Total Symptom Score; SE: standard error; SoC: standard of care.

B.3.3.4. Treatment discontinuation

The probability of treatment discontinuation with dapagliflozin was derived from the DELIVER clinical trial and was applied as a constant probability of discontinuation to all patients receiving treatment with dapagliflozin in each modelled cycle. The annual probability of treatment ******* ******* discontinuation was (SE: ).[78] Following discontinuation of dapagliflozin, patients were assumed to continue receiving SoC alone, and experienced the same event rates, mortality and costs as patients in the SoC alone arm. This approach assumes that all treatment effect of dapagliflozin is instantly lost upon discontinuation and may therefore be considered a conservative assumption.

B.3.3.5. CV mortality and all-cause mortality

Evaluation of survival

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 88 of 134

The DELIVER trial provided observed survival data over a median follow-up of ** months with ** end-of-trial overall survival of %.[78] To adopt a lifetime time horizon in the model, extrapolation beyond the trial period was required. The approach to survival modelling followed the extensive methods advocated by the NICE Decision Support Unit (DSU) Technical Support Documents (TSD) and published guidelines.[113-115]

Non-parametric evaluation of the DELIVER trial data was demonstrated with treatment-stratified Kaplan-Meier (KM) survival curves for CV mortality and all-cause mortality as illustrated in Figure 19 and Figure 20 respectively. The data were considered immature, as only a minority of patients died over the course of the trial, and median survival was *** ******* for CV or all-cause mortality.[76, 78]

When stratified by treatment arm, the KM curves for dapagliflozin in addition to SoC versus SoC alone followed a similar trajectory with overlapping and crossing of curves, possibly indicating a trial entry effect, before later differences emerged. The KM curves for dapagliflozin in addition to SoC versus SoC alone then demonstrated clear separation after one year for both CV and allcause mortality (a larger separation was observed for CV mortality).

Figure 19: KM curves for CV mortality in the DELIVER trial, stratified by treatment

==> picture [386 x 247] intentionally omitted <==

Abbreviations: CV: cardiovascular; KM: Kaplan-Meier.

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 89 of 134

Figure 20: KM curves for all-cause mortality in the DELIVER trial, stratified by treatment

==> picture [387 x 247] intentionally omitted <==

Abbreviations : KM: Kaplan-Meier.

For both CV mortality and all-cause mortality, dapagliflozin in addition to SoC was generally associated with a lower hazard than SoC alone (some overlap occurs in the early phases of the trial).

Hazard plots (presented in Appendix N.1.1) showed a general trend towards an increasing hazard of mortality over the course of the DELIVER trial, with a greater increase apparent for allcause mortality compared with CV mortality, as expected with an aging trial population.

As such, it was considered that the parametric models used for CV and all-cause mortality should broadly reflect this trend in increasing hazards over time. An inflection point in the hazard trajectory was observed after approximately one year, with the hazards of mortality generally appearing to increase beyond this point.

Evaluation of relational models

Based on the evaluation of the survival and hazard profiles and in line with NICE DSU TSD14 guidance, the data were taken to be too complex to be represented with a single statistical model and therefore a piecewise approach was adopted. Diagnostic assessment informed the placing of a single split at one year to address the major inflection point and the change in hazard profile at this time point, while maximising the use of available data to inform the extrapolations. Suitability of the approach to address the proportional hazards (PH) assumption was confirmed by visual inspection and inferential testing, with p-values greater than 0.05 taken to indicate results consistent with the PH assumption. Full details of the PH assessments are presented in Appendix N.1.2.

Visual inspection of the log-cumulative hazard plots stratified by treatment arm showed general vertical parallelisation, which indicates that the PH assumption was valid. Log-cumulative hazard plots stratified by KCCQ-TSS quartiles were piecewise parallel, with deviations as described above, only in distinct follow-up phases. After application of the piecewise approach past one year, diagnostic plots of Schoenfeld residuals for models stratified by treatment and time-varying

Company evidence submission template for dapagliflozin for treating chronic heart failure with preserved or mildly reduced ejection fraction [ID1648] © AstraZeneca (2022). All rights reserved Page 90 of 134

KCCQ-TSS quartile suggested results were not inconsistent with the PH assumption as a function of time across the duration of the trial follow-up (p=***** for CV mortality and p=***** for all-cause mortality).

In addition to assessing for the PH assumption, accelerated failure time (AFT) models were also assessed using visual and statistical diagnostics. Visual inspection of the log-cumulative hazard plots stratified by treatment arm showed parallelisation on the horizontal plane being suggestive of AFT. No major deviations from linearity on the quantile-quantile plot suggested the data were not inconsistent with the AFT assumption. Where deviations did occur, these were observed at the extremes of the follow-up period, either within the first or last few months of the trial period.

Based on the assessment of PH and AFT, and in line with NICE DSU TSD14 guidance, a series of parametric models were deemed suitable to fit to the trial data.[15] The exponential, generalised gamma, Gompertz, log-logistic, log-normal and Weibull distributions were all explored. Both adjusted and unadjusted models were considered in order to determine which would be most appropriate.

Unadjusted models

Initially, unadjusted survival models were explored, including only dapagliflozin as a variable in separating the survival extrapolations. The Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) values for each of the unadjusted survival models for CV and allcause mortality are presented in Table 35 and Table 36, below.

Table 35: AIC and BIC values of the unadjusted parametric survival model distributions for CV mortality derived from the DELIVER ITT population

Abbreviations: AIC: Akaike Information Criterion; BIC: Bayesian Information Criterion; CV: cardiovascular; ITT: intention-to-treat.

Table 36: AIC and BIC values of the unadjusted parametric survival model distributions for all-cause mortality derived from the DELIVER ITT population