==> picture [215 x 22] intentionally omitted <==

Single Technology Appraisal

Empagliflozin for treating chronic kidney disease [ID6131]

Committee Papers

==> picture [215 x 22] intentionally omitted <==

NATIONAL INSTITUTE FOR HEALTH AND CARE EXCELLENCE

SINGLE TECHNOLOGY APPRAISAL

Empagliflozin for treating chronic kidney disease [ID6131]

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 Boehringer Ingelheim: a. Full submission

• b. Summary of Information for Patients (SIP)

2. Clarification questions and company responses

• a. Initial clarification response

• b. Further clarification response

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

• a. Diabetes UK

• b. Kidney Care UK

• c. Kidney Research UK

• d. UK Kidney Association

4. External Assessment Report prepared by Warwick Evidence

5. External Assessment Group response to factual accuracy check of EAR

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

Empagliflozin for treating chronic kidney disease [ID6131] Document B

Company evidence submission

29[th] June 2023

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 1 of 160

Contents

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 2 of 160

==> picture [451 x 494] intentionally omitted <==

----- Start of picture text -----

----- End of picture text -----

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 3 of 160

List of Tables

Table 1. The decision problem .............................................................................................................. 10 Table 2. Technology being evaluated ................................................................................................... 14 Table 3 . NICE criteria for ungraded CKD diagnosis ............................................................................. 17 Table 4 . Classification of CKD based on eGFR and uACR categories ................................................ 18 Table 5. UK Renal Registry (UKRR) CKD by aetiology in 2020 ........................................................... 19 Table 6 . Mortality rate per 100 person-years among CKD patients by eGFR and uACR .................... 22 Table 7. CV event rate per 100 person-years in CKD patients by eGFR & uACR ............................... 23 Table 8 . Incidence rate of cause-specific hospitalisations for patients with and without CKD ............. 23 Table 9. Annual eGFR and uACR monitoring requirement for patients with or at risk of CKD ............ 29 Table 10. UKKA Quick Reference Guide for implementation in people with CKD with or without T2DM .............................................................................................................................................................. 31 Table 11. Clinical effectiveness evidence ............................................................................................. 39 Table 12. Inclusion and exclusion criteria of the EMPA-KIDNEY trial .................................................. 45 Table 13. The EMPA-KIDNEY trial drugs ............................................................................................. 46 Table 14. Pre-specified primary and secondary outcomes ................................................................... 46 Table 15. Demographic and baseline characteristics of randomised patients in the EMPA-KIDNEY trial ......................................................................................................................................................... 48 Table 16. Summary of statistical analysis in the EMPA-KIDNEY trial .................................................. 49 Table 17. Results of the critical appraisal of EMPA-KIDNEY trial ........................................................ 53 Table 18: Causes of CKD in EMPA-KIDNEY vs UKKR report ............................................................. 54 Table 19 Summary of key differences in baseline characteristics between EMPA-KIDNEY and DAPACKD trials .............................................................................................................................................. 69 Table 20. Summary of the trials used to conduct the indirect treatment comparison ........................... 73 Table 21. Overall summary of AE - TS ................................................................................................. 80 Table 22. Serious and pre-specified non-serious AE with frequency ˃2% - TS ................................... 80 Table 23. Summary of AESI and specific AE – TS ............................................................................... 81 Table 24. KDIGO classification health states incorporated in the model .............................................. 88 Table 25. Complications incorporated in the model .............................................................................. 89 Table 26. Features of the economic analysis ....................................................................................... 91 Table 27. Baseline characteristics for full cohort, diabetics, and non-diabetics ................................... 93 Table 28. Key articles identified in targeted literature search for eGFR and uACR ............................. 95 Table 29. Annual eGFR change in patients receiving empagliflozin and SoC ..................................... 96 Table 30. Percentage of patients reporting specific eGFR slopes........................................................ 97 Table 31. Change in the uACR values of empagliflozin and SoC over one year ................................. 99 Table 32. Coefficients of the Framingham risk progression equations for TC, HDL, SBP (Framingham progression) ........................................................................................................................................ 101 Table 33. Coefficients of the risk progression equations for HbA1c and BMI in patients with DM (UKPDS 90)......................................................................................................................................... 102 Table 34. Coefficients of the risk progression equation for HbA1c in patients without DM (Framingham Offspring cohort) ................................................................................................................................. 102 Table 35. Lower limb amputation rates per 100 patient-years ............................................................ 104 Table 36. Summary of utility values for cost-effectiveness analysis ................................................... 107 Table 37. Calculation of SoC weighted average cost ......................................................................... 109 Table 38: Cost of treatment with empagliflozin and SoC .................................................................... 110 Table 39. Maintenance costs per health state as per KDIGO classification (hospitalisation and critical care costs excluded) ........................................................................................................................... 111 Table 40. Acute event and follow-up costs for management of CVD complications .......................... 112 Table 41. Cost of BMD ........................................................................................................................ 113 Table 42. Cost of AKI .......................................................................................................................... 113 Table 43. Cost of infections................................................................................................................. 114 Table 44. Cost of incident cancer ....................................................................................................... 114 Table 45. Cost of other complications ................................................................................................. 114 Table 46. Cost of ESKD ...................................................................................................................... 116 Table 47. Cost of managing lower limb amputations .......................................................................... 116 Table 48 Submodule-wise application of costs in the model .............................................................. 116 Table 49. Base-case model settings ................................................................................................... 119 Table 50. Base-case variables applied in cost-effective analysis ....................................................... 120

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 4 of 160

Table 51. Base-case: deterministic cost-effectiveness analysis results for empagliflozin as add on to SoC compared to SoC only................................................................................................................. 132 Table 52. Net health benefit ................................................................................................................ 132 Table 53: Cost-comparison for empagliflozin (intervention) versus dapagliflozin (comparator) ......... 133 Table 54 . Base-case: probabilistic cost-effectiveness analysis results for empagliflozin as add on to SoC compared to SoC only................................................................................................................. 134 Table 55: Deterministic one-way sensitivity analysis inputs and results ............................................ 135 Table 56: Scenario analyses: ICERs for empagliflozin on top of SoC compared to SoC alone ......... 139 List of Figures

Figure 1. Patient population addressed in the submission, according to KIDGO categories ................. 9 Figure 2. Current clinical pathway for CKD in the UK ........................................................................... 28 Figure 3. Expected positioning of empagliflozin in the UK treatment pathway ..................................... 34 Figure 4. The EMPA-KIDNEY trial design ............................................................................................ 43 Figure 5. CONSORT diagram of patient flow in each stage of the EMPA-KIDNEY trial ...................... 53 Figure 6. Time to the first event of kidney disease progression or adjudicated CV death, estimated cumulative incidence function (considering non-CV/renal death as a competing risk) – RS ............... 55 Figure 7. Sensitivity analyses of the time to the first event of kidney disease progression or adjudicated CV death – RS ................................................................................................................... 56 Figure 8. Time to events of ACH, mean cumulative function – RS ....................................................... 57 Figure 9. Sensitivity analyses of time to ACH and adjudicated death – RS ......................................... 57 Figure 10: Time to death comparing patients with and without hospitalisations – RS ......................... 58 Figure 11. Time to the first event of adjudicated HHF or adjudicated CV death, estimated cumulative incidence function (considering non-CV death as a competing risk) – RS ........................................... 59 Figure 12. Forest plot of sensitivity analyses of time to the first event of adjudicated HHF or adjudicated CV death – RS ................................................................................................................... 59 Figure 13. Time to adjudicated death from any cause, Kaplan-Meier estimate – RS .......................... 60 Figure 14. Forest plot of sensitivity analyses of time to adjudicated death from any cause – RS ........ 61 Figure 15. Time to the first event of kidney disease progression, estimated cumulative incidence function (considering non-renal death as a competing risk) – RS ........................................................ 62 Figure 16. Estimated cumulative incidence function for time to adjudicated CV death (non-CV death as competing risk)-RS ........................................................................................................................... 63 Figure 17. Time to adjudicated CV death or ESKD, estimated cumulative incidence function (considering non-CV death as a competing risk) – RS ......................................................................... 64 Figure 18. Change from Baseline in the eGFR ..................................................................................... 65 Figure 19. Forest plot for key pre-specified subgroup analyses of time to the first event of kidney disease progression or adjudicated CV death ...................................................................................... 67 Figure 20. Effect of SGLT2 inhibitors on kidney disease progression by DM status and eGFR .......... 78 Figure 21. Effect of SGLT2 inhibitors on kidney disease progression by DM status and uACR .......... 79 Figure 22. CKD Markov microsimulation model schematic .................................................................. 90 Figure 23. Histogram showing distribution of eGFR slope in the CKD-PC population (Naimark et al) 97 Figure 24. Histogram showing uACR fold change in the CKD-PC population ................................... 100 Figure 25. Distribution of patients per uACR fold change ................................................................... 101 Figure 26. Mean annual per patient healthcare costs for the overall CKD cohort, stratified by KDIGO classification group .............................................................................................................................. 111 Figure 27. PSA cost-effectiveness plane for empagliflozin on top of SoC ......................................... 134 Figure 28. PSA cost-effectiveness acceptability curve for empagliflozin on top of SoC .................... 135 Figure 29: Deterministic OWSA tornado diagram for ICER results .................................................... 138

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 5 of 160

Abbreviations

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 6 of 160

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 7 of 160

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

B.1.1 Decision problem

This submission covers a population of adults with chronic kidney disease (CKD) (patients with or without type 2 diabetes mellitus [T2DM], with a broad range of estimated glomerular filtration rate [eGFR] from 20 to 90 mL/min/1.73m[2] , and varying levels of albuminuria). This population falls within the full anticipated marketing authorisation for empagliflozin in this indication i.e., for the treatment of adults with CKD.

EMPA-KIDNEY is the pivotal randomised controlled trial (RCT) assessing the effect of empagliflozin 10mg oral once daily (OD) versus matching placebo on top of standard of care (SoC) on the progression of kidney disease in a broad population of CKD patients at risk of further disease progression. This submission further addresses the cost-effectiveness, comparative effectiveness, clinical efficacy, and safety of empagliflozin versus SoC in adult patients with CKD in alignment with the final National Institute for Health and Care Excellence (NICE) scope as outlined in Table 1.

The intention to treat (ITT) population comprised of 6,609 randomised CKD patients with heterogenous baseline characteristics, including 46% and 54% with or without diabetes (DM), and 26% and 74% of patients with or without history of cardiovascular disease (CVD) at baseline respectively (1, 2). Mean eGFR at baseline was 37.4 mL/min/1.73m[2] , with 34% of patients having eGFR <30 mL/min/1.73m[2] ; 44% with eGFR ≥30 to <45 mL/min/1.73m[2] ; and 22% with eGFR ≥45 mL/min/1.73m[2] . Median urinary albumin-to-creatinine ratio (uACR) at baseline was 329 mg/g (37.2 mg/mmol), with 20% of patients having uACR <30 mg/g (3 mg/mmol; A1); 28% having uACR ≥30 to ≤300 mg/g (≥3 to ≤30 mg/mmol; A2); and 52% having uACR >300 mg/g (30 mg/mmol; A3) (2). EMPA-KIDNEY is the first CKD trial to include patients with low or no albuminuria (uACR <22.6 mg/mmol).

Empagliflozin is currently recommended by NICE for the treatment of adult patients with T2DM [NICE appraisal TA336] (3) and adult patients with heart failure (HF) with reduced ejection fraction (HFrEF), i.e., HF with left ventricular ejection fraction (LVEF) ≤40% [NICE appraisal TA773] (4), within the National Health Service (NHS) England. This means empagliflozin is already recommended for adult patients with comorbid CKD within these populations within the relevant marketing authorisations. Other disease-modifying sodium-glucose cotransporter 2 (SGLT2) inhibitors are already recommended in CKD/diabetic kidney disease (DKD), as per NICE Clinical Guideline NG203, NICE Clinical Guideline NG28, and NICE appraisal TA775 (dapagliflozin) (5-7).

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 8 of 160

However, recommendations are limited to patients with certain eGFR and uACR thresholds and/or T2DM, based on available evidence from pivotal RCTs (8, 9).

This submission provides evidence to support the inclusion of empagliflozin in NICE Clinical Guideline NG203 as an SGLT2 inhibitor treatment option for a broader population of adults with CKD (patients with or without T2DM, with a broad range of eGFR from 20 to 90 mL/min/1.73m[2] and varying levels of albuminuria) in line with the EMPA-KIDNEY ITT population and supporting data from EMPA-REG OUTCOME, thus addressing an important unmet need for patients who fall outside the scope of current recommendations (Figure 1).

Figure 1. Patient population addressed in the submission, according to KIDGO categories

==> picture [471 x 182] intentionally omitted <==

Abbreviations: CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; T2DM, type 2 diabetes mellitus

Source: Adapted from KDIGO 2013 (10)

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 9 of 160

Table 1. The decision problem

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131]

© Boehringer Ingelheim Ltd (2023). All rights reserved Page 10 of 160

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131]

© Boehringer Ingelheim Ltd (2023). All rights reserved Page 11 of 160

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131]

© Boehringer Ingelheim Ltd (2023). All rights reserved Page 12 of 160

==> picture [83 x 105] intentionally omitted <==

A positive NICE recommendation for empagliflozin in CKD that facilitates broad access for patients across primary and secondary care settings and the multidisciplinary care team can help alleviate health inequalities, as CKD patients may be seen by a variety of specialists in clinical practice. Broad access is important for alleviating any health inequalities in terms of access to nephroprotective treatments for CKD patients.

Abbreviations: ACE, Angiotensin-converting enzyme; ARB, Angiotensin receptor blocker; CI, confidence interval; CKD, chronic kidney disease; CV, cardiovascular; CVD, cardiovascular disease; eGFR, estimated glomerular filtration rate; ESKD, end-stage kidney failure; HR, hazard ratio; ITT, intention to treat; NICE, National Institute for Health and Care Excellence; NHS, National Health Service; N/A, not applicable; T2D, Type 2 diabetes; uACR, urine albumin-to-creatinine ratio.

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131]

© Boehringer Ingelheim Ltd (2023). All rights reserved

B.1.2 Description of the technology being evaluated

A description of the technology being evaluated is presented in Table 2. The current summary of product characteristics (SmPC) for empagliflozin is available in Appendix C.

Table 2. Technology being evaluated

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 14 of 160

Abbreviations: CKD, chronic kidney disease; EMA, European Medical Agency; HbA1c, glycated haemoglobin; MHRA, Medicines and Healthcare products Regulatory Agency; OD; Once daily; SGLT2, Sodium-glucose cotransporter 2; SmPC, Summary of product characteristics; T2DM; Type 2 diabetes Mellitus; UK, United Kingdom.

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 15 of 160

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

treatment pathway

Summary of health condition and position of the technology in the treatment pathway

• CKD is a chronic progressive condition in which kidney function deteriorates over time, potentially leading to end-stage kidney disease (ESKD) (17, 18).

• The 2021 NICE CKD guidelines (NG203) define CKD as a condition in which kidney structural or functional abnormalities are present for ≥3 months and has implications on health. This includes all people with markers of kidney damage and those with an eGFR of <60 mL/min/1.73m[2] on at least two occasions separated by a period of ≥90 days (with or without markers of kidney damage) (7).

• CKD may result from various underlying systemic conditions and primary kidney diseases, with T2DM and hypertension cited as the most common causes (19).

• The prevalence of CKD in the UK is estimated at 3.5 million (37), although a new report by Kidney Research UK (KRUK) suggests up to 7.19 million may be affected (38).It is projected that CKD may affect 14% of the total UK population by 2025 (20).

• The NHS Quality and Outcomes Framework (QoF) 2021-2022 estimated the prevalence of CKD stages G3a to G5 as 3.98% (1.9 million) among adults aged ≥18 years in England (21).

• • CKD is associated with a considerable number of comorbidities and complications that increase the disease burden. Patients with CKD are at a higher risk of experiencing adverse CV (HF, coronary artery disease, stroke, MI) and non-CV events (anaemia, infections, metabolic and bone mineral disorders [BMD]) as compared to non-CKD patients (22, 23).

• • Patients with CKD demonstrate higher rates of hospitalisation as compared to patients without CKD (24). In the UK, patients with CKD have reported two-to-three-fold increased rates of emergency hospitalisation as compared to non-CKD patients, and a 2-fold higher risk of hospital readmission or death within 30 days of discharge across all disease stages (25).

• This burden of disease imposes substantial costs to the NHS, with RRT costs estimated to be £32,259 per patient per year (PPPY) for dialysis and £27,033 for the initial cost of renal transplantation (20).

• • A 2009-2010 study estimates direct healthcare costs for CKD at £1.45 billion per annum (half of which accruing to RRT), representing approximately 1.3% of the total NHS budget (26).

• • CKD also impacts patients’ QoL, with the extent of decrement varying by disease stage, treatment modality, and patients’ comorbidity profile (27).

• Patients with CKD’s risk of all-cause and CV mortality increases with eGFR decline and increasing albuminuria (28). This risk is evident in early stages of CKD and increases as the disease progresses to later stages with particularly poor outcomes for patients with ESKD receiving RRT (29).

• • Management of CKD in the NHS is informed by NICE clinical guidelines (NG203 and NG28) and NICE-accredited guidelines from the UK Kidney Association (UKKA), which can be considered as being informed by the latest clinical evidence available (5, 7).

• • NICE CKD guidelines (NG203) currently recommend SGLT2 inhibitors in select CKD patients, who meet uACR thresholds and/or have T2DM. It is evident that the NG203 (Chronic kidney disease: assessment and management) guideline needs to be revised to incorporate a thorough and concise overview of all relevant recommendations on SGLT2 inhibitors in patients with CKD, with or without T2DM (7).

• • Not all patients with CKD are prescribed ACE inhibitors, ARBs, and statins in UK clinical practice, and patients receiving individually optimised SoC inclusive of these treatments remain at residual risk of CKD disease progression and adverse outcomes (2).

• • The recent 2023 UKKA guidelines recommend SGLT2 inhibitors to slow the rate of kidney function decline in patients with CKD with and without T2DM at broader ranges of eGFR and

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 16 of 160

lower uACR thresholds than earlier NICE guidelines, which is more reflective of evidence from the EMPA-KIDNEY ITT population than other comparator trials (5, 7, 30).

• Based on the existing evidence, empagliflozin should be positioned as a disease-modifying treatment option in combination with individually optimised SoC at the earliest opportunity in adult patients at risk of CKD disease progression.

B.1.3.1 Overview of CKD

CKD is a chronic progressive condition in which kidney function deteriorates over time, potentially leading to end-stage kidney disease

CKD is caused by abnormalities of kidney function or structure that are present for ≥3 months (7, 10, 31). This definition of CKD includes all individuals with markers of kidney damage or those with an eGFR < 60 mL/min/1.73m[2] on at least two separate occasions 90 days apart (with or without markers of kidney damage). Markers of kidney damage can include albuminuria, haematuria, and structural abnormalities detected by imaging, or a history of kidney transplantation (Table 3).

Table 3 . NICE criteria for ungraded CKD diagnosis

Abbreviations: uACR, urine albumin-creatinine ratio; AER, albumin excretion rate; CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate. Source: NICE 2021 (7)

Structural and functional abnormalities of the kidney in CKD lead to progressive deterioration of renal function as measured by eGFR, eventually causing nephron loss and potentially leading to ESKD, also referred to as a kidney failure, necessitating RRT constituted by either dialysis or kidney transplant depending on disease severity (17, 18).

Disease severity in CKD is classified according to Kidney Disease Improving Global Outcomes (KDIGO) group recommendations, which incorporate eGFR and uACR categories into a two-dimensional framework to “stratify (CKD) risk, focus management priorities and guide referral to specialist care” (17). There are 6 eGFR categories ranging from normal to ESKD which are further subdivided by three albuminuria categories (Table 4). NICE CKD guidelines [NG203] highlight that the risk of adverse outcomes in CKD (e.g., all-cause mortality and cardiovascular [CV] events) elevates with increasing uACR and eGFR categories as summarised in Table 4.

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 17 of 160

Table 4 . Classification of CKD based on eGFR and uACR categories

Abbreviations: CKD, chronic kidney disease; ESKD, end-stage kidney disease; eGFR, estimated glomerular filtration rate; uACR, urine albumin-creatinine ratio *No CKD if there are no other markers of kidney damage Source: KDIGO 2012 clinical practice guidelines (10)

CKD has a heterogenous aetiology; however, a common pathophysiology is found across all causes of the condition

CKD results from various systemic conditions and primary kidney diseases, with T2DM and hypertension cited as the most common causes (19). Other less frequent causes include polycystic kidney disease (PKD), obstructive uropathy, and various glomerular nephrotic and nephritic syndromes (32)). While UK data on causation across the eGFR spectrum is limited, a 2020 analysis by UK Renal Registry (UKRR) in patients with eGFR <30 mL/min/1.73m[2 ] receiving RRT demonstrated that the most common identifiable causes were T2DM (30.5%), glomerulonephritis (12.3%) and hypertension (7.1%). T2DM was the most common cause of CKD in all age groups except 18-34, where glomerulonephritis was predominant (Table 5).

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 18 of 160

Table 5. UK Renal Registry (UKRR) CKD by aetiology in 2020

Abbreviations: CKD, chronic kidney disease; T2DM, Type 2 diabetes mellitus Source: UK renal registry 24[th] annual report (33)

Irrespective of the origin of any renal injury, all CKD aetiologies follow a common pathophysiology characterised by progressive, and irreversible loss of nephrons (the core structural and functional units of the kidneys). Nephrons are composed of glomeruli that filter the blood, and tubules that return useful compounds into circulation and excrete waste products via urine. In CKD, nephrons undergo glomerular hypertension and hyperfiltration, triggering a proinflammatory, profibrotic cascade that causes interstitial scarring and nephron failure (34).

Post initial nephron failure, ‘remnant’ functional nephrons also experience glomerular hypertension in an effort to maintain homeostasis, which eventually leads to further nephron failure. Without intervention, this cycle can lead to ESKD where the kidneys are no longer able to perform an adequate level of glomerular filtration at which point RRT may be necessitated (35, 36).

CKD incidence and prevalence increases with age, with evidence of underdiagnosis in earlier disease stages

Estimates of the prevalence of CKD in the UK vary. Kidney Care UK (KCUK) have reported this as an estimated at 3.5 million (37), although a new report by Kidney Research UK (KRUK) suggests this could be much higher (38). It is projected that CKD will affect 14% of the UK population by 2025 (20). Furthermore, an increase of approximately 7% is expected in more advanced stages of CKD (3b-5) relative to the total CKD population (39). The NHS QoF 20212022 estimates the prevalence of CKD at stages G3a to G5 as 3.98% (1.9 million) amongst all adults aged ≥18 years in England (21). Health Survey for England (HSE) (2016) reported a higher prevalence of CKD at stages G3a to G5 of 7% in adults aged ≥35 years specifically (40). HSE (2016) further reported an estimated prevalence of CKD at stages G1 to G5 of 15% overall in this population (40). A 2020 English cohort study estimates a CKD prevalence of 18.2% across adults ≥60 years (41), and HSE 2016 also reported a 34% prevalence in adults aged ≥70 years (21).

Studies from 2020-2022, estimate that 44% of patients with CKD aged ≥60 years, and 48% of patients with CKD aged ≥18 years are undiagnosed in clinical practice respectively (41, 42).

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 19 of 160

Underdiagnosis is common in early CKD stages as patients are often asymptomatic. The diagnostic criterion for such patients is a raised uACR ≥3 mg/mmol or other markers of kidney damage; however, the National CKD Audit reports sub-optimal use of uACR testing among people at high risk of CKD (41, 43), limiting the potential to identify these patients early in clinical settings.

At a national level, an aging UK population is expected to contribute to increased CKD prevalence, while at an individual patient-level, a confluence of increasingly observed risk factors such as hyperglycaemia, hypertension, history of CVD, and obesity are also increasing the risk of developing CKD (44-46). However, there is promising but limited evidence demonstrating trends of improved management in some patient-level risk factors in UK clinical practice (20, 47).

B.1.3.2 Burden of disease in CKD

• CKD is associated with considerable comorbidities and complications that contribute to the disease burden. Patients with CKD are at a higher risk of experiencing adverse CV ( e.g., HF, coronary artery disease, stroke, myocardial infraction [MI]) and non-CV events (e.g., anaemia, infections, and metabolic and bone mineral disorders [BMD]) as compared to non-CKD patients (22, 23).

• • Patients with CKD demonstrate higher rates of hospitalisation as compared to patients without CKD (24). In the UK, patients with CKD have reported two-to-three-fold increased rates of emergency hospitalisation as compared to non-CKD patients, and a 2-fold higher risk of hospital readmission or death within 30 days of discharge across all disease stages (25).

• • This burden of disease imposes substantial costs to the NHS, with RRT costs estimated to be £32,259 PPPY for dialysis and £27,033 for the initial cost of renal transplantation (20).

• • A 2009-2010 study estimates direct healthcare costs for CKD at £1.45 billion per annum (half of which accruing to RRT), representing approximately 1.3% of the total NHS budget (26). This is supported by a recent report by KRUK which projects that NHS costs for CKD stages 1-5 (excluding RRT) will reach £1.95 billion by the end of 2023, representing 1% of the total NHS budget (38).

• • CKD also impacts patients’ QoL, with the extent of decrement varying by disease stage, treatment modality, and patients’ comorbidity profile (27).

• • Patients with CKD’s risk of all-cause and CV mortality increases with eGFR decline and increasing albuminuria (28). This risk is evident in early stages of CKD and increases further as the disease progresses to later stages with particularly poor outcomes for patients with ESKD receiving RRT (29).

CKD is associated with a considerable number of comorbidities and complications which exert an increased disease burden as patients progress towards ESKD and require RRT

Patients with CKD are at higher risk of various complications including anaemia, infections, metabolic disorders (e.g., acidosis, hyperkalaemia) and BMD as compared to patients without CKD (22, 23). The Renal Risk in Derby (RRID) study reported that 19.9% of their cohort of Stage 3 patients with CKD had anaemia (48), the rates of which was reported to increase with advanced disease (49). Metabolic disorders such as hyperphosphatemia, hypocalcaemia, and secondary hyperthyroidism lead to an increased risk of fractures in patients with CKD (22, 23, 50). The RRID study reported that 19.9% of their cohort of Stage 3 CKD patients had anaemia (48), the rates of

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 20 of 160

which was reported to increase with advanced disease (49). Metabolic disorders such as hyperphosphatemia, hypocalcaemia, and secondary hyperthyroidism lead to an increased risk of bone fractures in patients with CKD (22, 50), with comorbidities further contributing to healthcare resource utilisation (HCRU) (24).

Once CKD has progressed to ESKD, RRT becomes the predominant treatment option (10). Since 1990, the global age-standardised incidence of patients requiring RRT has increased by 43.1% (dialysis) and 34.4% (transplant) (51). A patient-level microsimulation modelling study estimated a prevalence of 9.2 million people (diagnosed and estimated undiagnosed) with CKD by 2027 in UK and projected a 5% increase in costs for diagnosed CKD and RRT (52). The treatment of CKD and ESKD imposes substantial societal costs, which are highest for RRT, particularly in-hospital haemodialysis (HD) (53). In 2019, of the 57,510 ESKD patients receiving RRT in England, it was found that 32,367 (56.3%) of them received a kidney transplant, 20,759 (36.1%) received in-centre haemodialysis (ICHD), 3,175 (5.5%) received peritoneal dialysis (PD) and 1,209 (2.1%) home HD, further contributing to extensive HCRU and thus costs (29). In many western countries, at home HD, PD, and self-care dialysis are more cost-effective than in-hospital haemodialysis. However, in Europe, 89% of patients still underwent dialysis in secondary care settings in 2013, and 14% of UK patients received hospital PD (53).

A comprehensive analysis of the costs of dialysis undertaken in the UK in 2018 reported the annual direct cost per patient for home-based modalities to be £16,395 for continuous ambulatory peritoneal dialysis (CAPD), £20,295 for automated PD (APD), and £23,403 for home-based HD. The cost of dialysis was increased at £28,931 for satellite units and £32,678 for hospital units, including costs for transportation (25).Beyond the economic burden, patients with CKD undergoing RRT are also subjected to potential risks associated with treatment, which include haemolysis due to cannulas, air embolisms due to HD and continuous RRT, vascular access related infections, risk of severe blood loss, risk from fluid overload during APD, and risk of medicine-induced nephrotoxicity or systemic side effects caused by poor metabolite excretion resulting from reduced kidney function (54).

The 2023 KRUK report also highlights existing NHS resource constraints for RRT, whereby maximum capacity has likely been reached. This was compounded by a suspension of nonelective surgical care during COVID-19 in 2020, impacting kidney transplantation appointments and causing an estimated loss of 1,600 opportunities for kidney transplant surgeries. This led to concomitant increases in waiting lists and the proportion of patients receiving dialysis as a form of RRT (increased from 91.7% to 94.1%). Future growth in demand for RRT led by ongoing epidemiological trends in CKD has the potential to exacerbate current capacity constraints (38).

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 21 of 160

Patients with CKD are at a higher risk of all-cause mortality, poor CV outcomes, and CV death than the general population

Increased risk of all-cause mortality and CV mortality is evident from early-stage CKD and increases further as the disease progresses to later stages, with particularly poor mortality outcomes for patients with ESKD receiving RRT (28, 29). The RRID study reported an all-cause mortality rate of 41.2% by the end of the 5-year follow-up for Stage 3 CKD patients (55). UKRR (2019) reports a total death rate of 91 per 1,000 prevalent RRT patients (29). Further, DISCOVER CKD (2022), an international observational cohort study in patients with CKD reported an increased risk of all-cause mortality and CV mortality with decreasing eGFR levels and increasing uACR in English adults not receiving dialysis (Table 6) (28).

Table 6 . Mortality rate per 100 person-years among CKD patients by eGFR and uACR

Abbreviations: CKD, Chronic kidney disease; CV, cardiovascular; eGFR, estimated glomerular filtration rate; N/A, not applicable; uACR, urine albumin-creatinine ratio Source: James et al. 2022 (28)

Patients with CKD also face increased risk of adverse CV events and outcomes including HF, coronary artery disease (CAD), peripheral vascular disease (PVD), stroke and MI (23, 28, 56), with DISCOVER CKD reporting on the increasing risk of stroke and MI in particular with higher uACR and lower eGFR categories (Table 7) (28).

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 22 of 160

Table 7. CV event rate per 100 person-years in CKD patients by eGFR & uACR

Abbreviations: CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; MI, myocardial infarction; N/A, not applicable; uACR, urine albumin-creatinine ratio Source: James et al. 2022 (28)

High CV and non-CV disease burden contributes to a higher rate of higher rate of allcause hospitalisations (ACH) in CKD patients

Multimorbidity in CKD leads to complex treatment regimens. Progressive eGFR decline in CKD often becomes a limiting factor in the management of comorbidities, which increases patients’ risk of hospitalisation compared to the general population. Iwagami et al. (2018) examined cause-specific hospitalisation rates for English CKD patients stages 3-5 from 2004 to 2014 to demonstrate a higher incidence rate of hospitalisation for patients with CKD compared to those without CKD (Table 8) (24). Hospitalisation for heart failure (HHF) demonstrated the largest difference with patients with CKD having a 9.7% hospitalisation incidence compared to 3.1% for patients without CKD.

Table 8 . Incidence rate of cause-specific hospitalisations for patients with and without CKD

Abbreviations: CKD, chronic kidney disease; HHF, hospitalisation for heart failure Source: Iwagami et al. 2018 (24).

In the UK, two-to-three-fold increased rates of emergency hospitalisation have been reported in patients with CKD as compared to those without the condition (25). CKD patients have almost a 2-fold higher risk of hospital readmission or death within 30 days of discharge across all disease

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 23 of 160

stages and approximately 28% of patients with CKD stages 4 and 5 are readmitted or die within 30 days of discharge (57). Similar results have been observed globally:

In the United States of America (US), ACH were found to be 4 times higher in CKD patients aged 18-64 years at any stage and 12 times higher in those with CKD Stage 4 or 5 compared to people without CKD (57). German adults with Stage 3 CKD demonstrated hospitalisation rates four times higher than those without CKD (58), and a Canadian study reported that more than 40% of patients with CKD aged >65 with unplanned hospitalisation die within 5 years regardless of admission cause (59). ACH has proven to be an indicator for long-term risk of death; thus, hospitalisations represent a substantial morbidity and mortality risk in CKD (59).

Increased HCRU and hospitalisations in CKD lead to a substantial economic burden on the NHS

CKD is a progressive disease that is associated with significant morbidity and increased risk of hospitalisations, which imposes a substantial economic burden on the healthcare system. A 20092010 estimate suggests that £1.45 billion was spent by the NHS on CKD, which comprised approximately 1.3% of the total NHS budget. A June 2023 report by KRUK projects that CKD stages 1-5 will cost the NHS £1.95 billion by the end of 2023 (excluding RRT). This alone represents approximately 1% of the total NHS budget, with 91% (£1.79 billion) of these costs attributable to CKD stages 3-5, and 9% (£167 million) attributable to CKD stages 1-2 (38).

Increased hospitalisations observed in CKD patients compared to the general population are a key driver of this economic burden (26). Morbidities can lead to a 3-fold surge in total cost of care as compared to CKD without additional conditions. At each stage of the disease, CKD in combination with comorbidities such as DM, CVD, and HF significantly amplify healthcare expenditures, with this effect steadily increasing as the disease progresses to later stages (60, 61). Hospitalisations are the key drivers for annual healthcare costs for CKD in the US and account for up to 65% of the total cost of CKD associated care (60, 61).

DISCOVER CKD assessed HCRU and costs stratified by CKD severity to report that hospitalisation rates were three times higher in the A3 uACR category than for A1. Hospitalisations increased to 889.7/1000 patient-years for those with stages 4 and 5 CKD (20). Mean annual per patient costs ranged from £4,966 (A1) to £9,196 (A3), and from £4,997 (G2) to £7,595 (G5) in the UK, establishing that the economic burden of healthcare in CKD is weighted towards a small proportion of patients with late-stage CKD, including those with kidney failure and/or albuminuria (21). DISCOVER CKD study further found that across all disease stages mean NHS healthcare costs PPPY were £5,401 for patients with CKD (21). Costs were found to increase as CKD stage increases, with G2A1 averaging £4,654 and G4A3 £11,419.

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 24 of 160

CKD progression to ESKD and RRT further increase the substantial economic burden on the NHS

Once patients with CKD progress to ESKD, RRT is the remaining treatment option (10) and although only a smaller proportion of patients advance to ESKD, RRT is associated with a substantial economic burden on the health system. In the US, dialysis and kidney transplants accounted for approximately $36 billion of total healthcare expenditure as per the US Renal Data System (USRDS) 2022 annual report. Per this report, PPPY inpatient and outpatient costs for patients with CKD were comparable at approximately $25,000 and $28,000 respectively, however dialysis contributed to the majority (92.7%) of total outpatient costs at $9.88 billion per year. Further, average PPPY costs for RRT were considerably higher, and estimated at $98,410 and $442,500 PPPY for dialysis and kidney transplants respectively (62).

Similarly, RRT is associated with a substantial economic burden on the NHS at £32,259 PPPY and £27,033 PPPY for dialysis and initial kidney transplant respectively (20). Of the £1.45 billion spent on treatment of CKD stages 3-5 in England in 2009-2010, more than 50% was spent on RRT, which was required for just 2% of the CKD population (26). KRUK (2023) further project that the cost of dialysis for people with ESKD will reach £1.05 billion (or 0.53% of the NHS budget), and that the cost of kidney transplants will reach £293 million by the end of 2023 (38). These estimates collectively demonstrate the need to prevent or delay CKD progression to reduce the economic burden on the NHS that is associated with later disease stages.

CKD is associated with a high symptomatic burden and decrements in health-related quality of life (HRQoL), particularly in later disease stages

CKD leads to decrements in patients’ HRQoL that increase depending on disease stage, treatment modality, and patients’ comorbidity profile (27). A study by Nguyen et. al (2018) used 2010 HSE data to show statistically significant decrements in health utility index of 0.11, 0.19 and 0.28 for patients with G2, G3a and G4/5 respectively, compared to patients with normal kidney function or at stage G1 (56). This is aligned with a 2020 systematic literature review (SLR) of HRQoL in CKD that reported a decline in UK-specific EQ-5D-3L-derived utilities of 0.85 at G2 to 0.73 at G5.

Nguyen et. al (2018) demonstrated that pain/discomfort and mobility were the HRQoL domains for which patients with CKD reported the most problems (56). This is supported by findings from a 2020 English prospective cohort study, where the EQ-5D-5L domains reported most common problems to be pain/discomfort, mobility, and usual activities (70.6%, 57.7%, and 46.2% of patients reporting any problem respectively). This study also found that higher comorbidity count and obesity are independently associated with patient-reported problems in these domains, demonstrating the importance of comorbidity management in improving the HRQoL of patients with CKD (55).

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 25 of 160

HRQoL is lowest for CKD patients requiring RRT, with utilities of 0.44 and 0.53 reported for patients on HD and PD respectively (63). Dialysis’ detrimental impact on HRQoL has been attributed to the travel and dietary restrictions it imposes, as well as the copious amount of time patients need to spend in dialysis units. Such patients often require support with mobility and transportation, as well as with personal care, exerting a considerable burden on their carers and support networks (64). Kidney transplants have conversely been associated with positive effects on HRQoL (65).

CKD also impacts patients’ and carers lived experiences, patients’ perspective on their health, and their ability to perform normal activities of daily life

A 2022 survey conducted by the European Kidney Patients Federation (EKPF) on the impact of CKD on patient’s lives revealed that 88% experienced some kind of life change due to their condition (66). Negative impacts reported included reduced energy for doing things they enjoyed earlier (42%), anxiety (34%), worry about losing their independence (33%), worry about the burden inflicted on friends/family (32%), feeling sick (27%) and being more irritable around friends/family (24%). Nearly 77% of survey respondents mentioned an impact of CKD on their work life and career, with 25% reporting reduced productivity, concern about their future earnings and loss of drive and ambition. The most common symptom experienced by patients was fatigue (62% of all respondents increasing to 75% by Stage 4 or 5). Additionally, more than half (55%) with severe CKD experienced at least five symptoms, reflecting the increased impact of the disease in advanced stages.

Overall, most patients (96%) had concerns about their future due to CKD; the most common being poor QoL (40%), fear of overall health getting worse (39%), and being a burden on their family/friends (37%). In the UK, 55% found it hard to cope emotionally with the impact of the disease and 63% had lower self-esteem, thus negatively impacting their relationships. Providing optimal care for both young and older patients with any chronic disease is challenging and in turn imposes substantial burden on their caregivers, especially immediate family (67, 68). Additionally, factors such as relationship between caregiver and patient; behavioural, and psychological symptoms displayed by the patient; patient’s gender; and adverse events (AEs) impact caregiver burden (69).

B.1.3.3 Current clinical pathway of CKD in the UK

• Management of CKD in the NHS is informed by NICE clinical guidelines (NG203 and NG28) and NICE-accredited guidelines from the UKKA. Being the most recent, UKKA guidelines can be interpreted as being informed by the latest clinical evidence available (5, 7, 30).

• CKD screening is recommended in adults using eGFR and uACR if applicable risk factors exist, and annual eGFR monitoring is recommended for patients taking calcineurin inhibitors (e.g., ciclosporin), lithium, or on long-term anti-steroidal anti-inflammatory drugs (Nonsteroidal anti-inflammatory drugs [NSAIDs]) (7, 70).

• However, diagnosis typically occurs as an incidental finding in primary care during basic metabolic panels or routine eGFR and uACR testing as early CKD is often asymptomatic (7, 70).

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 26 of 160

• The primary goal of CKD management is to prevent or delay progression to higher disease stages and ESKD, therefore reducing the risk of complications and adverse mortality outcomes (5, 7).

• Current established clinical practice in the NHS is individually optimised for each patient, wherein clinical management includes treatments directly modifying treatment progression such as renin-angiotensin-aldosterone system (RAAS) inhibitors and SGLT2 inhibitors, along with CVD risk management and comorbidity management, as well as management of CKD complications if required (5, 7, 30).

• NICE CKD guidelines (NG203) currently recommend SGLT2 inhibitors in selected CKD patients, who meet uACR thresholds and/or have T2DM. It is evident that the NG203 (Chronic kidney disease: assessment and management) guideline needs to be revised to incorporate a thorough and concise overview of all relevant recommendations on SGLT2 inhibitors in patients with CKD, with or without T2DM, and with or without albuminuria (7).

• • If CKD progresses to stages 4 or 5 (7); assessments for RRT should be made at least 1 year before the patient reaches ESKD as per NICE guidelines NG107 (71).

• • The 2023 UKKA guidelines make new recommendations for SGLT2 inhibitors to slow the rate of kidney function decline in CKD patients with and without T2DM at broader ranges of eGFR and lower uACR thresholds than existing guidelines, which is more reflective of evidence from the EMPA-KIDNEY ITT population than comparator trials (5, 7, 30).

Management of CKD in the NHS is informed by NICE clinical guidelines (Figure 2) and NICEaccredited guidelines from the UKKA. NICE Clinical Guideline 203 [NG203] for the assessment and management of people with or at risk of CKD was published in August 2021 (7), and an update of NICE Clinical Guideline 28 [NG28] was published in November 2021 which included recommendations for T2DM patients with comorbid CKD (5). UKKA guidelines for CKD management were published in May 2023. While these three guidelines were published within a 2-year period, they make differing recommendations for the role of SGLT2 inhibitors in CKD due to the fast-evolving publication and interpretation of clinical trial evidence for the drug class in this indication. As they are most recent, UKKA guidelines can be interpreted as being informed by the latest clinical evidence available.

CKD is commonly diagnosed as an incidental finding in primary care, and there is evidence of low adherence to uACR testing and monitoring guidance in UK clinical practice.

CKD screening is recommended in adults using eGFR and uACR if applicable risk factors are found, and annual eGFR monitoring is recommended for patients taking calcineurin inhibitors (e.g., ciclosporin), lithium, or on long-term anti-steroidal anti-inflammatory drugs (NSAIDs) (7, 70). However, diagnosis commonly occurs in primary care setting as an incidental finding during basic metabolic investigations or routine eGFR and uACR testing as early CKD is often asymptomatic Figure 2. Nearly 50-70% of CKD patients are diagnosed during screening for other conditions (31, 32, 70). The remaining 30-50% of CKD patients are then usually identified via specific CKD screening.

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 27 of 160

Figure 2. Current clinical pathway for CKD in the UK

==> picture [452 x 332] intentionally omitted <==

Abbreviations: ACEis, Angiotensin-converting enzyme inhibitors; ARB, Angiotensin receptor blockers; CKD, chronic kidney disease; CV, cardiovascular; eGFR, estimated glomerular filtration; ESA, Erythropoiesis-stimulating agents; ESKD, end-stage kidney disease; HTN, hypertension; NSAIDs, Nonsteroidal anti-inflammatory drugs; RRT, Renal replacement therapy; SoC, standard of care; T2DM, Type 2 diabetes mellitus; uACR, urine albuminto-creatinine ratio,

a Abnormalities of kidney function or structure present for more than three months, with implications for health. This includes all people with markers of kidney damage and those with an eGFR <60 mL/min/1.73m[2] on at least two occasions separated by a period of at least 90 days (with or without markers of kidney damage).

b The 2021 draft NICE guidelines for the treatment of CKD also recommend the use of SGLT2 inhibitors in patients with T2DM, if they meet the criteria in the relevant marketing authorisation.

c Measured using the 4-variable Kidney Failure Risk Equation. Source: Adapted from DAPA appraisal [TA775] (6)

A CKD diagnosis is made if patients are found to have repeated measures of reduced eGFR <60 mL/min/1.73m[2] or increased uACR ≥3 mg/mmol at least 3 months apart (7). Further investigations may be performed to establish aetiology and KDIGO stage (10) and support individually optimised treatment plans to mitigate against the risk of adverse outcomes at each stage (7), including CT scans, renal biopsies, urea nitrogen tests, and Cystatin C (7, 10).

As referral criteria are strict, patients typically do not consult a specialist until their CKD has progressed to at least stage 3 (72). Scenarios where specialist referral is appropriate are a 5-year risk of needing RRT >5% (measured by the 4-variable ESKD Risk Equation); a uACR ≥70 mg/mmol (unless caused by DM and already treated), a uACR >30 mg/mmol with haematuria; a

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 28 of 160

sustained decrease in eGFR ≥25% and a change in eGFR category within 12 months; a sustained decrease in eGFR of ≥15 mL/min/1.73m[2] per year; poorly controlled hypertension; known or suspected rare or genetic cause of CKD; and suspected renal artery stenosis (7).

Post diagnosis, individualised annual monitoring plans are required to identify any further disease progression. NG203 suggests minimum number of annual eGFR and uACR tests dependent on KDIGO category based on KDIGO 2012 recommendations, with patients with more advanced CKD subject to more frequent monitoring based on risk (7, 10) (Table 9).

Table 9 . Annual eGFR and uACR monitoring requirement for patients with or at risk of CKD

Abbreviation: CKD, chronic kidney disease; eGFR, estimated Glomerular filtration rate; uACR, urine albumin-tocreatinine ratio Source: NG203 guidelines (7)

However, rates of uACR testing and monitoring are low in UK clinical practice: A 2023 Hospital Episode Statistics (HES) linked Clinical Practice Datalink (CPRD) Aurum observational study found only 45% of adults with CKD had a uACR measurement between January 2010 to December 2019. DISCOVER CKD further reported a low frequency of uACR testing in clinical practice with less than 10% of patients in the base cohort with 2 eGFR measures having a uACR measurement (28). Low rates of uACR testing contribute to most CKD diagnoses occurring at Stage 3 or above, as diagnosis of early disease is dependent on uACR while eGFR remains ≥60 mL/min/1.73m[2] .

Treatment goals in CKD are to prevent or delay disease progression and reduce the risk of complications and AEs

The primary goal of CKD management is to prevent or delay progression to higher disease stages and ESKD, therefore reducing the risk of complications and adverse mortality outcomes (5, 7). UK clinical practice focusses on management of comorbidities including CVD (statins, anticoagulants, antiplatelet therapy), T2DM (antidiabetic medication), and hypertension (antihypertensive medication) leading to a complex treatment pathway with polypharmacy due to the need to meet multiple treatment goals. Disease-modifying therapy for CKD includes renin-angiotensin-aldosterone system (RAAS) inhibition (angiotensin-converting enzyme [ACE] inhibitors and angiotensin-receptor blockers [ARB]) and more recently SGLT2 inhibition as per NG203, NG28, and UKKA guidelines (Figure 2) (5, 7, 30).

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 29 of 160

SGLT2 inhibitors represent a significant advancement in disease-modifying treatment options for CKD patients, and 2023 UKKA guidelines reflect the latest clinical evidence

SGLT2 inhibitors are currently recommended as an adjunct to individually optimised SoC, in order to slow disease progression. Of note, NG203 (Chronic Kidney Disease: assessment and management) does not make direct recommendations on the use of SGLT2 inhibitors in patients with CKD, but instead directs readers to follow recommendations made in NG28 (Type 2 diabetes in adults: management) or TA775 (NICE technology appraisal guidance on dapagliflozin for treating chronic kidney disease). It is evident that the NG203 guidelines need to be revised to incorporate a thorough and concise overview of all relevant recommendations on SGLT2 inhibitors in patients with CKD, with or without T2D.

NG28 recommends that adults with T2D and CKD, who are taking an ARB or an ACE inhibitor (titrated to the highest licensed dose that they can tolerate), are offered an SGLT2 inhibitor (in addition to the ARB or ACE inhibitor) if their uACR is over 30 mg/mmol and they meet the criteria in the marketing authorisation (including relevant eGFR thresholds) (5). The guidelines also suggest that an SGLT2 inhibitor is considered for patients with both T2D and CKD (who are on an ARB or an ACE inhibitor at the highest tolerated dose), provided their uACR is between 3 and 30 mg/mmol (subject to meeting the criteria in the marketing authorisation).

TA775 recommends the SGLT2 inhibitor dapagliflozin (Forxiga[®] ) as an option for treating CKD in adults if it is an add-on to optimised SoC including the highest tolerated licensed dose of ACE inhibitor or ARB, unless contraindicated, and people have an eGFR of 25 mL/min/1.73m[2] to 75 mL/min/1.73m[2] at the start of treatment and have T2D or have a uACR of 22.6 mg/mmol or more (6).

NG203, NG28 and TA775 do not recommend SGLT2 inhibitor use in CKD patients without either T2DM or albuminuria. However, the new UKKA Clinical Practice Guideline: Sodium-Glucose Cotransporter-2 (SGLT-2) Inhibition in Adults with Kidney Disease (published May 2023) makes recommendations to use SGLT2 inhibitors to slow the rate of kidney function decline in CKD patients with and without T2DM at broader ranges of uACR and eGFR (30).

In summary, the UKKA guidelines recommend SGLT2 inhibitors are initiated in CKD patients with/without T2D in the following scenarios:

• An eGFR 20-45 mL/min/1.73m[2, ] irrespective of albuminuria

• An eGFR >45 and uACR of ≥25 mg/mmol.

Furthermore, the guidelines suggest:

• SGLT2 inhibitors are initiated in patients with T2DM and eGFR >45–60 mL/min/1.73m without albuminuria (uACR <25 mg/mmol)

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 30 of 160

• Clinicians consider SGLT2 initiation in patients with eGFR <20mL/min/1.73m[2] to slow progression of kidney disease.

The UKKA guidelines offer the following Quick Reference Guide for implementation in people with CKD with or without T2DM (Table 10).

Table 10. UKKA Quick Reference Guide for implementation in people with CKD with or without T2DM

Abbreviations: eGFR, estimated glomerular filtration rate; SGLT2, sodium glucose co-transporter 2; UKKA, United Kingdom Kidney Association *People with type 1 diabetes, polycystic kidney disease, or kidney transplant excluded from the definitive trials. † In this guideline we do not make recommendations on the use of SGLT2 inhibition to reduce kidney disease progression for people with eGFR ≥60 mL/min/1.73m[2] and uACR <25 mg/mmol as this is outside the scope of this guideline. However, we support the use of SGLT2 inhibitors in this population for relevant indications, including treatment of people with heart failure and reduction of cardiovascular risk in people with T2D at high CV risk. ‡ Further research recommended in people with kidney replacement therapy to establish the role SGLT2 inhibition in these patients. Source: UKKA Guideline 2023 (30)

UKKA guidelines reflect the latest clinical evidence for SGLT2 inhibitors in CKD, and data from both the pivotal EMPA-KIDNEY and DAPA-CKD trials were considered in their development. EMPA-KIDNEY incorporated a broader range of CKD patients as part of its inclusion criteria (eGFR ≥20 and <45mL/min/1.73m[2] ; or eGFR ≥45 and <90mL/min/1.73m[2 ] with uACR ≥22.6 mg/mmol) compared to DAPA-CKD (eGFR ≥25 and <75mL/min/1.73m[2] with uACR ≥22.6 to <565 mg/mmol). The recommendations made by UKKA are closely aligned with evidence of the efficacy of empagliflozin in the ITT population of EMPA-KIDNEY, indicating scope to prescribe empagliflozin in patients who would not be eligible for dapagliflozin but have residual risk of CKD disease progression.

RRT is required as a life-supporting treatment if CKD progresses to ESKD

A cohort of patients will still progress to ESKD despite individually optimised SoC and SGLT2 inhibition. If CKD disease cannot be managed appropriately and the disease progresses to stages 4 or 5 (usually regarded as pre-dialysis) (7); assessments for RRT to maintain normal homeostatic processes should begin before they reach Stage 5 or ESKD. NICE guidelines NG107 outline that assessments should be made at least 1 year before the patient reaches ESKD when RRT is likely to be needed (71). RRT options include HD, haemofiltration, haemodiafiltration, PD, and kidney

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 31 of 160

transplantation. Patients may also be given the choice of conservative management which may reflect patient preferences in cases where HRQoL or life expectancy are poor (7, 71).

B.1.3.4 Limitations of current pathway and unmet need

• There is no standard response criterion in CKD management and all CKD patients receiving individually optimised SoC maintain a residual risk of disease progression and adverse outcomes, irrespective of disease stage (2).

• ACE inhibitors and ARBs are not universally included in the SoC across all patients in UK clinical practice, and intolerability precludes prescription in some patients with CKD (5, 7, 30).

• Current NICE recommendations for SGLT2 inhibitors in the management of CKD exclude a broad range of CKD patients, including those without T2D or albuminuria, and/or with certain eGFR ranges (5, 7, 30).

• There is an unmet need for disease-modifying treatment options in CKD patients who remain at risk of disease progression, in particular those without access to SGLT2 inhibitors and those who cannot tolerate an ACE inhibitor or ARB.

Patients receiving individually optimised SoC including ACE inhibitors, ARBs and statins remain at residual risk of CKD disease progression and adverse outcomes

There is no standard response criterion in CKD management and all patients with CKD maintain a residual risk of disease progression and adverse outcomes irrespective of disease stage. This residual risk exists for patients receiving individually optimised SoC, as demonstrated in the EMPA-KIDNEY trial (which included patients receiving ACE inhibitors, ARBs and statins in both arms), where progression of CKD or death from CV causes occurred in 16.9% of patients in the placebo on top of SoC group (2).

There is evidence of lack of universal uptake of ACE inhibitors, ARBs, and statins in CKD patients in UK clinical practice

ACE inhibitors, ARBs and statins constitute part of SoC, with RAAS inhibitors classed as diseasemodifying CKD treatment options as per TA775 (Figure 2) (6). However, DISCOVER CKD demonstrated that only 51.5% of patients with CKD were prescribed an ACE inhibitor or ARB in the UK CPRD cohort (28) due to issues with tolerability and contraindications (e.g., in patients at risk of hypotension or hyperkalaemia) (73), with Boehringer Ingelheim (BI) prescription data suggesting that ********************************************************************************************

******************* (74). Statins are recommended for reducing CV risk in patients with CKD who have a 10-year risk of developing CVD of ≥10%, and for secondary prevention in patients with CKD and established CVD (7, 75). However, DISCOVER CKD also reported only 52.9% of patients with CKD as receiving this drug class (28). As such, a significant proportion of patients with CKD are not receiving the SoC in UK clinical practice.

NG203 and TA775 restrict access to SGLT2 inhibitors for CKD patients with T2DM or albuminuria meaning a proportion do not have access to disease-modifying therapy

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 32 of 160

There is a substantial proportion of patients to whom SGLT2 inhibitors are currently not available. BI prescription data (Q1 2023) suggest ***************************************************************** ****************************************************************** (74). This demonstrates that use of SGLT2 inhibitors is not yet well established in clinical practice, with scope to expand the use of this drug class in this indication to slow CKD disease progression.

Prior to EMPA-KIDNEY, SGLT2 inhibitor trials did not include a broad range of CKD patients (8, 9). Thus, based on available evidence, dapagliflozin is recommended for CKD patients with an eGFR between 25 to 75 mL/min/1.73m[2] , with comorbid T2DM or a uACR ≥22.6 mg/mmol (6). Further, the CREDENCE trial restricted inclusion only to CKD patients with comorbid T2DM, uACR >30 mg/mmol and eGFR of 30 to <90 mL/min/1.73m[2] . In contrast, EMPA-KIDNEY included patients with or without comorbid T2DM, eGFR of ≥20 to <45 mL/min1.73m[2] , or eGFR of ≥45 to <90 mL/min/1.73m[2] and a uACR of ≥ 22.6 mg/mmol, representing an opportunity to extend SGLT2 inhibitors to a broader range of CKD patients not investigated in previous RCTs or covered by existing NICE recommendations.

The 2023 UKKA guidelines recommend SGLT2 inhibitor use in CKD patients with and without T2DM over a broader range of eGFR and lower uACR thresholds than NG203 and TA775, which is more reflective of evidence from the EMPA-KIDNEY ITT population than comparator trials. A NICE recommendation for empagliflozin in line with the ITT population will address an important unmet need for patients currently not able to access SGLT2 treatment for CKD and reflect UKKA guideline recommendations.

B.1.3.5 Expected positioning of empagliflozin in the UK treatment pathway

Empagliflozin should be positioned as a disease-modifying treatment option in combination with individually optimised SoC at the earliest opportunity in adult patients at risk of CKD disease progression (

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 33 of 160

Figure 3).

CKD is usually treated and monitored in primary care settings (7, 70, 76), however, empagliflozin should be made available in both primary and secondary care to ensure the broadest population of patients are able to gain access.

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 34 of 160

Figure 3. Expected positioning of empagliflozin in the UK treatment pathway

==> picture [357 x 252] intentionally omitted <==

Abbreviations: ACE, Angiotensin-converting enzyme; ARB, Angiotensin receptor blockers; CKD, chronic kidney disease; CV, cardiovascular; HTN, hypertension; SoC, standard of care; T2DM, Type 2 Diabetes mellitus Source: Adapted from NICE guideline 2021 NG203 (7)

BI seek a recommendation for empagliflozin in adults with CKD in line with the pivotal EMPAKidney trial ITT population, plus a broader recommendation in CKD patients with comorbid T2DM, as an add-on to individually optimised standard care including ACE inhibitors or ARBs, unless these are contraindicated or not tolerated, and people have either, at the start of treatment:

• an eGFR of 20 mL/min/1.73m[2] up to 45 mL/min/1.73m[2] or

• an eGFR of 45 mL/min/1.73m[2] up to 90 mL/min/1.73m[2] and either:

  • a uACR of 22.6 mg/mmol or more or

  • T2DM.

A positive recommendation for empagliflozin in the above CKD population will provide additional options for patients and address an important unmet need for a disease-modifying option for CKD patients in England and Wales not currently covered by existing NICE recommendations.

B.1.4 Equality considerations

A positive recommendation for empagliflozin enabling broad access for patients across primary and secondary care settings and the multidisciplinary care team can help address inequalities in access to care for CKD patients, particularly in areas with limited presence of secondary and

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 35 of 160

specialist care facilities. A 2016 review outlined the pivotal role that primary care has in preventing the progression of disease and complications amongst CKD patients (48).

Principle 9 of NICE’s Social Value judgements as part of its statement highlights the goal to reduce health inequalities across protected characteristics as well as considering those arising from socioeconomic factors (11). Socioeconomic disparities are associated with health inequalities in England, with more socially advantaged patients often receiving better access to secondary and specialist care in the NHS (12). There are demonstrable distributional health inequalities in terms of the risk and impact of CKD, with some groups more disadvantaged by the condition than others:

There is evidence of reduced access to dialysis services in rural areas, and people from areas with higher social deprivation are more likely to develop CKD, progress to more severe disease stages, and experience lower life expectancy with the condition compared to those from less socially deprived areas. Further, people from Black and South Asian backgrounds are three to five times more likely to require dialysis and on average wait between 168 to 262 days longer for kidney transplantation than those from a Caucasian background (77).

Resource constraints in a post-COVID-19 healthcare system may further exacerbate pre-existing inequalities in access to secondary and specialist care in the NHS. Secondary care in CKD is largely focussed on patients with ESKD, and barriers in ease and affordability of travel may further complicate access to these settings.

A NICE recommendation should allow broad access to empagliflozin for adults with CKD across primary and secondary care settings, as well as across the full multidisciplinary team as CKD patients may be seen by a variety of specialists in the NHS. Broad access is important for potentially improving the quality of care (thus reducing the burden of complications) and for alleviating any health inequalities in terms of access to nephroprotective treatments for CKD patients. Further, the availability of an additional SGLT2 inhibitor for patients with CKD could help to overcome potential supply issues that may limit access to treatments (78). Importantly, the availability of an additional SGLT2 inhibitor enables patient and physician choice.

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 36 of 160

B.2 Clinical effectiveness

B.2.1 Identification and selection of relevant studies

A SLR was conducted to identify data on RCTs reporting on the efficacy and safety of potential comparators, namely SGLT2 inhibitors and finerenone for the treatment of adult patients with CKD/ DKD. The SLR was supplemented by a targeted literature review (TLR) to identify relevant observational studies that could supplement RCT evidence. The methods used to identify the relevant clinical evidence are described in Appendix D.

MEDLINE[®] , Embase[®] and Cochrane Central Register of Controlled Trials and Cochrane Database of Systematic Reviews (via Ovid) were performed on 03 October 2022. This was supplemented with a desk search of conference proceedings from last 5 years (2019 to 2022 meetings). The TLR Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 37 of 160

searches were performed on the same date as SLR, using MEDLINE[®] and Embase[® ] (via Ovid) as well as manual checking of bibliography lists of any relevant SLRs identified in the database search.

The eligible studies encompassed all RCTs evaluating efficacy of pharmacological interventions (SGLT2 inhibitors and finerenone) used in the treatment of adults (age ≥18 years) with CKD. The search strategy was designed to be broad and to encompass all interventions that currently comprise the SoC, as well as recently approved interventions and investigational agents for the management of CKD (eligibility criteria are shown in Table 8 and Table 9 of Appendix D). All studies meeting the pre-specified population, intervention, comparator, outcomes, and study (PICOS) eligibility criteria were retained. Data extraction was done in a pre-defined data extraction template (DET) in MS Excel[®] to capture the data elements of interest from each included study.

The SLR included three relevant trials of empagliflozin (EMPEROR-Reduced, EMPERORPreserved and EMPA-REG OUTCOME). The pivotal trial, EMPA-KIDNEY trial, compared oral empagliflozin 10mg OD on top of SoC versus matching placebo on top of SoC, in patients with established CKD. This is the primary source of clinical evidence in the economic model. At the time of performance, the EMPA-KIDNEY trial publication had not yet been published, thus information for EMPA-KIDNEY was taken from confidential clinical study reports. A full list of studies that were included and excluded during the SLR is provided in Table 10 and Table 11 of Appendix D.

B.2.2 List of relevant clinical effectiveness evidence

As discussed in above Section B.2.1 Identification and selection of relevant studies , the SLR included four trials investigating the efficacy of empagliflozin in CKD (Table 11). The pivotal trial for empagliflozin in this indication is EMPA-KIDNEY, a phase III, randomised, double-blind, placebo-controlled trial that compared empagliflozin 10mg OD on top of SoC (n=3,304) to matched placebo on top of SoC (n=3,305) for the treatment of CKD in patients with and without comorbid T2DM (2, 79, 80). EMPA-KIDNEY is described in full in the following sections.

The EMPA-REG OUTCOME (81), EMPEROR-Reduced (82), and EMPEROR-Preserved (83) trials also evaluated the efficacy of empagliflozin; however in these trials, patients with CKD were subpopulations of the main population (Table 11 Error! Reference source not found. ) and included patients with uACR and eGFR values outside of the EMPA-KIDNEY eligibility criteria who were at less advanced stages of CKD, i.e. moderate to high risk according to KDIGO. Results of a supplementary efficacy analysis on a large population of patients from these trials complement the findings from the EMPA-KIDNEY trial (predominantly in patients at very high risk according to KDIGO) by providing evidence in renal endpoints and support the generalisability of the EMPAKIDNEY results to a broad population of patients at various stages of CKD. Thus, these trials

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 38 of 160

provide additional information and evidence relevant to the decision problem. For the full supplementary analysis, refer to Appendix M.

Despite the usefulness of these trials in demonstrating efficacy in the broad CKD population, the primary objectives of these trials were not directly relevant to the original decision problem, therefore were not included in economic model. The pivotal trial, EMPA-KIDNEY, was the main source of clinical efficacy evidence in the cost-utility model described in Section Error! Reference source not found. .

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 39 of 160

Table 11. Clinical effectiveness evidence

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131]

© Boehringer Ingelheim Ltd (2023). All rights reserved Page 40 of 160

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131]

© Boehringer Ingelheim Ltd (2023). All rights reserved Page 41 of 160

Abbreviations: ACH, all-cause hospitalisations; ARN, angiotensin receptor-neprilysin; CKD, chronic kidney disease; CV, cardiovascular; DM, diabetes mellitus; EF, ejection fraction; eGFR, estimated glomerular filtration rate; ESKD, end-stage kidney disease; EQ-5D, EuroQol- 5 Dimension; HbA1c, glycated haemoglobin, HF, heart failure; HHF, hospitalisation for heart failure; LVEF, left ventricular ejection fraction; MACE, major adverse cardiovascular events; MI, myocardial infraction; MRA, aldosterone receptor antagonist; N/A, not applicable; OD, once daily; PICO, patient intervention comparator outcome; RAS, renin-angiotensin system; SoC, standard of care; T2DM, type 2 diabetes mellitus; uACR, urine albumin-creatinine ratio.

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 42 of 160

B.2.3 Summary of methodology of the relevant clinical effectiveness

evidence

The EMPA-KIDNEY trial investigated the effect of empagliflozin 10mg oral OD on top of SoC versus matching placebo on top of SoC on the combined risk of kidney disease progression and CV death in patients with CKD. The rationale for testing empagliflozin in the full range of people at risk of CKD progression was based on findings from the EMPA-REG OUTCOME trial (84) and other clinical and experimental data, which showed SGLT2 inhibitors induce glycosuria and lower BP and albuminuria (85-87). Empagliflozin had also been shown to have haemodynamic effects in the kidney in the absence of elevated blood glucose i.e., in people without DM (88). In summary, it was hypothesised that empagliflozin may have beneficial effects on kidney disease progression and CV risk among those with CKD, irrespective of the presence of DM. Thus, the EMPA-KIDNEY trial aimed to include large numbers of patients without DM, patients with an eGFR <30mL/min/1.73m[2] , and patients with low levels of proteinuria (measured by the uACR) to assess the safety and efficacy of empagliflozin in a broad range of people with CKD with and without DM (2).

The enrolled patients were broadly representative of the population of patients with CKD who are at risk for disease progression (2). The applicability of the trial results to NHS clinical practice is discussed further in Section B.2.5.1 Applicability to clinical practice .

B.2.3.1 EMPA-KIDNEY trial design

EMPA-KIDNEY was a large, international, multicentre, phase III, double-blinded, randomised placebo-controlled trial conducted between February 2019 and July 2022 assessing the effect empagliflozin on top of addition to SoC on the progression of kidney disease and CVD in a wide range of patients with CKD with or without DM (2). Patients were allocated using a minimised randomisation algorithm to receive either oral empagliflozin 10 mg OD or matching placebo in a 1:1 ratio. The algorithm helped ensure balance between the treatment groups with respect to the following prognostic variables: age, sex, prior DM, eGFR and uACR (both based on local laboratory results at screening) (79).

In total, 8,544 potentially eligible patients were screened. Post screening, 8,184 patients entered an 8-12 week ‘Run-in’ period prior to randomisation, during which they received single-blind placebo tablets. This was to ensure that only those patients likely to continue taking study treatment for an extended period were randomised and provided investigators with an opportunity to review and approve the participation of each participant to ensure they were on appropriate background therapy (including RAAS blockade). In total 6,609 patients were ultimately randomised to

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 43 of 160

empagliflozin 10 mg OD (n=3,304) or matching placebo (n=3,305). See Figure 4 for schematic depiction of the trial design and Figure 5 for a CONSORT diagram illustrating the patient flow (2).

Figure 4. The EMPA-KIDNEY trial design

==> picture [468 x 170] intentionally omitted <==

Abbreviations: RAS, Renin-angiotensin system; SoC, Standard of care *Guideline directed medical therapy. Source: Adapted from EMPA-KIDNEY Clinical trial report (CTR) (79)

The pre-specified composite primary outcome was the first occurrence of progression of kidney disease or death from CV causes. Progression of kidney disease was defined as:

1. ESKD; the initiation of maintenance dialysis or receipt of a kidney transplant,

2. a sustained decrease in the eGFR to <10 mL/min/1.73m[2] , a sustained decrease from baseline in the eGFR of at least 40%, or

3. death from renal causes.

Key secondary outcomes were:

• ACH (first and recurrent combined),

• First occurrence of HHF or CV death, and

• All-cause mortality (2).

A formal interim analysis to decide whether to stop the trial for benefit was planned to be made after 150 ESKD events (chronic dialysis, transplant) had occurred, by which time it was expected that approximately 60% of all first primary outcomes had occurred. At the time of the interim analysis, 624 (58% of 1070) primary outcome events had occurred (interim database lock 22[nd] Feb 2022) and two conditions were required to be met:

• Hazard ratio for the primary outcome of <0.778 with a two-sided p-value of <0.0017 and

• Hazard ratio for the outcome of ESKD (chronic dialysis, transplant) or CV death (other secondary outcome) of <0.778 with a two-sided p-value of <0.05.

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 44 of 160

Furthermore, in March 2022, the Independent DMC recommended to stop the trial early due to positive efficacy after both of these pre-specified conditions were met (79).

B.2.3.2 Eligibility criteria for study participants

Key inclusion and exclusion criteria of EMPA-KIDNEY are listed in Table 12. Eligible participants were adult CKD patients with or without comorbid DM. The number of patients with or without DM (of any type) was intended to be at least one-third of each, and the number of patients with an eGFR ≥45 mL/min/1.73m[2 ] was limited to about one-third. All patients were required to provide written informed consent (79).

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 45 of 160

Table 12. Inclusion and exclusion criteria of the EMPA-KIDNEY trial

Abbreviations: ACE, Angiotensin-converting enzyme; ALT, alanine transaminase; ARB, angiotensin receptor-II blocker; AST, aspartate transaminase; CKD, chronic kidney disease; CKD-EPI, chronic kidney disease epidemiological collaboration; DRI, direct renin inhibitor; eGFR, estimated glomerular filtration rate; RAS, renin-angiotensin system; SGLT, sodium-glucose cotransporter; T1DM, type 1 diabetes mellitus; T2DM, type 2 diabetes mellitus; uACR, urine albumin-to-creatinine ratio, ULN, upper limit of normal. *Based on self-reports at screening and randomisation visits.

†Myocardial infarction, angina, stroke, or peripheral arterial disease (including lower limb amputation)

‡ As of January 2020, the protocol was amended to allow currently enrolled patients with T1D to continue in the study and limit screening of new patients with T1D due to a sponsor decision. At that time, the Data Monitoring Committee (DMC) did not report any safety concerns in patients with T1D. Source: EMPA-KIDNEY Collaborative Group 2022 (1)

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131]

© Boehringer Ingelheim Ltd (2023). All rights reserved

B.2.3.3 Settings and locations where data were collected

EMPA-KIDNEY was a multicentre study conducted in 241 centres in eight countries across North America, Europe, and Asia (2). Of the 6,609 randomised patients, 2,648 (40%) were from Europe; 1,717 (26%) from North America, 1,632 (25%) from China and Malaysia, and 612 (9%) from Japan (79).

B.2.3.4 Trial drugs and concomitant medications

Study interventions are summarised in Table 13. Disallowed concomitant medications included any SGLT2 inhibitors or combined sodium-glucose cotransporter 1 (SGLT1) and 2 inhibitors, except the blinded trial medication. Women of childbearing potential had to agree to use highly effective contraception throughout the trial and for 7 days after the end of the trial (79).

Table 13. The EMPA-KIDNEY trial drugs

Source: The EMPA-KIDNEY Collaborative Group (2)

B.2.3.5 Pre-specified primary and secondary outcomes of EMPA-KIDNEY

The primary and secondary outcomes of the EMPA-KIDNEY trial are shown in Table 14 below.

Table 14. Pre-specified primary and secondary outcomes

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 47 of 160

Abbreviations: ACE, all-cause hospitalisations; AE, adverse events; CV, cardiovascular; eGFR, estimated glomerular filtration rate; ESKD, end-stage kidney disease; EQ-5D, EuroQol 5 dimensions; EQ-VAS, EuroQol visual analogue scores; HHF, hospitalisation for heart failure; HRQoL, health-related quality of life; SAE, serious adverse event *ESKD - defined as the initiation of maintenance dialysis or receipt of a kidney transplant #The HRQoL results are presented in Appendix N Source: The EMPA-KIDNEY Collaborative Group 2023 (2); EMPA-KIDNEY CTR (79)

Central, blinded adjudication was performed by trained clinicians based at the Central Coordinating Office, Clinical Trial Service Unit and Epidemiological Studies Unit (CTSU) at the University of Oxford for death events and events initially reported as HHF, myocardial infarction, stroke, liver injury, ketoacidosis, lower limb amputation, acute kidney injury (AKI), and genital infections. Only events that have been confirmed or not refuted by adjudication were included in the relevant analyses (79).

B.2.3.6 Pre-specified subgroup analyses

The subgroups of key interest were DM status, baseline eGFR, and baseline uACR. Furthermore, the primary outcome was subject to subgroup analyses for age, sex, region, ethnicity, baseline blood pressure, baseline body mass index (BMI), history of prior disease (including CVD), cause of CKD, baseline laboratory values, and medication use at randomisation (2).

B.2.3.7 Demographics and baseline characteristics

Patients in the empagliflozin and placebo groups were well balanced with respect to demographic and clinical characteristics at baseline (Table 15). The patient population broadly represented those with CKD who are at risk for disease progression. Approximately two-thirds of the patients (66.8%) were men, 54.6% of the patients were ≥65 years old, including 23.0% of patients ≥75 years old. The majority of patients had a baseline eGFR equivalent to Stage 3 CKD (≥30 to <60 mL/min/1.73m[2] ; 44.3% and 13.4% had an eGFR of ≥30 to <45 and ≥45 to <60 mL/min/1.73m[2] respectively), followed by Stage 4 (34.5%; eGFR <30 mL/min/1.73m[2] ). Mean eGFR at baseline was 37.37 ± 14.48 mL/min/1.73m[2] for the empagliflozin group and 37.26 ± 14.42 mL/min/1.73m[2] for the placebo group. A marginal majority of patients in both the groups (approximately 52% in each) had macroalbuminuria. Median uACR at baseline was 329.35 mg/g (37.22 mg/mmol);

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 48 of 160

330.58 mg/g (37.36 mg/mmol) for the empagliflozin group and 327.26 mg/g (36.98 mg/mmol) for the placebo group (2, 79).

The majority (54.0%) of patients were non-diabetic while, 44.4% had comorbid T2DM (empagliflozin: 44.5%, placebo: 44.4%) achieving the protocol requirement of including at least one-third with DM and one-third without DM. Over a quarter of patients had comorbid CVD (empagliflozin: 26.1%, placebo: 27.4%) and over 10% had comorbid myocardial infarction (10.6%, in each empagliflozin and placebo groups). The use of concomitant medications was generally well balanced across the treatment groups. The most common non-study medication of interest at baseline was RAS-inhibitors (empagliflozin: 85.7%, placebo: 84.6%) (2, 79).

Table 15. Demographic and baseline characteristics of randomised patients in the EMPAKIDNEY trial

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 49 of 160

Abbreviations: CI, confidence interval; DM, diabetes mellitus; eGFR, estimated glomerular filtration rate; IQR, interquartile range; no/N, number; NT pro-BNP, N-terminal prohormone of brain natriuretic peptide; SD, standard deviation.

*Percentages may not total 100 because of rounding. † Race was reported by the patients. The “other” category indicates that the race was not specified, or the patient preferred not to answer. ¶ The body mass index is the weight in kilograms divided by the square of the height in metres.

‡ History of DM was defined as a patient-reported history of DM of any type, use of glucose-lowering medication, or a glycated haemoglobin level of at least 48 mmol per mole (6.5%) at the randomisation visit.

§ History of cardiovascular disease was defined as a patient-reported history of myocardial infarction, heart failure, stroke, transient ischaemic attack, or peripheral arterial disease.

‖ The values represent the measurement recorded at the randomisation visit or the most recent local laboratory result recorded before randomisation. ** For the urinary albumin-to-creatinine ratio, albumin was measured in milligrams and creatinine was measured in grams. Source: The EMPA-KIDNEY Collaborative Group 2023 (2).

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

relevant clinical effectiveness evidence

The statistical analysis methods and definitions of study groups used in the pivotal EMPA-KIDNEY trial are described in Table 16.

B.2.4.1 Statistical methods and analysis sets

Table 16. Summary of statistical analysis in the EMPA-KIDNEY trial

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131]

© Boehringer Ingelheim Ltd (2023). All rights reserved

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 51 of 160

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 52 of 160

Abbreviations: ACH, all-cause hospitalisations; AE, adverse event; AESI, adverse events of special interest; CI, confidence interval; CV, cardiovascular; DM, diabetes mellitus; eGFR, estimated glomerular filtration rate; ESKD, end-stage kidney disease; HHF, hospitalisation for heart failure; HF, heart failure; HR, hazard ratio; ITT, intention to treat; LCC, Local clinical centre; LLA, lower limb amputation; MMRM, mixed model repeated measure analysis; OS-OT, observed case on treatment; OC-AD, random-set observed case including data after treatment discontinuation; OC-OT, treated set observed case on treatment; uACR, urine albumin-to-creatinine ratio Source: EMPA-KIDNEY CTR 2022 (79)

B.2.4.2 Participant flow in the relevant randomised controlled trials

A total of 8,544 potential participants attended a screening visit; 8,184 patients (95.8%) entered the pre-randomisation run-in phase, and 6,609 underwent randomisation (2). Of the 6,609 randomised patients (3,304 and 3,305 patients to empagliflozin and placebo arms, respectively), 6,568 patients (99.4%) completed the trial, including 315 patients who died (79). Of the 6,609 patients treated with study medication, 1,603 patients prematurely discontinued treatment (24.3%, including patients who died). The most common reasons for premature discontinuation of study medication were AEs (7.3%) and unknown, i.e., largely consisting of patients who provided no early treatment discontinuation reason but attended the final follow-up with a treatment stop date >1 day prior to the visit date (9.6%). Participant flow in the EMPAKIDNEY trial is shown in Figure 5 (79).

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 53 of 160

Figure 5. CONSORT diagram of patient flow in each stage of the EMPA-KIDNEY trial

==> picture [326 x 234] intentionally omitted <==

Abbreviation: AE, adverse event

Note: Complete follow-up defined as death before April 01, 2022 (start of final follow-up window) or completed final follow-up with last known alive after April 01, 2022 Source: Adapted from the EMPA-KIDNEY Collaborative Group 2022 (2)

B.2.5 Critical appraisal of the relevant clinical effectiveness evidence

A summary of the critical appraisal of EMPA-KIDNEY, a parallel group RCT, is shown in Table 17. The complete critical appraisal is provided in Appendix D.

Table 17. Results of the critical appraisal of EMPA-KIDNEY trial

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 54 of 160

B.2.5.1 Applicability to clinical practice

EMPA-KIDNEY is the largest phase III trial investigating SGLT2 inhibitors in CKD so far and had broad eligibility criteria encompassing adult patients with CKD previously underrepresented in other SGLT2 inhibitor trials. Prior SGLT2 inhibitor trials have excluded the broad range of CKD patients, particularly those with lower eGFR and uACR (8, 9), whereas in contrast, EMPA-KIDNEY included patients with or without comorbid T2DM, eGFR of ≥20 to <45 mL/min1.73m[2] , or eGFR of ≥45 to <90 mL/min/1.73m[2] and a uACR of ≥200mg/g (22.6 mg/mmol) (2).

In EMPA-KIDNEY, SoC was more aligned to clinical practice than other SGLT2 inhibitor trials in CKD. Physicians were responsible for ensuring individually optimised SoC was in place for each participant, including management of CV risk factors and other existing comorbidities (e.g., hypertension, T2DM etc.), and to ensure appropriate RAAS-inhibition was in place (i.e., ACE inhibitor or ARB), unless such treatment was either not tolerated or not indicated, making EMPAKIDNEY’s results applicable to patients with CKD seen in clinical settings.

As reported in section B.1.3.4, approximately 52–*** of CKD patients in the NHS receive an ACE inhibitor or ARB. On the contrary, all or most patients in prior SGLT2 inhibitor CKD trials to date have received an ACE inhibitor or ARB (100% in CREDENCE and 97% in DAPA-CKD) (8, 9, 89). As a result, at approximately 85%, the rate of ACE inhibitor or ARB use in EMPA-KIDNEY is more representative of UK clinical practice than SoC in other SGLT2 inhibitor trials (2). EMPA-KIDNEY also included patients with a broad range of underlying causes of CKD, which reasonably correspond to causes of CKD as reported in the UKRR (amongst patients with eGFR <30 mL/min/1.73m[2 ] receiving RRT) (Table 18).

Further, clinical expert opinion concluded that whilst there are inherent differences between the EMPA-KIDNEY population and the real-world NHS CKD population, the patient population in EMPA-KIDNEY is similar to those seen in NHS clinical practice, and the results of the trial are generalisable to the UK clinical setting (Appendix O). Of note, the target CKD population in this submission is patients who meet the EMPA-KIDNEY renal inclusion criteria.

In summary, the EMPA-KIDNEY trial represents a highly relevant extension of the evidence of SGLT2 inhibition across a broad range of CKD patients, representative of those seen in NHS clinical practice.

Table 18: Causes of CKD in EMPA-KIDNEY vs UKKR report

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 55 of 160

Abbreviations: DKD, diabetic kidney disease; PKD, polycystic kidney disease Sources: UK renal registry 24[th] annual report (33), EMPA-Kidney Collaborative Group (2022) (2)

B.2.6 Clinical effectiveness results of the relevant studies

B.2.6.1 Primary outcome: kidney disease progression or CV death

Empagliflozin significantly reduced the relative risk of the primary outcome by 28% compared with placebo (Absolute Risk Reduction [ARR] 3.6%)

Empagliflozin significantly reduced the risk of progression of kidney disease or death from CV causes, which occurred in 432 patients (13.1%) of the empagliflozin group and 558 patients (16.9%) of the placebo group (HR 0.72; 95% CI 0.64 to 0.82; p<0.001). Due to the pragmatic design and limited follow-up visits, the first clear evidence of separation of the estimated cumulative incidence of kidney disease progression or CV death between empagliflozin and placebo became evident approximately 1 year after randomisation, however may have been sooner if data were collected at earlier visits, and continued to separate throughout the 2½ years of follow-up time until the number of patients at risk became too low to provide stable estimates (Figure 6) (2).

Figure 6. Time to the first event of kidney disease progression or adjudicated CV death, estimated cumulative incidence function (considering non-CV/renal death as a competing risk) – RS

==> picture [405 x 259] intentionally omitted <==

Abbreviations: CV, cardiovascular; PY, patient year; RS, randomised set. Source: EMPA-KIDNEY CTR 2022, Figure 11.1.1.1: 1 and Table 11.1.1.1: 1 (79)

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 56 of 160

Results of the individual components of the primary outcome, kidney disease progression and CV death, are detailed in Section B.2.6.3 Tertiary/exploratory outcomes below.

The results of the sensitivity analyses were consistent (i.e., the HR and CIs were similar) with the results of the primary analysis (Figure 7). There was no meaningful effect on the primary analysis results with respect to the presence of COVID-19 AEs.

Figure 7. Sensitivity analyses of the time to the first event of kidney disease progression or adjudicated CV death – RS

==> picture [452 x 135] intentionally omitted <==

Abbreviations: AE, adverse event; CI, confidence interval; eGFR, estimated glomerular filtration rate; RS, randomised set.

*There is no single definition of number of patients with an event as each imputation can produce a different number of events.

Source: EMPA-KIDNEY CTR 2022, Figure 11.1.1.2: 1 (79)

B.2.6.2 Key secondary outcomes

B.2.6.2.1 Time to occurrence of ACH (first and recurrent combined)

Empagliflozin significantly reduced the risk of ACH (first and recurrent combined), including those attributable to renal and CV reasons, compared with placebo

The rate of first and subsequent hospitalisations from any cause was lower in the empagliflozin group than in the placebo group (24.8 vs. 29.2 hospitalisations per 100 patient-years; HR 0.86; 95% CI, 0.78 to 0.95; p=0.0025). The total number of hospitalisation events (first and recurrent) was also lower in the empagliflozin group than in the placebo group (1,611 vs. 1,895). The mean cumulative incidence of ACH in the empagliflozin and placebo groups started to diverge shortly after randomisation and continued to separate over time (Figure 8) (79).

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 57 of 160

Figure 8. Time to events of ACH, mean cumulative function – RS

==> picture [432 x 298] intentionally omitted <==

Abbreviations: ACH, all-cause hospitalisations; PY, patient year; RS, randomised set. Source: The EMPA-KIDNEY Collaborative Group 2023 (2); EMPA-KIDNEY CTR 2022, Figure 11.1.2.3: 1 (79)

The results of the sensitivity analyses were consistent with the overall results (Figure 9).

Figure 9. Sensitivity analyses of time to ACH and adjudicated death – RS

==> picture [452 x 103] intentionally omitted <==

Abbreviations: ACH, all-cause hospitalisations; AE, adverse event; CI, confidence interval; RS, randomised set. Source: EMPA-KIDNEY CTR 2022, 11.1.2.3: 2 (79)

To further explore the impact of hospitalisation, post hoc analysis of survival time from first hospitalisation in EMPA-KIDNEY was performed. The analysis showed that mortality rate in patients with at least one hospitalisation accounted to 25% over the remaining trial duration period. The risk of death was almost ten times higher in patients with hospitalisations compared to those without hospitalisations (HR 9.53; 95% CI 7.18, 12.64; p <0.0001) (Figure 10).

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 58 of 160

Figure 10: Time to death comparing patients with and without hospitalisations – RS

==> picture [396 x 265] intentionally omitted <==

Abbreviations: RS, randomised set. Source: Data on File. Boehringer Ingelheim Ltd. (90)

The reasons for hospitalisations were analysed using AEs leading to hospitalisations per system organ class. AEs leading to hospitalisations were reported most frequently in the system organ classes of infections and infestations, surgical and medical procedures, cardiac disorders, renal and urinary disorders, and investigations, all of which comprised approximately two-thirds of all hospitalisations. Renal and cardiovascular reasons for hospitalisations were analysed based on the respective the system organ classes and additionally based on a user-defined list of renal and cardiovascular AEs leading to hospitalisations. In all analyses, the total number of renal/CV hospitalisations and the proportion of patients with events were lower in the empagliflozin group than in the placebo group. The HR is generally consistent across all analyses.

B.2.6.2.2 Time to first occurrence of HHF or CV death

There was a numerical reduction in the risk of HHF or CV death

There were 131 (4.0%) and 152 (4.6%) patients with any event of this composite outcome in the empagliflozin and placebo groups, respectively; however, the reduction in the risk of HHF or CV death with empagliflozin as compared with placebo did not reach statistical significance (HR, 0.84; 95% CI, 0.67 to 1.07; p=0.1530). This was due to sample power: the trial had a lower absolute risk of CV events and mortality as the enrolled patients were less likely to have these events (>50% patients did not have baseline DM or severe albuminuria and >70% did not have baseline CV

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 59 of 160

disease) (for details see Section B.2.4.1 Statistical methods and analysis sets ). The cumulative incidence of HHF or CV death between empagliflozin and placebo is displayed in Figure 11 (79).

Figure 11. Time to the first event of adjudicated HHF or adjudicated CV death, estimated cumulative incidence function (considering non-CV death as a competing risk) – RS

==> picture [432 x 288] intentionally omitted <==

PY, patient year; RS, randomised set. Source: EMPA-KIDNEY CTR 2022, Figure 11.1.2.2: 1 and Table 11.1.2.2: 1 (79)

The results of sensitivity analyses were consistent with the overall results (Figure 12).

Figure 12. Forest plot of sensitivity analyses of time to the first event of adjudicated HHF or adjudicated CV death – RS

==> picture [452 x 121] intentionally omitted <==

Abbreviations: AE, adverse event; CI, confidence interval; CV, cardiovascular; HHF, hospitalisation for heart failure; RS, randomised set.

Events confirmed or unrefuted by adjudication are considered as an outcome event

*There is no single definition of number of patients with an event because each imputation can produce a different number of events

Source: EMPA-KIDNEY CTR 2022, Figure 15.2.2.1.2: 1 (79)

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 60 of 160

B.2.6.2.3 Time to adjudicated death from any cause

There was a numerical reduction in all-cause mortality for empagliflozin vs. placebo

There were 148 (4.5%) deaths in the empagliflozin group and 167 (5.1%) deaths in the placebo group. However, the reduction in the risk of all-cause death with empagliflozin treatment as compared with placebo did not reach statistical significance (HR 0.87; 95% CI, 0.70 to 1.08; p=0.2137). Historically, it is very difficult to demonstrate all-cause mortality in randomised clinical trials as it can take a long observation time, or a very high-risk population, to achieve enough events to reach statistical significance. Here, a nominal p-value of 0.2137 and a nominal relative risk reduction of 13% trending in the direction of reduced risk in those randomised to empagliflozin versus placebo is observed, which can be explained by a lower than anticipated rate of all-cause death due to the enrolled population, too few events, and too little time, overall contributing to having limited power. The hazard ratios were consistent with the totality of evidence for empagliflozin. The Kaplan-Meier estimates of all-cause death between empagliflozin and placebo is displayed in Figure 13 (79).

Figure 13. Time to adjudicated death from any cause, Kaplan-Meier estimate – RS

==> picture [468 x 313] intentionally omitted <==

Abbreviations: PY, patient-year; RS, randomised set. Source: EMPA-KIDNEY CTR 2022, Figure 11.1.2.1: 1 and Table 11.1.2.1: 1 (79)

The results of sensitivity analyses were consistent with the overall results (Figure 14).

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 61 of 160

Figure 14. Forest plot of sensitivity analyses of time to adjudicated death from any cause – RS

==> picture [452 x 113] intentionally omitted <==

Abbreviations: AE, adverse event; CI, confidence interval; RS, randomised set. Events confirmed or unrefuted by adjudication are considered as an outcome event.

*There is no single definition of number of patients with an event because each imputation can produce a different number of events Source: EMPA-KIDNEY CTR 2022, Figure 15.2.2.3.2: 1 (79)

B.2.6.3 Tertiary/exploratory outcomes

B.2.6.3.1 Time to first occurrence of kidney disease progression

Empagliflozin significantly reduced the risk of kidney disease progression compared with placebo

Kidney disease progression occurred in 384 patients (11.6%) in the empagliflozin group and 504 patients (15.2%) in placebo group; the risk of kidney disease progression was significantly lower with empagliflozin treatment vs. placebo (HR, 0.71; 95% CI, 0.62 to 0.81; p<0.0001). The separation of the cumulative incidence of kidney disease progression became evident approximately 1 year after randomisation and continued over time until the number of patients at risk became too low to provide stable estimates (Figure 15) (2).

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 62 of 160

Figure 15. Time to the first event of kidney disease progression, estimated cumulative incidence function (considering non-renal death as a competing risk) – RS

==> picture [468 x 306] intentionally omitted <==

Abbreviations: CI, confidence interval; RS, randomised set. Source: EMPA-KIDNEY CTR 2022, Figure 15.2.1.1: 1 and Table 11.1.2.4: 1 (79)

B.2.6.3.2 Time to adjudicated CV death

There was a numerical reduction in time to adjudicated CV death for empagliflozin vs. placebo

Adjudicated CV death occurred in 59 patients (1.8%) in the empagliflozin group and 69 (2.1%) patients in the placebo group. There was no statistically significant treatment difference between empagliflozin and placebo (HR, 0.84; 95% CI, 0.60 to 1.19; p=0.3366). This was due to sample power: the trial was under powered to detect a difference in CV death and mortality between arms as the enrolled patients were less likely to have these events (>50% patients did not have baseline DM or severe albuminuria and >70% did not have baseline CV disease) (for details see Section B.2.4.1 Statistical methods and analysis sets ), The cumulative incidence of adjudicated CV death is displayed in Figure 16.

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 63 of 160

Figure 16. Estimated cumulative incidence function for time to adjudicated CV death (non-CV death as competing risk)-RS

==> picture [468 x 314] intentionally omitted <==

Abbreviations: CI, confidence interval; CV, cardiovascular; RS, randomised set. Note - Death from cardiovascular causes occurred in 59 patients (1.8%) in the empagliflozin group and 69 patients (2.1%) in the placebo group.

Source: EMPA-KIDNEY CTR 2022, Figure 15.2.3.2: 1 and Table 11.1.2.5: 1 (79)

B.2.6.3.3 Time to first occurrence of adjudicated CV death or ESKD

Empagliflozin significantly reduced the risk of adjudicated CV death or ESKD compared with placebo

CV death or ESKD occurred in 163 (4.9%) patients in the empagliflozin group and 217 (6.6%) patients in placebo group. The risk of CV death or ESKD was significantly reduced with empagliflozin treatment vs. placebo (HR, 0.73; 95% CI, 0.59 to 0.89; p=0.0023). The separation of the estimated cumulative incidence of CV death or ESKD between empagliflozin and placebo became evident approximately 1 year after randomisation and continued over time until the number of patients at risk became too low to provide stable estimates (Figure 17) (79).

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 64 of 160

Figure 17. Time to adjudicated CV death or ESKD, estimated cumulative incidence function (considering non-CV death as a competing risk) – RS

==> picture [468 x 298] intentionally omitted <==

Abbreviations: CI, confidence interval; CV, cardiovascular; ESKD, end-stage kidney disease; RS, randomised set. Source: EMPA-KIDNEY CTR 2022, Figure 11.1.2.6: 1 and Table 11.1.2.6: 1 (79)

B.2.6.4 Further outcomes

B.2.6.4.1 Annual rate of change in eGFR

The eGFR decline was lower for the empagliflozin group compared with placebo group

An expected acute decrease in the eGFR was seen in the empagliflozin group at the start of the trial regimen due to haemodynamic effects; however, the rate of annual decline slowed after the initial decrease. When the results are fitted on a slope analysis, on average patients in EMPA-KIDNEY progressed at a rate of 2.75 mL/min/1.73m[2] per year. Overall, the between-group difference in the eGFR slope from randomisation to the final follow-up visit was 0.75 mL/min/1.73m[2] (95% CI, 0.54 to 0.96; p<0.0001) per year, favouring empagliflozin. With respect to the decline in eGFR from 2 months to the time of the final follow-up visit, there was a between-group difference of 1.37 mL/min/1.73m[2] (95% CI, 1.16 to 1.59; p<0.0001) per year and relative difference to placebo of -50% (95% CI -0.56%, -0.44%; p<0.0001) (Figure 18) (2).

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 65 of 160

Figure 18. Change from Baseline in the eGFR

==> picture [432 x 357] intentionally omitted <==

Abbreviations: CI, confidence interval; eGFR, estimated glomerular filtration rate. The values shown as “Total slope” represent the mean (±SE) changes from randomisation to the final follow-up visit. The values shown as “chronic slope” represent the mean (±SE) changes from 2 months after the first dose of empagliflozin or placebo to the final follow-up visit.

Source: Adapted from The EMPA-KIDNEY Collaborative Group 2023 (79)

B.2.7 Subgroup analysis

The pre-specified subgroups of key interest at baseline for the efficacy outcomes were:

• DM status (present/absent),

• eGFR (<30 mL/min/1.73m[2] , ≥30 to <45 mL/min/1.73m[2] , ≥45 mL/min/1.73m[2] ), and

• uACR (<30 mg/g [3 mg/mmol], ≥30 to ≤300 mg/g [3 to 30 mg/mmol], >300 mg/g [30 mg/mmol]).

HRs and CIs were estimated with the use of Cox proportional hazards regression models, with adjustment for age, sex, history of DM, eGFR, uACR (with albumin measured in milligrams and creatinine measured in grams), and geographic region.

The pre-specified subgroup for people with DM was requested by NICE in the final scope; this was a pre-specified subgroup of key interest in the EMPA-KIDNEY trial. The final scope additionally requested subgroup analyses for people with CVD and people with other causes of CKD

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 66 of 160

(interpreted to mean people without a history of DM and CVD) if evidence allows. 'People with CVD’ was a prespecified subgroup in EMPA-KIDNEY; ‘people without DM and without CVD’ was not. Cause of CKD was a prespecified subgroup analysis, which includes patients with ‘other/unknown’ causes, however this is not mutually exclusive with CVD or T2D. The heterogenous nature of CKD should be noted; of patients with diabetes enrolled in EMPA-KIDNEY, one third of them had a primary cause of kidney disease other than diabetes (e.g., glomerular and hypertensive/renovascular).

Available results for the requested subgroups, plus the two other pre-specified subgroups of key interest, are detailed below.

The results of the subgroup analyses of the primary outcome for baseline DM status and baseline eGFR were consistent (interaction p-values >0.05), with the upper bound of the 95% CI for the HR for each subgroup <1. There was some evidence that the proportional risk reduction may have been larger among patients with higher uACR.

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 67 of 160

Figure 19 shows the HRs for the primary outcome in pre-specified subgroups of key interest, defined according to baseline characteristics, and the additional pre-specified subgroup of people with CVD and causes of CKD (2). Additional results of prespecified subgroups of key interest for primary and key secondary endpoints can be found in Appendix E.

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 68 of 160

Figure 19. Forest plot for pre-specified subgroup analyses of time to the first event of kidney disease progression or adjudicated CV death

==> picture [452 x 307] intentionally omitted <==

Abbreviations: CI, confidence interval; GFR, glomerular filtration rate. Source: Adapted from the EMPA-KIDNEY Collaborative Group 2023 (79)

B.2.8 Meta-analysis

Pairwise meta-analysis was not feasible for this submission as there are no head-to-head trials comparing empagliflozin to the comparators in the NICE scope, specifically dapagliflozin. Indirect treatment comparison (ITC) via network meta-analysis (NMA) has therefore been performed. Please see Section B.2.9 Indirect and mixed treatment comparisons and Appendix D for details.

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 69 of 160

B.2.9 Indirect and mixed treatment comparisons

• There were differences in trial design, studied population and endpoint definition that did not allow a meaningful direct comparison of empagliflozin vs. dapagliflozin based on CKD trials. Therefore, ITC (NMA) based on an expanded evidence network including patients with prevalent CKD from multiple trials was performed to examine the relative efficacy of empagliflozin to potential comparators for the treatment of patients with CKD/DKD.

• • Thirteen RCTs assessing safety and efficacy of empagliflozin, canagliflozin, dapagliflozin and finerenone in a mix of CKD or DKD populations were selected for inclusion in the NMA. These included studies where CKD/DKD diagnosis was a primary inclusion criterion, as well as studies in T2DM or HF patients with subgroup analyses of CKD/DKD patients.

• • Where more than one study was available to assess a treatment effect of a given comparator, assessments of heterogeneity were undertaken prior to conducting the NMA. The majority of the tests for heterogeneity yielded non-significant results.

• • All analyses were conducted in a Bayesian framework applying a fixed effects (FE) model in case of non-heterogeneity. The model parameters were estimated using a Markov Chain Monte Carlo (MCMC) algorithm implemented in the Just Another Gibbs Sampler (JAGS) software package.

• • The efficacy of interventions did not differ meaningfully for most outcomes. The SGLT2 inhibitors had better efficacy than finerenone for most included outcomes but the difference was non-significant.

• • Based on NMA results, the economic assessment assumes equivalence of treatment effects between empagliflozin and the comparator dapagliflozin. This is further supported by an independent source, reporting similar treatment benefits and safety across SGLT2 inhibitors with their use for modifying risk of kidney disease progression and AKI not only in patients with T2DM at high CV risk but also in patients with CKD or HF irrespective of diabetes status, primary kidney disease, or kidney function (91).

• • For the full SLR report and NMA feasibility assessment, refer to Appendix D. For the full results of the NMA, refer to Appendix N.

B.2.9.1 Objective of the indirect comparison

The primary objective of the NMA was to examine the relative efficacy of empagliflozin to comparators for the treatment of patients with CKD/DKD, with or without other comorbidities such as T2DM or HF. The secondary objective was to assess the appropriateness of the NMA to inform the economic model for the assessment of the cost-effectiveness of empagliflozin relative to comparators in CKD/DKD. Dapagliflozin is the comparator identified in the NICE Scope, however canagliflozin and finerenone were also included in the NMA and are therefore reported here.

B.2.9.2 Evidence base and comparators

A clinical SLR and TLR was conducted to identify all the relevant RCTs and observational studies, respectively, related to the treatment of CKD/DKD (Section B.2.9.5 Results and Appendix D). From the evidence base identified in the SLR, studies were selected for inclusion in a feasibility assessment for the NMA according to a narrower set of criteria that included only Phase III (or II/III) studies evaluating approved doses of included treatments, and studies of duration of >52 weeks.

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 70 of 160

Thirteen RCTs conducted in a mix of CKD or DKD populations were selected for inclusion in the NMA; these 13 trials studied the efficacy and safety of empagliflozin, canagliflozin, dapagliflozin and finerenone. Five RCTs (CREDENCE (9), DAPA-CKD (8), EMPA-KIDNEY (79), FIDELIO-DKD (92) and FIGARO-DKD (93)) included CKD or DKD patients as the target population. Five RCTs (CANVAS program [CANVAS (94), CANVAS-R (95)], DECLARE-TIMI 58 (96), Dekkers 2018 (97), EMPA-REG OUTCOME (98), MB102029 (99)) included patients with T2DM, some of whom had CKD or DKD. The final three RCTs included patients with HF, some of whom had CKD or DKD (DAPA-HF (100), EMPEROR-Reduced (101), EMPEROR-Preserved (102)). Please see Table 20 for further details on the characteristics of included trials.

B.2.9.3 NMA framework

An anchored Bayesian NMA was chosen for this analysis, rather than a matching adjusted indirect comparison (MAIC), due to inherent differences in the trials of EMPA-KIDNEY and DAPA-CKD. Key differences between the two trials are summarised in Table 19 below.

Table 19 Summary of key differences in baseline characteristics between EMPA-KIDNEY and DAPA-CKD trials

Abbreviations: CV, cardiovascular disease; eGFR, estimated glomerular filtration rate; IQR, interquartile range; SD, standard deviation; SGLT2, sodium-glucose cotransporter 2; T2DM, type-2 diabetes mellitus; uACR, urine albumin-to-creatinine ratio. Sources: The EMPA-KIDNEY Collaborative Group 2023 (79), Heerspink 2020 (8)

A MAIC weights the patients in the index trial so that the means of the baseline characteristics in the index trial match the means of the same characteristics in the comparator trial; the population of interest for this ITC is that of the EMPA-KIDNEY population, which is broader than several comparator trials. The subset of EMPA-KIDNEY patients who met DAPA-CKD renal inclusion criteria differs largely from the DAPA-CKD ITT population, thus absolute risk of outcomes in these two trials is not purely determined by CKD status. Other confounders must be present, but these are largely unknown and thus cannot be adjusted for in the analysis.

Patient identifiable data (PID) for comparator trials is unavailable to match to the DAPA-CKD trial population to that of the EMPA-KIDNEY trial; it is only possible to match EMPA-KIDNEY data to the population of comparator trials. Furthermore, a MAIC can only match on observed and reported characteristics, so cannot account for all possible sources of heterogeneity.

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 71 of 160

Bayesian NMA preserves randomisation and therefore controls for reported and unreported sources of heterogeneity; thus, effect modifiers are the only concern, not prognostic factors. A MAIC is more sensitive to differences in prognostic factors, and there are differences in baseline eGFR between EMPA-KIDNEY and DAPA-CKD. An anchored NMA was selected as there is a connected network of interventions with a common comparator, and the use of NMA preserves randomisation by using relative versus absolute effects.

Moreover, MAIC ignores correlations between covariates, which may affect the performance of the method if correlations differ between studies. In this case, the correlation of covariates within DAPA-CKD were not known, so it was not possible to assess how they may differ from that of EMPA-KIDNEY.

All analyses were therefore conducted in a Bayesian framework, applying both a FE and random effects (RE) model, with the former being preferred in the absence of heterogeneity. For binomial outcomes, the NMA was performed on the proportion of patients experiencing each outcome of interest. A regression model with binomial likelihood and the logit link function was used. For rate outcomes, the NMA was performed using count data and the number of person-years at risk. A Poisson likelihood and log link was used. The model parameters were estimated using a MCMC algorithm implemented in the JAGS software package. All analyses were performed using R version 4.0.5 and JAGS version 4.3.0.

B.2.9.4 Assessment of heterogeneity

The evidence base included both studies with CKD/DKD as a primary inclusion criterion as well as broader studies in T2DM or HF patients with a reported subgroup of CKD/DKD patients. Across the 13 trials, various definitions for the target population were used: KDIGO risk score of High or Very High, CKD Stage 3 or higher (according to the National Kidney Foundation [NKF] definition), or eGFR <60 mL/min/1.73m[2] were considered eligible groups of patients for this analysis. It should be noted that the KDIGO definition classifies patients according to both eGFR and albuminuria, while the NKF definition uses only eGFR levels; use of NKF CKD Stage 3+ excludes patients with normal to mildly decreased eGFR despite severely increased albuminuria.

In addition to differing definitions of CKD, the included studies showed differences with respect to a number of other factors other than CKD. CREDENCE, FIDELIO-DKD and FIGARO-DKD enrolled only patients with CKD and T2DM; EMPA-KIDNEY and DAPA-CKD enrolled CKD patents with and without T2DM. CANVAS, CANVAS-R, DECLARE-TIMI 58, Dekkers 2018 analysed published studies of T2DM patients; DAPA-HF, EMPEROR-Preserved, EMPEROR-Reduced enrolled patients with HF. Two studies, FIDELIO-DKD and EMPA-KIDNEY were explicitly designed to include CKD patients under-represented in prior SGLT2 inhibitors trials. In particular, EMPA-KIDNEY recruited a high percentage (54%) of patients without T2DM, with eGFR

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 72 of 160

<45 mL/min/1.73m[2] (78%), and with uACR <300 mg/g [30 mg/mmol] (48%); FIDELIO-DKD enrolled patients at lower eGFR levels (maximum of 60 mL/min/1.73m[2] for uACR 30 to 300 mg/g [3 to 30 mg/mmol] and maximum 75 mL/min/1.73m[2] for patients with uACR 300-5000 mg/g [30 to 565 mg/mmol]).

Patient populations in the included studies were broadly similar in terms of distribution of age, sex, and BMI. The proportion of Asian patients varied widely among studies; EMPA-KIDNEY included the largest percentage of Asian patients (36.2%). Distributions of eGFR and uACR at baseline varied widely, however, driven by differing inclusion criteria and different CKD/DKD subgroup definitions with respect to these measures; these measures are known prognostic factors so differences in these values are likely to affect renal outcomes. Studies also differed notably in the proportion of patients with history of CVD; in particular, EMPA-KIDNEY and the CANVAS program included lower proportions of patients with CVD (22.4% and 26.8% respectively), compared to CREDENCE, FIDELIO-DKD, and EMPA-REG OUTCOME (50.4%, 54.1%, and 100% respectively).

Assessments of heterogeneity were undertaken prior to conducting the NMA. All but one test for heterogeneity yielded non-significant results; however, estimates of relative treatment effects may still be affected by known differences between trials. As with any NMA, the validity of the estimates of relative efficacy depends on the comparability of the trials included in the analysis. As no closed loops exist in the network of evidence, it was not possible to evaluate inconsistency.

B.2.9.5 Results

A summary of characteristics of included trials is presented in Table 20 Table 20 . The following efficacy outcomes were assessed in the NMA: composite renal outcomes, progression to ESKD/ESRD, HHF, CV death, a composite of HHF or CV death, 3P-MACE+, all-cause mortality, and ACH. The composite renal outcomes were defined as follows: 1.) eGFR decline, ESKD, or renal death or 2.) eGFR decline, ESKD, or CV or renal death; for both composite outcomes eGFR decline thresholds of 40%, 50%, and 57% were considered. Further details on outcome definitions can be found in Appendix D.

Significant heterogeneity was only observed for the comparison between dapagliflozin and placebo for the outcome “3P-MACE+ and 3P-MACE” (see Appendix N for details). All other comparisons showed non-significant results for heterogeneity. Generally, the efficacy of the interventions did not differ meaningfully for most outcomes. Empagliflozin was associated with a lower rate of ACH admissions than finerenone (OR 0.92 [0.85-1.00]) and dapagliflozin was associated with a lower rate of HHF than finerenone (OR 0.64 [0.41-0.98]). No other statistical differences were found between interventions. However, the SGLT2 inhibitors showed numerically better efficacy than

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 73 of 160

finerenone for most included outcomes, with generally similar SGLT2 inhibitor treatment effects. Detailed results for each outcome are presented in Appendix N.

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 74 of 160

Table 20. Summary of the trials used to conduct the indirect treatment comparison

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 75 of 160

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 76 of 160

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 77 of 160

Abbreviations: CKD, chronic kidney disease; CVD, cardiovascular disease; DKD, diabetic kidney disease; DM, diabetes mellitus; eGFR, estimated glomerular filtration rate; g, gram; HbA1c, glycated haemoglobin; HDL, high-density lipoprotein; HF, heart failure; KDIGO, Kidney Disease: Improving Global Outcomes; LVEF, left ventricular ejection fraction; m, metre; mg, milligram; min, minute; mL, millilitre; mmol, millimole; NT-proBNP, N-terminal pro b-type natriuretic peptide; SGLT2-i, sodium-glucose cotransporter two inhibitor; T2DM, type 2 diabetes mellitus; uACR, urine albumin-to-creatinine ratio.

aThe broadest subgroup(s) meeting inclusion criteria are listed here. Trials may report data for additional subgroups (e.g., eGFR <45 mL/min/1.73m2 or eGFR <60 mL/min/1.73m2 and uACR >300 mg/g [30 mg/mmol])

*Creatinine clearance <60mL/min based on Cockroft-Gault equation listed as exclusion criteria.

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 78 of 160

B.2.9.6 Uncertainties in the indirect and mixed treatment comparisons

There are several limitations to this evidence base in terms of suitability for an NMA. Firstly, the definition of CKD varied across included studies, both in terms of study inclusion criteria and reported subgroups of T2DM or HF trials. In terms of composite renal event, differences in use of eGFR reduction thresholds between studies prevented indirect comparisons of all four interventions in a single network for this outcome. Additionally, estimation of relative treatment effects for ACH was limited by a lack of reported data for canagliflozin and dapagliflozin. Finally, the follow-up time of reported outcomes differed across studies; while the NMA assumes that event rates of each outcome are constant over time.

B.2.9.7 Conclusion

Despite the uncertainties, the strengths of this analysis include the derivation of an evidence base from an SLR. Furthermore, the evidence base available for the NMA consisted of a connected network of placebo controlled RCTs, allowing for an anchored indirect comparison of the interventions of interest.

In summary, the efficacy of the interventions included in the NMA network did not differ meaningfully for most outcomes. Compared to finerenone, empagliflozin was associated with a significantly lower rate of ACH admissions and dapagliflozin was associated with a significantly lower rate of HHF. No other statistical differences were found between interventions. Results suggested the SGLT2 inhibitors had better efficacy than finerenone for most included outcomes, but the difference was non-significant. Moreover, clinical expert opinion supports the conclusion that there is no difference in treatment effect between empagliflozin and dapagliflozin in similar eligible populations (see Appendix O).

Since there are no meaningful differences observed for most of the outcomes, the economic assessment versus dapagliflozin assumes equivalence of treatment effects between empagliflozin and SGLT2 comparators, thus justifying the decision to perform a cost-comparison analysis for empagliflozin versus dapagliflozin in this appraisal. This is further supported by the entirety of the evidence which has been generated over the years for SGLT2 inhibitors that supports a consistent kidney protective effect across several compounds and in various disease populations and clinical CKD phenotypes. A recent meta-analysis systematically investigated outcomes from 13 trials with SGLT2 inhibitors, which included patients with DM (n = 74,804) and without DM (n = 15,605); triallevel mean baseline eGFR ranged from 37 mL/min/1.73m[2 ] to 85 mL/min/1.73m[2 ] (91). Overall, SGLT2 inhibitors reduced the risk of kidney disease progression by 37% (RR 0.63, 95% CI 0.58, 0.69), with similar effects in patients with DM (RR 0.62, 95% CI 0.56,0.68) and without DM (RR 0.69, 95% CI 0.57, 0.82), (heterogeneity by DM status p = 0.31) and consistency across baseline eGFR levels (Figure 20). Likewise, consistent treatment effects on kidney disease

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 79 of 160

progression were observed in both DM and non-DM patients across a broad range of baseline uACR values (Figure 21).

Figure 20. Effect of SGLT2 inhibitors on kidney disease progression by DM status and eGFR

==> picture [452 x 459] intentionally omitted <==

Abbreviations: CI, confidence interval; DM, diabetes mellitus; eGFR, estimated glomerular filtration rate; SGLT2, Sodium-glucose transporter 2; RR, relative ratio

Kidney disease progression was defined as a sustained decrease in eGFR (≥50%) from randomisation, a sustained low eGFR, end-stage kidney disease, or death from kidney failure in all presented trials.

*One participant without diabetes in DELIVER was missing a baseline creatinine measurement and was excluded. Source: Herrington et al. (91)

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 80 of 160

Figure 21. Effect of SGLT2 inhibitors on kidney disease progression by DM status and uACR

==> picture [478 x 476] intentionally omitted <==

Abbreviations: CI, confidence interval; eGFR, estimated glomerular filtration rate; SGLT2, Sodium-glucose transporter 2; RR, relative ratio; uACR, urine albumin to creatinine ratio.

Kidney disease progression was defined as a sustained decrease in eGFR (≥50%) from randomisation, a sustained low eGFR, end-stage kidney disease, or death from kidney failure in all presented trials. Source: Herrington et al. (91)

B.2.10 Adverse reactions

For the full adverse reactions results of the trial, refer to Appendix F. Median exposure to study medication was approximately 22 months in both treatment groups, with 91% of patients treated for at least 1 year. Safety was assessed descriptively based on AE, adverse events of special interest (AESI), and specific AEs (2).

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 81 of 160

Empagliflozin and placebo groups had similar frequencies of patients with reported SAEs and pre-specified non-serious AEs (Table 21). The frequency of patients reported with AEs leading to discontinuation of study medication was also similar between treatment groups. The frequency of patients with investigator-defined drug-related AEs was low. The frequency of patients with SAEs overall was comparable between both groups. The frequency of patients with fatal AEs was similar in both groups (79).

Table 21. Overall summary of AE - TS

Abbreviations: AE, adverse event; SAE, serious adverse event; TS, treated set. Note: Percentages calculated using total number of patients per treatment as the denominator. A patient may be counted in more than one seriousness criterion. aOther medically important serious events were important medical events in the opinion of a responsible local investigator (i.e., not life threatening or resulting in hospitalisation, but could jeopardise the participant or require intervention to prevent one or other of the outcomes listed above). Source: EMPA-KIDNEY CTR 2022, Table 15.3.1.2.1: 1 (79).

The frequencies of SAEs in each system organ class were similar in the empagliflozin and placebo groups. The most frequently reported SAEs and pre-specified non-serious AEs were in the system organ class metabolism and nutrition disorders, followed by infections and infestations, investigations, and renal and urinary disorders. On the PT level, the most frequently reported SAEs and pre-specified non-serious AEs were gout, AKI, and coronavirus infection (Table 22). All other SAE were reported in less than 3.0% of patients per treatment group (79).

Table 22. Serious and pre-specified non-serious AE with frequency ˃2% - TS

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 82 of 160

Abbreviations: AE, adverse event; MedDRA, Medical dictionary for regulatory activities; MedDRA PT, Medical dictionary for regulatory activities preferred term; SAE, serious adverse event; TS, treated set. Source: EMPA-kidney CTR 2022, Table 12.1.2.1: 1 and Table 15.3.1.2.1: 4 (79).

AESIs were pre-specified in the protocol as liver injury, ketoacidosis, and AE leading to lower limb amputation. Specific AE were defined as severe hypoglycaemia, urinary tract infection, genital infection, bone fracture, urinary tract malignancy, volume depletion, AKI, gout, hyperkalaemia, and COVID-19 events (Table 23). The overall frequencies for liver injury, serious urinary tract infection, serious genital infection, severe hypoglycaemia, and urinary tract malignancy were comparable in the empagliflozin and placebo groups. Ketoacidosis and lower limb amputations occurred in higher number of patients in the empagliflozin group than in the placebo group. Otherwise within the individual categories of AESIs and specific AEs, generally similar proportions of patients in both treatment groups had serious AEs. Few AEs in any category of AESIs or specific AEs led to treatment discontinuation. Safety results in the subgroup of patients with a baseline eGFR <20 mL/min/1.73m[2] were consistent with the overall AE profile in the trial (79).

Table 23. Summary of AESI and specific AE – TS

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 83 of 160

Abbreviations: AE, adverse event; AESI, adverse event of special interest; BIcMQ, Boehringer Ingelheim customised MedDRA query; TS, treated set; SMQ, standardised MedDRA query; BIcMQ, Boehringer Ingelheim customised MedDRA query adjudication of events stopped at final follow-up period; any residual effect period afterwards was not considered for these events. Source: EMPA-KIDNEY CTR 2022, Synopsis Table 4; Table 12.1.3.2.7: 1 and Table 12.1.3.2.10: 1 (79).

The adverse event profile for empagliflozin is further similar to that other SGLT2 inhibitors, which has been shown by the above-mentioned meta-analysis published by Herrington et al. (2022) (109, (91).

B.2.11 Ongoing studies

There are no ongoing studies of empagliflozin relevant for this appraisal.

B.2.12 Interpretation of clinical effectiveness and safety evidence

In the EMPA-KIDNEY trial, treatment with empagliflozin 10 mg OD as an add-on to SoC in patients with CKD demonstrated superiority compared to placebo for the primary outcome, time to the first occurrence of kidney disease progression or CV death. The trial also demonstrated superiority of empagliflozin over placebo for the key secondary outcome of time to ACH. The reduction in risk of all-cause death and HHF or CV death was not statistically significant with empagliflozin treatment as compared with placebo; however, it should be noted these key secondary outcomes were based on a relatively small number of patients with events. The results of all sensitivity analyses were consistent with the results of the primary analysis (i.e., the HR was numerically similar).

For the primary outcome, treatment with empagliflozin lead to a clinically and statistically significant reduction in risk of kidney disease progression or CV death by 28% compared with placebo added to SoC. The treatment effect of empagliflozin became apparent approximately 1 year after randomisation and was maintained over time. The results of the primary outcome were consistent across the pre-specified key subgroups of DM status and baseline eGFR categories as well as other subgroups such as patients with and without CVD (the upper bound of the 95% CI for the HR for each subgroup <1).

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 84 of 160

Treatment with empagliflozin significantly reduced the risk of ACH by 14% compared with placebo. The treatment effect of empagliflozin was observed shortly after randomisation and maintained throughout the trial. Risk reduction for the time to first occurrence of kidney disease progression and the time to first occurrence of ESKD or CV death were of similar magnitude to that of the primary outcome. Few adjudicated CV deaths occurred in both treatment groups; the treatment difference was not statistically significant because of the low number of events.

Overall, empagliflozin was well tolerated in CKD patients and had similar frequencies of patients with reported SAE and pre-specified non-serious AE as placebo. The overall frequencies of AESI and specific AE such as liver injury, serious urinary tract infection, serious genital infection, severe hypoglycaemia, and urinary tract malignancy were also comparable in the empagliflozin and placebo groups. The frequency of patients reported with AE leading to discontinuation of study medication was also similar between treatment groups.

In addition to direct evidence, the relative efficacy of empagliflozin versus competing interventions for the treatment of patients with CKD was assessed in an ITC (NMA). The NMA results showed that the SGLT2 inhibitors canagliflozin, dapagliflozin and empagliflozin show consistent benefit for the treatment of CKD/DKD and may offer benefit over finerenone, though study heterogeneity from differing inclusion criteria prevents statistical differentiation between these interventions. Our findings are further supported by the meta-analysis published by Herrington et al. (2022) (109, (91), which showed consistent benefits and safety of SGLT2 inhibitors in modifying risk of kidney disease progression and acute kidney injury, not only in patients with type 2 diabetes at high cardiovascular risk, but also in patients with chronic kidney disease or heart failure irrespective of diabetes status, primary kidney disease, or kidney function. Together, our findings combined with those of independent sources support the assumption of equivalence of treatment effects between empagliflozin and the comparator dapagliflozin in the economic assessment (see section Error! Reference source not found. ).

In conclusion, data presented in this section demonstrate that compared to placebo, empagliflozin 10mg led to a significant reduction in risk of progression of kidney disease or death from CV causes as well as a significant reduction in ACH amongst a broad range of patients with CKD who were at risk for disease progression. The data therefore supports addition of empagliflozin to the guideline directed medical therapy for patients with CKD.

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 85 of 160

B.3 Cost-effectiveness

• A de novo Markov state patient-level microsimulation model was developed in Microsoft Excel[®] to estimate the cost-effectiveness of oral empagliflozin 10 mg OD on top of individually optimised SoC versus placebo on top of SoC for the treatment of adult patients with CKD.

• The model is structured around a set of 18 mutually exclusive and collectively exhaustive KDIGO health states defined by eGFR and uACR chosen to represent the natural history of CKD and economically important events in the disease progression. Death was an absorbing state in the model.

• • Bootstrapping/sampling was used to randomly select individual patients entering the model based on a combination of eGFR and uACR states, baseline demographic characteristics and comorbidities representative of the ITT population of the EMPA-KIDNEY trial.

• • Patients’ disease progression through KDIGO health states was modelled through annual transition probabilities derived from eGFR slopes and uACR changes over time. Treatment specific transition probabilities were applied while patients were alive and remained on treatment (i.e., up until treatment discontinuation).

• • No treatment effect was assumed after treatment discontinuation, following which disease progression (through KDIGO health states) is modelled using observational data reported in the Chronic Kidney Disease Prognosis Consortium (CKD-PC) or Chronic Renal Insufficiency Cohort Study (CRIC) registries.

• • Patients entering the model were at risk of common CKD complications including CVD, hypertension, infections, BMD, anaemia, DM, and AKI among others. Individual risk of experiencing each complication was determined by risk equations that incorporated predictor variables including eGFR and uACR, or probabilities sourced from published literature.

• • The model incorporates costs involved in the management of CKD and associated complications, over time. Costs were obtained primarily through a structured literature search (Appendix I) and are assigned for the specific health state/ complication/ event in each cycle of the model engine.

• • Health state utility values and clinical event disutility values were obtained through a structured literature review and prior NICE technology appraisals (TAs), and relevant health care state utilities from EMPA-KIDNEY trial were utilised in the scenario analysis.

• • The cost-effectiveness analysis was consistent with the NICE reference case and performed from an NHS and PSS perspective. Costs and benefits were discounted at a rate of 3.5%.

• • A time horizon of 50 years was adopted to reflect a lifetime analysis in older and younger individuals (mean age of the full ITT cohort was 63.8 years).

• • In the deterministic base-case economic analysis, treatment with empagliflozin, compared with placebo, as an add-on therapy to SoC was dominant, with an incremental cost-effectiveness ratio (ICER) of -£6,431.37/quality-adjusted life year (QALY) gained and a net health benefit (NHB) of 1.12 at the £20,000 willingness to pay (WTP) threshold. Results of the cost comparison versus dapagliflozin resulted in no difference in costs associated with SGLT2 treatment acquisition and its management.

• The probabilistic ICER was also dominant (-£5,998.34/QALY gained) and highly comparable with the deterministic ICER, with 99% probability of cost-effectiveness at a WTP threshold of £4,000.

• The key drivers of the deterministic sensitivity analysis were the age limit for patients being eligible for RRT (80 years limit in the base-case), and health state utilities for patients in G+15_A-300.

• The scenario analyses also demonstrated that ICER remained dominant in all cases.

• In summary, the cost-utility analysis demonstrated oral empagliflozin 10 mg OD is a highly costeffective treatment option for CKD.

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 86 of 160

B.3.1 Published cost-effectiveness studies

An SLR of published UK full economic evaluations assessing the cost-effectiveness of CKD treatments was conducted in October 2022. Full details of the SLR search strategy, study selection process, results and a Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram detailing studies that were included and excluded at each stage are presented in Appendix G. MEDLINE[®] (Ovid SP[®] ), EMBASE[®] (Ovid SP[®] ) and EconLit[®] (Ovid SP[®] ) were searched in addition to hand searching of conference proceedings and health technology assessment (HTA) websites. Eligible records for inclusion were those reporting novel economic evaluations of the cost-effectiveness of CKD treatments.

In total, 2,462 unique records were identified in the SLR database searches. 2,315 records were excluded following title and abstract review. 125 records were excluded following full text review, and 22 remaining records were found to meet the eligibility criteria. Hand searching found an additional 4,583 records, of which 4,570 records were excluded following review, meaning 13 records were ultimately included from hand-search results. A total of 35 publications reporting on 33 unique economic evaluations were ultimately included in the SLR as relevant to UK clinical practice.

B.3.1.1 Summary of published cost-effectiveness studies

The summary of the cost-effectiveness studies meeting inclusion and exclusion criteria for the SLR is presented in Table 7 and 8 of Appendix G. Of the 33 unique UK economic evaluations in CKD identified through the searches, three were patient-level simulation models (103-105), twenty-three were Markov models (106-127), one was a pathway model (128), one was a proportional hazards risk prediction model (129), and five did not report the model type (130-133). Three of these evaluations assessed the SGLT2 inhibitors dapagliflozin (DAPA) (6, 122) and canagliflozin (131). The strengths and limitations of the published CKD models including the NICE committee critiques of those submitted for Single Technology Appraisal (DAPA-CKD model) were reviewed to inform the eventual structure of the model for empagliflozin in CKD (130-133).

In the dapagliflozin CKD model (TA775) (6), a Markov model was chosen to synthesise the economic evidence submitted for the economic assessment. The following critiques were received from the NICE external assessment group (EAG) in relation to this model choice:

1. Some of the estimated transition probabilities applied in the DAPA-CKD model did not appear to be clinically plausible. For example, the same transition probabilities were used for subgroups of patients without albuminuria, patients with DM and patients without DM. It is unlikely to be appropriate since CKD progression would be different for all these subgroups

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 87 of 160

2. The DAPA-CKD model estimated state specific mortality risks using a “mean of covariates” approach. The EAG considered that this reflected a misinterpretation of the outputs of the multivariable survival model, which had been shown to lead to bias when estimating survival distributions

3. The EAG believed that resolving the poor model fit may require a different modelling approach (e.g., a time-homogeneous multi-state model which jointly estimates all transition probabilities between model states using a single dataset) and that it might be possible to achieve a better model fit to OS using an alternative modelling approach

B.3.2 Economic analysis

To address the limitations of published models and those critiqued by previous NICE committees appraising dapagliflozin (6, 122), a de novo Markov state patient-level microsimulation model was developed in Microsoft Excel[®] to assess the cost-effectiveness of empagliflozin in patients with CKD. A microsimulation approach was selected over a cohort-level method for the following reasons:

• Microsimulation methodology offers advantages over a cohort-level model, including the flexibility to randomly allocate baseline characteristics and risk factors, and to track individual disease histories over time. This approach was deemed necessary to reliably model a broad range of progression paths across CKD patients with different eGFR and uACR levels (103105)

• A microsimulation approach offers an advantage in cases where state transition probabilities depend on baseline characteristics and past medical history (e.g., time since disease onset, the occurrence of previous events, or time-varying response to treatment), in addition to timedependent risk factors such as eGFR and uACR. This approach also enables modelling of the occurrence of complications based on established risk equations or probabilities linked to eGFR and uACR, which are particularly relevant in CKD (134)

• A microsimulation model also helps in modelling of multiple complications and comorbidities independently which allows to capture the burden of CKD patients in a more aligned way to the routine clinical care practice

• Microsimulation approaches also facilitate transition probabilities that are a function of any number of attributes and offer flexibility to capture a greater scope of outputs since the model can return estimates of the entire distribution of events, rather than just expected values (135)

• Microsimulation approaches further facilitate flexibility in allowing random allocation of baseline characteristics to important sources of heterogeneity, allowing for continuous, dynamic risk factors that can be modified over time and their ability to track individual disease histories over time (46)

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 88 of 160

• In the empagliflozin microsimulation model, mortality risk was predicted based on UK National life tables and evidence identified through literature searches (specifically, CVD mortality risk equations [e.g., CKD-PC registries]) and all-cause mortality reported on patients receiving RRT (UKRR registries). Also, a patient-level microsimulation model utilises individual patient characteristics (including age, eGFR and uACR) to determine mortality risk. This patient-level approach negates the need for a “mean of covariates” approach and hence the associated bias

• In the empagliflozin microsimulation model, treatment effects are applied directly on the progression of eGFR and uACR (to capture benefit derived from slowing disease progression). Additionally, treatment effect was also applied on HHF and AKI to allow capturing of treatment benefit on these outcomes beyond the benefit derived from slowing disease progression. No treatment effect on mortality is applied, eliminating risk of double counting

• The appropriateness of patient-level microsimulation methodology for modelling CKD is supported by published literature identified in the cost-effectiveness SLR (103-105), as well as further literature published since SLR completion (46, 135).

• NICE Scientific Advice was also obtained on the appropriateness of the microsimulation model presented in this submission. This concluded that the structure of the model and approach to modelling CKD disease progression and management were broadly appropriate, and that the model’s internal validity was high. Recommendations resulting from NICE Scientific Advice were applied to the final empagliflozin microsimulation model (136).

These factors collectively contribute to why a microsimulation approach was deemed necessary and appropriate to reliably model the progression of complex and multi-morbid CKD patients as seen in the broad, heterogenous CKD population included in the pivotal EMPA-KIDNEY trial. An extensive internal and external validation of the model was carried out prior to conducting the costutility analysis, with internal model validation performed by comparing outcomes observed in source studies used in the derivation of risk equations and studies and cohorts not directly used in the development of risk equations being utilised for external validation (section B.3.14).

B.3.2.1 Patient population

The base-case analysis evaluated the cost-effectiveness of empagliflozin 10mg oral OD on top of individually optimised SoC in adult patients with CKD, in line with the final NICE scope of this appraisal. The baseline characteristics of the patients were derived from the ITT population of the EMPA-KIDNEY trial (see section Error! Reference source not found. ). Truncation of minimum and maximum baseline values was applied in cases of extreme values, as per Appendix P. In line with the EMPA-KIDNEY trial, three patient populations were considered: all CKD patients, CKD patients with DM and CKD patients without DM. The patient population derived from the trial was

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 89 of 160

broadly representative of the CKD population in UK clinical practice (see section B.2.5.1 Applicability to clinical practice).

B.3.2.2 Model structure

The model is structured around a set of 18 mutually exclusive and collectively exhaustive KDIGO health states chosen to represent the natural history of CKD and economically important events in the disease progression. KDIGO stages are also in line with disease severity classification in NG203 and KDIGO guidelines (7, 10). Patients may progress through the 18 health states, which are based on a combination of eGFR and uACR values (Table 24).

Table 24. KDIGO classification health states incorporated in the model

Abbreviations: CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; KDIGO, Kidney Disease Improving Global Outcomes; mg/mmol, milligrams per millimole; mL/min, millilitre per minute; uACR, urine albumincreatinine ratio.

Patients move between health states within discrete annual cycles. Bootstrapping/sampling method is used to randomly distribute patients based on an initial set of characteristics including demography, risk factors, baseline comorbidities and background medications. The eGFR and uACR of each patient are then individually tracked in the model (along with further risk factors discussed in section B.3.3.2.3 Progression of other risk factors, defining which CKD KDIGO health state patients occupy at any given point in time. While on treatment, progression of eGFR and uACR is informed by observations in the EMPA-KIDNEY trial. Following the treatment discontinuation clinical data from large observational studies/patient registries published in peerreviewed journals replicated using bootstrapping inform the progression of eGFR and uACR. Annual cycles are used with a maximum time horizon of 50 years to capture lifetime outcomes. All

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 90 of 160

mortality events are captured in the death health state, defined as CVD death plus renal death plus non-specific death.

Patients are at risk of experiencing most adverse events and CKD complications at all times (i.e., CVD, BMD, anaemia, T2DM, hypertension, AKI, infections, incident cancer, metabolic complications, hospitalisations, and ESKD), however, only patients reaching ESKD are at risk of experiencing peritonitis, AV access thrombosis, and BSI. The risk of experiencing these at any point in time varies according to past medical history, eGFR, uACR and further risk factors as discussed in section B.3.3.2. The choice of the events and complications included in the model was based on published literature, or by their presence in the previously published commonly recognised predictive risk equations (see Appendix P) and verified by clinical expert opinion (Appendix O). The acute events and long-term complications (hereafter referred to as complications) incorporated in the model are described in Table 25.

Table 25. Complications incorporated in the model

Abbreviations: ACH, all-cause hospitalisations; AKI, acute kidney injury; BMD, bone and mineral disorder; BSI, bloodstream infections; AV, Arteriovenous; CV, cardiovascular; CVD, cardiovascular disease; DM, diabetes mellitus; ESKD, end-stage kidney disease; HD, haemodialysis; HF, heart failure; KT, kidney transplant; MI, myocardial infarction; PAD, peripheral arterial disease; PD, peritoneal dialysis; TIA, transient ischaemic attack

Further details on the complications sub-modules incorporated in the model are provided in Appendix P. The interaction between baseline characteristics, eGFR, uACR, health states and associated complications are demonstrated in Figure 22.

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 91 of 160

Figure 22. CKD Markov microsimulation model schematic

==> picture [452 x 216] intentionally omitted <==

Abbreviations: AKI, acute kidney injury; AV, arteriovenous; BMD, bone and mineral disorder; BMI, body mass index; BSI, bloodstream infections; CKD, chronic kidney disease; CVD, cardiovascular; eGFR, estimated glomerular filtration rate; ESKD, end-stage kidney disease; HbA1c, glycated haemoglobin; HD, haemodialysis; HF, heart failure; HS, health state; KDIGO, Kidney Disease Improving Global Outcomes; MI, myocardial infarction; PAD, peripheral arterial disease; PD, peritoneal dialysis; RT, renal transplant; TIA, transient ischaemic attack; uACR, urine albumin-to-creatinine ratio.

The key features of the cost-effectiveness analysis used in the model are summarised in Table 26.

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 92 of 160

Table 26. Features of the economic analysis

Abbreviations: CKD, chronic kidney disease; EMPA, empagliflozin; KDIGO, Kidney Disease Improving Global Outcomes; NHS, National Health Service; NICE, National Institute for Health and Care Excellence; PSS, Personal Social Services; QALY, quality-adjusted life years; TLR, targeted literature review

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 93 of 160

B.3.2.3 Intervention technology and comparators

The intervention in the cost-utility analysis is oral empagliflozin 10mg OD in addition to individually optimised SoC in adult patients with CKD, and the comparator is matching placebo with individually optimised SoC, in line with the NICE final scope and the EMPA-KIDNEY trial design.

A full incremental cost-effectiveness analysis versus dapagliflozin as a comparator has not been performed due to inherent differences in the EMPA-KIDNEY and DAPA-CKD populations, outcome definitions, and trial duration that could not be fully adjusted for as per section B.2.9 Indirect and mixed treatment comparisons. Expanded NMA and ITC results did not show any clinically meaningful differences, supporting a conclusion of similar health benefits (and therefore costs). Clinical expert opinion supports this conclusion (see Appendix O). Therefore, a cost-comparison has been presented for empagliflozin and dapagliflozin.

B.3.3 Clinical parameters and variables

B.3.3.1 Baseline characteristics

This section describes the baseline characteristics of the three different populations for which costutility analysis is performed i.e., the full ITT population (hereafter referred to as full cohort), patients with DM, and patients without DM. Baseline characteristics for all populations were obtained from the EMPA-KIDNEY trial.

Table 27 outlines the baseline characteristics of the studied populations. Baseline characteristics determine the KDIGO health state at which the patient enters the model. Disease progression (based on eGFR and uACR) and the risk of occurrence of complications is also dependent on baseline characteristics as well as the evolution of further risk factors over time (see section B.3.3.2).

The EMPA-KIDNEY trial did not report data on some parameters that are required for the model, thus alternative sources were identified. The proportion of patients on anti-hypertension therapy was assumed to be the same as the percentage of patients with hypertension. Baseline values for HDL, TC, and proportions of patients with hypertension, metabolic acidosis, family history of DM, and receiving anti-hypertensive, atypical antipsychotic and corticoid steroid medication were obtained from published articles based on CRIC Study data (137, 138), and data used in development and validation of QDiabetes-2018 risk prediction algorithm (139).

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 94 of 160

Table 27. Baseline characteristics for full cohort, diabetics, and non-diabetics

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 95 of 160

Abbreviations: AKI, acute kidney injury; BAD, bipolar affective disorder; BMI, Body mass index; cm, centimetres; CHF, congestive heart failure; CVD, cardiovascular disease; DM, diabetes mellitus; eGFR, estimated glomerular filtration rate; HbA1c, glycated haemoglobin; HDL, high-density lipoprotein; PCOS, polycystic ovary syndrome; TC, Total Cholesterol; LDL, Low-density lipoprotein; NGT, normal glucose tolerance; SBP, Systolic blood pressure; uACR, urine albumin-to-creatinine ratio.

Sources: 1 EMPA-KIDNEY Trial data on file - Clinical trial documentation; 2 Lash et. al 2009 (137); 3 Grams et. al 2020 (138); 4 Hippisley-Cox 2018 (139)

B.3.3.2 Risk of disease progression

As discussed in section B.3.2.2 Model structure, disease progression for individual patients through KDIGO health states is determined by progression of eGFR and uACR. Data informing the rate of disease progression in treatment and comparator arms was taken from two sources:

1. EMPA-KIDNEY trial: Patients’ disease progression through KDIGO health states in the treatment and comparator arms was modelled through annual treatment specific transition probabilities derived from observed eGFR slopes (Table 29) and uACR observed changes over time in EMPA-KIDNEY while patients remain alive and on treatment as per Table 31.

2. Literature: A TLR of observational studies was performed for evidence on eGFR and uACR progression in CKD to inform model parameterisation for risk of disease progression following treatment discontinuation. The TLR retrieved 21 relevant studies – 15 studies for eGFR progression and six studies for uACR progression (

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 96 of 160

3. Table 28). Data from Naimark et al. 2016 (140) (from the CKD-PC registry) was ultimately used in the base-case of the model for eGFR progression, as it reported the absolute changes in eGFR required for modelling inputs. For uACR progression, data from Coresh et al. 2019 (141) was used in the base-case of the model, as it provided three-year uACR fold change which fitted to a lognormal distribution.

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 97 of 160

Table 28. Key articles identified in targeted literature search for eGFR and uACR

Abbreviations: CKD, chronic kidney disease; CKD-PC, Chronic Kidney Disease Prognosis Consortium; CRIC, Chronic Renal Insufficiency Cohort Study; DM, diabetes mellitus; eGFR, estimated glomerular filtration rate; EMR, electronic medical record; ESKD, end-stage kidney disease; HK, Hong Kong; JDCS, Japan diabetes complication study; KDIGO, Kidney Disease Improving Global Outcomes; KPNC, Kaiser Permanente Northern California; T2D, type 2 diabetes mellitus; US, United States

B.3.3.2.1 eGFR progression

In the EMPA-KIDNEY trial, the annual progression of eGFR in patients receiving empagliflozin (on top of SoC) were reported per KDIGO categories and overall (across all health states, irrespective of KDIGO class) (Table 29). In general, eGFR progression was calculated in the model taking eGFR slope change values per KDIGO class. Further, patients with health states G1 and G5 at baseline were not included in the EMPA-KIDNEY trial.

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 98 of 160

Table 29. Annual eGFR change in patients receiving empagliflozin and SoC

Abbreviations: CI, confidence interval; eGFR, estimated glomerular filtration rate; NA: not available; SoC, standard of care Significant differences are in bold. Source: EMPA-KIDNEY Trial data on file – Clinical trial documentation

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 99 of 160

Following treatment discontinuation: Figure 23 and Table 30 both show the distribution of eGFR decline (mL/min/1.73m[2] ) over 3 years compared to baseline observed in the CKD-PC cohort in Naimark et al. 2016 (140). Annual eGFR decline was estimated and fitted to a normal distribution to apply to annual cycles in the microsimulation model. The normal distribution is used to randomly sample and determine the annual eGFR for each patient per cycle in the model, thus simulating the heterogeneity in eGFR decline seen in a typical CKD cohort, with eGFR estimated in a prior cycle informing eGFR decline in the proceeding cycle.

Figure 23. Histogram showing distribution of eGFR slope in the CKD-PC population (Naimark et al)

==> picture [276 x 192] intentionally omitted <==

Abbreviations: CKD-PC, Chronic Kidney Disease Prognosis Consortium; eGFR: estimated glomerular filtration rate; HR, hazard ratio Source: Naimark et al. 2016 (140)

Table 30. Percentage of patients reporting specific eGFR slopes

*Proportion of patients obtained from Naimark et al. 2016 (140) have been changed to percentages Abbreviations: CKD-PC, Chronic Kidney Disease Prognosis Consortium; eGFR: estimated glomerular filtration rate Source: Naimark et al. 2016 (140)

The normal distribution applied in the engine has a mean and a standard deviation of -0.6 and 1.43, respectively. To determine the new eGFR value of the following cycle, the eGFR value of the previous cycle and the random eGFR decline are summed. When a patient initiates RRT, with either HD or PD, the progression of eGFR is assumed to remain constant. In cases where patients

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 100 of 160

have a successful kidney transplant, patients move to the KDIGO stage G3A1 and reinitiate eGFR and uACR decline and disease progression from that health state.

B.3.3.2.2 uACR progression

The change in the uACR values compared to baseline over time were derived from the EMPAKIDNEY trial and reported as ratios. uACR values were measured at 2, 18, 24, 30, and 36 months during the trial, values at 18 months were utilised to describe annual uACR progression in the model as 12 months uACR values were not available from the trial. uACR values were not one of the top 20 variables impacting the ICER in the one-way sensitivity analysis (OWSA), thus this assumption has limited impact on the results. The change in the uACR values in one year used in the model, are presented in Table 31.

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 101 of 160

Table 31. Change in the uACR values of empagliflozin and SoC over one year

Abbreviations: CI, confidence interval; NA: not available; SoC, standard of care; uACR, urine albumin-to-creatinine ratio. Significant differences are in bold. Source: EMPA-KIDNEY Trial data on file - Clinical trial documentation

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 102 of 160

Following treatment discontinuation: Coresh et al. 2019 (141) used patient-level data from eligible patients in the CKD-PC registry to assess the change in uACR values, with follow-up periods of one, two and three years. The study measured uACR changes by comparing the three-year values versus baseline.

The histogram presented in Figure 24 shows the uACR progression over time. The distribution of uACR fold change (shown in Figure 25) was fitted to a lognormal distribution in the model. This distribution enables random sampling and quantification of the uACR fold change for an individual patient, thus simulating the heterogeneity in uACR decline seen in a typical CKD cohort. To employ these random changes to annual cycles of the model, the cubic root of the value was determined. Each annual uACR fold change is then multiplied by the uACR value of the previous cycle to obtain the uACR value of the following cycle.

Figure 24. Histogram showing uACR fold change in the CKD-PC population

==> picture [289 x 191] intentionally omitted <==

Abbreviations: CKD-PC, Chronic Kidney Disease Prognosis Consortium; uACR: urine albumin-to-creatinine ratio Source: Coresh et al. 2019 (141)

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 103 of 160

Figure 25. Distribution of patients per uACR fold change

==> picture [363 x 226] intentionally omitted <==

----- Start of picture text -----
Histogram of uACR fold change, observed and predicted
0.3
0.25
0.2
0.15 Observed
Predicted
0.1
0.05
0
0 5 10 15 20
----- End of picture text -----

Abbreviations: uACR: urine albumin-to-creatinine ratio Source: Coresh et al. 2019 (141)

B.3.3.2.3 Progression of other risk factors

Progression of total cholesterol (TC), high-density lipoproteins (HDL), systolic blood pressure (SBP), glycated haemoglobin (HbA1c) and BMI is also incorporated in the model. These risk factors (along with eGFR and uACR) impact patient transition from one health state to another, as well as the probability of occurrence of complications in the model. Risk factor progression equations from the Framingham Heart Study [Wilson 1993 (158)] and United Kingdom Prospective Diabetes Study (UKPDS) 90 were used (159) to map the progression of these risk factors in the model.

For TC, HDL and SBP (irrespective of DM status), the Framingham progression equations were used which are based on two populations: the original 1948 Framingham cohort and offspring cohort. The individuals in original and offspring cohort returned for regular clinic visits after every two years and four years, respectively. Both these equations are applied up to 70 years in the model and no further progression is assumed after that. Table 32 demonstrates the coefficients of risk progression equations for TC, HDL and SBP by gender.

Table 32. Coefficients of the Framingham risk progression equations for TC, HDL, SBP (Framingham progression)

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 104 of 160

Note: The main driver of these equations is age. Abbreviations: HDL: high-density lipoprotein; SBP: systolic blood pressure; TC: total cholesterol. Source: Framingham Heart Study (158)

5,102 patients with newly diagnosed DM were followed up for a total of 30 years in UKPDS 90. UKPDS 90 progression equations were used to model HbA1c and BMI in patients with T2DM (159). The coefficients of these risk factor progression equations are described in Table 33.

Table 33. Coefficients of the risk progression equations for HbA1c and BMI in patients with DM (UKPDS 90)

Abbreviations: BMI: body mass index; DM, diabetes mellitus; Hb1Ac: glycated haemoglobin; SE: standard error * Three-year lag of Y for risk factors collected every three years. Source: UKPDS (159)

HbA1c is a variable included in QDiabetes-2018 prediction algorithm applied in the microsimulation to predict the risk of developing T2DM in patients with normal glucose tolerance (see Appendix P). As HbA1c change over time was not reported in EMPA-KIDNEY, the Framingham Offspring cohort [Pani et al. (160)] was used to model HbA1c over time in patients without DM. Based on this study, an increase of 0.014% in HbA1c per year is applied in the model. Table 34 shows the coefficients of risk progression equation for HbA1c. For patients without DM, it is considered that BMI progression follows a constant natural increase over time. In the model, a constant increase of 0.296 kg per year in the body weight until age 66 years is applied (161). After 66 years, a gradual decrease of 0.296 kg per year in the body weight is applied (162).

Table 34. Coefficients of the risk progression equation for HbA1c in patients without DM (Framingham Offspring cohort)

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 105 of 160

Abbreviations: DM, diabetes mellitus; SE: standard error Source: Pani et al. (160)

B.3.3.2.4 Assumptions used in the model for disease progression

The following assumptions were made for disease progression, which were adopted in consultation with clinical experts who were directly involved in the process of developing the empagliflozin microsimulation model:

• It was assumed that the eGFR change over time can be characterised by annual eGFR decline sampled within an observed distribution range would be applicable as an adequate representation of the disease progression to be applied in every cycle

• Following initiation of RRT, no further eGFR change was assumed in the model.

• It is assumed that patients having a successful kidney transplantation would move to the KIDIGO health state G3a A1

• For simplicity, it is assumed that the annual uACR fold change data collected by Coresh et al. 2019 (141) could also be applied at other time points, i.e., it is assumed that the effect of the time-variant exposure equals the effect of the time-invariant exposure.

• The progression of BMI is calculated in the model with the assumption that the height of the patients remains constant over time.

• The Framingham progression equations for TC, HDL and SBP are applied up to 70 years and no further progression is assumed after that.

• For individuals without DM, it is considered that BMI progresses following a constant natural increase in body weight of 0.296 kg per year. This increase is applied until the age of 66 years after which a gradual decrease in the body weight is noted. The decrease in weight applied after 66 years is again 0.296 kg per year. The progression of BMI is calculated using these parameters with the assumption that the height of the patients remains constant over time.

A list of other assumptions used in the model is provided in section B.3.9.2 Error! Reference source not found. .

B.3.3.3 Risk of complications

The risk of complications is based on the initial baseline characteristics and clinical risk factors of the patients. Patients are at risk of the same set of complications in any health state except death. The probability of patients experiencing any complication/ event per cycle is predicted by using either clinical data from literature (using transition probabilities or incidence rates) or commonly recognised predictive risk equations. A detailed account on the risk of complications considered in the model is provided in Appendix P. Risk of complications is modelled along with the risk of disease progression over a lifetime horizon until death.

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 106 of 160

B.3.3.4 Risk of all-cause mortality (death)

In the base-case of the model, the risk of all-cause mortality (ACM) was predicted using nonspecific cause of death plus CV death plus renal death:

1. Non-specific cause death

The non-specific cause death was estimated by subtracting CVD and renal failure deaths from the UK general population all-cause mortality. The general population mortality by age and gender was taken as per UK Office for National Statistics (ONS) life tables (163) and ONS lifetables selected are presented in Appendix Q.

2. CVD mortality

Predictive risk equations developed by the Matsushita et al 2020 study (164) were used to estimate risk of CVD mortality. Further details are discussed in Appendix P.

3. Renal death (mortality in patients on RRT)

The “renal death” tracker traces all the fatal events associated to PD, HD, and kidney transplant and moves patients to the death health state every time such an event occurs. The data for the renal death events has been sourced from the UKRR annual report (33). It is assumed that all renal deaths occurring are due to PD, HD, or kidney transplant as the UKRR annual report data does not indicate the cause of renal death. Appendix P details the risk of death as calculated from the UKRR report (33).

B.3.3.5 Adverse events

Lower limb amputations (leg, toe, and foot) sourced from the EMPA-KIDNEY trial are included in the base-case of the model. Lower limb amputations occurred more frequently in empagliflozin group (0.43 versus 0.29 events per 100 patient-years with placebo) (2). However, there were no statistically significant difference between the incidence rates in the empagliflozin group versus placebo. Table 35 shows the lower limb amputation events used in the model.

Table 35. Lower limb amputation rates per 100 patient-years

Source: EMPA-KIDNEY Trial data on file - Clinical trial documentation

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 107 of 160

B.3.3.6 Discontinuation rates

The discontinuation rate used for empagliflozin and placebo in the model was sourced from the EMPA-KIDNEY trial. The annual discontinuation rate of 12.56 and 14.16 per 100 patient-years was applied while on treatment with empagliflozin and placebo respectively (79). As described in section B.3.3.2 Error! Reference source not found. , after discontinuation of treatment in EMPAKIDNEY trial, the progression of patients was modelled using observational data reported in the CKD-PC or CRIC registries.

B.3.4 Measurement and valuation of health effects

B.3.4.1 HRQoL data from clinical trials

EMPA-KIDNEY trial utilities are utilised in scenario analysis and discussed in Appendix N. The base-case cost-effectiveness analysis utilises HRQoL data obtained from the HRQoL SLR, discussed in section Error! Reference source not found. and as described in detail in Appendix H.

A qualitative comparison of the HRQoL outcomes between trials of empagliflozin and dapagliflozin was conducted to further substantiate the cost comparison approach (only a comparison to dapagliflozin is relevant to the decision problem). As HRQoL data from DAPA-CKD was not reported publicly, this was performed using HRQoL data available from the pivotal trials supporting the HF with LVEF≥40% indications for both medications.

Both empagliflozin and dapagliflozin resulted in improvements in HRQoL in the HF with LVEF≥40%. The mean difference in change from baseline to 12 months in KCCQ-CSS was 1.32 (95%CI: 0.45-2.19) for empagliflozin (102), and the mean difference in change from baseline to 8 months in KCCQ-CSS was 2.3 (95%CI: 1.5-3.2) for dapagliflozin (12-month KCCQ data was not available) (165).

The comparison shows consistency between HRQoL results between empagliflozin and dapagliflozin and is also reflected by the similar efficacy and safety of the interventions, as described above in section B.2.9 Indirect and mixed treatment comparisons and independent sources (91).

B.3.4.2 Health state utilities derived from published evidence

Structured literature searches were conducted in October 2020 to identify and collate the utility and disutility inputs for the health states and events/complications associated with CKD to be used in the model. Full details of the search strategy, study selection process and results are presented

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 108 of 160

in Appendix H. MEDLINE[®] , Embase[®] , EconLit™ (via Ovid platform), NICE website, International Society for Pharmacoeconomics and Outcomes Research (ISPOR) website and Google Scholar were searched. The study selection process used pre-specified eligibility criteria to identify the relevant NICE TAs and studies using the PICOS framework, presented in Appendix H. A total of six TAs and 12 journal articles met the eligibility criteria. A full list of the included health state utility studies can be found in Table 14 and Table 15 in Appendix H.

In the base-case cost-effective analysis, utility weight inputs per health state (as per KDIGO classification) were required. Jesky 2016 (166) was selected for usage in the model as it met the criteria among all published evidence identified in the HRQoL SLR. As such, utilities are identical for the health states with same eGFR class in the base-case scenario (Table 36).

B.3.4.3 Mapping

No utility mapping was performed in the base-case scenario as the utilities are sourced from literature, specifically a study by Jesky et al (166) in which data were collected from participants using the EQ-5D-3L, and health states were converted into an EQ-5D[index] score using a set of weighted preferences produced from the UK population. In a scenario analysis [Section Error! Reference source not found. ], the literature derived health state utilities were replaced by baseline EMPA-KIDNEY trial utilities. The EQ-5D-5L descriptive system data collected from EMPA-Kidney trial were mapped onto the EQ-5D-3L value set using the mapping algorithm developed by the DSU based on the EEPRU dataset (167, 168). The UK-specific value sets proposed by Hernández-Alava et al. (2020) (168) were used to convert the five-digit EQ-5D health states into utility scores taking into account societal preferences for health (see Appendix N).

B.3.4.4 Complication related disutilities

The structured literature review, described in Appendix H, provided the disutility values for CVD, BMD, anaemia, AKI, and incident cancer shown in Table 36Table 36. Summary of utility values for cost-effectiveness analysis. These values were applied in the model only for the year in which the event occurs. The same disutility is applied in the event of recurrence. Disutilities for peritonitis, blood stream infections, and AV access thrombosis were not retrieved from literature. No disutility is applied for these complications, and they are assumed to be captured within utility for peritoneal or haemodialysis. Further disutilities for infections, metabolic events, DM, and hypertension have not been applied as they have been assumed to be included in the respective health state utility. Moreover, for events with a long-term impact, like stroke or myocardial infarction, a conservative approach is taken to only apply the disutility in year the event occurs, and not in the following years.

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 109 of 160

B.3.4.5 ACH related disutilities

Upon selection of the ACH module, the disutilities associated with acute events are superseded by a unique disutility applied every time an ACH event occurs in the model. In absence of data, the value used for ACH disutility was assumed equal to the utility loss with an MI (-0.06), applied during a 1-year cycle (169).

B.3.4.6 ESKD related utilities and disutilities

These utility inputs have mainly been sourced from Liem 2008 (170), Peasgood 2016 (171), and the technology assessment of dapagliflozin for treating CKD (6), as identified in the SLR described in Appendix H. The utility values used for ESKD in the model are provided in Table 36 and are applied as follows:

• For conservative therapy it is assumed to be same as G5 from the KDIGO classification

• For PD, HD and the first year of kidney transplant a state-specific annual utility is applied

• As patients move to G3A1 post kidney transplant, the utility of that group is used after year 1, however a disutility is applied to account for the immunosuppressive therapy in the follow-up years

B.3.4.7 AE related disutilities

These utility inputs have been sourced from Peasgood 2016 (171), as identified in the SLR described in Appendix H. Disutilities are applied for the acute event in the cycle it occurs in.

B.3.4.8 HRQoL data used in the cost-effectiveness analysis

Table 36Table 36 presents all HRQoL data used in the base-case of the cost-effectiveness analysis.

Table 36. Summary of utility values for cost-effectiveness analysis

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131]

© Boehringer Ingelheim Ltd (2023). All rights reserved Page 110 of 160

Abbreviations: ACH, all-cause hospitalisations; AE, adverse event; AKI, acute kidney injury; BMD, bone-andmineral disease; CHF, congestive heart failure; CI; confidence interval; CVD, cardiovascular disease; ESKD, endstage kidney disease; KDIGO, Kidney Disease Improving Global Outcomes; PAD, peripheral arterial disease; PVD, peripheral vascular disease; TIA, transient ischaemic attack

B.3.5 Cost and healthcare resource use identification, measurement, and valuation

A structured literature search was conducted to identify cost and resource use for the CKD health states and the relevant complications included in the cost-effectiveness model (CEM). Full details of the search strategy, study selection process and results are presented in Appendix I.

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 111 of 160

Data for cost inputs was searched in a five-step process (see Figure 3 in Appendix I): NICE TAs, NHS Reference Costs 2020/21, and databases MEDLINE[®] , Embase[®] , and EconLit™ (via Ovid platform) were searched, in addition to hand searching of ISPOR website and Google Scholar. The study selection process used pre-specified eligibility criteria to identify the NICE TAs and studies relevant to the decision problem and NICE reference case, using the PICOS framework presented in Table 18 in Appendix I.

A total of nine TAs and 16 journal studies were included. A full list of included TAs and studies are presented in Table 19 and Table 20 of Appendix I respectively. Costs for ACH, hyperphosphatemia, secondary hyperparathyroidism, and hypocalcaemia were not obtained from the structured literature search and expert clinical opinion informed costings (Appendix O).

All costs applied in the model were inflated to a 2020/21 cost-year, based on the Hospital and Community Health Services (HCHS) Pay and Prices inflation index (up to and including 2007/08), the HCHS index (between 2008/09 and 2014/15), and the NHS Cost Inflation Index (NHSCII) (from 2015/2016 onwards), as reported in the relevant Personal Social Services Research Unit (PSSRU) publications. No inflation was applied to NHS Reference Costs 2020/21. Please see Appendix Q for details on the inflation indices used.

B.3.5.1 Intervention and comparators’ costs and resource use

The cost-effectiveness analysis compares empagliflozin 10mg OD on top of SoC against matching placebo on top of SoC, in line with the EMPA-KIDNEY trial. As per sections B.1.3.3 Current clinical pathway of CKD in the UK and B.1.3.4 Limitations of current pathway and unmet need, SoC is individually optimised and can include ACE inhibitors, ARBs, and CVD medications dependant on patients individual characteristics.

The average annual SoC cost was based on TA775 for dapagliflozin (6), which calculated a weighted average of the most frequently prescribed ACE inhibitor (ramipril), ARBs (losartan and irbesartan), statin (atorvastatin), and antiplatelet medication (aspirin) in CKD. Prices were updated using the British National Formulary (BNF) 2022 (177-181) to determine annual cost of SoC (Table 37).

Table 37. Calculation of SoC weighted average cost

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 112 of 160

Source: TA775 (6); Ramipril BNF (177); Losartan BNF (178); Irbesartan BNF (179); Atorvastatin BNF (180); Aspirin BNF (181)

The treatment cost values used in the base-case analysis are summarised below in Table 38. Treatment costs are applied in the first and subsequent years until discontinuation due to 1) treatment discontinuation, 2) initiation of RRT, or 3) death.

Table 38: Cost of treatment with empagliflozin and SoC

Abbreviations: BNF, British National Formulary; CKD, chronic kidney disease; SoC, standard of care

B.3.5.2 Health state unit costs and resource use

The model incorporates annual CKD health state costs, as well as event-driven costs for CKD complications. The cost components are broadly categorised as:

1. Health state costs – annual maintenance/monitoring costs per KDIGO category

2. Complications – include first year and follow-up costs for the long-term complications

   • First year costs apply to the initial resources used in the year the acute complication event occurs

   • Follow-up costs apply to ongoing resource usage in successive annual cycles whilst the complication persists (e.g., the CV events)

3. Event costs – for one-time acute events considered in the model (e.g., fractures, infections)

4. Chronic costs – apply to chronic CKD complications like anaemia with a constant cost applied

In the base-case analysis, health state costs, first year, and follow-up complication costs are added together to get the total costs per submodule. Costing methodology is summarised below.

Annual health state costs

Annual health state costs were sourced from Pollock et. al 2022 (20), which reported annual healthcare cost per patient by KDIGO stage (G2 to G5) and cost category (i.e., critical care, outpatient visits, general physician visits, emergency room visits, hospitalisation, and ambulance use) (Figure 26). Hospitalisation and critical care costs were excluded from annual health state costs in the cost-effectiveness model, as these are already accounted for in complications submodules. G1 costs were assumed be equal to G2 costs across albuminuria levels, as G1 costs were not directly reported in this paper (20). See Table 39 for the values used in the base-case analysis, which were applied as maintenance costs per health state. Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 113 of 160

Figure 26. Mean annual per patient healthcare costs for the overall CKD cohort, stratified by KDIGO classification group

==> picture [400 x 286] intentionally omitted <==

Abbreviations: CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; ER, emergency room; GP, general physician; KDIGO, Kidney Disease Improving Global Outcomes; uACR, urine albumin-to-creatinine ratio. Source: Pollock 2022 (20)

Table 39. Maintenance costs per health state as per KDIGO classification (hospitalisation and critical care costs excluded)

Abbreviations: KDIGO, Kidney Disease Improving Global Outcomes; SE, standard error Source: Pollock 2022 (20)

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 114 of 160

Cost of CVD

For all CVD complications, a first year (acute) cost is applied in the cycle when the complication occurs, and follow-up costs applied in successive annual cycles whilst the complication persists. Unit costs for acute and follow-up CVD complications are outlined in Table 40. Acute costs for CVD complications were sourced from NHS 2020/21 reference costs (latest available version in Q4 2022), where weighted averages of non-elective long stay inpatient costs were calculated using the relevant Healthcare Resource Group (HRG) codes (182).

Follow-up costs for CVD complications were sourced as follows: Danese 2016 which considered the cost of drugs, hospitalisations, and visits for stroke, TIA and unstable angina (183); TA773 which considered the cost of GP visits, cardiologist visits, and A&E referral for HF (4); Sundström 2022 which considered the cost of annual hospital healthcare for PAD and PVD (184); and TA10820 for MI (185).

Table 40. Acute event and follow-up costs for management of CVD complications

Abbreviations: CHF, congestive heart failure; CVD, cardiovascular disease; MI, myocardial infarction; PAD, peripheral arterial disease; PVD, peripheral vascular disease; TIA, transient ischaemic attack

Cost of BMD

An event cost is applied in each cycle a patient has a BMD event. BMD costs are summarised in Table 41. The event costs of hyperphosphatemia, secondary hyperparathyroidism, and hypocalcaemia were calculated from the BNF 2022 (188-195) using the average prices of the drugs administered for treatment (phosphate binders for hyperphosphatemia/hypocalcaemia, Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 115 of 160

vitamin D analogues and calcimimetics for secondary hyperparathyroidism) and average prescribing distributions based on UK clinical guidelines and verified by clinical expert opinion (Appendix O). Costs of managing fractures were sourced from NHS 2020/21 reference cost data using a weighted average of the total cost for the relevant HRG codes (182).

Table 41. Cost of BMD

Abbreviations: BMD, bone and mineral disorders; BNF, British National Formulary; NHS, National Health Service

Cost of AKI

An event cost is applied in each cycle a patient has an AKI event and includes first and recurrent hospitalisations. A weighted average of relevant HRG codes from the NHS 2020/21 reference cost data was calculated for this input (182) (Table 42). Based on interviews with UK nephrologists, it was assumed reasonable to only include an AKI hospitalisation cost since AKI not requiring hospitalisation (i.e., outpatient AKI) is unlikely to require additional appointments and tests outside of what is normal for patients with CKD (Appendix O).

Table 42. Cost of AKI

Abbreviations: AKI, acute kidney injury; NHS, National Health Service.

Cost of infections

An event cost is applied in each cycle an infection event occurs. Cost of infections are listed in Table 43. Gastrointestinal, muscular, nervous system, and sepsis infection costs were sourced from the NHS 2020/21 reference cost data (182). Gastrointestinal infections costs considered nonadmitted face-to-face attendance and follow-up visits for gastroenterology. Muscular infection, nervous system infection and sepsis considered the weighted non-elective long stay costs. Respiratory infection costs were sourced from Kohli 2021 which considered outpatient care of respiratory infection in at risk patients aged 65-74 (196). Skin and soft tissue infection costs were sourced from Humphreys 2023 which considered the mean total HCRU cost of cellulitis for patients aged 61-74 (197). The event cost of urinary tract infection was sourced from TA775 which considered the cost of one GP consultation lasting 9.22 minutes (6).

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 116 of 160

Table 43. Cost of infections

Abbreviations: CKD, chronic kidney disease; NHS, National Health Service

Cost of incident cancer

A one-time event cost is applied in the first annual cycle an incident cancer complication occurs. Costs of renal or urothelial cancer (which patients with CKD are at an increased risk of) are shown in Table 44. The cost for renal cancer was sourced from Amdahl 2017 which considered the annual cost for metastatic renal cell carcinoma (198). The cost for urothelial cancer was sourced from Sangar 2004 which considered the annual cost for prostate and bladder cancer (199).

Table 44. Cost of incident cancer

Cost of other complications

For metabolic acidosis, hyperkalaemia, and hyperuricaemia (gout), an event cost is applied in each cycle a patient has the event. For anaemia, annual cycle costs are applied from its occurrence until renal transplant or death as this is a chronic condition. Costs of other CKD related complications are shown in Table 45. The annual cost of metabolic acidosis was sourced from Witham 2020 who considered the annual cost of sodium bicarbonate therapy (130). The annual cost of hyperuricemia/gout was sourced from Morlock 2016 who considered the total healthcare costs in gout patients (130). The event cost of hyperkalaemia was calculated as a weighted average of non-elective long/short stay costs using the relevant HRG codes from NHS 2020/21 reference cost data (182). The annual cost of hyperuricemia/gout was sourced from Morlock 2016 who considered the total healthcare costs in gout patients (130). The chronic cost of anaemia was sourced from TA481 (200).

Table 45. Cost of other complications

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 117 of 160

Abbreviations: NHS, National Health Service

Cost of ESKD

ESKD costs include costs of RRT (PD, haemodialysis, or kidney transplant) and dialysis complications, or conservative therapy. PD includes CAPD or APD. For PD and haemodialysis, annual cycle costs are applied whilst patients remain on therapy. An event cost is applied in each cycle patients experience a dialysis complication (peritonitis, AV access thrombosis, bloodstream infections). For kidney transplant, costs are applied in the annual cycle of the transplant and followup immunosuppression costs are applied in subsequent annual cycles until graft rejection or death due to any cause. For PD and haemodialysis, the cost per cycle is applied every year that the patient is on therapy. For kidney transplant, the intervention cost is applied in the year of the event and follow-up costs are applied in subsequent years until graft rejection or death due to any cause. The cost of conservative therapy is applied annually whist patients remain on therapy. To avoid double counting, costs of RRT or conservative therapy are not applied to patients in the G5 health state.

Unit costs for ESKD are listed in Table 46. Annual cost of CAPD and APD were sourced from NHS 2020/21 reference cost data using relevant HRG codes and frequencies of sessions (182, 202). Cost of haemodialysis was sourced from TA10820 which considered the cost and frequency of haemodialysis sessions (185). Costs of AV access thrombosis and peritonitis were sourced from the NHS 2020/21 reference cost data using the relevant HRG codes (182). For AV thrombosis a weighted average of the total costs was calculated and for peritonitis the weighted average of nonelective long stay costs was calculated. Cost of bloodstream infections was sourced from Manoukian 2021 (203).

Kidney transplant costings followed assumptions from TA775 i.e., HRG codes from 2020/21 NHS reference cost data for pre-transplant, transplant, and post-transplant care, however costs for living and deceased donor transplants were considered separately. The weighted averages of the relevant HRG codes for the three components of kidney transplant costs were summed to determine a total cost (6, 182). Cost of immunosuppressive therapy to avoid transplant rejection is also applied for annual cycles post successful kidney transplant (6). The annual cost of conservative therapy for ESKD was sourced from Agus et. al 2017, which considered management costs for ESKD in patients refusing dialysis (204).

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 118 of 160

Table 46. Cost of ESKD

Abbreviations: AKI, acute kidney injury; APD, automated peritoneal dialysis; AV, arteriovenous; CAPD, continuous ambulatory peritoneal dialysis; CKD, chronic kidney disease; NHS, National Health Service; PD, peritoneal dialysis.

B.3.5.3 Adverse events unit costs and resource use

As described in section B.3.3.5 three adverse events (leg, foot, and toe amputation) were included in the model. An event is applied in each cycle patients experience adverse events. Amputation costs were sourced from relevant HRG codes from NHS 2020/21 reference costs (182) (Table 47).

Table 47. Cost of managing lower limb amputations

Abbreviations: NHS, National Health Service

B.3.5.4 Summary of cost application in the model

The cost application method for each health state and complication submodule in the base-case cost-effectiveness analysis are presented in Table 48.

Table 48 Submodule-wise application of costs in the model

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 119 of 160

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 120 of 160

Abbreviations: ACE, angiotensin-converting enzyme; AKI, acute kidney injury; ARB, angiotensin receptor blockers; AV, arteriovenous; CKD, chronic kidney disease; DM, diabetes mellitus; EMPA, empagliflozin; ESKD, end-stage kidney disease; HD, haemodialysis; HS, health state; KDIGO, Kidney Disease: Improving Global Outcomes; KT, kidney transplant; MI, myocardial infarction; PAD, peripheral arterial disease; PD, peritoneal dialysis; PVD, peripheral vascular disease; RRT, renal replacement therapy; TIA, transient ischaemic attack; UA, unstable angina

B.3.5.5 Miscellaneous unit costs and resource use

The empagliflozin cost-effectiveness model includes an alternative ACH scenario in which cost of acute events including hospitalisations are turned off and replaced by costs for ACH (Appendix P). An ACH cost of £4,554 was calculated as a weighted average of all HRG non-elective long stay costs, excluding codes for patients under 18 and for obstetric procedures, using NHS 2020/21 reference cost data (182). This method was deemed appropriate through clinical expert opinion (Appendix O). The ACH option is not included in the base case scenario.

B.3.6 Severity

It is not anticipated that empagliflozin would qualify for a severity modifier in this indication.

B.3.7 Uncertainty

BI believes there is no aspect of the condition or technology presented in this submission that would impact the ability to generate high-quality evidence.

B.3.8 Managed access proposal

Not applicable. No patient access scheme is expected as the cost per QALY is low (dominant).

B.3.9 Summary of base-case analysis inputs and assumptions

B.3.9.1 Summary of base-case analysis inputs

An overview of the base-case cost-effectiveness analysis settings and variables is provided in Table 49 and Table 50 respectively.

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 121 of 160

Table 49. Base-case model settings

Abbreviations: BC, base-case; CVD, cardiovascular disease; DM, diabetes mellitus; eGFR: Estimated glomerular filtration rate; ESKD, end-stage kidney disease; KDIGO, Kidney Disease Improving Global Outcomes; PSA, probabilistic sensitivity analysis; RRT, renal replacement therapy; RT, renal transplant; SoC, standard of care; uACR, urine albumin-to-creatinine ratio

Company evidence submission for empagliflozin for treating chronic kidney disease [ID6131] © Boehringer Ingelheim Ltd (2023). All rights reserved Page 122 of 160

Table 50. Base-case variables applied in cost-effective analysis